FRACTURE OF NANO AND ENGINEERING MATERIALS

AND STRUCTURES
Fracture of Nano and Engineering
Materials and Structures
Proceedings of the 16th European Conference of
Fracture, Alexandroupolis, Greece, July 3-7, 2006

Edited by

E. E. GDOUTOS
Democritus University of Thrace,
Dept of Civil Engineering, Xanthi, Greece
A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN-10 1-4020-4971-4 (HB)

ISBN-13 978-1-4020-4971-2 (HB)
ISBN-10 1-4020-4972-2 ( e-book)
ISBN-13 978-1-4020-4972-9 (e-book)

Published by Springer,
P.O. Box 17, 3300 AA Dordrecht, The Netherlands.

www.springer.com

Cover picture
Fracture and Delamination of Oxide: Fracture and delamination of 1µm (1x10–6 m)
SiO2 on Si with 1µm conical probe tip. Courtesy of Hysitron Inc., Minneapolis, Minnesota, USA

Printed on acid-free paper

All Rights Reserved

© 2006 Springer
No part of this work may be reproduced, stored in a retrieval system, or transmitted
in any form or by any means, electronic, mechanical, photocopying, microfilming, recording
or otherwise, without written permission from the Publisher, with the exception
of any material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work.

Printed in the Netherlands.

 Contents

Editor’s Preface............................................................................................................ xliii

Organizing Committees ................................................................................................ xlv
ECF16 TRACKS......................................................................................................... xlvii
ECF16 SPECIAL SYMPOSIA/SESSIONS .............................................................. xlvix

A. INVITED PAPERS ........................................................................................... 1

Deformation and Fracture at the Micron and Nano Scales.............................................. 3
E. C. Aifantis
Statistical Mechanics of Safety Factors and Size Effect in Quasibrittle Fracture............ 5
Z. P. Bazant and S.-D. Pang
Nanoreliability – Fracture Mechanics on the Way from Micro to Nano ......................... 7
B. Michel
Fracture Mechanics and Complexity Sciences ................................................................ 9
A. Carpinteri and S. Puzzi
Failure of Composite Materials...................................................................................... 11
I. M. Daniel
Interactions of Constrained Flow and Size Scale on Mechanical Behavior .................. 13
W. W. Gerberich, W. M. Mook, M. J. Cordill and D. Hallman
Space Shuttle Columbia Post-Accident Analysis and Investigation ............................. 15
S. McDanels
The Role of Adhesion and Fracture on the Performance of Nanostructured Films....... 17
N. Moody, M. J. Cordill, M. S. Kennedy, D. P. Adams, D. F. Bahr
and W. W. Gerberich
Assessment of Weldment Specimens Containing Residual Stress ................................ 19
K. M. Nikbin
MEMS: Recent Advances and Current Challenges ....................................................... 21
R. J. Pryputniewicz
Fracture, Aging and Disease in Bone and Teeth ............................................................ 23
R. O. Ritchie and R. K. Nalla
Laboratory Earthquakes ................................................................................................. 25
A. J. Rosakis, K. Xia and H. Kanamori
A Historical Retrospective of the Beginnings of Brittle Fracture Mechanics -
The Period 1907-1947.................................................................................................... 27
H. P. Rossmanith
Dynamic Crack Propagation in Particle Reinforced Nanocomposites and Graded
Materials......................................................................................................................... 29
A. Shukla
Spatial and Temporal Scaling Affected by System Inhomogeneity: Atomic,
Microscopic and Macroscopic. ...................................................................................... 31
G. C. Sih
vi Contents

B. TRACKS
B1. Nanomaterials and Nanostructures ........................................................ 33

1T1. Fracture and Fatigue of Nanostuctured Materials .......................................... 35

Channeling Effect in Fracture of Materials with Nanostructured Surface Layers......... 35
V. E. Panin and A. V. Panin
Atomistics and Configurational Forces in Gradient Elasticity ...................................... 37
P. Steinmann and E. C. Aifantis
Tensile Behavior and Fracture of Carbon Nanotubes Containing Stone-Wales Defects 39
K. I. Tserpes and P. Papanikos
Atomic-Scale Investigation on Fracture Toughness in Nano-Composite
Silicon Carbide............................................................................................................... 41
M. Ippolito, A. Mattoni, L. Colombo and F. Cleri
Multiscale Modeling and Computer Simulation of Stress-Deformation Relationships
in Nanoparticle-Reinforced Composite Materials ......................................................... 43
L. V. Bochkaryova, M. V. Kireitseu, G. R. Tomlinson, V. Kompis
and H. Altenbach
The Mechanical Parameters of Nanoobjects (Theory and Experiment) ........................ 45
E. Ivanova, N. Morozov and B. Semenov
Advanced Manufacturing Design Concepts and Modelling Tools of the Next
Generation Nanoparticle-Reinforced Damping Materials ............................................ 47
M. V. Kireitseu, G. R. Tomlinson, R. A. Williams and V. Kompis
Fracture of Nanostructured Ionomer Membranes.......................................................... 49
Yue Zou, X. Huang and K. L. Reifsnider

1T2. Failure Mechanisms ............................................................................................ 51

Deformation and Limit States of Carbon Nanotubes under Complex Loading............. 51
A. V. Chentsov and R. V. Goldstein
Interaction of Domain Walls with Defects in Ferroelectric Materials ........................... 53
D. Schrade, R. Mueller, D. Gross, T. Utschig, V. Ya. Shur, D. C. Lupascu
Microstructure and Internal Stresses in Cyclically Deformed Al and Cu Single
Crystals........................................................................................................................... 55
M. E. Kassner
Determination of Equilibrium Configurations of Atomic Lattices at Quasistatic
Deformation ................................................................................................................... 57
S. N. Korobeynikov
Multiscale Mechanics of Carbon Nanotubes and their Composites .............................. 59
X.-Q. Feng

1T4. Fatigue and Fracture of MEMS and NEMS..................................................... 61

In-Situ Scanning Electron Microscope Indentation of Gallium Arsenide ..................... 61
C. Pouvreau, K. Wasmer, J. Giovanola, J. Michler, J. M. Breguet
and A. Karimi
Fracture of Nanostructured Lithium Batteries ............................................................... 63
K. E. Aifantis, J. P. Dempsey and S. A. Hackney
Analytical and Experimental Characterization of a Micromirror System ..................... 65
E. J. Pryputniewicz, C. Furlong and R. J. Pryputniewicz
 Contents vii

A Metal Interposer for Isolating MEMS Devices from Package Stresses ..................... 67
R. J. Pryputniewicz, T. F. Marinis, J. W. Soucy, P. Hefti and A. R. Klempner
Computational Modeling of Nanoparticles in Biomicrofluidic Devices ....................... 69
R. J. Pryputniewicz, Z. Sikorski, M. Athavale, Z. J. Chen and A. J. Przekwas
Characterization of a MEMS Pressure Sensor by a Hybrid Methodology .................... 71
R. J. Pryputniewicz and C. Furlong
New Approach to Synthesis of Laser Microwelding Processes for Packaging ............. 73
R. J. Pryputniewicz, W. Han and K. A. Nowakowski
Thermal Management of RF MEMS Relay Switch....................................................... 75
R. J. Pryputniewicz

1T7. Thin Films ............................................................................................................ 77

Buckling and Delamination of Thin Layers on a Polymer Substrate ............................ 77
A. A. Abdallah, D. Kozodaev, P. C. P. Bouten, J. M. J. den Toonder
and G. De With
Carbide Coated Cutting Tool Properties Investigation by Nano-Mechanical
Measurements under 250-500°C.................................................................................... 79
B. Vasques, D. Joly, R. Leroy, N. Ranganathan and P. Donnadieu
Diamond Coating Debounding in Tool Application ...................................................... 81
D. Moulin, P. Chevrier, P. Lipinski and T. Barré
Interfacial Strength of Ceramic Thin Film on Polymer Substrate ................................. 83
M. Omiya and K. Kishimoto
Delaminate Behavior of PVD/CPVD Thin Film ........................................................... 85
S. Doi and M. Yasuoka
Experimental Study of Microhardness and Fracture of Implanted Gállium
Nitride Films .................................................................................................................. 87
P. Kavouras, M. Katsikini, E. Wendler, W. Wesch, H. M. Polatoglou,
E. C. Paloura, Ph. Komninou and Th. Karakostas

1T9. Failure of Nanocomposites ................................................................................. 89

Crack Tip Strain Field and its Propagation Characteristics in a Polymer Foam............ 89
F.-P. Chiang, S. Chang and Y. Ding
How to Toughen Ceramics – Nanocomposites .............................................................. 91
H. Awaji and S.-M. Choi
Deformation and Fracture Behaviour of Nanocomposites ............................................ 93
S. Dunger, J. K. W. Sandler, K. Hedicke and V. Altsadt
Fracture Mechanisms in Carbon Nanotube-Reinforced Composites ............................ 95
E. T. Thostenson and T.-W. Chou
viii Contents

B. TRACKS
B2. Engineering Materials and Structures .................................................. 97

2T1. Physical Aspects of Fracture .............................................................................. 99

Fractal Approach to Crack Problems with Non-Root Singularity ................................. 99
A. Kashtanov
New Method for Analysing the Magnetic Emission Signals During Fracture ............ 101
Gy. B. Lenkey, N. Takacs, F. Kun and D. L. Beke
Electromagnetic Radiation Method for Identification of Multi-Scale Fracture .......... 103
Yu. K. Bivin, A. S. Chursin, E. A. Deviatkin and I. V. Simonov
Micromechanical Modeling of Grain Boundary Resistance to Cleavage Fracture
Propagation .................................................................................................................. 105
M. Stec and J. Faleskog
Microstructure of Reactor Pressure Vessel Steel Close to the Fracture Surface.......... 107
M. Karlik, P. Hausild and C. Prioul

2T2. Brittle Fracture.................................................................................................. 109

Brittle Fracture in Heat-Affected Zones of Girth Welds of Modern Line Pipe Steel
(X100) .......................................................................................................................... 109
A. S. Bilat, A. F. Gourgues-Lorenzon, J. Besson and A. Pineau
Cleavage Fracture of Steels at Very Low Temperatures .............................................. 111
R. Rodriguez-Martin, I. Ocana and A. Martin-Meizoso
New formulation of the Ritchie, Knot and Rice Hypothesis ....................................... 113
A. Neimitz, M. Graba and J. Galkiewicz
The Effect of the Rate of Displacement on Crack Path Stability............................... 115
D. A. Zacharopoulos and P. A. Kalaitzidis
Scratching and Brittle Fracture of Semiconductor In-Situ Scanning Electron
Microscope................................................................................................................... 117
K. Wasmer, C. Pouvreau1, J. Giovanola and J. Michler
Cracks in Thin Sheets: when Geometry Rules the Fracture Path ................................ 119
P. M. Reis, B. Audoly and B. Roman
Cleavage Mechanisms in a Ship Plate Steel ................................................................ 121
R. Cuamatzi, I. C. Howard and J. Yates

2T3. Ductile Fracture................................................................................................. 123

Failure Behavior of Hybrid-Laser Welds..................................................................... 123
A. Bajric and W. Dahl
Fracture of Plastic Bodies. Deformations Concentrators............................................. 125
A. I. Khromov, A. A. Bukhanko, S. L. Stepanov and E. P. Kocherov
3D Ductile Tearing Analyses of Bi-Axially Loaded Pipes with Surface Cracks ........ 127
A. Sandvik, E. Ostby and C. Thaulow
New Model Materials for Ductile Fracture Studies .................................................... 129
A. Weck and D. S. Wilkinson
Fatigue Threshold Computation Model Based on the Shakedown Analysis .............. 131
M. A. Belouchrani, D. Weichert and A. Hachemi
Void Coalescence in Metals Involving Two Populations of Cavities .......................... 133
D. Fabregue and T. Pardoen
 Contents ix

Effects of Characteristic Material Lengths on Ductile Crack Propagation.................. 135

E. Radi
Ductile Fracture by Void Nucleation at Carbides ........................................................ 137
J. Giovanola, D. Cannizzaro, R. Doglione and A. Rossol
The Significance of Maximum Load on a Load-Displacement Curve with Stable
Crack Extension ........................................................................................................... 139
J. R. Donoso and J. D. Landes
3D Visualization of Ductile Fracture using Synchrotron X-Ray Computer
Tomography ................................................................................................................ 141
L. Qian, H. Toda, T. Ohgaki, K. Uesugi, M. Kobayashi and T. Kobayashi
Non-Local Plastic-Damage Model for Failure Analysis of Sheet-Metals ................... 143
M. Brunet, F. Morestin and H. Walter-Leberre
A Novel Technique for Extracting Stretch Zone Features From Fractographs ........... 145
M. Tarafder, Swati Dey, S. Sivaprasad and S. Tarafder
Simulation of Fatigue Crack Growth by Crack Tip Blunting ...................................... 147
P. Hutar and M. Sauzay
Loading Rate Effect on Ductile Fracture ..................................................................... 149
R. Chaouadi
Experimental Investigation of Slant Crack Propagation in X100 Pipeline Steel......... 151
S. H. Hashemi, I. C. Howard, J. R. Yates, and R. M. Andrews

2T4. Nonlinear Fracture Mechanics......................................................................... 153

Esis TC8 – Numerical Round Robin on Micro Mechanical Models : Results of
Phase III for the Simulation of the Brittle to Ductile Transition Curve....................... 153
C. Poussard and C. Sainte Catherine
Closure of a Rectangular Skin Defect via the Advancement Flap............................... 155
C. Antypas, C. Borboudaki, V. Kefalas and D. A. Eftaxiopoulos
Similarity Solutions of Creep – Damage Coupled Problems in Fracture Mechanics.. 157
L. V. Stepanova and M. E. Fedina
Impact Fracture Toughness Determination of Ductile Polymers by SPB Method ...... 159
J. Wainstein , L. A. Fasce and P. M. Frontini
A Micro-Toughness Model for Ductile Fracture ......................................................... 161
K. Srinivasan, T. Siegmund and O. Kolednik

2T5. Fatigue and Fracture......................................................................................... 163

Crack Coalescence Modelling of FSW Joints.............................................................. 163
A. Ali, M. W. Brown and C. A. Rodopoulos
Fatigue Crack initiation in a Two Phase B-Metastable Titanium Alloy: Influence
of Microstructural Parameters..................................................................................... 165
A. Lenain, P. J. Jacques and T. Pardoen
Effects of Specimen Type, Size and Measurement Techniques on FCGR .................. 167
B. Kumar and J. E. Locke
The Effect of Stress Ratio on Fatigue Short Cracking ................................................. 169
C. A. Rodopoulos and S.-H. Han
Dwell-fatigue Behaviour of a Beta-Forged Ti 6242 Alloy .......................................... 171
P. Lefranc, C. Sarrazin-Baudoux and V. Doquet
Investigation into Fatigue Life of Welded Chemical Pipelines ................................... 173
x Contents

Cz. Goss and L. Sniezek

Different Analytical Presentations of Short Crack Growth under Rotation-Bending
Fatigue ......................................................................................................................... 175
D. Angelova and A. Davidkov
Variable Amplitude Load Interaction in Fatigue Crack Growth for 2024-T3
Aluminium Alloy ........................................................................................................ 177
D. Kocanda, S. Kocanda and J. Torzewski
An Investigation on the Fatigue Performance of Hydraulic Gate Wheels................... 179
D. Polyzois and A. N. Lashari
A Micromechanical Model for Crack Initiation in High Cycle Fatigue of Metallic
Materials ...................................................................................................................... 181
V. Monchiet, E. Charkaluk and D. Kondo
Comparative Analysis of Two Models for Evaluating Fatigue Data ........................... 183
E. Castillo, A. Ramos, M. Lopez-Aenlle, A. Fernandez-Canteli and R. Koller
Assessment of Damage at Notch Root of Thick Plates ............................................... 185
E. C. G. Menin and J. L. de A. Ferreira
Fatigue Strength Prediction of Spot-Welded Joints Using Small Specimen Testing... 187
E. Nakayama, M. Fukumoto, M. Miyahara, K. Okamura, H. Fujimoto
and K. Fukui
A Thermo-Mechanical Model for Random Braking of Machine Components ........... 189
F. Loibnegger, H. P. Rossmanith and R. Huber
Lifetime Calculation of Railway Wheel Steels Based on Physical Data ..................... 191
F. Walther and D. Eifler
Fatigue Crack Propagation of Super-Duplex Stainless Steel at Different
Temperatures ................................................................................................................ 193
G. Chai and S. Johansson
Transitions of Fatigue Crack Initiation From Surface, Subsurface to SNDFCO......... 195
G. Chai
Surface Fatigue of Gear Teeth Flanks.......................................................................... 197
G. Fajdiga, M. Sraml and J. Flasker
Fatigue and Fracture Processes in High Performance PM Tool Steels ........................ 199
G. Jesner, S. Marsoner, I. Schemmel and R. Pippan
Notch and Defect Sensitivity of ADI in Torsional Fatigue.......................................... 201
B. Atzori and G. Meneghetti
Multi Axial Fatigue in Welded Components ............................................................... 203
G. Mesmacque, B. Wu, C. Robin, D. Zakrzewski and X. Decoopman
Enhanced Fatigue Life by Mechanical Surface Treatments – Experiment and
Simulation .................................................................................................................... 205
H. P. Gaenser, I. Goedor, H. Leitner and W. Eichlseder
Analysis of Repaired Aluminum Panels in General Mixed-Mode Conditions............ 207
H. Hosseini-Toudeshky, M. Saber and B. Mohammadi
Effect of Strain Rate on Fatigue Behavior of Ultrafine Grained Copper..................... 209
P. Gabor, H. J. Maier and I. Karaman
Lubricant Effects on Propagation of Surface-Breaking Cracks under Rolling
Contact Loading........................................................................................................... 211
J. Lai and S. Ioannides
 Contents xi

Computational Modelling of Crack Initiation in a Mixing Tee Subjected to Thermal

Fatigue Load ................................................................................................................ 213
I. Varfolomeyev
Estimation of Critical Stress Intensity Factor in Steel Cracked Wires......................... 215
J. Toribio, F. J. Ayaso, B. Gonzalez, J. C. Matos and D. Vergara
Low-Cycle Fatigue of Din 1.2367 Steels in Various Treatments................................. 217
C. C. Liu, J. H. Wu and C. C. Kuo
Impact Testing a Capable Method to Investigate the Fatigue Resistance.................... 219
K. David, P. Agrianidis, K. G. Anthymidis and D. N. Tsipas
Comparative Assessment of Fatigue-Thresholds Estimated by Short and
Long Cracks ................................................................................................................. 221
K. K. Ray, N. Narasaiahb and S. Tarafderb
Scanning Electron Microscope Measurements of Crack-Opening Stress on Fatigue
Cracks Exposed to Overloads ...................................................................................... 223
L. Jacobsson and C. Persson
Propagation Path and Fatigue Life Predictions of Branched Cracks under Plane
Strain Conditions.......................................................................................................... 225
M. A. Meggiolaro, A. C. O. Miranda, J. T. P. Castro and L. F. Martha
Short Crack Equations to Predict Stress Gradient Effects in Fatigue .......................... 227
M. A. Meggiolaro, A. C. O. Miranda, J. T. P. Castro and J. L. F. Freire
Fatigue Behaviour of Pre-Strained Type 316 Stainless Steel...................................... 229
M. Akita, M. Nakajima, K. Tokaji and Y. Uematsu
The Influence of Constraint on Fitting Fatigue Crack Growth Data ........................... 231
M. Carboni and M. Madia
Atomic Force Microscopy of Local Plastic Deformation for Tempered Martensite ... 233
M. Hayakawa, S. Matsuoka and Y. Furuya
Improvement of Fatigue Strength due to Grain Refinement in Magnesium Alloys .... 235
M. Kamakura, K. Tokaji, H. Shibata and N. Bekku
A Unified Fatigue and Fracture Model Applied to Steel Wire Ropes ......................... 237
M. P. Weiss, R. Ashkenazi and D. Elata
Correlation Between Paris’ Law Parameters Based on Self-Similarity and Criticality
Condition...................................................................................................................... 239
A. Carpinteri and M. Paggi
Thermo-Mechanical Fatigue Lifetime Assessment with Damage-Parameters,
Energy-Criterions and Cyclic-J-Integral Concepts ...................................................... 241
M. Riedler, R. Minichmayr, G. Winter and W. Eichlseder
Predicting Fatigue Crack Retardation Following Overload Cycles............................. 243
M. V. Pereira, F. A. Darwish, A. F. Camarao and S. H. Motta
Fatigue Crack Growth at Notches Considering Plasticity Induced Closure ................ 245
J. Bruening, O. Hertel, M. Vormwald and G. Savaidis
Influence of Microstructure on Fatigue Properties of Ni-Base Superalloy at
Elevated Temperature .................................................................................................. 247
Qy. Wang , Y. Matsuyama, N. Kawagoishi, M. Goto and K. Morino
Moddelling Fatigue Crack Closure using Dislocation Dipoles ................................... 249
P. F. P. de Matos and D. Nowell
Comparison Between Fatigue Crack Growth Modelled by Continous Dislocation
Distributions and Discrete Dislocations ...................................................................... 251
xii Contents

P. Hansson, S. Melin and C. Persson

Fatigue Evaluation Considering the Environmental Influence Using a Monitoring
System.......................................................................................................................... 253
R. Cicero, I. Gorrochategui and J. A. Alvarez
Thermal Fatigue Crack Initiation and Propagation Behavior of Steels for Boiler....... 255
S. Aoi, T. Marumiya, R. Ebara , T. Nishimura and Y. Tokunaga
Recent Developments in Fatigue Crack Growth ......................................................... 257
R. Jones , S. Pitt, and E. Siores
Crack Closure Effects in a Cracked Cylinder under Pressure...................................... 259
J. Zhao, R. Liu, T. Zhang and X. J. Wu
An Experimental Study of Tearing-Fatigue Interaction............................................... 261
P. Birkett, M. Lynch and P. Budden
Sif Solutions for Cracks in Railway Axles under Rotating Bending........................... 263
S. Beretta, M. Madia, M. Schode and U. Zerbst
Mechanical Characterization of Single Crystal Bars with Capacitor Discharge
Welding and Laser Cladding........................................................................................ 265
S. Chiozzi, V. Dattoma and F.o W. Panella
Fractal Dimension Analysis of Fracture Toughness Used High Strength Cast Iron.... 267
S. Doi and M. Yasuoka
Investigating Gap Effects in Fatigue Life of Spot Welded Joints ................................ 269
M. Zehsaz and S. Hasanifard
Fatigue of Pmma Bone Cement ................................................................................... 271
S. L. Evans
Influence on Thermal Barrier Coating Delamination Behaviour of Edge Geometry .. 273
H. Brodin, X. H. Li and S. Sjoestroem
Low Cycle Fatigue and Fracture of a Coated Superalloy CMSX-4 ............................ 275
S. Stekovic
Thermomechanical Fatigue of Open-Cell Aluminium Sponge ................................... 277
T. Guillen, A. Ohrndorf, U. Krupp, H. J. Christ, S. Derimay, J Hohe
and W Becker
The Influence of Alternate Block Loading on the Fatigue Lifetime............................ 279
M. Kohut and T. Lagoda
Fatigue Design and Inspection Planning of Welded Joints Based on Refined
Physical Modelling ..................................................................................................... 281
T. Lassen and N. Recho
A Mixed Mode Fatigue Crack Growth Model Including the Residual Stress Effect
Due to Weld.................................................................................................................. 283
S. Ma, X. B. Zhang, N. Recho and J. Li
Effects of Shot Peening on Fatigue Property in SICP/Al-MMC ................................. 285
Y. Ochi, K. Masaki , T. Matsumura and T. Hamaguchi
Fatigue Behaviour of Friction Stir Welded 6061-T6 Aluminium Alloy...................... 287
Y. Uematsu, K. Tokaji, Y. Tozaki and H. Shibata
Transformation of a Nonproportional Multiaxial Loading to an Equivalent
Proportional Multiaxial Loading.................................................................................. 289
A. Chamat, Z. Azari, M. Abbadi and F. Cocheteux

2T8. Polymers and Composites................................................................................. 291

 Contents xiii

Acoustic Emission Monitoring of Delamination Growth in Fiber-Reinforced

Polymer-Matrix Composites ........................................................................................ 291
A. J. Brunner and M. Barbezat
Fracture Mechanics Versus Strength Concepts for Evaluation of Adhesion Quality 293
B. Lauke
Alternative Approaches for the Evaluation of the Slow Crack Growth Resistance
of Polyethylene Resins Used in the Production of Extruded Water Pipes................... 295
F. M. Peres and C. G. Schon
A Stereoscopic Method for Fractographic Investigations of Ordinary Ceramics........ 297
C. Manhart and H. Harmuth
Modelings of Fiber Deformation During Machining Aramid-FRP ............................. 299
E. Nakanishi, M. Fukumori, Y. Sawaki and K. Isogimi
Quality Control and the Strength of Glass ................................................................... 301
F. Veer, C. Louter and T. Romein
Experimental Study of Cracked Laminate Plates by Caustics ..................................... 303
G. A. Papadopoulos and E. Sideridis
Fracture of Composites in Military Aircraft ................................................................ 305
R. Pell, N. Athiniotis and G. Clark
Analysis of 7005/AL2O3/10P MMC Sheets Joined by FSW by Thermoelasticity..... 307
P. Cavaliere, G. L. Rossi, R. di Sante and M. Moretti
Surface Modification of Lightweight Aggregate and Properties of the Lightweight
Aggregate Concrete...................................................................................................... 309
T. Y. Lo and H. Z. Cui
Finite Element Based Prediction of Failure in Laminated Composite Plates .............. 311
H. Hosseini-Toudeshky, B. Hamidi, B. Mohammadi and H. R. Ovesi
An Embedded Cylindrical PZT with Electroded Imperfect Interface ......................... 313
H. M. Shodja and S. M. Tabatabaei
Characterization of Composites for the Maeslant Storm Surge Barrier ...................... 315
J. Degrieck, W. van Paepegem, L. van Schepdael, P. Samyn, P. de Baets,
E. Suister and J. S. Leendertzc
Weight Function, J-Integral and Material Forces Approach to Ceramic Multilayers.. 317
J. Pascual, C. R. Chen, O. Kolednik, F. D. Fischer, R. Danzer and T. Lube
Assessment of Matrix Fatigue Damage in CFRP ........................................................ 319
K. J. Cain and A. Plumtree
Progressive Failure of Composite Materials under Dynamic Loading........................ 321
L. Xing, X. Huang and K. Reifsnider
Aging Aircraft Transparencies: A Case History from Italian Air Force Fleet............. 323
C. M. Bernabei, D. Caucci and C. L. Aiello
Fatigue Crack Growth in Quenched Amorphous Polymers PC and PET.................... 325
M. Kitagawa and D. Nishi
Thermo-Mechanical State of Bimaterial with an Interface Crack ............................. 327
R. Martynyak, M. Matczynski and K. Honchar
Thermo-Mechanical State of Bimaterial with an Interface Crack ............................... 329
M. Tarfaoui, S. Choukri, A. Neme and M. Mliha-Touati
Indentation Response of Fibre Reinforced Composite Laminates............................... 331
P. Bourke and I. Horsfall
Analysis of Tubular Composite Cylindrical Shells...................................................... 333
R. M. Gheshlaghi and M. H. Hojjati and H. R. M. Daniali
xiv Contents

Analysis of Composite Pressure Vessels...................................................................... 335

R. M. Gheshlaghi, M. Hassan H. and H. R. M. Daniali
Gradients Influence on Damage and Cracking in Crystalline Polymers ..................... 337
S. Castagnet and J.-C. Grandidier
An Elasto-Plastic Shear-Lag Model for Single Fiber Composite ................................ 339
S. Kimura, J. Koyanagi and H. Kawada
Progressive Failure of Composite Laminates; Analysis vs Experiments .................... 341
V. Skytta, O. Saarela and M. Wallin
A Temperature Dependent Viscoelastic-Damage Model for Ceramics Failure........... 343
V. P. Panoskaltsis, L. M. Powers and D. A. Gasparini

2T11. Fracture Mechanics Analysis ......................................................................... 345

Avalanche Mechanics: Lefm vs. Gradient Model........................................................ 345
A. Konstantinidis, N. Pugno, P. Cornetti and E. C. Aifantis
Influence of Austempering on Fracture Mechanics Parameters of 65 Si 7 Steel......... 347
D. Pustai, F. Cajner and M. Lovreni
Modelling the Evolution of Elastic Symmetries of Growing Mixed-Mode Cracks .... 349
H. Schutte and K. M. Abbasi
Effect of Aging on the Microstructure and Fracture of Aluminum-Lithium ............... 351
J. M. Fragomeni
Buckling of Multicracked Columns............................................................................. 353
C. Carloni, C. Gentilini and L. Nobile
Experimental and Numerical Analysis of Interactions Between Stress
Corrosion Cracks.......................................................................................................... 355
M. Lamazouade, M. Touzet and M. Puiggali
An Improved Upper Bound Limit Load Solution for Weld Strength anisotropic
Overmatched Cracked Plates in Pure Bending ............................................................ 357
N. Kontchakova and S. Alexandrov
Fracture Parameter Estimation of Alloy Steel Reinforced with Maraging Steel ......... 359
S. Bhat, V. G. Ukadgaonker, M Jha and S. M. Nirgude
Incorporation of Length Scales in Plane Stress Fracture Analysis .............................. 361
V. P. Naumenko
Mode III Crack in a Functionally Graded Piezoelectric/Piezomagnetic Half Plane.... 363
W.-H. Hsu and C.-H. Chue
Electro-Mechanical Field of a Piezoelectric Finite Wedge under Antiplane Loading. 365
W.-J. Liu and C.-H. Chue
Sensitivity of Crack Nucleation Parameters to the Geometric Imperfection............... 367
V. P. Naumenko and Yu. D. Skrypnyk

2T13. Probabilistic approaches to Fracture Mechanics ......................................... 369

An Experimental Evaluation of a Local Approach Model for Graded Materials........ 369
B. Bezensek, J. Flasker and J. W. Hancock
A Stochastic Model for Crack Growth......................................................................... 371
C.-R. Chiang
Stochastic Evaluation of Fatigue Crack Initiation and Propagation ............................ 373
G. S. Wang
 Contents xv

A Weibull-Based Method to Predict the Strength of Adhesively Bonded Joints of

Pultruded FRPS............................................................................................................ 375
T. Vallée, J. R. Correia and T. Keller

2T14. Computational Fracture Mechanics.............................................................. 377

The Lateral Constraint Index as a New Factor to Assess the Influence of the
Specimen Thickness..................................................................................................... 377
A. Fernandez-Canteli, D. Fernandez-Zuniga and E. Castillo
Analysis of Crack Propagation in Alumina-Glass Functionally Graded Materials ..... 379
V. Cannillo, L. Lusvarghi, T. Manfredini, M. Montorsi, C. Siligardi
and A. Sola
Numerical Solution of Integro-Differential Equations for Fracture Mechanics
Problems....................................................................................................................... 381
A. V. Andreev
Analytical Method of Generating DA/DN Curve for Aerospace Alloys..................... 383
B. Farahmand
Thermo-Elastic Fracture of Edge Cracked Plate under Surface ‘Shock’ Loading ...... 385
B. P. Fillery, X. Hu and G. Fisher
Failure Prediction of IC Interconnect Structures Using Cohesive Zone Modelling .... 387
B. A. E. van Hal, R. H. J. Peerlings, M. G. D. Geers and G. Q. Zhang
Non-Local Damage Simulation in Composites Using Crack Propagation and Mesh
Adaptivity..................................................................................................................... 389
F. Reusch, C. Hortig and B. Svendsen
Elastic Wave Motion in a Cracked, Multi-Layered Geological Region under
Transient Conditions .................................................................................................... 391
P. S. Dineva, T. V. Rangelov and G. D. Manolis
Wood Beam Strengthened with Glass/Epoxy Composite Sheets ............................... 393
G. E. Papakaliatakis, G-S. P. Diamantopoulos, P.A. Kalaitzidis
and E. M. Marinakis
Computation of Dynamic Stress Intensity Factors Using Enriched Finite Elements .. 395
M. Saribay and H. F. Nied
Partly Cracked Xfem Interface for Intersecting Cracks............................................... 397
J. L. Asferg, T. Belytschko, P. N. Poulsen and L. O. Nielsen
On the Evaluation of Elastic Compliance Tensor Due to Growing Mixed-Mode
Microcracks.................................................................................................................. 399
K. M. Abbasi and H. Schutte
On the Problem of Determination of Safety Factors for Machine-Building Parts
Using the Finite Element Computations ...................................................................... 401
L. B. Getsov, B. Z. Margolin and D. G. Fedorchenko
Dynamic Explicit Cell Model Simulations in Porous Ductile Metals ......................... 403
L. Siad and M. O. Ouali
Numerical Evaluation of Energy Release Rates for Bimaterials Interface Cracks...... 405
M. Belhouari, B. B. Bouiadjra, B. Boutabout and K. Kaddouri
Inclusion Effect on the Plastic Zone Size in Confined Plasticity................................. 407
M. Benguediab, M. Elmegueni, M. Nait-Abdelaziz and A. Imad
Modified Key-Curve-Method for Determination of Dynamic Crack Resistance
Curves .......................................................................................................................... 409
xvi Contents

U. Muhlich, A. Emrich and M. Kuna

A Coupled Computational Framework for Ductile Damage and Fracture .................. 411
R. H. J. Peerlings, J. Mediavilla and M. G. D. Geers
Marble Discs under Distributed Loading: Theoretical, Numerical and Experimental
Study ........................................................................................................................... 413
Ch. Markides, E. Sarris, D. N. Pazis, Z. Agioutantis and S. K. Kourkoulis
Simulation of the Mechanical Behaviour of the Lumbar Intervertebral Disc.............. 415
M. Satraki, E. A. Magnissalis, G. Ferentinos and S. K. Kourkoulis
The Pull-Out Strength of Transpedicular Screws in Posterior Spinal Fusion.............. 417
P. Chazistergos, G. Ferentinos, E. A. Magnissalis and S. K. Kourkoulis
Mechanical Behavior Simulation of Hip Prostheses Stress Distributions Analysis .... 419
M. Kadi, R. Boulahia, K. Azouaoui, N. Ouali, A. Ahmed-Benyahia
and T. Boukharouba
Dbem Analysis of Axisymmetric Crack Growth in a Piston Crown .......................... 421
T. Lucht
Residual Shear Stresses and KII Computation ............................................................. 423
W. Cheng and I. Finnie

2T15. Experimental Fracture Mechanics ................................................................ 425

Quantitative Interpretation of Crack Tip Strain Field Measurements.......................... 425
A. M. Korsunsky
Mixed Mode (I+II) Stress Intensity Factor Measurement Using Image Correlation... 427
A. Shterenlikht, P. López-Crespo, P. J. Withers, J. R. Yates
and E. A. Patterson
Fracture of Turbine Blades under Self-Exciting Modes .............................................. 429
C. A. Sciammarella, C. Casavola, L. Lamberti and C. Pappalettere
Predicting Crack Arrest Behaviour of Structural Steels Using New Procedures......... 431
C. Gallo, J. A. Alvarez, F. Gutierrez-Solana and J. A. Polanco
Mechanical Properties of Large Plastic-Mold Steel Blooms. ...................................... 433
M. Chiarbonello, D. Firrao, R. Gerosa, A. Ghidini, M. G. Ienco, P. Matteis,
G. Mortarino, A. Parodi, M. R. Pinasco, B. Rivolta, G. Scavino, G. Silva,
E. Stagno and G. Ubertalli
Non-Linear Photoelastic Method for Study Fracture Problems................................... 435
G. Albaut
Fatigue Crack Length Measurement Method with an Ion Sputtered Film .................. 437
G. Deng, K. Nasu, T. D. Redda and T. Nakanishi
Individual Fracture Events in Cellular Foods .............................................................. 439
H. Luyten, E. M. Castro-Prada, E. Timmerman, W. Lichtendonk
and T. van Vliet
Exfoliation Fracture Mode in Heavily Drawn Pearlitic Steels..................................... 441
J. Toribio and F. J. Ayaso
Investigation of Crack Closure by Using Thermoeastic Stress Analysis..................... 443
L. Marsavina, R. A. Tomlinson, E. A. Patterson and J. R. Yates
Fracture Toughness Investigations of Severe Plastic Deformed Tungsten Alloys ...... 445
M. Faleschini, W. Knabl and R. Pippan
Photoelastic Analysis of Mode I Stress Intensity Factor in Beams with Angular
Notches......................................................................................................................... 447
M. Tabanyukhova and V. Pangaev
 Contents xvii

Full-Scale Experimental Investigations on Pressure Tubes Rupture of RBMK .......... 449

N. Yu. Medvedeva, A. V. Andreev, S. V. Timkin, I. A. Peshkov, V. N. Zhilko,
D. Ye. Martsiniouk and O. A. Poshtovaya
Study of Fracture Mechanism of Composite Material Buildings by Photoelasticity
and Photoelasitc Coating Methods............................................................................... 451
O. Ivanova, G. Albaut, V. Mitasov, V. Nikiforovskij and M. Tabanyukhova
Fracture Energy in Mode I and Mode II of Textile Reinforced Wood......................... 453
R. Putzger and P. Haller
Measurement Based Performance Prediction of the Europabrucke Against Traffic
Loading ....................................................................................................................... 455
R. Veit and H. Wenzel
The Effect of the Laboratory Specimen on Fatigue Crack Growth Rate..................... 457
S. C. Forth, W. M. Johnston and B. R. Seshadri
Validity of the Caustics Method for Plates with Circular Hole.................................... 459
P. Tsirigas, G. Kontos, D. N. Pazis, S. K. Kourkoulis and Z. Agioutantis
An Enhanced Normalization Method for Dynamic Fracture Toughness Testing ........ 461
S. M. Graham and D. J. Stiles
The Potential Drop Technique for Measuring Crack Growth in Shear........................ 463
V. Spitas and P. Michelis
A Modified DCB Geometry for CTOA Measurement in Thin Sheet 2024-T3
Aluminium Alloy ........................................................................................................ 465
Y. H. Tai, S. H. Hashemi, R. Gay, I. C. Howard and J. R. Yates
Could Cod Serve as Fracture Criterion in Case of Marble? ....................................... 467
A. Marinelli, S. K. Kourkoulis and I. Vayas

2T16. Creep Fracture ................................................................................................ 469

Creep Rupture of a Lead-Free Sn-Ag-Cu Solder......................................................... 469
C.-K. Lin and D.-Y. Chu
Quantitative Evaluation of Acceleration Creep in Magnesium-Aluminum
Alloys at 0.65tm........................................................................................................... 471
H. Sato
Long-Term Creep Rupture Prediction in Unidirectional Composites ........................ 473
J. Koyanagi, F. Ogawa and H. Kawada
A Computational Model for Cardboard Creep Fracture ............................................. 475
J. Schonwalder, G. P. A. G. van Zijl and J. G. Rots
Creep Fracture of Binary and Ternary Commercial Aluminum Alloys....................... 477
K. Ishikawa
Analysis of Creep Crack initiation and Growth in Laboratory Specimens.................. 479
K. Wasmer
Temperature Gradient Effects on the Creep Behaviour of Structures.......................... 481
F. Vakili-Tahami and S. Hasanifard

2T17. Environment Assisted Fracture ..................................................................... 483

A Surgical Implant Crevice-Assisted Corrosion Fatigue In-Body Failure .................. 483
H. Amel-Farzad, M.-T. Peivandi and S. M.-R. Yosof-Sani
Asymptotically Stable Growth of Delaminations under Hydrogen Embrittlement
Conditions .................................................................................................................... 485
xviii Contents

A. V. Balueva
Corrosion and Mechanical Strength of Russian Light Water Reactors ....................... 487
B. T. Timofeev
Corrosion Fatigue Characteristics of CF8A Steel Degraded at High Temperature ..... 489
S.-C. Jang, D.-H. Bae, G.-Y. Lee, and S.-Y. Baek
Modeling Environment-Assisted Fatigue Crack Propagation ..................................... 491
J.-A. Ruiz-Sabariego and S. Pommier

2T18. Dynamic, High Strain Rate, or Impact Fracture ......................................... 493

Measuring the Fracture Resistance of Composites and Adhesively Bonded Joints
at High Test Rates. ....................................................................................................... 493
B. R. K. Blackman, D. D. R. Cartie, A. J. Kinloch, F. S. Rodriquez-Sanchez
and W. S. Teo
Quasistatic and Dynamic Fracture of Pearlitic Steel.................................................... 495
B. Strnadel, P. Hausild and M. Karlik
Fragmentation in the Expanding Ring Experiment...................................................... 497
H. Zhang and K. Ravi-Chandar
Influence of Friction on Results of an Instrumented Impact Test ............................... 499
I. V. Rokach
Influence of Moisture Content on the Dynamic Behaviour of Concrete ..................... 501
I. Vegt and J. Weerheijm
Strength and Toughness Properties of Steels under Dynamic Loading ....................... 503
J. Fang
Rubber Particle Size Effect on Impact Characteristics of PC/ABS (50/50) Blends .... 505
M. Nizar Machmud, Masaki Omiya, Hirotsugu Inoue and Kikuo Kishimoto
Effect of Strain Rate on Mechanical Properties of Reinforced Polyolefins................. 507
M. Schossig, C. Bieroegel, W. Grellmann, R. Bardenheier
and T. Mecklenburg
Fracture Related Mechanical Properties of Aircraft Cast Aluminum Alloy A357...... 509
N. D. Alexopoulos
Shear Failure of TI-6AL-4V by Direct Impact and analyse of the Process of
Elastic and Plastic Wave Propagation ......................................................................... 511
P. Chwalik, A. Rusinek and J. R. Klepaczko
Evaluating of Fracture Mechanics Properties at Intermediate Strain Rates,
Transferable to Components ........................................................................................ 513
P. Trubitz, A. Ludwig, G. Pusch and H.-P. Winkler
Crack Resistance Determination From the Charpy Impact Test.................................. 515
R. Chaouadi
A Stochastic Interface Model for the Fracture of Bars ................................................ 517
S. Nagy and F. Kun
The Anti-Penetration Properties of Space Armor ....................................................... 519
Tso-Liang Teng, Cho-Chung Liang and Cheng-Chung Lu
Key Curve Methods for Dynamic Fracture Mechanics of Cast Iron ........................... 521
W. Baer
Dynamic Tensile Behavior of Aramid Frp Using Split Hopkinson Bar Method......... 523
Y. Sawaki, J. Watanabe, E. Nakanishi and K. Isogimi

2T19. Damage Mechanics.......................................................................................... 525

 Contents xix

Detection of Low-Velocity Impact Damage in Carbon-Epoxy Plates using NDT ..... 525
A. M. Amaro, M. F. M. S. de Moura and P. N. B. Reis
Damage Accumulation at High Temperature Creep of a Single-Crystal Superalloy... 527
A. Staroselsky and B. Cassenti
Asymptotic Homogenisation for Heterogeneous Media with Evolving Microcracks 529
E. K. Agiasofitou, C. Dascalu and J. L. Auriault
On the Analysis of Damage Localization as Precursor of Macro-Cracks ................... 531
H. Stumpf and K. Hackl
Fatigue Assessment Based on Statistical Analysis of Theoretical Parameters ............ 533
J. Cacko
Determination of Ductile Damage Parameters by Local Deformation Fields ............. 535
M. Kuna and M. Springmann
Fracture of Concrete Due to Corrosion........................................................................ 537
N. Thanh, A. Millard, Y. Berthaud, S. Care and V. L’Hostis

2T21. Concrete and Rock .......................................................................................... 539

Experimental Study of Sprayed Concrete Strength Using Marble Aggregates ........... 539
A. Sotiropoulou and Z. G. Pandermarakis
Analysis of the Behaviour of Interface Cracks in Gravity Dam ................................. 541
B. B. Bouiadjra, A. B. Bouiadjra, M. Belhouari and B. Serier
Application of Composite Mechanics to Composites Enhanced Concrete Structures 543
C. C. Chamis and P. K. Gotsis
Initiation and Coalescence of Locals Damages on Blanco de Macael Marble............ 545
K. Mehiri, P. Vieville, P. Lipinski, A. Tidu and V. Tijeras
Influence of Concrete´s Mineralogical Components on Fracture Compressive and
Tractive ........................................................................................................................ 547
M. P. Morales Alfaro and F. A. I. Darwish
Constitutive Model for Description of High-strain Rate Behavior of Concrete ......... 549
I. R. Ionescu and O. Cazacu
Hydraylic Fracturing in Weak Rocks .......................................................................... 551
P. Papanastasiou
Application of Fracture Mechanics on Unreinforced Concrete Walls ......................... 553
T. Eck, B.-Gu Kang and W. Brameshuber
Subcritical Crack Growth in Rocks under Water Environment ................................... 555
Y. Nara, H. Kurata and K. Kaneko

2T22. Sandwich Structures ....................................................................................... 557

Stress Analysis and Prediction of Failure in Structurally Graded Sandwich Panels ... 557
A. Lyckegaard, E. Bozhevolnaya and O. T. Thomsen
Debonding and Kinking in Foam-Core Sandwich Beams ........................................... 559
D. A. Zacharopoulos, V. D. Balopoulos, Z. S. Metaxa, P. A. Kalaitzidis
and E. E. Gdoutos
Modeling Core Failure by the Tsai–Wu Criterion in the Design of Foam-Core
Sandwich Beams .......................................................................................................... 561
E. E. Gdoutos, V. D. Balopoulos, P. A. Kalaitzidis and M. Konsta
Numerical Investigation of Crack Propagation in Sandwich Structures...................... 563
E. E. Theotokoglou
xx Contents

Local Effects in Sandwich Beams: Modelling and Experimental Investigation.......... 565

M. Johannes, J. Jakobsen, V. Skvortsov, E. Bozhevolnaya
and O. T. Thomsen
Typical In-Plane Response Surfaces for Prismatic Foam-Core Sandwich Beams ...... 567
V. D. Balopoulos, P. A. Kalaitzidis, D. A. Zacharopoulos
and E. E. Gdoutos

2T23. Novel Testing and Evaluation Techniques.................................................... 569

Non-Destructive Evaluation of Yield Strength Using a Novel Miniature Dumb-Bell
Specimen-An Empirical Approach ............................................................................. 569
G. Partheepan, D. K. Sehgal and R. K. Pandey
3D Measurement of the Strain Field Surrounding Crack Tip ...................................... 571
D. Vavrik, J. Bryscejn, J. Jakubek and J. Valach
Radiographic Observation of Damage Zone Evolution in High Ductile Specimen .... 573
D. Vavrik, T. Holy, J. Jakubek, M. Jakubek and Z. Vykydal
Calibration of Fracture Parameters by Instrumented Indentation and Test
Simulation .................................................................................................................... 575
M. Bocciarelli, G. Bolzon and G. Maier
Internal Crack Detection and Analysis Using Thermoelastic Stress Analysis ............ 577
N. Sathon and J. M. Dulieu-Barton
Ultrahigh-Resolution Transversal Polarization-Sensitive Optical Coherence
Tomography: Structural Analysis and Strain-Mapping ............................................... 579
K. Wiesauer, M. Pircher, R. Engelke, G. Ahrens, G. Grutzner, R. Oster,
C. K. Hitzenberger and D. Stifter
Application of Digital Shearography in Determining Opening Mode SIF in
Edge Cracks ................................................................................................................. 581
M. Ghassemieh, A. Ghazavizadeh and N. Soltani
Finite Element Modeling of Pulse Transient IR Thermography.................................. 583
M. Krishnapillai, R. Jones, I. H. Marshall, M. Bannister and N. Rajic
A New Technique for the Machining of Natural Cracks ............................................. 585
N. P. Andrianopoulos and A. Pikrakis
Displacements Measurement in Irregularly Bounded Plates Using Mesh Free
Methods........................................................................................................................ 587
N. P. Andrianopoulos and A. P. Iliopoulos
Biaxial Strength Testing on Mini Specimens............................................................... 589
R. Danzer, P. Supancic, W. Harrer, T. Lube and A. Borger
Numerical Simulation of a Fracture Test for Brittle Disordered Materials ................. 591
T. Auer and H. Harmuth

2T26. Structural Integrity ......................................................................................... 593

Unification of the Out-of-Plane Constraint Loss in Centre-Cracked Panels ............... 593
B. Bezensek, A. Baron and J. W. Hancock
High Temperature Failure Assessment of Weldments ................................................ 595
B. Dogan, B. Petrovski and U. Ceyhan
Post-Tensioned Glass Beams ....................................................................................... 597
C. Louter, J. van Heusden, F. Veer, J. Vambersky, H. de Boer
and J. Versteegen
 Contents xxi

Structural Integrity of a NPP Using the Master Curve Approach................................ 599

D. Ferreno, I. Gorrochategui, M. Scibetta, R. Lacalle, E. van Walle
and F. Gutierrez-Solana
FRP Consolidation for Masonry Arches by Using Bridged Crack Model .................. 601
G. Ferro, M. Ipperico, V. Pignata and A. Carpinteri
Structural Reliability Analysis of Pipe Subjected to Reeling ...................................... 603
H. A. Ernst, R. E. Bravo and F. Daguerre
Network Seismic Capability Assessment of Power High Voltage
Electric Equipment....................................................................................................... 605
I. Manea, C. Diaconu, C. Radu and M. Negru
FKM Guideline “Fracture Mechanics Proof of Strength for Engineering
Components” – Overview and Extension topics.......................................................... 607
B. Pyttel, I. Varfolomeyev and M. Luke
Static and Dynamic Behavior of a 3D-Periodic Structure ........................................... 609
J. Rishmany, L. Renault, C. Mabru, R. Chieragatti and F. Rezaï Aria
Environmental Effect on Pipeline Steels: A Fitness for Service Perspective .............. 611
J. A. Alvarez, F. Gutierrez-Solana and S. Cicero
Finding the Australian Railway Load Spectrum Design and Assessment of Light
Weight & Durable Railway Structural Components .................................................... 613
R. Jones and J. Baker
Structural Integrity Assessment of Componets with Low Constraint.......................... 615
S. Cicero, F. Gutierrez-Solana and J. A. Alvarez
Life Assessment of Superheater Tubes Fabricated From 2.25CR-1MO Steel ............ 617
S. Fujibayashi
Predicting Cleavage Fracture in Presence of Residual Stresses; A Numerical
Case Study.................................................................................................................... 619
S. Hadidi-Moud, C. E. Truman and D. J. Smith
A Necessary Condition for Cleavage on Laboratory Specimens and Structures......... 621
V. le Corre, S. Chapuliot, S. Degallaix and A. Fissolo
Safety Assessment of Components with Crack-Like Defects .................................... 623
Yu. G. Matvienko and O. A. Priymak
Numerical Analysis of Surface Cracks in Steam Generator Tubes ............................. 625
Z. Tonkovi, I. Skozrit and J. Sori

2T28. Mesofracture Mechanics................................................................................. 627

Tensile Simulation of Polymeric Material Considering the Meso-Scale Structure .... 627
A. Shinozaki, K. Kishimoto and I. Hirotugu
Microfracture and Strain Localization: A Computational Homogenization Approach 629
C. Dascalu, G. Bilbie and R. Chambon
Strain and Fracture at Mesoscale of Coated Materials ................................................ 631
S. Panin

2T32. Micromechanisms in Fracture and Fatigue.................................................. 633

Relating Cleavage Crack Nucleation to Cracked Carbides in A533B Steel................ 633
A. Kumar and S. G. Roberts
xxii Contents

Micro-Energy Rates for Damage Tolerance and Durability of Composite Structures 635
C. C. Chamis and L. Minnetyan
Micromechanical Observation of Fracture Process in Mortars ................................... 637
E. Schlangen and O. Copuroglu
Micro-fracture Maps in Progressively Drawn Pearlitic Steels .................................... 639
J. Toribio and F. J. Ayaso
A Brief History of Fractography.................................................................................. 641
S. P. Lynch and S. Moutsos

C. SPECIAL SYMPOSIA/SESSIONS
C1. Nanomaterials and Nanostructures ..................................................... 643

1. Fracture and Fatigue at the Micro and Nano Scales .......................................... 645
Size Effects in Lead Free Solder-joints........................................................................ 645
A. Betzwar-Kotas, G. Khatibi, A. Ziering, P. Zimprich, V. Groeger,
B. Weiss and H. Ipser
Micro-Scale Simulation of Impact Rupture in Polysilicon MEMS ............................. 647
A. Corigliano, F. Cacchione, A. Frangi and B. de Masi
Nanoindentation of CNT Reinforced Epoxy Nanocomposites.................................... 649
D. C. Lagoudas, P. R. Thakre and A. A. Benzerga
Diffusion Kinetics and Multivariant Phase Transformation in Shape Memory Alloys 651
D. R. Mahapatra and R. V. N. Melnik

3. Nanoscale Deformation and Failure..................................................................... 653

EBSD Analysis on Deformation of Nanocrystals in ECAP-Processed Copper .......... 653
H. Kimura, Y. Akiniwa, K. Tanaka and T. Ishida
The Effect of Extensional Strains on Molecular Orientation, Polymer Free Volume
Distribution and Crystallization ................................................................................... 655
H. Dong, R. Guo and K. I. Jacob
Microrotation-augmented Energy-Minimization for 3D Nanocrystalline Cu
Structures ..................................................................................................................... 657
M. A. Tschopp and D. L. McDowell
Mechanics and Electromechanics of Single Crystalline Piezoelectric Nanowires ...... 659
M.-F. Yu, Z. Wang, J. Hu and A. Suryavanshi
Multiscale Simulation for High Spped Propagation of Disordered Regions .............. 661
W. Yang, X. Li and Z. Guo
Surface-Stress-Driven Pseudoelasticity and Shape Memory Effect at the Nanoscale. 663
W. Liang and M. Zhou
Thermomechanical Behavior of Zinc Oxide Nanobelts .............................................. 665
A. Kulkarni and M. Zhou
Natural Modes of C60 Cage via Carbon-Carbon Bonding Element............................ 667
P. Zeng, X.-G. Y. and J. Du

11. Deformation and Fracture at the Nano Scale.................................................... 669

Fracture of Nanocrystalline Aluminum ....................................................................... 669
 Contents xxiii

C. San Marchi, S. L. Robinson, N. Y. C. Yang and E. J. Lavernia

Wear and Fatigue in Silicon Structural Films for MEMS Applications ...................... 671
D. H. Alsem, R. Timmerman, E. A. Stach, C. L. Muhlstein, M. T. Dugger
and R. O. Ritchie
Indentation Induced Through Thickness Film Fracture on Engineering Alloys ......... 673
D. F. Bahr, K. R. Morasch and A. Alamr
Surface Nanostructured Aluminum by Severe Plastics Deformation.......................... 675
E. I. Meletis, K. Y. Wang and J. C. Jiang
Contribution of Localized Deformation to IGSCC and IASCC in Austenitic
Stainless Steels ............................................................................................................. 677
G. S. Was, Z. Jiao and J. T. Busby
A Study of Crack-Dislocations Interaction with 3D Discrete Dislocation Dynamics . 679
I. N. Mastorakos and H. M. Zbib
Numerical Simulations and Measurements of Cracks Parallel and Near Interfaces
in Graded Structures..................................................................................................... 681
I. Reimanis, K. Rozenburg, J. Berger, M. Tilbrook and M. Hoffmann
Deformation and Failure Processes Operating in Ultra-Fine Grain Metals................. 683
K. Hattar, I. M. Robertson, J. Han, T. Saif, S. J. Hearne and D. Follstaedt
Simulation of Cross-Sectional Nanoindentation in Interconnect Structures with
Cohesive Elements....................................................................................................... 685
D. Gonzalez, J. Molina, I. Ocana, M. R. Elizalde, J. M. Sanchez.
J. M. Martinez-Esnaol, J. Gil-Sevillano, G. Xu, D. Pantuso, T. Scherban,
B. Sun, B. Miner, J. He and J. Maiz
Fracture Between Two Self-Assembled Monolayers .................................................. 687
K. M. Liechti and D. Xu
Nanotube Nanoactuator................................................................................................ 689
M.-F. Yu, J. Hu, Z. Wang and A. Suryavanshi
Nanocrack Detection in Vibrating Nanowires ............................................................. 691
R. Ruoff, L. Calabri, N. Pugno, X. Chen, W. Ding and K. Kohlhaas
Fracture of atomic Layer Deposited Nanolaminate Films ........................................... 693
N. R. Moody, J. M. Jungk, T. M. Mayer, R. A. Wind, S. M. George
and W. W. Gerberich
Influence of Microstructure, Strength and Adhesion on Au Electrodeposits ............. 695
N. Yang, J. Kelly, T. Headley and C. S. Marchi
Fracture of Submicron Thin Metal Films During Cyclic Loading .............................. 697
S. Eve, D. Wang, C. Volkert, N. Huber and O. Kraft
Micromechanics of Damage Evolution in Solid Propellants ....................................... 699
N. Aravas, F. Xu and P. Sofronis
Deformation and Failure Mechanisms in Metallic Nanolayered Composites ............. 701
R. G. Hoagland, J. P. Hirth, and A. Misra
Dislocation Source Sensitivity of Plasticity and Fracture in Tungsten........................ 703
J. E. Talia and R. Gibala
Delamination of Thin Metal Films on Polymers ......................................................... 705
A. Pundt, E. Nikitin, and R. Kirchheim
Fracture Mechanics of One-Dimensional Nanostructures ........................................... 707
W. Ding, L. Calabria, K. M. Kohlhaas, X. Chen and R. S. Ruoff
Effects of Structure and Bonding at Surfaces and Interfaces on Fracture .................. 709
S. P. Lynch, S. Moutsos, B. Gable, S. Knight, D. P. Edwards
xxiv Contents

and B. C. Muddle

29. Reliability and Failure Analysis of Electronics and Mechanical Systems ...... 711
Application of the New Static Photoelastic Experimental Hybrid Method with New
Numerical Method to the Plane Fracture Mechanics................................................... 711
J.-S. Hawong , J.-H. Nam, O.-S. Kwon and K. Tche
Risk Analysis of Buried Pipeline using Probabilistic Method..................................... 713
O. S. Lee, D. H. Kim and N. H. Myoung
Reliability Estimation of Solder Joint by Accelerated Life Tests ................................ 715
O. S. Lee, N. H. Myoung and D. H. Kim
Analysis of Engineering Plastic Behaviors in Thermal Stress Condition.................... 717
S. I. Ham, D. J. Choi and S. D. Park
A Mechanistic Model for the Thermal Fatigue Behavior of the Lead-Free
Solder Joints................................................................................................................. 719
I. Kim, T.-S. Park and S.-B. Lee
Mechanical Behavior of Metallic Thin Film on Polyimide Substrate ......................... 721
D.-C. Baek, S.-Y. Kim and S.-B. Lee

31. Multiscaling in Molecular and Continuum Mechanics - Scaling in Time

and Size From Macro to Nano .................................................................................. 723
Macro-, Meso- and Micro-damage Model Based on Singularity Representation
for Anti-plane Deformation ......................................................................................... 723
G. C. Sih and X. S. Tang
Multiscaling Effects in Trip Steels............................................................................... 725
G. N. Haidemenopoulos and N. Aravas
A Hyper-Surface for the Combined Rate and Size Effects on the Material Properties 727
Z. Chen, L. Shen, Y. Gan and H. E. Fang

34. Cracks in Micro- and Nanoelectronics............................................................... 729

A New Method for Local Strain Field Analysis Near Cracks in Micro- and
Nanotechnology Applications...................................................................................... 729
B. Michel, D. Vogel, N. Sabate and D. Lieske
Experimental Investigations for Fracture Analysis of Solder Joints in
Microelectronic and MEMS Applications ................................................................... 731
H. Walter, C. Bombach, R. Dudek, W. Faust and B. Michel
Simulation of Interface Cracks in Microelectronic Packaging .................................... 733
J. Auersperg, B. Seiler, E. Cadalen, R. Dudek and B. Michel
AFM Based Fracture Analysis in Micro- and Nanomaterials...................................... 735
J. Keller, A. Gollhardt, D. Vogel and B. Michel
Simulation of Deformation and Fracture Behaviour in Microelectronic Packaging ... 737
O. Wittler, H. Walter, J. Keller, R. Dudek, D. Vogel and B. Michel

43. Interfacial Fracture in Composites and Electronic Packaging Materials ...... 739
Mixed-Mode Fracture Modeled Through a Discrete Cohesive Zone Model-DCZM.. 739
D. Xie and A. M. Waas
Signifince of K-Dominance in Delamination Cracking in Composite Laminates....... 741
C. T. Sun and Z. Yang
 Contents xxv

Evaluation of Interface Toughness Between Submicron Island and Substrate............ 743

H. Hirakata, T. Kitamura, S. Matsumoto and Y. Takahashi
Three-Dimensional Thermal Stress Analysis Considering the Stress Singularity for
Bonded Structures ........................................................................................................ 745
H. Koguchi
Center of Dilatation and Penny-Shaped Crack in Viscoelastic Bimaterial .................. 747
K. T. Chau, R. C. K. Wong and Y. Z. Sun
Fracture Analysis on Popcorning of Plastic Packages During Solder Reflow ............ 749
S. W. R. Lee and D. C. Y. Lau
Delamination of PB-Free Flip Chip Underfill During 2nd Level Interconnect
Reflow.......................................................................................................................... 751
S. Chung, Z. Tang and S. Park
Reliability of Interfaces Between Components in Advanced Electronic Packages
under Solder Reflow Process ....................................................................................... 753
T. Ikeda and N. Miyazaki
Three-Dimensional Stress Intensity Factors Analyses of Interface Cracks Between
Dissimilar Anisotropic Materials ................................................................................. 755
M. Nagai, T. Ikeda, N. Miyazaki
Molecular Dynamics of Interfacial Fracture................................................................ 757
T. E. Tay, V. B. C. Tan and M. Deng

C. SPECIAL SYMPOSIA/SESSIONS
C2. Engineering Materials and Structures................................................ 759

4. Fracture and Fatigue of Elastomers..................................................................... 761

Nucleation, Growth and Instability of the Cavitation in Rubber ................................. 761
E. Bayraktar, K. Bessri and C. Bathias
Engineering Fracture Mechanics for Crack Toughness Characterisation of
Elastomers.................................................................................................................... 763
K. Reincke, W. Grellmann and G. Heinrich
Multiaxial Fatigue Crack Initiation on Filled Rubbers : Statistical Aspects................ 765
L. Laiarinandrasana, A. Bennani and R. Piques
Fracture Criteria of Rubber-like Materials under Plane Stress Loadings .................... 767
A. Hamdi, M. Nait-Abdelaziz, N. Ait Hocine and P. Heuillet
Prediction of Rubber Fatigue Life under Multiaxial Loading ..................................... 769
A. Zine, N. Benseddiq, M. Nait-Abdelaziz and N. Ait Hocine
Modeling of Biaxial Fatigue of Natural Rubber ........................................................ 771
S. Dong, C. Bathias, K. le Gorjo, F. Hourlier and J. F. Vitorri
Modeling of Crack Propagation in Elastomeric Materials Using Configurational
Forces ........................................................................................................................... 773
T. Horst and G. Heinrich
Determination of Inter-Fibre-Failure in Complex, Reinforced Composites................ 775
V. Trappe and H. Ivers
The Test Frequency Dependence of the Fatigue Behavior of Elastomers ................... 777
Z. Major, Ch. Feichter, R. Steinberger and R. W. Lang
xxvi Contents

5. Integrity of Dynamical Systems ............................................................................ 779

Nonlinear Model for Reinforced Concrete Frames Loaded by Seismic forces ........... 779
D. Kovacevic
Monitoring the Durability Performances of Concrete and Masonry Structures by
Acoustic Emission Technique ..................................................................................... 781
A. Carpinteri and G. Lacidogna
Bifurcation Control of Parametric Resonance in Axially Excited Cantilever Beam .. 783
H. Yabuno and M. Hasegawa
Adaptive Properties of Dynamic Objects..................................................................... 785
I. I. Blekhman and L. A. Vaisberg
Influence of Addendum Modification Coefficient on the Gear's Load Capacity ........ 787
I. Atanasovska and V. Nikoli-Stanojevi
Micromechanical Modelling of Fracture-induced Anisotropy and Damage in
Orthotropic Materials .................................................................................................. 789
V. Monchiet, I.-C. Gruescu, D. Kondo and O. Cazacu
Vibration Control Devices and their Application ....................................................... 791
K. Nagaya
Measurements of Dynamical System Integrity and Fracture Mechanics .................... 793
K. S. Hedrih
Modeling of the Surface Cracks and Fatigue Life Estimation..................................... 795
K. Maksimovic, S. Maksimovic and V. Nikolic-Stanojevic
Structural Damage Detection via the Subspace Identification Method ....................... 797
M. Trajkovic, D. Sumarac and M. Mijalkovic
Clock Mechanism as Base of Artillery Safety and Arming Devices........................... 799
M. Ugrci c
Twisting Deformation Evolution of Drilling Ropes .................................................... 801
N. P. Puchko
Hereditary Strain Theory of Syntetic and Steel Ropes ................................................ 803
O. O. Goroshko
Brittle and Ductile Failure in Thermoviscoplastic Solids under Dynamic Loading.... 805
R. C. Batra and B. M. Love
Some Aspects of Dynamic interfacial Crack Growth ................................................. 807
R. R. Nikolic and J. M. Veljkovic
On Stability Problems of Periodic Impact Motions ..................................................... 809
S. Mitic
Dynamical Integrity of Nonlinear Mechanical Oscillators .......................................... 811
S. Lenci and G. Rega

8. Modelling of Material Property Data and Fracture Mechanisms..................... 813

Fatigue Crack initiation and Propagation at High Temperature in a Softening
Martensitic Steel ......................................................................................................... 813
B. Fournier, M. Sauzay, M. Mottot, V. Rabeau, A. Bougault and A. Pineau
Transferability of Cleavage Fracture Parameters Between Notched and Cracked
Geometries ................................................................................................................... 815
C. Bouchet, B. Tanguy, J. Besson, A. Pineau and S. Bugat
Relation Between Crack Velocity and Crack Arrest .................................................... 817
M. Hajjaj, C. Berdin, P. Bompard and S. Bugat
 Contents xxvii

Mechanisms of Damage and Fracture in Trip Assisted Multiphase Steels.................. 819

G. Lacroix, Q. Furnemont, P. J. Jacques and T. Pardoen
The Role of Sub-Boundaries in the Brittle Fracture of Polycrystalline Materials....... 821
G. Hughes, P. Flewitt, F. Sorbello, G. Smith and A. Crocker
Three-Dimensional Modelling of Fracture in Polycrystals.......................................... 823
G. Smith, A. Crocker, G. Hughes and P. Flewitt
Anti-Wing Crack Growth from Surface Flaw in Real Rock under Uniaxial
Compression................................................................................................................. 825
R. H. C. Wong, Y. S. H. Guo , L. Y. Li, K. T. Chau , W. S. Zhu and S. C. Li
Mechanical Behavior Modeling in the Presence of Strain Aging................................ 827
J. Belotteau, C. Berdin, S. Forest, A. Parrot and C. Prioul
On the Local Conditions for Cleavage Initiation in Ferritic Steels.............................. 829
J. Hohe, V. Friedmann and D. Siegele
Unified Constitutive Equations to Describe Elastoplastic and Damage Behavior of
X100 Pipeline Steel...................................................................................................... 831
T. T. Luu, B. Tanguy, J. Besson, A. Pineau and G. Perrin
Estimation of Lower Bound Engineering Fracture Toughness in the Ductile
to Brittle Transition Regime......................................................................................... 833
R. Moskovic and R. A. Ainsworth
Cleavage Fracture Micromechanisms Related to WPS Effect in RPV Steel............... 835
S. R. Bordet, B. Tanguy, S. Bugat, D. Moinereau and A. Pineau
Modelling of Fatigue Damage in Aluminum Cylinder Heads..................................... 837
R. Salapete, B. Barlas, E. Nicouleau, D. Massinon, G. Cailletaud
and A. Pineau
Local Approach to High Temperature Ductility Modeling in 6XXX Aluminium
Alloys ........................................................................................................................... 839
D. Lassance, D. Fabregue, F. Delannay and T. Pardoen

9. Micromechanisms in Fracture and Fatigue......................................................... 841

Small Fatigue Crack Growth in Steel-Compressor Disks of Aircraft Engines............ 841
A. A. Shanyavskiy and A. Yu. Potapenko
Micromechanisms of Damage in Multiaxial Fatigue of an Austenitic-Ferritic
Stainless Steel............................................................................................................... 843
A. el Bartali, V. Aubin, S. Degallaix and L. Sabatier
Multiscale Modeling of Fracture and Plasticity in Layered Structures........................ 845
A. Hartmaier, N. Brodling and H. Gao
Critical and Fracture Planes of 18G2A Steel under Non-Proportional Combined
Bending and Torsion .................................................................................................... 847
A. Karolczuk and E. Macha
Slip Processes and Fracture in Iron Crystals................................................................ 849
V. Pelikan, P. Hora, A. Machova1 and M. Landa
A Discussion of the Applicability of DK-Values to Explain the Fatigue Crack
Growth Behaviour of Short Cracks.............................................................................. 851
A. Tesch, H. Doker, K. H. Trautmann, R. Pippan and C. Escobedo
Simulation of Crack Growth under Low Cycle Fatigue at High Temperature in a
Single Crystal Superalloy............................................................................................. 853
B. Fedelich, Y. Kiyak, T. May and A. Pfennig
xxviii Contents

Fatigue Crack Growth for Different Ratios of Bending to Torsion in ALCU4MG1 .. 855
D. Rozumek and E. Macha
Ductile Damage Models Applied to Anisotropic Fracture of Al2024 T351 ............... 857
D. Steglich, W. Brocks and T. Pardoen
Fatigue and Fracture Processes in Severe Plastic Deformed Rail Steels .................... 859
F. Wetscher, R. Pippan and R. Stock
Damage Evolution in Torsion Specimens Deformed at Forging Temperatures .......... 861
G. Trattnig, R. Pippan and S. Kleber
Microstructural Effects on Short Fatigue Crack Propagation and their Modelling ..... 863
H. J. Christ, O. Duber, W. Floer, U. Krupp, C. P. Fritzen, B. Kunkler
and A. Schick
Micromechanical Aspects of Transgranular and Intergranular Failure Competition .. 865
I. Dlouhy and M. Holzmann
Defect in Ultra-fine Grained Mg-based Alloys Deformed by High-Pressure Torsion 867
J. Cizek, I. Prochazka, B. Smola, I. Stulikova, R. Kuzel, Z. Matej
and V. Cherkaska
Modelling Crack-Tip Shielding Effects in Particle Reinforced Composites .............. 869
J. Hornikova, P. Sandera and J. Pokluda
Early Stages of Fatigue Damage in 316l Steel............................................................. 871
J. Man, K. Obrtlik, J. Polak and P. Klapetek
AB Initio Study of Elasticity and Strength of Nano-Fibre Reinforced Composites .... 873
M. Cerny and J. Pokluda
Strength and Fracture of Ultra-Fine Grained Aluminum 2024 ECAP Metal .............. 875
K. B. Yoon, Y. W. Ma, J. W. Choi and S. H. Kim
Fatigue Lifetime of Bearing Steel in Ultra-High-Cycle Region .................................. 877
L. Kunz, P. Lukas, M. Cincala and G. Nicoletto
Calculation of K-Factor and T-Stress for Crack at Anisotropic Bimaterials .............. 879
M. Kotoul, T. Profant and O. Sevecek
Interaction of Microcracks with Grain Boundaries: Systematical Investigation of
the Mechanisms............................................................................................................ 881
M. Marx, W. Schaf and H. Vehoff
Dislocation Arrangements in Cyclically Strained Inconel 713LC............................... 883
M. Petrenec, K. Obrtlik and J. Polak
Crack Initiation and Fracture of Metal Matrix Composites......................................... 885
K. Unterweger and O. Kolednik
Mechanical Behaviour of Ultra-Fine Grained Austenitic Stainless Steel .................... 887
S. Brochet, A. Poulon-Quintin, J.-B. Vogt , J.-C. Glez and J.-D. Mithieux
Tribological Properties and Wear Mechanisms of Wear Resistant
Thermally Sprayed Coatings........................................................................................ 889
Sa. Houdkova, F. Zahalka and R. Enzl
Crack Propagation Resistance and Damage Mechanisms in Nuclear Graphite........... 891
A. Hodgkins, J. Marrow, P. Mummery, A. Fok and B. J. Marsden
Environment-assisted Cracking of High-Strength Magnesium Alloys WE43-T6....... 893
A. Ahmad and T. J. Marrow
Effects of Surface Finish on the Fatigue Limit in Austenitic Stainless Steels
(Modelling and Experimental Observations) ............................................................... 895
M. Kuroda, T. J. Marrow and A. Sherry
 Contents xxix

Intergranular Stress Corrosion Crack Propagation in Sensitised Austenitic

Stainless Steel (Microstructure Modelling and Experimental Observation)................ 897
T. J. Marrow, L. Babout, A. P. Jivkov, P. Wood, D. Engelberg, N. Stevens, P.
J. Wither and R. C. Newman
Ideal Strength of Nanoscale Thin Films ...................................................................... 899
T. Kitamura, Y. Umeno and A. Kushima
Toughness Variability................................................................................................... 901
R. Bouchard, G. Shen and W. R. Tyson
Thermo-Mechanical Behaviour of Nanostructured Copper......................................... 903
C. Duhamel, S. Guerin, M. J. Hytch and Y. Champion
Some Insights into Fatigue Crack Initiation Stage....................................................... 905
H. Alush and Y. Katz
Fatigue Behaviour of Metallic Materials Exposed to High Pressure Hydrogen
Environments ............................................................................................................... 907
Y. Mine, S. Matsuoka, Y. Murakami C. Narazaki and T. Kanezaki
In-Situ Investigations of the Fracture Mechanisms at Various Length Scales............. 909
Z. Pakiela, W. Zielinski and K. J. Kurzydlowski

12. Interface Fracture and Behavior of Joints ........................................................ 911

Environmental Attack at Polymer/Metal Interfaces..................................................... 911
A. J. Kinloch, D. Bland, K. T. Tan and J. F. Watts
Modelling of Elastic-Plastic Peel Tests for Structural Adhesives................................ 913
A. J. Kinloch, H. Hadavinia, L. Kawashita, D. R. Moore and J. G. Williams
An Alternating Crack Growth in Adhesively Bonded Joints ...................................... 915
A. R. Akisanya
Measurements of Interface Fracture and Mechanical Properties of Low-K Dielectric
Thin Films ................................................................................................................... 917
F. Atrash and D. Sherman
Initiation of Fracture Mechanisms at the Fibre/Matrix interface................................. 919
E. Martin, B. Poitou and D. Leguillon
Effects of Plasticity and Residual Stress for Cracks Near Interfaces........................... 921
I. Reimanis, K. Rozenburg, M. Tilbrook and M. Hoffmann
Toughness of a ±45o Interface ..................................................................................... 923
L. Banks-Sills, Y. Freed, R. Eliasi and V. Fourman
Residual Stress Influence on Dissimilar Material Weld Junction Fracture.................. 925
P. Gilles and M.-F. Cipiere
Fracture Mechanisms of a Thin Elastic Plastic Laminate............................................ 927
C. Bjerken, S. Kao-Walter and P. Stahle
Crack-Tip Parameters in Polycrystalline Plates with Compliant Grain Boundaries.... 929
R. Ballarini and Y. Wang
Extended Fe Simulations of Crack Growth in Layered and Functionally Graded
Materials....................................................................................................................... 931
C. Comi and S. Mariani

13. Computational Fracture Mechanics................................................................... 933

Simulation of Plastic Fatigue Crack Growth by a Two Scale Extended Finite
Element Method........................................................................................................... 933
xxx Contents

A. Gravouil, T. Elguedj and A. Combescure

Accurate Determination of Cohesive Crack Tip Fields using Xfem and Admissible
Stress Recovery............................................................................................................ 935
B. L. Karihaloo, Q. Z. Xiao and X. Y. Liu
A New Generation of Boundary Element Method for Damage Tolerance
Assessment of Aerostructures...................................................................................... 937
M. H. Aliabadi
Robust Stress Intensity Factors Evaluation for 3D Fracture Mechanics with
X-FEM ....................................................................................................................... 939
H. Minnebo, E. Bechet and N. Moes
A Micro-Macro Partition of Unity Method for Crack Propagation ............................. 941
P. A. Guidault, O. Allix, L. Champaney and C. Cornuault
A Dynamic Crack Propagation Criteria for XFEM, Based on Path-Independent
Integral Evaluation....................................................................................................... 943
I. Nistor, S. Caperaa and O. Pantale
Truss Model as Simple Computational Tool in Fracture Mechanics ........................... 945
P. G. Papadopoulos, D. Plasatis and P. Lambrou
Finite Element Modeling of Cohesive Cracks by Nitsche’s Method........................... 947
P. Hansbo and P. Heintz
Computing Crack Growth in Quasiperiodic Alloys..................................................... 949
P. M. Mariano and F. L. Stazi
X-FEM for 3D Cracks in Shaft with Contact .............................................................. 951
S. Geniaut, P. Massin and N. Moes
Some Improvements for Extended Finite Element Methods in Fracture Mechanics .. 953
P. Laborde, J. Pommier, Y. Renard and M. Salaun

14. Cohesive Models of Fracture .............................................................................. 955

Failure Prediction of Adhesively Bonded T-Peel Joints .............................................. 955
A. Pirondi
An Approach for the Determination of Mixed Mode Cohesive Laws......................... 957
B. F. Sorensen and T. K. Jacobsen
The Use of CZM for Coupled Fatigue/Plasticity Crack Propagation Simulation........ 959
Jl. Bouvard, F. Feyel and Jl. Chaboche
Dynamic Crack Growth : Analytical and Numerical CZM Approaches ..................... 961
G. Debruyne, J. Laverne and P. E. Dumouchel
Simulation of Pre-Critical Cracking in Concrete Using 3D Lattice Model................. 963
H.-K. Man and J. G. M. Van Mier
Effect of Cohesive Law and Triaxiality Dependence of Cohesive Parameters in
Ductile Tearing............................................................................................................. 965
I. Scheider, F. Hachez and W. Brocks
Modeling Quasibrittle Material Cracking with Cohesive Cracks: Experimental
and Computational Advances ...................................................................................... 967
J. Planas, J. M. Sancho, A. M. Fathy, D. A. Cendon and J. C. Galvez
Pinwheel Meshes and Branching of Cohesive Cracks................................................. 969
P. Ganguly and K. D. Papoulia
A Dynamic Crack Growth Simulation Using Cohesive Elements .............................. 971
M. Anvari and C. Thaulow
 Contents xxxi

A New Cohesive Zone Model for Mixed-Mode Decohesion ...................................... 973

M. J. van Den Bosch , P. J. G. Schreurs and M. G. D. Geers
Cohesive-Zone Modeling of Crack Growth in Specimens with Different Constraint
Conditions .................................................................................................................... 975
C. R. Chen, O. Kolednik and F. D. Fischer
Effect of Anisotropic Plasticity on Mixed Mode Interface Crack Growth .................. 977
V. Tvergaard and B. N. Legarth

16. Environment Assisted Fracture.......................................................................... 979

Characterisation of TG-SCC in Pure Magnesium and AZ91 Alloy ............................ 979
N. Winzer, G. Song, A. Atrens, W. Dietzel and C. Blawert
Hydrogen Embrittlement and Cracking of 18MN-4CR Steels .................................... 981
A. Balitskii
Transient Stress and EAC of Steam Turbine Disc Steel .............................................. 983
A. Turnbull and S. Zhou
Irreversible Hydrogen Trapping in Welded Beta-21S Titanium Alloy ........................ 985
D. Eliezer, E. Tal-Gutelmacher, C. E. Cross and Th. Boellinghaus
EAC in High Strength Steels for Gas Transportation .................................................. 987
G. Gabetta and R. Bruschi
High Temperature Fatigue Crack Growth in Titanium Microstructures...................... 989
H. Ghonem
Corrosion Damaging and Corrosion Fatigue Assessment in Three-Layered
Metallic Material.......................................................................................................... 991
I. M. Dmytrakh and V. V. Panasyuk
Simulation of Hydrogen Assisted Stress Corrosion Cracking Using a Time
Dependent Cohesive Model ......................................................................................... 993
I. Scheider, M. Pfuff and W. Dietzel
Environmental Stress Cracking of Polyethylene Pipes in Water Distribution
Networks ...................................................................................................................... 995
J. P. Dear, N. S. Mason and M. Poulton
Fatigue Crack Propagation in 2XXX Aluminium Alloys at 223K .............................. 997
C. Gasqueres, C. Sarrazin-Baudoux, D. Dumont and J. Petit
Hydrogen Assisted Cracking Paths in Oriented Pearlitic Microstructures .................. 999
J. Toribio and E. Ovejero
Effect of Residual Stress-Strain Profile on Hydrogen Embrittlement Susceptibility
of Prestressing Steel Wires......................................................................................... 1001
J. Toribio and V. Kharin
Hydrogen Embrittlement of Austenitic Stainless Steels at Low Temperatures ......... 1003
L. Zhang, M. Wen, M. Imade, S. Fukuyama and K. Yokogawa
Hydrogen Diffusion and EAC of Pipeline Steels under Cathodic Protection .......... 1005
M. Cabrini and T. Pastore
Initiation of Environmentally Assisted Cracking in Line Pipe Steel ......................... 1007
M. Elboujdaini
Fatigue Crack Growth Behaviour Depending on Environment in Magnesium
Alloys ........................................................................................................................ 1009
M. Nakajima, K. Tokaji, Y. Uematsu and T. Shimizu
xxxii Contents

Assessment of High-Temperature Hydrogen Degradation of Power Equipment

Steels .......................................................................................................................... 1011
H. M. Nykyforchyn and O. Z. Student
Stress Corrosion Cracking of 18MN-4CR Generator Rotor End-retaining Ring
Steel............................................................................................................................ 1013
N. Mukhopadhyay and U. K. Chatterjee

17. SIM, Philosophy, Instrumentation and Analysis ............................................ 1015

Non Contacting Stress Monitoring ............................................................................ 1015
W. D. Dover, R. F. Kare and N. Stone
Rapid Calculation of Stress Intensity Factors ............................................................ 1017
A. J. Love and F. P. Brennan
Variable Amplitude Corrosion Fatigue of High Strength Weldable Steel ................. 1019
S. S. Ngiam and F. P. Brennan
Crack Monitoring using ACFM................................................................................. 1021
R. F. Kare

18. Fracture of Biomaterials ................................................................................... 1023

Fatigue Behaviour of Fiber Reinforced Bone Cement .............................................. 1023
B. Kumar and F. W. Cooke
Fracture and Fatigue of Bone and Bone Cement: The Critical Distance Approach .. 1025
D. Taylor, D. Hoey, L. Sanz and P. O’Reilly
Fatigue Failure in Reconstracted Acetabula – a Hip Simulator Study....................... 1027
J. Tong, N. P. Zant and P. Heaton-Adegbile
Deformation and Fracture of Bioactive Particulate Composites Developed for
Hard Tissue Repair..................................................................................................... 1029
M. Wang
Failure of Biomaterials in Implant Fixation............................................................... 1031
P. J. Prendergast, J. R. Britton, P. T. Scannell and A. B. Lennon

19. Structural integrity Assessment in Theory and Practice ............................... 1033

Stress Analisys of High Pressure Steamlines in Thermal Power Plants .................... 1033
A. Jakovljevic
Laminar Composite Materials Damage Monitoring by Embedded Optical Fibers ... 1035
A. Kojovi, I. Zivkovi, L. Brajovi, D.Mitrakovi and R. Aleksi
SOL GEL Synthesis and Structure of Hybrid Nanomaterials with Strong Chemical
Bonds ......................................................................................................................... 1037
B. Samuneva, P. Djambaski, E. Kashchieva and G. Chernev
An Alternative Approach to Conventional Data Presentation of Fatigue.................. 1039
D. Angelova
Absorbers of Seismic Energy for Damaged Masonary Structures............................. 1041
D. Sumarac, Z. Petraskovi, S. Miladinovic, M. Trajkovi, M. Andjelkovic
and N. Trisovic
Numerical Analysis of Tensile Specimen Fracture with Crack in HAZ ................... 1043
G. Adziev, A. Sedmak and T. Adziev
Determination of JR-Curve by Two Points Method .................................................. 1045
I. Blacic and V. Grabulov
 Contents xxxiii

Monitoring of Stress-Strain State of Boiler During Pessure Test ............................. 1047

J. Kurai, Z. Burzic, N. Garic, M. Zrilic and B. Aleksic
Local Variation of Crack Driving Force in a Mismatched Weld ............................... 1049
J. Predan, N. Gubeljak and O. Kolednik
Strength Recovery of Machined Alumina by Self Crack Healing............................. 1051
K. Ando, K. Takahashi, W. Nakao, T. Osada and S. Sato
Crack Initiation and Growth in HAZ of Microalloyed Steel ..................................... 1053
K. Geric and S. Sedmak
Structural Integrity at Elevated Temperatures - Residual Service Life Evaluation ... 1055
L. Milovic and S. Sedmak
The Analysis of Supporting Structure of Planetary Gear Box Satellite .................... 1057
M. Arsi, V. Aleksi and Z. Anelkovi
Failure Probability of Gear Teeth Wear ..................................................................... 1059
M. Ognjanovic
Some Aspects of Engineering Approach to Structural Integrity Assessment............ 1061
M. Kiric and A. Sedmak
Structural Integrity Assessment Applying Ultrasonic Testing................................... 1063
M. Kiric
(Crack-Healing + Proof-Test): Methodology to Guarantee the Reliability of
Ceramics..................................................................................................................... 1065
M. Ono, W. Nakao, K. Takahashi, K. Ando and M. Nakatani
Risk Based Integrity Assessment of Concrete Structures .......................................... 1067
M. Pavisic
Structural Integrity Assessment by Local Approach to Fracture ............................... 1069
M. Zrilic, M. Rakin, Z. Cvijovic, A. Sedmak and S. Sedmak
Brittle and Ductile Fracture in Service of Pressure Vessels....................................... 1071
N. Filipovic and K. Geric
Mechanisms of Fracture in Medium Carbon Vanadium Microalloyed Steels ........... 1073
N. Radovic, Dj. Drobnjak and H. Hraam
Computation and Experimental Investigations of Notched Components Fatigue
Life Estimation .......................................................................................................... 1075
S. Maksimovic, Z. Burzic and K. Maksimovic
Failure Analysis of Layered Composite Structures: Computation and Experimental
Investigation............................................................................................................... 1077
S. Maksimovic
Loading Rate Effect on HSLA Steel Welded Joints Fracture Resistance ................. 1079
V. Grabulov, I. Blai, A. Radovi and S. Sedmak
Case Study of Supporting Tubes Failure.................................................................... 1081
V. S. Zeravcic, M. Djukic, G. Bakic, B. Andjelic and B. Rajicic
Structure Integrity of Pressure Vesels Repair Welding Joints.................................... 1083
V. S. Zeravcic, G. Bakic, M. Djukic and B. Rajicic
Effect of Microalloyed Steel Welding Procedure on Fatigue Crack Growth ............ 1085
Z. Burzic, V. Grabulov, M. Burzic, M. Manjgo, V. Gliha and T. Vuherer
Fracture Resistance of High-Strength 7000 Forging Alloys...................................... 1087
Z. Cvijovic, M. Rakin and M. Vratnica

20. Critical Distance Theories of Fracture............................................................. 1089

xxxiv Contents

Does a Characteristic Crack-Tip Distance Imply Discontinuous Crack

Propagation?............................................................................................................... 1089
A. P. Kfouri
A Multiaxial Criterion for Notch Fatigue Using a Critical-Distance Method .......... 1091
A. Carpinteri, A. Spagnoli, S. Vantatori and D. Viappiani
Size Effects for Crack Initiation at Blunt Notches or Cavities in Brittle Materials... 1093
D. Leguillon, E. Martin, D. Picard and C. Putot
The Theory of Critical Distances ............................................................................... 1095
D. Taylor
Strength Analysis of Composite Pinned Joints .......................................................... 1097
H. A. Whitworth, O. Aluko and N. Tomlinson
Application of the Theory of Critical Distance to Fretting Fatigue .......................... 1099
J. A. Araujo, L. Susmel, D. Taylor and L. H. M. Lopes
The Theory of Critical Distances: Applications in Fatigue ....................................... 1101
L. Susmel
Fatigue Assessment using an Integrated Threshold Curve Method - Applications... 1103
M. D. Chapetti
Anaytical Approaches vs Atomistic Simulations in Fracture ................................... 1105
N. Pugno, A. Carpinteri, M. Ippolito, A. Mattoni and L. Colombo
A Coupled Stress and Energy Criterion within Finite Fracture Mechanics............... 1107
P. Cornetti, N. Pugno, A. Carpinteri and D. Taylor
Local Strain Energy Density and Fatigue Strength of Welded Joints ........................ 1109
P. Lazzarin, P. Livieri and F. Berto
An Implicit Gradient Application to Fatigue of Notches and Weldments .................. 1111
R. Tovo and P. Livieri
Use of JVR to Predict Static Failures in Notched Components................................. 1113
P. Livieri
Standardization of Strength Evaluation Methods Using Critical Distance Stress ..... 1115
T. Hattori, N. Nishimura and M. Yamashita
Application of Point Stress Method to Hydro-Fracturing Tectonic Stress
Measurement.............................................................................................................. 1117
T. Ito
A Unified Failure Criterion for Brittle or Quasi-Brittle Materials under Arbitrary
Stress Concentration .................................................................................................. 1119
J. Li and X. B. Zhang

22. New Investigations on Very High Cycle Fatigue of Materials ....................... 1121
Morphology of Step-Wise S-N Curves Depending on Notch and Surface
Roughness in High Strength Steel.............................................................................. 1121
H. Itoga, K. Tokaji, M. Nakajima and H. N. Ko
Very High Cycle Fatigue Behaviour under Cyclic Torsion Loading ......................... 1123
H. Mayer and S. Stanzl-Tschegg
Modelling of Fatigue Crack Growth From Exfoliation and Pitting Corrosion.......... 1125
G. Clark, P. K. Sharp and R. Jones
Does Copper Undergo Surface Roughening during Fatigue in the VH Regime?...... 1127
S. Stanzl-Tschegg, H. Mughrabi and R. Schuller
Crack Initiation Mechanism of Bearing Steel in High Cycle Fatigue ....................... 1129
 Contents xxxv

T. Sakai
Very High Cycle Fatigue Behavior of High Strength Steels...................................... 1131
Y. Akiniwa, N. Miyamoto, H. Tsuru and K. Tanaka

23. Deformation and Fracture of Engineering Materials..................................... 1133

Fracture Toughness of Hydrided Zircaloy-4 Experimental and Numerical Study .... 1133
C. Langlade, P. Bouffioux and M. Clavel
Crack Growth Behavior in a Highly Filled Elastomer............................................... 1135
C. T. Liu , R. Neviere and G. Ravichandran
Crack Tip Behavior in TiAl when Approaching Grain Boundary ............................. 1137
F.-P. Chiang, S. Chang and K. Wang
Effect of Loading Rate on the Energy Release Rate in a Constrained Elastomeric
Disk ............................................................................................................................ 1139
H. K. Ching, C. T. Liu and S. C. Yen
Analyses of Progressive Damage and Fracture of Particulate Composite Materials
Using S-FEM Technique............................................................................................ 1141
H. Okada, S. Tanaka, Y. Fukui and N. Kumazawa
Fracture Mechanics on PVDF Polymeric Material : Specimen Geometry Effects.... 1143
L. Laiarinandrasana and G. Hochstetter
Fracture Toughness of Alloyed Austempered Ductile Iron (ADI)............................. 1145
O. Eric, D. Rajnovic, Z. Burzic, L. Sidjanin and M. T. Jovanovic
Prediction of Crack Growth under Random Load in Railway Wheel ....................... 1147
R. Hamam, S. Pommier and F. Bumbieler

24. Materials Damage Prognosis and Life Cycle Engineering............................. 1149

Predicting the Evolution of Stress Corrosion Cracks From Pits ................................ 1149
A. Turnbull, L. N. Mccartney and S. Zhou
Corrosion Problems in Nuclear Industry : Lessons Learned and Perspectives.......... 1151
J. M. Boursier, F. Foct, F. Vaillant and E. Walle
Aluminium Alloys Fatigue Evaluation Method......................................................... 1153
S. Rymkiewicz

25. Mixed-Mode Fracture........................................................................................ 1155

Singular Stress Fields Situations in Mode-II and Mixed-Mode Loaded Cracks ...... 1155
D. Fernández-Zúñiga, J. F. Kalthoff, A. Blázquez and A. Fernández-Canteli
Evaluation of M-Integral for Rubbery Material Problems Containing Multiple
Cracks......................................................................................................................... 1157
J.-H. Chang and D.-J. Peng
Use of a Crack Box Technique for Crack Bifurcation in Ductile Material................ 1159
D. Lebaillif, X. B. Zhang and N. Recho
Mixed Mode Fracture of Linear Elastic Materials with Cubic Symmetry ................ 1161
D. E. Lempidaki, N. P. O’Dowd and E. P. Busso
Three-Dimensional Experimental and Numerical SIFs and Crack Growth ............. 1163
D. M. Constantinescu, B. Bocaneala and L. Marsavina
An Arbitrarily Oriented Crack Near a Coated Fiber.................................................. 1165
H. M. Shodja and F. Ojaghnezhad
Simulation of the Mixed Mode Fracture of Concrete with Cohesive Models ........... 1167
xxxvi Contents

J. C. Gálvez, D. A. Cendón, E. Reyes, J. M. Sancho and J. Planas

Micromechanical Analysis of Rupture Mechanisms in Mixed Mode Ductile
Fracture ...................................................................................................................... 1169
I. Barsoum and J. Faleskog
Mode I Preloading-Mode II Fracture in Warm Pre-Stressing .................................... 1171
M. R. Ayatollahi and M. Mostafavi
Predictions of Mixed Mode I/II Fracture toughness for Soft Rocks.......................... 1173
M. R. Ayatollahi and M. R. M. Aliha
An Interface Model for Mixed-mode, Buckling-Driven Decohesion of Superficial
Layers......................................................................................................................... 1175
S. Bennati and P. S. Valvo
MXED-Mode Fracture Analyss of Orthotropc Functonally Graded Materals ......... 1177
S. Dag, B. Yildirim, D. Sarikaya
New Scheme for Fea of Mixed Mode Stable Crack Growth ..................................... 1179
S. K. Maiti , S. Namdeo and A. H. I. Mourad
Numerical Simulation of Nonlinear Crack Propagation under Mixed-Mode Impact
Loading ...................................................................................................................... 1181
T. Fujimoto and T. Nishioka
Elastic-Plastic Behaviour of Crack Propagation under Biaxial Cyclic Loading ....... 1183
T. Hoshide
Numerical Analysis of Mixed-Mode Cracking in Concrete Dams............................ 1185
Z. Shi

26. Fracture Mechanics Characterization of Wood.............................................. 1187

Species and Other Physical Effects on Parameters Describing a Wood Toughness
Test. ............................................................................................................................ 1187
B. Thibaut and J. Beauchene
Yew and Spruce Wood: Mechanical Properties and Fracture Surface Studies .......... 1189
D. Keunecke, C. Marki and P. Niemz
Critical Crack Lengths in FRP Reinforced Glulam Beams ....................................... 1191
J. Desjarlais, W. G. Davids and E. N. Landis
Failure Analysis of Engineering Wood Products ....................................................... 1193
I. Smith, M. Snow and A. Asiz
Modelization of Slow Crack Growth in Wood Considered as a Damage
Viscoelastic Material.................................................................................................. 1195
M. Chaplain and G. Valentin
Mode I Crack Propagation in Softwood, Microanalyses and Modeling.................... 1197
P. Navi and M. Sedighi-Gilani
Fracture Properties of Pine and Spruce in Mode I ..................................................... 1199
N. Dourado, S. Morel, M. F. S. F. De Moura, G. Valentin and J. Morais
Influence of the Specimen Geometry on R-Curve: Numerical Investigations. ......... 1201
C. Lespine, S. Morel, J.-L. Coureau and G. Valentin
Fracture Behaviour and Cutting of Small Wood Specimens in RT-Direction ........... 1203
S. Koponen and P. Tukiainen
Fracturing of Wood under Torsional Loading: Fracture Mechanisms and
Mechanics .................................................................................................................. 1205
E. K. Tschegg and S. E. Stanzl-Tschegg
 Contents xxxvii

On the Influence of Humidity Cycling on Fracture Properties of Wood ................... 1207

S. Vasic and S. Tschegg
Determination of Cohesive Fracture Parameters for Wood ....................................... 1209
T. Astrup, J. F. Olesen, L. Damkilde and P. Hoffmeyer
The Role of Fracture Toughness in the Cutting of Wood .......................................... 1211
T. Atkins

28. Short Fatigue Crack Growth under Multi-Axial Loading Conditions ......... 1213
Short Fatigue Cracks of In-Service Fatigued Turbine Blades ................................... 1213
A. A. Shanyavskiy, M. A. Artamonov, A. L. Tushentsov
and Yu. A. Potapenko
Short Crack Growth under Cyclic Torsion with Static Tension................................. 1215
I. Ohkawa, S. Hirano, T. Negishi and M. Misumi
Resistance-Curve Method for Predicting Fatigue Thresholds under Combined
Loading ...................................................................................................................... 1217
K. Tanaka, Y. Akiniwa and M. Wakita
The Growth of Short Cracks From Defects under Multi-Axial Loading................... 1219
M. Endo and A. J. Mcevily
Short Fatigue Cracks in Notched and Unnotched Specimens under
Non-Proportional Loading ......................................................................................... 1221
O. Hertel, T. Seeger, M. Vormwald, R. Doring and J. Hoffmeyer
Microcracks Growth in Push-Pull and Reversed Torsion in Stainless Steel.............. 1223
V. Doquet and G. Bertolino
Hydrogen and Notch Effects on Torsional Fatigue of Stainless Steel ....................... 1225
Y. Kondo, M. Kubota and K. Ohguma

30. Integrity of Gears ............................................................................................... 1227

Influence of Moving Tooth Load on Gear Fatigue Behaviour................................... 1227
D. T. Jelaska and S. Podrug
Comparison of Solid Spur Gear Face Load Factors .................................................. 1229
G. Marunic
Prediction of Contact Fatigue Internal Crack Propagation in Hypoid Gears............. 1231
M. Vimercati, M. Guagliano, L. Vergani and A. Piazza
Fatigue Crack Initiation Along Inclusion Interfaces of Contacting
Mechanical Elements ................................................................................................. 1233
S. Glodez, M. Ulbin and J. Flasker
Energy Based Gear Fault Diagnostics........................................................................ 1235
S. J. Loutridis
Crack Propagation in Gear Tooth Root...................................................................... 1237
S. Pehan, B. Zafosnik and J. Kramberger
Experimental Evaluation of Stress Intensity Factors in Spur Gear Teeth .................. 1239
V. Spitas , G. Papadopoulos, Th. Costopoulos and C. Spitas

35. High Temperature and Thermomechanical Fatigue ...................................... 1241

Isothermal and Thermomechanical Fatigue Behavior of the ODS Superalloy
PM1000...................................................................................................................... 1241
W. O. Ngala, G. Biallas and H. J. Maier
xxxviii Contents

Fatigue-Creep-Environment Interactions in a Directionally-Solidified Ni-Base

Superalloy .................................................................................................................. 1243
A. P. Gordon, M. M. Shenoy, R. W. Neu and D. L. McDowell
The Effects of Microstructure, Deformation Mode and Environment on Fatigue .... 1245
S. D. Antolovich and B. F. Antolovich
Comparing Fatigue Behaviour of TI6242 and Novel TIAL Intermetallics ............... 1247
T. K. Heckel, A. Guerrero-Tovar and H. J. Christ
A TBC Failure Model Based on Crack Number Density ......................................... 1249
X. Wu, Z. Zhang and R. Liu

36. Impact Failure of Laminated and Sandwich Composite Structures............. 1251

Impact Induced Composite Delamination: State and Parameter Identification via
Unscented Kalman Filter ........................................................................................... 1251
A. Corigliano, A. Ghisi and S. Mariani
Modelling Impact Damage in Sandwich Concept Structures .................................... 1253
A. Johnson and N. Pentecote
Punch Shear Behavior of Composites at Low and High Rates.................................. 1255
B. A. Gama and J. W. Gillespie Jr.
Repeated Impact Behaviour and Damage Progression of Glass Reinforced Plastics 1257
G. Belingardi, M. P. Cavatorta and D. S. Paolino
Impact Behaviour Modelling of a Composite Leading Edge Structure..................... 1259
G. Labeas and Th. Kermanidis
Bending Strength of Sandwich Panels with Different Cores After Impact ............... 1261
W. Goettner and H. G. Reimerdes
Energy Absorbing Ability of Sandwich Composite Structures ................................. 1263
J. P. Dear, W. Maruszewska, S. T. Oh and H. Lee
Impact Behaviour of Metal Foam Cored Sandwich Beams....................................... 1265
S. Mckown and R. A. W. Mines

37. Mesofracture and Transferability .................................................................... 1267

Stress Gradient at Notch Roots Using Volumetric Method ....................................... 1267
H. Adib and G. Pluvinage
Local Approach Use at Solution of Fracture Parameters Transferability .................. 1269
L. Jurasek, M. Holzmann and I. Dlouhy
Damage in Rubber-Modified Polymers : Experimental, Modelling and
Computational Aspects .............................................................................................. 1271
N. Belayachi, F. Zaïri, N. Benseddiq and M. Naït Abdelaziz
Failure Assessment Diagrams Based on the Criterion of Average Stress ................. 1273
Y. G. Matvienko

38. Damage in Composites - Damage Development in Composite Materials

& Structures - Models of Prediction ............................................................... 1275
Material Models for Damaged Composite Laminates ............................................... 1275
J. Varna
Raman Spectroscopy Assessment of Stiffness Reduction and Residual Strains
due to Matrix Cracking in Angle – PLY Laminates .................................................. 1277
P. Lundmark, D. G. Katerelos, J. Varna and C. Galiotis
 Contents xxxix

Physical Modelling of Failure Processes in Composite Materials............................. 1279

P. W. R. Beaumont
NCF Cross-PLY Laminates: Damage Accumulation and Degradation of Elastic
Properties ................................................................................................................... 1281
R. Joffe and D. Mattsson
Matrix Crack Initiation and Propagation in Laminates with Off-Axis PLIES........... 1283
N. Vrellos, S. L. Ogin and P. A. Smith
Stress Oscillation and Instability of Yielding in Polymers and Nanocomposites...... 1285
D. E. Mouzakis, G. Kandilioti, S. Tzavalas and V. Gregoriou
Prediction of Cyclic Durability of Woven Composite Laminates ............................. 1287
V. Tamuzs and K. Reifsnider

39. Aging Aerostructures......................................................................................... 1289

Repair of Corroded Aerospace Aluminium Panels Using Ultrasonic Impact
Treatment ................................................................................................................... 1289
C. A. Rodopoulos, S. Pantelakis, M. Liao and E. Statnikov
Fatigue Crack Initiation in Stress Concentration Areas............................................. 1291
C. Schwob, F. Ronde-Oustau and L. Chambon
Hydrogen Trapping: Deformation and Heat Treatment Effects in 2024 Alloy ......... 1293
H. Kamoutsi, G. N. Haidemenopoulos, V. Bontozoglou , P. V. Petroyiannis
and Sp. G. Pantelakis
An Integrated Methodology Assessing the Aging Behaviour of Aircraft Structures 1295
G. Labeas and I. Diamantakos
Numerical Investigation on the Tensile Behaviour of Pre-Corroded 2024
Aluminium Alloy ....................................................................................................... 1297
P. V. Petroyiannis, G. Labeas, Sp. G. Pantelakis, E. Kamoutsi,
V. Bontozoglou and G. N. Haidemenopoulos

40. Residual Stress and its Effects on Fatigue and Fracture................................ 1299
Assessment of Defects under Combined Primary and Residual Stresses .................. 1299
A. H. Sherry and M. R. Goldthorpe
Effect of Residual Stresses on the Crack Growth in Aluminum ............................... 1301
B. Kumar and J. E. Locke
Effect of the Cryogenic Wire Brushing on the Surface Integrity and the Fatigue
Life Improvement of the AISI 304 Stainless Steel Ground Components ................. 1303
N. B. Fredja, H. Sidhoma and C. Brahamb
Interaction of Residual Stress with Mechanical Loading in Ferritic Steels ............... 1305
A. Mirzaee-Sisan, C. E. Truman and D. J. Smith
Evaluation of Novel Post Weld Heat Treatment in Ferritic Steel Repair Welds
Based on Neutron Diffraction .................................................................................... 1307
C. Ohms, D. Neov, R. C. Wimpory and A. G. Youtsos
Surface Crack Development in Transformation Induced Fatigue of SMA Actuators 1309
D. C. Lagoudas, O. W. Bertacchini and E. Patoor
Finite Element Simulation of Welding in Pipes: a Sensitivity Analysis.................... 1311
D. Elias Katsareas , C. Ohms and A. G. Youtsos
Residual Stress Prediction in Letterbox-Type Repair Welds ..................................... 1313
L. Keppas, N. K. Anifantis, D. E. Katsareas and A. G. Youtsos
xl Contents

Effect of Reflection Shot Peening and Fine Grain Size on Improvement of Fatigue
Strength for Metal Bellows ....................................................................................... 1315
H. Okada, A. Tange and K. Ando
Viscosity Effect on Displacements and Stresses of a Two-Pass Welding Plate ........ 1317
W. El Ahmar and J. F. Jullien
Surface Integrity in High Speed Machining of TI-6WT.%Al-4WT.%V Alloy ......... 1319
J. D. P. Velasquez, B. Bolle, P. Chevrier and A. Tidu
Phase Transformation and Damage Elastoplastic Multiphase Model for Welding
Simulation .................................................................................................................. 1321
T. Wu, M. Coret and A. Combescure
The Present Sans Instrument and the New HFR-Petten Sans Facility Based
on a Cold Neutron Source.......................................................................................... 1323
O. Ucaa,B, C. Ohmsa, D. Neova and A. G. Youtsosa
Residual Stress Numerical Simulation of Two Dissimilar Material Weld Junctions. 1325
P. Gilles, L. Nouet and P. Duranton
Identification of Weld Residual Stress Length Scales for Fracture Assessment ....... 1327
P. J. Bouchard and P. J. Withers
High-Resolution Neutron Diffraction for Phase and Residual Stress Investigations 1329
P. Mikula and M. Vrana
Sensitivity of Predicted Residual Stresses to Modelling Assumptions...................... 1331
S. K. Bate, R. Charles, D. Everett, D. O’Gara1, A. Warren
and S. Yellowlees
Welding Effects on Thin Stiffened Panels ............................................................ 1333
T. T. Chau
Evaluation of Residual Stresses in Ceramic Polymer Matrix Composites Using
Finite Element Method............................................................................................... 1335
K. Babski, T. Boguszewski, A. Boczkowska, M. Lewandowska,
W. Swieszkowski and K. J. Kurzydlowski

41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) .. 1337

Towards Data-Driven Modeling and Simulation of Multiphysics Degrading
Systems ..................................................................................................................... 1337
J. G. Michopoulos and C. Farhat
Mathematical Modelling of Piezoceramic Transducer Performance in the
Presence of Material Defects ..................................................................................... 1339
T. A. Christensen, N. L. Andersen, and M. Willatzen
A Continuum Approach for Identifying Elastic Moduli of Composites.................... 1341
J. G. Michopoulos and T. Furukawa
Regularized Identification of Material Constants Using Multi-Objective
Gradient-Based Method ............................................................................................. 1343
T. Furukawa and J. G. Michopoulos
Loading and Material Features Influence on Piezoelectric Material Performance.... 1345
V. G. Degiorgi and S. A. Wimmer
Modeling of Plasma Chemical Deposition and Degradation of Silicon Thin Films . 1347
V. V. Krzhizhanovskaya, P. M. A. Sloot and Y. E. Gorbachev

42. Scaling and Size Effects ..................................................................................... 1349

 Contents xli

A Fractal Approach Interpretation for the Indentation Size Effect............................ 1349

A. Carpinteri and S. Puzzi
Description of Multi-Scaling Power Laws in Fracture and Strength......................... 1351
A. M. Korsunsky
The Spalling Failure Around Deep Excavations in Rock Masses ............................ 1353
A. P. Fantilli and P. Vallini
Scaling in Multiaxial Compressive Fracture.............................................................. 1355
A. S. Elkadi and J. G. M. van Mier
Fracture of Antarctic FY Sea Ice ............................................................................... 1357
J. P. Dempsey, S. Wang and D. M. Cole
Mixed Mode Fracture of Brickwork Masonry........................................................... 1359
J. C. Galvez, E. Reyes, M. J. Casati, J. M. Sancho, J. Planas
and D. A. Cendon
Geometric Scaling and Instability in FRP-Concrete Debonding ............................... 1361
K. V. Subramaniam, M. Ali-Ahmad and C. Carloni
A Simplified MCFT for Shear Capacity Scaling of R/C Beams ............................... 1363
M. T. Kazemi and V. Broujerdian
Interplay of Sources of Size Effects in Concrete Specimens..................................... 1365
M. Vorechovsky and D. Matesova
Scale Effect in Elastic and Strength Properties of Nanostructures ............................ 1367
O. S. Loboda , A. M. Krivtsov and N. F. Morozov
Fracture Toughness Assessment of a C-MN Steel Using Miniature Specimens ....... 1369
P. J. Apps, W. Geary, J. W. Hobbs and G. Wardle
Size Effect in the Bonding of Smooth and Deformed Bars: NSC versus HPC ......... 1371
P. Bamonte, D. Coronelli and P. G. Gambarova
Size Effect in the Cracking of Drying Soil ................................................................ 1373
P. C. Prat, A. Ledesma, and M. R. Lakshmikantha
Modelling of the Volume Effects Related to the Unixial Behaviour of Concrete.
From a Discontinuous to a Macroscopic Approach................................................... 1375
P. Rossi, J. L. Tailhan, J. Lombart and A. Deleurence
Size Effect and R-Curve in Quasibrittle Fracture ...................................................... 1377
S. Morel, E. Bouchaud and G. Valentin
Bifurcation and Size Effect in a Viscoelastic Non-Local Damageable Continuum... 1379
Th. Baxevanis, G. Pijaudier-Cabot and F. Dufour
Ultiscale Necessary and Sufficient Strength Criteria................................................. 1381
V. M. Kornev
Size Effects: Moving forwards................................................................................... 1383
X. Hu and K. Duan
An Experimental Study on Rapid Setting Concrete Repair Materials....................... 1385
J. P. Richards and Y. Xi

44. Multiple Cracking and Delamination .............................................................. 1387

Hierarchical Failure Modeling and Related Scale-Invariant Probability
Distributions of Strength ............................................................................................ 1387
D. A. Onishchenko
Interaction of Two Adhesively Bonded Weak Zones .............................................. 1389
I. V. Simonov and B. L. Karihaloo
xlii Contents

Multiple Cracking in Surface-Hardened Tensile Specimens and their Fracture

Mechanisms ............................................................................................................... 1391
L. S. Derevyagina, V. E. Panin, R. V. Goldstein, N. A. Antipina
and I. L. Strelkova
Fracture Criterion of Cracks Initiation and Growth................................................... 1393
M. Perelmuter
Interfacial Cracks Emanating from Partially Debonded Subsurface Circular Elastic
Inclusions ................................................................................................................... 1395
P. B. N. Prasad
Mechanics of Block Structures and its Applications to Geodynamics ...................... 1397
P. V. Makarov
Static and Dynamic Response of Multiple Delaminations ........................................ 1399
M. G. Andrews and R. Massabo
Modeling Crack Growth in Structure- Nonhomogeneous Medium under Complex
Stress State ................................................................................................................. 1401
R. V. Goldstein, Y. V. Zhitnikov and N. M. Osipenko
Nonideal Interface of a Bimaterial with Defects under Thermal Load...................... 1403
V. E. Petrova and K. P. Herrmann
Multiple Cracking Development at the Prefructure Stage of Ion Crystals ................ 1405
Y. Y. Deryugin, V. E. Panin, V. Hadjicontis, K. Mavromatou

Author Index ............................................................................................................ 1407

 Editor’s Preface

This volume contains two-page abstracts of the 698 papers presented at the “16th
European Conference of Fracture,” (ECF16) held in Alexandroupolis, Greece, July 3-7,
2006. The accompanying CD attached at the back cover of the book contains the full
length papers.
The abstracts of the fifteen plenary lectures are included in the beginning of the book.
The remaining 683 abstracts are arranged in 25 tracks and 35 special symposia/sessions
with 303 and 380 abstracts, respectively. The papers of the tracks have been contributed
from open call, while the papers of the symposia/sessions have been solicited by the
respective organizers. Both tracks and symposia/sessions fall into two categories,
namely, fracture of nanomaterials and structures and engineering materials and
structures with 88 and 595 papers, respectively.
Started in 1976, the European Conference of Fracture (ECF) takes place every two
years in a European country. Its scope is to promote world-wide cooperation among
scientists and engineers concerned with fracture and fatigue of solids. ECF16 was under
the auspices of the European Structural Integrity Society (ESIS) and was sponsored by
the American Society of Testing and Materials, the British Society for Stain
Measurement, the Society of Experimental Mechanics, the Italian Society for
Experimental Mechanics, and the Japanese Society of Mechanical Engineers. ECF16
focused in all aspects of structural integrity with the objective of improving the safety
and performance of engineering structures, components, systems and their associated
materials. Emphasis was given to the failure of nanostructured materials and
nanostructures and micro and nanoelectromechanical systems (MEMS and NEMS). The
technical program of ECF16 was the product of hard work and devotion of more than
150 world leading experts to whom I am greatly indebted. The success of ECF16 relied
solely on the dedication and titanic work of the members of the Scientific Advisory
Board, the pillars of ECF16. As chairman of ECF16 I am honored to have them on the
Board and have worked closely with them for a successful conference.
Fracture mechanics analysis has been successful for many years in the prevention of
failures of engineering materials and structures. It is based on the realistic assumption
that all materials contain crack-like defects from which failure initiates. New
technological developments, however, raise new challenges for fracture mechanics
research and development. Quasi-brittle materials including concrete, cement pastes,
rock, soil, etc. are being extensively used in engineering applications. Layered materials
and especially thin film/substrate systems are becoming important in small volume
systems used in micro and nanoelectromechancial systems (MEMS and NEMS).
Nanostructured materials are being introduced in our every day life. In all these problems
fracture mechanics plays a major role for the prediction of failure and safe design of
materials and structures. Failure of materials and structures at the micro and nano scale
levels are adequately addressed at ECF16 with 93 papers referred to in this area.
xliv Editor’s Preface

More than nine hundred participants attended ECF16, while more than eight hundred
fifty papers were presented, far more than any other ECF over a thirty year period. The
participants of ECF16 came from 49 countries. Roughly speaking 66% came from
Europe, 17% from the Americas, 8% from the Far East and 9% from other countries. I
am happy and proud to have welcomed in Alexandroupolis well-known experts who
came to discuss problems related to the analysis and prevention of failure in structures.
The tranquility and peacefulness of this small town provided an ideal environment for a
group of scientists and engineers to gather and interact on a personal basis. Presentation
of technical papers alone is not enough for effective scientific communication. It is the
healthy exchange of ideas and scientific knowledge, formal and informal discussions,
together with the plenary and contributed papers that make a fruitful and successful
meeting. Informal discussions, personal acquaintance and friendship play an important
role.
I am proud to have hosted ECF16 in the beautiful town of Alexandroupolis, site of
the Democritus University of Thrace and I am pleased to have welcomed colleagues,
friends, and old and new acquaintances.
I very sincerely thank the authors who have contributed to this volume, the symposia/
sessions organizers for their hard work and dedication and the referees who reviewed the
quality of the submitted contributions. Our sponsors’ support, give in various forms, is
gratefully acknowledged. The tireless effort of the members of the Organizing
Committee as well as of other numerous individuals, and people behind the scenes is
appreciated. I am deeply indebted to the senior students of the Department of Electrical
and Computer Engineering of the Democritus University of Thrace Messrs. N.
Tsiantoulas and S. Siailis for their hard work and dedication in the preparation of the
ECF16 website in a timely and efficient manner and the organization of the conference,
and for their efforts in helping me compile this volume. Finally, a special word of thanks
goes to Mrs. Nathalie Jacobs of Springer for the nice appearance of this book and her
kind and continuous collaboration and support.

January 2006 Emmanuel E. Gdoutos

Xanthi, Greece Editor

 ORGANIZING COMMITTEES

Scientific Advisory Board

Emmanuel E. Gdoutos (Chairman)

Track 1 (Nanomaterials and Nanostructures)

Awaji, H. (Japan), Bahr, D., (USA), Ballarini, R., (USA), Batra, R. (USA), Belytschko,
T. (USA), Berndt, C. (USA), Bhushan, B. (USA), Espinosa, H. (USA), Friedrich, K.
(Germany), Karimi, A. (Switzerland), Kouris, D. (USA), Lagoudas, D. (USA), Meletis,
E.I. (USA), Michel, B. (Germany), Moody, N. (USA), Plumbridge, W.J. (UK),
Pluvinage, G. (France), Ruoff, R. (USA), Sih, G.C. (China), Zhang, Z. (Germany), Zhou,
M. (USA).

Track 2 (Engineering Materials and Structures)

Akid, R. (UK), Aliabadi, M.H. (UK), Andrianopoulos, N. (Greece), Angelova, D.
(Bulgaria), Aravas, N. (Greece), Atkins, A.G. (UK), Banks-Sills, L. (Israel), Bartolozzi,
F. (Italy), Barton, J. (UK), Bartzokas, D. (Greece), Bathias, C., (France), Bazant, Z.
(USA), Beaumont, P. (UK), Beretta, S. (Italy), Blackman, B. (UK), Brocks, W.
(Germany), Bunch, J. (USA), Cacko, J. (Slovakia), Carpinteri, Alberto (Italy),
Carpinteri, Andrea (Italy), Chona, R. (USA), Daniel, I.M. (USA), Danzer, R. (Austria),
Dietzel, W. (Germany), Dover, B. (UK), Exadaktylos, G. (Greece), Fernando, U.S.
(UK), Ferro, G. (Italy), Finnie, I. (USA), Fischer, F.D. (Austria), Fleck, N. (UK), Freitas,
M.J.M. (Portugal), Gabetta, G. (Italy), Galiotis, C. (Greece), Georgiadis, H.G. (Greece),
Goldstein, R. (Russia), Gosz, M. (USA), Gubeljack, N. (Slovenia), Hawong, J.S.
(Korea), Hedrich, K. (Republic of Serbia), Hopkins, S. (USA), Ingraffea, A. (USA),
Isogimi, K. (Japan), Jelaska, D. (Croatia), Jirasek, M. (USA), Johnson, E. (Sweden),
Jones, R. (Australia), Kalluri, S. (USA), Kalthoff, J.F. (Germany), Karalekas, D.
(Greece), Karihaloo, B. (UK), Kassner, M. (USA), Kermanidis, Th. (Greece), Kienzler,
R. (Germany), Kocanda, D. (Poland), Konsta-Gdoutos, M. (Greece), Kostopoulos, V.
(Greece), Kourkoulis, S. (Greece), Landes, J. (USA), Lee, B-L (USA), Lee. O.S.
(Korea), Lee, S.B. (Korea), Liolios, A. (Greece), Liu, C.T. (USA), Luxmoore, A.R.
(UK), MacGillivray, H. (UK), Mai, Y.W. (Australia), Maier, H.J. (Germany), Marioli-
Riga, Z. (Greece), Markenskoff, X. (USA), Massabo, R. (Italy), Matczynski, M.
(Poland), Matysiak, S. (Poland), Mayer, H.R. (Austria), McEvily, A. (USA),
Michopoulos, J. (USA), Mines, R. (UK), Mitchell, M. (USA), Moscovic, R. (UK),
Murakami, Y. (Japan), Needleman, A. (USA), Neimitz, A. (Poland), Neu, R. (USA),
Nied, H. (USA), Nilsson, F. (Sweden), Nurse, A. (UK), Nykyforchyn, H. (Ukraine),
Paipetis, S.A. (Greece), Panoskaltsis, V. (USA), Pantelakis, S. (Greece), Papadopoulos,
G. (Greece), Papakaliatakis, G. (Greece), Pappalettere, C. (Italy), Patterson, E. (UK),
Pavan, A. (Italy), Petit, J. (France), Pineau, A. (France), Pokluda, J. (Czech Republic),
xlvi Organizing Committees

Pook, L. (UK), Prakash, V. (USA), Prassianakis, J. (Greece), Rajapakse, Y.D.S. (USA),

Ravi-Chandar, K. (USA), Ravichandran, R. (USA), Rodopoulos, Chr. (UK), Rosakis, A.
(USA), Rossmanith, H.P. (Austria), Saxena, A. (USA), Sciammarella, C. (Italy),
Sedmak, S. (Republic of Serbia), Shah, S. (USA), Shukla, A. (USA), Soboyejo, W.
(USA), Sotiropoulos, D. (Greece), Spyropoulos, C. (Greece), Stanzl-Tschegg, S.
(Austria), Staszewski, W.J. (UK), Steen, M. (Netherlands), Subhash, G. (USA),
Sumarac, D. (Republic of Serbia), Sun, C.T. (USA), Sutton, M. (USA), Tamuzs, V.
(Latvia), Taylor, D. (Ireland), Theotokoglou, S. (Greece), toor, P. (USA), toth, L.
(Hungary), Tsamasphyros, G. (Greece), Tvergaard, V. (Denmark), Unger, D. (USA),
van Mier, J.G.M. (Switzerland), Vardoulakis, Y. (Greece), Vodenicharov, S. (Bulgaria),
Wallin, K. (Finland), Wardle, G. (UK), Williams, J.G. (UK), Withers, P. (UK), Yates, J.
R. (UK), Youtsos, A.G. (Netherlands), Zacharopoulos, D.A. (Greece).

LOCAL ORGANIZING COMMITTEE

Emmanuel E. Gdoutos (Chairman)

Z. Adamidou, P. Kalaitzidis, M.S. Konsta-Gdoutos, G. Papakaliatakis, S. Sailis, N.

Tsiantoulas, D.A. Zacharopoulos
 ECF16 TRACKS

B: TRACKS

B1: Nanomaterials and Nanostructures

1T1. Fracture and Fatigue of Nanostuctured Materials
1T2. Failure Mechanisms
1T4. Fatigue and Fracture of MEMS and NEMS
1T7. Thin Films
1T9. Failure of Nanocomposites

B2: Engineering Materials and Structures

2T1. Physical Aspects of Fracture
2T2. Brittle Fracture
2T3. Ductile Fracture
2T4. Nonlinear Fracture Mechanics
2T5. Fatigue and Fracture
2T8. Polymers, Ceramics and Composites
2T11. Fracture Mechanics Analysis
2T13. Probabilistic Approaches to Fracture Mechanics
2T14. Computational Fracture Mechanics
2T15. Experimental Fracture Mechanics
2T16. Creep Fracture
2T17. Environment Assisted Fracture
2T18. Dynamic, High Strain Rate, or Impact Fracture
2T19. Damage Mechanics
2T21. Concrete and Rock
2T22. Sandwich Structures
2T23. Novel Testing and Evaluation Techniques
2T26. Structural Integrity
2T28. Mesofracture Mechanics
2T32. Micromechanisms in Fracture and Fatigue
 ECF16 SPECIAL SYMPOSIA/SESSIONS

C: SPECIAL SYMPOSIA/SESSIONS

C1: Nanomaterials and Nanostructures

1. Fracture and Fatigue at the Micro and Nano Scales (Organized by H.D. Espinosa and
I.M.Daniel)
3. Nanoscale Deformation and Failure (Organized by M. Zhou)
29. Reliability and Failure Analysis of Electronics and Mechanical Systems (O.S. Lee)
31. Multiscaling in Molecular and Continuum Mechanics – Scaling in Time and Size from Macro
to Nano (Organized by G.C. Sih)
34. Cracks in Micro- and Nanoelectronics (Organized by B. Michel)
43. Interfacial Fracture in Composites and Electronic Packaging Materials (Organized by C.T. Sun
and T. Ikeda)

C2: Engineering Materials and Structures

4. Fracture and Fatigue of Elastomers (Organized by C. Bathias and E. Bayraktar)
5. Integrity of Dynamical Systems (Organized by K. Hedrih)

8. Modelling of Material Property Data and Fracture Mechanisms (Organized by R. Moskovic)

9. Micromechanisms in Fracture and Fatigue (Organized by J. Pukluda and R. Pippan)
12. Interface Fracture and Behavior of Joints (Organized by L. Banks-Sills)
13. Computational Fracture Mechanics (Organized by T. Belytschko and A. Gravouil)
14. Cohesive Models of Fracture (Organized by W. Brocks)
16. Environment Assisted Fracture (Organized by G. Gabetta, W. Dietzel and H. Nykyforchyn)
17. SIM, Philosophy, Instrumentation and Analysis (Organized by W. D. Dover)
18. Fracture of Biomaterials (Organized by J. tong)
19. Structural Integrity Assessment in Theory and Practice (Organized by S. Vodenitsarov and S.
Sedmak)
20. Critical Distance Theories of Fracture (Organized by D. Taylor)
22. New Investigations on Very High Cycle Fatigue of Materials (Organized by H. Mayer and S.
Stanzl-Tschegg)
23. Deformation and Fracture of Engineering Materials (Organized by C.T. Liu)
24. Materials Damage Prognosis and Life Cycle Engineering (Organized by R. P. Wei, G. Harlow,
A. Ingraffea and J. Larsen)
25. Mixed-Mode Fracture (Organized by M. Gosz)
l ECF16

26. Fracture Mechanics Characterization of Wood (Organized by S. Stanzl-Tschegg)

28. Short Fatigue Crack Growth under Multi-axial Loading Conditions (Organized by Y.
Murakami and A.J. McEvily)
30. Integrity of Gears (Organized by D. Jelaska)
33. Fracture and Failure of Natural Building Stones Applications in the Restoration of Ancient
Monuments (Organized by S. Kourkoulis)
35. High Temperature and Thermomechanical Fatigue (Organized by R.W. Neu, S. Kalluri and
H.J. Maier)
36. Impact Failure of Laminated and Sandwich Composite Structures (Organized by R. Mines)
37. Mesofracture and Transferability (Organized by G. Pluvinage)
38. Damage in Composites - Damage Development in Composite Materials & Structures - Models
of Prediction (Organized by C. Galiotis)
39. Aging Aerostructures (Organized by S. Pantelakis)
40. Residual Stress and its Effects on Fatigue and Fracture (Organized by A.G. Youtsos and P.J.
Withers)
41. Computational Modeling of Multiphysics Degrading Systems (CMMDS), (Organized by J.
Michopoulos)
42. Scaling and Size Effects (Organized by Z.P. Bazant and M. Jirasek)
44. Multiple Cracking and Delamination (Organized by R. Goldstein and R. Massabo)
A. INVITED PAPERS
 Invited Papers 3

DEFORMATION AND FRACTURE AT THE MICRON AND NANO SCALES

E. C. Aifantis
Laboratory of Mechanics and Materials, Polytechnic School, Aristotle University of Thessaloniki,
54124, Thessaloniki, Greece
mom@mom.gen.auth.gr

The progressively increasing demands of new science and technology to understand the behavior
of materials/components and processes at the micrometer and nanometer regimes has led in the
mid seventies to the development of micromechanics. In the mid nineties a new term
nanomechanics was used by the author to indicate the forthcoming excessive activity in this field
and point out to the need for new constitutive equations and mechanics tools to be developed in
relation to the emerging fields of nanotechnology. In fact, it was only a few years earlier, that the
first carbon nanotubes were produced in Japan - a unique example for the use of elasticity theory
at the nanoscale - and the first bulk nanopolycrystals were produced in Russia. At the same time
the first experimental observations on deformation and fracture mechanisms of nanopoly-
crystalline thin films were reported by the author and his co-workers in US. It was reported, among
other things, that plastic deformation at the nanoscale does not take place through lattice
dislocation activity but through grain boundary processes including material rotation and mass
diffusion. Moreover, fracture processes occur through nanovoid nucleation and coalescence. Some
of these experimental observations were numerically verified a few years later through molecular
dynamics (MD) multimillion atom simulations.
A first attempt to develop constitutive equations for describing deformation and fracture at the
nanoscale is outlined by the author and co-workers with two specific concepts being advanced: the
use of a mixture argument for “bulk” and “grain boundary” states and the resort to non-locality for
describing the state of stress and strain at the nanoscale. In very recent years, due to the most
promising developments in nanosciences/nanotechnologies in conjunction with the rapidly
evolved computational advances, coupled ab initio/atomistic/molecular dynamics/finite element
calculations have been employed to simulate the mechanical response of matter at the nanoscale in
thin film and bulk configurations within the so-called multiscale modeling approach. A variety of
nano-objects, including multilayered films, nanocomposites, proteins, as well as other metal or
non-metal and biological nanostructures are modeled within such multiscale modeling framework
and new nano-testing procedures and nano-apparatuses have been developed to capture this
response experimentally. HRTEM, STM, AFM, micro/nano tensile machines and micro/nano
indenters are among some of the new experimental tools for probing the mechanical response of
materials and structures at the micro/nano scale and designing MEMS/NEMS devices for a variety
of electromechanical and biomedical applications.
In an effort to have a general micro/nano mechanics framework as useful as the continuum
mechanics model that has been employed so successfully for the understanding of the mechanical
behavior at the macroscale and the design of macroscopic components and structures, a proposal
will be presented by a straightforward extension/generalization of the macroscopic continuum
model. This generalization is based on the concept of a micro/nano continuum which is capable of
exchanging mass, momentum and energy with its bounding surface. While such a model has been
suggested by the author more than twenty five years ago, it was not until recently that its
implications to elasticity, plasticity, and other continuum theories of structural defects was fully
explored. The two new basic ingredients of the model are: a) the appearance of deterministic
higher order spatial and time derivatives in the governing equations of mechanical fields; and b)
4 E. C. Aifantis

the appearance of stochastic terms due to random effects associated with the nucleation and
evolution of deformation events at the micron/nano scale regime.
Several benchmark problems are considered to illustrate the applicability of the proposed
framework, as follows: (i) Elimination of Singularities. Gradient elasticity models are shown to
eliminate the strain and stress singularities form dislocation/disclination lines and crack tips. On
the basis of these solutions new relations can be obtained for the strength, energy and interaction of
defects in nanocrystals and new fracture criteria can be derived at the nanoscale. (ii) Internal
Stress, Elastic Constants and Yield Strength: Size Effects. Gradient elasticity and gradient
plasticity are shown to produce new formulas for the determination of internal stress and elastic
moduli in micro/nano multilayers and micro/nano plates under bending. A modification of the
well-known Stoney formula is an example. Also the dependence of the elastic modulus on the
nanoplate thickness is another interesting example. In this connection, it is pointed out that the size
dependence of yield strength of micro/nano columns and the dependence of Young’s modulus/
failure stress on nanotube diameter has also been documented. (iii) Micro/Nano Indentation.
Various basic formulas that have been used for determining material properties at the macroscale
during indentation are revisited by employing gradient elasticity/plasticity with or without
stochastic terms. Displacement bursts, load-depth serrations, and size-dependent hardness are all
phenomena that are often observed during micro/nano indentation and their proper interpretation
can assist in the determination of deformation and fracture properties at these scales. (iv)
Localization of Deformation and Multiple Shear Banding. The interesting features of deformation
and fracture at the micro/nano scale are concerned with the determination of the critical grain sizes
where a plasticity transition mechanism takes place. At the nanoscale (1-100 nm) the critical grain
size determines the transition from grain rotation/sliding to massive dislocation motion, which
often manifests itself by the appearance of an inverse Hall-Petch behavior. At the ultra-fine grain
size regime (100-1000 nm) another plasticity mechanism that occurs is the so-called multiple shear
banding which often manifests itself by the appearance of a perfectly plastic behavior in the
corresponding stress-strain curve. These two plasticity mechanism transitions will be discussed
within the proposed unified material micro/nano mechanics framework.
In concluding, it will be pointed out why new techniques such as fractals, wavelets and time-
series are often necessary for capturing details and additional features of micro/nano deformation
and fracture. Information on some of the above topics can be found in references [1]-[3] and
articles quoted therein.

References
1. Aifantis, E.C., In Recent Advances in Applied Mechanics (Honorary Volume for Academician
A.N. Kounadis), edited by T. Katsikadelis, D.E. Beskos and E.E. Gdoutos, NTUA, Athens,
2000, 243-254.
2. Aifantis, E.C., Mech. Mat., vol. 35, 259-280, 2003.
3. Konstantopoulos, I., Tragoudaras, D., Mokios, G., Konstantinidis, A., Zaiser, M. and
Aifantis, E.C., Research in Progress.
 Invited Papers 5

STATISTICAL MECHANICS OF SAFETY FACTORS AND SIZE EFFECT IN

QUASIBRITTLE FRACTURE

Z. P. Bazant and S.-D. Pang

McCormick Institute Professor and W.P. Murphy Professor of Civil Engineering
and Materials Science, Northwestern University,
Graduate Research Assistant
z-bazant@northwestern.edu

Throughout most of the 20th century, it was widely believed that the size effect on structural
strength has a purely statistical origin, explained by extreme value statistics based on the weakest
link model, and described by Weibull statistical theory of random strength. However, beginning
with the first suggestions made already in the early 1970s, it gradually transpired that, in
quasibrittle materials (i.e. heterogeneous brittle materials with a non-negligible fracture process
zone), the mean size effect is essentially deterministic, stemming from energy release caused by
stress redistribution in a structure prior to maximum load. The quasibrittle energetic scaling
bridges three simple asymptotic power-law scalingsthose of linear elastic fracture mechanics,
plasticity, and Weibull theory. Renormalization group transformation does not suffice to handle
the transitional nature of this quasibrittle size effect, often spanning several orders of magnitude of
size. As is now widely accepted, quasibrittle materials including concrete, rock, tough ceramics,
sea ice, snow slabs and composites exhibit major size effects on the mean structural strength that
are largely or totally deterministic in nature, being caused by stress redistribution and energy
release associated with stable propagation of large fractures or with formation of large zones of
distributed cracking.
The lecture begins by reviewing the general asymptotic properties of size effect implied by the
cohesive crack model or crack band model, and highlights the use of asymptotic matching
techniques as a means of obtaining scale-bridging size effect laws representing a smooth transition
between two power laws. Asymptotic matching is a range of diverse techniques widely used in
fluid mechanics, but overlooked in solid mechanics. Presented is a method of asymptotic matching
which is suitable for structural strength problems. The method is based on power series expansion
of the governing equation written as a function of dimensionless variables. The key idea is to
choose these variables in such a way that, at each asymptotic state, all of them vanish except one.
Attention is then focused on the size effects observed in fiber-polymer composites failing
either by tensile fracture or by propagation of compression kink bands with fiber micro-buckling.
The size effects in polymeric foams and sandwich structures are also discussed. Nonlocal
probabilistic analysis of the size effect on the statistical distribution of nominal strength of
structures is outlined and discussed from the viewpoint of the extreme value statistics. Implications
for the design of hulls, bulkheads, decks, masts and antenna covers for very large ships, and for the
design of large load-bearing aircraft fuselage panels, are pointed out.
The problem of estimating loads of extremely small failure probability, such as 10-7, required
for design is investigated. Attention is focused on the type 1 size effect, occurring in structures
failing at crack initiation, which is the only type for which material randomness affects not only the
scatter but also the mean of nominal strength. It is shown that a reform of structural reliability
concepts is necessary because of a strong effect of structure size (or brittleness) on failure
probability in the far-out tail of the cumulative probability distribution function (cdf) of structural
strength. The cdf is modeled by a chain of representative volume element (RVE) of the material,
each of which is represented by a hierarchical hybrid series-parallel coupling model. Each micro-
6 Z. P. Bazant and S.-D. Pang

element of this model simulates one of the micro-bonds within the RVE. From Maxwell-
Boltzmann distribution of thermal energies of atoms and the effect of applied stress on the
activation energy, it is deduced that the left tail of cdf of a RVE (for failure probabilities < 0.0001
to 0.01) must be a power law, while there must be a broad Gaussian core because of parallel
couplings within the RVE. The amplitude of the power law tail is obtained as a function of
temperature and load duration.
A chain-of-RVEs model is proposed to model a gradual transition of strength cdf from a
mostly Gaussian pdf with a short Weibull tail for small structures to a purely Weibull cdf for very
large structures. The equivalent number of RVEs for which the chain represents the cdf of strength
of a structure with nonuniform stress field is expressed in terms of Weibull integral according to
nonlocal Weibull theory. The center of the Weibull-Gaussian transition moves along the cdf from
left to right as a function of the equivalent number Neq of RVEs in the chain-of-RVEs model.
Matching of this model serves to calibrate a smooth analytical expression for the transitional cdf,
varying from purely Gaussian for zero size (or zero brittleness) to purely Weibull for infinite size
(or perfect brittleness) as a function of the structure size as well as geometry. The distance from the
mean to a point of a tolerable failure probability such as 10-7 or 10-6 on this transitional cdf is
shown to be strongly size and geometry dependent, and nearly doubles while passing from very
small to very large structures. To capture this major effect, it is necessary to introduce a correction
into Cornell's and Hasofer-Lind's reliability indices (known from the classical first-order reliability
method, or FORM).
To make reliability assessments realistic, it is further necessary that the 'covert' understrength
factors implied in brittle failure provisions of concrete design codes be made overt, and that a
'covert' size effect implied by excessive load factor for self-weight acting alone be eliminated.
Adaptation of the stochastic finite element method to cope with extreme value statistics of
energetic-statistical size effect is described, and its importance is demonstrated by analysis of some
famous disasters, particularly Malpasset Dam. To improve design safety and efficiency, experts in
statistical reliability and fracture mechanics will need to collaborate to tackle these problems in a
comprehensive manner
To improve design safety and efficiency, experts in statistical reliability and fracture
mechanics will need to collaborate to tackle these problems in a comprehensive manner

References
1. Bažant, Z.P., "Scaling theory for quasibrittle structural failure." Proc., National Academy of
Sciences, 101 (37), 13397-13399, 2004.
2. Bažant, Z.P., Scaling of Structural Strength. Hermes Penton Science, London, 2002. (French
translation, Hermes, Paris 2004); 2nd ed., Elsevier 2005, in press.
3. Bažant, Z.P., "Probability distribution of energetic-statistical size effect in quasibrittle
fracture." Probabilistic Engineering Mechanics, 19(4), 307-319, 2004.
 Invited Papers 7

“NANORELIABILITY” – FRACTURE MECHANICS ON THE WAY FROM

MICRO TO NANO

B. Michel
Fraunhofer Micro Materials Center at IZM Berlin
Gustav-Meyer-Allee 25, D-13355 Berlin, Germany
bernd.michel@izm.fraunhofer.de

“Thermo-mechanical compatibility” (TMC) of different micro- and nanomaterials in electronic

and mechatronic systems is one mayor reason for troubles in the field of reliability and life-time of
“high-tech” applications. Problems of thermal misfit (or mismatch) have become more and more
important in the recent years because more than 60 percent of failure events in modern
microelectronics are more or less directly connected with thermal misfit problems. The interface
regions between the different materials, e.g. of a chip interconnection layer within a microsensor or
a microactuator, is very important for the reliability of the component and of the whole system.
Interface cracks, therefore, have to be dealt with in details in modern microsystem and
nanotechnology as well. The author is going to present a survey on modern crack, fracture and
reliability concepts of Microsystems in the chip interconnection region, where a lot of very
different materials are “playing the concert”. Polymers, metals, ceramic materials and composites
of very different kind will be shown to excert their specific influence on the local deformation and
crack behaviour. Fatigue, creep and moisture effects and their complicated interactions have to be
taken into account for reliable lifetime prognoses.
The author presents some new experimental techniques based on modern digital image
correlation techniques (DIC) which enable to take into account local and global effects as well.
The so-called microDAC and nanoDAC deformation analyses provide very good means for local
crack field analysis. These methods are combined with numerical field calculation and advanced
reliability and lifetime concepts and lead to very good results for time-dependent failure analysis of
Microsystems. In electronic packaging of MEMS and sensor components nanomechanics effects
have been shown to become more and more important. The author presents recent results of his
group to include focus ion beam technique (FIB), AFM analysis and advanced methods of
materials testing (e.g. nanoindentation, nanoDMA, local stress and strain analysis) into the crack
evaluation procedure in the micro-nano interface regions. Besides the above mentioned techniques
the author also is going to outline his opinion about the perspectives of modern fracture mechanics
in the field of micro- and nanotechnologies applied to the electronics, automotives and above all IT
branches.

References
1. Michel, B., Testing at Micro and Nanoscale, EuroSIME, European Conf. On Thermal,
Mechanical and Multiphysics Simulation and Experiments in Microelectronics, Berlin, 18-20
April 2005.
2. Michel, B., Keller, J., NanoDAC – A New Technique for Micro- and Nanomechanical
Reliability Analysis of Lead-free Solder Interconnects, Int. Conf. on Lead-free Soldering,
Toronto, Canada, 24-26 May 2005.
3. Walter, H., Dudek, R., Michel, B., Fracture and Fatigue Behaviour of MEMS Related Micro
Materials, Int. Conf. on Fracture (ICF 11), Turin, Italy, 20-25 March 2005.
8 B. Michel

4. Michel, B., Fracture Electronics and Thermo-Mechanical Compatibility (TMC) of

Microcomponents in High-Tech Systems, Int. Conf. Micro Materials 2000, Berlin, 17-19
April 2000.
5. Michel, B., Experimental Mechanics on the Way from Micro to Nano, Exp. Technique 29
(2005) 2, 3-5.
 Invited Papers 9

FRACTURE MECHANICS AND COMPLEXITY SCIENCES

A. Carpinteri and S. Puzzi

Politecnico di Torino, Department of Structural and Geotechnical Engineering
Corso Duca degli Abruzzi 24, 10129 Torino, Italy
alberto.carpinteri@polito.it

The so-called Complexity Sciences are a topic of fast growing interest inside the scientific
community. Actually, researchers did not come to a definition of complexity, since it manifests
itself in so many different ways [1]. This field itself is not a single discipline, but rather a
heterogeneous amalgam of different techniques of mathematics and science. In fact, under the
label of Complexity Sciences we comprehend a large variety of approaches: nonlinear dynamics,
deterministic chaos theory, nonequilibrium thermodynamics, fractal geometry, intermediate
asymptotics, complete and incomplete similarity, renormalization group theory, catastrophe
theory, self-organized criticality, neural networks, cellular automata, fuzzy logic, etc.
Complex systems lie somehow in between perfect order and complete randomness –the two
extreme conditions that are likely to occur only very seldom in nature– and exhibit one or more
common characteristics, such as: sensitivity to initial conditions, pattern formation, spontaneous
self-organization, emergence of cooperation, hierarchical or multiscale structure, collective
properties beyond those directly contained in the parts, scale effects.
Aim of this paper is to provide an insight into the role of complexity in the field of Materials
Science and Fracture Mechanics. The included examples will be concerned with the snap-back
instabilities in the structural behaviour of composite structures (Carpinteri [2-3]), the occurrence of
fractal patterns and self-similarity in material damage and deformation of heterogeneous materials,
and the apparent scaling on the nominal mechanical properties of disordered materials (Carpinteri
[4,5]). Further examples will deal with criticality in the acoustic emissions of damaged structures
and with scaling in the time-to-failure (Carpinteri et al. [6]). Eventually, results on the transition
towards chaos in the dynamics of cracked beams will be reported (Carpinteri and Pugno [7]).

References
1. Garrido, M.S. and Vilela Mendes, R., Complexity in physics and technology, World
Scientific, Singapore, 1992.
2. Carpinteri, A., In Application of Fracture Mechanics to Cementitious Composites
(Proceedings of a NATO Advanced Research Workshop, Evanston, USA, 1984), edited by
S.P. Shah, Martinus Nijhoff Publishers, Dordrecht, 1985, 287-316.
3. Carpinteri, A., J. Mech. Phys. Solids, vol. 37, 567-582, 1989.
4. Carpinteri, A., Mech. Mater., vol. 18, 89-101, 1994.
5. Carpinteri, A., Int. J. Solids Struct., vol. 31, 291-302, 1994.
6. Carpinteri, A., Lacidogna, G. and Pugno, N., In Fracture Mechanics of Concrete and
Concrete Structures (Proceedings of the 5th International FraMCoS Conference, Vail,
Colorado, USA, 2004), edited by V.C. Li et al., 2004, vol. 1, 31-40.
7. Carpinteri, A. and Pugno, N., J. Appl. Mech., in print.
 Invited Papers 11

FAILURE OF COMPOSITE MATERIALS

I. M. Daniel
Robert R. McCormick School of Engineering and Applied Science
Northwestern University, Evanston, IL 60208, USA
imdaniel@northwestern.edu

The failure of composites has been investigated extensively from the micromechanical and
macromechanical points of view. On the micromechanical scale, failure mechanisms and processes
vary widely with type of loading and are intimately related to the properties of the constituent
phases, i. e., matrix, reinforcement, and interface-interphase. Failure predictions based on
micromechanics, even when they are accurate with regard to failure initiation at critical points, are
only approximate with regard to global failure of a lamina and failure progression to ultimate
failure of a multi-directional laminate. For these reasons a macromechanical approach to failure
analysis is preferred.
Numerous failure theories have been proposed and are available to the composite structural
designer[1].They are classified into three groups, limit or noninteractive theories (maximum
stress, maximum strain); interactive theories (Tsai-Hill, Tsai-Wu); and partially interactive or
failure mode based theories (Hashin-Rotem, Puck). The validity and applicability of a given
theory depend on the convenience of application and agreement with experimental results. The
plethora of theories is accompanied by a dearth of suitable and reliable experimental data, which
makes the selection of one theory over another rather difficult. Considerable effort has been
devoted recently to alleviate this difficulty. The problem can be divided in two parts, one being the
prediction of failure of a single lamina and the second dealing with prediction of first-ply-failure
and damage progression leading to ultimate failure of a multi-directional laminate.
C. T. Sun [2] reviewed six failure theories and showed comparisons of theoretical predictions
with experimental results for six different composite material systems and various loading
conditions. A round robin exercise was initiated by Hinton, Soden, and Kaddour for the purpose of
assessing the predictive capabilities of current failure theories [3]. The difficulty in evaluating
failure theories is much greater in the case of a multidirectional laminate. The scope of the
proposed laminate failure analysis comprises the following [1]:
1 A selected or adapted lamina failure theory for prediction of failure initiation, i.e., first-ply
failure (FPF) in the laminate.
2 A failure mode discrimination rule and a scheme of ply discounting and failure progression
in the laminate after FPF.
3 A criterion or definition of ultimate laminate failure (ULF).
In general, a wide variation has been observed in the prediction of laminate failures by the
various theories. The divergence in the predictions is greater for FPF than for ULF; also is greater
for matrix dominated failures than for fiber dominated ones. The divergence observed may be
attributed primarily to the following factors:
1 The different ways in which curing residual stresses are introduced in the predictions,
especially in the case of first-ply-failure.
2 The concept of in-situ behavior of a lamina within the laminate which is still debated.
3 The different methods of modeling the progressive failure process and the definition of
ultimate laminate failure.
12 I. M. Daniel

4 The nonlinear behavior of matrix-dominated laminates, e.g., angle-ply laminates.

Under uniaxial loading of a laminate the deciding factor in predicting ultimate failure is
whether it is fiber or matrix dominated. In the case of matrix dominated angle-ply laminates,
predictions by the limit or interactive theories are not usually in agreement with each other and
with experimental results. Failure is governed by the lamina transverse normal stress V 2 and the
in-plane shear stress W 6 . When V 2 ! 0 , as in the case of [r45]s and the limit theories predict
higher strengths in agreement with the experiment [1]. When V 2  0 , as in the case of the
[r20]2 s laminate under tension, the Tsai-Wu criterion comes closer to the experimental results
[1]. In the more general cases of biaxial loading it is not easy to establish fiber or matrix
dominance in failure as that varies with the loading biaxiality.
In view of the multitude of failure theories, the divergence of their predictions and the lack of
definitive general conclusions regarding their applicability, a practical approach is recommended
as follows [1]:
1 Select a classical representative theory from each category, i.e., non-interactive (maximum
stress), fully interactive (Tsai-Wu), and partly interactive (Hashin-Rotem).
2 Compute and plot stress-strain relations of the laminate under representative mechanical
and hygrothermal loading.
3 Use a newly proposed failure mode discrimination rule and define ULF
4 Compute safety factors for FPF and ULF and compute and plot failure envelopes for the
selected failure theories for the two failure levels (FPF and ULF).
5 Select prediction according to degree of conservatism desired. For the most conservative
approach, limit the state of stress (loading) to within the common domain of the selected
failure envelopes.
All computations and plots can be performed by a newly developed computer program [4]. The
approach above is adequate for conservative structural design. More sophisticated theories and
approaches exist as discussed before incorporating nonlinear behavior and in-situ effects.

References
1. I.M. Daniel and O. Ishai, Engineering Mechanics of Composite Materials, Second Edition,
Oxford University Press, 2005.
2. C.T. Sun, “Strength Analysis of Unidirectional Composites and Laminates,” in
Comprehensive Composite Materials, ed. by A. Kelly and C. Zweben, Ch. 1.20, Elsevier
Science, Ltd., Oxford, UK, 2000.
3. M.J. Hinton, P.D. Soden, and A.S. Kaddour, Failure Criteria in Fibre-Reinforced-Polymer
Composites, Elsevier, Oxford, 2004.
4. J.J. Luo and I.M. Daniel, “Webcomp: Stress and Failure Analysis of Laminate Composites,”
http:www.composites.northwestern.edu/awebcomp, 2004.
 Invited Papers 13

INTERACTIONS OF CONSTRAINED FLOW AND SIZE SCALE ON

MECHANICAL BEHAVIOR

W. W. Gerberich, W. M. Mook, M. J. Cordill and D. Hallman

Chemical Engineering and Materials Science Department
University of Minnesota
421 Washington Ave SE
Minneapolis, MN 55455
wgerb@umn.edu

Coupled effects between constrained flow, increased strength as a function of decreased sample
size, and resulting high stresses affect both modulus and fracture toughness. For submicron size
crystalline spheres [1,2], boxes [3], and cubes [4], we have recently shown that dislocation by
dislocation events can be followed using a combination of AFM/nanoindentation. This has led to
at least three proposed strengthening mechanisms for hardening of small constrained volumes
under compression[4]. With the increased stresses, this can produce increased moduli of elasticity
in confined volumes small in three dimensions. With increased constrained plasticity this
produces increased strength in volumes small in three, two or one dimensions.
Hardening mechanisms for the gigapascal strengths observed in small volumes are addressed.
These are explored for Si and Ti nanospheres as well as Ni, Co, and Ni80Fe20 (permalloy) thin
films in the 10-40 nm particle radius or thin film thickness[4,5] regime. Because of trapped
dislocations between the upper and lower loading surfaces being typically diamond or sapphire,
strengths approaching theoretical are often reached with deformation of only tens of nanometers or
less. Of course, this is highly dependent upon the length scale of the confining structure. For
oxide-covered nanospheres, the length scale can be taken as the volume to contact surface area as
given elsewhere[5] by (2/3)r3/a2 where r is the sphere radius and a is the contact radius at the
upper and lower platen. Alternatively, from contact geometry this length scale is (2/3)r2/G. For
deposited thin films, it was previously shown that a measured volume of plastic deformation to
contact surface area was empirically given by Dh2/a where h is the film thickness and a is the
contact radius[6]. These two measures of length scale show that the flow strength of the
nanospheres and the hardness of the thin films scale with G1/2/r and G1/4/h, respectively. Here, G is
the displacement the sphere has been squeezed or the penetration depth of the indenter into the
film. It is significant that both of these show strength to be inversely proportional to the smallest
dimension of the volume. Additionally, it is interesting that the more constrained nanosphere has
strength increasing as G1/2 compared to film hardness, small in only one dimension, increasing as
G1/4.
With this same length scale approach we have shown that the fracture toughness of a
delaminating thin film conforms to R-curve behavior with GR representing the resistance
equivalent to the strain energy release rate at fracture. This has been given by[6]

1/ 2
f
DV ys2 h § ' b ·
G ~
R
¨¨ ¸¸
E © bo ¹ (1)
where Vys/E is the strength to modulus ratio, 'b/bo is the incremental growth to initial crack size
and D is a constant on the order of 10. In the present paper we use the same approach by equating
14 I. M. Daniel

the local deformation length scale to the fracture process zone length scale. This gives the fracture
resistance for the sphere to be

6V ys2 r § t ox · 1 / 2
G RS ~ ¨ ¸ (2)
E ©G ¹
Here tox is the oxide film thickness on the sphere, and G is the vertical displacement the sphere has
been squeezed to the point of fracture. Note that the triggering event is assumed to be a crack
nucleated in the less robust oxide film around the sphere such that the defect size is tox. In both
cases for Eqs. (1) and (2), the leading term is the strain energy density times the smallest
dimension of the constrained volume.
We also discuss relationships such as Eq. (2) in terms of the overall size dependence if Vys
obeys a Hall-Petch type relation giving strength proportional to G1/2/r for the nanosphere. It is
emphasized that this is an evolutionary length scale which decreases with increasing displacement,
i.e., (2/3)r2/G. Incorporating the length scale dependence of yield strength actually gives Gg’
proportional to G1/2/r so that both strength and fracture resistance scale with G1/2/r for very small
volumes. Due to the first term in the Hall-Petch relation dominating at very large volumes for both
thicker films and larger spheres, this trend would be predicted to reverse. That is, at larger
volumes one should find an increase in fracture resistance with scale according to Eqs. (1) and (2)
as Vys becomes more nearly constant.
Most of these relationships are still in their formative stages but have some corroboration with
respect to Cu and Au film delamination studies. Regarding fracture experiments on nanospheres,
we have only recently fractured a silicon particle in situ in the transmission electron microscope.
However, this corroborates some previous indirect analysis using atomic force microscopy based
nanoindentation to determine the fracture toughness of silicon nanospheres with radii in the range
of 20 to 110 nm. In the full paper we will derive the above relationships and discuss increased
strength and fracture resistance of constrained volumes under compression or pressure as having
ramifications to friction, wear, and microelectromechanical systems.

REFERENCES
1. Gerberich, W.W., Cordill, M.J., Mook, W.M., Moody, N.R., Perrey, C.R., Carter, C.B.,
Mukherjee, R., and Girshick, S.L., Acta Mater., vol. 53, 2215-2229, 2005.
2. Gerberich, W.W., Jungk, J.M., Cordill, M.J., Mook, W.M., Boyce, B., Friedmann, T., Moody,
N.R. and Yang, D., Intern. J. Fracture, 2005 (accepted).
3. Mook, W.M., Jungk, J.M., Cordill, M.J., Moody, N.R., Sun, Y., Xia, Y., and Gerberich,
W.W., Z. Metallkd., vol. 95, 416-424, 2004.
4. Cordill, M.J., Chamber, M.M., Hallman, D., Lund, M., Perrey, C.R., Carter, C.B.,
Kortshagen, U. and Gerberich, W.W., “Plasticity Responses in Ultra-Small Confined Cubes
and Films,” 2005 (in preparation).
5. Gerberich. W.W., Mook. W.M., Perrey. C.R., Carter. C.B., Baskes. M.I., Mukherjee. R.,
Gidwani. A., Heberlein. J., McMurry. P.H. and Girshick. S.L., J. Mech. Phys. Solids, vol. 51,
979-992, 2003.
6. Gerberich. W.W., Jungk, J.M., Li, M., Volinsky, A.A., Hoehn, J. W. and Yoder, K., Intern. J.
Fracture vol. 119/120, 287-405, 2003.
 Invited Papers 15

SPACE SHUTTLE COLUMBIA POST-ACCIDENT ANALYSIS AND

INVESTIGATION

S. McDanels
NASA
YA-F1 Kennedy Space Center, FL, 32899 USA
steve.mcdanels@nasa.gov

Although the loss of the Space Shuttle Columbia and its crew was tragic, the circumstances offered
a unique opportunity to examine a multitude of components which had experienced one of the
harshest environments ever encountered by engineered materials: a break up at a velocity in
excess of Mach 18 and an altitude exceeding 200,000 feet (63 KM), resulting in a debris field 645
miles/1,038 KM long and 10 miles/16 KM wide. Various analytical tools were employed to
ascertain the sequence of events leading to the disintegration of the Orbiter and to characterize the
features of the debris. The testing and analyses all indicated that a breach in a left wing reinforced
carbon/carbon composite leading edge panel was the access point for hot gasses generated during
re-entry to penetrate the structure of the vehicle and compromise the integrity of the materials and
components in that area of the Shuttle.
The analytical and elemental testing utilized such techniques as X-Ray Diffraction (XRD),
Energy Dispersive X-Ray (EDX) dot mapping, Electron Micro Probe Analysis (EMPA), and X-
Ray Photoelectron Spectroscopy (XPS) to characterize the deposition of intermetallics adjacent to
the suspected location of the plasma breach in the leading edge of the left wing, Fig.1.

FIGURE 1. Micrograph of a Left Wing Carrier Panel Slag Deposit. Cummings [1]

Fractographic and metallographic analyses of several pieces of debris, Fig. 2, were

performed to evaluate such fracture characteristics as broomstrawing and feathering of aluminum
alloys, suspected stress-assisted grain boundary oxidation (SAGBO) of Inconel components, and
intergranular high temperature fracture features observed on several A286 stainless steel spar
fittings from the left wing structure of the Orbiter. Likewise, depositional characteristics such as
composition, directionality and orientation, and the sequence and order of layering, were evaluated
to assist in re-tracing the path of the plasma flow into the wing structure.
16 S. McDanels

FIGURE 2. Interior view of aluminum drag chute canister. Parker [2]

The examination of the debris’ fracture surfaces, and of metallographically-prepared

specimens harvested from the debris, was performed via scanning electron microscope. The
resultant features and characteristics were compared to those of laboratory exemplars of similar
base materials. The observed features, along with the results from the elemental analytical testing,
helped the Space Shuttle Columbia accident investigation team reconstruct the mishap and
determine the sequence of events which ultimately led to the loss of the vehicle.
The debris from the Columbia accident now resides in the Vehicle Assembly Building (VAB)
at the Kennedy Space Center. A portion of the debris is on display, while the majority is in storage
in the VAB. A process is in place whereby universities and professional societies can request
pieces of debris for educational and research purposes, Fig.3.

FIGURE 3. Debris sample loaned out for educational purposes.

References
1. Cummings, V. J., XPS, Metallographic and SEM X-Ray elemental Dot Map Analysis of STS-
107 Debris Sample 24543-1, National Aeronautics and Space Administration, KSC-MSL-
2003-149, 2003.
2. Parker, D. S., Optical and SEM/EDS Analysis of STS-107 Debris Sample 58693-1, National
Aeronautics and Space Administration, KSC-MSL-2003-208, 2003.
 Invited Papers 17

THE ROLE OF ADHESION AND FRACTURE ON THE PERFORMANCE OF

NANOSTRUCTURED FILMS

Neville Moody, Megan J. Cordill1, Marian S. Kennedy2, David P. Adams3, David F. Bahr2 and
William W. Gerberich1
Sandia National Laboratories, Livermore, CA 94550
1University of Minnesota, Minneapolis, MN 55455
2Washington State University, Pullman, WA 99164
3Sandia National Laboratories, Albuquerque, NM 87185

nrmoody@sandia.gov

Nanostructured materials are the basis for emerging technologies, such as MEMS, NEMS, sensors,
and flexible electronics, that will dominate near term advances in nanotechnology. These
technologies are often based on devices containing layers of nanoscale polymer, ceramic and
metallic films and stretchable interconnects creating surfaces and interfaces with properties and
responses that differ dramatically from bulk counterparts. The differing properties can induce high
interlaminar stresses that lead to wrinkling, delamination, and buckling in compression [1,2], and
film fracture and decohesion in tension. [3] However, the relationships between composition,
structure and properties, and especially adhesion and fracture, are not well-defined at the
nanoscale. These relationships are critical to assuring performance and reliability of
nanostructured materials and devices. They are also critical for building materials science based
predictive models of structure and behavior.
Gold films are of special interest in applications from MEMS mirrors to nanoscale
interconnects. In these applications, the need to minimize stress effects requires deposition of very
thin films. Figure 1 shows that the fracture energies for these films decrease to work of adhesion
values as the films become very thin. [2] While these fracture energies are fully capable of forming
plastic zones, there is no evidence of deformation or ductile fracture processes on the fracture
surfaces. This strongly suggests that dislocations were not emitted from the crack tips during
interfacial fracture. This is a concept supported by elastic strip and dislocation free zone models
where the only requirement for fracture is for the energy release rate to attain the true work-of-
adhesion. It also shows that adhesion controls performance and reliability of nanoscale films
through its effect on fracture.
The composition of gold-on-oxidized silicon can change markedly with increases in
temperature from any post-deposition processing. The changes can be even more dramatic for
films such as scandium, where strong reactions during deposition markedly alter composition,
structure and functionality (Figure 2). For nanoscale films, any film-substrate interactions, or when
using adhesion promoting interlayers, any film-film interactions lead to a complete change in film
composition and properties. Understanding these interactions is therefore critical for assuring
performance and reliability as there are no plastic energy dissipative processes to mitigate the
effects of low adhesive energies. These interactions and their effect on adhesion and fracture are
the focus of current work and of this presentation.
18 S. McDanels

Figure 1. (a) Steady state, *ss, and mode I, *I, fracture energies for thin gold films on sapphire.
Gold-on-oxidized silicon values are superimposed for comparison. The shaded region corresponds
to measured gold-on-sapphire work of adhesion values.

(a) (b)
Figure 2. The composition of gold-on-oxidized silicon can change markedly with post-deposition
processing as shown for (a) as deposited and (b) 300C/1hr annealed films.

References
1. R. Huang, Z. Suo, Journal Applied Physics, 91, 1135, 2002.
2. N. Moody, D. Adams, A. Mudd, M. Cordill, D. Bahr, Plasticity Effects on Interfacial Fracture
of Thin Gold Films, Proceedings ICM9, Geneva, Switzerland 2003 (SAND2003-8146)
3. T. Ye, Z. Suo, A. G. Evans, Int. J. Solids Structures, 29, 2639, 1992.

This work is supported by Sandia National Laboratories, a multiprogram laboratory operated by

Sandia Corporation, a Lockheed Martin Company for the United States Department of Energy's
National Nuclear Security Administration under contract DE-AC04-94AL85000.
 Invited Papers 19

ASSESSMENT OF WELDMENT SPECIMENS CONTAINING RESIDUAL

STRESS

K. M. Nikbin
Mechanical Engineering Department, Imperial College,
Lonndon SW7 2BX., UK
k.n@ic.ac.uk

Weldments in components are regions where failures are most likely to occur either by fast
fracture, creep or fatigue. These regions could exhibit microstructural inhomogeneity as well as the
presence of micro-cracks and residual stresses. Understanding their behaviour is of major source of
interest for a range of industries. Creep and creep/fatigue crack growth models as well as residual
defect assessment codes need reliable and verifiable material properties data and validated fracture
mechanics parameters for use in their predictive methodologies. The research to develop an overall
methodology for deriving acceptable data and validated parameters for life assessment analysis has
been developing in Europe through a number of collaborative European projects. These have
covered both parent as well as weld material of a range of alloys and conditions. Although there is
substantial information and data on weld tests available, due to an absence of validated information
the industrial community cannot easily use this information with confidence. The present
standards, also, do not deal directly with testing of welded specimens and therefore. The Analysis
of the results from these projects have been continuing and furthermore have been used to assist in
the development of testing standards as well as life assessment codes of practice.
As a result a VAMAS (Versailles Agreement for Materials And Standards) Round Robin weld
testing programme has been initiated to address the issues of testing and analysis of weld related
materials. The paper presents the methodology that will lead to recommendations for a code of
practice (CoP) of welded materials. The overall aims are presented below;
Objectives
The following are the objectives of the new programme which adds emphasis on both the
Round Robin testing as well as the predictive modelling, and measuring of residual stresses in
components.
• Undertake a review of the information available on cracking of weld specimens and
components at high temperatures.
• Partners to link this initiative to their own projects
• Initiate a Round Robin testing programme of welded specimens within the TWA
collaboration based on the available information from the review. The round robin will
cover four different steels namely (347 weld, 316H stainless, P22, P91 and P92) steels
which have been offered for testing by partners in UK, Germany and Japan.
• The tests to be carried out will consist of mainly Compact Tension (CT) specimens of
parent, weld, and x-weld crack growth tests. The testing and analysis will be performed on
a Round Robin basis. Details will be set out at the kick-off meeting
• FE Modelling of residual stresses and identifying the role of stress relaxation during the
testing of the component at elevated temperatures.
• Measurement of residual stresses in a number of CT crack growth tests before and after
crack initiation using deep-hole drilling, neutron and X-ray diffraction of a number of tests
20 K. M. Nikbin

from the Round Robin. In addition collection of information on residual stress

measurements from partners involved in other research in this field.
• Identify the appropriate fracture mechanics parameters and materials and weld conditions
for different geometries, to cover the majority of cases for testing of weldments.
• Provide recommendations on weld testing and analysis plus the effects of stress relaxation
at high temperatures.
Pre-standardisation Needs
At present there is excessive conservatism in treating weldments and residual stresses.
Therefore it is envisaged that the results from this collaboration can be incorporated quickly into
existing or planned standards so that
• Relevant existing standards such as ASTM E1457 and the industrial codes of practice such
as R5, BS7910, ASME and API are likely to be improved as a result of this work.
• The recommendations from the work will improve predictive methods in High
Temperature Life Assessment procedures.
• The range of industries using this will be the power, aerospace, chemical and the electrical
industries
• Collaboration and sharing of information with active standardisation committees such as
ASTM, ‘ESIS TC11: HTMT Working Group On High Temperature Testing of
Weldments' and the 'Net European Network: Network On Neutron Techniques
Standardization For Structural Integrity'.
Deliverables
• Define the criteria and methodology for dealing with weldment testing and analysis.
• Provide relevant crack growth data from available tests of specimens and feature tests and
provide residual stress measurements and relaxation data from these tests.
• Using the appropriate fracture mechanics parameters analyse the crack growth data and
identify the effects of residual stresses.
• Make recommendations of welded component testing and analysis in the draft Code of
Practice for dealing with component creep crack growth testing and analysis of industrial
feature specimens derived from the initial phase of VAMAS TWA 25.
• Dissemination of results to the wider industrial audience and future implementation of
results in an ISO document.
 Invited Papers 21

MEMS: RECENT ADVANCES AND CURRENT CHALLENGES

R. J. Pryputniewicz
Worcester Polytechnic Institute, Mechanical Engineering Department / CHSLT-NEST
rjp@WPI.EDU

Recent advances in MEMS technology have led to development of a multitude of new devices.
However applications of these devices are hampered by challenges posed by their integration and
packaging (Wei et al., 2005). Current trend in micro/nanosystems is to produce ever smaller,
lighter, and more capable devices at a lower cost than ever before. In addition, the finished
products have to operate at very low power and in very adverse conditions while assuring durable
and reliable performance (Pryputniewicz et al., 2001).
Some of the new devices were developed to function at high rotational speeds, others to make
accurate measurements of operating conditions of specific processes. Regardless of their
application, the devices have to be packaged to facilitate their use. MEMS packaging, however, is
application specific and, usually, has to be developed on a case by case basis (Pryputniewicz et al.,
2006). To facilitate advances of MEMS, educational programs have been introduced addressing
all aspects in their development (Pryputniewicz et al., 2003). This presentation will address
various aspects in a development of MEMS including, but not limited to, design, analysis,
fabrication, characterization, packaging, and testing. The presentation will be illustrated with
selected examples, Figs 1 to 5.
22 R. J. Pryputniewicz

References
1. Pryputniewicz, R. J., T. F. Marinis, D. S. Hanson, and C. Furlong, 2001, “New approach to
development of MEMS packaging for inertial sensors,” Paper No. IMECE2001/MEMS-
22906, Am Soc. Mech. Eng., New York, NY.
2. Pryputniewicz, R. J., E. Shepherd, J. J. Allen, and C. Furlong, 2003, “University – National
Laboratory alliance for MEMS education,” Proc. 4th Internat. Symp. on MEMS and
Nanotechnology (4th-ISMAN), Charlotte, NC, pp. 364-371.
3. Pryputniewicz, R. J., T. F. Marinis, J. W. Soucy, P. Hefti, and A. R. Klempner, 2006, “A
metal interposer for isolating MEMS devices from package stresses,” in press, Proc. EFC-16,
Alexandoupolis, Greece.
4. Wei, J., Wong, C. K., and Lee, L. C., 2005, “Wafer-level micro/nanosystems integration and
packaging,” Proc. 6th Internat. Symp. on MEMS and Nanotechnology (6th-ISMAN),” pp. 1-12,
Portland, OR.
 Invited Papers 23

FRACTURE, AGING AND DISEASE IN BONE AND TEETH

R. O. Ritchie and R. K. Nalla

University of California, Berkeley
Department of Materials Science and Engineering, Berkeley, CA 94720, USA
roritchie@lbl.gov

Biological materials comprising mineralized tissues, such as bone and dentin in teeth, have
hierarchical structures with characteristic length scales ranging from nanometers to millimeters. In
this presentation, in vitro fracture toughness and fatigue-crack propagation properties of dentin and
human cortical bone are examined from a perspective of discerning how these properties depend
upon such microstructural hierarchies. The motivation for this is that although there is substantial
clinical interest in their fracture resistance, there is relatively little mechanistic information
available on how bone and teeth derive their resistance to cracking and how this is affected by
cyclic loads. Specifically, in vitro experiments are described that establish that the initiation of
fracture is locally strain-controlled (Nalla et al. [1]) and that subsequent crack growth
(characterized by resistance-curve behavior) is associated with a variety of extrinsic toughening
(crack-tip shielding) mechanisms, most importantly crack bridging (from individual collagen
fibrils and especially “uncracked ligaments”), macroscopic crack deflection and to a lesser extent
diffuse microcracking (Fig. 1) (Kruzic et al. [2], Nalla et al. [3]).

FIGURE 1. Schematic illustrations of some of the toughening mechanisms possible in cortical

bone: (a) crack deflection (by osteons), (b) crack bridging (by collagen fibers), (c) uncracked-
ligament bridging, and (d) constained microcracking.

Quantitative estimates for the relative contributions of these mechanisms to the overall
toughness are derived from simple micromechanical models [3]. In a manner not unlike ceramic
materials, it is shown that such extrinsic mechanisms act to toughen bone by lessening the
magnitude of stresses experienced at the tip of any cracks. Although macroscopic crack deflection
along cement lines provides a principal source of toughening in the transverse orientation, crack
bridging by intact regions in the crack wake (so-called uncracked ligaments) is the primary
toughening mechanism in longitudinal orientations; such bridges act to sustain load that would
otherwise be used to propagate the crack. In vitro fatigue experiments that seek to examine time- or
cycle-dependent crack-growth behavior, which pertain to stress fractures in bone, are also
described [4].
24 R. O. Ritchie and R. K. Nalla

Finally, we show that the role of biological aging, which causes a marked deterioration in the
fracture toughness of bone (Fig. 2), can be attributed to an age-related deterioration in the potency
of crack bridging [5], a phenomenon that we believe is associated with the role of excessive
remodeling in increasing the density of secondary osteon structures. However, the mechanistic
aspects of this age-related degradation in bone quality is additionally characterized at multiple
dimensions, including molecular (using deep UV Raman spectroscopy), sub-micron (using pico-
force atomic force microscopy) and tens of micron scale (using X-ray computed tomography). We
attempt to discriminate between possible age-related changes in the constitutive properties of the
hard tissue and age-related changes in its microstructure.

FIGURE 2. Resistance curves for stable in vitro crack extension in human cortical bone, showing
the deterioration in both the crack-initiation and crack-growth fracture toughness with age from 30
to 99 years.

References
1. Nalla, R.K., Kinney, J.H. and Ritchie, R.O., Nature Materials, vol. 2, 164-68, 2003.
2. Kruzic, J.J, Nalla, R.K, Kinney, J.H. and Ritchie, R.O., Biomaterials, vol. 24, 5209-21, 2003.
3. Nalla, R.K., Stölken, J.S., Kinney, J.H. and Ritchie, R.O., Journal of Biomechanics, vol. 38,
1517-25, 2005.
4. Nalla, R.K., Kruzic, J.J., Kinney, J.H. and Ritchie, R.O., Biomaterials, vol. 26, 2183-95,
2005.
5. Nalla, R.K., Kruzic, J.J., Kinney, J.H. and Ritchie, R.O., Bone, vol. 35, 1240-46, 2004.
 Invited Papers 25

LABORATORY EARTHQUAKES

A. J. Rosakis, K. Xia1 and H. Kanamori2

Graduate Aeronautical Laboratories
1Graduate Aeronautical Laboratories and Seismological Laboratory
2Seismological Laboratory

California Institute of Technology, Pasadena, CA

rosakis@atlantis.caltech.edu

The goal of the present study is to create model laboratory experiments mimicking the dynamic
shear rupture process. We hope to use such experiments to observe new physical phenomena and
to create benchmark comparisons with existing analysis and numerics. The experiments use high-
speed photography, photoelasticity, and infrared thermography as diagnostics. The fault systems
are simulated using two photoelastic plates (Homalite) held together by friction. The far field
tectonic loading is simulated by pre-compression and the triggering of dynamic rupture
(nucleation) is achieved by an exploding wire technique. The fault forms an acute angle with the
compression axis to provide the shear driving force necessary for continued rupturing.
Earthquake dynamics and, in particular, the mechanics of dynamic shear rupture are two
relatively under-investigated sub-fields of seismology. Most efforts to date have focused on
analytical studies (Rice 2001) and on the numerical modeling of dynamic rupture processes using
finite element, finite difference, and boundary element methods (e.g., Ben-Zion and Andrews,
1997). As clearly elucidated by Rice (2001), the nature and stability of the predicted process
depends very strongly on the choice of frictional laws employed in the modeling and, as a result,
validation of the fidelity of such calculations becomes of primary importance.

FIGURE1. A homogeneous system composed of two frictionally held Homalite plates is shown.
Purely subRayleigh (D=25°, P=7 MPa) (2A) and purely supershear (D=25°, P=15 MPa) (2B)
rupture at the same time (28 Ps) after triggering.

Our goals are to investigate the dependence the characteristics of rupturing, such as rupture
speed, rupture mode on experimental conditions such as far-field biaxial compression, tilt angle of
the fault to the compression axis, as well as on the frictional properties of the fault interface. (Fig.
1.)
Results on both homogeneous and bimaterial interfaces are reported. For bimaterial interfaces,
various combination of dissimilar materials, including Homalite/polycarbonate pairs, are chosen to
26 R. O. Ritchie and R. K. Nalla

mimic wave speed mismatch conditions that are reported to exist across mature, crustal faults (Xia,
Rosakis, Kanamori and Rice 2005).
In the present lecture we concentrate on the experimental observation of the phenomenon of,
spontaneously unrelated, supershear rupture on the visualization of the mechanics of sub Rayleigh
to supershear rupture transition in such frictionally held interfaces. The results suggest that under
certain conditions supershear rupture propagation can be facilitated during large earthquakes (e.g.
the 2001 central Kunlunshan earthquake in Tibet, (Lin, Fu, Guo, Zeng, Dang, He and Zhao 2002,
Bouchon and Vallee 2002); the 2002 Denali earthquake in Alaska, (Ellsworth, Celebi, Evans,
Jensen, Nyman and Spudich 2004).

References
1. Rice, J. R., Lapusta, N., Ranjith, K., J. of the Mech. Phys. Solids, vol. 49 (9) 1865-1898,
2001.
2. Andrews, D., Ben-Zion, Y. J. of Geophysical Research, vol. 102, 553, 1997.
3. Xia, K. W., Rosakis, A. J., Kanamori, H., Science vol. 308, 681-684, 2005.
4. Lin, A. M., Fu, B. H., Guo, J. M., Zeng, Q. L., Dang, G. M., He, W. G., and Zhao,Y., Science,
296 (5575), 2015-2017, 2002.
5. Bouchon, N and Vallee, M, Science, vol. 301, 824-826, 2002.
6. Ellsworth, W. L., M. Çelebi, J. R. Evans, E. G. Jensen, D.J. Nyman, P. Spudich, Eleventh
Int’l Conference of Soil Dyn. and Earthquake Engineering, Berkeley, CA, 2004.
 Invited Papers 27

A HISTORICAL RETROSPECTIVE OF THE BEGINNINGS OF BRITTLE

FRACTURE MECHANICS - THE PERIOD 1907-1947

H. P. Rossmanith
Institute of Mechanics and Mechatronics, Vienna University of Technology
Wiedner Hauptstr. 8-10/325, A-1040 Vienna, Austria
hans.peter.rossmanith@tuwien.ac.at

This contribution presents a historical retrospective of the early years of brittle fracture mechanics.
Looking back 99 years, the year 1907 is important for two reasons:
1 the German physicist Karl Wieghardt, a disciple of the famous physicist Arnold
Sommerfeld derived the complete stress field around the tip of a crack and gave the correct
order of the elastic crack tip singularity, and in solving the so-called Bach-Problem of
fracturing of roller bearings, presented the first mixed-mode fracture criterion, and
2 the Father of Engineering Fracture Mechanics, George Rankin Irwin, was born in El Paso,
Texas.
In the first part, this paper will highlight and detail the achievements in fracture mechanics by
Karl Wieghardt which for a long time – and for various reasons - were completely unknown or
disregarded by the fracture mechanics community at large. It will be shown that Wieghardt’s work
anticipated later developments by nearly fifty years [Wieghardt 1907; Rossmanith 1995a,b]. The
reasons for this neglect will be unveiled.
The second part of the paper will shed light on the very beginnings and struggles for
recognition of fracture mechanics at the time when the Griffith theory was extended to include
small scale plasticity by Irwin and Orowan. The role of research at the Naval Research Laboratory,
at North Carolina University at Chapel Hill, at several U.S. West Coast aircraft manufacturing
plants, and at Lehigh University in the founding of engineering fracture mechanics will be outlined
[Paris, P.C. 1995; Rossmanith 1997].
With the advent of a new generation of fracture mechanics researchers the author feels that the
history of fracture mechanics is not a dead branch of a huge tree but should be brought to the
attention of the young academics and professionals, and those who are interested [Cotterell B.
2002]. In fact, the future will only be mastered by looking at and understanding the past.

References
1. Cotterell, B. (2002) The Past, Present, and Future of Fracture Mechanics. Engineering
Fracture Mechanics 69:533-553.
2. Paris, P.C. (1995) Reflections on Progress in Fracture Mechanics Research. ASTM STP
1207:5-17.
3. Rossmanith, H.-P. (1995) An introduction to K. Wieghardt’s historical paper on splitting and
cracking of elastic bodies. Fatigue Fract Engng Mater Struct 12: 1367-69.
4. Rossmanith, H.-P. (1997) G.R. Irwin – The Father of Fracture Mechanics: A Biographical
Sketch. In: Fracture Research in Retrospect. An Anniversary Volume in Honour of
G.R.Irwin’s 90th Birthday. 3-36, A.A. Balkem, Rotterdam.
28 H. P. Rossmanith

5. Rossmanith, H.-P. (1997) The Struggle for Recognition of Engineering Fracture Mechanics.
In: Fracture Research in Retrospect. An Anniversary Volume in Honour of G.R.Irwin’s 90th
Birthday. 37-94, A.A. Balkem, Rotterdam.
6. Wieghardt, K. (1907) Ueber das Spalten und Zerreissen elastischer Koerper. Z. Mathematik
und Physik 55:60-103; Translation Rossmanith H.P. On splitting and cracking of elastic
bodies. Fatigue Fract Engng Mater Struct 1995, 12: 1371-1405.
 Invited Papers 29

DYNAMIC CRACK PROPAGATION IN PARTICLE REINFORCED

NANOCOMPOSITES AND GRADED MATERIALS

A. Shukla
Dynamic Photomechanics Lab, Department of Mechanical Engineering
University of Rhode Island, Kingston, RI 02881, USA
shuklaa@egr.uri.edu

An experimental investigation has been conducted to evaluate the mechanical properties of novel
materials fabricated using nano and micron sized particles in polymer matrix. Experiments were
also conducted to investigate the dynamic crack propagation in theses particle reinforced materials.
High-speed digital imaging was employed along with dynamic photoelasticity to obtain real time,
full-field quantification of the dynamic fracture process. Birefringent coatings were used to
conduct the photoelastic study due to the opaqueness of these materials [1].

Unsaturated polyester resin specimens embedded with small loadings of 36 nm average

diameter TiO2 particles were fabricated using a direct ultrasonification method to study the effects
of nanosized particles on nanocomposite bulk mechanical properties [2]. The ultrasonification
method employed produced nanocomposites with excellent particle dispersion as verified by TEM.
The presence of the particles had the greatest effect on fracture toughness. Both static and dynamic
fracture toughness showed marked improvements with addition of small volume fractions (up to
1%)of nano particles. Dynamic fracture experiments were conducted with various specimen
geometries to study the complete history of dynamic crack propagation from initiation to crack
branching. Results from several of these experiments were compiled in order to establish a
relationship between dynamic stress intensity factor, KI, and crack tip velocity, a , and the
behavior of the nanocomposites is compared with that of the virgin polyester matrix. This
relationship is shown in Fig. 1. The specimens used in this study and the nano- particle distribution
is also shown in the figure. Crack arrest toughness increased by 64% in the nanocomposite relative
to the virgin polyester. Also, Crack propagation velocities in nanocomposites were found to be
50% greater than those in the virgin polyester
A detailed analytical and experimental investigation has also been conducted to understand the
behavior of a rapidly moving crack-tip in functionally graded materials (FGMs). First, an
30 A. Shukla

elastodynamic solution for a crack propagating at an angle from the direction of property variation
in an FGM is developed. Subsequently, the elastic stress, strain and displacement fields around the
crack-tip are obtained. This is followed by a comprehensive series of experiments to get more
insight into the behavior of propagating cracks in FGMs. The full-field stress data around the
propagating crack was analyzed using the stress field expansions developed in the first part of this
study. Dynamic fracture experiments were also conducted to study the behavior of a crack moving
at an arbitrary angle from the direction of property variation. It was found that when crack
propagation is inclined from the direction of property variation, crack-tip experience mixed mode
loading even if the far field loading is pure opening mode. Also, dynamic fracture experiments
were performed with different specimen geometries (modified compact tension and singe edge
notch tension) to develop a dynamic constitutive fracture relationship between the mode I dynamic
stress intensity factor (KID) and crack-tip velocity ( a ) for FGMs with crack moving in increasing
fracture toughness direction. This relationship is shown in Fig. 2.

The transient nature of crack growth in FGMs has also been investigated both analytically and
experimentally. It was concluded that not including the transient higher order terms in the analysis
of highly transient crack propagation experiments might give rise to high errors.

References
1. Der, V. K. and Barker, D. B., Mech. Res. Comm., vol. 5, 313-318, 1978.
2. Evora, V. F., Jain, N. and Shukla, A., Exp. Mech., vol. 45, 153-159, 2005.
 Invited Papers 31

SPATIAL AND TEMPORAL SCALING AFFECTED BY SYSTEM

INHOMOGENEITY: ATOMIC, MICROSCOPIC AND MACROSCOPIC.

G. C. Sih
School of Mechanical Engineering,
East China University of Science and Technology, Shanghai 200237, China
Department of Mechanical Engineering and Mechanics,
Lehigh University, Bethlehem PA 18015, USA
gcs@ecust.edu.cn Fax: +86 (21) 6425-3500

The impetus of nanotechnology has shed light on the direction of new research for many fields and
continuum mechanics is no exception. The current trend is to reach down the scale such that all
disciplines would meet and benefit from one another. However, there are overwhelming
difficulties associated with the discontinuities of results from the various scales. The prevailing
gaps in materials and continuum theories are being referred to “Mesomechanics” [1].
Mesoelectronics has in fact discovered that the heat transfer behavior of small bodies is dinstinctly
different and requires fundamental studies of electronics at the subatomic scale. New physical laws
may have to be discovered to fill in the gaps. The final answer lies in multiscaling [2,3] where the
results at the smaller scales must be translated to the macroscopic level.
Modeling of multiscale material damage theories raises several basic issues. To begin with,
conditions must be invoked to connect the results observed at the different temporal and spatial
scales. Up to now, discussions seem to be confined to a very narrow range of size and time.
Furthermore, the models seldom address the effect of the initial or residual state as differentiated
from the performance sate. The former becomes increasingly more important as the size of device
is reduced to sub-microns. This is the rule rather than the exceptions in microelectronics. It is in
this nanoelectronics region that the electron transport behavior does not strictly obey quantum
mechanics nor classical physics. It has been referred to as the mesoscopic electronics region,
particularly with reference to power dissipation. Micro-chips should be kept sufficiently cool so
that they will operate in a stable manner. And yet the density of the transistors must also be high
and closely packed. The optimum balance can be achieved only by knowing the limits of how
effectively a very small device can dissipate heat. At the mesoscopic scale, the non-equilibrium
isoenergy density approach [4] can be applied. It has solved many problems with mesoscopic
phenomena [5]. The non-equilibrium theory [4] stresses in particular the power dissipation that is
derived directly by considering the mutual interaction of mechanical and thermal effects without
invoking artificial dissipation laws and/or constitutive relations. The dissipation has been shown to
depend sensitively on the temporal and spatial characters of the local deformation. Under
extension, a cooling period has been observed [5] that precedes heating for solid, liquid and gas.
This fundamental feature is not considered in classical physics. It can be very important for the
design of microchips in electronics.
The objective of this work is directed towards the development of physical models that can
relate results at different scale ranges with account for change of system homogeneity as the region
of interest is reduced in size. A corresponding increase in the time scale follows automatically. In
order to preserve the use of equilibrium mechanics in the ranges referred to as atomic, microscopic
and macroscopic, attention will be focused in the region where damage is concentrated in the form
of a singularity for the stress and energy density fields. The displacement field is required to
remain finite and continuous even though its cyclic value may become multi-valued. Cross scale
transition is made possible by imposing scale invariant criterion based on the “force” and/or
“energy” quantities.
32 G. C. Sih

The singularity representation approach [6] will be applied to illustrate how disorders in the
system at the microscopic and atomic scales can interact. The former and latter will be associated,
respectively, with micro-cracking and dislocations. Non-linear equation are solved for the coupling
of the micro-energy and dislocation-energy designated by Wmicro and W disln in normalized form,
respectively. They will be used to derive the length of inhomogeneity for the system.The
formulation entails several orders of magnitude extending from 10-11 to 10-1 cm on the lineal scale.

Examples will be presented for short time and long time effects where the Lower scale
chemical effects can have significant higher scale mechanical effects. They correspond to the
problems of solid rocket propellant explosionsand the stress corrosion of low alloy metals in high
temperature environment such as the nuclear reactors.

References
1. Prospects of Mesomechanics in the 21st Century: Current Thinking on Treatment of
Multiscale Mechanics Problems, in: G. C. Sih and V. E. Panin, J. of Theoretical and Applied
Fracture Mechanics, 37(1-3)(2001) 1-410.
2. G. C. Sih and X. S. Tang, Dual scaling damage model associated with weak singularity for
macroscopic crack possessing a micro/mesoscopic notch tip, J. of Theoretical and Applied
Fracture Mechanics, 42(1) (2004) 1-24.
3. G. C. Sih and X. S. Tang, Simultaneity of multiscaling for macro-meso-micro damage model
represented by strong singularities, J. of Theoretical and Applied Fracture Mechanics, 42(3)
(2004) 199-225.
4. G. C. Sih, Thermomechanics of solids: nonequilibrium and irreversibility, J. of Theoretical
and Applied Fracture Mechanics, 9(3) (1988) 175-198.
5. G. C. Sih, Some basic problems in nonequilibrium thermomechanics, in: S. Sienietyez and P.
Salamon, (eds.), Flow, Diffusion and Rate Processes, Taylor and Francis, New York, (1992)
218-247.
6. G. C. Sih and X. S. Tang, Singularity representation of multiscale damage due to
inhomogeneity with mesomechanics consideration, G. C. Sih, T. Kermanidis and Sp.
Pantelakis, eds., Sarantidis Publications, Patras, Greece (2004) 1-15.
B. TRACKS

B1. Nanomaterials and Nanostructures

 1T1. Fracture and fatigue of nanostuctured materials 35

CHANNELING EFFECT IN FRACTURE OF MATERIALS WITH

NANOSTRUCTURED SURFACE LAYERS

V. E. Panin and A. V. Panin

Institute of Strength Physics and Materials Science, SB RAS
2/1, Academichesky pr., Tomsk, 634021, Russia
pav@ms.tsc.ru

1 The “chessboard” distribution of normal and tangential tensile and compressive stresses on
the “nanostructured surface layer – substrate” interface is revealed for metal materials (Fig.
1).

FIGURE 1. The “chessboard” model of the conjugation between nanostructured surface layer and
substrate (“+” – compressive stress, “-” – tensile stress) and channeling propagation of localized
shear.

2 A stochastic two-level model of the chessboard stress distribution on the interface is

worked out. The simulation results correlate well with appropriate experimental data.
3 In tension of metal materials with nanostructured surface layers one can observe the effects
of strain channeling in the nanostructured surface layer and fracture of the specimen as a
whole. This is related to shear localization and subsequent propagation of the main crack
along “chessboard squares” with normal tensile stresses (Fig. 2).
4 On the stage of uniform specimen elongation in the nanostructured surface layer localized
deformation bands evolve in the conjugate directions of maximum tangential stresses
W max. On the prefracture stage a localized deformation macroband in the form of an
extended neck propagates in the conjugate W max directions zigzagging along the whole
specimen length. On the stage of localized neck formation one can see the development of
two macrobands self-consistent by the scheme of a dipole or a cross (Fig. 3).
5 We have measured experimentally the evolution of main plastic shear in localized
deformation macrobands in the neck and rotational deformation modes associated with
localized shears. With non-compensated rotational deformation modes, the main crack is
generated in two localized deformation macrobands in the neck and then the specimen
fails.
6 Methods of governing the channeling effects of plastic flow and fracture of the material
with nanostructured surface layers are proposed.
36 V. E. Panin and A. V. Panin

FIGURE 2. Asymmetrical (a-d) and symmetrical (e-h) necking and fracture pattern of the tensile
cold-rolled titanium specimen with the nanostructured surface layer: optical image (a,e);
displacement-vector field (b,f); distribution of main plastic shear (c,g); fracture pattern (d,h); H =
17 %. u15.

FIGURE 3. Self-organization pattern of shears in interactions of macroscale localized-shear bands

of the dipole (a) and cross configurations (b).
 1T1. Fracture and fatigue of nanostuctured materials 37

ATOMISTICS AND CONFIGURATIONAL FORCES IN GRADIENT

ELASTICITY

P. Steinmann and E. C. Aifantis

Chair of Applied Mechanics, University of Kaiserslautern, Dept. of Mechanical and
Process Engineering, P.O. Box 3049, D-67653 Kaiserslautern, Germany
Laboratory of Mechanics and Materials, Polytechnic School,
Aristotle University of Thessaloniki, 54124, Thessaloniki, Greece
mom@mom.gen.auth.gr

The term Gradient Elasticity was introduced in the early 1990’s by Aifantis and co-workers (e.g.
Aifantis [1]) to denote a particular form of higher-order elasticity involving one extra
phenomenological coefficient in addition to the usual elastic Lamé constants. It distinguished itself
from an excessively large number of generalized elasticity and Cosserat type theories that were
advanced in the 1960’s and 1970’s in the sense that the stress remains symmetric and that the extra
gradient coefficient or internal length parameter involved may be directly related to the underlying
microstructure and determined experimentally. The theory allows for the formulation of
conveniently solvable boundary value problems by their reduction to inhomogeneous Helmholtz
equation with the source term being the solution of corresponding boundary value problems of
classical elasticity (e.g. Ru-Aifantis theorem [2]). It was shown (see also [3-6]) that the special
form of gradient elasticity can lead, among other things, to the elimination of singularities from
dislocation lines and crack tips. Motivated by this success, a large number of publications on wave
propagation, cracks, dislocations and other inhomogeneities (e.g. [7-10] and references quoted
therein) have been published recently based, more or less, on the special gradient elastic model
introduced in [1].
The aforementioned special linearized gradient elasticity model of [1] was motivated by a
simple gradient generalization of hyperelasticity theory elaborated upon by Triantafyllidis and
Aifantis [11] in relation to the “loss of ellipticity” in the governing differential equations of
nonlinear elasticity and the associated problem of localization of deformation. The gradient
hyperelastic model discussed in [11] and its linearized counterpart discussed in [1] are revisited
here in view of recent results obtained by Steinmann and co-workers (e.g. [12-14]) in relation to
atomistic modeling and configurational forces.
In relation to atomistic modeling, the Cauchy-Born rule is first extended to incorporate the
second-order deformation gradient. In addition to the Lennard-Jones and Morse interatomic
potentials a new interaction potential based on a Gaussian type nonlocal kernel of the form
(1 ʌ ʌ m 3 ) e x p [ (  r m b ) 2 ] , where mb is an intrinsic length scale (b denotes as usual the
Burger’s vector), will be used. Such nonlocal kernels can be used for the calibration of the gradient
coefficient and for obtaining corresponding new expressions for the Peierls stress. The coupling
between the molecular dynamics and the finite element method, the so-called hybrid model, will be
illustrated by considering the growth of a crack in two dimensions at the nanoscale. The difficulties
associated with the correct treatment of the transition between the atomistic and the continuum
regions and the role of higher-order gradients to settle the fundamental incompatibility between the
nonlocal character of the atomistic description and the local character of the continuum description
will be discussed.
Next, expressions are developed for the Peach-Koehler force and the force driving a crack in a
gradient elastic solid (J-integral) for the special gradient elasticity model of [1]. The notion of null
Lagrangeans and the boundary conditions are of relevance here and both topics will be addressed.
38 P. Steinmann and E. C. Aifantis

The special gradient elasticity model is presented as an example of the use of configurational
mechanics to deal with problems of defects within a linearized theory of gradient elasticity and set
up the stage for a general nonlinear treatment.
Finally, a general framework of the spatial and material settings of geometrically nonlinear
gradient hyperelasticity, allowing in particular for the consideration of material defects is
presented. Continua which are described within this constitutive class are characterized for the
spatial setting by a dependence of the stored energy density per unit volume undeformed
configuration on the first and second gradient of the spatial motion. Based on this description the
formulation of the corresponding material motion problem or rather of configurational mechanics
is developed. The material setting is particularly suited to compute the so-called material motion
problem, including the movement of material defects (dislocations, cracks, inclusions, phase
boundaries, etc.) relative to the ambient material. We highlight the duality of the spatial and the
material setting, in particular, the existence of a stored energy density per unit volume deformed
configuration depending on the first and second gradient of the material motion (i.e. the inverse
motion, deformation map); provide transition rules between them; and compare the results
obtained to those derived for continua without higher-order gradients.

References
1. Aifantis, E.C., Int. J. Engng. Sci., vol. 30, 1279-1299, 1992.
2. Ru, C.Q. and Aifantis, E.C., Acta Mechanica, vol. 101, 59-68, 1993.
3. Altan, B.S. and Aifantis, E.C., J. Mechan. Behav. Mats., vol. 8, 231-282, 1997.
4. Aifantis, E.C., J. Eng. Mater. Techn., vol. 121, 189-202, 1999.
5. Gutkin, M.Yu. and Aifantis, E.C., Scripta Mat., vol. 40, 559-566, 1999.
6. Aifantis, E.C., Mech. Mat., vol. 35, 259-280, 2003.
7. Georgiadis, H.G. and Vardoulakis, I. and Velgaki, E.G., J. Elasticity, vol. 74, 17-45, 2004.
8. Georgiadis, H.G., J. Appl. Mech., vol. 70, 517-530, 2003.
9. Lazar, M., Maugin, G.A. and Aifantis, E.C., Phys. Stat. Sol., vol. 242, 2365-2390, 2005.
10. Zhang, X. and Sharma, P., Int. J. Sol. Struct., vol. 42, 3833-3851, 2005.
11. Triantafyllidis, N. and Aifantis, E.C., J. Elasticity, vol. 16, 225-238,1986.
12. Sunyk, R. and Steinmann, P., Int. J. Sol. Struct., vol. 40, 6877-6896, 2003.
13. Steinmann, P. and Elizondo, A., RTN DEFINO Report No. 3, 2005.
14. Kirchner, N. and Steinmann, P., Phil. Mag., in press, 2006.
 1T1. Fracture and fatigue of nanostuctured materials 39

TENSILE BEHAVIOR AND FRACTURE OF CARBON NANOTUBES

CONTAINING STONE-WALES DEFECTS

K. I. Tserpes and P. Papanikos

Laboratory of Structural Mechanics, Department of Rural & Surveying Engineering,
National Technical University of Athens
Department of Product and Systems Design Engineering, University of the Aegean
Zografou Campus, 9 Iroon Polytechniou St., 15780, Athens, Greece
Ermoupolis, Syros, 84100, Greece
kitserpes@hol.gr, ppap@syros.aegean.gr

The effectiveness of carbon nanotubes (CNTs) as reinforcements is designated by their mechanical

behavior as stand alone units. One of the most commonly present topological defects, whose effect
on the mechanical behavior of CNTs needs to be clarified, is the Stone-Wales (SW) defect. In this
paper, the effect of SW defect on the tensile behavior and fracture of armchair, zigzag and chiral
single-walled carbon nanotubes (SWCNTs) was investigated using an atomistic-based progressive
fracture model (PFM).
Following the concept of Li and Chou [1], CNTs are treated as space-frame structures by
assuming that the C-C bonds act as load-carrying members and the carbon atoms as joints of the
members. The PFM utilizes the FE model developed in Ref.[2] for analyzing the nanotube
structure and the Morse interatomic potential, as modified by Belytschko et al. [3], for simulating
the non-linear behavior of the C-C bonds. The FE models of the SWCNTs were created using the
ANSYS FE code. For the modeling of the C-C bonds, the 3D elastic ANSYS BEAM4 beam
element was used. The specific beam element, as all 3D beam elements of the ANSYS FE code, is
linear elastic and does not have the ability to model non-linear behavior. This restriction is
surpassed by the stepwise procedure of progressive fracture modeling. The nanotube is loaded by
an incremental displacement at one of each ends with the other end fixed. At each load step, the
stiffness of each beam element is redefined using its axial strain, as evaluated from the FE model,
and the force-strain relationship of the modified Morse potential. Ôï optimize the accuracy of
computational results, in each analysis, the number of load steps was chosen from convergence
tests. Before using the PFM to accomplish the present study, its ability to accurately simulate the
tensile behavior of defected SWCNTs was successfully verified through the comparison with the
molecular dynamics simulations of Belytschko et al. [3].
The SW defect is the 90o rotation of a bond, which transforms 4 hexagons to 2 pentagons and 2
heptagons. The modeling of the defect is performed during the creation of the FE mesh of the
nanotubes.
Fig.1 displays the predicted stress-strain curves of the SWCNTs. In all cases, SW defects
served as nucleation sites for fracture and their presence reduced the failure stress and failure strain
of the nanotubes. The percentage reduction of failure stress and strain depends on the chirality of
tubes. On the other hand, the nanotube stiffness was found to be unaffected. Fig.2 shows the
evolution of fracture in the (12,12) SWCNT. Fracture initiated from the longitudinal bond
connecting the two heptagons of the SW defect and propagated in the r45o direction. The fracture
process was completed when all bonds around the circumference failed.
40 K. I. Tserpes and P. Papanikos

FIGURE 1. Predicted stress-strain curves of the SWCNTs.

FIGURE 2. Evolution of fracture in the (12,12) SWCNT containing a SW defect.

References
1. Li, C. and Chou, T-W., International Journal of Solids & Structures, vol. 40, 2487-2499,
2003.
2. Tserpes, K.I. and Papanikos, P., Composites Part B: Engineering, vol. 36(5), 468-477, 2005.
3. Belytschko, T., Xiao, S.P., Schatz, G.C. and Ruoff, R.S., Physical Review B, vol. 65, 235430,
2002.
 1T1. Fracture and fatigue of nanostuctured materials 41

ATOMIC-SCALE INVESTIGATION ON FRACTURE TOUGHNESS IN NANO-

COMPOSITE SILICON CARBIDE

M. Ippolito, A. Mattoni, L. Colombo and F. Cleri1

INFM-SLACS Sardinian LAboratory for Computational materials Science and Department of
Physics, University of Cagliari, Cittadella Universitaria, I-09042 Monserrato (Ca), Italy
1
ENEA, Unita` Materiali e Nuove Tecnologie, and INFM Centro Ricerche della Casaccia, CP
2400, I-00100 Roma, Italy
mariella.ippolito@dsf.unica.it

Ceramic materials are attractive for structural applications because of their low density, chemical
inertness, high strenght, high hardness and high-temperature stability. However they have
inherently low fracture toughness, so that plastic deformation in ceramics is found to be extremely
limited. Ceramic matrix composites (CMC) have been therefore developed to overcome the
intrinsic brittleness and lack of reliability of monolithic samples. CMC's consist of a ceramic
matrix reinforced with inclusions, such as particles, whiskers or chopped fibers (fiber
thoughening).
Although the macroscopic toughening is the result of several complex microscopic process,
such as the formation of subgrain boundaries and frontal process zone enhancement, the key issue
is represented by the interaction between the tip of possible crack front and the fiber.
In this work we present an atomistic investigation of the stress states of a crack inclusion pair
in nano-composite -SiC. We show that atomistic simulations (AS) are a powerful tool to test the
reliability of possible continuum approximations and moreover provide a predictive modelling of
stress intensification phenomena in regions where the continuum models fail to offer a unique
picture. Our simulations provide a rather general and simple constitutive equation, valid over a
range of crack-inclusion elastic mismatch and for a wide range of relative distances.
We consider two kind of coherent inclusion: carbon and silicon fibers. Because of the lattice
mismatch with respect to the matrix, both the carbon and silicon inclusions induce stress which
turns out to be, respectively, tensile or compressive.
The interatomic forces are described by the Tersoff potential for Si-C systems [1] and the load
is applied by means of the constant traction method[2].
We have compared our AS results [3] with two recent (and conflicting) continuum solutions: a
first one by Li and Chen[4], based on Eshelby equivalent inclusion approach, and a second one by
Helsing[5], showing that the available continuum models are not able to properly describe the
stress intensification phenomena at arbitrary values of distance between the crack tip and the
inclusion, and for different matrix-inclusion elastic mismatch.
The atomistic results, instead, provide a simple law for the effective variation of the crack
toughness, valid in both the silicon and carbon case, i.e. for very different matrix-inclusion
mismatch, and provides a simple and robust constitute equation for stress intensification
phenomena at any crack-inclusion distance in a ceramic composite.

References
1. J. Tersoff, Phys. Rev. B, vol. 39, 5566, 1989.
2. F. Cleri, Phys. Rev. B, vol. 65, 014107, 2002.
42 M. Ippolito et al.

3. M. Ippolito, A. Mattoni, L. Colombo, F. Cleri, submitted for publication.

4. Z. Li and Q. Chen, Int. J. Fracture, vol.118, 29, 2002.
5. J. Helsing, Engng. Fracture Mech., vol.64, 245, 1999
 1T1. Fracture and fatigue of nanostuctured materials 43

MULTISCALE MODELING AND COMPUTER SIMULATION OF STRESS-

DEFORMATION RELATIONSHIPS IN NANOPARTICLE-REINFORCED
COMPOSITE MATERIALS

L. V. Bochkaryova, M. V. Kireitseu, G. R. Tomlinson1, V. Kompis2 and H. Altenbach3

National Academy of Sciences of Belarus, Lesnoe 19 – 62, Minsk 223052
1Rolls-Royce UTC and Dynamics Group, the University of Sheffield, UK
2Department of Mechanics, University of Zilina, Slovakia
3Faculty of Engineering Sciences, Martin-Luther-Universitet Halle-Wittenberg

l_silver@rambler.ru

A novel concept of nanoparticle vibration damping [1, 2] shows the effect that molecule-level
mechanism can have on the damping and that nanoparticles/fibres/tubes-reinforced materials can
provide enhanced strength and vibration damping properties. It is particularly worth noting that
carbon nanotubes can act as a simple nanoscale spring where rheological modelling can play a
significant role while understanding of mechanics of novel systems across the scale length. The
mechanisms involved in such materials need to be understood and the relevance to strength/
damping identified. Additionally adequate modelling techniques for the next generation of
vibration damping systems are technology gaps that need urgent consideration.
The computational models simulating mechanical properties range many scales starting from
nano-scale (Molec. Dynam.), investigating the atomic interaction of the nano-particles in a very
reinforcing material alone and its interaction with the atoms of matrix [3, 4], up to the description
of such composite in large structures with dimension of several meters. The mechanisms involved
in such materials need to be understood and the relevance to damping identified.
Our aim was to investigate and develop advanced computational FEM-based technique to
estimate performance and mechanical/damping properties of nanoparticle/fiber-reinforced
engineering materials and then, if successful, assemble them into viable engineering workbench.
Carbon nanotube-reinforced structures are principally used at all phases of modeling and
simulation. Thus we will form a bridge from the very basic research done by physicists and
chemists to the practical applications by engineers.
To achieve the objective we performed the development of characterization and modeling
technique of the fundamental phenomena that describe relationships between structure and
mechanical properties of the materials, formalize the set of structural mechanical approaches to
build a bridge between macro and nanoscales. Carbon-nanotube-reinforced composite material
was simulated via advanced finite element and meshfree codes, using a hollow shell representation
of the individual nanotubes. It’s noted that the recently developed meshfree techniques for shells
do not present undesirable membrane and shear locking phenomena, while these locking
phenomena are inherent in the more commonly used finite element methods. Our computational
approach is fundamentally equal simulation and modelling of materials with combined molecular
dynamics (MD) and FEM technique, where an equivalency of each other is shown by
mathematical analysis. This conclusion proves a joint application of MD and FEM principles in
2D/3D modeling of materials where stress-deformations are described in angle coordinates of sin
and cos found through application of force field methodology and structural mechanic approach [3,
4].
A continuum model for deformation of material reinforced with nanoscale hard fibers/tubes
has been developed. A model using the fast multi-pole method (FMM) is presented. The FMM
44 L. V. Bochkaryova et al.

uses the Taylor expansion of the integral equations describing the interaction of rigid inclusion
with the closest neighbors and with the flexible matrix. The FMM models reduce drastically both
computation time and storage requirements so that models, which were not possible to solve with
present computational technique, are investigated. The method enables to solve the continuum
containing up to millions of such inclusions in a computer by parallel algorithms.
In conclusion, it is worth noting that for a realistic simulation of the stability behavior of the
nanoparticle-reinforced material, the nonlinear intramolecular inter-actions between neighboring
atoms have to be taken into account. In order to reduce computational costs, it is necessary to
develop advanced homogenization technique so as to apply shell elements in the model.
Comparing to the FEM, the new technique will introduce further reduction of both computer time
and storage requirement. The results will potentially create fundamentals for investigation and
development of 3-D reinforced composite structures with high nanoscale structures volume
content, using nano-scale reinforcement architecture to reduce component mass and dimension.
Although the process of verification and validation is somewhat circular, the entry point into
this process is clearly through experiments that help determine the validity of theory and
assumptions while also helping to quantify the state variables associated with the problem. It is,
therefore, necessary that the Computational Materials approach must use experimental data to
establish the range of performance of a material and to validate predicted behaviour. Even at the
atomistic level, methods such as molecular dynamics require careful parameterization (fit) to
empirical data. Therein, it gives a challenge to Computational Materials: validation of methods
across the complete range of length and time scales. To achieve this validation requires advances
in measurement sciences as well as advances in theory and models, coupled with integrated,
interdisciplinary research. It is imperative that research laboratories maintain a focused effort to
develop new programs that provide for the simultaneous growth of all the critical elements that are
required for validation of multi-scale methods.
Research work of Dr. Kireitseu has been supported by the Royal Society in the UK and
WELCH scholarship administered through the Amer. Vac. Soc. / Int.-l Union for Vac. Sc., Tech.
and Appl. in the US and Europe. Dr. Bochkareva is continuing her research work under EU INTAS
2005-2007 postdoctoral fellowship Ref. Nr 04-83-3067. It should be noted however that the views
expressed here are those of the authors and not necessarily those of any institutions.

References
1. Rivera J.L., McCabe C., and Cummings P.T. Nanoletters, Vol. 3, No. 8, 2003.
2. Li J., Ye Q., A. Cassell, H. T. Ng, R. Stevens, J. Han, M. Meyyappan, Appl. Phys. Lett., Vol.
82 (No.15), 2003.
3. Li C. and Chou T.W. Physical Review B Vol. 68, 2003.
4. Kireitseu M., Hui D., Bochkareva L., Eremeev S. and Nedavniy I. In Proceedings of the 106th
Annual American Ceramic Society Meeting & Exp., Symposium 17 - Innovative Processing
and Synthesis of Ceramics, Glasses and Composites. - April 18-21, 2004, edited by G. Geiger.
- Indianapolis, IN, USA, 250-264.
 1T1. Fracture and fatigue of nanostuctured materials 45

THE MECHANICAL PARAMETERS OF NANOOBJECTS (THEORY AND

EXPERIMENT)

E. Ivanova, N. Morozov1 and B. Semenov1

Dep. of Teoretical Mechanics, St.-Petersburg State Polytechnical University,
Politechnicheskaya str., 29, St.-Petersburg, Russia, 195251
1
Dep. of Elasticity theory, St.-Petersburg State University, Bibliotechnaya sq., 2, Staryi Petergof,
St.-Petersburg, Russia, 198904
morozov@NM1016.spb.edu

Advances in high technologies, using nanometer-size structures, requires calculation of

mechanical properties for the objects of the nanosize scale level. Majority of the theoretical
mechanical models for nanoobjects is based on the macroscopic equations of theory of elasticity.
However, a lot of researchers have noted inconsistency between the values of the elastic moduli
obtained from micro- and macroexperiments.
The presented paper is devoted to theoretical and experimental investigation of the influence of
the scale effects on the bending stiffness of a nanocrystal, which is extended in one direction and
has a limited number of atomic layers in another direction. It is theoretically shown that for small
number of atomic layers the bending stiffness of the nano-crystal substantially depends on the
number of layers and tends to its elasticity-theory value for large number of layers [1].

FIGURE 1.

The problem of the experimental determination of elastic moduli of nanoscale objects is of

present interest. The determination of the elastic moduli of thin macroscopic shells is usually based
on experiments with plates. It is known that, when grown using certain techniques, nanoobjects are
obtained only in the form of shells. Therefore, it is necessary to develop a method for determining
the elastic moduli of nanoobjects on the basis of experiments with shells. Experimental
determination of the bending stiffness of nanosize shells presents a serious problem, because for
such widespread nanoobjects as nanotubes and fullerenes under arbitrary deformation, the material
is subjected to both bending and tension. Therefore, all parameters (e.g., natural frequencies) that
can be measured directly are complicated functions of both bending and tension stiffness. In recent
years, together with nanotubes and fullerenes, nanoobjects of a more intricate configuration have
been obtained [2–4]. Nanosize cylindrical helices [2] (see Fig. 1) are of particular interest in
connection with the possible experimental determination of bending stiffness.
This is due to the fact that in helical shells under arbitrary deformation, the material is mainly
bent, so that the material tension effect can be neglected when interpreting experimental data; and
the natural oscillation shapes of helical shells are much more easily observed than those of
46 E. Ivanova et al.

cylindrical shells associated with pure bending of the material. The latter statement is illustrated in
Fig. 2, which presents the first four helical shell oscillation shapes. The analysis of helical shell
dynamics may be a theoretical foundation for experimental testing of the applicability of the
continuum theory to (a) the calculation of mechanical characteristics of nanoobjects and (b) the
experimental determination of the bending stiffness of nanoshells [5].

FIGURE 2.

Acoustical and optical methods of measuring of the eigenfrequencies of micro-objects are

based on using homodyne laser vibrometers and adaptive photodetectors. Technique for vibration
analysis of the laser vibrometer using adaptive photodetectors is based on the effect of non-steady-
state photoelectromotive force. The technique enables efficient direct conversion of high-
frequency phase modulation of speckle-like optical waves reflected from the vibrating object into
output electrical signal with concomitant setting of the optimal operation point of the
interferometer and suppression of amplitude laser noise. Methods of measuring of the
eigenfrequencies of nano-objects are based on using atomic force microscope. Technique for
vibration analysis is based on the effect of mechanical interaction of cantilever needle with nano-
object.
References
1. Ivanova, E.A., Krivtsov, A.M., Morozov, N.F., Dokl. Phys. 47, 620. 2002.
2. Golod, S.V., Prinz, V.Ya., Mashanov, V.I., Gutakovsky, A.K., Semicond. Sci. Technol. 16,
181. 2001.
3. Vorob’ev, A.B., Prinz, V.Ya., Semicond. Sci. Technol. 17, 614. 2002.
4. Prinz, V. Ya., Microelectron. Eng. 69 (2/4), 466. 2003.
5. Ivanova, E.A., Morozov, N.F., Dokl. Phys. 50, 83. 2005.
 1T1. Fracture and fatigue of nanostuctured materials 47

ADVANCED MANUFACTURING DESIGN CONCEPTS AND MODELLING

TOOLS OF THE NEXT GENERATION NANOPARTICLE-REINFORCED
DAMPING MATERIALS

M. V. Kireitseu, G. R. Tomlinson1, R. A. Williams2 and V. Kompis3

National Academy of Sciences of Belarus, Lesnoe 19 – 62, Minsk 223052
1Rolls-Royce UTC and Dynamics Group, the University of Sheffield, UK
2Institute for Particle Science & Engineering, the University of Leeds, UK
3Department of Mechanics, University of Zilina, Slovakia
indmash@yandex.ru

Vibrations and noise exist in almost every aspect of our life and are usually undesirable in
engineering structures [1]. Vibrations are of concern in large structures such as aircraft either civil
(airbus A 380) or military, as well as small structures such as electronics [2]. It is now accepted
that nanotechnology can help solve vibration damping and high noise issues through the utilisation
of nanomaterials (or media) that dissipate a substantial fraction of the vibration energy that they
receive. The mechanisms involved in such materials need to be understood and the relevance to
damping identified via both computational and experimental benchmarks.
The main issue is the need to match applications to technologies/materials being developed [3-
5]. Wide-ranging application of damping materials in real-life products is one of the best ways to
ensure future development. Commercial utilisation of a damping technology depends from both
technical performance and business environment for that. There must be a business support for
extensive technology implementation during its life-cycle cost. This has often been seen to be a
limiting factor in the utilisation of novel material/technology. In this situation, a broader
understanding of the material and its potential application is of great benefit since a “secondary”
feature can make the damping system more attractive than its predecessor.
Major technical barriers also prevent greater use of damping nanoparticle-reinforced materials.
These include sensitivity to temperature (particularly viscoelastic materials based on polymer
matrix), ability to manufacture (nanoparticle is expensive to manufacture and disperse in a matrix),
increased complexity of both nanoparticle damper and a whole damping system (affecting weight
and size) as well as effort required to predict/model performance and to optimize a technology for
a specific application. The current drive for increased efficiency and reduced cost of machinery,
usually results in a requirement for mass to be minimised. To avoid vibration related problems
therefore, considerable effort has to be expended to increase the stiffness or damping of the
structure. In terms of development of materials to achieve these goals, damping has probably
received the least attention.
The focus in this paper is directed toward to the investigation into carbon-based nanoparticle/
fibre/tube-reinforced materials and coating systems and their dynamic/damping characterization.
Computational work is concentrated on hierarchical multiscale modelling of damping behaviour as
a function of frequency, amplitude and temperature. A computational model is formulated in terms
of meso- and nanoscopic ideas of damping behavior and could provide an approach to predict
vibration damping properties and optimize some manufacturing design concepts of those material
systems so as to enable the efficient synthesis of these novel damping solutions.
The novel concept of nanoparticle-based damping technology shows that a molecule-level
mechanism can considerably enhance vibration damping and dynamic of aerospace components
48 M. V. Kireitseu et al.

(fan blades) via enhanced energy dissipation because of large surface-to-volume aspects in
nanoparticle-reinforced composite material, large damping energy sources for friction and slip-
stick motion at interfaces of matrix and nanoparticle. Thus carbon nanotube can act as a simple
nanoscale damping spring in aerospace materials and is suggested for aerospace damping materials
of the next generation. The materials offer the potential to further reduce the mass and dimension,
increase performance, and reduce vibrations.
As a result the nanoparticle/tube/fibre-reinforced composite material gains advanced damping
properties compared with conventional materials reinforced through available technologies as well
as other types of commercialized damping solutions. The damping properties of the material can
be further enhanced by designing unique aerospace components based on personal manufacturer
needs of either civil or military aircrafts. Personalized set of nanoparticles can be introduced into
material matrix, for example, through the CVD-based technology combined with conventional
particle technologies (thermal spraying, PVD, etc.).
The principal conclusions are that by invoking the properties of nano-auxetics/nanostructures
it is possible to control the wave/sound/vibration propagation in the material and enhance the
energy dissipation that can assist in improving the inherent damping of materials, but an
experimental/theoretical environment is required to apply it. Nanoparticles/tubes can be used as a
reinforcement of a matrix to provide multi-functionality, and thus we need to create an
environment (knowledge) to introduce nanomaterials widely in industry. Developed computational
tools and workbench is important part of next generation aerospace design.
Topics that could yield particular success for damping in the future include 1) Micro and nano-
scaled damping materials where nanoparticle/tube/fibre reinforcement concept might give
exceptional, temperature independent damping and negligible added weight; 2) Optimisation
methods where structural mechanics/dynamics approaches could be combined with advanced
numerical FEM modeling and statistical variation to estimate dispersion of nanoparticles in a
matrix; and 3) Low-cost damping systems that is cheaper than current polymer-based coatings.
The outcome of the research work is expected to have wide-ranging technical benefits with
direct relevance to industry in areas of transportation (aerospace, automotive, rail) and civil
infrastructure development, but the goal is aerospace turbine applications.

References
1. Hollkamp J.J. and R. W. Gordon, Smart Mater. & Struct. Vol. 5(5), 1996.
2. Chung T.R. Journal of Materials Science. vol. 36, 2001.
3. Rivera J.L., McCabe C., and Cummings P.T. Nanoletters, Vol. 3, No. 8, 2003.
4. Li J., Ye Q., A. Cassell, H. T. Ng, R. Stevens, J. Han, M. Meyyappan, Appl. Phys. Lett., Vol.
82 (No.15), 2003.
5. Siegel R.W., Hu E., and Roco M.C. in Proceedings of WTEC Workshop, May 8-9, 1997
Workshop, Washington, DC, 1998.
 1T1. Fracture and fatigue of nanostuctured materials 49

FRACTURE OF NANOSTRUCTURED IONOMER MEMBRANES

Yue Zou, X. Huang and K. L. Reifsnider

CT Global Fuel Cell Center
University of Connecticut, Storrs, CT 06269
Xinyu@engr.uconn.edu

Thin ionomer membranes showing nanostructure phase separation are widely used in low
temperature fuel cells as electrolytes. Examples include perfluorinated sulfonic acid (PFSA)
ionomers, better known as NafionTM produced by E. I. du Pont de Nemours and Company. Their
electrochemical performance and mechanical strength properties are determined by their molecular
morphology, which resembles water-filled hydrophilic micellae (4~5nm in size) dispersed in a
hydrophobic matrix. Temperature and humidity are two major parameters in an internal fuel cell
working environment, which have significant influence on this nanostructure and consequently
influence the material properties. The transport properties of such ionomers have been extensively
studied. In this paper, we will discuss fracture and strength of such membranes, as related to
humidity and temperature.
To understand fracture of the membrane, it is necessary to measure its strength and determine
the failure mode(s) under all relevant conditions. Measuring strength at different humidity and
temperatures for very thin membranes is difficult. A new method using optical strain measurement
(even for submerged specimens) has recently been used in the authors’ laboratory. Some example
data and fracture patterns are shown in figure 1.

FIGURE 1. Left, stress-strain response of Nafion 111 membrane under different status of
hydration and temperatures; Right, fracture patterns for fully hydrated specimens in 25 ഒ, 65 ഒ
and 80 ഒ (in (a), (b) and (c), respectively).

The material tested was NafionTM NR111. The specimens were boiled in deionized (DI) H2O
for 1 hour to fully hydrate them. With an increase of temperature, there is a significant increase of
strain to break, but a decrease of stress to break, as shown in Figure 1. Although this material is
ductile, the broken surfaces show a clean severance resembling those formed in brittle fractures.
With an increase of water content, the yield behaviour becomes less distinct. It is believed these
features are due to the evolution of the nanostructure as a function of temperature and humidity.
The micro/nanostructual nature of these phenomena is very complex. We offer some hypotheses
that qualitatively explain some of these changes.
It is widely accepted that this kind of ionic material contains two immiscible phases, i.e., the
hydrophobic fluorocarbon and the hydrophilic ionic phase, and that the ionic groups tend to
50 Yue Zou et al.

aggregate to form tightly packed regions as clusters with a size on the nanoscale [1]. It has been
claimed by G. Gebel, that these clusters can vary from spherical structures to a rod-like (phase-
inverted) structure [2] with increasing water content in the membrane. For a known hygrothermal
history of the membrane, it is possible to determine the water content in a membrane by weight[3];
therefore, the structure can be then determined in principle. With a known structure, we can talk
about strength and the failure mechanism of the membrane. For low water content, the membrane
behaves like a typical tough thermoplastic material with a very distinct yielding point, which
corresponds to the initiation of molecular chain movements over each other as the applied load
reaches a critical point that balances out the resistance to chain movement due to entanglement and
ionic “cross-linking” through cations. For high water content, we believe that the increase of water
reduces the strength of ionic “cross-linking” as the separation distance of the polar groups is
increased due to the swelling of the ionic clusters. As a result, chains sliding over each other may
occur at relatively low load levels; hence we observe a “spreading out” of the yield point. When
the temperature approaches the Tg, the semicrystalline hydrophobic backbone softens; as such we
see a reduction of initial modulus and also an increase of failure strain. The large strain observed at
high temperature may be due to the drawing out of the chains from the folded semi-crystalline
phase. Fracture occurs when the accumulated strain energy propels the complete breaking of the
primary bonds and the entanglement points.
Models in the literature used to explain the behavior of elastomers attempt to express stress-
strain in general and temperature effects in particular, but do not explain the mechanism of
fracture. For ionomeric membranes, nanostructure variations caused by changes in material
hydration play a major role. This paper will offer some interpretations of the data based on
understandings of this behavior and interpretations using multiphysics concepts and
representations.

Reference
1. Gierke, T. D., Munn, G.E. and Wilson, F.G., J. Polym. Sci., Polym. Phys. ED 19, 1687-1704,
1981.
2. Gebel, G., Polymer, vol. 41, 5829-5838, 2000.
3. Choi, P. and Datta, R., J. Electrochem. Soc., 150 (12) E601-E607, 2003.
 1T2. Failure mechanisms 51

DEFORMATION AND LIMIT STATES OF CARBON NANOTUBES UNDER

COMPLEX LOADING

A. V. Chentsov and R. V. Goldstein

117526 Russia, Moscow, Vernadskogo prosp. 101 k.1, RUSSIA
chentsov@ipmnet.ru

For real experiments with nanotubes the complex and precision equipment is required. Being an
alternative method of research, the computer modelling allows to perform practically any
experiments at a level of atoms. These problems are solved by the methods of molecular dynamics.
All atoms of such model are in constant thermal motion and their interaction is described by a
known potential. At the same time the combination of geometrical scales of the objects like a
nanotube (the relation of length to radius is 100-1000, diameter to wall thickness - 15-150) allows
to model the nanotube deformation by some models of continuous medium.
The regularity of atomic structure of nanotubes enables us to replace the system of atoms with
equivalent model of elastic isotropic truss (linear or nonlinear), and the truss system in macroscale
by a continuous medium, passing to continuous model of a nanotube. The choice of model of
either the continuous cylinder or a cylindrical shell depends on a solved problem.
Just as at direct modelling of interatomic interactions in molecular dynamics, within the
framework of the deformation description for a nanotube elastic truss model the energetic
approach appears to be effective. From the computative point of view the truss model gives the big
advantage of a time of calculation.
Nanotubes possess an exclusive relationship length / radius. Tubes with such hexagonal cell
will be exposed to small deformations of atomic bonds even at significant axial deformations and
bends. Such assumption means, that as a potential of interatomic interaction it is sufficient to use
harmonic potential.
The harmonic potential corresponds to a potential strain energy of the elastic rod (spring)
connecting a pair of interacting atoms. Then for any graphene plane it is possible to build the truss
system equivalent to the atomic model at equal loading conditions. On the structures of nanometer
scale, this approach has been suggested by G.M.Odegard [1].
Covalent interactions of atoms of molecular structure can be quantitatively described by using
the methods of molecular dynamics. Forces of attraction and repulsion acting for each pair of
atoms depend on relative atomic positions and are presented by the chosen forcefield. These forces
give the contribution to full vibrational potential energy of molecular system which is equal to a
strain energy of a macroscopic body of equivalent geometry.
For nanotube modelling in most cases the only essential degrees of freedom are: tension,
change of the angle, nonbonded interactions.
Energy of torsion is small enough to be neglected. It only becomes essential in problems with
the large bending deformations. Nonbonded interactions were taken into account at interaction of
separate nanotubes and graphene planes.
To proceed from discrete atomic model to truss, it is necessary to take into account all essential
pairwise interactions in structure, and to replace with their equivalent truss.
Thus, having covered the entire hexagonal plane by two types of rods, we shall receive model
which will be strained just as discrete atoms in model of molecular dynamics.
52 A. V. Chentsov and R. V. Goldstein

For calculation of effective deformation characteristics of a nanotube model a condition is

used, that the strain energy at specifically chosen uniaxial loading depends only on one
characteristic of elasticity. Thus, to define the model property it is required to calculate only its
potential energy in the strained state and its dimensions.
By the results of numerical experiments it is possible to make a conclusion, that the Young's
modulus of a graphene plane model on chirality has a weak dependence, and is approximately
equal to experimental value for graphite.
When nanotubes are used as strengthening filler in polymers, the many atoms of nanotubes
interact with atoms of a polymeric matrix. These interactions can be covalent and Van-der-Waals
[2]. The constructed truss model enables to estimate the influence of external interactions on
elastic properties of a nanotube (nanofiber).
By the results it is possible to make a conclusion, that at non-covalent interactions with
participation of all atoms of a nanotube any significant loading transport onto a matrix does not
occur [3]. Only by introduction of covalent interactions significant improvement of elastic
properties of filler can be expected.
Modelling of the form of loss of stability was performed. The case of an axial compression was
investigated. The cases with the following boundary conditions were compared: the finite
displacements on the top side of a model, nodal compressive forces. In both cases the bottom row
of hexagons was fixed rigidly, the top flange was fixed in a plane, perpendicular to axis of a
nanotube. With tubes rather thin and long the form of loss of stability is close to classical for
compression of elastic rods. Shorter tubes showed behavior similar to a shell.
From comparison of results of calculation of loss of stability on the basis of nanotube truss
model with calculations by the theory of shells it follows, that with nanotube elongation the critical
loading decreases. However, stability rises with decrease of radius of a tube. It means, that tubes of
investigated diameters behave more likely as hollow shells, rather than as continuous rods.

References
1. Odegard G.M., Gates T.S., Nicholson L.M., NASA Langley Research Center, Technical
Memorandum NASA/TM-2002-211454, 2002.
2. Goldstein R.V., Chentsov A.V., Institute for Problems in Mechanics, Russian Academy of
Sciences. Preprint No. 739, 2003.
3. Goldstein R.V., Chentsov A.V., Mechanics of Solids, N.4, 2004.
 1T2. Failure mechanisms 53

INTERACTION OF DOMAIN WALLS WITH DEFECTS IN FERROELECTRIC

MATERIALS

D. Schrade, R. Mueller, D. Gross, T. Utschig1, V. Ya. Shur2, D. C. Lupascu1

Institute of Mechanics, TU Darmstadt, Hochschulstr. 1, D-64289 Darmstadt
1Institute of Material Science, TU Darmstadt, Petersenstr. 23, D-64287 Darmstadt
2Institute of Physics and Applied Mathematics, Ural State University, 620083 Ekatarinburg,

Russia
schrade@mechanik.tu-darmstadt.de

Experimental studies suggest that domain wall movement in ferroelectric materials is strongly
influenced by the presence of certain kinds of defects. The intention of this paper is to study this
phenomenon using concepts of continuum mechanics to describe ferroelectric material behaviour.
Closely following the concept introduced in Mueller et al. [1], a 180° domain wall is modelled as a
singular surface across which a jump in the spontaneous polarization occurs. The domain wall can
be thought of as an inhomogeneity which allows for the application of configurational or driving
forces (cf. Gross et al. [2]), which can also be applied to crack problems.
The material behaviour is characterized by linear coupled constitutive equations for the stress
and the electric displacement. With the solution of the field equations for the mechanical and
electric problem it is possible to calculate the driving force on a domain wall. Motivated by
experiments [3], a linear relation between the domain wall velocity and the driving force on the
domain wall is postulated.
Using 2D finite element simulations, the influence of different kinds of defects on the kinetics
of a domain wall in ferroelectric-ferroelastic gadolinium molybdate, Gd2(MoO4)3 (GMO), is
studied. Fig. 1 shows a sketch of the model.
Three types of defects are considered. The first one is a surface defect, reflecting an imperfect
electrode. The second one is a hole in one side of a sample where the electrodes remain intact. The
third type involves a polarization defect in one domain. For a single planar domain wall, the
following fundamental results were found:

FIGURE 1. Model of a GMO sample containing 2 domains with electrodes attached at top and
bottom providing an external field E.

• When approaching an electrode defect, the total driving force on a domain wall decreases
significantly resulting in deceleration or even stopping of the domain wall.
• If the domain wall enters the electrode defect region, it is trapped in the middle of the
defect. Considerably higher external fields are necessary to move the domain wall out the
54 D. Schrade et al.

defect region. This and the latter result are in agreement with experimental findings where
a domain wall was stopped in front of an electrode defect and trapped inside it.
• A defect in form of a hole in one side at which the electrodes remain intact has little
influence on a moving domain wall. Simulations and experiments show that a domain wall
cannot be stopped at such a defect. A slight slow-down is predicted by simulation and
observed in experiments.
• For a polarization defect only numerical results are available. It was found that a domain
wall can be slowed down and even stopped in front of such a defect. The effect on the
domain wall is comparable to that of the electrode defect, however the domain wall is not
trapped in the polarization defect.

References
1. Mueller, R., Gross, D. and Lupascu, D.C., Comp. Mat. Sci., accepted for publication and in
print, 2005
2. Gross, D., Kolling, S., Mueller, R., Schmidt, I., Europ. J. Mechanics A, vol. 22, 669-692,
2003
3. Flippen, R., J. Appl. Phys., vol. 46, 1068-1071, 1975
 1T2. Failure mechanisms 55

MICROSTRUCTURE AND INTERNAL STRESSES IN CYCLICALLY

DEFORMED AL AND CU SINGLE CRYSTALS

M. E. Kassner
Dept. of Aerospace and Mechanical Engineering, University of California
3650 McClintock OHE 430, Los Angeles, California 90089-1453
kassner@usc.edu

The concept of "long range internal stresses" (LRIS) is often utilized to explain various aspects of
the mechanical behavior of materials, including cyclic deformation and the Bauschinger effect.
These internal stresses are usually associated with the heterogeneous dislocation microstructure
(Straub et al. [1] and Tippelt et al. [2]). More specifically, it has long been suggested that long-
range internal stresses (LRIS) develop with plastic deformation in metals and alloys in association
with the development of a subgrain boundaries, cell walls, dipole bundles, etc. The evidence to
support the existence of LRIS has included Bauschinger experiments, dislocation radii
measurements, stress-dip tests and, especially, and more recently, asymmetry in x-ray diffraction
line profiles (XRD LPA) in association with the “composite model.” Convergent beam electron
diffraction, weak-beam dislocation dipole separation measurements and in-situ transmission
electron microscopy were all performed by the author on cyclically deformed Al and Cu single
crystals (Kassner et al. [3-5]). These provide insight into the mechanisms of cyclic plasticity but
especially indicate an absence of measurable LRIS.
More specifically, the dipole spacing measurements were performed by Kassner et al. [4,5],
and indicated that the dipole spacing and statistical distribution of spacings were independent of
the location in the heterogeneous substructure of cyclically deformed copper and aluminum single
crystals. The dipole spacings were the same in the channel and the veins and the maximum dipole
separations in both locations suggested that the stress to separate the dipoles was within a factor of
about one or two of the applied stress in both channels and veins. Independent work by Tippelt et
al. [6] on nickel also found dipole spacings that were independent of location. These experiments
suggest an absence of LRIS.
The extensive convergent beam electron diffraction measurements that were made on creep
deformed aluminum and copper, and cyclically deformed copper, by Kassner and co-workers,
showed that the lattice parameter was unchanged, at the equilibrium or stress-free value, within the
interior of the subgrains/dipole-bundles and along (within a one-beam diameter) the subgrain-
boundaries/dipole-bundles. If LRIS were present, a measurable variation in lattice parameter
between the interior of the subgrain, and near the subgrain boundary would be expected. It must be
considered possible that residual stresses were once present, but relaxed to an undetectable level
with sample thinning.
In-situ, reversed deformation of pre-cyclically deformed aluminum single crystals was
performed in the HVEM. These experiments allowed the imaging of dislocation motion during
“forward” and “reversed” deformation. Screws were imaged shuttling back and forth in the
channels, and their relaxation upon unloading was observed. However, no reversed motion or
bowing of dislocations was evident with unloading. If long-range internal stresses were present,
reversed motion of screws in the channels or reversed bowing of dipole bundle loops upon
unloading would be expected.
X-ray diffraction line profile analysis (XRD, LPA), specifically, the interpretation of
asymmetry in strain broadened Bragg diffraction peaks, has been extensively used to support
LRIS. A popular interpretation of asymmetry involves deconvoluting the asymmetric profile into
56 M. E. Kassner

two symmetric sub-profiles; one peak associated with elevated LRIS. This method follows from
the composite model and is used as evidence for the existence of LRIS (e.g. Straub et al. [1]).
However the author suggests that there is more than one reasonable explanation for x-ray line
asymmetry, and that LRIS is not required for asymmetry to be present. Computer simulation of x-
ray line profiles was undertaken by the author and co-workers in an attempt to determine what
arrangements may lead to asymmetry. The technique calculates atomic positions based on elastic
theory, then uses a kinematic scattering approximation in which reciprocal space intensities are
calculated using the squared Fourier transform (structure factor) of the real space atomic
arrangement. This technique was previously used successfully by Levine and Thomson [6]. In this
study, diffraction from screw dislocation dipoles was simulated, in an attempt to test the analytical
asymmetry predictions of Gaal [7]. These simulations were successful in replicating the behavior
predicted by Gaal, as the x-ray line profile from the polarized dipole ensemble was asymmetric,
and had a peak offset relative to the randomly polarized dipole ensemble. Other computer
modelling of x-ray diffraction from dislocated crystals is being performed using standard and
novel approaches, and these will be discussed.

References
1. Straub, S., Blum, W., Maier, H.J., Ungar, T., Borbely, A. and Renner, H., Acta Mater., vol.
44, 4337-4350, 1996.
2. Tippelt, B., Bretschneider, J. and Hahner, P., Phys. Stat. Sol. A, vol. 163, 11-26, 1997.
3. Kassner, M.E., Wall, M.A., and Delos-Reyes, M.A., Metall. Mater. Trans. A, vol. 28, 595-
609, 1997.
4. Kassner, M.E. and Wall, M.A., Metall. Mater. Trans. A, vol. 30, 777-779, 1999.
5. Kassner, M.E., Perez-Prado, M.-T. and Vecchio, K.S., Mater. Sci. Eng. A, vols. 319-321, 730-
734, 2001.
6. Levine, L. E. and Thomson, R., Acta Cryst. A, vol. 53, 590-602, 1997.
7. Gaal, I., In Proceedings of the 5th International Riso Symposium on Metallurgy
and Material Science, edited by N.H. Anderson et al., Roskilde, Denmark, 1984, 249-254.
 1T2. Failure mechanisms 57

DETERMINATION OF EQUILIBRIUM CONFIGURATIONS OF ATOMIC

LATTICES AT QUASISTATIC DEFORMATION

S. N. Korobeynikov
Lavrentyev Institute of Hydrodynamics, 630090, Novosibirsk, Russia
korob@hydro.nsc.ru

Currently many researches simulate both initiation and propagation of cracks in a solid at the
atomic level making use of nanomechanics equations. There exist two commonly used approaches
to mathematical simulation. The first approach uses equations of molecular dynamics for
immediate simulation of initiation and propagation of cracks; the second one is based on solution
of problems of atomic lattice quasi-static deformation under the assumption that atomic lattice
buckling is a starter for initiation and growth of a crack. The present work is aimed at developing
the second approach to simulation of fracture of atomic lattices.
First solutions of problems of atomic lattice buckling were found by Novozhilov [1],
Thompson and Shorrock [2], Kornev and Tikhomirov [3]. However, the analytical methods for
solving the problems suggested by these authors cannot be applied for solving problems of
deformation and buckling of atomic lattices in the general case.
Numerical methods for solving general 2/D and 3/D problems of quasistatic deformation of
atomic lattices have been developed in Korobeynikov [4], Dluzewski and Traczykowski [5],
Korobeynikov [6]. In the typical case, the Cauchy problem is solved in the form

K ( U ) U R , U (0) U 0,
(1)

where U is the displacement vector of atoms in a lattice, R is the vector of internal forces acting
on lattice atoms, K is the symmetric tangential stiffness matrix of a lattice, a superimposed dot
denotes derivative of the magnitude with respect to the monotonically increasing deformation
parameter t . The matrix K is determined by summing tangential stiffness matrices of all atomic
pairs in the lattice. System (1) is solved by step-by-step integration with iteration refinement of
solution for each discrete value of the parameter t .
Atomic lattice buckling takes place when singular points on the integral curve are attained, i.e.,
at the points of the matrix K degeneracy:

det K 0. (2)
Singular points can be turning points or bifurcation points as well as turning and bifurcation points
simultaneously. At the bifurcation points there exist several continuations (branches) of solutions.
In the vicinity of singular points, the arc-length method of step-by-step integration of equations (1)
is most appropriate. In this case, the solution process continues through the turning points without
any difficulties. However when bifurcation of solution of problem (1) occurs, it is desirable that all
branches of the solution be determined. To do this, we solve the auxiliary problem on
determination of eigenvectors W :

KW 0 (3)
To determine equilibrium configurations, we propose to introduce small perturbations into
potential parameters of atom pairwise interaction in conformity with the form of eigenvector W .
58 S. N. Korobeynikov

This method of continuation of solutions by the side branch is proposed as alternative to more
exact but, at the same time, more complicated method presented by Sokol and Witkowski [7].
Presented procedures of solution continuation through singular points of integral curves
require as exact definition of a matrix K as possible in order to improve a convergence as well as
for prevent a divergence of iteration processes applied for the solution refinement. The expressions
refined in comparison with expressions given in [4,6], which account for both tension/compression
of segments connecting atomic pairs and their rotations have been proposed by Korobeynikov [8].
The approaches to solution of problems of atomic lattices deformation in the presence of
singular points on integral curves are realized in the PIONER code (cf., Korobeynikov et al. [9]).
Using this code, solutions of some problems of atomic lattice buckling are obtained, which show
the efficiency of algorithms mentioned above.
The supports from Russian Foundation for Basic Research (04-01-00191) and Integration
Project of Russian Academy of Science No. 3.11.1 are gratefully acknowledged.

References
1. Novozhilov, V.V., Prikladnaja Matematika i Mekhanika, vol. 33, 797-812, 1969 [in Russian].
2. Thompson, J.M.T. and Shorrock, P.A., J. Mech. Phys. Solids, vol. 23, 21-37, 1975.
3. Kornev, V.M and Tikhomirov, Yu.V., J. Appl. Mech. Techn. Phys., vol. 34, No. 3, 439-448,
1993.
4. Korobeynikov, S.N., Application of the FEM to solving nonlinear problems of deformation
and buckling of atomic lattices, Preprint No 1-97, Inst. Hydrodynamics, Sib. Div., Russ.
Acad.of Sci., Novosibirsk, Russia, 1997 [in Russian].
5. Dluzewski, P. and Traczykowski, P., Arch. Mech., vol. 55, 393-406, 2003.
6. Korobeynikov, S.N., Int. J. of Fracture, vol. 128, 315-323, 2004.
7. Sokol, T. and Witkowski, M., In Proceedings of the Second International Conference on
Computational Structures Technology: Advances in Non-linear Finite Element Methods,
edited by M. Papadrakakis and B.H.V. Topping, Civil-Comp Press, Edinburgh, 1994, 35-45.
8. Korobeynikov, S.N., Arch. Mech. (to be published).
9. Korobeynikov, S.N., Agapov, V.P., Bondarenko, M.I. and Soldatkin, A.N., In Proceedings of
the International Conference on Numerical Methods and Applications, edited by B. Sendov et
al., Publ. House of the Bulgarian Acad. of Sci., Sofia, 1989, 228–233.
 1T2. Failure mechanisms 59

MULTISCALE MECHANICS OF CARBON NANOTUBES AND THEIR

COMPOSITES

X.-Q. Feng
Department of Engineering Mechanics, Tsinghua University,
Beijing 100084, China
fengxq@tsinghua.edu.cn

Since the discovery of carbon nanotubes (CNTs) by Iijima in 1991, interest in carbon
nanotechnology has grown very rapidly because of the unique, often enhanced, properties of
nanoscale materials. Owing to their superior mechanical and physical properties, carbon
nanotubes seem to hold a great promise as an ideal reinforcing material for advanced composites of
high-strength and low-density. However, in most of the experimental results, only modest
improvements in the strength and stiffness have been achieved by incorporating carbon nanotubes
in polymers. In this talk, the mechanical properties of carbon nanotubes and their composites will
be investigated by multiscale mechanics methods.
There are many factors that influence the overall mechanical property of CNT-reinforced
composites, e.g., the weak bonding between CNTs and matrix, the curviness and agglomeration of
CNTs. Even though the adhesion strength between the CNTs and the matrix may significantly
affect the failure behavior of composites (e.g., the ultimate tensile strength and fracture toughness),
its influence on the effective elastic modulus of composites can be negligible. Therefore, it is
thought that two most significant reasons that limit the effective elastic property of CNT
composites are the curve shape and agglomeration of CNTs due to their nanometer diameters and
large aspect ratios. In the present paper, the effects of the widely observed waviness and
agglomeration of carbon nanotubes are examined theoretically. The Mori-Tanaka effective-field
method is first employed to calculate the effective elastic moduli of composites with aligned or
randomly oriented straight CNTs. A novel micromechanics model is then developed to consider
the waviness or curviness effect of CNTs which is assumed to have a helical shape. Finally, the
influence of agglomeration of CNTs on the effective stiffness is analyzed, and analytical
expressions are derived for effective elastic stiffness of CNT-reinforced composites accounting for
the effects of waviness and agglomeration. It is established that these two mechanisms may
significantly reduce the stiffening effect of CNTs. The present study not only provides the
important relationship between the effective properties and the morphology of CNT-reinforced
composites, but also may be useful for improving and tailoring their mechanical properties.
It is of great interest to gain a deep understanding of fracture behaviors of CNTs. We use here
a hybrid atomistic/continuum mechanics method to simulation defect nucleation and facture of
CNTs under tension or torsion. Under a lower tensile strain, a CNT undergoes uniform
deformation, with the positions of atoms being determined by using the modified Cauchy-Born
rule and the Tersoff-Brenner potential. When the tensile strain reaches a critical value, defects may
nucleate as a result of the so-called Stone-Wales transformation. We use the atomistic-based
continuum mechanics theory to determine the displacements of atoms far from the defect, and an
atomistic mechanics method to calculate the positions of atoms in a local subregion around the
defect. It is found that the critical strain of defect nucleation and the subsequent fracture modes of
CNTs are sensitive to the chiral angle, and that the critical strain of a zig-zag CNT is about two
times of that of an armchair one. At low temperature, both armchair and zigzag CNTs fracture in a
brittle manner. Our numerical results on the fracture strains and tensile curves compare well with
experimental results.
60 X.-Q. Feng

The exceptional mechanical properties of CNTs make this new form of carbon an excellent
candidate for composite reinforcement. However, studies on deformation and fracture of CNTs in
composites are difficult both in experimental observation and in theoretical modeling. The above
method is also extended to analyze the fracture problem of CNTs embedded in a composite matrix.
In our hybrid atomistic/continuum method, the unit cell is divided into three zones, A, B and C,
which are dealt with in different manners according to their deformation features and the numbers
of atoms in them. The deformation of CNTs is constrained partly by the surrounding matrix.
Interaction among CNTs is considered by the Mori-Tanaka method. The calculation results of the
critical strains of defect nucleation of CNTs are given in Fig. 1. Our results show that both
armchair and zigzag CNTs in a composite are easier to fracture due to the effect of CNT-matrix
interaction.

FIGURE 1. Critical strains of defect nucleation upon the chiral angles and diameters of CNTs.

References
1. Shi, D.L., Feng, X.Q., Huang, Y., Hwang, K.C. and Gao, H., Trans. ASME, J. Eng. Mater.
Tech., vol. 126, 250-257, 2004.
2. Shi, D.L., Feng, X.Q., Jiang, H.Q., Huang, Y. and Hwang, K.C.., Int J. Fracture, (in press).
 1T4. Fatigue and Fracture of MEMS and NEMS 61

IN-SITU SCANNING ELECTRON MICROSCOPE INDENTATION OF

GALLIUM ARSENIDE

Pouvreau C., Wasmer K., Giovanola J.1, Michler J., Breguet J. M.1 and Karimi A.1
Swiss Federal Laboratories for Materials Testing and Research (EMPA)
Feuerwerkerstrasse 39, 3602 Thun, Switzerland
1
Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland
cedric.pouvreau@empa.ch

Indentation is a commonly used method to evaluate the fracture toughness of brittle materials
through crack length measurements. Many models for toughness estimations are available and rely
on the assumption that surface traces of the cracks are either of the Half Penny type, Lawn and
Evans, [1], Lawn, et al., [2], or of the radial type, Niihara, et al., [3], Niihara, [4]. However, it has
been shown that, especially at low loads, the crack morphology depends strongly on material and
thus this dependence affects calculated toughness values, Cook and Pharr, [5].

Figure 1: SEM micrographs taken during an in-situ indentation experiment showing the cracking
sequence : a) Half-load (250 mN); b) Maximum load (500 mN); c) Half unload (250 mN); d) Full
unload.

This paper reports on the results of indentation experiments on Gallium Arsenide (GaAs) with
two types of indenters (conical with 60° or 120° apex angles and cube corner) using two novel
experimental techniques, (1) in-situ indentation in the Scanning Electron Microscope (SEM) while
recording indentation load and displacement and (2) sectioning of indent zones by cleavage to
reveal the subsurface crack morphology. The first technique, described by Rabe, et al., [6] serves
to establish the cracking kinetics by correlating surface cracking observations with load-
displacement histories. The second technique is used to establish accurately crack shapes,
orientations and sizes. In this cleavage technique, a starter crack is generated by scratching the
sample along the [110] direction with an indenter. Then, five series of five indents each are made
with a Nanoindenter XP at a given load level, each series shifted by 10 µm with respect to the
preceding one, perpendicularly to the scratch direction in the indentation plane. Finally, the starter
crack is propagated through the whole set of indents with a home made cleaving device.
The indenter geometries were chosen to study the influence on crack generation of
axisymmetric and singular stress fields, respectively. Moreover, for the wedge and conical
indenters, 60° and 120° apex angles have been used to evaluate the indentation fracture
dependence on indenter geometry. One edge of the cube corner indenter was aligned with the [10]
direction.
The following observations were made combining the two novel techniques.
62 C. Pouvreau et al.

The cracking sequence of an in-situ indentation with the 60° conical indenter is presented on
Figure 1a) to d).
For conical indentations, radial cracks are initiated at loads lower than 50 mN. Manufacturing
imperfections on the conical indenter strongly influence the crack initiation site and conditions. At
500 mN, chipping out occurs at the end of the loading cycle when at loads lower than 250 mN it
generally occurs at the final stage of the unloading. Generally, this chipped region has the shape of
a Maltese cross. This pattern is commonly observed in glasses and is generally associated with the
residual stress field.
During indentations with the cube corner, important cracks emanate from the indenter corner
during the loading part of the cycle, followed by secondary cracks from the indenter faces. These
cracks have an angle of 45° with the [110] direction and could be the traces of cracks propagating
in a {110} plane. Additionally, traces of slip bands are easily visible at the surface of the indented
material as previously observed (Fujita, et al., [7]).
These results demonstrate that the combination of in-situ SEM indentation and transverse
cleaving through indents opens new perspective for crack investigation in brittle single crystals
since crack morphologies, cracking sequence and cracking conditions can all be accurately
determined and correlated, for a broad range of indentation parameters.

References
1. Lawn, B. R. and Evans, A. G., J Mater Sci, 12,2195-2199, 1977
2. Lawn, B. R., et al., J Am Ceram Soc, 63,574-581, 1980
3. Niihara, K., et al., Journal of Materials Science Letters, 1,13, 1982
4. Niihara, K., Journal of Materials Science Letters, 2,221, 1983
5. Cook, R. F. and Pharr, G. M., J Am Ceram Soc, 73,787-817, 1990
6. Rabe, R., et al., Thin Solid Films, 469-70,206-213, 2004
7. Fujita, S., et al., Philos Mag A, 65,131-147, 1992
 1T4. Fatigue and Fracture of MEMS and NEMS 63

FRACTURE OF NANOSTRUCTURED LITHIUM BATTERIES

K. E. Aifantis, J. P. Dempsey1 and S. A. Hackney2

Univ. of Cambridge & Univ. of St Petersburg
Wilberforce Road, Cambridge CB3 0AL, UK
1
Clarkson Univ., 8 Clarkson Avenue, Potsdam, NY 13699-5710, USA, jdempsey@clarkson.edu
2
Michigan Technological Univ., 1400 Townsend Dr., Houghton, MI 49931, USA,
hackney@mtu.edu
kaifantis@mom.gen.auth.gr, ka279@damtp.cam.ac.uk

Lithium-ion batteries have gained considerable attention during the past twenty years due to their
favourable properties over nickel-cadmium and nickel-metal-hydride batteries. Unlike the
aforementioned material systems Li is non-toxic and it is the third lightest element. Moreover, the
high energy density of Li batteries reduces their weight by half and their volume by 20% to 50%
compared to the same capacity of nickel-cadmium and nickel-metal-hydride batteries. However,
the use of pure Li as a negative electrode is not feasable due to safety and reliability concerns.
Therefore, extensive research is being performed to find suitable material canditates that form
alloys with Li. Unfortunately, many materials that form Li alloys with the required high volumetric
density of Li also show over 100% volume change during the discharge and charge of the battery,
leading to fracture of the electrode material and associated loss of capacity. Thus, materials designs
are being explored which can stabilize the mechnical integrity of the electrode material while still
providing large volumetric Li density.
In particular, battery developers have suggested that composite materials which contain
domains of electrochemically active and inactive materials may provide resistance to mechanical
degradation while still providing capacities between 900 and 4000 mAh g-1, Beaulieu et al. [1].
Graphitic carbon, which is the material that is presently being used for the anode by Li ion battery
manufacturers, gives a much lower Li capacity of 372 mAh g-1, Graetz et al. [2]. These active/
inactive composites typically comprise of a metal inclusion (such as Si, Sn, Al, Bi) that forms rich
Li compounds and is surrounded by an inert ceramic or glass, as shown in Fig. 1.

FIGURE 1. Idealized geometry of the electrode: Li-insertion particles (shaded) embedded in a

glass (blank) matrix. A unit cell is defined by a circle of radius b surrounding a circular particle of
radius a.

However, these composite materials designs have not been optimized for fracture resistance
and Li capacity. We consider an idealized analytical model to consider some major design
parameters for this problem. In particular, to model fracture of the anodes, the unit cell of Fig. 1 is
divided in three zones: the active site, the damage zone that develops at the active site/matrix
interface, and the undamaged matrix (see Fig. 2). It should be noted here that experimental studies
suggest that if the active sites are of nanometre size and are surrounded by a ceramic, the surface of
64 K. E. Aifantis et al.

the active site does not fracture, and hence the damage zone consists only of ceramic nanoparticles.

FIGURE 2. Damage in unit cell of anode: a and b denote the radii of the active site and matrix,
respectively, ' denotes the free expansion the active site would undergo during charging, and U-a
denotes the crack length.

The present authors have modelled the internal stresses that develop during the charge and
discharge of the battery, Aifantis and Hackney [3], and they have also predicted the stability of
these systems, Aifantis and Dempsey [4]. The purpose of this study is to determine the critical
crack length at which the electrode will fracture. Furthermore, the effect that the size of the active
site has to overall capacity of the battery will explored and finally, predictions will be made for the
values of a and b that will result in no cracking.

References
1. Beaulieu, L. Y., Eberman, K. W., Turner, R. L., Krause, L. J., Dahn, J. R., Electrochem.
Solid-State Lett., vol. 4 (9), A137, 2001
2. Graetz, J., Ahn, C.C., Yazami, R., Fultz, B., Electrochem. Solid-State Lett., 6(9), A194, 2003
3. Aifantis, K.E., Hackney, S.A., J. Mech. Behav. Matls, 14, 413, 2003
4. Aifantis, K.E., Dempsey, J.P, J. Power Sources, 143, 203-211, 2005
 1T4. Fatigue and Fracture of MEMS and NEMS 65

ANALYTICAL AND EXPERIMENTAL CHARACTERIZATION OF A

MICROMIRROR SYSTEM

E. J. Pryputniewicz, C. Furlong1, and R. J. Pryputniewicz1

Current address: Department of Homeland Security, Science and Technology, Alexandria, VA
1Worcester Polytechnic Institute, Mechanical Engineering Department / CHSLT-NEST

rjp@wpi.edu

Continued demands for miniaturization and increased functionality of microelectromechanical

systems (MEMS) require construction of complex three-dimensional structures. Recent
developments at Sandia National Laboratories allow construction of such structures from parts
fabricated in a plane of a wafer, by actuating them to extend hundreds of microns in the direction
normal to the fabrication plane. One such structure is a hinged positionable micromirror system
actuated by an electrostatic microengine, Fig. 1. It should be emphasized that this structure is
batch fabricated with no piece part assembly required and that it is actuated using only the on-chip
microengine; the torque delivered by the microengine is amplified by the transmission, which, in
turn, pushes a linear rack that positions the hinged micromirror, Fig. 2. In the configuration
shown, the micromirror can be used to reflect light beam capable of triggering, or activating,
sensors and other circuitry when it is in its elevated position. Therefore, repeatability of a reflected
beam impinging on a target, changes in performance of the micromirror over time, and distortion
of its surface during activation are just some of the characteristics that must be determined with
high accuracy and precision. Until recently, this characterization was hindered by lack of suitable
methodologies. However, building on advances in photonics, electronics, and computational
analyses, we have developed a new methodology Pryputniewicz et al. [1], using fiber-based laser
optoelectronic holography system, Fig. 3, to quantitatively characterize micromirror systems in
motion, Figs 4 and 5. In addition, in order to determine forces acting on various components of the
system, we have also developed analytical models to study kinematics and kinetics of these
components, based on vector calculus Pryputniewicz [2], Pryputniewicz [3]. According to this
model, magnitudes of forces, acting on the smallest gear (60 Pm diameter) in the system, range
from 4 nN to 27 Pm, as a function of rotational speed. The corresponding displacements of the
micromirror are 113 Pm.
In this paper, the methodology for analytical and experimental characterization of a
micromirror system is described and its use is illustrated with a representative case study. By
characterizing performance of the micromirror system, we can make specific suggestions for their
future improvements and we can verify the effect of these improvements.
The micromirror systems used in this study were fabricated at and provided by Sandia National
Laboratories. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed
Company, for the United States Department of Energy under Contract DE-AC04-94AL85000.
66 E. J. Pryputniewicz et al.
 1T4. Fatigue and Fracture of MEMS and NEMS 67

A METAL INTERPOSER FOR ISOLATING MEMS DEVICES FROM

PACKAGE STRESSES

R. J. Pryputniewicz, T. F. Marinis1, J. W. Soucy1, P. Hefti, and A. R. Klempner

Worcester Polytechnic Institute, Mechanical Engineering Department / CHSLT-NEST
1Draper Laboratory, Electronic Packaging Department, Cambridge, MA

rjp@wpi.edu

Many classes of MEMS devices, such as those with resonant structures, capacitive readouts, and
diaphragm elements, are sensitive to stresses that are exerted by their surrounding package
structure. Such stresses can arise as a result of changes in temperature, ambient pressure, or
relative humidity. We have demonstrated a dramatic reduction in scale factor bias over
temperature for a tuning fork gyroscope by mounting it on an interposer structure within a
conventional chip carrier, Fig. 1. Optimization of a MEMS sensor package for high performance
subject to various constraints cannot be accomplished by analysis alone Hanson et al. [1]. There
are too many unknown parameters, e.g., material properties, process conditions, and components/
package interface conditions, to make this feasible. Extensive performance evaluation of packaged
sensors is also prohibitively expensive and time consuming. However, recent advances in
optoelectronic laser interferometric microscope (OELIM) methodology Furlong and
Pryputniewicz [2] offer a considerable promise for effective optimization of the design of
advanced MEMS components and MEMS packages. Using OELIM, sub-micron deformations of
MEMS structures are readily measured with nanometer accuracy and very high spatial resolution
over a range of environmental and functional conditions. This greatly facilitates characterization
of dynamic and thermomechanical behavior of MEMS components, packages for MEMS, and
other complex material structures. In this paper, the OELIM methodology, which allows
noninvasive, remote, full-field-of-view measurements of deformations in near real-time, is
presented and its viability for development of MEMS is discussed. Using OELIM methodology,
sub-micron displacements of sensors can be readily observed and recorded over a range of
operating conditions, Fig. 2. In addition, detailed mapping of deformation fields due to process
conditions can also be made, Fig. 3. In this case, the OELIM results clearly show that the proof
masses are not flat, but rather exhibit curvature, with the maximum deviation from planarity of
1.05 Pm. This curvature, which affects performance of the sensor, is due to residual stresses
generated during fabrication of the device. Where applicable, the OELIM measurements are
coupled with the corresponding analytical and computational modeling results, in order to validate
and refine quantitative models of packages and complex material structures Przekwas et al. [3] and
Pryputniewicz et al. [4] as well as to develop operational relationships for MEMS structures,
which offer considerable promise for effective optimization of design of advanced sensor packages
and their fabrication processes.
68 R. J. Pryputniewicz et al.

References
1. Hanson, D. S., T. F. Marinis, C. Furlong and R. J. Pryputniewicz, “Advances in optimization
of MEMS inertial sensor packaging,” Proc. Internal Congress on Applied Mechanics for
Emerging Technologies, Portland, OR, pp. 821-825, 2001.
2. Furlong, C., and R. J. Pryputniewicz, “Absolute shape measurements using high-resolution
optoelectronic holography methods,” Opt. Eng., 39:216-223, 2000.
3. Przekwas, A. J., M. Turowski, M. Furmanczyk, A. Hieke, and R. J. Pryputniewicz
“Multiphysics design and simulation environment for microelectromechanical systems,”
Proc. Internat. Symp. on MEMS: Mechanics and Measurements, Portland, OR, pp. 84-89,
2001.
4. Pryputniewicz, R. J., P. Galambos, G. C. Brown, C. Furlong, and E. J. Pryputniewicz, “ACES
characterization of surface micromachined microfluidic devices,” Internat. J.
Microelectronics and Electronic Packaging (IJMEP), 24:30-36, 2001.
 1T4. Fatigue and Fracture of MEMS and NEMS 69

COMPUTATIONAL MODELING OF NANOPARTICLES IN BIOMICROFLUIDIC

DEVICES

R. J. Pryputniewicz, Z. Sikorski1, M. Athavale1, Z. J. Chen1 and A. J. Przekwas1

Worcester Polytechnic Institute, ME/CHSLT-NEST, Worcester, MA USA
1CFD Research Corporation, Huntsville, AL, USA

rjp@wpi.edu

Focused laser beam can be used to trap and manipulate small particles without mechanical contact.
Particles trapped by laser tweezer can be used as local force probes in biomicrofluidic
environment. Optical tweezer forces are in the pico-Newton range, which makes them suitable for
study of physics of biological objects like cells, viruses, bacteria, and research in genetics.
Theoretical models of optical forces acting on particles much smaller than wavelength of light
(Rayleigh regime) are based on dipole model of interaction of light with a dielectric particle. Total
optical force can be split into gradient force, which drags particle toward maximum of laser beam
intensity in the focal region, and scattering force, which pushes particle in the direction of Poynting
vector of laser light. Ratio of scattering force to gradient force increases with a particle radius.
Stable trap requires gradient force to be larger than scattering force. If potential well produced by
optical force field is higher than other forces acting on particle (e.g., viscous, Brownian, biological
interactions) it remains in the trap. Optical manipulation of nano-beads attached to a complex
biomolecule allows to rotate, stretch, and cut it, measure kinetic constants of binding, and study
behavior of molecular motors.
This paper presents a model of optical manipulation of nanoparticles in a scanning laser beam
Sikorski et al. [1] Laser beam is fast scanned in one direction and slow scanned in the other.
Optical field time averaged over the fast scan builds an optical trap. Slow scan in the second
direction allows dragging particles by this elongated optical trap. Formulas for optical forces are
derived for this case. CFDRC multiphysics solver ACE+ has been used for transient simulation of
particle manipulation in biomicrofluidic devices. Figure 1 shows an H-filter which is 1 mm high.
In this filter, the channel width is 20 Pm and beads of 70 nm radius, made of material with index of
refraction of 1.56, are used. The slow scan velocity, transverse to the channel, is 10-5 m/s.
Observed is particle rotation and sweeping for different shapes of particles. Device heating and
influence of temperature distribution on particle and fluid movement are presented. Figure 2
shows an enlarged snapshot of the H-filter separation region, after 1500 time steps of 50 ms each.
Although the laser beam is not shown in Fig. 2, its effects are visible: beads change from RHS
stream to the LHS stream. If the laser were not present, the beads would continue in the RHS
stream. Since this effect depends on the size and properties of the beads (i.e., particles, molecules,
etc.) it can be used to separate them. Continued work will result in additional examples.
70 R. J. Pryputniewicz et al.

References
1. Sikorski, Z., M. Athavale, Z. J. Chen, A. J. Przekwas, and R. J. Pryputniewicz, “Modeling of
optical trapping and manipulation of nanoparticles in biomicrofluidic devices,”, Proc. 4th
Internat. Symp. on MEMS and Nanotechnology (4th-ISMAN), Charlotte, NC, pp. 350-357,
2003.
 1T4. Fatigue and Fracture of MEMS and NEMS 71

CHARACTERIZATION OF A MEMS PRESSURE SENSOR BY A HYBRID

METHODOLOGY

R. J. Pryputniewicz and C. Furlong

Worcester Polytechnic Institute, Mechanical Engineering Department / CHSLT-NEST
rjp@wpi.edu

Recent advances in surface micromachining and microelectromechanical systems (MEMS)

technology have led to development of a MEMS pressure sensor, Fig. 1a. In this sensor, pressure
changes are detected by deformations of a diaphragm, Fig. 1b. This diaphragm is about 400 Pm
long, 100 Pm wide, and about 2 Pm thick, made of several different materials, each having
different properties, especially different coefficients of thermal expansion (CTE). As a result, as
the sensor is exposed to the environment, where in addition to pressure changes temperature
changes also, its measurements will be affected by thermomechanical response.
The objective of this paper is to develop a hybrid, computational and experimental,
methodology for characterization of thermomechanical behavior of MEMS pressure sensors. The
computational development is based on a finite element method (FEM). The experimental
development is based on the state-of-the-art optoelectronic laser interferometric microscope
(OELIM) method Pryputniewicz et al. [1].
Accurate and precise pressure measurements by MEMS sensors are limited by the effects that
environmental temperature variation has on their performance. As the sensor is exposed to a
changing pressure, the diaphragm deforms. Deformations of the diaphragm cause changes in
resistance of the bridge circuit, which is an integral part of the MEMS sensor. In this study the
sensors were subjected to a differential pressure of 0.2 MPa and a temperature ranging from 10qC
to 50q C. The analyses were performed for the combined pressure and thermal loads.
Representative results are shown in Figs 2 and 3. Figure 2 shows typical fringe patterns due to
changes in the differential pressure. Ideally, there should be no fringes when the sensor is at rest,
i.e., at atmospheric pressure. However, because of residual stresses due to fabrication, the
diaphragm is deformed by about 40 nm. Also, deformation nonlinearity, due to loadings by a
positive and a negative pressure differences of the same magnitude, is vividly displayed by the
corresponding fringe patterns; the negative pressure difference yields the maximum deformation of
792 nm, while the positive one yields the magnitude of 813 nm, resulting in a difference of 21 nm,
which is significant while interpreting the results. Computationally determined deformations and
stresses due to the positive pressure difference are shown in Fig. 3. Comparison of computational
and experimental results indicates good correlation and shows that the hybrid methodology is very
effective for characterization of MEMS pressure sensors.
72 R. J. Pryputniewicz and C. Furlong
 1T4. Fatigue and Fracture of MEMS and NEMS 73

NEW APPROACH TO SYNTHESIS OF LASER MICROWELDING

PROCESSES FOR PACKAGING

R. J. Pryputniewicz, W. Han and K. A. Nowakowski

Worcester Polytechnic Institute, Mechanical Engineering Department / CHSLT-NEST
rjp@wpi.edu

Microelectronics and packaging industry has ever-increasing requirements for miniaturized

features with high edge acuity and negligible thermal damage zone. Laser micromachining is a
unique way of processing materials with less thermal distortion and minimum metallurgical
damage to the workpiece, therefore it is a good alternative to conventional micromachining
processes. Typical applications of laser micromachining include laser bonding of a wafer, or of a
microsystem chip, to one or two parts Mescheder et al. [1], and laser micromachining of 3D
microchannel systems in chemical, biomedical, DNA, and environmental science analyses Qin and
Li [2].
In laser micromachining processes, such as laser welding, drilling, and cutting, the materials
experience heating, melting, evaporating, and re-solidifying stages. As a result, the laser
micromachined components can be affected by a number of factors such as laser beam properties,
system-cooling conditions, and surface roughness/reflectivity of the areas exposed to laser light.
This paper presents synthesis of the laser microwelding processes for packaging. More
specifically, various parameters are evaluated individually and then together to determine their
influence on the finished product. This evaluation is done analytically, computationally, and
experimentally Han and Pryputniewicz [3]. Analytical and computational results yield
temperatures in the heat affected zones (HAZs), Fig. 1. In order to make these results valid, they
are correlated with time dependent temperature measurements, Fig. 2; to facilitate this correlation,
the results shown were normalized. Thermal gradients developed during laser microwelding cause
deformations of surfaces exposed to laser beam. Successful development of advanced packaging
depends on accurate knowledge of these deformations. In order to quantify surface deformations
due to laser microwelding, we have developed an optoelectronic methodology based on the use of
CCD cameras for high-resolution imaging of the affected areas. Figure 3 shows a representative
fringe pattern recorded for one of the HAZs. Interpretation of this fringe pattern produces detailed
spatial distribution of deformations, Fig. 4; the maximum deformations shown are 2.5 Pm, which
may lead to fractures. In fact, in order to understand the influence that a laser beam has on the
materials, we make these measurements before, during, and after the laser microwelding, based on
which we will optimize the processes to obtain minimal deformations.
Comparison of preliminary analytical, computational, and experimental results shows good
correlation and indicates viability of the approach we have developed for synthesis of laser
microwelding processes for packaging.
74 R. J. Pryputniewicz et al.
 1T4. Fatigue and Fracture of MEMS and NEMS 75

THERMAL MANAGEMENT OF RF MEMS RELAY SWITCH

R. J. Pryputniewicz
Worcester Polytechnic Institute, Mechanical Engineering Department / CHSLT-NEST
rjp@wpi.edu

Radio frequency (RF) switches are one of many MEMS devices that make it possible to
communicate, sense, and measure while using minimal amount of space and very low power. The
RF microswitches have either capacitive or resistive configuration. The capacitive switches use a
flexible membrane design, in which capacitance between two electrodes is induced via electrical
voltage; reaching threshold capacitance activates the switch, which enables transmission of a
signal. The resistive switches, on the other hand, make direct metal to metal contact. Such design
usually uses a cantilever beam that bends as voltage is applied to the two electrodes.
In the RF MEMS contact switch configuration, a moving component is a cantilever beam with
a shorting bar at its free end, Fig. 1. The shorting bar comes into contact with the traces on the
substrate as the beam deflects under the electrostatic force loading. As the shorting bar makes
contact with the traces, i.e., as it closes the path, electric signals pass from one trace to the other.
Because of finite internal resitances of the components as well as resistance of the contact
interfaces, Joule heat is generated, affecting performance of the switch.
In this paper, we investigate the Joule heating effects computationally using thermal analysis
system (TAS) software. TAS models the structure to be analyzed by geometrically simple finite
elements, Fig. 2, convert these elements into resistors representing all three modes of heat transfer:
conduction, convection and radiation, and then solves the resistor network using the finite
difference solver, which performs heat balance at each node of the model Pryputniewicz et al. [1].
This entails calculating a node temperature based on the resistances and the temperatures of all
nodes attached to the node in question. During TAS model execution, temperature, or time
dependencies are interpolated for each computation time step.
Representative results obtained for 300 mA current are displayed in Figs 3 and 4. Figure 3
shows that the shorting bar reaches temperatures from 680qC to 717qC, for the geometry,
dimensions, and the material properties considered in this case; for other set of parameters the
results would be different. Figure 4 shows that temperatures in the traces range from 25qC to
45qC. It should be observed that the 665qC temperature difference between the components shown
in Figs 3 and 4 is due to effective heat transfer between the individual components of the switch as
well as with its environment.
Results generated in this study show that TAS provides an effective approach for thermal
management of RF MEMS switch designs.
76 R. J. Pryputniewicz
 1T7. Thin films 77

BUCKLING AND DELAMINATION OF THIN LAYERS ON A POLYMER

SUBSTRATE

A. A. Abdallah, D. Kozodaev1, P. C. P. Bouten2, J. M. J. den Toonder2 and G. de With

Laboratory of Materials and Interface Chemistry, Eindhoven University of Technology
Den Dolech 2, 5600 MB Eindhoven, The Netherlands
1
NT-MTD Co./NTI-Europe. Arnhemseweg 34 d, 7331 BL Apeldoorn, The Netherlands
2
Philips Research Laboratories, Prof. Holstlaan 4, 5656 AA Eindhoven, The Netherlands
a.abdallah@tue.nl

Flexible displays are multi-layered structures consisting of a polymer-based substrate sandwiched

between a numbers of functional layers that act as a gas barrier layer, e.g. inorganic thin silicon
nitride layer (Si3N4), and a transparent conducting oxide layer, e.g. indium tin oxide layer (ITO).
For the realization of flexible displays, a study of the mechanical integrity of multi-layered
structures and their reliability is required.
High compressive stresses in the thin layer in combination with insufficient adhesion at the
interface can cause the layer to buckle and to delaminate from the substrate (see Fig.1). These
buckling phenomena are undesirable from a functional and reliability point of view and therefore
should be avoided.

FIGURE 1: Atomic force microscope image obtained for buckling and delamination of a 400 nm
thin Si3N4 layer on a polymer substrate. Circular buckles with layer cracking due to uniaxial-
compressive external stress.

In the framework of buckling theory for a beam clamped on both edges, numerous buckle
morphologies have been studied for various thin layer structures on substrates, often neglecting the
effect of the substrate’s deformation. Recently, attention is paid to buckling on compliant
substrates [1, 2, 4]. When the substrate deformation is taken into account, significant effects of the
energy release rate on buckles are described, leading to lower critical strains for buckling onset.
Bouten and Van Gils [4] present buckling maps for compliant substrates; buckle sizes (height and
width) are related to adhesion energy and internal strain (prior to buckling). Yu and Hutchinson [2]
define a characteristic length l, adjacent to the buckle, contributing to the energy release upon
buckling. This length is given as

2 hf
l=
1D (1)
78 A. A. Abdallah et al.

where hf is the layer thickness, D is the elastic mismatch (Dundurs’ parameter [3]).
Assuming that only the elastic energy release of the buckling layer contributes to the buckling
process, the energy balance for buckling can be written as [4]

2bGss=2(b  l )(Go  Gc) (2)

where Go is the elastic energy stored in a thin layer per unit width, Gc is the remaining energy in
the buckled portion of the layer and b is characteristic buckle size. The right hand term describes
the energy release in the thin layer upon buckling as a function of the effective buckle length (b +
l), which depends on the layer thickness and the elastic mismatch. As a consequence significantly
more elastic energy is available for buckling formation. The steady-state energy release rate of the
advancing buckle Gss is an important parameter to describe layer buckling and delamination. The
left hand term represents the energy required for delamination.
In this paper attention is paid to the influence of loading mechanism, layer thickness, and level
of adhesion on buckle initiation and propagation. Based on the results of the experimental work, a
numerical model has been formulated to describe buckling and delamination phenomena in these
structures using the J-integral and cohesive zone elements. Different types of buckle profiles are
presented and the buckle evolution mechanism associated with the applied stress is discussed.

The results showed that the buckle onset strain defined at a given quality of adhesion for a
compliant substrate is lower than the one predicted by the model for a rigid substrate [5]. Apart
from that, the critical buckling strain Hc is found to be reduced for a stiff layer on a compliant
substrate due to the rotation and the displacement at the buckle edges. The buckle size is found to
increase with the layer thickness and the uniaxial-compressive strain.

References
1. B. Cotterell and Z. Chen, Int. J. Fracture 104, 169-179, 2000.
2. H.H. Yu and J.W. Hutchinson, Int. J. Fracture 113, 39-55, 2002.
3. J. Dundurs, J. App. Mech. 36, 650-652, 1969.
4. P.C.P. Bouten and M.A.J van Gils, In Proceedings of MRS, Boston, 2004. T4.9
5. J.W. Hutchinson and Z. Suo, Adv. Appl. Mech. 29, 63-191, 1992.
 1T7. Thin films 79

CARBIDE COATED CUTTING TOOL PROPERTIES INVESTIGATION BY

NANO-MECHANICAL MEASUREMENTS UNDER 250-500°C.

B. Vasques, D. Joly, R. Leroy, N. Ranganathan and P. Donnadieu1,2

LMR, Productique Department, University François Rabelais de Tours,
7 rue Marcel Dassault, 37204 Tours, France
1
SAFETY Production SA, rue Henri Garih, 37230 Fondettes, France
2
CEROC Cutting Tool Research Centre, rue Henri Garih, 37230 Fondettes, France
brigitte.vasques@groupe-safety.com, rene.leroy@univ-tours.fr, ranganathan@univ-tours.fr

Since most mechanical properties are temperature dependent. This one of the reasons that usual
mechanical tests at room temperature are insufficient to characterize advanced PVD coatings
designs. Many of these coating inserts are exposed to high temperatures in either processing or
working environments such as high speed machining or dry cutting operations [1, 2, 3, 4]. Wear
reaction between tool and work piece takes place at high temperature [5]. We need to study these
materials under such environment to really understand and predict there behavior. That’s why wear
resistance of PVD hard coatings are investigated with a particular attention to their critical
mechanical properties at higher temperature.
These properties are measured by a Nano Test pendulum set up specially equipped for high
temperature measurements. The hot stage consists of a thermally insulating ceramic block that is
attached to the specimen holder. A thermal shield is placed between the pendulum and the stage to
prevent thermal instability of the probe (Fig. 1).

Figure 1: Nano Test pendulum modified for high temperature measurements.

Indentations are performed at room temperature, 250 and 500°C using a Berkovich indenter.
The following parameters are used for the measurements: maximum load (50-250mN), loading
rate (2-5 mN/s, dwell time at maximal load (10-30 s). Scratch and impact experiments are
employed at the same room temperature to analyze critical load to adhesion failure and fatigue
properties allowing identification of operating limits. Microstructure analysis of film is done by
Scanning Electronic Microscopy (Jeol JSM6480).
Investigation of thin films and surface mechanical properties at high temperature will allow
optimizing the “Couple Tool Material “projects in metal cutting.
Keywords: Berkovich indenter, coated inserts, PVD films, High Temperature, Nano Test
techniques
80 B. Vasques et al.

References
1. Beake D., Garcia M. and Smith F., Thin Solid Films, 398-399 (2001) 438-443.
2. Beake D. and Ranganathan, ICMAT, Singapore, 3-8 July 2005.
3. Karimi A., Wang Y., Morstein M., Thin Solid Films, 420-421 (2002) 275-280.
4. Gong J., Miao H., Peng, Acta Materialia, vol. 52, 785-793, 2004.
5. Morant C., Prieto, Forn A., Surface and Coated Technology, 180-181 (2004) 512-518.
 1T7. Thin films 81

DIAMOND COATING DEBOUNDING IN TOOL APPLICATION

David Moulin, Pierre Chevrier, Paul Lipinski and Thierry Barré1

Laboratoire de Physique et Mécanique des Matériaux (LPMM) /
Ecole Nationale d’Ingénieurs de Metz (ENIM)
ENIM, ile du Saulcy, 57045 Metz, France
1Balzers Luxembourg, ZI Hanoesbesh, Luxembourg

d.moulin@enim.fr, chevrier@enim.fr, lipinski@enim.fr, thierry.barre@balzers.com

Machining of recent engineered complex materials such as aeronautical ones always require new
tools to lower production cost and in the same time increase quality. One promising solution is the
use of Chemical Vapour Deposition (CVD) diamond coatings on carbide tools, which is studied
here. Diamond is used because of its extreme hardness, which made it a perfect candidate to reduce
wear of the cutting edge. It is synthesized in a batch at low pressure and medium temperature
directly on the tools. Deposition process on tungsten carbide tool bound with cobalt, which is the
main industrial material for tool manufacture, and especially the adhesion of the diamond layer to
the substrate has to be improved, and many research are focused on this point [1]. The whole
process of deposition is described elsewhere [2]. This paper proposes an approach to study the
decohesion of the diamond layer via a combination of experiments and simulation.
Many techniques to study adhesion of the diamond layer were developed, such as indentation
[3], sand blasting [4]. However, these techniques have a high limitation: they are only qualitative,
and don’t take into account the physic of adhesion. Results obtained with these techniques depend
of the test procedure and of the machine used to apply the load. Results of a series of tests cannot
be compared to others. A better comprehension of the physic involved can lead to a quantitative
test, with a good repeatability and a measured value, which do not depend of the test process.
Experimentally, it can be observed that diamond delamination occurs because of a stress along
the interface between carbide and diamond (Fig. 1).

Figure 1 : Diamond fracture on rake face of a turning insert

This stress is of different natures. When coating flaked during machining, pressure and
temperature induced by the machining process, which are external loads, are responsible of the
degradation of the interface. However, debounding can occurs into the batch, without any external
loading, because high thermal stresses during cooling (see Fig. 2). These high thermal stresses are
due to by the difference of thermal dilatation coefficients and can be calculated with a finite
element model. This stress has to be taken into account together with that due to the external
loading. Observation of the insert after decohesion (fig 1) confirms that fracture occurs at the
interface of the two materials.
82 D. Moulin et al.

Figure 2 : Normal stresses induced by thermal cooling along the interface Carbide – diamond
depending on thickness and edge radius (centred on the edge)- Results on FEM calculation

In this work, adhesion is studied by completing experimental loading of the system by

simulations with a Finite Element Model (FEM) in order to obtain the stress state responsible for
the de-bounding of the diamond layer. A criteria is then written to characterize the behavior of the
system based on a combination of the normal and shearing stress responsible for de-cohesion.
Different loading configurations are imagined to obtain the form of the criteria: 3 or 4 points
flexion tests, impact tests, and variation of the radius edge. Flexion tests induce pure traction-
compression stress in both the coating and the substrate but with two different levels, leading to
shearing of the interface and to debounding of the coating.
Indentation and impacts tests are more complex loads, respectively quasi-static and dynamic in
nature.
Deposition on different edge radii leads to different thermal residual stresses, which play the
role of the load. A critical radius can be found, which is the smallest radius before debonding.
Stress state can be calculated according to this radius.
These different techniques, combined with the simulation, give a better understanding of the
phenomenon that control the adhesion. They lead to a debounding criteria useful to characterize
adhesion.

References
1. S. Kamiya, U.H. Takahashi, R. Polini and E. Traversa, Diamond Relat. Mater. , vol 9, 191-
194, 2000
2. H.G. Prengel, W.R.Pfouts, and A.T. Santhanam, Surf. Coat. Technol. , Vol. 102, 183 – 190,
1998
3. G.Jörgensen, M.Lahres, and J.Karner, Surf. Coat. Technol. , vol 97, 238-243, 1997
4. Friedrike Deuerler, Heiko Gruner, Michael Pohl and Ladji Tikana, J. Mater. Process.
Technol. , Vol. 99, 266 – 274, 2000
 1T7. Thin films 83

INTERFACIAL STRENGTH OF CERAMIC THIN FILM ON POLYMER

SUBSTRATE

M. Omiya and K. Kishimoto

Department of Mechanical and Control Engineering,
Tokyo Institute of Technology,
2-12-1, O-okayama, Meguro-ku, Tokyo, 152-8552, Japan
oomiyam@mep.titech.ac.jp

In this paper, we aim to evaluate the interfacial adhesion strength between ITO (Indium Tin Oxide)
coating layer and PET (Poly-Ethylene Terephthalate) substrate. To evaluate the interfacial
adhesion strength, we focused on the buckling phenomena of the coating layer during simple
tensile tests. During the tensile tests of PET/ITO specimen, the buckling induced delamination
occurred on the coating layer. These phenomena are related to the interfacial adhesion strength and
we evaluated the interfacial adhesion strength by considering the energy balance during the
buckling induced delamination and cracking.
During tensile tests, two types of cracks are formed on the coating layer: vertical cracks and
parallel cracks. Parallel cracks are induced by the buckling of thin coating film on the polymer
substrates and these phenomena are strongly related to the interfacial adhesion strength. Therefore,
we estimate the strain energy stored in the coating film by modified shear-lag model[1] and
considering the energy balance to evaluate the interfacial adhesion strength between thin film and
substrate.
Due to the tensile loading, the coating layer is assumed to be segmented as shown in Fig.1. In
the coating layer, the compressive loads are induced by the interfacial shear force. From modified
shear-lag model, when the compressive stress in the substrate is V c , the compressive stress in the
coating layer is,

(1)

where, E f , Es are Young’s modulus for coating layer and substrate, respectively. h, H are the
thickness of the coating layer and the substrate. L is the segment length of the coating layer
andăis the load transfer length. The energy balance before and after buckling induced
delamination and cracking is,

We Wd  Wec  W f
(3)

where the strain energy before buckling crack is W e , the delamination energy W d , the strain
energy of undelaminate coating layer is W and the energy for crack formation is W f . Note that
ec
the change of the strain energy in the substrate is small because the compressive stress is
continuously acting on the substrate from the applied load and it constrains the elastic recover of
the substrate due to the delamination and cracking. Then, the interfacial strength, * d , can be
described as,
84 M. Omiya and K. Kishimoto

(4)

where is the fracture toughness of the coating layer, l d is the delamination length and b is the
width of the segmented film. In this study, the dimensions needed to evaluate the interfacial
strength from by Eq.(4) are measured by atomic force microscope. From this observation, we
obtained the delamination length, l d , and evaluate the interfacial adhesion strength, * d . The
2
average value of the interfacial strength for PET/ITO specimen is 19.6 J/m and PET/ITO(UV)
specimen is 9.6J/m2. The obtained results agree well with the peel test results[2] as shown in Fig.2.
During the buckling and delamination process of the coating layer, the phase angle continuously
changes. Therefore, the obtained results are considered to be the averaged interfacial strength for
several phase angles. However, this simple approach is useful for the first estimation of the
interfacial strength and it is valuable for quick interfacial evaluation at manufacturing premises.

FIGURE 1. Buckling induced delamination and cracking.

FIGURE 2. Comparison with the results measured by Multi-stages Peel Test.

References
1. Yanaka, M., Kato, Y., Tsukahara, Y. and Takeda, N., Thin Solid Films, Vol.355-356, 337-
342, 1999.
2. Omiya M., Inoue H., Kishimoto, K., Yanaka, M. and Ihashi, N., Journal of the Society of
Materials Science, Japan, Vol.52, 856-861, 2004.
3. Omiya M., Inoue H., Kishimoto, K., Yanaka, M. and Ihashi, N., Key Engineering Materials,
Vol.297-300, 2284-2289, 2005.
 1T7. Thin films 85

DELAMINATE BEHAVIOR OF PVD/CPVD THIN FILM

S. Doi and M. Yasuoka

700 Dannoharu Oita-shi,JAPAN
doi@cc.oita-u.ac.jp

Accompanied with sophistication of the technology of creating surface, multi-layered structures of

film tend to increase for improving the accuracy with a large degree of the freedom playing an
important role. In particular, various kinds of methods for producing film have been proposed and
the application range of film has become wider. In this research, focusing on coating film materials
fabricated using PVD, CVD and a combined method of PVD+CVD, grasping their surface
morphology is aimed. On the other hand, the reliability of such a new material with a different
property between the surface function and the interface function is required. Accordingly, different
techniques from conventional methods of evaluating functions of laminated material are needed.
Therefore, in this research, so-called textured materials with new structure coated by a series of
carbon film or titanium nitride film as described in the following, were selected. The effect of
applied impact loading on the adhesion and the interface quality of their films was investigated by
observation using laser confocal microscope and fractal dimensional analysis of cracks or
exfoliation surfaces.
The molecular structure of DLC attracting attention recently differs from one of diamond and
graphite and constitutes a crystal structure with no aligning. This structure is said to be a kind of
amorphous structure with a similar characteristic to one of diamond showing high hardness and a
partially similar to one of graphite. The basic arrangement of the carbon includes the atomic
arrangement of SP‚R and this causes a different property of hardness of DLC. Also, TiN has a
structure with substrate of M35, that is to say, a two-layered structure and this leads to a crystalline
textured structure constituting the interface and the surface, respectively. As a method of
examining surface characteristics in fractal dimensional analysis, the box count method was
adopted because this is an adequate method for observing the local change.
86 S. Doi and M. Yasuoka

Fig.2 Relationship fractal dimension and cyclic impact numbers

 1T7. Thin films 87

EXPERIMENTAL STUDY OF MICROHARDNESS AND FRACTURE OF

IMPLANTED GÁLLIUM NITRIDE FILMS

P. Kavouras, M. Katsikini, E. Wendler1, W. Wesch1, H. M. Polatoglou, E. C. Paloura, Ph.

Komninou and Th. Karakostas
Physics Department, Aristotle University of Thessaloniki
541 24 Thessaloniki, Greece
1Institut für Festkörperphysik, Friedrich-Schiller-Universität Jena, Max-Wien-Platz 1, Jena D-

07743, Germany
pkavo@skiathos.physics.auth.gr, katsiki@auth.gr, wendler@pinet.uni-jena.de,
Werner.Wesch@uni-jena.de, hariton@auth.gr, paloura@physics.auth.gr, komnhnoy@auth.gr,
karakost@auth.gr

The investigation of radiation effects in III-V semiconductors is of current interest because of the
potential application of ion implantation in the production of electronic and photonic devices. Of
the plethora of III-V semiconductors, gallium nitride (GaN) has attracted keen interest, during past
decade, as wide gap semiconductor for numerous applications, including high-power or high
frequency devices and high-power switches. Although there is considerable interest in determining
the influence of ion implantation on the mechanical properties of GaN films, the matter has
received only scant attention, see for example Kavouras et al. [1]. Indeed, studies of the processes
controlling hardness, contact damage and cracking of epitaxially grown GaN films have significant
technological importance.
In this work, we present the results of the application of the Static Indentation (SI) technique,
using Vickers and Knoop indenter geometries, in the investigation of implantation induced effects
on the mechanical properties and fracture of epitaxially grown GaN films. More specifically, we
compare a number of physical characteristics that govern the elasto-plastic behavior before and
after implantation. The implanted species used to produce implantation induced damage were Au,
Xe, Ar and O ions.

TABLE 1. Structure and microhardness of as-grown and implanted GaN films. The errors
represent the standard deviations.
Material Name Structure Microhardness / GPa
As-grown GaN GaN - 13.73r0.12
Au-implanted GaN:Au Amorphous 12.28r0.16
Xe-implanted GaN:Xe Amorphous 10.19r0.10
Ar-implanted GaN:Ar Amorphous 9.41r0.14
O-implanted GaN:O Heavily damaged 14.53r0.26

A Knoop indenter was utilized to obtain the microhardness values (Table 1). The structure of
GaN implanted epilayers was obtained by X-ray Absorption Fine Structure (XAFS) technique. It
was found that in the case where implantation produces heavily damaged structure microhardness
is increased, while implantation induced amorphisation lowers microhardness value. Additionally,
the influence of the indentation load on the microhardness value was studied, i.e. the Indentation
Size Effect (ISE) curves were obtained in all cases. It was observed that GaN and GaN:O showed a
normal ISE behavior, i.e. microhardness value increases with decreasing indentation load, while
GaN:Au, GaN:Xe and GaN:Ar showed a Reverse ISE (RISE), i.e. microhardness value decreases
with decreasing indentation load. The magnitude of the post-indentation healing of indentation
88 P. Kavouras et al.

prints was evaluated by measuring (a) the relative length of the two normal diagonals of Knoop
indentation prints and (b) the depth of residual indentation prints by means of Atomic Force
Microscopy (AFM). The above information can give an estimation of the elastic behavior and
elastic modulus, as it was firstly recognized by Marshall et al. [2].

FIGURE 1. Optical micrographs, under transmitted illumination, of fracture formation sequences

for GaN (a, b and c) and GaN:O (d, e and f) films.

Fracture sequence was also observed in all cases, utilizing Vickers indenter geometry
according to the formalism elaborated by Cook and Pharr [3]. GaN did not show initiation of a
specific microcrack type. Only cumulative fracture events and/or film detachment were observed
for indentation loads higher that #1.5 N. In all other cases, fracture occurred in lower loads,
indicating that implantation has an effect analogous to embrittlement. Fig. 1 shows characteristic
sequences of fracture formation in the cases of GaN and GaN:O.

References
1. Kavouras, P., Katsikini, M., Kehagias, Th., Paloura, E.C., Komninou, Ph., Antonopoulos, J.,
Karakostas, Th., J. Phys. Cond. Mat., vol. 14, 12953-12959, 2002.
2. Marshall, D.B., Noma, T., Evans, A.G., J. Amer. Cer. Soc., vol 65, C175-C176, 1982.
3. Cook, R.F. and Pharr, G.M., J. Amer. Cer. Soc., vol 73, 787-817, 1990.
 1T9. Failure of nanocomposites 89

CRACK TIP STRAIN FIELD AND ITS PROPAGATION CHARACTERISTICS IN

A POLYMER FOAM

Fu-pen Chiang, Sheng Chang and Yi Ding

Department Of Mechanical Engineering, Stony Brook University
Fu-Pen.Chiang@sunysb.edu, fchiang@notes.cc.sunysb.edu

In recent years there has been an increasing interest in using foam composites as shipbuilding
materials. A sandwich panel built of fiberglass face sheets and a foam core has strong rigidity and
bending strength. It’s advantage for ship construction is obvious. Furthermore when a small
amount of nanoparticles are added to the material, it tends to increase its stiffness and retards fire
as well. In this paper we employ a unique micro/nano speckle technique to investigate the crack tip
deformation field of two different polymer foams at a length scale that has never been studied
before. The principle of the technique is described in [1]. Figure 1 shows the micrograph recorded
by a scanning electron microscope of a crack propagating through a NEAT foam specimen under
uniaxial tension. The void in front of the crack tip was kind of spontaneously generated as a result
of the load. The main crack then tends to link itself towards the void as demonstrated in the
sequence of pictures shown.

Fig 1. Crack propagation in a NEAT foam specimen under uniaxial tension

While in the macro scale the crack will largely show propagation in the mode I characteristics
at a micro/nanoscale the crack path is far from being a straight line. It tends to circle around a foam
cell and then advance. But under the right circumstance it will break across the cell-to-cell
interface. The nature of the propagation characteristics is revealed in the next example. Figure 2
shows the two micrographs (again recorded by an scanning electron microscope) at 50X and 100X
magnifications, respectively, of a nano-phased foam beam specimen under 3-point bending with a
single edge crack. The crack is perpendicular to the longitudinal axis of the beam and the tip of the
crack is visible in the SEM micrograph at 50X.
Figure 3 shows the displacement contours surrounding the crack tip at 50X magnification. The
deformation field is rather complicated and it bears almost no relationship to the classical
displacement field in the region around a crack tip. At a higher magnification, the deformation
field is even more complicated, as shown in Figure 4.
90 F.-P. Chiang et al.

Fig 2. SEM micrograph of a nano-phased PVC foam beam under 3 point bending with a single
edge crack

Fig 3. u and v displacement fields at 50x. Contour constant = 0.002Pm

Fig 4. u and v displacement fields at 100x. Contour constant 0.001Pm, Nano-phased PVC foam
specimen with an edge crack under 3 point bending.

As can be seen this is not a classical crack tip deformation field (for an isotropic and
homogeneous material) at all. Of particular interest is the dense contour lines (at 1 nm/contour)
surrounding a foam cell in front of the crack tip, indicating a stress concentration region. This
implies that the cell-to-cell interface is strong and the crack tends to propagate around the foam cell
as observed.
Acknowledgement
The author would like to thank Dr. Yapa D.S. Rajapakse, Manager of the Solid Mechanics
Program, US Office of Naval Research for supporting this work through grant #N000140410357
and Professor Hazzan Mahfuz of Tuskepee University for providing the foam materials.

Reference
1. Chiang F.P., 2003a, “Evolution of white light speckle method and its application to micro
nanotechnology and heart mechanics”, Optical Engineering Vol. 42 no.5, 1288-1292.
 1T9. Failure of nanocomposites 91

HOW TO TOUGHEN CERAMICS – NANOCOMPOSITES

H. Awaji and S.-M. Choi

Nagoya Institute of Technology
Gokiso-cho, Showa-ku, Nagoya, 466-8555 Japan
awaji@nitech.ac.jp

Ceramic-based nanocomposites with intra-type nano-structure have high strength with moderate
fracture toughness. In this research, the toughening and strengthening mechanisms of the ceramic-
based nanocomposites will be explained by dislocation activities even in brittle materials. Based
on these mechanisms, we fabricated toughened and strengthened alumina-nickel nanocomposites.
Fracture toughness of the annealed nanocomposites was 7.6 MPam1/2, which was two times higher
than that of monolithic alumina.
Dislocation activities
In intra-type nano-structure, nano-sized second-phase particles are embedded within the matrix
grains. The highest strength or fracture toughness is mostly achieved when only a few volume
percent of the second-phase particles are dispersed in ceramic materials. The structural
characteristic of these nanocomposites results in the generation of thermally induced residual
stresses around the dispersed particles in matrix grains [1]. We analyzed residual stresses using a
simplified model consisting of a spherical particle within a concentric matrix sphere with axial
symmetry [2]. Residual stresses numerically calculated on the particle-matrix boundary for
alumina/nickel nanocomposites are shown in Table 1, where the symbols with suffix p indicate the
properties of the particle (nickel) and the symbols with suffix m are the properties of the matrix
(alumina). It is noted that there is a large maximum shear stress on the boundary.
Figure 1 shows the temperature dependence of the residual shear stress on the boundary of
alumina/nickel, and critical resolved shear stresses for basal and prism plane slips in a single á-
alumina crystal measured by Lagerlöf et al. [3] This figure indicates that dislocation movements
are possible in the alumina grains at temperatures ranging from 600 to 1400 ºC, suggesting that this
temperature range is quite important in creating dislocations in the alumina matrix during the
cooling process.
Strengthening mechanism
The grains of sintered alumina contain tensile residual stresses resulting from anisotropic
thermal expansion, etc. Therefore, it is conceivable that the large crack along a grain boundary
created by the synergetic effect of both residual stresses and processing defects, will be equivalent
to the grain size of the materials and that the weakest crack generated along a boundary in the
specimen will dominate the strength of the specimen. Nanocomposites, however, will yield
dislocations around the particles, and the dislocations release residual stresses in the matrix.
Consequently, the defect size along the grain boundaries is reduced in nanocomposites.

TABLE 1. Residual stress along the particle-matrix boundary in alumina-nickel nanocomposites.

System Į m /Į p ×10 -6 E m /E p IJ m ax
(K -1 ) (GPa) (GPa)
Al 2 O 3 -Ni 8.8/13.3 380/207 1.0
92 H. Awaji and S.-M. Choi

Also, the dislocations are difficult to move in ceramics at room temperature, serve as origins of
small stress concentrations, and create nano-cracks around the propagating crack tip. These nano-
cracks reduce the strength of the alumina matrix slightly and change the fracture mode from
intergranular fracture in monolithic alumina to transgranular fracture in nanocomposites.
Toughening mechanism
To improve the intrinsic fracture toughness of ceramics, the fracture energy consumed in the
frontal process zone (FPZ) must be increased [4], which is called the FPZ toughening mechanism.
Because ceramics are brittle, the FPZ of ceramics is considered to be constructed by many nano-
cracks rather than dislocations in metals. Therefore, we must consider how to create many nano-
cracks in the FPZ.
Assuming that we obtained the nano-structure with dispersed dislocations within the matrix
grains after annealing, these dislocations become sessile dislocations at room temperature. In this
situation, when the tip of a propagating large crack reaches this area, these sessile dislocations will
operate as nano-crack nuclei in the vicinity of the propagating crack tip. The highly stressed state
in the FPZ is then released by nano-crack nucleation, and the nano-cracks expand the FPZ size,
enhancing the intrinsic fracture toughness of the materials.

References
1. T. Matsunaga, et al., J. Ceram.Soc., Japan, vol. 113, 123-125, 2005
2. H. Awaji, et al., Mechanics of Materials, vol. 34, 411-422, 2002.
3. K. P. D. Lagerlöf, et al., J. Am. Ceram. Soc., vol. 77, 385-397, 1994.
4. S-M. Choi and H. Awaji, Science & Tech. of Advanced Mater., vol. 6, 2-10, 2005.
 1T9. Failure of nanocomposites 93

DEFORMATION AND FRACTURE BEHAVIOUR OF NANOCOMPOSITES

S. Dunger, J. K. W. Sandler, K. Hedicke and V. Altsadt

University of Bayreuth, Department of Polymer Engineering, FAN A - Universitatsstrase 30
95448 Bayreuth, Germany
Siljana.Dunger@uni-bayreuth.de

Polymer nanocomposites are of great scientific interest due to the potential of improving the
resulting physical and mechanical properties of components, which cannot be modified using
conventional reinforcements. However, the successful development and industrial implementation
of such novel materials, especially of structural components, pose unique challenges.
Nanocomposites with oriented nanoparticles and good adhesion between both phases can offer a
stiffness and strength increase comparable to conventional fibre-reinforced composites [1 - 3];
however, only moderate filler weight fractions are possible due to the large specific surface area of
the particles [4]. Nevertheless, at the same time, nanoscale fillers can maintain or even increase the
composite thoughness [5-7]. For certain applications such as delicate structures the use of
nanoscale reinforcements therefore offers a unique potential [8].
For a given matrix and nanoscale reinforcement system, processing leading to a good
dispersion and distribution as well as orientation of the nanophase is crucial. The dispersion of
nanoparticles especially depends on a magnitude of processing parameters, e.g. shear rate and
processing temperature, as well as on the selection of appropriate materials. In addition,
interactions between the nanoscale reinforcement and the polymer matrix during processing
significantly influence the orientation of individual particles and the molecular morphology in the
vicinity of the particles.
The presentation will highlight the resulting deformation and fracture behaviour of different
polymer nanocomposites, taking into account variations in the matrix morphology as a function of
nanofiller type, geometry, surface chemistry and processing. Given that nanofillers have been
shown to lead to distinct variations in the resulting interphase, an investigation of the fracture
mechanics is especially important in order to understand the commonly occurring transition from a
ductile to a brittle behaviour of such nanocomposites. In order to analyse the influence of the
matrix ductility on the resulting nanocomposite fracture behaviour, polystyrene, polyamide 6, and
polystyrene-polybutadiene-polystyrene block copolymer were chosen as matrix materials. The
influence of the nanofiller geometry is investigated by using silicate clays, silicium dioxide
nanoparticles and tubular clay nanostructures. Moreover, the resulting interphase properties are
modified by different silicate surface treatments. The mechanical properties of the nanocomposites
as well as of comparative glass fibre-reinforced composites were characterised by tensile tests, KIc
tests and crack propagation tests. The fractured surfaces of the specimens were extensively
analysed by SEM. In order to investigate the fracture behaviour during deformation, TEM analyses
help to observe and explain the crack propagation of the nanocomposites as presented in fig 1.
Finally, a comparison between traditional fibre-reinforced systems and the novel nanocomposites
is made.
94 S. Dunger et al.

Fig. 1: Crack propagation and deflection in polyamide 6 nanocomposites

References
1. Fornes, T. D., Paul, D. R.; Polymer 44, (2003), 4993-5013
2. Galgali, G., Agarwal, S., Lele, A.; Polymer 45, (2004), 6059-6069
3. Fisher, F. T., Bradshaw, R. D., Brinson, L. C.; Composite Science and Technology 63,
(2003), 1689-1703
4. Luo, J.-J., Daniel, I. M.; Composite Science and Technology 64, (2003), 1607-1616
5. Zerda, A. S., Lesser, A. J.; Journal of Polymer Science: Part B 39, (2001), 1137-1146
6. Chen, L., Wong, S.-C., Liu, T., Lu, X., He, C.; Journal of Polymer Science: Part B 42, (2004),
2759-2768
7. Gojny, F. H., Wichmann, M. H. G., Kˆpke, U., Fiedler, B., Schulte, K.; Composite
8. Science and Technology 64, (2004), 2363-2371
9. Nam, P. H., Maiti, P., Okamoto, M., Kotaka, T.; Polymer Engineering and Science 42,
(2002), 1907-1918
 1T9. Failure of nanocomposites 95

FRACTURE MECHANISMS IN CARBON NANOTUBE-REINFORCED

COMPOSITES

E. T. Thostenson and T.-W. Chou

Department of Mechanical Engineering and
Center for Composite Materials
University of Delaware
Newark DE 19716 USA
chou@me.udel.edu

Carbon nanotubes have been targeted for potential applications ranging from the next generation of
computers and flat-panel displays to structural and functional materials. In addition to their well-
known stiffness (> 1 TPa) and strength (~30 GPa) properties, carbon nanotubes also possess
exceptionally high electrical and thermal conductivities, with the axial thermal conductivity near
that of crystalline diamond. The unique mechanical and physical properties of nanotubes offer
tremendous opportunity for the development of multi-functional composites [1, 2]. Full
understanding of the thermo-mechanical behavior of nanotube-based composites, requires
knowledge of the elastic and fracture properties of carbon nanotubes as well as interactions at the
nanotube/matrix interface. Although this requirement is no different from that in conventional
fiber composites [2], the scale of the reinforcement phase diameter has changed from micrometer
(e.g. glass and carbon fibers) to nano-meter. The change in reinforcement scale poses new
challenges in the development of processing techniques for these composites as well as
characterization techniques and methodologies to measure their elastic and fracture behavior.
A fundamental knowledge of the process/structure/property relationships is required to enable
the design of multi-functional materials by structuring at the nanoscale. A novel technique to
produce continuous nanocomposite ribbons of aligned multi-walled carbon nanotubes has been
developed [3]. This model nanocomposite system serves as a basis for the investigation of
structure/property relationships through characterization of their elastic and fracture behavior. The
elastic and fracture behavior of the model nanotube composites indicate the anisotropy in the load
transfer and confirm that nanotubes are able to carry load that is transferred via shear stresses at the
nanotube/matrix interface and through characterization of this model system a fundamental
knowledge of their structure/property relations has evolved [4]. The tensile fracture behavior of
carbon nanotube composites show similar mechanisms as in traditional fiber composites including
nanotube fracture, pullout, and crack bridging [3]. For compressive deformation, critical
nanoscale buckling behavior of carbon nanotubes was observed where small diameter nanotubes
deform through global bending analogous to Euler-type buckling and large diameter nanotubes
show locally sharp kinking [5]. These deformation behaviors suggest a critical diameter may exist
for the change in buckling modes and could have significant implications on the nanoscale design
of composite compressive properties.
Recent research by Gojny et al. [6] has shown that very low concentration of double-walled
carbon nanotubes (0.1 wt%) can result in substantial improvements in fracture toughness. In order
to evaluate the influence of multi-walled carbon nanotubes on the fracture toughness of epoxy
nanocomposites we fabricated composites with nanotube contents ranging between 0.1 wt% and 1
wt% in an EPON 862 epoxy matrix. The nanotubes were first dispersed in the epoxy resin and the
curing agent (Epi-Cure W) was added. The nanocomposites were then placed in a mold and cured
for 6 hours at 130oC. Fracture toughness measurements were conducted using the single-edge-
notch bending (SENB) method. Specimens were notched with a tapered diamond blade and a pre-
crack was introduced by tapping with an ultra-sharp carbon steel razor blade.
96 E. T. Thostenson and T.-W. Chou

The fracture toughness of the epoxy nanocomposites was significantly improved as compared
to the unreinforced resin. This indicates that nanotubes provide a reinforcing effect in improving
the fracture toughness through crack deflection or nanotube fracture and pullout. SEM
micrographs of the composite fracture surface show a change in the micron-scale surface
roughness and also the presence of nanotube pullout.

References
1. Thostenson, E.T., Li C.Y. and Chou T.W. Compos. Sci. Technol., vol. 65, 491-516, 2005
2. Thostenson, E.T., Ren Z.F. and Chou T.W. Compos. Sci. Technol., vol. 61, 1899-1912, 2001
3. Thostenson, E.T. and Chou T.W. J Phys D: Appl. Phys. vol. 35, L77-L80, 2002
4. Thostenson, E.T. and Chou T.W. J Phys D: Appl. Phys. vol. 36, 573-582, 2003
5. Thostenson, E.T. and Chou T.W. Carbon vol. 42, 3015-3018, 2004
6. Gojny F., Wichmann M and Kopke U, et al. Compos. Sci. Technol., vol. 64, 2363-2371, 2004
B. TRACKS

B2. Engineering Materials and Structures

 2T1. Physical aspects of fracture 99

FRACTAL APPROACH TO CRACK PROBLEMS WITH NON-ROOT

SINGULARITY

A. Kashtanov
St.-Petersburg State University
Universitetsky pr. 28, 198504 St.-Petersburg, Russia
arsen@ak1340.spb.edu

Fractal approach to the problems of fracture mechanics with non square root singularity of the
stress field is discussed. Fractal generalization of Griffith energy balance equation is proposed and
analysis of the fracture process at the sharp angular notch in a plate is performed. Good
correspondence between suggested approach and experiments is observed.
It is obvious that propagation of fracture surface is much more complex process than a simple
spreading of rectilinear crack with smooth faces and usually the crack surface has a lot of
irregularities of different sizes. The roughness of fracture surface can be accounted with the help of
fractal correction for the calculation of specific properties suggested by Mandelbrot [1]. It allows
us to receive more precise model of fracture process and to solve the problems of crack mechanics,
which do not have an adequate solution within the frameworks of traditional theory. One way is
the simulation of the crack by fractal with dimension determined from some theoretical reasons at
the fixed scale level. Then, solving a specific problem for simulated “fractal” crack it is possible to
determine some physical magnitudes, which values can be easily checked experimentally, for
example, the value of critical loading.
In particular it is suggested to evaluate the fractal dimension by sufficing to the energy balance
equation. In order to construct this equation for the case of “fractal” cracks the following
generalized energy balance concept was proposed by Kashtanov and Petrov [2]:
• The work of crack opening ' W is the integral parameter of stated problem and is
determined at the macroscopic scale level.

• The surface energy of crack '3 is calculated using the crack fractal length L to take into
account the microstructure of crack surface.
• The fractal dimension D of simulated crack is defined from the validity of energy balance
equation ' W '3 at the macroscale.
• At the macroscopic scale level the fracture process is characterized by the size of an
elementary fracturing cell

2 K 12c
d
S V c2 (1)

where V c is the ultimate strength of defectless material and K 1c is the static fracture toughness.
Therefore it is convenient to use the fractal low at the macroscale in the form

L l d  D d (2)

where l is the crack length at the macroscale.

100 A. Kashtanov

As an example of such approach the plane problem about an angular notch can be considered.
This problem is characterized by the non square root singularity of the stress field and it has no
solution in Griffith–Irwin theory of fracture. Nevertheless, if we suppose that the crack formed in
the notch vertex is a fractal then the connection between the value of fractal dimension of modeled
crack and actual structural parameters of fracture process can be constructed analytically from the
generalized energy balance equation. Then the critical load required for fracture at the crack tip can
be found and good coincidence with experiments is observed (Fig. 1).

FIGURE 1. The dependence between the critical loading and the hole length.

At the Fig. 1 the dependence between fracturing loading and the notch angular is displayed.
The solid line corresponds to the solution of generalized energy balance equation; the dotted line
displays the same dependence calculated from Neuber – Novozhilov fracture criterion; and
experimental data are pointed by circles. All the experiments have been conducted in
St.-Petersburg University by I. Bugakov and I. Demidova.

References
1. Mandelbrot, B. B., The Fractal Geometry of Nature, Freeman, Berlin, New York, 1983.
2. Kashtanov, A.V., Petrov, Y.V., Int. J. of Fracture, vol. 128, N 1, 271-276, 2004
 2T1. Physical aspects of fracture 101

NEW METHOD FOR ANALYSING THE MAGNETIC EMISSION SIGNALS

DURING FRACTURE

Gy. B. Lenkey, N. Takacs1, F. Kun2 and D. L. Beke3

Bay Zoltán Foundation for Aplied Research, Institute for Logistics and Prod. Systems, Miskolc,
Hungary
1
Metalelektro Co. Budapest, Hungary
2
Department of Theoretical Physics, University of Debrecen, Debrecen, Hungary
3
Department of Solid State Physics, University of Debrecen, Debrecen Hungary
lenkey@bzlogi.hu, takacsn@delfin.klte.hu, feri@dtp.atomki.hu, dbeke@delfin.klte.hu

In dynamic fracture testing the precise determination of the onset of crack initiation is crucial in
order to obtain characteristic quantities of the material. This task can be solved easier in case of
brittle fracture since brittle fracture is usually accompanied by a sudden drop in the force signal.
But in case of ductile fracture or if stable crack propagation occurs before unstable one, the instant
of crack initiation cannot be determined directly from the force signal. In these cases additional
measurement techniques should be applied. The magnetic emission technique has been proved for
detecting brittle crack initiation of ferromagnetic materials [1-3].
Two physical phenomena contribute to the magnetic emission signal [1]: (a) mechanically
induced Barkhausen signals appear when the internal magnetic structure changes during loading,
and (b) a propagating crack causes the internal magnetic field to emerge from the solid into the gap
between the two crack surfaces, thereby changing the external magnetic field. These field
variations can be observed locally by a magnetic transducer which basically consists of a coil. The
transducer's output voltage is the magnetic emission (ME) signal which is proportional to the
derivative of the magnetic field (MF).
The aim of the present work was to apply new statistical methods [4-9] to analyze the magnetic
emission signal to extract more information about the fracture process.
Magnetic emission spectra recorded in dynamic fracture experiments on ferromagnetic
materials are composed of more or less well-separated voltage peaks which provide direct
information about the microscopic processes involved in fracture. It will be illustrated, that before
crack initiation sudden movements of domain walls under external mechanical loading results in
low peaks. After crack initiation the opening of the growing crack provides the dominating
contribution for the voltage signals and is responsible for the higher peaks of the spectrum. After
removing a background voltage level the ME spectrum is characterised by the distribution of the
height, area, and width of peaks, furthermore, correlations among these quantities. One example
for the evaluated peak parameters is shown in Fig. 1.
While for the “low peaks” part (before crack initiation) no typical functional form can be
extracted, for the height distribution of peaks in the crack propagation regime the distributions
show power law behavior over a range of one order of magnitude. The analyses of magnetic
emission signals recorded during brittle and ductile failure showed that the value of the exponent
of the power law regime is characteristic for the failure mode: an exponent significantly higher
than for brittle failure characterizes ductile failure. The distribution of the peak areas has a similar
overall character, however, the power law regime spans practically two orders of magnitude and
the area distribution is universal in the sense that it does not change with the impact velocity: it
depends solely on the failure mode.
102 Gy. B. Lenkey et al.

FIGURE 1. Area distribution of peaks p(A) for different impact velocities and failure modes.

References
1. Gy. B. Lenkey, S. Winkler.: Fatigue and Fracture of Engineering Materials and Structures,
Vol. 20., No. 2., pp. 143-150., 1997
2. Equation Section 1 Gy. B. Lenkey, L. Tóth: Mat.wiss. und Werkstoffmech., 32, 1-6, pp. 1-6.,
2001
3. Gy. B. Lenkey: ASTM STP 1380, T. Siewert and M. P. Manahan, Sr., Eds., American Society
for Testing and Materials, West Conshohocken, PA, pp. 366-381., 1999
4. M. C. Miguel, A. Vespignaid, S. Zapperi, J. Weiss and.J. Grasso, Nature 410, 667, 2001
5. A. Maes, C. Van Moffaert, H. Frederix, and H. Strauven, Phys. Rev. B 57, 4987, 1998.
6. A. Guarino, S. Ciliberto, A. Garcimartin, M. Zei, and R. Scorretti, Eur. Phys. Jour. B 26, 141,
2002
7. R. A. White and K. A. Dahmen, Phys. Rev. Lett. 91, 085702, 2003
8. L. I. Salxninen, A. I. Tolvanen, and M. J. Alava, Phys. Rev. Lett. 89, 185503, 2002
 2T1. Physical aspects of fracture 103

ELECTROMAGNETIC RADIATION METHOD FOR IDENTIFICATION OF

MULTI-SCALE FRACTURE

Yu. K. Bivin, A. S. Chursin, E. A. Deviatkin and I. V. Simonov

Institute for Problems in Mechanics of RAS
Vernadskogo av. 101-1, 119526 Moscow, Russia
simonov@ipmnet.ru

The electromagnetic radiation (EMR) as well as acoustic emission (AE) is the statistical
phenomenon in fracture. Measurements of EMR can serve the purpose of identification. The
experimental data on this point are known for ion crystals, metals, rocks. The explanation of
physical mechanism has been given ideally, for example, in the case of shock wave in an “ideal”
metal.
Aim of the present work is to develop this promising method of EMR and to show its
possibilities for registration of multi-scale fracture mainly within non-conductors, structural
materials. A review of experimental results includes the following events: the fast micro-fracture
in ice, the single crack propagation in glass, the velocity decay during ball penetration in a ground
medium, the single free fiber break. The EMR signals were recorded by an antenna over radio
range of frequency spectrum. It is noteworthy that they reflect adequately the mentioned dynamic
fracture processes. So, when a single crack propagates with a constant or varying velocity, the
EMR signal behaves in like manner. Moreover, the high frequencies appear also due to micro-
cracking on the crack surfaces, and even their vibrations with low frequencies after the crack stop
is designed clearly on the oscillogram.
Series of tests on edge splitting of plates has been also conducted. Sizes of specimens from
polymetilmetacrilate (PMMC), glass-epoxy (GE) and carbon-carbon composite (CCC) (the last
two are the unidirectional fiber-reinforced composites) were approximately uu
mm³. The artificial initial 10-15 mm length cut along the central plane was sharpened. Each
specimen was loaded step by step with the incremental normal force on the split plate edge. The
determination of fracture parameters was performed by a compliance calibration technique, in
particular, for the purpose of validation of the tests. The theoretical predictions base on the
principles of linear fracture mechanics and the two conventional methods are demonstrated in the
tests. To determine the fracture toughness and Young’s module, the limit load and edge
displacement were measured at the crack start. These values are in agreement with the tabular
ones. Another concerns the calculation of new crack surface and the loss of work per the "loading-
unloading" cycle on the "force-displacement" quarter-plane during a crack jump. Then the energy
release rate can be determined. The acoustic emission and the outward electric field near to crack
tip were simultaneously recorded.
The load and displacement as a mechanical field, along with the AE and EMR, say, the
physical fields, exhibit different feature before the crack start and during its galloping motion. The
AE and EMR appear at the sharp increasing the load when a damage (micro-cracking, fiber break)
apparently occurs nearby the stationary crack front. The downward load jump corresponds to the
crack start and a maximum of the physical fields activity. The analysis of correlation of these
fields and comparison their amplitude and frequencies for different structures with each other are
the new elements in the failure/fracture. It was revealed that the GEC gives maximum the AE and
EMR activity among the materials tested. This reflects the intense micro-cracking of GEC which
has been proved by visual inspection. The minimum accounts for the CCC that is rather more
“perfect” structure than GEC. The EMR pulses turn out to possess the good sensitiveness and
104 Yu. K. Bivin et al.

correlate with AE data in the main. But the electric activity minimum in the case of CCC is also
explained by good carbon conductivity.
It can be concluded that the EMR record is the very simple, non-destructive, temperature
independent, and cheaper method for detection of different dynamic fracture stages into structural
materials and determination of limit loads. To use this method in combination with the AE
technique will lead to essential increasing reliability for the event identified, especially, in the
cases when the noise influence is expected. The future experimental data will also be presented.
This work was partly supported by the Program N13 of the Russian Academy of Sciences and
the Russian Foundation of Basic Researches, N 05-01-00628.
 2T1. Physical aspects of fracture 105

MICROMECHANICAL MODELING OF GRAIN BOUNDARY RESISTANCE TO

CLEAVAGE FRACTURE PROPAGATION

M. Stec and J. Faleskog

KTH Solid Mechanics, Royal Institute of Technology
Osquars Backe 1, SE-100 44 Stockholm, SWEDEN
mateusz@hallf.kth.se, jonasf@hallf.kth.se

Failure of polycrystalline structures due to cleavage fracture can be summarized in three critical
steps: (i) fracture of a brittle particle, e.g. a carbide, cause rapid growth of a micro-crack, (ii) which
must generate sufficient energy release to propagate into the much tougher surrounding matrix
material. In order to become critical, (iii) the micro-crack must propagate over grain boundaries
into neighbouring grains. An initiated microcrack may arrest at several instances. In a
micromechanical analysis by Kroon and Faleskog [1] step (ii) was examined and they found that a
micro-crack can arrest before the first grain boundary is reached. Based on a weakest link concept
Anderson et al. [2] studied cleavage crack growth across hexagonal grains and showed that grain
boundary resistance have a direct influence on the fracture toughness threshold. Moreover, they
also showed that if the microcrack arrests, it will most likely arrest at the first grain boundary
encountered. In this study the conditions for a microcrack to arrest at the first encounter with a
grain boundary is investigated using micromechanical analysis.
The cleavage planes in ferritic steels is one of the three {100} planes (Miller index notation).
At a grain boundary, the orientation of cleavage planes typically changes, as illustrated in Figure
1(a), where the cleavage crack has propagated from grain A into grain B. The relative difference in
orientation between adjacent cleavage planes can be characterized by two angles: a tilt-angle and
a twist-angle , see Figure 1(a). A grain usually has about 5-6 neighboring grains. The distance
along the grain boundary, where grain A meets primary cleavage planes of grain B, is here on the
average taken as w, see Figure 1(c). Experimental estimations of w can be found in a series of
fracture tests carried out by Qiao and Argon [3]. The micromechanical analysis is in the present
study inspired by the experimental observations made by Qiao and Argon [3] and based on an
idealized model of two grains, where grain A is modeled as a circular cylinder entirely surrounded
by grain B, see Figure 1(b). Furthermore, the primary cleavage planes in grain A and grain B are
assumed to intersect each other with a periodicity of w, as is illustrated in Figure 1(c). In order to
understand to what extent the grain boundary can act as a barrier and obstacle, an effective energy
release rate for a microcrack that has propagated to radius R ( t R G B ) an be formulated as

w
*e K 2 * A  (1  K 2 ) * B f B (\ , M )  K 2 * GB f GB (\ , M )
R GB . (1)

Here, *A , *B and *GB are critical energy release rates associated with grain A, grain B, and the
grain boundary, K RGB R (see Figure 1(a)), and f B and f GB are functions of the
misalignment angles ( ȥ , ij ) and is of order unity.
Both grain A and grain B are modeled as elastic viscoplastic materials. The crack growth is
modeled using cohesive surfaces, where the tractions are governed by a modified exponential
cohesive law in order to control the initial slope. The micromechanical model was numerically
analyzed by use of finite element modeling. The analysis consisted of two phases. In the first one,
true (Cauchy) tractions were applied quasi-statically on the remote boundary (see Fig. 1(b))
106 M. Stec and J. Faleskog

corresponding to the axial and radial stresses Ȉ Z and Ȉ R , respectively. In the following dynamic
phase both Ȉ Z and Ȉ R were held constant. The cleavage crack was then initiated in grain A and
started to grow in the radial direction until it encountered the grain boundary. The crack plane
either penetrated the grain boundary and continued into grain B or arrested. The main question in
the analysis was: given the overall stress state—how large can grain A be in order for the cleavage
crack not to penetrate grain B and possible cause catastrophic failure. The critical stress state
required to propagate the cleavage microcrack across grain boundary into grain B and further
depends primarily on: stress ratio [6R 6Z ] and grain boundary features
[ ȥ , ij , w * G B ( R G B * B )].It should also be pointed out that the characteristics of the cohesive
surfaces and the elastic viscoplastic behavior affects the results.

Figure 1. Grain boundary model characterization: (a) orientation of cleavage planes, (b) idealized
micromechanical model and (c) crack plane definition at grain boundary.

References
1. Kroon M., Faleskog J., J. Mech. Phys. Solids, vol. 53, 171-196, 2005
2. Anderson T., Stienstra, D. and Dodds, Jr. R. H., In Fracture Mechanics: 24th volume, ASTM
STP 1207, edited by J.D. Landes, et al. ASTM, Philadelphia, 1994, 186-214.
3. Qiao Y., Argon A.S., Mech. Mat., vol. 35, 313-331, 2003.
 2T1. Physical aspects of fracture 107

MICROSTRUCTURE OF REACTOR PRESSURE VESSEL STEEL CLOSE TO

THE FRACTURE SURFACE

M. Karlik, P. Hausild and C. Prioul1

Czech Technical University, Faculty of Nuclear Sciences and Physical Engineering,
Department of Materials, Trojanova 13, 120 00 Praha 2, Czech Republic
1
Ecole Centrale Paris, LMSS-Mat,Grande Voie des Vignes, 92295 Châtenay–Malabry, France
Miroslav.Karlik@fjfi.cvut.cz, Petr.Hausild@fjfi.cvut.cz, Prioul@mssmat.ecp.cz

A transmission electron microscopy study of microstructure in the vicinity of fracture surface was
carried out in French tempered bainitic nuclear reactor pressure vessel steel 16MND5 (equivalent
to the American standard A508 Class 3). Cross-section thin foils from nickel electroplated Charpy
V-notch specimens fractured in impact and quasi static loading at -30°C (ductile to brittle
transition temperature region) were prepared in the ductile tearing and cleavage regions of the
fracture surface (Fig. 1).

FIGURE 1. Slice cutting in the central part of a Charpy V-notch sample and the TEM disc
positions in cleavage – C and ductile – D fracture zones.

In the ductile tearing zone, the microstructure was very heterogeneous. Dislocation cells, shear
bands, and fine heavily deformed subgrains were found. The deformation in the impact specimen
was often localized only in the vicinity of the fracture surface, where long thin cells formed due to
dynamic recovery (Fig. 2). In the quasi static three-point bend specimen, the localization was
found also in deeper areas under the fracture surface. There were shear bands (bundles of long thin
cells) mostly aligned in parallel to the fracture surface. Numerous areas of the shear band
intersections (at ~45° to the main shear band direction) were also observed.
Comparing the microstructure in the ductile tearing zone of the Charpy impact specimens with
the corresponding part of the fracture of the CT25 fractured in quasi-static loading in previous
research [1], there are two similar features. The deformation is very heterogeneous - heavily
deformed material with very small and dislocation arranged in cells. On the other hand, there were
no twins, neither shear bands found in Charpy specimens. The shear bands were transformed by
dynamic recovery into long thin subgrains.
108 M. Karlík et al.

FIGURE 2. Microstructure of the steel in the ductile tearing region - very thin subgrains (50 to 100
nm) formed due to the local increase in temperature leading to dynamic recovery of dislocations
during rupture of impact specimen.

In the cleavage zones of both types of specimens, only an increased dislocation density was
found, no twins were observed. The reason for the absence of twins could be that the material of
the much smaller Charpy specimen was plastically deformed (bending of the test specimen) before
the cleavage crack propagation and the twins could not form due to an increased density of
dislocations.

References
1. Karlík, M., Nedbal, I., Siegl, J., Mater. Sci. Eng. A 357, 2003, 423-428.
 2T2. Brittle fracture 109

BRITTLE FRACTURE IN HEAT-AFFECTED ZONES OF GIRTH WELDS OF

MODERN LINE PIPE STEEL (X100)

A. S. Bilat, A. F. Gourgues-Lorenzon, J. Besson and A. Pineau

Centre des Matériaux, Ecole des Mines de Paris,
UMR CNRS 7633, B.P.87, F-91003 Evry cedex
anne-sophie.bilat@mat.ensmp.fr, anne-francoise.gourgues-lorenzon@ensmp.fr

Gas field development requires a cost reduction of gas transport. For that purpose, one solution is
to increase gas pressure inside pipes. High strength steels of API grade X100 (yield stress above
690 MPa) are potentially good candidates for these new applications [1], but the toughness of
X100 welded joints involves characterization. The aim of this project is to improve the accuracy of
prediction tools of the toughness and defect acceptability in such a GMAW welded joint, by
building a quantitative prediction tool of the risk of brittle fracture, adapted to the girth welds of
new X100 steels, and by determining and validating a crack initiation criterion in the most critical
zones of this welded joint. This study is conducted according to the local approach to fracture
procedure [2]: first, metallurgical and mechanical characterization of the girth weld, then, thermal
welding cycle simulation of the most critical zones and characterization of their fracture properties
and, finally, finite element modelling of brittle fracture initiated at these critical zones.
Hardness and microstructures were thoroughly characterized in each zone (Fig. 1). The weld
consists of three main zones: the base metal (BM), the weld metal (WM) and the Heat-Affected
Zone (HAZ). The BM microstructure is a textured bainitic matrix with ferrite and some secondary
phases; the WM is acicular ferrite and the Heat-Affected Zone is upper bainite. The HAZ is itself
divided in three zones: the coarse-grained (CG)HAZ (austenite grain size around 30 µm, spread
over 100 µm) near from the WM, the fine-grained (FG)HAZ (austenite grain size around 10 µm,
spread over 100 µm), and the intercritical (IC)HAZ (MB tempered between Ac1 and Ac3 spread
over 1 to 2 mm).
The mechanical strength of the joint was characterized by using tensile and impact tests (Fig.
2). Specimens were cut from the BM, and the WM, but the HAZ is too small to machine
homogeneous specimens, so that it is necessary to reproduce some HAZ microstructures. Two
microstructures (CG and FG) are reproduced by applying a simulated weld thermal cycle on base
metal blanks with a thermal-mechanical simulator (Gleeble). The welding thermal cycles are
determined experimentally by inserting a thermocouple in the heat affected zone of the joint during
a real welding operation and the Rykaline model [3]. All these specimen are then machined and
mechanically tested in tension at various temperatures. The weld is matched in yield stress
(YSWM=YSBM), and overmatched in tensile strength (TSWM>TSBM), because the base metal of
the pipe exhibits little work-hardening. The impact tests show low fracture energy at –20 °C, when
the specimen is taken from the inner subsurface part of the pipe wall and the notch is located on the
fusion line (FL i). The brittle fracture of these specimens is systematically localized in the CG
zone, which consists of upper bainite with Martensite-Austenite constituents.
The mechanical properties of the various zones of the weld will be used in a numerical finite
element model together with a statistical brittle fracture criterion to describe the behaviour of this
“multi-material” assembly.
110 A. S. Bilat et al.

Acknowledgments: Technical support and constant interest from Serimer-Dasa, Gaz de France
and Europipe as well as financial support from CEPM are gratefully acknowledged.

References
1. Glover, A., In Proceedings of the international Pipe Dreamer’s Conference, co-organized by
M.Toyoda and R. Denys, Yokohama, Japan, 2002, 33-52.
2. Besson, J., Local approach to fracture, Presses de l’Ecole des Mines, Paris, France, 2004.
3. Rykaline, N. N., Calcul des processus thermiques de soudage, Soudage et Techniques
connexes, 1961 (in French)
 2T2. Brittle fracture 111

CLEAVAGE FRACTURE OF STEELS AT VERY LOW TEMPERATURES

R. Rodriguez-Martin, I. Ocana and A. Martin-Meizoso

CEIT, Centro de Estudios e Investigaciones Técnicas de Guipúzcoa and
TECNUN, Escuela Superior de Ingenieros, University of Navarra
Paseo Manuel de Lardizábal, 15, 20018 San Sebastián, Spain
rarodriguez@ceit.es, iocana@ceit.es, ameizoso@ceit.es

The mechanical strength of polycrystalline metals is frequently associated with the propagation of
defects present in their structure as cracks or dislocations. Nevertheless, when these defects are
absent, the crystal lattice becomes decisive and the strength is determined by the stress at which the
lattice losses its stability. This stress is known as the ideal strength, Clatterbuck et al. [1], Krenn et
al. [2], and it is especially relevant in experimental situations where there are few mobile defects.
This paper proposes that, in bcc metals, brittle fracture at very low temperatures is caused
when an ideal cleavage strength is reached in their structure. At such low temperatures, the
movement of the defects is frozen and the material will behave as if it were free from defects.
Under these conditions, a common bcc metal would cleavage on (100) planes under an uniaxial
tensile stress. The mechanism proposed lies in the initial breakage of the grain having the best
oriented (100) plane with respect to the applied load. Subsequently, other grains with a favourable
orientation take its load and may break, causing the macroscopic failure of the test-piece or
component. Figure 1 shows an example of a random grain distribution with their respective
orientation for their (100) planes.

FIGURE 1. Schematic of a fatigue pre-cracked body with a random grain distribution at the fatigue
crack front. The orientation of a (100) plane is showed for some grains.

The value of the ideal cleavage strength on (100) has been investigated for many bcc metals on
the basis of ab-initio electronic structure calculations, [1-2], Friak et al. [3]. From these
calculations, cleavage fracture of the high strength steels of which chemical compositions are
listed in Table 1, has been analysed.
112 R. Rodriguez-Martin et al.

TABLE 1. Chemical composition of steels (%).

Previously precracked compact tensile specimens extracted from these materials have been
used in fracture tests. These tests were conducted at very low temperatures (liquid nitrogen).
Subsequent fractographic analyses showed the presence of multiple initiation sites situated close to
the fatigue crack in the fractured specimens of all the materials. Also a low-scatter in the obtained
results suggested that a similar mechanism causes the brittle failure for the three materials studied.
Taking into account all these experimental observations, the possibility of ideal cleavage
strength to control the brittle fracture makes sense. In order to predict this fracture behaviour at low
temperatures, a simulation program has been developed in Matlab£ code. The program generates
a random grain distribution for a precracked body. The crack is assumed to blunt under load,
following the behaviour proposed by McMeeking [4]. From this point forward, the program
calculates the applied load required for the failure of the random grain distribution by cleavage.
This is achieved by taking into account that the ideal cleavage strength is reached in the grain with
the best oriented (100) plane and subsequently transmitted to other grains with a favourable
orientation.

References
1. Clatterbuck, D.M., Chrzan, D.C. and Morris, J.X., Acta Mater., vol. 51, 2271-2283, 2003.
2. Krenn, C.R., Roundy, D., Morris, J.W., Cohen, Marvin L., Mat. Sci. Eng. A319-321, 111-114,
2001.
3. Friak, M., Sob, M. and Vitek, V., Phil. Mag., vol. 83, 3529-3537, 2003.
4. McMeeking, R.M., J.Mech.Phys.Solids., vol. 25, 357-381, 1977.
 2T2. Brittle fracture 113

NEW FORMULATION OF THE RITCHIE, KNOT AND RICE HYPOTHESIS

A. Neimitz, M. Graba and J. Galkiewicz

Kielce University of Technology
Al. 1000 lecia P.. 7, 25-314 Kielce, Poland
neimitz@tu.kielce.pl

Several authors adopted a local fracture criterion to assess the constraint influence on fracture
toughness. They assumed that the cleavage fracture to happen requires that the opening stress
reaches critical value, Vc at certain distance from the crack tip, rc or within the certain volume in
front of the crack tip.
O’Dowd at al [1] adopted modified small strain HRR solution, henceforth called the OS
model, to derive simple, approximate formula in order to predict the influence of in-plane
constraint on fracture toughness of structural element. According to O’Dowd’s model the critical
conditions must be satisfied independently the level of constraint, which was quantified by the
actual value of the Q-stress, utilizing the OS theory.
O’Dowd’s final formula to compute the actual value of fracture toughness is as follows:
n 1
§ Q ·
JC J IC ¨¨ 1  ¸¸
© V c /V 0 ¹ (1)
where J is the J-integral, Vo is a yield stress, subscript c or Ic denotes critical state, Q is a parameter
“measuring stress level in the modified OS model [1]
Eq. 1 follows from the Ritchie, Knot, Rice (RKR) hypothesis provided one assumes that the
stress field in front of the crack is singular (small strain assumption). In such a case the opening
stress is greater than critical one over the distance rc from the crack front. However, the critical
stress Vc is simply the parameter which adjusts theoretical prediction to experimental results.
Almost all other theories concerning the local approach to fracture use the stress distribution
in front of the crack characteristic for the finite strain. In real elastic-plastic materials the crack tip,
originally sharp, is blunted by plastic deformation. This process modifies the small-strain HRR
solution in front of the crack. Instead of singular stress field at the crack tip the opening stress
decreases towards the blunted crack tip after reaching a maximum at the normalized distance
r=MJ/V0 from the crack tip. Here, M is a function of constraint level, and the Ramberg-Osgood
power exponent n at given external loading.
In such a case the RKR hypothesis should be based on another assumption.
Hypothesis: It is assumed that the cleavage fracture may happen if the opening stress in front
max
of the crack exceeds the critical level over the distance l greater than lcrit ( V 22 >Vc and ltlc ).
max
If V 22 >Vc but l<lc fracture may proceeds due to void’s growth and coalescence, provided
the plastic strains are sufficient large. From simple geometrical consideration it follows:

V max  V c
l # 2 rmax
V max  EV 0 (4)
where EVo is equal to the opening stress component at the blunted crack tip.
114 A. Neimitz et al.

In order to apply this model to estimate fracture toughness at arbitrary in-plane constraint level
we assume that lc is characteristic length for material of interest.
To apply proposed hypothesis detailed numerical investigations of the stress field in front of
the blunted crack have been performed. They lead to general observations:
The maximum, value of the opening stress component saturates to a constant level which does
not depend on external loading and depends weakly on the in-plane characteristic specimen size a/
W. It depends on the deformation properties of the material, n and V0/E. It happens when plastic
zone covers an essential part of the ligament in front of the crack tip. However, for a very short
crack, e.g. a/W=0.05 or less the maximum stress at the crack front becomes more sensitive to the
change of the a/W ratio. In this case the maximum value decreases slightly with decreasing a/W.
The maximum, saturated value of the opening stress is more sensitive to the change of the a/W
ratio for the larger values of the n exponent.
The stress maximum is located at the distance r=MJ/V0 from the crack tip. It can be shown that
the coefficient M is a function of n, V0/E and the in-plane constraint measure Q when the maximum
stress saturates to a constant value. In fact, this function does not change with external loading but
changes with the ratio a/W. When the stress maximum still changes with external loading the
max
coefficient M depends on the V 22 also.
Basing on above observations the analytical formulas were derived to assess the influence of
the in-plane constraint on fracture toughness of structural element. One of them which satisfies
deformation pattern in front of the crack in the Sumpter and Forbes experiments [2] is as follows:
1 n
ª 1 1 º
«§ 1 · 1 n Q ª E º 1 n »
JC J IC «¨ ¸  ~ « » » MQ 0
«¨© M Q 0
¸
¹ V 22 ¬ DV 0 I n ¼ »
«¬ »¼
(5)
or
(1 n )
§ Q ·
JC J IC ¨¨1  max ¸¸
© V 22 / V o ¹ (6)
max
Where MQ=0, or V 22 must be computed numerically. Above equations fit experimental results
very well. More details of the numerical results as well as another equations will be presented in
the paper and presented during the conference.

References
1. O’Dowd, N.P. and Shih, C.F., J. Mech. Phys. Solids, vol. 49, 939-963, 19
2. Sumpter, J.D.G. and Forbes, A.T., Black, P.A., In Proceedings of TWI/EWI/IS Int. Conf.
Shallow Crack Fracture Mechanics Tests and Applications, Cambridge, U.K., paper 7, 1992.
 2T2. Brittle fracture 115

THE EFFECT OF THE RATE OF DISPLACEMENT ON CRACK PATH

STABILITY

D. A. Zacharopoulos and P. A. Kalaitzidis

School of Engineering
Democritus University of Thrace
GR 671 00, Xanthi, Greece
dzachar@civil.duth.gr, pkalait@civil.duth.gr

The search for methods for forecasting the crack path and its type, finally became an issue of
particular interest due to the increasing use of advanced materials like composites or coated
materials (substances) for improvement of their thermomechanical properties. With the increasing
use of structural adhesives in construction, aerospace, and automotive industries the need for an
estimate of the locus of failure and the crack path propagation is essential in ordrer to improve the
durability of the bonded joints. In addition, the prediction of the crack path can be beneficial in the
design of safe structures and gives answers on the possible initial conditions of loading in the case
of a destructive fracture.
We will try to clarify the concepts of ‘stable’ and ‘unstable’ crack path in the fracture of a solid
or a structure.
Part of the path of the propagated crack can be characterized as ‘stable’ if and only if this part
resulting from repeated experiments onto bodies with the same geometry and under the same
loading conditions, appears with identical geometrical characteristics. Specifically, when a stable
crack path situation prevails at the spread of a fracture in specimen and leads to the catholic
destruction of the specimen, the corresponding broken pieces have the same shape.
The verification of a theory or a method for the approach of a problem is usually based on the
experimental process. However, if the results are exceptionally sensitive on the initial and on
general conditions that exist during the experiment, then a scattering of the results is very likely.
The evaluation of this scattering in phenomena of "instability" can lead to erroneous conclusions.
In other words, in each experimental process we should compare the scattering of the results owed
to the problem’s endogenous instability factors with the magnitude of the experimental conditions’
divergence, like the constrained displacements, or the geometry of the specimens. Observations on
experimental results show that the propagation trajectory of an extended crack depends on the
material properties, the geometry of the specimen, the rate of loading or displacement , the
dynamic loading and the temperature. Furthermore, the control of the load’s increase or the
displacement’s increase on the specimen by the loading machine also plays an important role. On
the interface, the propagation trajectory also depends on the tensile strength and fracture toughness
of the bonded materials. Approaching the multifactor problem of stability for a part of a crack path
we could verify that all the above factors participate in the configuration of a situation that will
determine the stability of the crack path, adding the history of the crack path and the magnitude of
the applied forces.
A more developed study on the method, determining the crack path stability can be found at
[1]. For any two-dimensional problem with an arbitrary geometry of the body and arbitrary
constraints, the stress and strain fields can be calculated by developing the displacement and
loading results, by using a finite elements program. Using that data, we can draw the contour map
of strain energy density with a graphical program. This map has common features with a
geographic map: the point where the failure begins, according to the basic hypotheses, is the hilltop
116 D. A. Zacharopoulos and P. A. Kalaitzidis

of the topographic map respectively and it is surrounded from closed contours that may have U
shape. The higher contours are always enclosed by lower ones.
In the present work, we will demonstrate how we can estimate the crack path stability with the
elaboration of this map, when a very slow rate of displacement is imposed on the specimen. When
the critical locus on the plane body is the point O, which can be a crack tip, according to the above
hypotheses, the beginning, the initial direction of crack growth as well as the crack path of
propagation will occur. According to the strain energy density criterion (see [2,3]), the initiation of
r >
fracture from the O, takes place along the direction OL, (OL)= c , when the dW / dV max min L @
reaches the critical value (dW/dV)c. The predicted crack path during propagation is the curve that
starts from point L passes the points with the maximum gradient of (dW/dV) and ends up at point
G, where the global minimum value of (dW/dV) develops.
On the map, the crack path is indicated by V shaped contours of the strain energy density. If
the apicals of the Vs points are joined by a line, then the resulted plot curve, starts from the peak O
and arrives in the vicinity of the point G. In the geographic map the curve OG, represents a gorge
or a riverbed, which starts from the hilltop O. The drawing of this gorge can give additional
information for the estimation of the crack path’s stability according to following hypothesis:
The stability of the crack path can be deduced from the degree of the sharpness with which the
curve of the "gorge" is drawn.
By collecting the results for different geometric characteristics of the specimens, we create a
classification diagram of crack path stability.
Our experiments were performed on double cantilevering beam (DCB) specimens made from
PMMA material. The tests were performed by using a testing machine with controlled
displacement at a rate of 0.5mm/min or less. These slow rates were chosen to reveal the specimens
behavior while transiting across regions with different crack path stability of the classification
diagram.

References
1. Zacharopoulos, D.A., “Stability Analysis of Crack Path Using the Strain Energy Density
Theory”, Theoretical and Applied Fracture Mechanics, vol. 41, 327-337, 2004.
2. Sih G.C., “Introductory chapters in Mechanics of Fracture”, Vols. I to VII, ed. G.C. Sih,
Martinus Nijhoff: The Netherlands, (1972-1982).
3. Gdoutos, E.E., “Fracture mechanics”, Kluver Academic Publishers, 1993.
 2T2. Brittle fracture 117

SCRATCHING AND BRITTLE FRACTURE OF SEMICONDUCTOR IN-SITU

SCANNING ELECTRON MICROSCOPE

K. Wasmer, C. Pouvreau1, J. Giovanola1 and J. Michler

Swiss Federal Laboratories for Materials Testing and Research (EMPA)
Feuerwerkerstrasse 39, 3602 Thun, Switzerland
1
Swiss Federal Institute of Technology Lausanne (EPFL), ME A1 400, Station 9, 1015 Lausanne,
Switzerland
kilian.wasmer@empa.ch

Mastering the brittle facture (also called cleavage) of semiconductors is important when separating
complex devices produced on epitaxially grown multi-layers or when processing III-V optical
devices. Cleavage is usually performed in two stages. The surface is first micro-scratched with a
sharp diamond tip, which induces structural defects. Secondly, by applying a load to the sample,
fracture propagation is initiated, starting from the defect area. However, the mechanisms leading to
the preparation of high quality cleavage surface are often not fully understood. The main objective
of this paper is to present the fundamental phenomena occurring during the scribing and
subsequent fracturing process of brittle semiconductors via in-situ Scanning Electron Microscope
(SEM).
In the literature, nano-indentation has been largely investigated also for semiconductors,
Bradby et al. and Grillo et al. [1-2], but very little information is available in comparison in the
field of scratching, Ballif et al. and Gassilloud et al. [3-4]. In this paper, the complete sequence of
scratching is dissected via in-situ SEM observations. In order to perform these observations, a
miniaturized microscratch device has been built for use inside a scanning electron microscope,
Rabe et al. [5]. Also performed are Transmission Electron Microscope (TEM) investigations to
visualise the network of dislocations. It has been found that the scratching operation can be divided
into 5 distinct regions which are dependent of several parameters such as the shape of the diamond
tip, the scratching direction and the applied loads. The first one is the elastic regime which cannot
be observed in normal condition due to the total recovery after the withdrawal of the load. The
second is the plastic regime where dislocation nucleation and / or slip bands are generated. This is
followed by the subsurface cracking regime where median cracks (MC) along crystallographic
planes are created. Then, cracks such as radial and lateral appear at the surface. The last regime
takes place when the radial and lateral cracks join together since at this point a large amount of
chipping out occurs. In order to obtain atomically flat cleavage surfaces, only the median crack,
which depth will influence significantly the crack initiation and propagation, is required. Based on
our experiments, it has been also seen that the depth of the MC can be assumed to be either linear
or to the power 3/2 to the applied load as described by Eq. (1). This latter expression has already
been reported to be valid for the nano-indentation Lawn [6]. Furthermore, the influence of the
different epilayer on GaAs wafer is perceived to influence the onset of the crack median as well as
its depth for loads ranging between 10mN and 50 mN.

(a) P v c ˜ d and (b) P v c ˜ d 3 / 2 (1)

The second part of the paper scrutinise the fracture process in brittle semiconductors such as GaAs.
This includes the crack initiation of a defect in GaAs wafer, GaAs + epilayer wafer and laser diode
processed on GaAs wafer with a thickness of around 150 m. The defect is assumed to be semi-
elliptic with a length of 400 m and a depth in ranging from 5 to 20 m depending on the applied
load. The wafer is then subjected to bending stress through two apparatus allowing in-situ cleavage
118 K. Wasmer et al.

of devices. The first one, integrated into the SEM chamber, has a stroke of several hundred of
micrometers and is "displacement-controlled". Its displacement control is better than 100 nm and
the force applied by the cleaving rod is acquired simultaneously. The second apparatus is very
similar to the first one with a major advantage which is to be able to follow the crack during its
propagation. With these tools, it has been demonstrated that although, semiconductors such as
GaAs are considered as very brittle, it is possible to control the crack propagation when a big
enough starter crack is present. For uncontrolled crack growth, features are visible on the cleaved
surface which is consistent with the dynamic fracture mechanics principle Sauthoff et al. [7].
Finally, the effects due to the different epitaxial layers are in line with those observed on the depth
of the median crack.

FIGURE 1: SEM pictures of a (a) scratch with a Cube Corner diamond tip; b) cleave showing
crack deviation

References
1. Bradby, J. E., Williams, J. S., Wong-Leung, J., Swain, M.W., MV, and P. Munroe, J. Mater.
Res. Vol. 16, pp: 1500-, 2001.
2. Grillo, S. E., Ducarroir, M., Nadal, M., Tourni, E. and Faurie, J.-P., J. Phys. D: Appl. Phys.
Vol. 36, pp: 5-, 2003.
3. Ballif, C., Wasmer, K., Gassilloud, R., Pouvreau, C., Rabe, R., Michler, J., Breguet, J.-M.,
Solletti, J.-M., Karimi, A. and Schulz, D. Advanced Engineering Materials, June 2005.
4. Gassilloud, R., Ballif, C., Michler, J. and Schmuki, P., Submitted at Journal of Material
Research.
5. Rabe, R., Breguet, J.-M., Schwaller, P., Stauss, S., Patscheider J. and J. Michler, Thin Solid
Films, Vol. 469-470, pp: 206-213, 2004.
6. Lawn, B., "Fracture of Brittle Solids", 2nd Ed., Cambridge University Press, 1997.
7. Sauthoff, K., Wenderoth, M., Heinrich, A. J., Engel, K. J., Reusch, T. C. G. and Ulbrich, R.
G., Phys. Rev. B, Vol. 60, pp: 4789-95, 1999.
 2T2. Brittle fracture 119

CRACKS IN THIN SHEETS: WHEN GEOMETRY RULES THE FRACTURE

PATH

Pedro M. Reis, Basile Audoly1 and Benoit Roman2

Levich Inst., City College of New York,
140th St. & Convent Av., New York, NY 10027, USA
1
LMM, UMR 7607 CNRS/UPMC, 4 place Jussieu, case 162, 75252 Paris cedex 05, France
2
PMMH, UMR 7636 CNRS/ESPCI, 1O rue Vauquelin 75231 CEDEX 5 Paris, France
preis@levdec.engr.ccny.cuny.edu

We study both experimentally and theoretically the propagation of brittle fracture coupled to large
out-of-plane bending, as when a brittle elastic thin sheet is torn by a rigid moving object [1-3].
Taking into account the separation of the film’s bending and stretching energies and using classical
fracture theory we have shown that such cracks can propagate according to a simple set of 2D
geometrical rules in the limit of vanishing thickness, h [3]. Numerical integration of our
geometrical model accurately reproduces both the shape of the fracture pattern and the detailed
time evolution of the propagation of the crack tip.
In our experiments, a rigid object, the cutting tool, is forced through a thin polymer film and
tears through the material as it advances. A schematic diagram of the process is presented in Fig.
1(a,b). The film is clamped at its lateral boundaries imposing no initial tension. The cutting tool is
oriented perpendicularly to the horizontal film and is driven through the material at constant
velocity, v, in a direction parallel to the film’s major length. We have used biaxially-oriented
polypropylene (BOPP) and acetate films (25<h<127Pm) which, being brittle, undergo negligible
irreversible deformation besides fracture.

FIGURE 1. Schematic of an oscillatory path, P, obtained when a cutting tool, I, is driven at

constant velocity through a thin polymer film clamped at two of its lateral boundaries. L and A are
the wavelength and amplitude of the pattern. (b) 2D projection of the cutting process. (c)
Photograph of an experimental crack path. 30mm BOPP film, v=1.6mms-1 , rectangular cross
section tool (w=6.35mm). Videos of the fracture process can be found in [4].

This setup yields surprisingly complex crack propagation: for tools considerably wider than
the film's thickness the crack tip, T, follows a highly reproducible non-sinusoidal oscillatory path
as shown in Fig. 1(c). Each single period of this path consists of two smooth curves separated by a
120 P. M. Reis et al.

kink, at which there is a sharp change in the direction of curvature. Propagation is primarily quasi-
static but is interrupted by periodic bursts of dynamic propagation immediately after each kink. By
decreasing the size of the cutting tool down to widths comparable to the film thickness, the crack
path eventually becomes straight, as reported in [1], but here we focus on the oscillatory behaviour
far above threshold. In this regime, the crack morphology is independent of v (as long as this
remains much smaller than the speed of sound in the material), h (provided h>>w) and the film's
width, D. The only relevant lengthscale in this scenario appears to be the width of the cutting tool,
w, with which both the pattern's amplitude, A, and wavelength, O, scale linearly [1].

FIGURE 2. Motion of the crack tip from the model compared to experiments for a cylindrical
cutting tool (w=20.4mm, h=50mm, v=1.2mms-1). (a) Time-series for the crack tip coordinates x(t)/
w and y(t)/w as a function of dimensionless tool advance vt/w. Note the periodic gaps in crack
position, corresponding to dynamic propagation. (b) Final pattern in the film plane (x,y).

We shall show that the classical equations for elastic plates and for crack propagation can be
reduced to a simple set of 2D geometrical rules, which explain the experimentally observed crack
behaviour. Following a common procedure in fracture theory our approach is as follows: i) we
first calculate the elastic energy of the system, taking into account the possible large out-of-plane
deformations of the film induced by the cutting tool, ii) we then apply Griffith's criterion for crack
propagation and iii) finally establish the direction of propagation of the crack tip, E tip according
to the principle of local symmetry. In our model, the mechanical properties of the film are captured
by the two physical parameters, D >* /(EL)@ and EaSwhere * is an effective toughness of
the film (dimensions of surface tension), E is the Young’s Modulus and L is a typical length scale
near the crack tip. We have numerically integrated our geometrical model and have found
excellent agreement with experiments, as shown in Fig. 2. Both the kinks and the subsequent
bursts of dynamic propagation arise from a simple set of rules and are therefore intrinsic feature of
the tearing process. To our knowledge this is the first example where a complex crack motion is
entirely ruled by geometry.

References:
1. Roman B., Reis P.M., Audoly B., De Villiers S., Viguié V., Vallet D., C.R. Mecanique 331
811 (2003).
2. Ghatak A. and Mahadevan L., Phys. Rev. Let. 91, 215507 (2003).
3. Audoly B., Reis P.M. and Roman B., Phys. Rev. Lett., accepted, in press (2005);
4. Audoly B., Reis P.M., Roman B., http://www.lmm.jussieu.fr/platefracture/
 2T2. Brittle fracture 121

CLEAVAGE MECHANISMS IN A SHIP PLATE STEEL

R. Cuamatzi, I. C. Howard and J. Yates

Mechanical Engineering Department, Sheffield University,
Mappin Street, Sheffield, S1 3JD, UK
R.Melendez@sheffield.ac.uk

The proportion of Grade A ship plate steel for the construction of merchant ships is huge, however
there is an increasing concern about the susceptibility of Grade A steel plate to brittle fracture. A
recent ship failure [1] illustrate the importance of the concern. At low enough temperature, the
fracture of structural steels is often dominated by cleavage. McMahon and Cohen [2] demonstrated
that the cracking of carbide particles located at ferrite grain boundaries represents a primary
initiation mechanism in mild steels. Further investigation led to the Smith model [3] which
assumed that a cleavage crack could initiate when a grain boundary carbide is fractured by
impingement of a dislocation pile-up, and the final fracture is controlled by the carbide-sized crack
propagation into the neighbouring ferrite grain. A number of studies [4-8] have used notched
specimens to investigate the mechanism controlling cleavage fracture and the fracture criterion. In
the present work, an investigation into the micro-mechanisms for cleavage fracture has been
undertaken using notched specimens. The material used for the investigation of cleavage fracture
is a Grade A ship plate steel. The chemical composition is given in table 1. The microstructure
reveals the ferrite matrix and bands of pearlite.

Table 1. Chemical composition of Grade A ship plate steel, 20 mm plate (in wt %)

A set of three blunt-notch four-point bend laboratory specimens were tested at test temperature
of -60oC, which is in the lower transition region close to the lower shelf, on a servo-hydraulic
Instron 8501 test machine under displacement control of 0.01mm/s to analyze the consistency of
the observations in each test. It is well known that in double-notched 4PB specimens, when
cleavage initiates from one notch, fracture will certainly propagate in that notch and the surviving
notch must be very close to the critical state due to the same stress condition developed in that
notch. The surviving notches were sectioned for subsequent examination using scanning electron
microscopy for the identification of microcracks and possible particles that induce those
microcracks. Metallographic samples were prepared and etched with 5% nital and numbered.

Figure 1 – Typical microcracks found about 50 m ahead of the notch root.

Observations were made of microcracks developed in front of the notch root of each slice.
Typical micrographs of microcracks are shown in Figure 1. Numbers were used to identify
122 R. Cuamatzi et al.

microcracks nucleated in lamellar pearlite microstructure and capital letters were used to identify
microcracks nucleated in the pearlite boundary. Figure 1 was taken from slice two 50m ahead of
the notch root; this figure shows microcracks nucleated in both, the pearlite lamellar
microstructure and the pearlite boundary microstructure. Microcrack 1 initiated in the lamellar
pearlite microstructure, grew and crossed the pearlite microstructure and finally stopped in the
boundary of the ferrite matrix. Microcrack 2 was nucleated in the centre of the pearlite
microstructure and grew about 5m. Microcrack A originated from pearlite microstructure and
stopped in the ferrite boundary. These two mechanisms of cleavage nucleation also appeared in
slices 3, 4 and 5. For most microcracks, cracked carbides are clearly visible. A microcrack of about
210m size was found in a pearlite colony, the distance of the tip of this microcrack to the notch root
is 184m, that crack was nucleated in the pearlite lamellar microstructure and grew along a pearlite
boundary. Once the tip reached the ferrite grain, it stopped. More microcracks closer to the notch
root were developed. For all these microcracks, cracked carbides are identified as the cleavage
nucleation site.
A JEOL Scanning Electron Microscope was used to obtain more detailed resolution to provide
evidence of cracked carbides as the controlling mechanism for cleavage initiation. Carbides with a
diameter of approximately 0.4m were found. In order to identify which of either mechanism is the
dominant one for nucleating microcracks, the microcracks were counted and it was found that the
domain mechanism for cleavage nucleation is that of microcracks nucleated in the lamellar pearlite
microstructure and that cracked carbides are the controlling mechanism for cleavage initiation
which are probably the critical event of cleavage for this steel. The critical length of the
microcracks for cleavage extension is related to the size of pearlite colonies, suggesting that the
weakest link for cleavage to take place is the pearlite microstructure. Cracks adjoining the initial
microcracks determine the catastrophic event.

References
1. Transportation Safety Board of Canada, Marine Investigation Report, Report Number
M02L0021, Hull Fracture Bulk Carrier Lake Carling Gulf of St. Lawrence, Quebec. March
2002.
2. C. J. McMahon and M. Cohen, Initiation of cleavage in Polycrystalline Iron, Acta Metall.,
1965. 13: p. 591-604.
3. E. Smith, The nucleation and growth of cleavage microcracks in mild steels, Proceedings of
the conference on physical basis of yield and fracture, Institute of Physics and Physical
Society, London, 1966: p. 36.
4. Hendrickson J.A., Wood D. S. and Clarke, D. S., The cleavage fracture of mild steel,
Transaction ASM, 1958(50): p. 656-676.
5. J. F. Knott, Some effects of hydrostatic tension on the fracture behavior of mild steel, Trans.
Iron steel institute, 1966(204): p. 104-111.
6. Eric M. Taleff, John J. Lewandowski, and Bamdad Pourladian, Microstructure-property
relationships in pearlitic eutectoid and hypereutectoid carbon steel, Metallurg. Trans.
2002.(6): p. 25-30.
7. G. Z Wang and J. H. Chen, A statistical model for cleavage fracture in notched specimens of
C-Mn steel, Fatigue Fract. Engng. Mater. Struc, 2001(24): p. 451-459.
8. S. R. Bordet, A. D. Karstensen, D. M. Knowles, C.S. Weisner, A new statistical local
criterion for cleavage fracture in steel. Part II: application to an offshore structural steel,
Engineering fracture mechanics, 2005(72): p. 453-474.
 2T3. Ductile fracture 123

FAILURE BEHAVIOR OF HYBRID-LASER WELDS

A. Bajric and W. Dahl

Department of Ferrous Metallurgy, Aachen University, Intzestr. 1, 52072 Aachen, Germany
aida.bajric@iehk.rwth-aachen.de

In recent years the failure behavior of power beam welds have been investigated by means of
Finite Element Analysis. In order to determine the local stress state around a crack tip before the
crack initiation occurs, the J-Q-M concept has been applied [1]. A Q-factor is added to the crack
driving force, expressed as a J integral and account is taken of the effect of strength mismatch by
the M factor. As a results the recommendations for the application of constraint corrections have
been derived. Additionally the stable crack growth and crack path deviation have been simulated
by using the GTN damage model.
Based on these results the further numerical and experimental investigations have been carried
out on the hybrid laser welds. The objective is to analyze the failure behavior of these welds, with
different geometry and mismatch, which lie between the conventional and pure laser weldments.
First the local stress state in front of the crack tip are quantified by stress triaxiality and Q+M
in dependence of material properties, thickness, mismatch, weld seam geometry and type of
loading. Hence the conclusions are drawn, which can be used for safety assessment of the
component with hybrid laser welds.
For simulation of the stable crack growth the damage models are applied The GTN damage
model and cohesive zone model are compared with regard to prediction of the crack resistance
curves and crack path. The identified parameter sets are verified on fracture mechanics and
component like specimens with different in-plane and out-of-plane constraints. The transferability
of GTN damage parameters have already been demonstrated in [2]. The FIGURE 1 shows
measured (Exp.) and calculated (GTN) load versus diameter reduction curves for round bar
specimens with different notch radii U and net diameter DN. The damage parameters are
determined both for base and weld metal. As regards the crack growth and load-deformation
behavior of CT specimen, the well agreement is achieved between experiments (Exp.) and
calculations (GTN_Model A) (see FIGURE 2). The model A represents due to symmetry one
fourth of the specimen. For the opening displacement greater then 3 mm the crack growth is
underestimated due to the advanced distortion of the elements. This problem can be avoided, when
one half of the specimen is modelled (GTN_Model B). Since the course of crack path depends
besides on the mismatch also on the weld seam geometry, the variation of the weld seam width
across the specimen thickness will be taken in account.
124 A. Bajric and W. Dahl

FIGURE 1. Load versus diameter reduction curves for the notched round bar specimens of
homogenous base metal

FIGURE 2. Load and crack growth versus opening displacement curves for 0.4CT specimens with
hybrid laser welds
References
1. Heyer, J., Lokale Beanspruchung in angerissenen strahlgeschweißten Stahlbauteilen, Berichte
aus dem Institut für Eisenhüttenkunde, vol. 9/2004, Shaker Verlag, August 2004.
2. Nègre, P., Steglich, D., Brocks, W., Crack extension in aluminium welds: a numerical
approach using the Gurson-Tvergaard-Needleman model, Engineering Fracture Mechanics,
Nr.7, S. 2365-2383, 2004.
 2T3. Ductile fracture 125

FRACTURE OF PLASTIC BODIES. DEFORMATIONS CONCENTRATORS

A. I. Khromov, A. A. Bukhanko, S. L. Stepanov and E. P. Kocherov

Institute of Machining and Metallurgy Far Eastern
Branch of the Russian Academy of Sciences
1, Metallurgov Str., Komsomolsk-on-Amur, 681005, Russia
khromov@imim.ru, bukhanko@imim.ru

On the basis of the theory of ideal rigid-plastic body the approach to definition of invariant
tensorial characteristics of fracture is formulated on the basis of standard mechanical tests on
uniaxial tensile of flat and cylindrical samples. Instead of experimentally determined
characteristics of materials fracture (relative extension and narrowing of a sample at fracture), two
are entered invariant tensorial characteristics of a degree of sample deformation. They correspond
to the moment of macrocrack initiation and critical deformation in the crack vertex determining
process of crack propagation.
One of problems of the theory of ideal rigid-plastic bodies is nonuniqueness of position, mode
of plastic area and together with it nonuniqueness of a field of displacement velocities which
define change of a body geometry. For practical use of theoretical solutions it is offered: criterion
of a choice of a preferable instant field of displacement velocities and the criteria determining
change of a velocities field in time (change of plastic area). Suggested criteria are based on
extreme principles of nonequilibrium thermodynamics.
Within the framework of the theory of ideal rigid-plastic body the approach to investigation of
areas of abrupt change of the form as strains concentrators is offered. Definition of strains fields in
the neighbourhood of the concentrator is reduced to integration of the ordinary differential
equations. The number of analytical solutions for concentrators as V-notched bar is received.
Dependence of strains fields on form change and plastic area position is investigated during plastic
current. The number of solutions of a problem about fracture of V-notched rigid-plastic bar is
offered.
The approach to investigation of fracture processes for more complex bodies models is
formulated on the basis of the obtained solutions: elasto-plastic, strengthened plastic bodies, etc. It
is offered to consider the material in the neighbourhood of vertex crack as rigid-plastic material. It
allows to carry out the analytical description of strains fields in the neighbourhood of vertex crack
and to apply new deformation and power fracture criteria.
 2T3. Ductile fracture 127

3D DUCTILE TEARING ANALYSES OF BI-AXIALLY LOADED PIPES WITH

SURFACE CRACKS

Andreas Sandvik, Erling Ostby1 and Christian Thaulow

Department of Engineering Design and Materials, The Norwegian University of
Science and technology, N-7491 Trondheim, Norway
1Department of Applied Mechanics and Corrosion, SINTEF Materials and Chemistry, N-7465
Trondheim, Norway
asand@statoil.com, erling.ostby@sintef.no, christian.thaulow@sintef.no

This paper concerns 3D ductile tearing analyses of outer surface cracked pipes subjected to tension
loading and internal pressure.
Many new offshore development projects are in ultra-deep water depths with high pressure and
temperature reservoirs. Consequently, the pipelines may be exposed to extreme loadings under
service conditions and/or laying. Today, the tensile side often limits the allowable strain, and
therefore might limit the material utilization. It is believed that existing procedures/standards are
rather conservative with regard to fracture assessment for scenarios with global plastic strain
occurring. As a consequence, more knowledge is needed about the governing fracture response
parameters.
In this work we present FEM analyses of 3D outer surface cracked pipes loaded in tension with
and without internal pressure. The outer diameter, D, is 400[mm] and thickness, t, 20[mm]. Three
different crack depths and crack lengths are chosen. All the dimensions of the cracks are chosen
such that they are representative examples for possible girth weld defects. The canoe type defect
geometry assumed is illustrated in Fig.1.
The analyses are performed with a material isotropic power law hardening on the form:
n
§ H ·
Vi V 0 ¨¨1  p ¸¸
© H0 ¹ (1)
where V i is the flow stress, V 0 is the stress at the proportional limit, H p is the plastic strain,
and n is the hardening exponent. H 0 = V 0 e E , is the strain at the proportional limit, and E is the
Young’s modulus. If V  V 0 the material behaviour is linear elastic. Further, n 7 and
V0 460 [MPa] are assumed in the current analyses.
Ductile tearing is taken into account using the so-called Gurson-Tvergaard-Needleman model,
[1,2]. This model accounts for void growth and void coalescence. The void coalescence is
modelled with the Tvergaard and Needleman’s [3] “effective void volume fraction”. The value for
the initial void volume fraction parameter was fitted to represents a realistic material resistance
found from X65 pipeline-steel.
128 A. Sandvik et al.

FIGURE 1. (a) The pipe geometry with an external circumferential constant-depth surface flaw.
(b) Details of the canoe type defect with arc length, 2c, depth, a, and end radius, r, equal to the
crack depth, a.
The FEM model was solved using Abaqus Explicit [4], which originally was developed to
solve dynamic events. However, it is possible to retrieve quasi-static solutions with this solver, as
long as you prevent significant mass effects.
The results show how variations in crack lengths and crack depths, in addition to the effect of
internal pressure, influences the crack driving force and the strain capacity of the pipe.

References
1. Gurson, A.L., Continuum theory of ductile rupture by void nucleation and growth: Part I –
Yield criteria and flow rules for porous ductile materials J. of Eng. Materials, vol. 99, 2-15,
1977.
2. Tvergaard V., Influence of voids on shear band instabilities under plane strain conditions, Int.
Journal of Fracture, vol. 17, 389-407, 1981.
3. Tvergaard, V, Needleman, A., Analyses of the cup-cone fracture in a round tensile bar”, Acta
Metallurgica, vol. 32, 157-169, 1984.
4. Abaqus User’s manual, Version 6.5, Abaqus Inc., 2004.
 2T3. Ductile fracture 129

NEW MODEL MATERIALS FOR DUCTILE FRACTURE STUDIES

A. Weck and D. S. Wilkinson

Department of Materials Science and Engineering, McMaster University
1280 Main St. West, Hamilton, ON, L8S 4L7, Canada
weckag@mcmaster.ca

Ductile fracture of metals involves a sequence of overlapping processes that include the
nucleation, growth and coalescence of voids. Of these, the last is the most important because it
dictates the ductility of metals but is however the less understood. As the coalescence is a
stochastic event occurring over very short strains, it is really difficult to capture experimentally.
Attempts to fabricate model materials that would simplify the analysis of the ductile fracture
process have already been made by Babout et al.[1], Gammage et al.[2], Magnusen et al.[3], Jia
and Povirk [4] and Nagaki et al.[5]. However, they are of limited help when one wants to study the
coalescence event in detail because of their complicated microstructures, of there limited ductility,
of the controlling effect of the nucleation event, of the lack of constraint (2D approaches instead of
3D) or simply because they do not reproduce the key features of the microstructure (like the void
size for instance).
The present work describes the fabrication and characterization of two new model materials
for ductile fracture studies that overcome the previously described limitations.
The first consists of a controlled two dimensional array of laser drilled holes in metallic sheets
(2D approach) and the second consists of a controlled three dimensional array of laser drilled holes
in the bulk of metallic samples (3D approach). To drill the holes, a femtosecond laser is chosen for
the small heat affected zone that is created around the hole and the small hole size (10microns) that
can be produced.
The first model material consists of an array of laser drilled holes in 100microns thick metallic
sheets (Fig.1).

The samples are then tested in-situ in a Scanning Electron Microscope to capture the
coalescence event (Fig.2). Different arrays of holes can be used and the preliminary results are that
different modes of coalescence are obtained depending on the array of holes.
For the second model material, the laser holes are drilled in thin (typically 15 microns in
thickness) metallic sheets and the sheets are then diffusion bonded to obtain the 3D structure
(Fig.3). These samples are then tested in situ in the X-Ray Computed Tomography set-up at the
synchrotrons in France (ESRF) and in Japan (SPring-8).
130 A. Weck and D. S. Wilkinson

The results from the tomography experiments (Fig.4) confirm the successful fabrication of the
model material. Furthermore, the holes growth and coalescence can be easily followed allowing
valuable experimental data to be collected to support subsequent modeling efforts.

References
1. L. Babout, E. M., J.Y. Buffière and R. Fougères, Acta Mater., 49, 2055-2063, 2001.
2. J. Gammage, D. S. Wilkinson, J. D. Embury and Y. Brechet, Acta Mater., 52, 5255-5263,
2004.
3. Magnusen, P. E., E. M. Dubensky, et al., Acta Metallurgica, 36, 1503-1509, 1988.
4. S. Jia, G. F. R., G.L. Povirk, International Journal of Solids and Structures, 39, 2517–2532,
2002.
5. Nagaki, S., Y. Nakayama, et al., International Journal of Mechanical Sciences, 40, 215-226,
1998.
 2T3. Ductile fracture 131

FATIGUE THRESHOLD COMPUTATION MODEL BASED ON THE

SHAKEDOWN ANALYSIS

M. A. Belouchrani, D. Weichert1 and A. Hachemi2

Laboratoire Génie des Matériaux, E.M.P. BP 17C Bordj El Bahri Alger Algérie
1Insa de Rouen, Place Emile Blondel, BP 8 76131, Mont-Saint-Aigan France
2Institut für Allgemeine Mechanik, RWTH-Aachen, D-52056, Aachen Germany

nbelouch@yahoo.fr

The fatigue thresholds remain very significant parameters, helping the designer and the
manufacturer in their decision to reform a mechanical structure or its dimensioning.
These last years, several models were advanced in order to propose a method of their
determination or the study of their influence factors such as the microstructure or the load ratio.
The established fatigue threshold models can be classified in two groups, the theoretical
models and the models based on the experimentation. The majority, not to say all the models, agree
on the fact that the fatigue threshold increases with the size of the grain and depends strongly on
the yield stress of the material.
The fatigue threshold is interpreted physically as the sum of two components, one
i c
microstructural 'Kth and the other physics 'K th caused by the crack closing effect. So that:

' K th ' K thi  ' K thc (1)

For a structure with a regular distribution of the grains and with various orientations, the relation
allowing to connect the values of the microstructural threshold with the microstructure dimension
[ (grain size) combined with the definition of the crack propagation force related to the
deformation energy located at the crack tip, provided the expression of the fatigue threshold
depending on R (load ratio) [1, 2]:

i c 1
' K th ' K th  ' K th ( C 1 (1  R )  )V FL max S[
2 (2)
For materials verifying the law of Petch, the fatigue threshold can be written like:

' K th C 2  C 3[ 0 . 5 (3)
Expression formulated in experiments by the majority of the researchers, who consider
nevertheless the values thresholds for propagation velocities of about 10 -8 mm/cycle [3].
In this work, we will present a synthesis of these models which will be compared with our
model based on the shakedown analysis [4]. This analysis consider an elastic-plastic structure,
subjected to variable mode I loading P(t) and occupying a volume : with a surface * consisting of
disjoint parts *V and *u, where statical and kinematical conditions are prescribed, respectively.
The values of P(t) vary arbitrarily with time t, but remains between a prescribed loads Pmin and
Pmax. One then looks for the maximum value of the load factor D, such that the structure shake
down under the loads DP(t). This load factor called shakedown load factor DSD is determined as
solution of the following optimisation problem:
132 M. A. Belouchrani et al.

D SD max$ D
D ,U
(4)

U ij$ , j 0 in :
(5)

n j U ij$ 0 on *V
(6)

F (DV c ( P )  U $ , V s )  0 in :  P  >P min , Pmax @ (7)

m
ªm 1 1 D 1 º ( m 1)
a lim a0  « ³ 
U ij$ L ijkl U kl$ d : »  ac
¬ m K D 1 : 2 ¼
(8)

Here, F is a convex yield surface, V s the yield stress, a0 and ac are, respectively, the initial
crack length and the critical one for which unstable crack propagation occurs, when alim is the
limit crack length which can be attained when the shakedown state is reached V c (t ) is the time-
dependent stress state for a purely elastic comparison problem, differing from the original problem
only by the fact that the material reacts purely elastically with the same elastic moduli as for the
elastic part of the material law in the original problem and U $ time-independent state of residual
stress.
With the shakedown load factor D SD computed for a cracked structures loaded in mode I. We
can compute the fatigue threshold corresponding to the shakedown state 'K th , given by :

'K th D SD .P. Sa (9)

The computation of 'Kth with different grain size shows that for materials verifying the law of
Petch, the fatigue threshold can be written like [5]:

'K th (C4  C5[ 0.5 )V s

(10)
Where C1, C2, C3, C4, C5, m and K are material constants. The comparison of all these models
shows that the shakedown analysis constitutes an effective tool in the determination of the fatigue
thresholds. It has the advantage of determining them at the initiation of the crack propagation and
correctly translated the effect of the yield stress of the material by the influence of the residual
stresses and the grain size.
References
1. Wasen J., Hamberg K. and Karlsson B., Mater. Sci. Engng. A, 102, 217-226, 1988
2. Ravichandran K. S. and Dwarakadasa E. S., Acta Metall. Mater., vol. 39, N° 6, pp. 1343-1357, 1991
3. Nakai Y., Tanaka K. and Nakanishi T., Engng. Fract. Mech., vol. 15, N° 3-4, pp. 291-302, 1981.
4. Belouchrani M. A., Weichert D., Int. J. Mech. Sci., vol. 41, N° 2, pp. 163-177, 1998.
5. Belouchrani M. A., Weichert D. and Hachemi A., Mechanics. Resea. Comm., vol. 27, N° 3, pp. 287-293,
2000
 2T3. Ductile fracture 133

VOID COALESCENCE IN METALS INVOLVING TWO POPULATIONS OF

CAVITIES

D. Fabregue and T. Pardoen

Unité de Matériaux et des Procédés (IMAP), Université Catholique de Louvain (UCL)
Place Sainte Barbe 2, B-1348 Louvain-la-Neuve, Belgium
fabregue@imap.ucl.ac.be, pardoen@imap.ucl.ac.be

Void coalescence is the final stage in the failure of ductile materials. It consists in the localization
of the plastic deformation in the intervoid ligament between neighbouring voids. Several
experimental evidences obtained from fractographic analyses of broken samples or metallographic
analyses of polished samples strained near fracture have shown that a second population of cavities
nucleated on small particles significantly affect the damage process controlled by the first
population of cavities nucleated on larger particles. Although a second population of voids is
considered for a while in the literature as very detrimental for the ductility (e.g. Marini et al. [1]),
only a limited number of studies have been devoted to the modelling of this phenomenon
(Tvergaard [2], Brocks et al. [3], Faleskog and Shih [4], Perrin and Leblond [5], Enakousta et al.
[6]). The aim of this work is to develop a constitutive model for the nucleation, growth and
coalescence of voids that incorporate the effect of this second population on the onset of
coalescence.
Firstly, FE unit cell simulations have been performed with a first population of cavities
explicitly modelled in the mesh while the second population is introduced through the use of the
Gurson response for the matrix surrounding the first population (Gurson [7]) such as in Brocks et
al. [3]. The underlying assumption of this approach is thus that the second population is made of
much smaller voids than the first population, with a volume fraction sufficient to smear out their
effect within the ligament between the big voids. Calculations have been carried out for different
volume fractions of the first population, different stress triaxialities, different volume fractions of
the second population and different nucleation conditions for the second population. Fig. 1 shows
that the onset of coalescence is about 0.76 for a first population of 1.5% without second population
whereas the coalescence occurs for a macroscopic strain of 0.46 if a second population is present
for the same overall amount of initial porosity. Cell calculations show that the presence of the
second population significantly cut down the strain at the onset of coalescence. The effect of the
nucleation strain is also exhibited in Fig. 1.
The cell calculations show that the presence of a second population essentially affects the onset
of coalescence and not the evolution of the first population before coalescence. The void
coalescence model of Thomason (Thomason [8]) has been modified to take into account the
presence of the second population of voids through its softening effect on the strength of the matrix
material. Cell calculations show that the onset of coalescence is dictated by the local softening
caused by the second population inside the ligament very close to the large void. This model
provides very accurate predictions of the onset coalescence when all the evolutions of the variables
(relative void spacing, void aspect ratio of first population and current yield stress near the large
void) are extracted from the FE cell calculations. As a second step, the coalescence model is
coupled to an enhanced Gurson model for the description of the first population (Pardoen and
Hutchinson [9]). The difficulty relies on approximating the local strains near the large voids to
solve the Gurson model locally in order to calculate the evolution of the second population of
voids introduced into the void coalescence conditions. This approach which assumes that the
second population mostly affect the void coalescence conditions must be contrasted with another
134 D. Fabregue and T. Pardoen

assumption which treat the second population as an extra contribution to growth rate of first
population by repetitive coalescence between the big and small cavities (Enakousta et al. [6]).

FIGURE 1. Influence of the presence of the second population in cell calculations (f1=fraction of
1st population, f2=fraction of 2nd population, ånucl=strain at nucleation of the 2nd population)

References
1. Marini, B., Mudry, F, Pineau, A., Engng. Fracture Mech., vol.22, 989-996, 1985
2. Tveergard V., Int. J. Solids Structures, vol.35, 3989-4000, 1998.
3. Brocks, W., Sun, D.Z., Hönig, A., International Journal of Plasticity, vol.11, 971-989, 1995.
4. Faleskog, J., Shih, F.C., Journal of the Mechanics and Physics of Solids, vol.45, 21-50, 1997.
5. Perrin, G., Leblond, J.B., International Journal of Plasticity, vol.6, 677-699, 1990.
6. Enakousta, K., Leblond, J.B., Audoly, B., In Proceedings of the 11th International
Conference on Fracture, CD-Rom, 2005.
7. Gurson, A.L., Journal of Engineering Materials and Technology, vol.99, 2-15, 1977.
8. Thomason, P., Ductile Fracture of Metals, Pergamon Press, Oxford, 1990.
9. Pardoen, T., Hutchinson, J.W., Journal of the Mechanics and Physics of Solids, vol.48, 2467-
2512, 2000.
 2T3. Ductile fracture 135

EFFECTS OF CHARACTERISTIC MATERIAL LENGTHS

ON DUCTILE CRACK PROPAGATION

E. Radi
Dipartimento di Scienze e Metodi dell’Ingegneria, Università di Modena e Reggio Emilia
Viale Allegri, 13. I-42100 Reggio Emilia, Italy
eradi@unimore.it

Classical plasticity theories fail in characterizing the constitutive behavior of ductile materials at
the micron scale, which is necessary to define in order to investigate the stress and strain fields
near a propagating crack tip. Experimental observation on the macroscopic fracture toughness and
atomic work of separation of an interface between a ductile crystal of niobium and a sapphire
single crystal performed by Elssner et al. [1] found that the interface between the two materials
remained sharp and not blunted up to the atomic scale. Moreover, the stress level required to
produce atomic decohesion of the lattice turns out to be about 10 times the tensile yield stress,
whereas fracture mechanics analyses based on classical plasticity theories (Drugan et al., [2])
provide a maximum stress level near a crack tip not larger than 4–5 times the tensile yield stress.
Classical continuum theories are also unable to predict the size effect arising at small scales, due to
the lack of a length scale. Therefore, in order to describe the stress and deformation fields very near
the crack-tip during its propagation, it become necessary to adopt enhanced incremental
constitutive models, which account for the non linear behaviour of the material as well as for the
microstructure and the presence of dislocations, by incorporating one or more characteristic
lengths, typically of the order of few microns for ductile metals.
The couple stress (CS) flow theory of strain gradient plasticity has been introduced, in a first
attempt, by Fleck and Hutchinson [3]. Their model involves an intrinsic material length " and an
elastic length scale "e arbitrarily introduced in order to partition the deformation curvature rate
tensor into its elastic and plastic parts. The incremental model of CS plasticity developed by
Ristinmaa and Vecchi [4] and Ottosen et al. [5] is based on the Koiter [6] theory of couple stress
elasticity and introduces two distinct intrinsic material length, namely " and "c, whose effect on the
propagating crack tip field is almost unexplored.
Several analyses, carried out to investigate the effects of microstructure in fracture mechanics,
showed that the incorporation of couple stress and rotation gradients in the constitutive description
of ductile materials improves considerably the estimation of the stress traction level ahead of the
crack-tip. In particular, for the problem of mode I steady-state crack propagation, Wei and
Hutchinson [7], analysed the effects of stretch gradients (SG plasticity) in a numerical simulation
and found a sensible amplification of the traction level ahead of the crack-tip. It must be observed
that former investigation of crack propagation in linear hardening ductile materials described by
classical plasticity theories (Ponte Castañeda, [8]; Bigoni and Radi, [9, 10]) predicted an extremely
weak stress singularity for the typically small values of the strain hardening coefficient.
Recently, Radi [11] performed an asymptotic analysis of the same problem adopting the CS
flow theory of plasticity developed in [3] and found that the couple stress field is dominant near to
the crack-tip and produces a remarkable increase of the stress singularity, even for low hardening.
However, these analyses showed that the contribution of the elastic strain gradients strongly affects
the asymptotic crack-tip fields, through the elastic length scale "e. The further analysis performed
by Radi and Gei [12] indicated that the stress singularity increase at the crack-tip also for the Mode
III crack problem, due to the sole effects of rotation gradients provided by the CS theory of
plasticity, with no need to take stretch gradients into consideration.
136 E. Radi

In the proposed paper, the effects of strain rotation gradients on steady-state crack propagation
are investigated by performing an asymptotic analysis of the crack-tip fields derived from the flow
theory of CS plasticity with two characteristic material lengths. Rotation gradients are expected to
become significant at a distance from a crack-tip smaller with respect to these characteristic
lengths, and negligible at larger distances, with a gradual transition in the intermediate region.
According to the results obtained for a single characteristic material length " in [11, 12], the
couple stress and the skew-symmetric stress field are expected to dominate the asymptotic field
under Mode I and Mode III loadings conditions, respectively, and to produce an increase of the
stress singularity at the crack tip, also for a small hardening coefficient. The role of both
characteristic lengths " and "c will be examined in detail and their influence on the stress
singularity will be explored by numerical investigations of the asymptotic crack tip fields.
The performed asymptotic analysis will provide predictions on the level of traction ahead of
the propagating crack-tip more realistic then the classical solution obtained for the J2-flow theory
or for a single characteristic length, allowing the detailed mechanisms by which fracture may grow
and propagate in ductile metals to be understood in more depth, up to the micron scale.

References
1. Elssner, G., Korn, D. and Ruehle, M., Scripta Metall. Mater. vol. 31, 1037-1042, 1994.
2. Drugan, W.J., Rice, J.R. and Sham, T.L., J. Mech. Phys. Solids, vol. 30, 447-473, 1982.
3. Fleck, N.A. and Hutchinson, J.W., J. Mech. Phys. Solids, vol. 41, 1825-1857, 1993.
4. Ristinmaa, M., and Vecchi, M., Comp. Meth. Applied Mech. Engnrg., vol. 136, 205-224,
1996.
5. Ottosen, N.S., Ristinmaa, M. and Ljung, C., European J. Mech. - A/Solids, vol. 19, 929-947,
2000.
6. Koiter, W.T., Proc. Roy. Netherlands Acad. Sci. vol. B67, 17-44, 1964.
7. Wei, Y. and Hutchinson, J.W., J. Mech. Phys. Solids, vol. 45, 1253¯1273, 1997.
8. Ponte Castañeda, P., J. Mech. Phys. Solids, vol. 35, 227-268, 1987.
9. Bigoni, D. and Radi, E., Int. J. Solids Struct., vol. 30, 899-919, 1993.
10. Bigoni, D. and Radi, E., Int. J. Fracture, vol. 77, 77-93, 1996.
11. Radi, E., J. Mech. Phys. Solids, vol. 51, 543-573, 2003.
12. Radi, E., Gei, M., Int. J. Fracture, vol. 130, 765-785, 2004.
 2T3. Ductile fracture 137

DUCTILE FRACTURE BY VOID NUCLEATION AT CARBIDES

J. Giovanola, D. Cannizzaro, R. Doglione1 and A. Rossoll

Ecole Polytechnique Fédérale de Lausanne
LCSM-IPR-STI, EPFL, Station 9, 1025 Lausanne, Switzerland
1Politecnico Torino

jacques.giovanola@epfl.ch

Predicting ductile fracture of initially undamaged structures still represents a challenge in most
applications such as for instance crashworthiness evaluations. This paper focuses on ductile
fracture by the process of void nucleation at tightly bonded and closely spaced inclusions in an
otherwise homogeneous ductile matrix. Particular attention is given to loading conditions
involving only low stress triaxialities (0 to 0.5).
As model material for the investigation, we selected a Vacuum Arc Remelted (VAR) steel:
DIN Standard 39NiCrMo4, AMS 6414, austenitized at 860 °C for 1 h, oil quenched, tempered at
704 °C for 1h. and air cooled. This heat treatment resulted in a material with a ferritic matrix (yield
strength of 700 MPa, ultimate strength 800 MPa, reduction of area 37.5%) and a dense distribution
of fine carbides (volume fraction of about 6% with a size distribution ranging from 0.1 to 1 Pm,
with an average size of 0.23 Pm).
We developed a simple torsion-tension test using a thin-walled tubular specimen (Fig. 1) that
allowed us to investigate void nucleation at carbides under low triaxiality loading (0 – 0.5, axial
load of 0, 1/3 and 2/3 of the limit load Plim). We were able to measure local strains and to interrupt
the test during the softening stage of deformation and thus gain access to the damage history in the
specimen.

FIGURE 1. (a) Thin-walled cylindrical specimen for low triaxiality tension-torsion tests. (b)
Torque and axial displacement versus twist angle curves.

By means of grid deformation and high speed photography techniques, we demonstrated that
in the investigated quenched and tempered VAR steel, two competing failure mechanisms lead to
fracture: shear localization, which dominates at very low triaxialities and micro-void nucleation
and coalescence, as soon as some significant tensile load is superimposed on shear deformation.
We modeled the observed void nucleation process by means of a critical interfacial stress
condition. We used a published interfacial stress model based on dislocation theory, the Brown
Stobbs [1] Kwon Asaro [2] model (BSKA) to calculate an average interfacial stress acting in a
138 J. Giovanola et al.

representative volume element from the continuum stress and strain fields applied to this element.
Using the BSKA model to estimate the critical interfacial stress from experimental data obtained in
low-triaxiality tension-torsion tests, as well as in notched tension tests with higher triaxialities, we
estimated a critical interfacial stress of 3000 MPa and were able to fit a single void nucleation
threshold curve over a range of triaxialities from 0 to 1.5 (Fig. 2)

FIGURE 2. Void nucleation threshold curve: fit of BSKA model to experimental data for a critical
interfacial stress value of 3000 MPa.

We provided a first qualitative validation of the predictive capabilities of the model by

combining stress and strain results of finite elements calculations with the BSKA model to
determine the location in various torsion and torsion-tension specimens, where voids nucleate first
i.e the location of the maximum interfacial stress, The results of these simulations indicate that, as
axial loading is increased, the site of first void nucleation moves from the outside edge toward the
center of the specimen section. These predictions were confirmed by detailed fractographic
observations of tested thin-walled cylindrical specimens.

References
1. Brown, L. M. and Stobbs, W. M. , Phil. Mag., 34, 351-372, 1976.
2. Kwon, D. and Asaro, R.J., Metallurgical Transaction A, 21A, 117-134, 1990.
 2T3. Ductile fracture 139

THE SIGNIFICANCE OF MAXIMUM LOAD ON A LOAD-DISPLACEMENT

CURVE WITH STABLE CRACK EXTENSION

J. R. Donoso and J. D. Landes

Universidad Técnica Federico Santa María, Valparaíso, CHILE
University of Tennessee, Knoxville, TN, USA
juan.donoso@usm.cl, landes@utk.edu

The ASTM standard method for the measurement of fracture toughness, E 1820 [1], covers
procedures and guidelines for the determination of this material property in metallic materials
using the parameters K, J or CTOD. The fracture toughness may be measured as a point value, or
as a complete fracture toughness resistance curve. In the latter option a J- or CTOD-based
resistance curve may be obtained from a single specimen fracture test, in which the crack length is
measured from compliance changes, and later verified by optical measurements. The single
specimen J-R curve construction procedure defined in E1820 involves several steps, which go
from obtaining and plotting the raw J-'a data, to calculating an interim value of J, termed JQ, to
finally qualifying JQ as JIc, a size-independent value of fracture toughness.
Recently, Donoso, Zahr and Landes proposed an alternative way of obtaining the specimen J-R
curve using the Concise and Common (C&C) Formats [2]. The Concise Format [3] and the
Common Format [4] developed by Donoso and Landes are calibration functions that relate the
load, P, to the displacement, v, and the crack length, a, of a fracture toughness specimen. The
Concise Format stands for the elastic regime, in which v = vel [3], whereas the Common Format
deals with the plastic component of the displacement, vpl [4].
The Common Format Equation, CFE, was originally proposed by Donoso and Landes [4] as an
extension of the load separation concept [5] to describe the load-plastic displacement relationship
for a blunt-notch fracture specimen. As such, it relates the load P with two variables representing
the non-linear deformation of a fracture specimen with a stationary crack: vpl/W, the plastic
component of the load-line displacement, normalized by the specimen width W, and b/W, the
normalized ligament size (ligament b in lieu of the crack length a). The CFE also includes a term
that denotes the out-of-plane constraint, :*, and is written as:

P = :*˜B˜C˜W˜(b/W)m ˜V*˜(vpl/W) 1/n (1)

where B is the specimen thickness; C and m are the geometry function parameters, and V* and n
are material properties, obtained from a stress-strain curve.
In order to obtain the specimen J-R curve using the Concise and Common (C&C) Formats,
Donoso, Zahr and Landes [2] proposed a “crack growth law” to account for the relation between
stable crack extension, 'a, and vpl. The assumed crack growth law is a two-parameter power law
equation relating the change in crack length, 'a, with normalized plastic displacement, vpl/W, that
has the form

l1
'a § v pl ·
l 0 ¨¨ ¸
¸
W © W ¹ (2)
140 J. R. Donoso and J. D. Landes

The crack extension, 'a, may also be written in terms of the change in ligament size, that is, 'a =
bo – b, where bo is the initial ligament size (equal to W minus the initial crack length, ao). Thus,
Eq. (2) gives the following expression for the current ligament size, b:

l1
b b § v pl ·
o
 lo¨ ¸
W W
©W ¹ (3)
Substitution of Eqs. (2) and (3) into Eq (1) yield the following expression in terms of only the
plastic displacement when there is stable crack growth, with D being the product of the parameters
V* and :*:

m
§
¨ bo § v pl ·l 1 ·¸ § v pl
1/ n
·
P DBCW ¨  lo ¨ ¸ ¸ ¨
¨W
¸
¸
¨W ¨ W ¸ ¸ © ¹
© © ¹ ¹
(4)
The use of Eq. (4) not only has made it possible to generate Jpl as a first step to obtaining the
complete J-R curve [2], but also to predict maximum load on a load vs. displacement (P-v) curve
for a specimen that undergoes stable crack growth [6]. The purpose of this paper is to show how
the maximum load on a P-v curve obtained with a 1T-C(T) fracture specimen showing crack
extension relates to the value of J at initiation of ductile cracking, as defined by JQ following the
construction procedure of E1820. Following this, the potential use of the maximum load on a P-v
curve to determine directly a value of JQ will be explored and discussed.

References
1. E1820-99, Standard Test Method for the Measurement of Fracture Toughness, Annual Book
of Standards, Vol. 03.01.
2. Donoso, J.R., Zahr, J., and Landes, J.D., Second International ASTM/ESIS Symposium on
Fatigue and Fracture Mechanics, Tampa, FL, USA, November 2003.
3. Donoso, J.R. and Landes, Fatigue and Fracture Mechanics: 32nd Volume, ASTM 1406, R.
Chona. Ed., American Society for Testing and Materials, West Conshohocken, PA, 2001,
261-278.
4. Donoso, J.R. and Landes, J.D., Engineering Fracture Mechanics, vol. 47, No. 5, 619-628,
1994.
5. Ernst, H.A, Paris, P.C., and Landes, J.D., Fracture Mechanics; Thirteenth Conference, ASTM
STP 743, Richard Roberts, Ed., American Society for Testing and Materials, 1981, 476-502.
6. Donoso, J.R. and Landes, J.D., An Instability Analysis for a Crack Growth Situation Based on
the Common Format, ECF-15, Stockholm, 2004.
 2T3. Ductile fracture 141

3D VISUALIZATION OF DUCTILE FRACTURE USING SYNCHROTRON X-

RAY COMPUTER TOMOGRAPHY

L. Qian, H. Toda, T. Ohgaki, K. Uesugi1, M. Kobayashi and T. Kobayashi

Department of Production Systems, Toyohashi University of Technology,
Toyohashi, Aichi, 441-8580, Japan
1Japan Synchrotron Radiation Research Institute, Sayo, Hyogo, 679-5198, Japan

qian@sp-mac4.pse.tut.ac.jp (L. Qian)

The initiation, growth and coalescence of voids during ductile fracture can be evidenced by means
of traditional methods such as optical microscope or SEM. However, the obtained information is
limited by the surface observations in that the surface stress-strain state differs from that of the
interior material. In addition, false images may arise from sample surface preparation. Recently,
synchrotron X-ray CT has been applied to investigate fatigue crack problems [1,2]. The purpose of
the present work is to use this advanced technique to visualize three-dimensionally the fracture
process in a practical ductile material, hence gaining a new insight into the ductile fracture.
Uni-axial tensile tests were conducted on a notched specimen made of Al-Si alloy using a test
rig specially designed for the X-ray tomography. The in-situ observations were performed during
tensile loading using a high resolution X-ray tomography at the X-ray imaging beamline BL47XU
of SPring-8. The reconstructed 3D images, with a resolution of 0.474Pm, of the damaged material
were quantitatively visualized.
It is found that at a smaller load, voids are nucleated in front of the notch. The voids occur
mainly within the eutectic region and are closely related to the eutectic particles. With increasing
load, the size and number of voids are increased, and some adjacent voids coalesce with each
other. Sectional images indicate that the number, size and shape of the voids differ so much from
slice to slice. Some slices demonstrate crack length longer than 60 Pm while other slices show only
a few small voids.
Fig. 1(a) and (b) presents the reconstructed 3D images after two loading stages, showing the
void distributions by removing other material components. On the whole, the high density of voids
is mainly concentrated near the notch, and increasing load leads to an increase in number and size
of the voids. Quantitative analyses indicate that the generated voids almost lie within a region of
about 150 Pm near the notch tip. Both the number and size become even larger with further
increasing the load. It is also indicated that more and larger voids are generated in the internal
region than on the side surfaces.
142 L. Qian et al.

FIGURE 1. Void distribution in front of notch after loading to (a) 51.6 N and (b) 55.0 N. Arrows
indicate crack growth direction.

In conclusion, the high resolution X-ray CT has been proved to be a feasible way to visualize
and quantify the ductile fracture, and provide much more information than traditional methods.
Voids are nucleated more easily in notch front than in remote regions. The distribution of voids is
not uniform in a real material, and the initiation, growth and coalescence of voids are not
simultaneous along the notch front. In addition, three-dimensional architecture of particles packing
in the eutectic region, and the 3D morphology of D-phases are also visible, which can be applied as
input data for numerical simulations of fracture behavior.
Finally, the synchrotron radiation experiments were performed at the BL47XU in the SPring-8
with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal No
2004B0457-NI-np).

References
1. Toda, H., Sinclair I., Buffiere J. –Y., Maire E., Connolley T., Joyce M., Khor K.H., Gregson
P., Philos. Mag., vol. 83, 2429-2448, 2003.
2. Toda, H., Sinclair I., Buffiere J. –Y., Maire E., Khor K.H., Gregson P., Kobayashi T., Acta
mater., vol. 52, 1305-1317, 2004.
 2T3. Ductile fracture 143

NON-LOCAL PLASTIC-DAMAGE MODEL FOR FAILURE ANALYSIS OF

SHEET-METALS

M. Brunet, F. Morestin and H. Walter-Leberre

LaMCoS, Institut National des Sciences Appliquées,
20 Avenue A. Einstein, 69621 Villeurbanne, France.
Michel.brunet@insa-lyon.fr

A new damage model which takes account of void shape effect and anisotropy of the matrix
material is integrated into the dynamic explicit finite element framework to predict the damage
evolution which occurs under crash or stamping process. For the strain localization and failure, the
pathological mesh dependence has been overcome by a non-local approach where the evolution
equation for the porosity and the equivalent plastic strain is modified by an additional term
containing a characteristic internal length. The non-local plastic-damage potential is written as:

q2 § NV H ·
) C  2q1 f cosh¨ ¸  (1  q f 2 )
3 0
p
V y2 (H loc , ’ 2H p ) ¨ V (H p , ’ 2H p ) ¸
© y loc ¹ (1)
The damage model can take into account the three main phases of damage evolution: growth,
nucleation and coalescence. To determine the critical porosity f c , the void coalescence failure
mechanism by internal necking is considered by using a modified Thomason’s plastic limit-load
model on the reference volume element such as:

­ N M ½ ª
° § RZ · §R · ° V1 º V1
®F / ¨¨ ¸
¸  G /¨ X ¸ ¾ An «1  T »d
°̄ © X  RX ¹ © X ¹ °¿ ¬ « V y »¼ Vy
(2)
Consistent with (2), the plastic-damage potential (1) is used to calculate the void and matrix
geometry changes using the current strain, void volume fraction f and shape factor S. Once the
inequality (2) is satisfied, the void coalescence starts to occur and the void volume fraction at this
point is the critical value f c .

To evaluate the non-local variables (f , H p ) and their gradients at a Gauss point, the partial
differential equation of the implicit gradient formulation as:

p
’ 2H p ( x)  P 2H p ( x)  P 2H loc ( x)
(3)
is solved using the integral equation method at the beginning of each time step employing the
overlay mesh defined by the Gauss integration points of the underlying shell-element mesh
restricted to be in the same plane in order to solved a two-dimensional problem.
Once the damage model has been calibrated by an inverse identification technique using the
non-local FEM analysis on tensile tests, it can be used to predict the appearance of tearing in any
stamping process. In particular, the comparisons between experimental and numerical FLDs
calculated with the proposed non-local damage model is shown in Figure 1. The occurrence of
ductile rupture has been recognised following the satisfaction of the modified Thomason’s
coalescence criterion while the necking (load instability) has been determined when the calculated
punch force reaches a maximum. For this Ni-based alloy, the material parameters found are:
144 M. Brunet et al.

anisotropy: r0 = 1.20; r45 = 1.12; r90 = 0.85, isotropic hardening: V 0 547 Mpa; Qf 1891 Mpa;
b 1.84, void growth: f 0 0.001 ; S0 0.12 ; nucleation: A 0 0.08 , void coalescence: F=0.31;
G=1.23 and O0 1.15 .

Figure 1: Experimental and numerical FLD’s (INCO 718)

References
1. Pardoen, T., and Hutchinson, J.W.: An extended model for void growth and coalescence, J.
Mech. Physics of Solids, Vol. 48(2000), 2467-2485.
2. Brunet M., Morestin F. and Walter H.: Damage Identification for Anisotropic Sheet-Metals
Using a Non-local Damage Model, International Journal of Damage Mechanics, Vol.
13(2004), 35-57.
 2T3. Ductile fracture 145

A NOVEL TECHNIQUE FOR EXTRACTING STRETCH ZONE FEATURES

FROM FRACTOGRAPHS

M. Tarafder, Swati Dey, S. Sivaprasad and S. Tarafder

National Metallurgical Laboratory, Jamshedpur-831007, India
mt@nmlindia.org, swati@nmlindia.org, shiva@nmlindia.org, star@nmlindia.org

The crack tip blunting process leaves an imprint in the form of stretch zone on fracture surfaces
during the event of ductile fracture. A schematic representation of the stretch zone in a fractured
specimen and a corresponding fractographs is shown in Fig. 1. A typical stretch zone has two
components, stretch zone width (SZW) and stretch zone depth (SZD). The SZW is basically the
virtual crack extension 'a and SZD is half of the CTOD, G. Stretch zones can be easily identified,
when observed under the SEM, since they have visually identifiable boundaries in between the
fatigue precracked region and the ductile fracture region

.
FIGURE 1. (a) Elevation profile of crack tip zone ( b ) SEM image of fracture surface
Previous authors [1-4] have estimated SZW and SZD and have attempted to establish their
correlation with the initiation fracture toughness of materials. A typical correlation of Ji with SZW
[1] is as following

Ji
SZW
2 k V f tan T

where Vf is the flow stress, 2T is the crack blunting angle and k is a material constant. Sivaprasad
et. al [2] have devised a correlation to compute SZD using

SZT  SZW cos D

SZD
sin D (2)
where SZT is the projection of the SZW after rotating the specimen at an angle Dand further used
SZD to obtain the fracture toughness of materials. They have estimated SZW and SZT by manual
estimation procedure, which is tedious, time consuming and lacks reproducibility and
repeatability. This paper reports an image analysis (IA) technique for on-line automatic
measurement of stretch zone on ductile fracture surfaces, and applies the technique on Cu-
strengthened HSLA steel for the determination of ductile fracture toughness.
146 M. Tarafder et al.

Fig. 2 shows a schematic diagram of the placement of fracture surface in the SEM at 0o and
tilted conditions with four key points identifying the locations of interest. Fig. 2 also includes a
typical image signal profile after processing. The schematic is an idealisation, which does not show
local uneven topographical features that are present in real specimens. By carrying out simple
addition and subtraction operations of the image signals, the spatial co-ordinates of the key points
P1, P2, P3 and P4 can be computed.

FIGURE 2. Schematic diagram of 0o and tilted positioning of fracture surface under SEM and the
identified key points in the image signals
Considering f0(x,y) and fT(x,y) as the post processed image signals of the fracture surface at 0o
and tilted conditions respectively and g0(x) and gT(x) are the corresponding vertical density
histograms, can be written as
y2 y2
g 0 ( x) ¦
y y1
f 0 ( x , y ), g T ( x ) ¦
y y1
f 45 ( x , y )
(3)
Assuming no translational shifts in x and z directions after the specimen is tilted, mathematically
the following conditions can be stated to identify the key points.

P 4 : Max ^A ( x , y ) g 0 ( x )  g T ( x )`
P 1 : ^S ( x , y ) g T ( x )  g 0 ( x ) ~ 0`
P 2 , P 3 : Min ^S ( x , y ) g T ( x )  g 0 ( x )` (4)
After identification of the stretch zone boundaries, linear profile in terms of pixels (between P1 and P2 for
SZW and P1 and P4 for SZT) are estimated using
SZW=F.Zand SZT= F.W
where F is the conversion factor from pixel to micron meter and Z and W is the width of the stretch zone in 0o
and tilted condition in terms of the number of pixels. It was found that this technique generates reliable data
which can be used for prediction of initiation fracture toughness of the HSLA material.
References
1. Bassim M.N., J. Material Processing Technology, vol 54, 109-112, 1995
2. Sivaprasad S., Tarafder S., Ranganath V.R., Das S.K. and Ray K.K., Met. Matallurgical and
Materials Transactions A, vol 33A, 2002.
3. Srinivas M., Malakondaiah G. and Rama Rao P., Acta Metall. Matter, vol 41:4, 1301-1312,
1993.
4. Rogerio de O. Hein L., Ammann J.J. and Nazar A.M.M., Material Characterization, vol 43,
21-30, 1999
 2T3. Ductile fracture 147

SIMULATION OF FATIGUE CRACK GROWTH BY CRACK TIP BLUNTING

P. Hutar and M. Sauzay

Commissariat a l’Energie Atomique (CEA)
Bat.455, 91 191 Gif-sur-Yvette, cedex France
hutar@ipm.cz

The crack tip blunting and re-sharpening on the crack tip is one of the basic mechanisms for
fatigue crack growth in ductile metals and alloys, e.g. Laird [1], Neumann [2]. Previous numerical
studies of crack tip blunting made by Mc Meeking [3], Needleman and Tvergaard [4] have been
carried out for monotonic loading. Subsequently, this kind of analysis has been made for cyclic
loading (Gu and Ritchie, Tvergaard and Hutchinson [5]), for the load ratio R=0. It is shown that is
possible to obtain fatigue crack growth, which can be comparable with Paris law. Perfect plasticity
and a finite element mesh with a small initial radius of the crack tip were used. Because of strong
mesh distortion they were applied only three full cycles.

FIGURE 1. Initial mesh of finite elements used for the calculations

Recently, Tvergaard [6] made a numerical simulations of the fatigue crack growth for 200 full
cycles, using remeshing at several stages of the cyclic plastic deformation, for three different
values of the loading ratio.
This study is mainly focused on the possibility of modelling this phenomenon in the case of
long and physically short cracks. The numerical modelling is based on elastic-plastic finite element
analysis with the code CASTEM. The possibility of comparing experimentally obtained results of
the fatigue crack growth with the numerical simulation of crack blunting in the first few cycles is
studied.
Because of practical interest, many authors are interested in finding the ratio between CTOD
and fatigue crack growth e.g. Tomkins [7]. Tvergaard [5], [6] proposed a relationship between the
cyclic change of CTOD and the growth of fatigue crack in this form:

da
ȕ .' CTOD,
dN (1)
148 P. Hutar and M. Sauzay

where ȕ # 0.33 for the chosen initial radius at the crack tip B and for the material model of
perfect plasticity (non-hardening material). The initial radius depends on the applied loading and
the yield stress of the material. Our simulation accurately studies this kind of relation, and shows
the limits of this ratio.
Technical materials have more complex behaviour than can be described by a material model
of perfect plasticity. Therefore, in our computations different models of material hardening were
used and the effect of the hardening on the crack tip blunting was discussed.
For illustration initial mesh of finite elements used for analysis can be seen in Fig.1. The
displacement fields are approximated in terms of plane strain 8-noded isoparametric elements. A
plate with a small edge crack with initial radius B is used. The crack length is approximately ten
5
times smaller than other dimensions of the specimen. The ratio a/B 10 is used. Around the
crack tip was a very fine mesh of finite elements (the size of the smallest elements was
approximately a/ 2. 10 7 ).
The results presented can contribute to a better understanding of one of the basic mechanisms
controlling fatigue crack growth in the case of ductile materials.

References
1. Laird, C., In Fatigue Crack Propagation, ASTM STP 415, ASTM, Philadelphia, 1967, 131
2. Neumann, P., Acta Metallurg., vol. 17, 1219, 1969
3. McMeeking, R. M., J. Mech. Phys. Solids, vol. 25, 357-381, 2004
4. Needleman, A., Tvergaard, V., ASTM STP 803, Philadelphia, PA, 1983, 80-115
5. Tvergaard, V., Hutchinson, J.W., Fatigue 2002, vol. 1/5, EMAS, UK, 2002, 107-116
6. Tvergaard, V., J. Mech. Phys. Solids, vol. 52, 2149-2166, 2004
7. Tomkins, B., Philosophical Magazine, vol. 18, 1041-1066, 1968
 2T3. Ductile fracture 149

LOADING RATE EFFECT ON DUCTILE FRACTURE

R. Chaouadi
SCK-CEN
Boeretang 200, 2400 Mol, Belgium
rchaouad@sckcen.be

The Charpy impact test is widely used to monitor the quality requirements of industrial processes.
It was also adopted by engineers and scientists to monitor material embrittlement resulting from
environmental effects like for example irradiation. In this work, we investigated the effect of
loading rate, namely quasi-static versus impact loading, on the ductile fracture behavior. Two low
alloyed steels used in the reactor pressure vessel industry were selected, namely A533B and
20MnMoNi55. These steels were extensively characterized from the flow, Charpy impact and
fracture toughness properties [1-2]. Figure 1 shows how the loading rate affects the ductile to
brittle transition curve for both materials. As can be seen, the major effect of loading rate is located
in the fully ductile regime where quasi-static loading requires significantly less energy to full
fracture than dynamic (impact) loading. Two temperatures, namely 25°C and 290°C were selected
to investigate these loading rate effects. At both temperatures, tensile and crack resistance
measurements were performed at both quasi-static and dynamic loading. For the A533B steel, at
25°C, the fracture is not fully ductile and therefore only tests at 290°C were considered for this
material. The results are shown in Table 1 for the various materials and conditions. These result
clearly show that empirical correlations [3] relating fracture toughness to Charpy impact energy
are not applicable without an in-depth analysis. Moreover, the loading rate effects on the crack
resistance cannot be solely attributed to the strain rate sensitivity of the material.

TABLE 1. Ductile crack initiation toughness.

150 R. Chaouadi

FIGURE 1. Effect of loading rate on the ductile-to-brittle Charpy transition curve.

References
1. Chaouadi, R. and Fabry, A., In From Charpy to Present Impact Testing, edited by D. François
and A. Pineau, Elsevier, 2002, 103-117.
2. Chaouadi, R., J. Test. and Eval., Vol. 32, No. 6, 469-475, 2004.
3. Rolfe S.T. and Barsom, J.M., Fracture and Fatigue Control in Structure – Application of
Fracture Mechanics, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1977.
 2T3. Ductile fracture 151

EXPERIMENTAL INVESTIGATION OF SLANT CRACK PROPAGATION IN

X100 PIPELINE STEEL

S. H. Hashemi, I. C. Howard1, J. R. Yates1, and R. M. Andrews2

The University of Birjand, Department of Mechanical Engineering, Birjand, IRAN
1The University of Sheffield, Department of Mechanical Engineering, Sheffield, UK
2Advantica, Loughborough, UK

shhashemi@birjand.ac.ir

Failure information from full-thickness burst experiments on long distance gas transportation
pipelines has shown that unstable fractures propagating in the pipeline axial direction are
dominated by ductile slant shearing [1-3]. Different test samples, e.g. Charpy impact [4], drop
weight tear test (DWTT) [5] and double cantilever beam (DCB) [6] have been proposed to study
the ductile shear crack growth in pipeline steels in a laboratory scale experiment. Because of the
design geometry of these specimens, slant crack growth is often preceded by flat fracture in the
specimen un-cracked ligament.

In this research the slant fracture characteristics of 52cc O.D x 21mm W.T pipeline made
from high-toughness steel of grade X100 was investigated using modified compact tension C(T)
$
specimens. Two sets of test samples were used in this study; smooth and novel side-grooved 45
slant notch C(T) specimens, see Fig. 1. The latter restraint specimen was used to maintain the
propagating crack in its original 45$ plane.

FIGURE 1. Design geometry of slant C(T) samples: a) smooth, and, b) side-grooved specimen

All test specimens were loaded in tensile opening mode I. In each test a crack initiated from an
initial 45$ through-the-thickness machined slit, and propagated in slant mode fracture through the
specimen ligament. This allowed simulation of the slant crack growth in the laboratory scale close
to the real structure.
The test data monitored in each experiment was load and crack mouth opening displacement
(CMOD). Crack length measurements were conducted using two methods; the current direct
potential drop (CDPD), and an optical technique. In the CDPD test the input and output leads were
located at the positions such that the Johnson's formula [7] for the relationship between crack
length and electric potential was applicable. In the latter a fine square mesh was scored on the side
surface of the test sample. The crack length measurements then were monitored using high-
resolution digital video camera. Fig. 2 is a photograph of the experimental set up
152 S. H. Hashemi et al.

.
FIGURE 2. Experimental set up used in slant C(T) testing of X100 pipeline steel

From the test results it appeared that the PD method (and Johnson formula, initially developed
for crack growth measurement in standard C(T) and 3PB specimens with straight cracks) could be
used for slant crack growth estimation. This was validated by comparing the final crack length of
the broken specimens in liquid nitrogen with those predicted by the Johnson formula, and with
those monitored by the video imaging system used.
The load-displacement plots of all test specimens showed substantial load drop after small
amounts of ductile slant crack propagation. The optical observation of broken test samples
revealed that the fracture surface of both smooth and side-grooved slant C(T) specimens contained
traces of quasi-cleavage fracture even though the test temperature was above the material transition
temperature. The sharp load drop was associated with this mode of fracture in the specimens
tested. From the test data the specific slant fracture energy (in terms of J / mm2 ) was estimated for
the X100 steel tested, and shown to be similar to that measured on DCB like specimen for the same
class of steel.

References
1. Rothwell, A. B., In Proceedings of Pipeline Technology, Edited by R. Denys, Elsevier
Science, 2000, 387-405.
2. Demofonti, G., Mannucci, G., Spinelli, C. M., Barsanti, L. and Hillenbrand, H. G., In
Proceedings of Pipeline Technology, Edited by R. Denys, Elsevier Science, 2000, 509-520.
3. Berardo, G. , Salvini, P., Mannucci, G. and Demofonti, G. , In Proceedings of the 2000
International Pipeline Conference, Vol. 1. New York, ASME, 2000, 287-294.
4. Leis, B. N., Eiber, R. J., Carlson, L. and Gilroy-Scott, A., In Proceedings of the International
Pipeline Conference, Vol. II, ASME, 1998, 723-731.
5. Wilkowski, G. M., Rudland, D. L., Wang, Y. Y., Horsley, D., Glover, A. and Rothwell, B., In
Proceeding of IPC’02 4th International Pipeline Conference, Alberta, Canada, 2002, 1-7.
6. Shterenlikht, A., Hashemi, S. H., Howard, I. C., Yates, J. R. and Andrews, R. M., Engineering
Fracture Mechanics, Vol. 71, 1997-2013, 2004.
7. Johnson, H. H., Materials Research & Standards, Vol. 5, 442-445, 1965.
 2T4. Nonlinear fracture mechanics 153

ESIS TC8 – NUMERICAL ROUND ROBIN ON MICRO MECHANICAL

MODELS : RESULTS OF PHASE III FOR THE SIMULATION OF THE
BRITTLE TO DUCTILE TRANSITION CURVE

C. Poussard and C. Sainte Catherine

Commissariat à l’Energie Atomique, DEN/DMN/SEMI
CEA Saclay, Bat. 625P, F-91191 Gif Sur Yevtte Cedex, France
Christophe.poussard@cea.fr, Claude.sainte-catherine@cea.fr

This paper presents the results of a round robin organized by CEA Saclay within the ESIS
(European Structural Integrity Society), TC8 (Technical Committee n°8) dedicated to the
numerical analysis and the comparison with experimental data. The objective of this round robin
(phase III) was to model the ductile to brittle transition curve of the 22NiMoCr37 German RPV
(Reactor Pressure Vessel Steel) using elastoplastic damage models. The principle of the round
robin is that initial sets of damage parameters are first imposed. Then, contributors are free to
adjust and justify their parameters so that a good latch to experimental data may be obtained.
A good participation was received for this round robin given the substantial technical efforts
that were required to contribute. Ten laboratories including nine disseminated in Europe as well as
one laboratory in Asia have contributed. Seven finite element codes have been used and three
participants performed full 3D fracture mechanics analyses accounting for ductile crack growth.
Six participants contributed to the three steps that were proposed in the specifications.
Differences have been obtained between the contributions for which explanations have been
suggested. The first step was dedicated to the prediction of a 1TCT specimen at -20°C. For that
step, no reference was made to experimental results and the purpose was to compare the codes with
each other. The mesh, boundary conditions and critical damage parameters were imposed to the
participants and resulted from previous phases of the round robin. Differences have been obtained
which have been primarily explained by the fact that participants used different elements, either 8
nodes reduced integration elements or 4 nodes fully integrated elements. For a given element size,
this appears to play a major role when predicting ductile crack growth. For the prediction of
cleavage fracture, differences were obtained for the plastic volume, due probably to the definition
of the threshold used to define the plastic strain but this had no influence when Weibull stresses
were compared. The differences were smoothed in step two when comparing the predictions to
experimental results and when adjusting critical damage parameters was made. Most contributors
agreed to the fact that the initial ductile damage parameters that were supplied to the participants
did not allow to predict the ductile crack growth behavior observed in the tests. Larger elements at
the crack tip and along the crack growth path and a smaller initial void volume fraction were
identified from the computation of 1TCT specimens. The mandatory Beremin parameters used
also in preceding phases of the round robin did not lead to satisfactory predictions of the failure
probabilities at higher temperatures (-40 and -20°C). The correlation was slightly improved when
accounting for a well identified ductile damage model but the results clearly showed that the
critical cleavage stress had to be diminished in order to satisfactorily predict the experimental data.
Finally, it was found that these models could lead to very good prediction of the transition
curve. The correlation was found initially poor when using the initially imposed mandatory ductile
and cleavage damage parameters but when identified parameters were used, a good correlation
could be obtained within the temperature range covered by the test matrix. The agreement between
the computation and the observation was further improved when using a linear dependant critical
cleavage stress with temperature. Figure 1 is an example of the quality of the results that can be
154 C. Poussard and C. Sainte Catherine

obtained with the local approach mechanical models over the transition. Between -150°C and -
20°C, a perfect match to the experimental data is obtained. At 0°C, the prediction is found to be
over conservative since in this example, only a 2D computation was made whilst a full 3D model
was required to model correctly ductile crack growth prior cleavage.

FIGURE 1. Prediction of the transition assuming a linearly dependant critical cleavage stress.

The results will now be used in order to support an ESIS guideline document entitled Guidance on
local approach of rupture of metallic materials, document that describes the state of the art to
apply the local approach to crack components.

Acknowledgements
The authors are very grateful to the round robin participants for the high quality level of the work
they have produced for that round robin and the useful discussions that arose when comparing the
results. Their patience for obtaining this synthesis is also gratefully acknowledged. The authors
also whish to express their gratitude to the GKSS staff in Hamburg and in particular to Wolfgang
Brocks, Juergen Heerens, Karl Schwalbe and Ingo Scheider not only for providing the data used in
that round robin but also for their continuous and fruitful collaboration within ESIS TC1 and TC8
over the past decade.
 2T4. Nonlinear fracture mechanics 155

CLOSURE OF A RECTANGULAR SKIN DEFECT VIA THE ADVANCEMENT

FLAP

C. Antypas, C. Borboudaki, V. Kefalas and D. A. Eftaxiopoulos

Department of Mechanics, School of Applied Mathematical and Physical Sciences,
National Technical University of Athens
Theocaris Building, Zografou Campus, 15773 Athens, Greece
eftaxiop@central.ntua.gr

The article is concerned with the closure of a rectangular skin defect, via the enlongation of the
advancement local skin flap. Local skin flaps are skin islands that are stretched, rotated or
transposed in order to cover a skin defect, which was created due to the removal of a diseased or
injured skin portion. The paper consists of two parts, namely the experimental part and the
computational part. In the experimental part, in – vitro tension experiments conducted on piglet
skin strips are described. The experimental results yielded the nonlinear stress – strain curves of
skin. In the computational part, a stress – strain curve obtained in the experimental part, is used in
a finite element model for the closure of a rectangular skin defect via the advancement flap. The
skin, the subcutaneous fatty tissue and the underlying muscle tissue are included in the model. The
muscle and the subcutaneous tissue mechanical responses, are obtained from the literature.
Nonlinear finite element analysis with large deformations is performed in the model. Several stress
distributions, within the skin layer, are obtained. Finite element results indicate that stress
concentrations at the points where the stitches are done, are created. On the rest of the area of the
advancement flap the stress distribution is almost uniform.
Borboudaki [1] did in vitro tension experiments on piglet skin strips and the nominal stress –
engineering strain curves for the skin were obtained. Antypas [2] did a nonlinear finite element
analysis of the closure of a rectangular skin defect, via the advancement skin flap. The skin, the
subcutaneous tissue and the underlying muscle were considered as separate materials. The article
is based on the work performed in [3] and [4].
For the tension experiments on skin, a male, two month old Duroc piglet, weighing 15 kg, was
sacrificed. Immediately after the animal sacrifice, a large single skin portion was removed from the
abdominal and the thoracic area of the animal. This skin island, was kept in Ringer’s solution. The
fatty tissue was removed from the skin island by using a surgical knife. Then four pairs of
dumbbell specimens, i.e. eight specimens overall, were cut from the skin island by using a
dumbbell cutter D412 of the ASTM standard. The specimens were kept in Ringer’s solution. Each
pair consisted of two strips, taken from symmetric positions, left and right from the two columns of
nipples, located in the chest and the abdomen of the animal. Two pairs consisted of longitudinal
specimens (long dimension along the spinal cord direction) and the other two pairs consisted of
transverse specimens (long dimension perpendicular to the spinal cord direction).
Eight simple tension experiments were performed, one for each specimen. The experiments
were performed by using an INSTRON 1140 machine. Prior to each experiment, the specimen was
placed flat on the working bench, and was regularly rinsed with Ringer’s solution. Two black
marker lines were ascribed on each specimen, perpendicular to the longitudinal axis of the
specimen. The lines were ascribed by using a drawing ink marker with 0.1mm thick tip and were
placed well inside the neck of the dumbbell specimens. The length between the two markers was
measured. Afterwards, eight rectangular pieces of emery cloth were glued on the ‘bell’ area of the
specimen. The moving grip of the INSTRON 1140 machine was removed and the specimen was
attached to the grip. The grip together with the specimen was attached to the machine and the other
156 C. Antypas et al.

end of the specimen was attached to the other grip. Then the upper grip was displaced such that the
length between the marker lines, became equal to the length between them as measured on the
working bench. Afterwards, the tension experiment was executed, with a grip displacement
velocity of 10mm/s. The experiment was terminated when the specimen slipped out of the grips. A
digital video recorder captured the displacement of the marker lines during the experiment and a
video editor was used for breaking down the movie into individual frames. By using the number of
pixels between the marker lines, the strain was measured at specific times. The strain was
corresponded to the tensile force, recorded by the testing machine at specific times. In turn the
force was converted to nominal stress via a division by the undeformed cross sectional area of the
specimen. For one pair of longitudinal and one pair of transverse specimens, the stress – strain
curves were quite close to each other. For the other two pairs of longitudinal and transverse
specimens, the stress – strain curves were not close to each other.
In the subsequent finite element analysis, a simplified flat three – dimensional model of the
advancement skin flap was constructed, by using the commercial package ANSYS 8.0. The skin
layer was meshed with shell elements, the subcutaneous tissue was meshed with one dimensional
nonlinear truss elements and the underlying muscle tissue was meshed with pentahedral solid
elements. The Multilinear Elastic (MELAS) material model of ANSYS was used, where points of
the experimental nonlinear stress - strain curve were introduced into the finite element model. Via
the MELAS option, only isotropic materials could be modeled. Large displacements were also
incorporated in the model. Appropriate displacement constraint boundary conditions were applied,
such that the distance between selected nodes on the edge of the skin flap and on the surrounding
skin edge, became equal to zero, i.e. these nodes were merged. In this way the stitching process for
the closure of a rectangular skin defect during plastic surgery, was simulated. In an area close to
the edges of the flap and the surrounding skin, the link elements were removed, for the
undermining of skin during plastic surgery to be simulated. Computational results indicated that
stress concentrations, at the points where the stitches were placed, were created. On the rest of the
area of the advancement flap the stress distribution was almost uniform.

References
1. Borboudaki C., Experimental and computational study of a skin strip under tension, Diploma
Thesis, School of Civil Engineering, National Technical University of Athens, 2004.
2. Antypas C., A computational study of skin as a composite material – The advancement flap,
Postgraduate Thesis, Interscholar Programme of Postgraduate Studies on ‘Computational
Mechanics’, National Technical University of Athens, 2005.
 2T4. Nonlinear fracture mechanics 157

SIMILARITY SOLUTIONS OF CREEP – DAMAGE COUPLED PROBLEMS IN

FRACTURE MECHANICS

L. V. Stepanova and M. E. Fedina

Department of Continuum Mechanics, Samara State University
Akad. Pavlov str., 1, 443011, Samara, Russia
lst@ssu.samara.ru

The asymptotic solution to Mode III and Mode I crack problems in a creeping solid in the
framework of Continuum Damage Mechanics is presented. The kinetic law of damage evolution is
the Kachanov – Rabotnov equation [1]. The damage parameter is incorporated into the power-law
creep constitutive equations. Thus the coupled system of damage mechanics – creep theory
equations is considered. It is to be expected during such a coupling process that the damage
involved gives a great effect on the stress field near the crack tip and when the damage parameter
reaches its critical value a totally damaged zone near the crack tip may occur [2-4].
The present contribution is an attempt to obtain asymptotic fields of stress, creep strain rate and
continuity near Mode III and Mode I cracks in creeping damaged materials as functions of the
similarity variable initially proposed by Riedel [5] under the assumption that the totally damaged
zone in the vicinity of the crack tip does really exist. The shape and the characteristic length of the
totally damaged zone are not a priori and should be obtained as a part of the solution. The
instantaneous response of the materials characterised by the creep power-law constitutive
equations and the power-law damage evolution equation is non-linear viscous and the evolution of
damage at short times after load application can be analysed under the remote boundary conditions
that the stress field must approach the Hutchinson – Rice – Rosengren field of non-linear fracture
mechanics. If the HRR-field represents the initial conditions and the remote boundary conditions
then dimensional analysis shows that the damage mechanics equations must have similarity
solutions [5]. Since the classical continuum mechanics equations can not be valid inside the totally
damaged zone the asymptotic solution to the problem is sought at large distances from the crack tip
(at large distances as compared with the characteristic length of the totally damaged zone but at yet
still small distances as compared with the crack length, with the characteristic length of the cracked
body).
The asymptotic expansions of the continuity (integrity) parameter \ and the effective stress
Vij /\ (the stress referred to the surface that really transmits the internal forces) for large
distances R o f from the crack tip, where R is the similarity variable, have the following form

\ ( R , M ) 1  R J g ( 0 ) (M )  R J g (1) (M )  ... ( J , J 1  0 ),
1

V ij
( R , M ) R s f ij( 0 ) (M )  R s f ij(1) (M )  ... ( s , s1  0 ).
1

\ (1)
The two-term asymptotic expansion for the effective stress and the three-term asymptotic
expansion for the continuity (integrity) parameter are obtained. The construction technique of the
far-field stress asymptotics is elucidated. The asymptotic fields allow to find the geometry of the
totally damaged zone. The geometry of the totally damaged zone for different values of the
material constants is given and analysed (Figures 1 and 2). It is found that the HRR-solution for the
creep constitutive equations can not be used as the remote boundary conditions and the actual
stress field at infinity is found.
158 L. V. Stepanova and M. E. Fedina

FIGURE 1. The geometry of the totally damaged zone in the crack tip region of a Mode I crack
under plane strain conditions: 1 – the configuration given by the two-term asymptotic expansion of
the continuity parameter, 2 – the configuration given by the three-term asymptotic expansion of the
continuity parameter.

FIGURE 2. The geometry of the totally damaged zone in the crack tip region of a Mode I crack
under plane stress conditions.

References
1. Kachanov L.M., Introduction to Continuum Damage Mechanics, Martinus Nijhoff,
Dordrecht, Boston, 1986.
2. Zhao J., Zhang X., Engn. Fracture Mechanics, vol. 50, 131-141, 1995.
3. Zhao J., Zhang X., Int. J. of Fracture, vol. 108, 383-395, 2001.
4. Astafjev V.I., Grigirova T.V., Pastuchov V.A., In Mechanics of Creep and Brittle Materials,
edited by A.C.F. Cocks, A.R.S. Ponter, Elsevier, London. 1991, 49 -61.
5. Riedel H., Fracture at High Temperature, Springer, Berlin, 1987.
 2T4. Nonlinear fracture mechanics 159

IMPACT FRACTURE TOUGHNESS DETERMINATION OF DUCTILE

POLYMERS BY SPB METHOD

J. Wainstein , L. A. Fasce and P. M. Frontini

Instituto de Investigaciones en Ciencia y Tecnología de Materiales –INTEMA-,
Univ. Nac. de Mar del Plata - CONICET;
Juan B. Justo 4302, B7608FDQ, Mar del Plata, Argentina.
pmfronti@fi.mdp.edu.ar

Fracture toughness of ductile materials is often characterized by the J parameter that was
developed from the J-integral concept. The determination of a critical value of the J-integral is
generally performed through the construction of the resistance curve J-'a of the material. The
commonly used method for this purpose is the multiple specimen technique [1] in which several
specimens are loaded to obtain different amounts of crack growth. However, this method is very
difficult to apply at high rate conditions because of the need of interrupting the test at different
crack growth levels. Recently, a new extremely simple single specimen method has been
developed from the separation parameter, Spb, and successfully applied in fracture toughness
characterization of metals and ductile polymers [1,2]. It consists on the assumption that the load
can be separated into two multiplicative functions: the geometry (G) and the deformation (H)
functions. The separation parameter, Spb, is defined as the load ratio of a sharp and a blunt notched
specimen of the same material, geometry and constraint. Assuming that the load falls in
correspondence with stable crack propagation, the load drop after maximum load can be taken as a
symptom of crack growth. From the Spb expression (Eq. 1) and counting with at least two
calibration points (in order to assess the m parameter in Eq. 1), a simple relationship between the
load and the crack growth length can be simply obtained and then the J-'a curve can be evaluated.
m
§ ap · § ap ·
G p ¨¨ ¸¸ ¨¨ ¸¸
P a ,Q 
m
p p ©W ¹ ©W ¹ § ap ·
S pb ¨¨ ¸¸
P a ,Q  §a · § ab ·
m
© ab ¹
b b Q Gb ¨ b ¸ ¨ ¸ Q
©W ¹ ©W ¹
Q (1)
The Spb method may appear very similar to the Normalization method; though, Spb method has the
appealing advantage of requiring the assumption of only one hypothesis, i.e. a geometry function,
which indeed is well known for several specimen configurations. In the proposed Spb method, the
load separation principle is applied to total displacement with the advantage of non-subtracting the
elastic component of the displacement, which results difficult to evaluate from the load-
displacement impact records.
In this work the capability of the Spb method in determining the impact fracture toughness of
ductile polymers is studied. The methodology is assessed for two commercial grade polymers that
exhibit ductile impact fracture behavior: acrylonitrile-butadiene-styrene terpolymer ABS (Lustran
ABS-640 HR 850) and polypropylene block copolymer PPBC (Tipplen CS 2-8000). Stes
containing two records: one of a sharp-notched specimen and one of a blunt notched specimen are
employed in the application of the Spb method. Fracture tests are performed at 1m/s in a Fractovis
Ceast Falling Weight type machine and they are conducted up to the point of complete failure of
the specimen so that the final crack length is not available as a calibration point. In order to
determine the “m” parameter in Eq. 1, two points of the load displacement record for which crack
lengths are known are used. The first one corresponds to the initial crack length where the Spb
160 J. Wainstein et al.

parameter is constant (solid arrow in Fig. 1-b and c); the initial crack length can be measured on
the fracture surface of the broken specimen. The second point is taken under the assumption that
when the sharp specimen achieves the same crack length as the blunt notched specimen, both are
bearing the same load and the Spb parameter is equal to one [3] (dash arrow in Fig. 1-a and c). The
steps of the proposed Spb method are illustrated in Fig1.

FIGURE 1. Steps in Spb method for PPBC at 1m/s. a) Load displacement records of sharp and
blunt notched samples; b) Spb parameter; c) Estimated crack length.

J-R curves arisen from the application of the Spb method for triplicate set of samples are
shown in Figure 2. Plane strain fracture toughness values (JIc) derived from the fitted average
curves are 7.5 N/mm and 5.4 N/mm for ABS and PPBC, respectively. These fracture parameters
will be compared with the fracture toughness parameters evaluated by the essential work of
fracture approach (EWF)

.
FIGURE 2. J-R curves determined by the Spb method for a) ABS and b) PPBC.

The appealing of the Spb method is basically its simplicity. It seems to be applicable for many
polymers and it will result helpful for that cases in which the multiple specimen technique is hard
to performed such as at high loading rates, high temperatures or aggressive environments.

References
1. Wainstein J., de Vedia L., Cassanelli A.. Eng. Fracture Mechanics, 70, 2489-2496, 2003.
2. Wainstein J., Frontini P., Cassanelli A., Polymer Testing, 23, 591-598, 2004.
3. Kobayashi T., Yamamoto I., Niinomi M., JTEVA, 21, 145-153, 1993.
 2T4. Nonlinear fracture mechanics 161

A MICRO-TOUGHNESS MODEL FOR DUCTILE FRACTURE

K. Srinivasan, Thomas Siegmund and Otmar Kolednik1

School of Mechanical Engineering
Purdue University, West Lafayette, IN 47907, U.S.A.
1Erich Schmid Institute of Materials Science, Austrian Academy of Sciences, A-8700 Leoben,

Austria
ksriniv@ecn.purdue.edu, siegmund@ecn.purdue.edu, kolednik@unileoben.ac.at

Ductile fracture occurs through growth and coalescence of micro-voids that originate at the
location of inclusions and precipitates. The plastic work dissipated in these micro-separation
processes leading to the creation of a unit fracture surface area is a measure of the micro-toughness
of the material. Furthermore, void growth and coalescence processes are accompanied by plastic
deformation of the material surrounding the voids. The energy dissipated by the micro-separation
processes of void growth and coalescence, and the plastic deformation in the bulk material
surrounding the voids together contribute to the overall fracture toughness of the material.
Conventional fracture toughness tests fail to individually measure these two very different
contributions. As a result, there is limited transferability of fracture toughness test data from the
laboratory to an actual structure.
To overcome this problem there is a need for measurements that allow for the determination of
the micro-toughness in ductile fracture. The transferability of material toughness and crack growth
resistance can then be solved with a non-linear analysis model like a cohesive zone model (CZM).
If the micro-toughness can be determined experimentally then only the cohesive strength remains a
free parameter, and the CZM can then be easily connected to experiments. Furthermore,
measurements of micro-toughness are of fundamental interest in the development of materials with
improved crack growth resistance.
A micro-toughness model using measurements of the ductile fracture surface topology to
estimate the plastic work dissipated in the process of the formation of a dimple fracture surface was
proposed in [1]. This model provides the micro-toughness, * , (also called the cohesive energy) as
* SV H where V is a flow stress, H is the dimple height, and S is a shape factor [1]. The
dimple height can be obtained by topographic measurements performed by taking stereo image
pairs in a scanning electron microscope and analyzing them using a digital image analysis system
[2]. Detailed investigations of void growth processes reveal that the model of [1] underpredicts the
micro-toughness in ductile fracture. This issue is overcome by the present work by developing a
new micro-toughness model.
The micro-toughness involves the use of accurate expressions for the effective plastic strain,
H , fields and a numerical integration procedure to calculate the micro-toughness. The model
incorporates the initial spherical expansion of the void as well as the subsequent lateral void
expansion stage. A criterion for the transition from the spherical void growth to a lateral void
expansion is incorporated [3], and a criterion on final void link-up based on a critical inter-
ligament plastic strain, H , is used. The micro-toughness is then given from the following
parameters

* *  H , ri , V Y , n, H c 
(1)
162 K. Srinivasan et al.

where ri is the inclusion size, V Y yield strength, n hardening. Figure 1(a) depicts predicted
values of the micro-toughness in dependence of the dimple height. Good agreement between the
model and corresponding FEM simulations are obtained for the individual stages of void growth,
* 1 and * 2 as well as for the total micro-toughness. Practical examples of the application of the
micro-toughness model are discussed for two high strength steels (St37 and V720), as well as an
aluminium metal matrix composite [4].
The micro-toughness model is also applied to study the effects of dimple size on micro-
toughness. For dimple sizes approaching the material internal length scale, l, of plasticity (set
mainly the dislocation spacing) micro-toughness is calculated through a strain-gradient plasticity
model [5] in combination with the micro-toughness model. The material flow stress is then given
as V f (H )  lK with f (H ) the hardening function, and K the strain gradient. Figure 1(b)
demonstrates that for small dimples (dimple radius R0 < 10 Pm) such strain gradient effects need
indeed be accounted for in order to accurately predict the micro-toughness. The implications of
the finding to inter-granular ductile fracture are discussed.

(a) (b)
Figure 1: (a) Micro-toughness in dependence of dimple size. (b) Dimple size dependence of
micro-toughness, comparison of conventional plasticity and strain gradient plasticity.

References
1. Stüwe, H.P., Eng. Fract. Mech., vol. 13, 231-236, 1980.
2. Stampfl, J., Scherer, S., Gruber, M. and Kolednik, O., Appl. Phys., vol. A63, 341-346, 1996.
3. Pardoen, T. and Hutchinson, J.W., J. Mech. Phys. Solids, vol. 48, 2467-2512, 2000.
4. Miserez, A., Rossoll, A. and Mortensen, A., Acta Mater. vol. 52, 1337-51, 2004.
5. Huang, Y., Qu, S., Hwang, K.C., Li, M., Gao, H., Int. J. Plast., vol. 20, 753-782, 2004.
 2T5. Fatigue and fracture 163

CRACK COALESCENCE MODELLING OF FSW JOINTS

A. Ali, M. W. Brown1 and Chris A. Rodopoulos2

Department of Mechanical and Manufacturing
Engineering Faculty, Putra University Malaysia, 43400, Serdang Selangor , Malaysia
1Department of Mechanical Engineering

University of Sheffield , Sir Frederick Mappin Building, Mappin Street

Sheffield S1 3JD,United Kingdom
2Structural Materials and Integrity Research Centre,

Materials and Engineering Research Institute, Sheffield Hallam University, City Campus, Howard
Street, United Kingdom
aidy@eng.upm.edu.my, M.Brown@sheffield.ac.uk, C.Rodopoulos@shu.ac.uk

In the present work, Friction Stir Welds (FSW) of 2024-T351 aluminium alloys is characterised in
terms of macrostructure, microstructure, hardness, precipitate distribution, and weld residual
stress. Cyclic properties and fatigue endurance of the FSW joints are also investigated and
discussed. Critical areas for natural fatigue crack initiation in FSW are pinpointed. The fatigue
mechanism in FSW is identified to follow a multiple crack coalescence nature (Figure 1). The
number of cracks participate in coalescence and the resulting crack growth rate is governed by the
dynamic distance between the crack tips from crack initiation to coalescence. The above represents
a complex condition for modelling.
Based on crack growth and characterisation of FSW joints, a modified version of the Hobson-
Brown [1-3] is adopted. The good correlation achieved between the experimental data and the
model predictions is shown in Figure 2. The model can successfully handle multiple cracking and
coalescence as well as modelling crack growth within residual stress fields.
The model follows the damage tolerance design philosophy, which is widely accepted in
aircraft manufacturing. The damage tolerance approach incorporates knowledge of fatigue crack
detection, initial crack length, propagation direction, fatigue crack growth rate and maximum crack
length that leads to inspection time intervals.

Figure 1. Replicas images for LCF of 2024-T351 AL Alloy FSW polished mirror specimen at
300MPa applied stress with load ratio R=0.1 showing the coalescence of cracks.
164 A. Ali et al.

Figure 2. Life prediction of FSW 2024-T351 Al Alloy polished mirror.

References
1. P. D. Hobson, M.W. Brown., E. R. de Los Rios, Two phases of short crack growth in a
medium carbon steel. Behaviour of shorth fatigue crack(Edited by K. J. Miller & Eduardo),
1986: p. 441-459.
2. Shaikh, Z., Initiation, Propagation and Coalescence of Short Fatigue Crack in AISI 316
Stainless Steel at 200C and 5500C. PhD. Thesis, Sheffield University, 1991.
3. Gao, N., M. W. Brown, Short Crack Coalescence and Growth in 316 Stainless Steel
Subjected to Cyclic and Time Dependent Deformation. Fatigue and Fracture Engineering
Material Structure, 1995. 18(12): p. 1423-1441.
 2T5. Fatigue and fracture 165

FATIGUE CRACK INITIATION IN A TWO PHASE %-METASTABLE TITANIUM

ALLOY: INFLUENCE OF MICROSTRUCTURAL PARAMETERS

A. Lenain, P.J. Jacques and T.Pardoen

Département des Sciences des Matériaux et des Procédés, Université catholique de Louvain,
IMAP, Place Sainte Barbe 2, B-1348, Louvain-la-Neuve, Belgium
lenain@imap.ucl.ac.be, jacques@imap.ucl.ac.be, pardoen@imap.ucl.ac.be

A new generation of DEtitanium alloys with enhanced performance over density ratio is currently
receiving lots of attention from the aerospace industry. One of the critical parameter is the fatigue
resistance in the both low and high cycle fatigue (LCF and HCF) regimes, in particular the crack
initiation. Fatigue initiation is strongly influenced by several characteristics of the microstructure
such as grain size, grain orientation and misorientation (e.g Jin et al. [1]). Up to now, the initiation
of fatigue cracks in the new low cost beta (LCB Ti alloys) has been characterised only in the case
of a fully E microstructure (Hu et al. [2], Krupp et al. [3], Floer et al. [4]). Furthermore, hardly
anything can be found in the literature on the fatigue crack initiation in multiphase Ti-based
microstructures.
This study aims at understanding the couplings between cracking initiation and microstructural
features in the DE TIMETAL LCB titanium alloy under cycling loading.
Fatigue tests were carried out on notched specimens presenting either equiaxed (Fig. 1 (a)) or
acicular (Fig. 1(b)) morphologies. The first microstructure is obtained by hot rolling and
subsequent annealing at 760°C and the second one is obtained by continuous cooling from 810°C
at a speed rate of 2°C/min. These two microstructures were first characterized by SEM, XRD,
EBSD and TEM.
Fatigue tests are regularly stopped and sheet specimens are observed with the FEG-SEM.
Results, obtained on the equiaxed microstructure (Fig. 1 (a)), show that slip lines appears, mostly
in the E grains (Fig. 2 (a)). The density increases with the number of cycles. These slip lines
constitute privileged damage sites. The orientation of the E grains in which slip lines first appear
has been determined. Fig. 2 (b) shows that there is a misorientation profile across two gliding lines
in a E grain which is explained by the lattice rotation due to a deformations gradient into the grain.
166 A. Lenain et al.

References
1. 1. Jin O., Mall S., Materials Science and Engineering A, vol 359, p. 356-367, 2003.
2. Hu Y.M., Floer W., Krupp U., Christ H.-J., Materials Science and Engineering A, vol. 278, p.
170-180, 2000.
3. Krupp U., Floer W., Lei J., Hu Y., Christ H.-J., Philosophical Magazine A, vol. 82, No 17/18,
p. 3321-3332, 2002.
4. Floer W., Krupp U., Christ H.-J., Schick A., Fritzen C.-P, In Proceedings of the Eighth
International Fatigue Congress, edited by A.F. Blom, Stockholm, 2002, p. 2369-2376.
 2T5. Fatigue and fracture 167

EFFECTS OF SPECIMEN TYPE, SIZE AND MEASUREMENT TECHNIQUES

ON FCGR

B. Kumar and J. E. Locke

National Institute of Aviation Research & Wichita State University
1845. N Fairmount St, Wichita KS 67260
brijesh.kumar@wichita.edu, James.locke@wichita.edu

The literature on fatigue and fracture is extremely extensive, mainly due to the large number of
variables involved in the process. It is not possible to look at all the possible variables involved in
the fatigue process. The present study is focussed on some of the aspects involved in generating
fatigue crack growth data namely: specimen type, size, measurement techniques and test methods.
Schjive [1] mentioned several reasons for conducting fatigue test programs such as:
comparative experiments; compiling basic material data; verification of prediction models and
research on fatigue and fracture phenomena. The choice of specimen type is usually determined
by specimen production; ease of carrying out experiments, reproducibility, accuracy of test results,
and comparability to other test programs.
The specimen types that have been and are being investigated can be classified into the
following types:
1 Symmetric specimens
a. Middle crack tension specimen M(t)
b. Double edge notch tension specimen
2 Asymmetric specimens
a. Compact tension C(t) specimens
b. Eccentrically loaded single edge coupon ESE (t)

The specimen width’s being considered are 2.00” & 3.00” for both the thin and thick sheet
materials. The thicknesses range from 0.040” to 0.25”. The material’s being investigated are are
2024-T3 & T3511, 7050-T7451, 7075-T6 &T7351 and 7475-T7351 Aluminum alloys. They were
so chosen as they are widely used by the aircraft industry.
The test data has been generated using several different measurement techniques: optical
microscope, clip gage, Fractomat™ based Krak gages, and the ACPD system to monitor crack
growth. All these instruements provide crack growth resolution higher than required by ASTM
647E. The figures 1 & 2 show the crack length measurement data for 7475 and 7050 Al alloys.
Though there seems to be some variation in the reading between the optical microscope and the
clip-gage. We are testing some thicker specimens to see if we still see a difference between them.
There was very good agreement with the Krak Gage and the optical reading. The differences as a
result of the measurements are being analyzed and the results will be presented in da/dn versus
Kmax or K effective.
168 B. Kumar and J. E. Locke

Figure 1 & 2: Crack length versus number of cycles for Al 7475 and 7050 (t = 0.10”) ESE (t)
specimen using clip gage, optical microscope and Krak Gages.

There are several crack initiation techniques recommended, and their subsequent effect on
crack propagation data is still a source of concern and debate. Initial testing using the guidelines
provided in ASTM 647E, which recommend crack initiation to occur in no less than 1.0 x105
cycles has been followed. Crack growth data shows considerable variability near the threshold
regime Marci [2 & 3], this is because of the effects of the R-ratio, testing methods and specimen
types. It would seem that the differences in the data generated is dependent on of the above
mentioned variables. Though all the variables are beyond the scope of this investigation we are
will conduct some limited testing to see the effects of the testing methods. Most of the testing used
in this investigation is based on the constant Pmax loading with two stress ratio’s R =0.5 and R
=0.1. The K reducing method to determine FCGR is also planned.
Finally results from the testing of the M(t) , C(t), ESE(t) and the DENT double edge notch
specimens using different techniques will be reviewed and presented.

References
1. J. Schjive, “Fatigue Specimens for Sheet and Plate Material”, Fatigue & Fracture of
Engineering Materials & Structures, vol 21, pp 347-357, 1998.”
2. G., Marci, D. E., Castro, V., Bachmann, “Fatigue Crack Propagation Threshold”, Journal of
Testing and Evaluation JTEVA, vol. 17, No. 1, Jan. 1989.
3. G., Marci, “Non-propagation conditions (Kth) and Fatigue Crack Propagation Threshold
(KT)”, Fatigue Fracture Engineering Mater. Struct. vol. 17, No 8, pp 891-907, 1994.
 2T5. Fatigue and fracture 169

THE EFFECT OF STRESS RATIO ON FATIGUE SHORT CRACKING

C. A. Rodopoulos and S.-H. Han

Structural Materials and Integrity Research Centre, Materials and Engineering Research Institute,
Sheffield Hallam University, City Campus, Howard Street, United Kingdom
C.Rodopoulos@shu.ac.uk
Structural Safety Group, Korea Institute of Machinery & Materials,
171 Jang-dong, Yusung-Gu, Daejeon, 305-343, Republic of Korea
seungho@kimm.re.kr

Short or Stage I crack growth represents a unique case where the Paris-Erdogan Linear Elastic
Fracture Mechanics model fails to predict crack growth and hence fatigue life. The propagation
rate of short cracks has been found in numerous works and for a variety of materials to be
significantly higher than that predicted by the aforementioned model [1,2]. Yet, there is a strong
relation between the material and the extent of short cracking. In general, aluminium alloys exhibit
a much larger short cracking potential to high strength steels. In addition, stress ratio has also been
found to play an important role on such phenomenon [3]. In brief, the extent of short cracking has
been found to decrease with the stress ratio and even disappearing after a particular value [4]. In a
number of papers Rodopoulos and co-workers argued that the phenomenon depends on the ratio
between the fatigue limit and the cyclic yield stress [5], as shown in Figure.1.

FIGURE.1: Extent of fatigue short cracking (iint) versus the ratio between fatigue limit and cyclic
yield stress.

Such argument was based on examining the point where the transition line between a Stage I
and a Stage II (long crack) crosses the line representing transition between a propagating and a
non-propagating crack (Kitagawa-Takahashi diagram), see Figure 2.
170 C. A. Rodopoulos and S.-H. Han

FIGURE.2: Condition for evaluating the extent of short cracking. The solid line represents the
transition between short and long crack growth and the dashed line represents conditions for crack
arrest.

In this work, a sensitivity analysis is presented after the above transition lines have been
modified to include the effect of stress ratio. The work concludes with the development of material
maps versus stress ratio and the tendency towards short cracking.

References
1. Pearson, S. Engng., Fract. Mech. Vol. 7, 235-247, 1975.
2. Morris, W. L. Metall. Trans., A84, 589-596, 1976.
3. Edwards, P. R., Newman, J.C. AGARD R-767, 1990.
4. C. A. Rodopoulos, E. R. de los Rios, Facta Universitätis, Series Mechanics, Vol. 3, 13, 647-
655, 2003.
5. C. A. Rodopoulos and E. R. de los Rios, Inter. J. of Fatigue, 24, 719-724, 2002.
 2T5. Fatigue and fracture 171

DWELL-FATIGUE BEHAVIOUR OF A BETA-FORGED TI 6242 ALLOY

P. Lefranc, C. Sarrazin-Baudoux and V. Doquet2

LMPM – UMR CNRS 6617, ENSMA, 86961 Futuroscope-Chasseneuil Cedex, France
2LMS, Ecole Polytechnique, Palaiseau, France

lefranc@lmpm.ensma.fr, baudoux@lmpm.ensma.fr, doquet@lms.polytechnique.fr

For over three decades, gas turbine industries have been confronted to the phenomenon of ambient
temperature dwell fatigue sensitivity in titanium alloys. This effect consists of a reduction of
fatigue lives due to the introduction of dwell periods at the peak stress of the cyclic loading. Up to
now, no real mechanism can explain such behaviour.
During the last three decades, many studies [1] were performed to investigate the influence of
parameters such as microstructure, temperature, dwell period's length and hydrogen content on this
dwell-fatigue behaviour. Characteristic features are a significantly increase in dwell sensitivity
with a coarse lamellar microstructure or with an increase in hydrogen content and the
disappearance of dwell susceptibility in the temperature range of 150°C and 200°C.
This paper deals with a study of the influence of microstructural anisotropy on the dwell-
fatigue behaviour and with the influence of the dwell period introduction at the peak stress of the
cyclic loading on the crack initiation. So, cyclic and two minutes dwell tests were performed at
room temperature on a near alpha Ti-6242 forged above the beta transus. This kind of forging plus
heat treatments lead to a coarse lamellar microstructure either aligned in prior beta grains or basket
weaved microstructure.
Cylindrical specimens used in this study have a utile length of fifteen millimetres and a
diameter of four millimetres. Tests were conducted at different stress levels leading to lives
ranging between 104 to 105 cycles. Some specimens were instrumented with acoustic detection
gages in order to detect initiation.
The results show that a same dwell life debit ranging between 3:1 and 5:1 can be noticed for
the both specimen orientations. But one of the specimen orientations has shown cyclic and dwell
lives two times greater than the other one. Specimens with the weaker cyclic and dwell lives have
also shown up a lot of cracks along its external surface.
Finally, MEB observations have been performed in order to precise the cracking process. The
fracture surface morphologies for cyclic and dwell loading are respectively characterised with a
particular attention to the localisation of the ignition sites. Then, these morphologies are related to
the crack growth rate as determined from striations spacing and using relationships from the
literature which relate the crack propagation rate to the stress intensity factor or the local plastic
deformation for short and long cracks.
The mechanisms controlling the dwell effect are finally discussed on the basis of these results
and observations confronted to the available literature [1-6].

References
1. M. R. Bache. A Review of Dwell Sensitive Fatigue in Titanium Alloys : The Role of
Microstructure, Texture and Operating Conditions, International Journal of Fatigue, Vol. 25,
pp 1079-1087, 2003.
172 P. Lefranc et al.

2. D. Eylon,, J. A. Hall., Fatigue behaviour of E-processed titanium alloy IMI 685, Metallurgical
Transactions, Vol. 8A, pp. 981-990, 1977
3. W. J. Evans, C. R. Gostelow. The effect of hold time on the fatigue properties of D/E-
processed titanium alloy, Metallurgical Transactions, Vol. 10A, pp .1837-1846, 1979.
4. J. E. Hack and G. R. Leverant. The influence of microstructure on the susceptibility of
titanium alloys to internat hydrogen embrittlement, Metallurgical Transactions, Vol. 13A, pp.
1729-1738, 1982.
5. M. R. Bache. M. Copet, H. M. Davies, W. J. Evans and G. Harrison Dwell sensitive fatigue in
a near D titanium alloy at ambient temperature, International Journal of Fatigue, Vol. 19, Sup.
No l, pp S83-S88, 1997.
6. M. E. Kassner, Y. Kosaka and J. A. Hall. Low-cycle dwell-time fatigue in Ti-6242,
Metallurgical and Materials Transactions, Vol. 30A, pp. 2383-2389, 1999.
 2T5. Fatigue and fracture 173

INVESTIGATION INTO FATIGUE LIFEOF WELDED CHEMICAL PIPELINES

Cz. Goss and L. Sniezek

Military University of Technology
00-908 Warszawa, ul. Kaliskiego 2, Poland
l.sniezek@wme.wat.edu.pl

Industrial pipelines, in particular the ground-based and overhead ones within any chemical works,
demand special attention owing to hazards to the employees, population in the surrounding area,
and environment. Hazards are greatest while transferring dangerous media between manufacturing
departments and during the storage thereof. Hence, there is a permanent need to deal with the
problems of pipeline strength [1-2].
Failures/damages are usually attributable to, e.g. corrosion, design errors, flaws, and welding
notches. The latter ones are closely related with any discontinuities and changes in the shape of the
pipeline’s cross-section, which often become potential locations of crack initiation due to stress
concentrations.
The states of strain and stress in the bottom of a welding notch, i.e. the point of fatigue
cracking initiation, are determined with account taken of the fact that the material can get
plasticized even if the rated stresses (beyond the notch) are lower than the yield point of the steel.
Such being the case, one should aim at defining the effort of the material in the region of the
welding notch [3-4].
In the course of dimensioning the welded joints within the range of low-cycle fatigue strength,
the so-called ‘notch-bottom method’ is used. In the bottom of the notch there are stress and strain
concentrations occurring at the same time. They can be produced owing to some change(s) in the
shape of the joint itself and the shape of the weld; also, due to welding defects in the forms of
undercuts, complete fusion, no weld penetration, and to residual stress that remains after
manufacturing processes needed to fabricate the structure, e.g. after cutting, bending, welding, etc.
The effort in the bottom of the notch (in a transverse weld this would be the so-called line of
fusion, i.e. the native material and that of the weld penetrating each other) is practically found by
means of computations only.
Estimated were pipelines made of the 1H18N9T steel: those after thirty years’ service, and the
‘new’ ones. Chemical compositions of the steel used to construct these pipelines are shown in
Table 1. Table 2 presents static and cyclic mechanical properties of the steel.

TABLE 1. Chemical composition of the 1H18N9T steel.

Pipeline C Si Mo Ti Cr Mn Ni
% by weight
„old” 0,05 0,64 0,21 0,63 18,5 1,2 9,8
„new” 0,04 0,8 0,4 0,4 17,88 1,8 8,89
174 Cz. Goss and L. Sniezek

TABLE 2. Material properties.

The computational model presented in the paper provides capability to determine the effort of
the material (i.e. of the steel used to manufacture the chemical pipeline) in the area of the welding
notch. On the grounds of both the literature data and experimental work carried out by the Authors,
analytical dependences have been formulated to calculate then the elastic-plastic range of strain in
welding notches in pipelines operated for thirty (30) years, and in the ‘new’ ones. Computations
were made with the strain criterion applied, i.e. the effort of the material in the welding-notch area
was determined. A comparison between the range of strain in the welding notch (with the range of
nominal stresses assumed) and the destructive strain of steel characterised with the computational
range of strain has been accepted as the basis to calculate load capacity of welded joints for the
low-cycle range.
Some formulae have been suggested in the paper to calculate computational, low-cycle fatigue
strength expressed with the range of total strain for a pre-set number of loading cycles. On these
grounds formulae have been introduced to facilitate computations of service lives of pipelines. The
service lives are defined with the number of loading cycles Nf up to the crack initiation, with the
range of nominal stresses in the pipeline coating assumed to be 'VN. The calculated values of
fatigue lives for pipelines after many years’ operational use, with R = 0,1 assumed, are lower by 60
- 80% as compared to ‘new’ pipelines.

References
1. DeWolf G. B.: Process Safety Management in the Pipeline Industry: Parallels and
Differences Between the Pipeline Integrity Management (IMP) rule of the Office of Pipeline
Safety and the PSM/RMP Approach for Process Facilities. Journal of Hazardous Materials.
vol. 104, 169-192, 2003.
2. Papadakis G. A.: Major Hazard Pipelines: a Comparative Study of Onshore Transmission
Accidents. Journal of Loss Prevention in the Process Industries. vol. 12, 91-107, 1999.
3. Samuelsson J.: Design and Analysis of Welded High Strength Steel Structures. EMAS.
Stockholm, Sweden, 2002.
4. Maddox S.J.: Fatigue Strength of Welded Structures. Second edition. Albington Publishing,
Albington, 1991.
The testing work has been done and will be continued under the Research Project No. 5 TO7B 031
25, with financial support of the Committee for Scientific Research (KBN) in the years 2003-2006.
 2T5. Fatigue and fracture 175

DIFFERENT ANALYTICAL PRESENTATIONS OF SHORT CRACK GROWTH

UNDER ROTATION-BENDING FATIGUE

D. Angelova and A. Davidkov

Donka Angelova – Professor, University of Chemical Technology and Metallurgy-Sofia (UCTM),
8 Kl. Ohridsky Blvd., 1756 Sofia, Bulgaria, e-mail: dona_ang@techno-link.com
Aleksander Davidkov – Research Fellow, Institute of Metal Science, Bulgarian Academy of
Sciences, 67, Shipchensky Prohod Blvd.,1574 Sofia, Bulgaria, e-mail: lda14042001@yahoo.com

Although the enormous progress in fatigue investigations and the understanding achieved during
the last years, fatigue phenomenon stays as an important problem concerning strength of metals,
their life and structural integrity of engineering constructions. One method which is very useful,
informative and easy for application is that of investigation of short fatigue-crack growth by
replica monitoring of surface crack propagation from the initiation to failure. In our case, this
method includes short-crack rotation-bending (R-B) experiments and length measuring of
propagating crack a at some cycles N on smooth hour-glass specimens subjected to different
frequency and symmetric cycling loading, Table 1. Material under investigation is a rolled low-
carbon, low alloyed steel (RLCLAS), mostly used for off-shore applications and in shipbuilding,
marked as 092 according to the Bulgarian Construction Steel Standard. Experimental conditions
under in-air R-B are the following: specimen 1 (Stress range , MPa – 620, Frequency f, Hz – 11),
and respectively: specimen 2 (620, 11), specimen 3 (580, 11), specimen 4 (620, 6.6).
The experiments are carried out on a table-model machine for R-B fatigue, FATROBEM–
2004, newly designed, constructed and assembled in the Laboratory of Plastic Deformation of
UCTM. The machine is shown on Fig. 1 and its work described in Davidkov [1].

FIGURE 1. Test apparatus scheme: electric engine 1, driving belt 2, ball-bearing unit 3, leading
shaft 4, corrosion box 5, specimen 6, leaded shaft 7, device for circulation and aeration of
corrosion agent 8, working box 9, device for loading and load changing 10, counter 11.

A model “short crack growth-rate, da/dN – crack length, a” under cyclic R-B fatigue – Eq. (1)
– comprises parabolic-linear presentation of the three regimes of crack propagation (I SFC – short
fatigue crack, II PhSFC – physically small and III LFC – long crack), and analytical determination
of microstructural barriers d1 and d2, at which cracks are slowing down and stop [1]; a0 and af –
the initial and final crack size, A, B, C – coefficients depending on metal nature

da
I SFC {a(a0 , d1]} and II PhSFC {a[d1 , d2]}: Aia 2  Bia  Ci , i=1, 2;
dN
176 D. Angelova and A. Davidkov

da
III LFC { a[d2 , af )}: A3 a B3 (1)
dN

The basic model functions from (1) are corrected for each regime, considering the highest crack
growth rates, and presented in Fig. 2 as a parabolic-linear family 1.

FIGURE. 2. Different presentation of fatigue data: a. Corrected dependences “Crack growth rate,
da/dN – crack length, a” o Family 1; and b. “Crack growth rate, da/dN – surface energy,
'W ” o Line 2

An alternative approach to already described, classical way of treating fatigue comprises a new
function W, and a new presentation of experimental data in parameters dN / da , 'W . The function
'W discussed for the first time in Angelova [2, 3] has a dimension of surface energy per second
and unity of crack size, and leads to a linear presentation of fatigue data, da/dN– 'W o Line 2
shown as thick line in Fig. 3. The scatter band is indicated by fore thin lines at a condition of
excluding two distant points from Family 1; the inner thin lines correspond to ½ and 2 folds of da/
dN, and the outside ones - to 0.4 and 2.5. Such a precision makes it possible to use the presentation
da/dN– 'W at a lesser number of the measurements (recordings) made under every fatigue test,
especially at high correlation coefficient which in our case is fcor>0.8. Afterwards a transition to
the presentation “da/dN – a” showing much larger scatter would not be difficult.

References

1. Davidkov A., Angelova D., In Proceedings of the 2-nd International Conference

“Deformation, Processing and Structure of Materials”, Belgrade, May 2005, 173-178.
2. Angelova, D., In CD Proceedings of ECF13, Abstracts, San Sebastian, September 2000, 128.
3. Angelova, D., In Proceedings of the ECF14, Vol. 1, Cracow, September 2002, 89-97.
 2T5. Fatigue and fracture 177

VARIABLE AMPLITUDE LOAD INTERACTION IN FATIGUE CRACK

GROWTH FOR 2024-T3 ALUMINIUM ALLOY

D. Kocanda, S. Kocanda and J. Torzewski

Military University of Technology
Kaliskiego 2 Str., 00-908 Warsaw, Poland
d.kocanda@wme.wat.edu.pl

Special needs that concern fatigue damage of aircraft components require undertaking the
investigations of fatigue crack growth in aerospace alloys and finding the correlation between the
images of fracture surface and the accumulated damage due to applied load. Importance of
fractography supporting the analysis of failure cases of real structures was pointed out among
others by J. Schijve in his recently published book [1]. In order to check the capability of the
reconstruction of load-time history on the basis of fractographic analysis extensive research
program was developed for 2024-T3 Aclad aluminium alloy sheet subjected to different variable
amplitude loads [2]. The paper presents the results both of fatigue crack growth response and the
microfracture analysis for the CCT specimens (400x100x3 mm) under two similar loads LHL
(low-high-low) and FAF (flight-after-flight) (Fig. 1a and b). These loads are employed when
simulating the flight loads of the lower skin aircraft wing structure.

FIGURE 1. Loading of type LHL (a) and type FAF (b)

Load sequence LHL contains 2400 cycles whereas FAF sequence counts 240 cycles of the
stress levels just the same the LHL load. However, these loads differ either from the number of
cycles in the particular load blocks, thereby the number of overloads, or the order of the load
blocks in the program. Employing these two load programs we can learn the influence of the load
shape on the crack growth rate and the lifetime of a component as well. Crack growth rate
behaviour in 2024-T3 alloy was analyzed either on the surface or in the depth of the samples and
performed on the diagrams. The extension of crack depth was estimated on the basis of
fractographic analysis with the help of SEM or TEM microscopes. In many cases the effect of
multiple overloads that intersperse the baseline cycles is not clearly visible on the typical
experimental plot da/dN as function of 'Keff. More details of crack retardation and acceleration
associated with the overloads provide the microfracture analysis done with the help of TEM
microscope. Figure 2 performs the micrographs which illustrate the systems of fatigue striations on
the fracture surface under LHL and FAF load sequences. Microfracture analysis revealed a big
variation of crack growth rate within the particular load sequences. The courses of local crack
growth rate, affected by one load program of LHL type (Fig. 3a) and FAF type (Fig. 3b) against
crack length are shown in Figure 3. Crack growth rates were estimated on the basis of fatigue
striations spacing referred to the particular blocks of loading. Further analysis proved that under
LHL program the period of crack propagation covers less than 10% of load program duration
178 D. Kocanda et al.

whereas in the case of FAF program 26% of total time is devoted to crack propagation. Well, most
time of the load program duration is associated with the crack arrest or crack growth at very low
rate. This crack behaviour results from the plastically induced crack closure effect as well as the
crack penetration through the plastic zones associated with the overloads.

FIGURE 2. TEM (a) and SEM (b) micrographs illustrate the fatigue striation systems on the
fracture surface that are associated with LHL program (a) and FAF program (b).

FIGURE 3. Variation of crack growth rates in microscopic scale for 2024-T3 alloy within one load
sequence of LHL (a) and FAF (b) loadings against crack length, respectively.

For prediction the crack growth under VA loading and estimation the fatigue life of a
component was elaborated a deterministic approach. The crack retardation model is based on the
Wheeler model.

References
1. Schijve , J., Fatigue of structures and materials, Kluwer Academic Publishers, 2001.
2. Kocanda D., Kocanda S. and Torzewski J., Archive Mech. Engineering, vol. LI, No 3, 361-
376, 2004.
 2T5. Fatigue and fracture 179

AN INVESTIGATION ON THE FATIGUE PERFORMANCE OF HYDRAULIC

GATE WHEELS

D. Polyzois and A. N. Lashari

Department of Civil Engineering, University of Manitoba
15-Gillson St., Winnipeg, MB, R3T 5V6, Canada
polyzoi@cc.umanitoba.ca

Manitoba Hydro currently owns fourteen hydro power generating stations with a total capacity of
over 7500 MW. Both emergency intake gates and spillway gates are used in each. These are fixed-
wheel gates with wheels mounted on both sides which roll on roller paths. Environmental
corrosion along with high wheel loads cause differences in the profile of the roller path surface.
Combined with the relatively high torsional stiffness of the gate end girders, a condition of wheel
load redistribution occurs where some wheels are relieved of load while others are loaded beyond
their maximum design values. These loads can be as high as two to three times larger as the
original design loadings. Failure of one wheel could jeopardize the overall operation of the gate.
Currently, the design of gate wheels and roller paths do not consider the fatigue life of these
elements. Metal fatigue is a process resulting in failure or damage of a component subjected to
repeated loading. Although failure by progressive fracture is often associated with localized tensile
stresses, fatigue cracks can also occur on the plane of maximum shearing stresses. In the case of
wheels or wheels used in spillway or control gates, contact stresses on or somewhat beneath the
surface of the contact surface can cause failure of one or both of the bodies. In this case, since the
contact point changes as the gates open and close, the contact stresses are repeated over many
times, a situation that could eventually lead to fatigue failure. Usually, failure starts as a localized
fracture just below the surface of contact where the shearing stress is high and progresses
outwardly under the influence of the repeated wheel loads. Contact stresses can also cause pitting
at the surface of contact, as shown in Fig. 1.
An experimental investigation was carried out at the University of Manitoba in Winnipeg,
Canada, which involved the testing of three wheels and six roller path plates under cyclic loading.
The wheels were 838.2 mm in diameter while the roller path plates were 381 x 177.8 x 50.8 mm.
One of the wheels was made of cast iron while the other two were made of forged steel. The
material in two of the roller path plates was ASTM C 1040 with no heat treatment. The material in
the other plates was ASTM C 1045 with heat treatment. A unique test set up, shown in Fig. 2, was
designed and constructed for this special fatigue type of testing. The average Brinell Hardness
Number (BHN) for plates P1 and P2, which were not heat treated, was 291, whereas, for the rest of
the roller path plates, which were heat treated, the average BHN was 364. The hardness for the cast
iron wheel varied from 391 BHN at the rolling surface to 219 BHN at 38 mm away from the
rolling surface while hardness for forged steel wheel (R2) varied from 373 BHN at the rolling
surface to 326 BHN at 63 mm away from the rolling surface. The hardness for the other forged
steel wheel (R3) varied from 473 BHN at the rolling surface to 428 BHN at 38 mm away from the
rolling surface.
180 D. Polyzois and A. N. Lashari

The wheels were subjected to a radial load of, approximately, 825 kN that remained fairly
constant while the wheels were “rolled” over the roller path for up to one million cycles. The
tensile strain on the cast iron wheel ranged from 103 µ to 2057 µ while the tensile strain on the
forged steel ranged from 4 µ to 206 µ. Tensile strains were observed in almost all the strain gauges
installed on all roller path plates.
Roller path plates P1 and P2, which were not heat treated, exhibited a maximum indentation of
1.48 mm and 1.21, respectively, after one million cycles. Roller path plates P3 and P4 recorded a
very low indentation depth of 0.03 mm and 0.11 mm, respectively. The other four roller plates
were heat treated and suffered a much smaller surface indentation which ranged from 0.02 to 0.11
mm after 400,000 cycles. Clearly the test results demonstrated that cast iron wheels performed
very poorly under fatigue loading while heat treated forged steel wheels performed well.
 2T5. Fatigue and fracture 181

A MICROMECHANICAL MODEL FOR CRACK INITIATION IN HIGH CYCLE

FATIGUE OF METALLIC MATERIALS

Vincent Monchiet, Eric Charkaluk and Djimedo Kondo

Laboratoire de Mécanique de Lille, UMR CNRS 8107, Université Lille 1, France
Vincent.Monchiet@ed.univ-lille1.fr, eric.charkaluk@univ-lille1.fr, kondo@univ-lille1.fr

Most of structural components resisting to high cycle fatigue are subjected to a multiaxial state of
stress. Since fatigue cracks generally initiate and propagate in a plane of maximal shear stress
(stage I), the first approaches of Crossland and Sines considered the octahedral plane and their
criteria are based on the amplitude of the second invariant of the deviatoric stress tensor. In order
to take into account the mean stress, these authors postulated also a linear combination of J 2 and
the hydrostatic part of the stress tensor. It is now widely recognized that under high cycle loadings
conditions, metals fracture is the result of the strain localization in some unfavorably oriented
grains undergoing plasticity. Dang Van, followed by Papadopoulos and others, had proposed a
multiscale framework which leads to a sufficient condition for non nucleation of cracks under high
cycle loading. This condition is ensured by elastic shakedown at the grain scale (see the
representive elementary volume on Fig. 1). Nevertheless, microplastic activity alone cannot
explain the role of pressure on the fatigue limit and a generalized multiaxial fatigue limit
depending on the hydrostatic pressure is due. So far, this is the proposal already suggested by
Papadopoulos in his generalized multiaxial fatigue limit.

FIGURE 1 : Representative Elementary Volume (RVE)

To support this type of modelling, we propose in the present study to incorporate some
observed damage mechanisms (reported in literature for materials involving faced-centred-cubic
structures) in the multiscale approach of Dang Van. The characteristics of the study lies in the
consideration of the scale of Persistent Slips Bands (PSB) at which appear the damage
micromechanisms. Actually, the PSB constitute preferential sites for fatigue cracks nucleation.
The outline of the study is as following :
1 we first propose a simple model for the grain behavior under high cycle loading :
microplasticity and microdamage mechanisms (induced by micro-cavities growth) are
incoprorated in this grain level model. The monocrystal plastic behavior is described by
the a Schmid's law with a linear isotropic hardening and non linear kinematics hardening
law, the later being based on the effective plastic slip. On the basis of experiments by
Antonopoulos et al. (1976) and Esseman et al. (1982), damage along PSB is associated to
the production of vacancies (which is responsible of irreversible volume change) and non
spherical microvoids nucleation and growth into these localized bands. The model is
completed by adopting the crack nucleation condition proposed by Antonopoulos et
182 V. Monchiet et al.

al.(1976) according to which a crack nucleates at the PSB-matrix interface when the total
strain along the PSB reaches a critical value. At the grain level, a crack nucleation criterion
is then derived and allows to determine the local condition for fatigue limit.
2 a second part of the study deals with non linear homogenization techniques allowing to
translate the local crack nucleation criterion into a macroscopic fatigue criterion. Different
non linear homogenization schemes are considered : Lin-Taylor model, Sachs scheme and
Kroner's accomodation law. The consideration of such simple schemes is justified by the
fact that in high cycle fatigue context, plasticity is localized in one or few grains.
An illustration of the micro-macro approach in the case of macroscopic affine loadings is
presented on Fig. 2 ; this clearly shows the capability of the model to take into account the effect
of the pressure. Besides, the fatigue limit appears to be independent of the mean shear stress.

Ta
FIGURE 2 : Affine torsion-tension loadings. Amplitude of the normalized shear stress as
W0
function of the normalized mean pressure. Comparison of the predictions of two homogenization
schemes (Sachs, Eshelby-Kroner).

References
1. J. G. Antonopoulos, L. M. Brown, A. T. Winter. Vacancy dipoles in fatigued copper.
Philosophical Magazine, Vol. 34, No. 4, pp. 549-563, 1976.
2. M. Berveiller, A. Zaoui. An extension of the self-consistent scheme to plastically flowing
polycrystals. J. Mech. Phys. Solids, Vol. 26, pp. 325-344, 1979.
3. U. Essmann, U. Gosele, H. Mugrhabi. A model of extrusions and intrusions in fatigued
metals. I. Point-defects production and the growth of extrusions. Philosophical Magazine A,
Vol.44, No. 2, pp. 405-426, 1991.
 2T5. Fatigue and fracture 183

COMPARATIVE ANALYSIS OF TWO MODELS FOR EVALUATING FATIGUE

DATA

Enrique Castillo, Antonio Ramos2, Manuel Lopez-Aenlle2, Alfonso Fernandez-Canteli2 and

Roland Koller3
ETSICCP, University Cantabria, Avda. de los Castros s/n, 39005 Santander, Spain.
2EPSIG, University Oviedo, Campus de Viesques, 33203 Gijón, Spain.
3EMPA, Überlandstr. 129, 8600 Dübendorf, Switzerland.

castie@unican.es, afc@uniovi.es, Roland.Koller@empa.ch

The optimization of fatigue programs comprising planning, testing and results evaluation is an
issue of paramount importance for material and testing laboratories and a recursive subject in the
literature which research groups dealing with fatigue design are long concerned with. There is a
general belief that no specific model fulfils in a satisfactory way all these functions, i.e. a model to
be considered acceptable by the community. Nevertheless, the up-and-down method, despite its
limitations, finds a good recognition by both, research groups related to the academy as well as to
the testing laboratories. This can be due to the simple test strategy and evaluation technique
implied in the practical application of the up-and-down method, which is easy to understand and to
apply in practice. The limited information supplied in this methodology, consists in the
determination of the probabilistic stress range for a certain limit number of cycles, here denoted as
pseudo endurance limit, which is, generally, considered, although unjustifiably, sufficient for
practical design [1,2].
Among the inconsequences of the up-and-down method we find the assumption of a normal
(or log-normal) distribution function for the variation of the stress range for given number of
cycles, which is not suppported by statistical analysis of real data [3]. Further, data results
corresponding to the long life fatigue region, i.e., that properly involved in the up-and-down
method, is to be completed by additional results in the medium life fatigue region if the whole S-N
field is required to be considered in the current fatigue life design.
In this paper, an alternative fatigue test strategy and the subsequent parameter estimation from
data results is proposed for use. The procedure is based on a regression Weibull model comprising
the whole S-N field, as developed by Castillo et al. [4]. The model enables us to incorporate
specimens of different lengths into the analysis [5]. Firstly, the two models considered here, that
based on the up-and-down methodology and the regression model proposed by the authors, are
introduced and discussed. The possibility of considering a Weibull distribution for the stress range
in the up-and-down methodology, leading to the so-called Weibull-up-and-down model, is
justified and proposed. The experimental program carried out at the EMPA (Swiss Federal
Laboratories for Materials Testing and Research) is also presented (see Fig. 1). Thereafter, a
comparative analysis of the fatigue results obtained in the frame of this program is performed using
both methodologies.
The applicability and reliability resulting from both procedures is then analyzed and the
advantages and shortcomings of both methods are discussed. Finally, the conclusions of this
investigation are presented.
184 E. Castillo et al.

FIGURE 1. Adjusting fatigue data of 42CrMo4 steel using the proposed regression model .

References
1. Deubelbeiss E., Materialprüfung, Vol. 16, 240-244, 1974
2. Hück, M., Schütz, W., Zenner, H., Industrieanlagen Betriebsgesellschaft mbH, ReportB-TF-
742B, 1974.
3. Castillo, E., Fernández Canteli, A., Esslinger, V., Thürlimann, B.,IABSE Periodica 1/1985,
IABSE Proceedings 82/85, 1-40, 1985.
4. Castillo, E., Fernández Canteli, A., Int. Journal of Fracture Nr. 107, 117-137, 2001.
5. Castillo, E., López-Aenlle, M., Ramos, A., Fernández-Canteli, A., Esslinger ,V., Kieselbach,
R., Int. J. of Fatigue. Submitted to evaluation.
 2T5. Fatigue and fracture 185

ASSESSMENT OF DAMAGE AT NOTCH ROOT OF THICK PLATES

E. C. G. Menin and J. L. de A. Ferreira

University of Brasília - UnB, Mechanical Engineering Department, GAMMA Research Group
Campus Univ. Darcy Ribeiro, Asa Norte, Brasília, Brazil,
70940-910; 55-61-3073706 R. 213
edumenin@unb.br, jorge@unb.br

Heavily loaded structural components may present local yield at stress concentrators, such as,
notches and geometrical discontinuities [1]. In cases involving cyclic loads, the presence of local
plasticity could lead to the nucleation and propagation of fatigue cracks and complete fracture of
the components [2, 3]. In such cases, the strain life approach should be used to evaluate fatigue
damage, making it necessary to perform an elastoplastic analysis of stresses and strains at the
proximities of these stress concentrators. Due to reduced computational effort requested,
approximated models are widely used with that purpose. The most used are the ones proposed by
Neuber [4], Seeger et al. [5], Glinka [6] and Ye et al. [7].
One factor that complicates the use of these approaches consists in the evaluation of stress
concentration problems, by most part of the specialized literature, as plane approximation
problems that do not take into account the variations in the stress-strain field configuration along
the thickness of the components, as these become larger in magnitude. To corroborate the
importance of such 3D analysis, several numerical and experimental studies have been published
indicating that the stress field near the geometrical discontinuities varies along the notch root, as
components become thicker [8]. In such cases, 2D values of Kt and stress/strain plane
approximations should be carefully evaluated when associated to local strain models in order to
correctly estimate local levels of solicitations at the notch root of thick components.
In that sense, this study presents examples of the difficulties associated to the use of local strain
approximated models when evaluating cases involving notched components presenting expressive
values for the ratio between thickness and notch root radius, or dimensionless thickness, t/r. A
preliminary linear elastic behavior analysis for the variation of Kt and the displacement constraint
level of points along the notch root was performed using 3D FEM. After that, elastoplastic
analyses were performed by means of the most used local strain models, as well as, finite element
3D models and 2D approximation. These simulated the local behavior at the root of different
geometries and types of steel associated to two cases of study, exemplified on Tables 1 and 2.

TABLE 1. Notch geometries and materials.

TABLE 2. Mechanical and fatigue properties

186 E. C. G. Menin and J. L. de A. Ferreira

The results obtained for the linear elastic domain analysis of the cases of study made it possible
to evaluate the stress constraint level of points at the notch root of the stress concentrators, as well
as, the behaviour of Kt for components with different values of t/r tending to zero and to determine
the reliability and coherence of the use of 2D values of this factor to describe the stress
concentration phenomena.
The results presented for the elastoplastic analysis of the two cases of study made evident the
difficulties associated to the use of local strain approach in cases involving components with
expressive values of the dimensionless thickness. Among these are the ones associated to the
determination of correct values of Kt and the constitutive curve constants to be used along with the
local strain models, the determination of the points subjected to the most severe conditions, as well
as, the type of bi-dimensional approximation to be used with FEM.
The analysis results suggested the necessity of 3D FEM evaluation to determine in a precise
way the local strain and fatigue life of components with expressive values of t/r. Moreover, the
results made evident that, in the impossibility to use such practice, one should be careful when
using plane stress and strain approximations and 2D values of Kt because these could present
divergences with respect to the real conditions evaluated for the models.

References
1. Peterson, R.E. (1997), Stress Concentration Factors, Ed. John Wiley & Sons, EUA.
2. Filippini, M. (2000), “Stress gradient calculations at notches”, Int. Journal of Fatigue, 22,
397-409.
3. Visvanatha, S.K., Straznicky, R.L. e Hewitt, R.L. (2000), “Influence of strain estimation
methods on life prediction using the local strain approach”, Int. Journal of Fatigue 22, 675-
81.
4. Neuber, H. (1961), “Theory of stress concentration for shear-strained prismatical bodies with
arbitrary nonlinear stress-strain law”, Journal of Applied Mec., 28.
5. Seeger, T.H. e Heuler, P. (1980), “Generalised application of Neuber’s rule”, Journal of
Testing and Evaluation, 8, 199-204.
6. Glinka, G. (1985), “Energy density approach to calculation of inelastic strain-stress near
notches and cracks”, Eng. Fract. Mechanics, 22(3), 485-508.
7. Ye, D., Matsuoka, S., Susuki, N. e Maeda, Y. (2003), “Further investigation of Neuber’s rule
and the equivalent strain energy density (ESED) method”, Int. J. of Fatigue, 22, 675-681.
8. Zhenhuan L., Wanlin, G. e Zhenbang, K. (2000), “Three-Dimensional elastic stress fields
near notches in finite thickness plates”, Int. Journal of Solids and Structures, 37, 7617-7631.
 2T5. Fatigue and fracture 187

FATIGUE STRENGTH PREDICTION OF SPOT-WELDED JOINTS USING

SMALL SPECIMEN TESTING

Eisuke Nakayama, Manabu Fukumoto, Mitsuo Miyahara, Kazuo Okamura, Hiroki Fujimoto and
Kiyoyuki Fukui1
Corporate Research & Development Laboratories, Sumitomo Metal Industries, Ltd.,
1-8, Fuso-Cho, Amagasaki, Hyogo 660-0891, Japan
1Steel Sheet, Plate, Titanium & Structural Steel Company, Sumitomo Metal Industries, Ltd.,

1-8-11, Harumi, Chuo-ku, Tokyo 104-6111, Japan

nakayama-eis@sumitomometals.co.jp

It is well known that fatigue strength of spot weld of high strength steel sheet is not improved,
compared with that of mild steel sheet. In this study, the governing factors and the effects of steel
grade on fatigue strength of spot weld is investigated. Firstly, small specimens with total length of
less than 3mm are taken from the spot weld of mild steel sheet (270MPa-grade) and high strength
steel sheet (590MPa-grade). And then, tensile and high cycle fatigue properties are individually
evaluated by newly-developed testing technique. Secondly, finite element analyses of tensile-shear
specimen of spot-welded joints under cyclic loading are carried out and fatigue limit of the joints
are predicted, using the above-mentioned local material strength properties and considering
welding residual stresses around spot weld. Predicted results are nearly equal in both steels, which
coincides with experimental results. It is found that fatigue strength of HAZ, which is the crack
initiation site in joint, of 590MPa-grade steel is higher than that of 270MPa-grade steel. However,
residual stress in 590MPa-grade steel is also higher and this may be one of the reasons why
590MPa-grade steel exhibits little improvement in fatigue strength of the joint over 270MPa-grade
steel.

FIGURE 1. Shape and dimensions of fatigue specimen (unit:mm).

FIGURE 2. S-N diagrams of HAZ specimen.

 2T5. Fatigue and fracture 189

A THERMO-MECHANICAL MODEL FOR RANDOM BRAKING OF MACHINE

COMPONENTS

F. Loibnegger, H.P. Rossmanith1 and R. Huber

Vienna University of Technology
Institute for Testing and Research in Materials Technology, Karlsplatz 13, A-1010 Vienna
1Institute of Mechanics and Mechatronics, Wiedner Hauptstr. 8-10/325, A-1040 Vienna

friedrich.loibnegger@tuwien.ac.at, hans.peter.rossmanith@tuwien.ac.at,
richard.huber@tuwien.ac.at

This contribution presents a hybrid analytical-numerical treatment of the thermo-mechanical

model for the analysis of the braking process of rotating machine parts such as wheels, etc. The
machine parts have the shape of a thick disk. These disks are repeatedly frictionally disk-braked at
random sequences of time instants. As the loading is rather severe, the stresses during braking and
subsequent cooling reach the plastic limits in compression and tension, respectively. Hence, the
loading may cause conditions similar to those encountered in shake-down phenomena.
The thick disk is assumed to contain initial compressive and tensile stresses as a result of
inappropriate heat treatment after forging and due to press-fitting the disk on the axis. In addition,
the disk is allowed to develop and extend radial surface cracks within the annular brake regions.
The shapes of these surface cracks initially are semi-elliptical, but vary largely, depending on the
loading sequences and the existing residual stress fields after cyclic straining [1].
Under certain loading conditions and random sequences of brake phenomena, the
circumferentially distributed and radial oriented cracks may exhibit accelerated crack growth in
axial direction and in a few cases cracks may even emerge on free rim surfaces.
A thermo-elasto-plastic fracture mechanical analysis was performed, based on a finite element
analysis. The crack was locally modelled as a semi-circular surface crack which was allowed to
propagate at variable velocities along the surface and into the material. Linear elastic fatigue type
crack advance was assumed during the cooling phase after braking.
It was detected, that the distribution of residual stresses had a decisive influence on the stability
behaviour of fatigue cracks. It could be shown that the extension of cracks on the surface along the
circumferential direction was controlled by the residual stresses, ultimately leading to crack arrest.
The large tensile circumferential stresses within the brake region are balanced by the residual
compressive stresses in the rim and hub regions.
Results indicate that the actual load is depending on the randomness and characteristics of the
braking events. The speed of directional crack advance also depends on the style and the loading
sequence.
Safe life – time diagrams could be constructed for the use of the rotating machine parts [2].
These safety diagrams clearly demonstrate the effect of the initial crack size on the arrest
behaviour of the observed crack and the remaining life time of the structure.
This paper discusses in detail the various steps of the very complex analysis [3] and at each
stage indicates alternative routes and avenues for improvement of the analysis while still keeping
the amount of numerical calculations at bay. A practical example will be demonstrated.
190 F. Loibnegger et al.

References
1. Rolfe S.T. & Barsoum, J.M., Fracture and Fatigue Control in Structures - Applications of
Fracture Mechanics, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1977.
2. Anderson T.L., Fracture Mechanics - Fundamentals and Applications. CRC Press, Boca
Raton, USA, 1991.
3. Gdoutos, E.E., Rodopoulos, C.A. & J.R. Yates : Problems of Fracture Mechanics and Fatigue
- A Solution Guide, Kluwer Academic Press Publishers, 2003.
 2T5. Fatigue and fracture 191

LIFETIME CALCULATION OF RAILWAY WHEEL STEELS

BASED ON PHYSICAL DATA

F. Walther and D. Eifler

Institute of Materials Science and Engineering
University of Kaiserslautern, P.O. Box 3049, D-67653 Kaiserslautern, Germany
walther@mv.uni-kl.de, eifler@mv.uni-kl.de

Higher speeds in passenger traffic and higher axle loads in freight traffic cause a substantial
increase of the mechanical load of railway wheels. To develop and verify fatigue criteria and
lifetime calculation methods, a reliable fatigue data base of highly loaded wheels steels, in
particular under near-service loading conditions, is required.
Fatigue tests with variable maximum stresses were performed at ambient temperature on
servohydraulic testing systems with specimens of the wheel steels R7 (SAE 1055) and B6 (SAE
1070) according to UIC standards. The specimens were taken from different depth positions of the
rim of original railway wheels. Due to the industrial heat treatment and the size of the components
microstructural gradients are unavoidable [1, 2]. In addition to the plastic strain amplitude Ha,p
from mechanical stress-strain-hysteresis measurements, changes of the specimen temperature 'T
and the electrical resistance 'R during cyclic loading were considered. The integral temperature
and resistance data can be used to evaluate continuously the fatigue behaviour under variable
amplitude loading. Besides the specimen geometry and temperature, the resistance depends on the
specific resistance of the material and changes in a characteristic manner as a function of the defect
density in the material. To describe proceeding fatigue damage, the resistance can be measured as
a reference value at the beginning and during a fatigue test as well as in load-free states [1, 2].

FIGURE 1. Variable amplitude loading with constant amplitude (measuring) sequences.

The basic idea of the new testing procedure developed at the Institute of Materials Science and
Engineering is to combine near-service load spectra with periodically inserted measuring
sequences with a constant amplitude below the endurance limit. With this method (Fig. 1) reliable
information about proceeding fatigue damage under near-service loading can be obtained on the
basis of Ha,p, 'T and 'R. In representative cyclic ‘deformation’ curves the mean values of these
physical quantities are plotted as a function of the number of cycles N*. N* was calculated by
multiplication of the frequency and the time of experiment. The random signal was generated by a
combination of a White Noise and a Gauss distribution. After normalising the random signal to
values between -1 and 1 it was multiplied with the maximum stress.
Cyclic softening and hardening processes were observed for random loading experiments with
maximum stresses of 400 MPa d Vmax d 800 MPa (Fig. 2), comparable to single-step tests. An
192 F. Walther and D. Eifler

increasing maximum stress leads to higher measured values in an unique manner and consequently
shorter lifetimes. The physical quantities Ha,p, 'T and 'R are directly correlated with
microstructural details. The test with Vmax = 400 MPa was stopped without specimen failure.

FIGURE 2. Representative cyclic Ha,p-, 'T- and 'R-curves for random loading, R7.

Furthermore, the new testing procedure allows to establish lifetime calculation methods for
near-service loading. To derive a physically-based lifetime calculation (PHYBAL), 'T and 'R
were used equivalently to Ha,p in Morrow and Manson/Coffin curves. Using the mathematical
descriptions of these curves, S,N curves according to the Basquin equation can be calculated. The
lifetimes calculated on the basis of the plastic strain amplitude, the temperature and the resistance
agree very well with the experimentally determined lifetimes [3].

References
1. Walther, F., Eifler, D., Mater. Sci. Eng. A 387-389, 481-485, 2004.
2. Walther, F., Eifler, D., In Proceedings of the 11th International Conference on Fracture,
edited by A. Carpinteri, Turin, No. 3944, 2005, 1-6.
3. Starke, P., Walther, F., Eifler, D., PHYBAL - A new method for lifetime prediction based on
strain, temperature and electrical measurements, Int. J. Fatigue, 2005, in press.
 2T5. Fatigue and fracture 193

FATIGUE CRACK PROPAGATION OF SUPER-DUPLEX STAINLESS STEEL

AT DIFFERENT TEMPERATURES

G. Chai and S. Johansson

Sandvik Materials Technology, R&D Centre
811 81 Sandviken, Sweden
Linköping University, Department of Mechanical Engineering
SE 581 83 Linköping, Sweden
guocai.chai@sandvik.com, stejo@ikp.liu.se

The fatigue crack propagation behaviour in an austenite-ferrite two phase super-duplex stainless
steel (SDSS) (UNS S32750) has been investigated at temperatures from –50ºC to 150qC. Two
material conditions (as delivered condition-AD and aged condition (475 ºC/4h)) and two stress
ratio values (R=0,1 and 0,5) were used. The fatigue crack propagation rate, threshold value,
closure and fracture behaviour have been studied.
The fatigue crack propagation of this alloy at –50ºC shows a tendency of lower crack
propagation rates at small stress intensity factor range, but higher rates at higher stress intensity
factor range, and a higher crack propagation threshold value comparing with those at RT. The
crack propagation rates at 150ºC are similar to those at RT, and the crack propagation threshold
value is also higher than that at RT, but smaller than that at –50ºC. The ageing of the alloy or an
increase in R value increases crack propagation rate, but decreases the threshold value (Figure 1).

FIGURE 1. Influence of temperature, material and stressconditions on crack propagation

behaviour of a DSS material

This alloy shows rather high fatigue crack propagation threshold values 'Kth (Table 1). It was
found that it is mainly due to the fact that this two phase material has large crack closure,
especially at low and high temperatures (Table 1). The following two phenomena observed may
give some explanations. One is type of “crack-bridges” occurred in the softer phase (Figure 2a).
Compressive residual stresses in this phase were measured after the fatigue testing. This may
indicate that fatigue crack closure can be induced by residual stresses. The other is type of
“fracture mismatch” observed near the threshold regime. The fracture changes its pattern or
directions when a crack propagates from one phase to another (Figure 2b). This may increase the
surface roughness induced closure.
194 G. Chai and S. Johansson

TABLE 1. Influence of temperature on the closure and effective threshold values.

*: from da/dN measurements; **: from closure measurements. '.cl: closure value; 'Keff th :
effective threshold value.

FIGURE 2. (a). Residual stress induced closure; (b). Fracture mismatch from austenite phase Jto
ferrite phase D

The higher effective threshold value 'Keff th at 150qC, which can not be explained by the
classic theories, was further investigated. It may be attributed to dynamic strain ageing.
A transition from cleavage fracture to facet fracture was observed near the threshold values in
the as delivered material at -50qC and in the aged material at RT. The possible mechanisms were
discussed.
Key words: Super duplex stainless steels, Fatigue crack propagation behaviour, Fatigue crack
closure, Temperature, Fracture.
 2T5. Fatigue and fracture 195

TRANSITIONS OF FATIGUE CRACK INITIATION FROM SURFACE,

SUBSURFACE TO SNDFCO

G. Chai
Sandvik Materials Technology, R&D Centre,
811 81 Sandviken, Sweden
guocai.chai@sandvik.com

As known, fatigue crack initiation mainly starts at surface defects at high stress or strain
amplitudes, but shifts to pre-existing subsurface defects such as inclusions or pores at low stress
amplitudes or in the very high cycle fatigue regime [1]. Recently, it was reported that fatigue crack
initiation could occur in non-defect existing areas and form a subsurface non-defect fatigue crack
origin (SNDFCO) [2]. Figure 1 shows a summary of these three phenomena.

FIGURE 1. Three types of fatigue crack initiation; (a). From a surface defect; (b). From a
subsurface inclusion; (c). From subsurface non-defect fatigue crack origin (SNDFCO).

This paper gives a discussion on the transition mechanisms of fatigue crack initiation from
surface defects, subsurface defects to subsurface non-defect fatigue crack origins in some two
phase or multiphase steels. It was found that surface crack initiation mainly is caused by extrusion
and intrusion of slip bands at the specimen surface, where the applied stresses are higher than the
elastic limit of the material (Figure 2a). The occurrence of subsurface crack initiation is caused by
localised cyclic plastic deformation or dislocation slipping processes around the subsurface defects
due to stress concentrations (Figure 2b). The formation of SNDFCO in some two phase alloys is
also a cyclic plastic deformation process. Although the stresses applied are lower than the elastic
limit of the material, but they can be higher than the elastic limit of the softer phase. This leads to
the occurrence of cyclic plastic deformation in this softer phase, which causes the damage and then
the formation of micro or short cracks in the softer phase (Figure 2c) or at the grain boundaries or
corners due to the stress concentration by dislocation pileups.
196 G. Chai

FIGURE 2. Transition mechanisms of the fatigue crack initiation from the surface, subsurface to
subsurface non-defect fatigue crack origin; (a). Extrusion and intrusion of slip bands at the
specimen surface; (b). Slip band formed near the subsurface defect; (c). Short cracks formed in the
softer phase.

The formation of SNDFCO is a crack propagation process, which is controlled by

microstructure mechanics. The crack propagation transition from stage I to stage II leads to the
formation of SNDFCO. The size and morphology of the SNDFCO strongly depend on the applied
stresses. The size of the SNDFCO increases and their fracture morphology changes from more
ductile to facet when the applied stresses decrease.
Key words: Fatigue crack initiation, Subsurface non-defect fatigue crack origin (SNDFCO),
Fracture, Two phase alloys, Dislocation slip band.

References
1. Murakami, Y., Metal Fatigue: Effect of small defects and non-metallic inclusion, Elsevier
Science Ltd, 2002.
2. Chai, G., In Proceedings of the Third International Conference on Very High Cycle fatigue,
edited by T. Sakai and Y Ochi, Kusatsu, 2004, 24-31 and 374-381.
 2T5. Fatigue and fracture 197

SURFACE FATIGUE OF GEAR TEETH FLANKS

G. Fajdiga, M. Sraml and J. Flasker

University of Maribor, Faculty of Mechanical Engineering
Smetanova ulica 17, SI-2000 Maribor, Slovenia
gorazd.fajdiga@uni-mb.si

Mechanical behaviour of various machine elements, such as gears, brakes, clutches, rolling
bearings, wheels, rails, and screw and riveted joints, are influenced by interaction between contact
elements and surfaces. Surfaces in rolling and/or sliding contact are exposed to material contact
fatigue. Contact fatigue can be defined as a kind of damage caused by changes in the material
microstructure which results in crack initiation followed by crack propagation, under the influence
of time-dependent rolling and/or sliding contact loads. The contact fatigue process can in general
be divided into two main parts:
• initiation of micro-cracks due to local accumulation of dislocations,high stressesat local
points, plastic deformation around inhomogeneous inclusions or other imperfections in or
under the contact surface;
• crack propagation, which causes permanent damage to a mechanical element.
In this paper the pitting phenomenon of gears is addressed and the developed numerical model
is used for determination of pitting resistance, i.e. the service life of gear teeth flanks. The
initiation of fatigue cracks represents one of the most important stages in the pitting process. The
position and mode of fatigue crack initiation depends on the microstructure of the material, the
type of the applied stress and micro- and macro-geometry of the specimen, Cheng [1]. The crack
initiation periods can be very different and cracks can be initiated either on or under the surface,
depending on a different combination of rolling and sliding contact conditions, Kaneta [2].
In general, gear teeth pitting may be initiated on sub-surface or surface. Sub-surface pitting
initiation can be observed in high quality gears made of alloy steel, with smooth contact surfaces
and good lubrication, where the large shearing sub-surface stresses due to contact loading initiate
substantial dislocation motion, which governs the crack initiation process, Glodež [3]. Surface
pitting initiation is common in industrial gears, which have rougher surfaces and are made of
ordinary construction steels. The surface pitting is strongly influenced by surface roughness and
other surface defects, like machining marks, large notches, inclusions, etc. The surface cracks may
also appear as a consequence of thermal treatment of the material due to residual stresses.
This paper considers only the second mechanism of gear pitting, i.e. surface pitting. The
process of surface pitting can be visualized as the formations of small surface initial cracks grow
under repeated contact loading. Eventually, the crack becomes large enough for unstable growth to
occur, which causes the material surface layer to break away. The resulting void is a surface pit
(Fig. 1).
The number of stress cycles N required for pitting of a gear teeth flank to occur can be
determined from the number of stress cycles Ni required for the appearance of the initial crack in
the material and the number of stress cycles Np required for a crack to propagate from the initial to
the critical crack length, when the final failure can be expected to occur:

N Ni  N p
(1)
198 G. Fajdiga et al.

FIGURE 1. Typical surface pits on gear tooth flank.

This paper describes a computational model for contact fatigue crack initiation and crack
propagation in the contact area of gear teeth flanks. The purpose of the present study is to present,
firstly, a model for prediction of contact fatigue initiation, which is based on continuum
mechanics, real cyclic contact loading and specific material fatigue parameters. The material
model is assumed as homogeneous, without imperfections such as inclusions, asperities,
roughness, residual stresses, etc., as often occur in mechanical elements. A moving contact load is
often used for simulation of the cyclic loading in fatigue crack initiation and propagation analyses
on mechanical elements (simulation of meshing of gears) Fajdiga [4]. The second part of the
computational model is a crack propagation model based on appropriate short fatigue crack growth
theories Navarro [5], Sun [6]. The model attempts to account for the different parameters
influencing the crack propagation process (Hertzian contact pressure, friction between contacting
surfaces, fluid trapped in the crack, meshing of gears, etc,) leading to pitting, starting from the
initial surface fatigue crack to the critical crack length, when the occurrence of a surface pit is
expected. The results of the computations can be used to predict the time required for the
development of pits, i.e. the service-life of gears with regard to pitting [4].

References
1. Cheng W., Cheng HS., Mura T., Keer LM. Micromechanics modelling of crack initiation
under contact fatigue. ASME Journal of Tribology; 116: 2-8, 1994.
2. Kaneta M., Yatsuzuka H., Murakami Y. Mechanism of crack growth in lubricated rolling/
sliding contact. ASLE Transactions; 28: 407-414, 1985.
3. Glodež S., Ren Z., Flašker J. Simulation of surface pitting due to contact loading.
International Journal for Numerical Methods in Engineering; 43: 33-50, 1998.
4. Fajdiga G., Flašker J., Glodež S. and Ren Z. Numerical simulation of the surface fatigue crack
growth on gear teeth flanks, Journal of Mech. Engineering 46(6), 359-369, 2000.
5. Navarro A. and Rios E.R. Short and long fatigue crack growth- a unified model,
Philosophical Magazine, 57, 15-36, 1988.
6. Sun Z., Rios E.R. and Miller K.J. Modelling small fatigue cracks interacting with grain
boundaries, Fatigue Fract. Engng Meter, 14, 277-291, 1991.
 2T5. Fatigue and fracture 199

FATIGUE AND FRACTURE PROCESSES IN HIGH PERFORMANCE PM

TOOL STEELS

G. Jesner, S. Marsoner1, I. Schemmel2 and R. Pippan3

Materials Center Leoben Forschung GmbH
Franz-Josef-Strasse 13, 8700 Leoben, Austria
gerhard.jesner@mcl.at
1Materials Center Leoben Forschung GmbH

Franz-Josef-Strasse 13, 8700 Leoben, Austria

marsoner@mcl.at
2Boehler Edelstahl Ges.m.b.H

Mariazeller Strasse 25, 8605 Kapfenberg, Austria

ingrid.schemmel@bohler-edelstahl.at
3Erich Schmid Institute for materials science of the Austrian academy of sciences

Jahnstrasse 12, 8700 Leoben, Austria

pippan@unileoben.ac.at

For cold forging applications often high performance PM – tool steels were applied. The tools are
cyclically repeated loaded to very high stresses by the forging process. The stresses are often
significantly larger than the yield stress. This will result in a local plastic deformation, crack
initiation, crack propagation and finally to the failure of the tools.
It is important to know the fatigue behaviour of these types of steels to get information about
the life time of such tools.
From microstructural point of view the investigated steels can be considered as MMC with
primary carbides with the size of a few microns and a martensitic matrix consisting of secondary
carbides. By heat treatments the strength of the matrix can be varied over wide range.
Microstructures with low and high primary carbide content and an extreme variation in
hardness – i.e. a large variation in the matrix microstructure are investigated.
The fatigue crack propagation curve and the threshold of stress intensity range fatigue crack
growth curves were determined for different stress ratios in such tool steels with different
microstructures and heat treatment. Fig. 1 shows as an example the micrographs of two selected
steels.
The different fatigue and fracture mechanism and the influence of the design of the
microstructure as well as the mechanical properties of the matrix will be discussed.

FIGURE 1. Microstructure of two selected high performance tool steels.

 2T5. Fatigue and fracture 201

NOTCH AND DEFECT SENSITIVITY OF ADI IN TORSIONAL FATIGUE

B. Atzori and G. Meneghetti

Department of Mechanical Engineering, University of Padova
Via Venezia, 1 – 35131 Padova (Italy)
giovanni.meneghetti@unipd.it

Recently Atzori et al. [1,2] analysed the sensitivity to defects and to standard notches in axial
fatigue of metallic materials commonly adopted in manufacturing engineering components and
structures, like steels and aluminium alloys. As a result of the proposed approach, the effect of any
kind of geometrical discontinuity on the fatigue limit can be treated, such as small defects, cracks,
sharp notches or crack-like notches and standard rounded notches characterised by an arbitrary
notch tip radius and notch opening angle. Two material parameters are needed, i.e. the material
fatigue limit and the threshold value of the Stress Intensity Factor.

FIGURE 1. plot of the torsional fatigue limit for components weakened by V-notches.

In the present work such an approach has been extended from axial fatigue to torsion fatigue
and can be summarised by means of the diagram presented in Fig. 1, where the fatigue limit in
terms of nominal shear stress range referred to the gross section 'Wg,th is reported as a function of
an effective dimension of the component aeff defined as:

1
§ K 3 · 1O3
aeff ¨ ¸
¨W ¸
© g ¹ (1)
being K3 the Notch-Stress Intensity Factor for mode III loading evaluated for the same
component but imposing a notch tip radius equal to zero, Wg the nominal shear stress evaluated on
the gross section and O3 is the degree of singularity of the local stress field, which depends on the
notch opening angle. Three regions can be schematically singled out divided by two length
parameters, namely a 0V t , *V
which is a material parameter, and a t , which depends also on the
V
elastic stress concentration factor Ktg for torsion. If aeff < a 0 t, then the presence of a (small) notch

does not lower the fatigue limit with respect to the material torsional fatigue limit. If a 0V t<aeff
202 B. Atzori and G. Meneghetti

*V
<a t , the fatigue limit is determined by the condition that the N-SIF equals the threshold value,
*V
which is thought of as a material property, i.e. 'K3='K3,th. If aeff >a t , then the fatigue limit can
be predicted by applying the classic Notch Mechanics criterion, i.e. 'Wg,th='W0/Ktg. The smooth
V *V
transition zones around a 0 t and a t (see figure 1) characterize the sensitivity to defects and
notches, respectively, and in this work an austempered ductile iron was studied.

TABLE 1. Geometries of the tested series

Torsion fatigue tests have been conduced on both smooth and notched specimens according to
the geometries listed in Table 1. A total number of 6 geometries were taken into account for a total
number of 72 fatigue tests. The adopted load ratio, defined as the ratio between the minimum and
the maximum applied torque, was equal to –1. Results have been summarised in the diagram,
which describes the sensitivity of the considered material to defects and notches in torsional
fatigue. Additionally, observations conduced by means of the Scanning Electron Microscope
enabled us to investigate the early crack propagation phase. As an example, fig. 2 shows small
cracks initiated after 2˜106 cycles from a graphite nodule at the tip of a 5-mm-deep notch having a
root radius equal to 0.1 mm.

References
1. Atzori, B., Lazzarin, P. and Meneghetti, G., Fatigue Fract. Engng Mater. Struct, vol. 26, 257-
267, 2003.
1. Atzori, B., Lazzarin, P. and Meneghetti, G., Int. J. Fract., vol. 133, 61-87, 2005.
 2T5. Fatigue and fracture 203

MULTI AXIAL FATIGUE IN WELDED COMPONENTS

G. Mesmacque, B. Wu1, C. Robin1, D. Zakrzewski1 and X. Decoopman

University of Science and Technology, Lille, France
1Ecole des Mines, Douai, France

gerard.mesmacque@univ-lille1.fr

In order to study the fatigue behavior of 6000 aluminium alloy welded structures and to take into
account the small size of the affected zone, a specific heat treatment was developped to get a
structure equivalent to that of the heat affected zone (simulated HAZ). Such an equivalent structure
was determined by comparing their hardness, microstructure and tensile properties. A new
cumulative damage law has been developed taking in account the loading history .

TABLE 1 Mechanical characteristics for simulated HAZ

FIGURE . Simulated Heat affected Zone Choice of Heat treatment

Several existing multiaxial fatigue criteria (Dang Van K.[1], Findley[2], Matake [3],
McDiarmid [4]) have been evaluated under Castem 2000. The misalignment, the local geometry
and the residual stress have been taken in account in the numerical model.
204 G. Mesmacque et al.

FIGURE 2. Numerical analysis and fatigue tests-Geometry of specimen real -simulated HAZ
Specimen were tested in the High Cycle Fatigue regime, under uniaxial and in-phase and out-
of-phase biaxial loadings. The results of all simulations were compared to the test data. The
position of the critical plane given by the simulations was also compared to the test results. Some
criteria (Dang Van, K. & McDiarmid) were found to be conservative. The criterion of McDiarmid
was found to be the best whatever the loading, while the criteria of Dang Van and Matake were
found to give out life predictions not enough precise, the proportion between the test results and
the life assessments being sometimes larger than 10. The criterion of Findley did not agree well
with the test results.
The cumulative damaging law [4] developed by G. Mesmacque and al [4] was used in blocks
loading history. Some results are presented in the frame of this work.
A general discussion in the light of the multi axial criterion in biaxial loading (in phase and out
phase) is made and compared to the IIW recommendations.
Keywords : (HCF) High Cycle Fatigue, Multiaxial Fatigue Criterion, Heat Affected Zone,
Aluminium Welded Joints

References
1. Dang Van K., Sciences et techniques de l'Armement, 1973, N° 47, 3ème fascicule
2. Findley W.N. Transaction of the American Society of Mechanical Engineers, 1957, Vol . 79.
3. D.L. Mc Diarmid Journal, July 1974,PP. 325-329.
4. G. Mesmacque International Journal of fatigue Volume 27, Issue 4, Pages 455-469 April
2005
 2T5. Fatigue and fracture 205

ENHANCED FATIGUE LIFE BY MECHANICAL SURFACE TREATMENTS –

EXPERIMENT AND SIMULATION

H. P. Gaenser, I. Goedor, H. Leitner and W. Eichlseder

Montanuniversität Leoben
A-8700 Leoben, Austria
{Hans-Peter.Gaenser|Istvan.Godor|Heinz.Leitner|Wilfried.Eichlseder}@mu-leoben.at

Although mechanical surface treatments are commonly used in engineering practice for their
beneficial effect on fatigue life, only few quantitative data are available from literature [1, 2]. This
contribution presents first results of a long-term program that aims at developing quantitative
design guidelines for the influence of mechanical surface treatment on fatigue life of components.
To this purpose, experimental work and numerical simulations are performed in parallel, giving a
detailed understanding of the effects of mechanical surface treatments on surface roughness,
residual stress distribution, and resulting fatigue strength.
Four-point rotating bending tests were performed on notched specimens of a quenched and
tempered low-alloy steel (34CrNiMo6) as well as of an aluminum wrought alloy (AlCuBiPb). The
specimens had a minimum diameter of 7.5 mm and a notch stress concentration factor of 1.8,
giving a generic geometry and stress gradient typical for usual rotating machine parts, cf. Fig. 1.
The heat-treated and machined specimens were subjected to two different mechanical surface
treatments: shot peening and deep rolling. In what follows, some preliminary results representative
for the underlying mechanisms and effects of surface treatments are presented.
Compared to the untreated specimens, deep rolling gives a very smooth surface and a fatigue
strength enhancement of as much as 27 %, cf. the S/N curve for 34CrNiMo6 in Fig. 1. Shot
peening results in a rough surface, but still in a fatigue strength enhancement of 18 %. In order to
test whether this discrepancy was due to the surface roughness of the shot peened specimens (Rmax
= 30 µm), some experiments with peened and subsequently polished specimens were conducted.
Contrary to the expected increase in fatigue strength due to polishing, and also contrary to the
findings reported in [2], the endurance limit of the peened and polished specimens was within the
scatter band of the peened, unpolished samples.

FIGURE 1. Rotating bending experiments (34CrNiMo6):

specimen (left), normalized S/N curves (right)

In order to gain a deeper understanding of the underlying mechanical processes, finite element
(FE) simulations of both surface treatments – deep rolling and shot peening – were performed. For
deep rolling, the entire specimen and process were modeled (Fig. 2). For shot peening, a
206 H. P. Gaenser et al.

representative volume element (RVE) was subjected to an impact of eight rigid particles, giving
approximately the same shot distribution as in experiment (Fig. 3).
For an assessment of the resulting residual stress distributions (Figs 2-3), the axial (rsp. in-
plane) stresses, compensating the bending stresses in rotating bending loading, are evaluated. The
maximum compressive residual stress is reached in a depth of approx. 0.7 mm in deep rolling of
aluminum, compared to 0.05 mm in shot peening of steel. One observes that, for the steel
specimens, the depth of the maximum residual stress due to shot peening is of the same order of
magnitude as the surface roughness, 0.03 mm. If the latter is removed by polishing, the residual
stress maximum will be reduced in order to restore overall stress equilibrium. In the present case,
the increase in fatigue strength due to a finer surface finish and the decrease due to a reduction of
the residual compressive stress are precisely compensating each other.

FIGURE 2. Finite element simulation of deep rolling (AlCuBiPb): FE mesh of specimen with tools
(left) and normalized resulting residual stress distribution in the notch ground (right)

FIGURE 3. Finite element simulation of shot peening (34CrNiMo6): representative volume

element with resulting surface structure (left) and normalized residual stress distribution (right)

References
1. Torres, M.A.S, and Voorwald, H.J.C., Int. J. Fatigue, vol. 24, 877-886, 2002
2. Schreiber, R., Untersuchungen zum Dauerschwingverhalten kugelgestrahlter Einsatzstähle –
Abschlussberich (Experimental investigation of the fatigue limit of shot-peened case-
hardened steels – final report) (in German), Universität Karlsruhe, 1975
 2T5. Fatigue and fracture 207

ANALYSIS OF REPAIRED ALUMINUM PANELS IN GENERAL MIXED-MODE

CONDITIONS

H. Hosseini-Toudeshky, M. Saber2 and B. Mohammadi3

Associate Professor, 2 & 3 Post Graduate Student
Aerospace Engineering Department, Amirkabir University of Technology, No. 424, Hafez Ave.,
Tehran, Iran
hosseini@aut.ac.ir

In the real applications, most of the cracked components are subjected to cracks in general mixed-
mode condition. A typical geometry and loading of single-side repaired panels in mixed-mode
conditions is shown in Fig. 1. A single-sided repair produces out-of-plane bending, which causes
stress variations over the thickness of the cracked panel. One of the most challenging aspects of
bonded composite repair technology has been the stress analysis of repaired panels and the
calculations of subsequent fracture parameters. The stress variations over the thickness of a
cracked plate in un-symmetric repair present a greater challenge in modeling due to the existence
of out-of-plane bending and developing highly complicated three dimensional stresses.
Non-uniform crack propagation along the panel thickness is expected due to the above reasons.
In the previously published works [1], the authors assumed that the crack-front remains
perpendicular to the panel surfaces during the crack propagation of the repaired panels in mixed-
mode conditions. But, experimental results show that crack growth rates are not uniform along the
thickness and the actual crack-front becomes curved shape as shown in Fig. 2. This figure also
shows that the crack surfaces are three dimensional. Therefore, fracture parameters analysis of the
repaired panels in general mixed-mode conditions are required to predict the crack re-initiation and
propagation lives and crack trajectories.
In this paper, three-dimensional finite element analysis of the repaired aluminum panels (AL
2024 T3) in general mixed-mode conditions using glass/epoxy composite patches with different
lay up configurations are performed. Theses analyses are based on linear elastic fracture mechanics
assumptions. Variation of stress intensity factors in repaired panels with composite patch along the
thickness in the mixed modes conditions is studied. The difficulty of this work arises from the
complexity of modeling and meshing. The sub-modeling technique is used and then a fine mesh is
generated at the crack-front area to calculate the fracture parameters.
The computational fracture analysis is based on the calculation of separated energy release
rates (SERRs) by the aid of the modified virtual crack closure technique (MVCCT) in order to
calculate the local SERR distributions along the crack-front. MVCCT has the advantageous of: 1-
ability to decouple of three modes of fracture 2- simplicity of formulation, but it imposes a regular
mesh around the crack-front.
Variation of stress intensity factors versus distance from the patch surface along the initial
crack-front has been shown in Fig. 3 for a typical patch lay up configuration. The variations and
magnitudes of the stress intensity factors in different modes show that none of them can be
neglected in crack re-initiation and propagation lives and crack-front configuration. These
behaviors are considerably changed with the changes of patch layers orientation and lay up
configuration which will be discussed in the full paper.
208 H. Hosseini-Toudeshky et al.

FIGURE 1. Typical geometry and loading of the repaired panels

FIGURE 2. Crack-front and crack surface configuration obtained from experiments

FIGURE 3. Variations of stress intensity factors along the crack-front (E=45°)

References
1. Hosseini-Toudeshky H., Mohammadi B. and Daghyani HR., “Mixed-Mode fracture analysis
of aluminum repaired panels using composite patches”, Accepted for publication in the
journal of composites science and technology (special issue dedicated to ETDCM-6), 2005.
 2T5. Fatigue and fracture 209

EFFECT OF STRAIN RATE ON FATIGUE BEHAVIOR OF ULTRAFINE

GRAINED COPPER

P. Gabor, H.J. Maier and I. Karaman1

Lehrstuhl für Werkstoffkunde, Universität Paderborn,
Pohlweg 47-49, 33098 Paderborn, Germany
1Department of Mechanical Engineering, Texas A&M University,

College Station, TX 77843, U.S.A.

hmaier@zitmail.uni-paderborn.de

In recent years, ultrafine grained (UFG) materials have received a great deal of attention, as they
exhibit superior combinations of high strength and enhanced ductility as compared to conventional
large grained alloys, c.f. Valiev et al. [1], Love and Valiev [2]. Until recently, most studies that
focused on rate sensitivity and work hardening behavior of UFG materials at different strain rates
were conducted under monotonic test conditions, e.g. Wang and Ma [3] and Li et al. [4]. However,
for many of the envisaged structural applications of these materials, cyclic properties and their
dependence on loading parameters will be a key issue.
The experiments were performed on oxygen-free high conductivity copper with 99.995%
purity acquired from Hitachi Cable Ltd..The UFG microstructures were obtained by equal channel
angular extrusion (ECAE). Small dog-bone type samples were electro-discharged machined from
the ECAE billets, which underwent up to 16 ECAE passes. In the low-cycle fatigue tests, strain
rate was varied between 1.2-1 s-1 and 1.2-5 s-1, and a total strain amplitude ('t/2) of 1 x 10-3 was
employed in all LCF tests. In addition to the as-ECAEd material, samples that had received an
additional heat treatments at 200 °C for 3 minutes and 170 °C for 2 hours, respectively, were
studied.
The ultrafine grained copper was found to be strain rate dependent at room temperature in all
tested microstructural conditions. The highest sensitivity to strain rate changes was observed for
the as-ECAEd material (Figure 1), where the stability of the microstructure is expected to be the
lowest. Interestingly, the stress amplitude and LCF life both increase at higher strain rate for all
specimens. The detrimental effect of lower strain rates on fatigue life is attributed to changes in the
deformation mechanisms.
210 P. Gabor et al.

FIGURE 1. Stress amplitude as a function of strain rate for LCF tests conducted at a total strain
amplitude of 1 x 10-3.

In order to shed light on the effect of the strain rate on the evolution of fatigue damage,
additional in-situ fatigue tests were performed using a small-scale load frame placed within an
environmental scanning electron microscope. The evolution of the surface features was inspected
periodically at given cycle intervals. The observed operating damage mechanisms were correlated
with the effect of strain rate on macroscopic stress-strain response and on resulting fatigue life.
Optimized heat treatments and ECAE process routes are discussed with respect to microstructural
stability and resulting fatigue life.

References
1. Valiev, R.Z., Kozlov, E.V., Ivanov, Y.F., Lian, J., Nazarov, A.A., Baudelet, B., Acta Metall.
Mater., vol. 42, 2467-2475, 1994.
2. Lowe, T.C., Valiev, R.Z., In: Investigations and Applications of Severe Plastic Deformation,
NATO Science Series, v.3 80, Kluwer Publishers, Dordrecht, 2000.
3. Wang, Y.M., Ma, E., Mater. Sci. Eng., vol A 375–377, 46–52, 2004.
4. Li, Y.J., Zeng, X.H., Blum, W., Acta Mater., vol. 52, 5009–5018, 2004.
 2T5. Fatigue and fracture 211

LUBRICANT EFFECTS ON PROPAGATION OF SURFACE-BREAKING

CRACKS UNDER ROLLING CONTACT LOADING

J. Lai and S. Ioannides

SKF RDC B.V., P.O. Box 2350, 3430 DT Nieuwegein, The Netherlands
Junbiao.Lai@skf.com

It has been well recognized that lubricants or fluids may have significant effects on propagation of
surface-breaking cracks under rolling contact loading. Various models have been proposed in the
literature [1,2] to describe different possible cracking mechanisms associated with the effects of
lubricant inside the crack, including fluid-lubricated crack faces, fluid seepage into the crack by
load, and fluid entrapment. To date, the solution to one fundamental problem is still lacking, i.e.
the amount of lubricant forced into the crack by load, and /or the amount of lubricant entrapped in
the crack. This is a problem of special technical relevance for bearings and gears, which operate
under oil- or grease-lubricated condition, since the viscosity of lubricant is largely dependent on
pressure and temperature. The lubricant pressure exerted on the crack faces depends, among
others, on the amount of lubricant seeping into or entrapped inside the crack. The amount of
lubricant seeping into the crack depends on the pressure at the crack mouth and the pressure-
dependent viscosity of the lubricant.

FIGURE 1. Schematic illustration of lubricant seeping into a slit by Hertzian pressure.

In the present paper an attempt is made to determine the amount of lubricant forced into a
surface-breaking crack by contact pressure based on the solution for viscous fluids in laminar flow.
Considering a 2D configuration as shown in Fig. 1, a close-form solution is derived of the lubricant
seepage depth L inside the slit under a moving Hertzian pressure p with speed V,

2 2b p( x)
L( x) B
3V ³0 P ( x)
dx
(1)
in which b is the half width of the Hertzian pressure, B is the opening displacement of the crack
mouth, and P is the viscosity of the lubricant. The temperature- and pressure-dependent viscosity
of fluids is represented by the Roelands relation. Equation (1) relates the amount of lubricant
seeping into a slit/crack to the load, rolling speed and viscosity of lubricant. Numerical
(integration) results show that the lubricant seepage depth decreases with load, owing to the
significant increase of the viscosity of the lubricant with pressure.
212 J. Lai and S. Ioannides

The seepage model (1) is implemented in finite element calculations to simulate lubricant
seeping, entrapment and pressurization in the crack during rolling of a roller over the crack. The
amount of lubricant seeping into the crack determines if the crack mouth will be closed or if the
lubricant will be entrapped in the crack. Traction on rolling contact surface and friction between
crack faces are taken into account in the FE calculations. Fig. 2 shows the FE simulation of
lubricant entrapment, pressurizing and releasing from the crack when a load is moving over the
crack. Stress intensity factors (KI and KII) at the crack tip are calculated, based on which crack
propagation in mixed mode is discussed.

FIGURE 2. Snapshots from FE simulation of lubricant entrapment in crack.

References
1. Bower, A.F., J. Trobology. vol. 110, 704-711, 1988.
2. Kaneta, M. and Murakami, Y., J. Tribology, vol. 113, 1991
3. R. Byron Bird, Warren E. Stewart and Edwin N. Lightfoot: "Transport Phenomena", John
Wiley & Sons, 1960, New York.
 2T5. Fatigue and fracture 213

COMPUTATIONAL MODELLING OF CRACK INITIATION IN A MIXING TEE

SUBJECTED TO THERMAL FATIGUE LOAD

I. Varfolomeyev
Fraunhofer-Institut für Werkstoffmechanik, Wöhlerstr. 11, 79108 Freiburg, Germany
igor.varfolomeyev@iwm.fhg.de

Thermal fatigue phenomenon is an important subject in the life-time management of piping

components of nuclear power plants. Especially in T-joints, high frequent temperature fluctuations
as a result of mixing of the hot and cold fluids may cause material degradation and thus affect the
component fatigue life. The European research project THERFAT was initiated to study the
significance of thermal fatigue issues and to improve analysis tools for structural integrity
assessments under thermal fatigue conditions, Metzner et al. [1].
Within the THERFAT consortium involving 16 research teams from 8 European countries,
several key aspects of the thermal fatigue evaluation were investigated. These included collation of
field experience, load determination, material characterization, integrity assessment, verification
tests. The overview and general approach to solving the problem are presented in [1], whereas
some results of investigations are provided in [2, 3].
This paper concerns with numerical computations of the stress and strain evolution in an
austenitic mixing tee subjected to thermal fluctuations, as produced by two mass flows with
different temperature and flow rates. Based on the stress and strain analysis results, predictions are
then made for fatigue crack initiation in the component.
Input data for the thermal-mechanical analyses were obtained through temperature history
measurements performed by another project partner (SPG, Dresden) on thin-walled models. Figure
1 gives an example of records for the fluid temperature at the inner surface in the mixing zone and
at the outer surfaces of the pipe. The temperature difference 'T in those experiments was about
90°C. In view of difficulties to define exact values of the heat transfer coefficient (D) and the
frequency of thermal fluctuations (f) for in-service conditions, the influence of D and f on the stress
state in a pipe was examined in a parameter study.
To investigate possible component behaviour, a fictitious mixing tee with the wall thickness of
10 mm, the outer diameter of 60 mm and the temperature difference up to 'T = 250°C was
investigated. Since no direct measurements of the thermal-hydraulic data for in-service conditions
are available, the above experimental results were scaled and used as boundary conditions in the
stress analyses.
Calculations revealed that local plastic deformations in the mixing area would be possible at
certain combinations of the thermal load parameters D, f and 'T. To properly describe elastic-
plastic stress and strain behaviour under cyclic loading, a Chaboche type material model after
Jiang [4] was employed taking into account combined isotropic and kinematic strain hardening, as
well as strain rate effects. The model has previously been adopted (Sester and Chauvot [5]) to
describe the ratchetting phenomena in a pipe subjected to thermal stratification.
Based on the results of the stress-strain calculations (Fig. 2) and making use of engineering
models describing short crack behaviour (see [5]), the crack initiation in the component can be
predicted.
214 I. Varfolomeyev

FIGURE 1. Fluid and outer wall temperature in the mixing zone.

FIGURE 2. Example of calculated stress versus strain evolution.

References
1. Metzner K.J., Braillard, O., Faidy, C., Marcelles, I., Solin, J., Stumpfrock, L., In Proceedings
of the 30th MPA-Seminar, Stuttgart, 2004, 17.1-17.12.
2. Ertem-Müller, S., Gerstner, L., Schippers, M., In Proceedings of the 3rd International
Conference on Fatigue of Reactor Components, Seville, 2004.
3. Paffumi, E., Nilsson, K.-F., Taylor, N.G., In Proceedings of the 2nd International Symposium
on Fatigue and Fracture Mechanics, Tampa, ASTM/ESIS, 2003.
4. Jiang, Y., Ph. D. Thesis, University of Illinois, Urbana-Champaign, 1993.
5. Sester, M., Chauvot, C., In Proceedings of the 26th MPA-Seminar, Stuttgart, 2000, 3.1-3.9.
 2T5. Fatigue and fracture 215

ESTIMATION OF CRITICAL STRESS INTENSITY FACTOR IN STEEL

CRACKED WIRES

J. Toribio, F. J. Ayaso, B. Gonzalez, J. C. Matos and D. Vergara

Department of Materials Engineering, University of Salamanca
E.P.S., Campus Viriato, Avda. Requejo 33, 49022 Zamora, Spain
Tel: (34-980) 54 50 00; Fax: (34-980) 54 50 02
toribio@usal.es

In the framework of damage tolerance analyses in structural engineering, a problem of major

concern is the formulation of a fracture criterion useful for engineering design against catastrophic
failure. This is particularly important in the case of high-strength structural members in the form of
bars, wires, strands, tendons and cables (e.g., high-strength steels for reinforcing and prestressing
concrete) which are axially loaded under very severe loads and frequently work in harsh
environments, so that they can suffer fatigue, stress corrosion cracking or corrosion-fatigue, thus
increasing the risk of catastrophic failure and reducing the structural life [1-3]
Two fracture criteria for prestressing steel cracked wires are considered. Firstly, a global
fracture criterion [4-6] is formulated on the basis of energetic considerations (strain energy release
rate concept) or compliance measurements. It only requires a single-parameter approach
accounting for the crack depth (related to the compliance), or an average value of the energy
release rate along the crack front, as follows:
G* = 12s-s+sKI2E'ds (1)
where s is the curvilinear coordinate over the crack line. Thus the SIF is:

KI* = [0.473 – 3.286 (a/D) + 14.797 (a/D)2]1/2 [(a/D) – (a/D)2]–1/4 V Sa (2)
where Vis the axial remote stress and a the crack depth.
A local fracture criterion [6-8] is also considered, according to which fracture takes place when
the maximum SIF along the crack line reaches a critical value, i.e., a local instability at a certain
point produces the catastrophic fracture of the overall flaw:

S up * K I (s) = K IC (3)
where * is the crack line. The SIF is in this case:
4 3
KI** = ¦ ¦
i 0 j 0
Cij (a/D)i (a/b)j V Sa 
iz1
(4)
where the coefficients Cij are given in [9] and a, b are the dimensions of the crack ( elliptical
shape).
In this paper an engineering estimation of the critical SIF is proposed for eutectoid steel
cracked wires subjected to axial loading. The method for estimating the critical SIF consists of
computing it by considering, in addition to the fatigue precrack, the extension of the subcritical
crack propagation before the final fracture situation of the catastrophic type.
Fig. 1 shows the critical SIF for the fracture tests in air using fatigue precracked samples taken
from eutectoid steel wires with different levels of cold drawing, where it is seen that cold drawing
216 J. Toribio et al.

clearly improves the fracture behaviour of the steels, in addition to the improvement of classical
mechanical properties (yield strength) which is the final aim of manufacturing.

FIGURE 1. Critical stress intensity factor for the fracture tests in air.

References
1. Elices, M., Llorca, J. and Astiz, M.A., in Handbook of Fatigue Crack Propagation in Metallic
Structures (edited by A. Carpinteri) pp. 191-220, Elsevier, Amsterdam (1994).
2. Barsom, J.M. and Rolfe, S.T., Fracture & Fatigue Control in Structures, Prentice-Hall, NJ,
(1987).
3. Toribio, J. and Valiente, A., Proceedings of the 13th European Conference on Fracture (ECF-
13), San Sebastián, Spain, 2000.
4. Elices M., in Fracture Mechanics of Concrete: Structural Application and Numerical
Calculation (edited by G.C. Sih and A. DiTommaso), pp. 226-271, Martinus Nijhoff,
Dordrecht (1985).
5. Athanassiadis A., Boissenot J.M., Brevet P., Francois D. and Raharinaivo A., Int. J. Fracture
vol. 17, 553-566, 1981.
6. Valiente, A., Ph. D. Thesis, Polytechnic University of Madrid, 1980.
7. Astiz M.A., Elices M. and Valiente A., in Fracture Control of Engineering Structures/ECF6
(edited by H.C. Van Elst and A. Bakker), pp. 65-74, EMAS, West Midlands (1986).
8. Bui H.D. and DangVan K., J. Mécanique Appliquée, vol. 3 (2), 205-225, 1979.
9. Astiz M.A., Int. J. Fracture, vol. 31, 105-124, 1986.
 2T5. Fatigue and fracture 217

LOW-CYCLE FATIGUE OF DIN 1.2367 STEELS IN VARIOUS TREATMENTS

C. C. Liu, J. H. Wu2 and C. C. Kuo

Department of Mechanical Engineering, Chien Kuo Technology University
No. 1, Jieshou N. Rd., Changhua City, Changhua County 500, Taiwan, R.O.C.
2Automotive Research and Testing Center
2No. 6, Lugong S. 7th Rd., Lugang Town, Changhua Country 505, Taiwan, R.O.C.

ccl@ctu.edu.tw, JuiHung@artc.org.tw, cckuo@ctu.edu.tw

DIN 1.2367 (X38CrMoV5) is currently one of the most often used chromium-molybdenum tool
steels, but which contains a higher molybdenum content. The higher molybdenum content gives
DIN 1.2367 better hot hardness and temper resistance compared to similar-type AISI H11 and H13
tool steels. The steel DIN 1.2367 has become of increasing importance for various applications,
such as gears, shear blades, mandrels punches and tooling dies, due to its high hot hardness and
toughness, and good resistance to softening, wear and thermal fatigue.
The mechanical, friction and wear properties of DIN 1.2367 tool steels have been extensively
investigated as a function of tempering treatment and associated microstructure [1,2]. Little work
has been done on the strain-controlled low-cycle fatigue (LCF) behaviour of this steel. As some
applications of DIN 1.2367 tool steels are subjected to high cyclic plastic deformation which
causes micro chipping and fracture in service, it is important to characterize the LCF behaviour of
this steel in order to better predict the fatigue lifetime for components made from this material.
The objective of the present study is, macroscopically and microscopically, to characterize the
effects of tempering and plasma-nitriding treatments on the LCF behaviour of DIN 1.2367 tool
steels by systematic experiments. Appropriate fatigue life prediction approaches are proposed to
correlate the LCF lifetime data obtained for this DIN 1.2367 steels under various treatment
conditions. Literatures related to this work are listed in references 1-7.
The LCF behaviour has been investigated for DIN 1.2367 tool steel in seven different
conditions including tempering and plasma-nitriding treatments. The LCF results indicated that
the fatigue life for the three tempered conditions took the following sequence: 580oC temper >
540oC temper > 620oC temper. For the three tempered conditions with plasma-nitriding, the
differences of fatigue life were significantly decreased and the strain-life curves almost merged
together. Furthermore, due to the formation of brittle Fe4N compounds on the specimen surface
easily resulting in crack initiation, the fatigue life was apparently decreased after plasma-nitriding
for each tempered condition. Fractography observations indicated that all of the fatigue fracture
surfaces for the plasma-nitrided specimens exhibited a common fracture mode dominated by brittle
cracking with no obvious crack initiation sites. In addition, the LCF lifetime data generated under
seven different treatment conditions for the given DIN 1.2367 tool steel are well correlated with a
yield-strength-normalized Smith-Watson-Topper (SWT) parameter in a log-log linear relationship,
proposed by Lin and Chu [7], for which the correlation coefficient, r2, is up to 0.83.

References
1. Mebarki, N., Delanges, D., Lamesle, P., Delmas, F. and Levaillant, S., Mater. Sci. Eng. A,
vol. 387-389, 171-175, 2004.
2. Barrau, O., Boher, C., Gras, R. and Rezai-Aria, F., Wear, vol. 255, 1444-1454, 2003.
218 C. C. Liu et al.

3. ASTM E606-80, Annual Book of ASTM Standards, vol. 3.01, 609-615, 1991.
4. Bannantine, J. A., Comer, J. J. and Handrock, J. L., Fundamentals of Mental Fatigue
Analysis, Prentice-Hall Press, New Jersey, 1990.
5. Nakagawa, H. and Miyazaki, T., J. Mater. Sci., vol. 34, 3901-3908, 1999.
6. Mar, S., Li, Y. and Xu, K., Surf. Coat. Tech., vol. 137, 116-121, 2001.
7. Lin, C. K. and Chu, C. C., Fatigue Fract. Eng. Mater. Struct., vol. 23, 545-553, 2000.
 2T5. Fatigue and fracture 219

IMPACT TESTING A CAPABLE METHOD TO INVESTIGATE THE FATIGUE

RESISTANCE

K. David, P. Agrianidis, K. G. Anthymidis1 and D. N. Tsipas2

Department of Mechanical Engineering, Technological University of Serres,
Terma Magnesias str., GR 62124, Serres, Greece
1Serres Applied Research Center, Terma Magnesias str., GR 62124, Serres, Greece
2Department of Mechanical Engineering, Aristotle University of Thessaloniki,

54006, Thessaloniki, Greece

kanth@teiser.gr

The modern power generation steam turbines are being designed to have higher efficiencies and to
meet the stringent environmental regulations, ensuring plant reliability, availability and
maintainability without compromising cost. High efficiencies can be achieved by higher
temperatures. Therefore, the operating temperature is expected to rise from 550oC to 650oC and
from the material perspective to implement turbine components protected by spallation and
oxidation resistant coatings.
To guarantee the reliability of coated steam turbines components used in power plants, the
lifetime assessment of the coatings and their failure prediction become very important.
Microhardness, scratch adhesion and pin-on-disc sliding tests are commonly used for rapid
evaluation of the mechanical properties of coatings. However, they do not model the dynamic
cyclic impact fatigue. Recently the impact test method has been introduced as a convenient
experimental technique to evaluate the fatigue strength of coatings being exposed in alternate
impact loads, Bouzakis et al. [1]. According to this method a coated specimen is exposed to a
cyclic impact load. The superficially developed Hertzian pressure induces a complex stress field
within the coating, as well as in the interfacial zone. Both these stress states are responsible for
distinct failure modes such as a cohesive or adhesive one. The exposure of the layered compounds
against impulsive stresses creates the real conditions for the appearance of coating fatigue
phenomena based upon structural transformation, cracking generation and cracking growth, which
are responsible for the gradual microchipping and the degradation of the coating.
The objective of this experimental study was to investigate the influence of the impact stress
fields on the performance and fatigue strength of thermal spray HVOF coatings. Furthermore, the
overall aim of the current research was to prove the reliability of the impact test, as a new testing
method, to assess the coating lifetime against fatigue, to interpret the failure modes of coatings,
and thereby to exam its capability, whether this non-standard test can be used as a useful method
for the development and optimisation of fatigue resistant coatings working under impact loading.
Figure 1 shows the impact test rig where the experiments have been conducted.
220 K. David et al.

FIGURE 1. The impact test rig.

The stress strain problem related to the impact test is the Herzian contact, which develops
between the spherical indentor (carbide ball) and the examined layered space. The contact load
leading to coating fatigue fracture was recorded in fatigue-like diagrams (endurance strength
curves) versus the number of impacts. Gradual intrinsic coherence release and coating
microchipping or abrupt coating fracture and consequent exposure of the substrate material
designate the coating failure. The coating failure mode and its extent were assessed by SEM
observations and EDX analysis. In case of relatively tough coating microstructure with high wear
resistance as the WC-CoCr thermal spray coating is, the coating layer sustains the cyclic impacts
without any sign of cohesive delamination failure. Instead of that, only superficial abrasive wear
and spalling failure has been observed (Fig 2). This behaviour can be attributed to the improved
fracture toughness of this coating.

FIGURE 2. WC-CoCr coating failure initiation (cohesive failure mode) and microhardness
measurement of the layered compound.

References
1. Bouzakis, K., Vidakis N. and David, K., J. of Thin Solid Films, vol. 351, 1-8, 1999.
 2T5. Fatigue and fracture 221

COMPARATIVE ASSESSMENT OF FATIGUE-THRESHOLDS ESTIMATED BY

SHORT AND LONG CRACKS

K. K. Ray, N. Narasaiahb and S. Tarafderb

Department of Metallurgical and Materials Engineering, Indian Institute of Technology,
Kharagpur-721302, India,
b National Metallurgical Laboratory, Jamshedpur-831007, India.

kkrmt@metal.iitkgp.ernet.in, nn@nmlindia.org, star@nmlindia.org

The safe life design philosophy of structural components is based on the knowledge of the fatigue
threshold for long cracks ('Kth). The emergence of the concept of short cracks, its numerous
experimental verification and the continued studies on its growth behaviour over the past years
have brought forward challenges to the basis of the safe life design philosophy. However, short
cracks interestingly exhibit either single or multiple-thresholds. But prior investigations related to
the determination and examinations of fatigue threshold for short cracks ('Kthsc) in structural
materials are limited in number. In an earlier communication [1], the largest value amongst the
detected thresholds associated with short cracks has been referred as “near long crack fatigue
threshold” (NLFTH) and the magnitude of NLFTH has been hypothesized as close to the fatigue
threshold obtained from long crack experiments as illustrated using Fig.1. This investigation aims
to explore the magnitudes of 'Kthsc and 'Kth for a few steels, to determine the NLFTH values for
these materials and finally to make a comparative assessment of NLFTH with their corresponding
'Kth values.

Fig.1 Schematic view of the critical crack length at the transition between short and long crack and
their fatigue thresholds. The points B and C are the thresholds for short and long cracks. The point
D (in part a) or B and C (in part b) indicates the transition length [1].
Experimental investigations on short crack growth behaviour and measurement of long crack
fatigue thresholds have been carried out using some recently developed specimens [1-3] which can
be coupled to rotating bending machine for the convenience of carrying out fatigue tests at
relatively high frequency. Schematic configurations of these specimens are shown in Fig.2.

Fig.2 Configuration of specimens used for (a) short crack and (b) long crack growth experiments.
222 K. K. Ray et al.

The experiments have been carried out on four plain carbon steels and the generated data of
crack growth rates against stress intensity factor range have been analyzed to examine the
threshold values, 'Kth, 'Kthsc and NLFTH. The magnitudes of 'Kth have been estimated by load
shedding procedure whereas the magnitudes of 'Kthsc have been evaluated from the short crack
growth behaviour. An alternative set of experiments has been conducted to examine the
appropriateness of the 'Kth values obtained using the new specimen configuration by comparison
with standard conventional tests using compact tension specimens. Some typical results on 'Kth,
'Kthsc and NLFTH are depicted in Fig.3. The fatigue crack growth studies have been
supplemented with characterization of the microstructure, hardness and tensile property of the
selected steels. In addition, the effect of the associated microstructures on the investigated short
crack paths has been carefully examined.
The transition of short to long crack in the investigated steels was found to be dependent on the
nature of the short cracks. But, the maximum fatigue threshold values obtained from short crack
growth experiments on various specimens were found to be independent of the nature of the short
cracks. This maximum values of 'Kthsc, referred here as NLFTH for short cracks are interestingly
found to be in good agreement with their corresponding 'Kth values. The short cracks are
considered to be un-influenced by the crack closure effect, whereas the long cracks considered in
these experiments are observed to have little effect due to crack closure at R =-1. These aspects are
considered to lead the above agreement. The current observations can be supported by some earlier
reports of James and Knott [4], Sadananda and Vasudevan [5] and Taylor [6].

References
1. Ray K. K., Narasaiah N., and Sivakumar R., Mater. Sci. Engg. A, vol. 372, 81-90, 2004.
2. Narasaiah N., Initiation and growth of cracks near fatigue threshold in plain carbon steels,
Ph.D Dissertation, IIT, Kharagpur, India, 2004.
3. Narasaiah N. and Ray K. K, Role of microstructure on short crack propagation and its
threshold, Paper No. 5075, ICF11, Turin , Italy, 2005
4. James M. N. and Knott J. F., Fat. Fract. Eng. Mater. Struct., vol.8, 177-183, 1985
5. Sadananda K. and Vasudevan A. K. , Int. J. Fatigue. vol.19, S99-103
6. Taylor, D., Fatigue Thresholds, Butterworths and Co., London, p-135 1989.
 2T5. Fatigue and fracture 223

SCANNING ELECTRON MICROSCOPE MEASUREMENTS OF CRACK-

OPENING STRESS ON FATIGUE CRACKS EXPOSED TO OVERLOADS

L. Jacobsson and C. Persson

Materials Engineering, Lund University, Sweden
Box 118, 221 00 Lund, Sweden
lars.jacobsson@material.lth.se

A fatigue crack exposed to an overload change the characteristics of propagation due to the
enlarged plastic zone at the crack tip. The residual stresses in the material surrounding the crack tip
lead to a crack closure phenomenon, and when the global applied stress gives a zero stress at the
crack tip, the crack start to open. The enlarged plastic zone changes the crack closure and opening
stresses that affect the effective stress intensity factor that is a measure of the stresses acting at the
crack tip.
To measure the global stress when the crack tip starts to open, different techniques are used.
The use of a travelling microscope gives visual observations of the crack, and the observations can
be done to detect when the crack opens. The compliance-offset method is used to measure the
crack closure stress from the reload compliance curve that is produced from an extensometer
measuring the displacement in the crack mouth. Variations in the potential drop signal throughout
a load cycle are used by Andersson et al. [1] to identify opening and closure stresses. Shimojo et al.
[2] uses an electrochemical technique to detect the creation of fresh surface within the crack. They
discovered that the plastic zone expand more in the region with plane stress condition close to the
surface after an overload, and when material was removed from the surface both the retardation of
the crack propagation rate and crack closure decreased. During in-situ fatigue crack experiments
Halliday et al. [3] observed differences in crack closure stresses and near tip COD for the load
conditions R = 0.05 and -1. They dismiss the influence of mechanisms such as crack tip blunting
and strain hardening.
The aim of this study were to make more exact measurements of crack opening and closure
stresses and crack shape very local around the crack tip to find relations to the crack propagation
rate after an overload. This was done from high-resolution scanning electron microscope images
that were analyzed with an image analysis program. Experiments with different R-values and
overload ratios were performed.
To detect the crack shape and the crack opening and closure stresses, an image analyzing
technique is used together with in-situ fatigue crack propagation experiments within a scanning
electron microscope. Throughout the load cycle, high-resolution images are taken of the crack. A
selected spot in one image can be re-found with a cross-correlation algorithm in images at other
stress conditions. With this technique the displacement field around the crack is determined
throughout the load cycle and the stresses when the crack opens and closes can be accurately
measured. Also, the crack shape and compliance curve were determined. In fig. 1 a) the
compliance curve is plotted where displacements are measured a few micrometers from the crack
tip and fig. 1 b) show the crack shapes, from the crack tip to 0.1 mm, for increasing loads.
224 L. Jacobsson and C. Persson

FIGURE 1. a) Compliance curve where the displacements are measured at the crack tip. (o)
increasing load, () decreasing load. b) Crack shapes for different applied loads.

References
1. Andersson, M., Persson, C., Melin, S., ICF 9, 2003
2. Shimojo, M., Chujo, M., Higo, Y., Nunomura, S., Int. J. Fatigue, vol. 20, no. 5, 365-371,1998
3. Halliday, M.D., Zhang, J.Z., Poole, P., Bowen, P., Int. J. Fatigue, vol. 19, no. 4, 273-282,
1997
 2T5. Fatigue and fracture 225

PROPAGATION PATH AND FATIGUE LIFE PREDICTIONS OF BRANCHED

CRACKS UNDER PLANE STRAIN CONDITIONS

M. A. Meggiolaro, A. C. O. Miranda1, J. T. P. Castro and L. F. Martha2

Dept. Mech. Eng., 1Tecgraf, 2Dept. Civil Eng., Pontifical Catholic University of Rio de Janeiro
Rua Marquês de São Vicente 225, Gávea – Rio de Janeiro, RJ, Brazil 22453-900
meggi@mec.puc-rio.br, amiranda@tecgraf.puc-rio.br, jtcastro@mec.puc-rio.br, lfm@tecgraf.puc-
rio.br

Fatigue cracks can significantly deviate from their Mode I growth direction due to the influence of
overloads, multi-axial stresses, micro structural inhomogeneities such as grain boundaries and
interfaces, or environmental effects, generating crack kinking or branching (Lankford and
Davidson [1]). The stress intensity factors (SIF) associated to branched fatigue cracks can be
considerably smaller than that of a straight crack with the same projected length, causing crack
growth retardation or even arrest, as discussed by Suresh [2]. This mechanism can quantitatively
explain retardation effects even when plasticity induced crack closure cannot be applied, e.g. in
high R-ratio fatigue problems under plane strain conditions. Analytical solutions have been
obtained for the SIF of some branched cracks, however numerical methods such as the ones
presented in Miranda et al. [3] are the only means to predict the subsequent curved propagation
behaviour.
In this work, a specialized Finite Element program is used to calculate the propagation path
and associated SIF of bifurcated cracks. The numerical calculations are validated through
experiments on 4340 steel ESE(T) specimens. A total of 6,250 FE calculations are used to fit
empirical equations to the process zone size and crack retardation factor along the curved crack
branches. The bifurcation simulations include several combinations of bifurcation angles, branch
asymmetry ratios, crack growth exponents, and even considers interaction between crack
branching and other retardation mechanisms such as crack closure, assuming the crack opening
level is well known.
It is shown that very small differences between the lengths of the bifurcated branches are
sufficient to cause the shorter one to eventually arrest as the longer branch returns to its pre-
overload conditions, see Fig. 1. The process zone size is found to be smaller for lower bifurcation
angles and for branches with greater asymmetry, in both cases due to the increased shielding
effects on the shorter branch. Higher crack closure levels also result in smaller process zones,
because the shorter branch is more easily arrested due to the reduction in its stress intensity range.
However, a competition between smaller process zone sizes and lower growth rates of the longer
branch takes place to determine the real effect of combined bifurcation and closure.
The proposed equations can be readily used to predict the propagation behaviour of branched
cracks in an arbitrary structure, as long as the process zone is small compared to the other
characteristic dimensions. From these quantitative results, it is shown that crack bifurcation may
provide a sound alternative mechanistic explanation for overload-induced fatigue crack retardation
on structural components.
226 M. A. Meggiolaro et al.

FIGURE 1. Bifurcated crack propagation behaviour.

References
1. Lankford, J., Davidson, D.L. Advances In Fracture Research, vol. 2, 899-906, 1981.
2. Suresh, S., Fatigue of materials, Cambridge University Press, U.K., 1998.
3. Miranda, A.C.O., Meggiolaro, M.A., Castro, J.T.P., Martha, L.F., Bittencourt, T.N., Engng.
Fracture Mechanics, vol. 70, 1259-1279, 2003.
 2T5. Fatigue and fracture 227

SHORT CRACK EQUATIONS TO PREDICT STRESS GRADIENT EFFECTS

IN FATIGUE

M. A. Meggiolaro, A. C. O. Miranda1, J. T. P. Castro and J. L. F. Freire

Dept. Mechanical Engineering and 1Tecgraf, Pontifical Catholic University of Rio de Janeiro
Rua Marquês de São Vicente 225, Gávea – Rio de Janeiro, RJ, Brazil 22453-900
meggi@mec.puc-rio.br, amiranda@tecgraf.puc-rio.br, jtcastro@mec.puc-rio.br,
jlfreire@mec.puc-rio.br

It is well known that the notch sensitivity factor q can be associated with the presence of non-
propagating cracks. Such cracks are present when the nominal stress range 'Vn is between 'V0/Kt
and 'V0/Kf, where 'V0 is the fatigue limit, Kt is the geometric and Kf the fatigue stress
concentration factors of the notch. Therefore, in principle it is possible to obtain expressions for q
if the propagation behaviour of small cracks emanating from notches is known.
Several expressions have been proposed to model the dependency between the threshold value
'Kth of the stress intensity range and the crack size a for very small cracks, see Chapetti [1]. Most
of these expressions are based on length parameters such as El Haddad-Topper-Smith’s a0 [2],
estimated from 'Kth and 'V0, resulting in a modified stress intensity range

2
'K I 'V ʌ( a  a 0 ) 1 § 'K th ·
, a0 ¨ ¸ (1)
S © 'V 0 ¹
which is able to reproduce most of the behaviour shown in the Kitagawa-Takahashi plot. Yu et al.
[3] and Atzori et al. [4] have also used a geometry factor D to generalize the above equation to any
specimen, resulting in

2
1 § 'K th ·
'K I D ˜ ' V ʌ( a  a 0 ) a ¨ ¸
, 0 S © D ˜ 'V 0 ¹ (2)

Peterson-like expressions are then calibrated to q based on these crack propagation estimates.
However, such q calibration is found to be extremely sensitive to the choice of 'Kth(a) estimate.
In this work, a generalization of El Haddad-Topper-Smith’s equation, which better correlates
with experimental crack propagation data collected from Tanaka et al. [5] and Livieri and Tovo
[6], is proposed:

1 / n
' K th ª § a ·n / 2 º
«1  ¨ 0 ¸ »
'K 0 «¬ © a ¹ »¼
(3)
where 'K0 is the threshold stress intensity factor for a long crack with load ratio R = 0. In the
above equation, n is typically found to be between 1.5 and 8.0, see Fig. 1. Clearly, Eqs. (1) and (2)
are obtained from Eq. (3) when n = 2.0.
228 M. A. Meggiolaro et al.

FIGURE 1. Ratio between long and short crack propagation thresholds as a function of a/a0.
Equation (3) is used in a Finite Element program to evaluate the behavior of cracks emanating
from circular and elliptical holes. For several combinations of notch dimensions, the smallest stress
range necessary to both initiate and propagate a crack is calculated, resulting in expressions for Kf
and therefore q. It is found that the q estimates obtained from this generalization better correlate
with experimental crack initiation data. Expressions for the maximum admissible flaw sizes at a
notch root are also obtained.

References
1. Chapetti, M.D., Int. J. of Fatigue, vol. 25, 1319–1326, 2003.
2. El Haddad, M.H., Topper, T.H., Smith, K.N., Engng. Fract. Mech., vol. 11, 573-584, 1979.
3. Yu, M.T., DuQuesnay, D.L., Topper, T.H., Int.J.Fatigue, vol. 10, 109-116, 1988.
4. Atzori, B., Lazzarin, P., Meneghetti, G., Fatigue Fract. Engng. Mater.Struct., vol. 26, 257-
267, 2003.
5. Tanaka, K., Nakai, Y., Yamashita, M., International J. of Fracture, vol.17, 519-533, 1981.
6. Livieri, P., Tovo, R., Fatigue Fract. Engng. Mater.Struct., vol. 27, 1037-1049, 2004.
 2T5. Fatigue and fracture 229

FATIGUE BEHAVIOUR OF PRE-STRAINED TYPE 316 STAINLESS STEEL

Masayuki Akita, Masaki Nakajima1, Keiro Tokaji2 and Yoshihiko Uematsu2

Faculty of Eng., Gifu Univ., 1-1 Yanagido, Gifu 501-1193, Japan
1 Dept. of Mech. Eng., Toyota National College of Technology,

2-1 Eisei-cho, Toyota 471-8525, Japan

2 Dept. of Mech. and Systems Eng., Gifu Univ., 1-1 Yanagido, Gifu 501-1193, Japan

akita@cc.gifu-u.ac.jp

Type 316 stainless steel has excellent corrosion resistance, which has been widely used for
machine components and structures. Materials are usually subjected to plastic deformation by
processing or forming. Therefore, it is very important to understand the effect of pre-strain on
fatigue behaviour (Radhakrishnan and Baburamani [1], Yokotsuka and Ikegami [2]). The present
paper describes the fatigue behaviour of pre-strained 316 stainless steel. Rotating bending fatigue
tests have been performed using specimens subjected to tensile pre-strains and the fatigue
behaviour was discussed.
The material used is a 316 stainless steel with a diameter of 16 mm. The chemical composition
(wt.%) is C: 0.04, Si: 0.23, Mn: 1.33, P: 0.33, S: 0.03, Ni: 10, Cr: 16.9, Mo: 2.01. The material was
solution treated at 1353 K for 1 h. After solution treatment, two different tensile pre-strains of 5 %
and 15 % were given to the material from which hourglass-shape fatigue specimens with a
diameter of 5 mm were machined. Vickers hardness was 137HV for the virgin specimen, 177HV
for the 5% pre-strained specimen and 214HV for the 15% pre-strained specimen. Before fatigue
test, specimens were mechanically polished by emery paper.

FIGURE 1. S-N diagram.

Fatigue tests were performed using cantilever-type rotating bending fatigue testing machine
operating at a frequency of 53 Hz. Crack initiation and small crack growth were monitored with
replication technique. After experiments, fracture surfaces were examined in detail by a scanning
electron microscope (SEM).
The S-N diagram is shown in Fig.1. It can be seen that fatigue strength increases with
increasing pre-strain level. Only a slight increase in fatigue strength can be seen in the 5% pre-
strained specimens, while a very large increase is attained in the 15% pre-strained specimens. The
fatigue limits of the virgin, 5% pre-strained and 15% pre-strained specimens are 300 MPa, 320
230 M. Akita et al.

MPa and 380 MPa, respectively. Fatigue tests were continued to 108 cycles, but no failure took
place.
Based on the observation of crack initiation and growth, the increase of fatigue strength due to
pre-strain was attributed to increased crack initiation resistance, because small crack growth was
not affected significantly by pre-strain.
Figure 2 illustrates the results of stress-incremental fatigue tests. Experiments were started
from the stress level of 20 MPa lower than the fatigue limit for each specimen. When the
specimens were not fractured until 107 cycles, then the stresses were raised by 20 MPa. As seen in
the figure, the fatigue limits are increased significantly in the virgin and 5% pre-strained specimens
(27% and 25% increase, respectively), while no increase is seen in the 15% pre-strained specimen,
suggesting that the coaxing effect depends on the initial condition of specimens. The hardnesses of
fractured specimens were increased remarkably in the virgin and 5% pre-strained specimens, while
a slight increase in the 15% pre-strained specimen. Therefore, the coxing effect is strongly related
to the ability of work hardening during stress cycling.

FIGURE 2. Stress-incremental fatigue test results.

References
1. 1. Radhakrishnan, V.M. and Baburamani, P.S., Mater. Sci. Eng., 17, 283-288, 1975.
2. 2. Yokotsuka, T. and Ikegami, K., J. Soc. Mat. Sci., Japan, 48, 38-43, 1999.
 2T5. Fatigue and fracture 231

THE INFLUENCE OF CONSTRAINT ON FITTING FATIGUE CRACK

GROWTH DATA

M. Carboni and M. Madia

Politecnico di Milano
Via La Masa 34, I-20156 Milano, Italy
carboni@mecc.polimi.it, mauro.madia@mecc.polimi.it

Fracture mechanics, in terms of crack propagation, is one of the most popular approaches for life
prediction of components and structures subjected to fatigue loads. Experimental tests (as an
example for a structural steel see Fig. 1) are carried out in order to quantify crack growth rates and
thresholds.

FIGURE 1. Typical experimental results obtained from fracture mechanics tests on a structural
steel.

The obtained results, together with dedicated numerical analyses, are used in the calibration
process of crack propagation analytical models, such as the Strip-Yield model (Newman [1]) able
to keep into account the crack closure phenomenon. These kind of empirical models, initially
introduced for aeronautical applications, are included in widespread life prediction softwares, such
as AFGROW [2] and NASGRO [3], which adopt, as propagation law, the so-called “NASGRO
Equation”
p
§ ' K th ·
1
º ¨© ' K ¸¹
n
da ª§ 1  f ·
C «¨ ¸ 'K » q
dN ¬© 1  R ¹ ¼ § K max ·
¨1  ¸
© K crit ¹
(1)
where the parameter “f” is named “Newman’s closure function” and is a function a the constraint
factor “D” [1] used to introduce in the calculations the real 3D stress field at the crack tip.
The existing software incorporate numerical procedures for fitting parameters of Eq. (1) by
using da/dN data without any explicit reference to test conditions. On the other hand, fatigue crack
growth data obtained with different experimental conditions (specimen type, maximum applied
stress) are also differing in D at the crack tip and so the direct fitting of Eq. (1) is not correct.
In this paper we examine a set of crack growth data obtained onto a mild structural steel with
different specimens (SE(B) and C(T)) from the point of view of constraint factor at the crack tip. In
particular, constraints have been analysed with a series of detailed FEM analyses (Fig. 2).
232 M. Carboni and M. Madia

Results have allowed us to show how apparently conflicting data (especially in terms of 'Kth)
can be re-analysed considering D factors at the crack tip in the different tests.

FIGURE 2. Analysis of the constraint factor at the tip of a SE(B) specimen

References
1. Newman, J. C. Jr., ASTM STP 748, 53-84, 1981.
2. Anonymous. 2001, NASA Technical Report JSC-22267B, Website: www.nasgro.swri.org,
2001.
3. Harter, J. A., U.S. Air Force Research Laboratory Technical Report AFRL-VA-WP-TR-2002-
XXX, Website: http://afgrow.wpafb.af.mil, 2002.
 2T5. Fatigue and fracture 233

ATOMIC FORCE MICROSCOPY OF LOCAL PLASTIC DEFORMATION FOR

TEMPERED MARTENSITE

Masao Hayakawa, Saburo Matsuoka1 and Yoshiyuki Furuya

Materials Information Technology Station, National Institute for Materials Science (MITS/NIMS),
1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan
1Faculty of Engineering, Kyushu University, Fukuoka, Japan

HAYAKAWA.Masao@nims.go.jp

An AFM technique established for observing two types of fine and complicated tempered
martensite structures was developed into a technique for quantitatively evaluating local plastic
deformation near tensile yield points. Surface steps with 5-40 nm were observed in locally
deformed martensite blocks near prior austenite grain boundaries.
The material used was JIS-SCM440 steel (containing of 0.40%C, 0.25%Si, 0.79%Mn,
1.12%Cr, 0.17%Mo; mass%). The 7.75 mm-diameter bar was austenized for 720 s at 1323 K, and
then rolled at 1048-1053 K to the total area reduction was 50% as a modified-ausforming treatment
[1]. The bar was tempered for 10 s at 803 K by induction heating. (This was called MAQT;
Modified-Ausformed-Quenced and Tmepered.) Another bar was austenized for 0.9 ks at 1153 K
and then tempered for 1.8 ks at 673 K. (This was called CQT; Conventional Quenched and
Tempered.) Tensile strength showed the same value of 1580 MPa.
AFM observation was conducted under the tapping mode in the atmosphere. The detail of an
AFM observation method is described in the reference [2]. Briefly, the method is: 1) taking AFM
images of an electropolished surface (see Figs. 1(a) and 2(a)); 2) taking AFM images of the same
location after the tensile test (tensile plastic strain: 0.2, 0.4, 0.6%, respectively) after finding the
position using a micro-Vickers indent as a reference mark (see Figs. 1(b) and 2(b)); 3) Finally,
prior austenite grain boundaries in the chemically corroded surface were identified.
Figures 1(a) and 2(a) show AFM images of the electropolished surfaces for MAQT and CQT
before the tensile test, respectively. The black and white contrast in the images show the level
differences on the surface attributable to differences in crystal misorientation with high-angle
boundaries. Therefore, each strip of uniform brightness corresponds to a martensite block that has
a high-misorientation-angle boundary. The mean block widths were 0.38 and 0.49 m for MAQT
and CQT, respectively. Cementite particles, which project out from the base metal, are shown as
white spots.
Figures 1(b) and 2(b) are images created by superimposing prior austenite grain boundaries on
the AFM images of the electropolished surfaces for MAQT and CQT after the tensile tests,
respectively. The horizontal direction is the tensile loading direction. The local plastic deformation
occurred in relative large blocks near prior austenite grain boundaries. The surface steps caused by
local deformations in MAQT were lower than in CQT, since the martensite blocks in MAQT were
finer than in CQT. The values of the surface steps corresponded to the number of pile-up
dislocation at block boundaries.
234 M. Hayakawa et al.

FIGURE 1. AFM images of the electropolished surface before and after the tensile test in (a)
plastic strain 0% and (b) 0.6% for MAQT. Prior Ȗ grain boundaries were shown in (b).

FIGURE 2. AFM images of the electropolished surface before and after the tensile test in (a)
plastic strain 0% and (b) 0.6% for CQT. Prior grain boundaries were shown in (b).

References
1. Yusa S., Hara T., Tsuzaki K. and Takahashi T., CAMP-ISIJ vol.12, 1296, 1999
2. Hayakawa M., Matsuoka S. and Tsuzaki K., Mater. Trans., vol. 43, 1758-1766, 2002
 2T5. Fatigue and fracture 235

IMPROVEMENT OF FATIGUE STRENGTH DUE TO GRAIN REFINEMENT IN

MAGNESIUM ALLOYS

Mitsutoshi Kamakura, Keiro Tokaji1, Hideaki Shibata2 and Norikatsu Bekku3

Graduate Student, Gifu Univ., Yanagido, Gifu 501-1193, Japan
1Dept. of Mech. Systems Eng., Gifu Univ., Yanagido, Gifu 501-1193, Japan
2Gifu Prefectural Research Institute, Oze, Seki 501-3265, Japan
3Honda Material Industries, Inc., Yokohira, Ooi-machi, Ena 509-7201, Japan

kamakura@metal.rd.pref.gifu.jp

Magnesium (Mg) alloys are very attractive materials for structural applications because of
excellent specific strength. However, their absolute strengths are insufficient, thus it is necessary to
further improve the mechanical properties, particularly fatigue strength. One of the methods for
improving mechanical properties is grain refinement, but studies on grain refinement in Mg alloys
are very limited (Yamashita et al. [1], Kumar et al. [2]). The purpose of the present study is to
achieve grain refinement due to controlled extrusion and associated improvement of fatigue
strength in wrought Mg alloys. First, grain refinement due to extrusion was studied and then the
fatigue behaviour of the extruded materials was discussed.
Billets (grain size: 200-250Pm) of AZ61A and AZ31B were extruded at an extrusion ratio of
67. The most important parameter influencing grain refinement is working temperature that was
controlled to be low (L), middle (M) and high (H) temperatures. The microstructures on the cross
section in AZ61A and AZ31B are shown in Figs1 and 2, respectively. In both alloys, grain size
decreases with decreasing working temperature and grain refinement is much more remarkable in
AZ31B. The average grain sizes are 12.1Pm, 12.7Pm and 5.8Pm for AZ61A-H, AZ61A-M and
AZ61A-L, 7.4Pm, 2.9Pm and 2.1Pm for AZ31B-H, AZ31B-M and AZ31B-L, respectively.

FIGURE 1. Microstructures in AZ61A: (a) AZ61A-H, (b) AZ61A-M, (c) AZ61A-L.

FIGURE 2. Microstructures in AZ31B: (a) AZ31B-H, (b) AZ31B-M, (c) AZ31B-L.

236 M. Kamakura et al.

FIGURE 3. S-N diagram.

FIGURE 4. Grain size dependence of proof stress and fatigue strength.

Rotary bending fatigue tests were performed in laboratory air using smooth specimens of the
extruded materials in both alloys. The S-N diagram is shown in Fig.3. In AZ61A, the grain size
dependence of fatigue strength is less remarkable, but the fine-grained material (AZ61A-L)
exhibits slightly higher fatigue strength in long life regime. On the other hand, in AZ31B, fatigue
strength increases with decreasing grain size. Based on observation of crack initiation and growth,
such grain size dependence of fatigue strength in AZ31B resulted from both improved crack
initiation resistance and small crack growth resistance. Proof stress and fatigue strength at 107
cycles are represented in Fig.4 as a function of grain size, where both strengths were expressed
properly with the Hall-Petch relation in AZ31B, but not in AZ61A.

References
1. 1. Yamashita, A., Horita, Z. and Langdon, T.G., Mater. Sci. Engng, A300, 142-147, 2001.
2. 2. Kumar, N.V.R., Blandin, J.J., Desrayaud, C., Montheillet, F. and Suéry, M., Mater. Sci.
Engng, A359, 150-157, 2003.
 2T5. Fatigue and fracture 237

A UNIFIED FATIGUE AND FRACTURE MODEL APPLIED TO STEEL WIRE

ROPES

M. P. Weiss, R. Ashkenazi and D. Elata

Faculty of Mechanical Engineering, Technion IIT
Haifa 32000, Israel
mweiss@technion.ac.il, matron@netvision.net.il, elata@technion.ac.il

The behavior of short and very short fatigue cracks, emanating from so called “smooth” specimens
with stress concentrations has been an intriguing research topic for a long time. It is well known
that micro-cracks are embedded in the surface of any smooth specimen and are created during the
manufacturing process. A few of these micro-cracks will eventually propagate due to fatigue loads
and will become major cracks. Due to a combination of high local stresses and stress
concentrations, the short cracks are the first to become the major cracks and eventually cause
failure.
A quantitative two-term model for a step by step simulation of crack propagation from very
short cracks to fracture has been proposed by one of the authors, Weiss [1] and Weiss and
Hirshberg [2]. The model is based on the Fatigue Diagram that segregates the whole fatigue and
fracture domain into 6 unique zones, and relates each zone to a known fatigue and/or fracture
regime. Zone 4 in the diagram is the most prevalent in industry, as it is so bounded that the stress
amplitudes are higher than the endurance stress, but lower than the elastic limit, and the stress
intensity factor ranges are higher than the threshold and lower than the fracture toughness. Most
metal parts and structures that are loaded by alternating loads, finally break down in this zone,
either by gross yielding or by critical crack propagation. In this zone the superposition of two
fatigue crack propagation rules is used.
In the current study, the model is described, discussed and enhanced and is shown to be
applicable to fatigue failure in single wires of steel wire ropes in general and specially in non-
rotating tower crane wire ropes.
In its service life a wire rope is subjected to fluctuating tension loading and to bending over
sheaves. The bending stresses along the individual double helix wire are strongly dependent on the
sheave-to-wire diameters ratio and on the arrangement of the reeving system. Predicting the
fatigue life of a wire rope is considered as an essential objective for users and manufacturers.
Generally, the failure of the individual wire is caused due to an initial crack that propagates to a
critical length. In order to improve the safety level in hoisting appliances, the development of a
model that will reliably predict the fatigue life and the safe service life of the rope, is needed.
Moreover, a reliable model based on all the stress affecting factors, will considerably reduce
complex and expensive experimental study.
In this study a new approach to simulate the fatigue life of individual wires within a wire rope,
that is subjected to tension-tension and bending over sheave is presented. The model is based on
the two-term fatigue propagation model [1] and on a new analysis of the stresses that simulates
bending and tension stresses that are generated along the double helix wire within a rope, as shown
by the authors, Elata et al. [3]. It is assumed that initial flaws that were generated during the cold
drawing manufacturing process exist in the individual wire. Several of these flaws (initial cracks)
are located at critical sites where tension and bending stresses are maximal. Once the hoisting
cycle is defined by means of the rain-flow method, it is converted to an “equivalent” loading cycle
with a completely reversed loading (R=-1) by employing the Gerber parabolic line. The growing of
the assumed initial crack is then simulated according to the appropriate zone on the fatigue
238 M. P. Weiss et al.

diagram [1]. Material parameters are calibrated according to experimental fatigue data that was
measured on eutectoid steel wires.
An un-lubricated rope is assumed. Accordingly, relative displacements between adjacent wires
are not permitted. The stress simulation model fully considers the configuration of the straight and
bent double helix wire. The fatigue life of individual wires within an 18u7 non rotating rope are
simulated for various reeving systems, D/d ratios and tension loads. This type of rope is mainly in
use in tower crane applications.
The theoretical results were tested on regular, industrial wires, using a specially built tension
apparatus, that is free to rotate on one side, so that the twisting moments and deformations, that are
generated in tension, can be measured. The results of the stresses were found to be close to the
analytical predictions. Fatigue life was found to be in good agreement with Feyrer 's [4; 5]
experimental prediction formula. Moreover, simulation results predict extensive fatigue
deterioration of the wires within the inner strand layer, under free end attachment using a swivel,
just as is customary in many tower cranes.
The model quantitatively emphasizes the strong dependence of the fatigue life on the quality of
the surface finish of the wires, namely on the residual micro-cracks that were generated during the
deep drawing plastic deformation, in the diameter reduction process of the single wires in
manufacturing.
The new approach may be applied to predict the fatigue life of a complicated cross section of a
rope within a hoisting system. Moreover, the model will extensively reduce expensive
experimental study and can also be used to design better wire configurations. The demonstrated
application in wire ropes is just one example for the use of the two-term model in fatigue life
prediction of complicated mechanical parts underd different stress combinations and histories.

References:
1. Weiss M.P., Int. J. of Fatigue, Vol. 14, No. 2, pp. 91-96, 1992
2. Weiss M.P. and Hirshberg Z., Fatigue & Fracture of Engineering Materials & Structures,
Vol. 19, No. 2\3, pp. 241-249, 1996
3. Elata D., Eshkenazy R and Weiss M.P., International Journal of Solids and Structures, 2004,
Vol 41 /5-6, pp. 1157-1172
4. Feyrer K., Drahtseile, Springer-Verlag Berlin, Heidelberg, 1994
5. Feyrer K., Wire, vol. 45 (2) pp.99-103, 1995
 2T5. Fatigue and fracture 239

CORRELATION BETWEEN PARIS’ LAW PARAMETERS BASED ON SELF-

SIMILARITY AND CRITICALITY CONDITION

A. Carpinteri and M. Paggi

Department of Structural and Geotechnical Engineering, Politecnico di Torino
Corso Duca Degli Abruzzi 24, 10129 Torino, Italy
alberto.carpinteri@polito.it, marco.paggi@polito.it

Fatigue crack growth data are usually presented in terms of the crack growth rate, d a / d N , and
the stress-intensity factor range, ' K . The typical fatigue crack propagation curve is shown in
Fig.1, where Region I is referred to as the near-threshold region, Region II as the power-law region
and Region III as the rapid crack propagation region where K max o K IC and crack growth
instability occurs. In Region II the Paris’ equation (Paris and Erdogan [1]) provides a good
approximation to the majority of experimental data:

da
C ('K )m
dN (1)
where C and m are empirical constants usually referred to as Paris’ law parameters.

FIGURE 1. Scheme of the typical fatigue crack propagation curve. Point CR corresponds to the
onset of crack growth instability.

From the early 60’s, research studies have been focused on the nature of the Paris’ law
parameters, demonstrating that C and m cannot be considered as material constants. In fact, they
depend on the testing conditions, such as the loading ratio R (Radhakrishnan [2]), on the size of the
specimen (Barenblatt [3]), and, as pointed out very recently, on the initial crack length (Spagnoli
[4]).
However, an important question regarding the Paris’ law parameters still remains to be
answered: are C and m independent of each other or is it possible to find a correlation between
them based on theoretical considerations? Concerning this point, it is important to take note of the
controversy about the existence of a correlation between C and m in the Literature (Radhakrishnan
[5], Cortie [6], Bergner and Zhouar [7]). However, a very consistent empirical relationship
between the Paris’ law parameters is usually represented by the following formula [5]:
240 A. Carpinteri and M. Paggi

C AB m
, (2)
where parameters A and B depend on the material being studied and are obtained from a best-fit
procedure on experimental data.
In the present paper, the correlation existing between the Paris’ law parameters is derived on
the basis of theoretical arguments. To this aim, both self-similarity concepts [3] and the condition
that the Paris’ law instability corresponds to the Griffith-Irwin instability at the onset of rapid crack
growth (see point CR in Fig.1) are profitably used. Comparing the functional expressions derived
according to these two independent approaches, a relation between the Paris’ law parameters C and
m formally similar to Eq.(2) is proposed. Parameter A is found to be dependent on the ultimate
size of the inelastic zone ahead of the crack tip and on the Carpinteri’s brittleness number
(Carpinteri [8]). On the other hand, parameter B turns out to be dependent on the loading ratio and
on the critical stress-intensity factor. The main consequence of these relations is that only one
macroscopic parameter is needed for the characterization of damage during fatigue crack growth.
An experimental assessment of this new correlation is proposed for a wide range of materials
including steels, Aluminium alloys, epoxy resins with liquid-filled urea-formaldehyde
microcapsules and polymer/silica interfaces. The effect of the loading ratio on the parameter C is
also deeply discussed. A good agreement with experimental data is achieved, showing the
effectiveness of the new proposed correlation.

References
1. Paris P.C. and Erdogan F., J. Basic Engineering, vol. 85D, 528-534, 1963.
2. Radhakrishnan V.M., Engineering Fracture Mechanics, vol. 11, 359-372, 1979.
3. Barenblatt G.I., Scaling, self-similarity and intermediate asymptotics, Cambridge University
Press, Cambridge, 1996.
4. Spagnoli A., Mechanics of Materials, vol. 37, 519-529, 2005.
5. Radhakrishnan V.M., Engineering Fracture Mechanics, vol. 13, 129-141, 1980.
6. Cortie M.B., Engineering Fracture Mechanics, vol. 30, 49-58, 1988.
7. Bergner F. and Zouhar G., International Journal of Fatigue, vol. 22, 229-239, 2000.
8. Carpinteri A., Materials and Structures, vol. 14, 151-162, 1981.
 2T5. Fatigue and fracture 241

THERMO-MECHANICAL FATIGUE LIFETIME ASSESSMENT WITH

DAMAGE-PARAMETERS, ENERGY-CRITERIONS AND CYCLIC-J-
INTEGRAL CONCEPTS

M. Riedler, R. Minichmayr, G. Winter and W. Eichlseder2

University of Leoben, Christian Doppler Laboratory for Fatigue Analysis
Chair Mechanical Engineering2
Franz-Josef-Str. 18, A-8700 Leoben
Martin.Riedler@notes.unileoben.ac.at

The simulation of the thermo-mechanical fatigue (TMF) behaviour of cylinder heads is an

important design step in the automotive industry. The steady rise of engine power and the demand
of lightweight construction with a concurrent enhanced reliability require an optimised
dimensioning process. The goal of this paper is to apply classical damage parameters, plastic and
total energy criterions and cyclic J-integral concepts for a thermo-mechanical lifetime assessment
of aluminium and cast iron alloys.
Many of the empirical models are strain based criterions like the Manson-Coffin [1-2] criterion
with numerous modifications. Criterions based on damage parameters try to find a correlation
between the number of cycles to failure and loading parameters. The fracture mechanical view
allows e.g. with cyclic J-integrals a description of the TMF lifetime. Cumulative models, as the
Chaboche [3] models, try to cumulate the damage for each cycle, therefore they need a lot of
computing time for complex structures. The use of submodels is one possibility of using them for a
TMF lifetime assessment of cylinder heads. Another method is the accumulation of the specific
damage parts (pure fatigue, oxidation, creep), as e.g. Miller [4] (accumulation of crack propagation
rates) or Neu/Sehitoglu [5] (accumulation of damage rates) do it. Microstructural methods have a
physical background, but are often difficult to use for a TMF lifetime estimation.
It is shown in previous work (Riedler [6]), that energy based criterions are qualified for TMF
lifetime approaches. Due to the interplay of strain and stress values at a TMF loading they are able
to account for more specific influences as well as their interactions. The general energy based
lifetime criterion is as follows:

'W e ˜ f ('We )  p ˜ f ('W p ) Nf A ˜ 'W  B

, (1)
The specific energy is the sum of an elastic and plastic energy function, whereas both are
characterised with strain and stress parameters, as e.g. amplitude, maximum, minimum or effective
values. A power law is used to describe the dependency of the lifetime.
The plastic energy criterion W1 [6] is best suited for the very ductile Al-alloy AlCuBiPb to
account for the influences of a varying maximum temperature, dwell time, mean strain, pre-ageing
and ageing in service time. The classical damage parameter according to Smith-Watson-Topper [7]
gives the best result for the ductile aluminium cast alloys AlSi7MgCu0.5 and AlSi8Cu3. For
inductile materials like alloy AlSi6Cu4 a criterion based on fracture mechanics in the manner of a
cyclic J-integral approach according to Tomkins [8] provides the best TMF lifetime estimation
result. A TMF lifetime assessment for the different materials investigated shows a reasonable
scatter of about 2.5, see Fig. 1. Since only the Tomkins approach as a sum of an elastic and plastic
energy part [8]:
242 M. Riedler et al.

2
§ S ˜ V max 2S ˜ V max ˜ 'H p · B
'J a ˜¨  ¸ N § 'J ·
¨ ¸ f A˜¨ ¸
© E 1  n' ¹ , © a ¹ (2)
which is a special type of Equ. (1), never gives a rank worse than five (of ten investigated
criterions) for the different materials researched, a standard deviation minimisation process to
determine an appropriate energy definition based on Equ. (1) is executed. This allows us to
describe the TMF life of different materials and material properties with an improved accuracy.

FIGURE 1. Simulation of the TMF lifetime.

References
1. Manson, S.S, Behaviour of materials under conditions of thermal stress, NACA Report No.
1170, 1954.
2. Coffin, L.F, Trans. ASME, vol. 76, 931-950, 1954.
3. Lemaitre, J. and Chaboche, J.L., Mechanics of solid materials, Cambridge Univ. Press, 1990.
4. Miller, M.P., McDowell, D.L., Oehmke, R.L.T. and Antolovich, S.D., In Proceedings of
Thermo-Mechanical Fatigue Behaviour of Materials, ASTM STP 1186, edited by H.
Sehitoglu, Philadelphia, 1993, 35-49.
5. Neu, R.W. and Sehitoglu, H., Metals Trans. A, vol. 20A, 1989, 1755-1767 and 1769-1783
6. Riedler, M.: PhD Thesis, University of Leoben, 2005.
7. Smith, K.N., Watson, P. and Topper, T.H., J. of Materials, vol. 5, No. 4, 767-778, 1970.
8. Tomkins, B., Sumner, G. and Wareing, J., In Proceedings of International Symposium on Low
Cycle Fatigue Strength and Elasto-Plastic Behaviour of Materials, edited by K.-T. Rie and E.
Haibach, DVM, Berlin, 1979, 495-508.
 2T5. Fatigue and fracture 243

PREDICTING FATIGUE CRACK RETARDATION FOLLOWING OVERLOAD

CYCLES

Marcos V. Pereira, Fathi A. Darwish1, Arnaldo F. Camarao2 and Sérgio H. Motta3

Department of Materials Science and Metallurgy / Catholic University of Rio de Janeiro
Rua Marquês de São Vicente 225, Rio de Janeiro / RJ, CEP 22453-900, Brazil
1Fluminense Federal University

Rua Passo da Pátria 156, Niterói / RJ, CEP 24210-240, Brazil

2GEP / ArvinMeritor CVS

Av. João Batista 825, Osasco / SP, CEP 06097-900, Brazil

3Technology Division / Brasilamarras

Rua Eng. Fábio Goulart 40, Niterói / RJ, CEP 24050-090, Brazil
marcospe@dcmm.puc-rio.br, fadarwish@poscivil.uff.br, arnaldo.camarao@arvinmeritor.com,
shamotta@brasilamarras.com

Dating back to the beginning of the seventies, a number of models have been proposed to predict
fatigue crack growth rate under variable amplitude loading. This effort was motivated by earlier
observations that the application of an overload is followed by crack growth retardation over a
crack length increment. The model of Willenborg, which belong to the group of yield zone models,
incorporate interaction effects and is characterized by introducing crack tip plasticity. Although
interaction models are generally considered to be applicable for high strength alloys with limited
ductility, empirical verification of the predictions made by these models is rather limited.
Accordingly, the present study was initiated in an effort to test the validity of the Willenborg
model for predicting fatigue crack growth retardation in an R3 grade structural steel following the
application of overload cycles. As this grade steel is used for fabricating offshore mooring chains,
the study was extended to also include flash welded joints taken from the chain links.
CT specimens were machined from both the base metal as well as from the welded joints and a
number of these specimens were subjected to a heat treatment that involved quenching and
tempering. The CT specimens, both heat treated and untreated, were fatigue tested under constant
amplitude (CA) loading in order to establish the typical da/dN versus K curves. During CA fatigue
loading, some specimens in different material conditions, were subjected to single and multiple
overloads applied at a given crack length, and crack growth rate da/dN was followed as a function
of K, evidencing the retardation in crack propagation over an interval of crack length. Crack
propagation rate da/dN within the delay period was predicted by Willenborg model and then
compared with the experimental data. Finally, the results are presented and discussed focusing on
the comparison between the predictions made by the model in light of the experimental data.
244 M. V. Pereira et al.

FIGURE 1. Variation of crack size (a) with the number of cycles (N) for flash welded joint after
overloading.
 2T5. Fatigue and fracture 245

FATIGUE CRACK GROWTH AT NOTCHES CONSIDERING PLASTICITY

INDUCED CLOSURE

Jutta Bruening, Olaf Hertel, Michael Vormwald and Georgios Savaidis

Technische Universität Darmstadt, Materials Mechanics Group
Petersenstraße 12, D-64287 Darmstadt, Germany
bruening@wm.tu-darmstadt.de
hertel@wm.tu-darmstadt.de
vormwald@wm.tu-darmstadt.de
Aristotle University Thessalonica, Division of Mechanical Engineering, Laboratory of Machine
Elements and Machine Design
GR-54124 Thessalonica, Greece
gsavaidis@meng.auth.gr

Classical fatigue analyses discriminate between technical crack initiation (crack length of about
1mm) and crack propagation stages. The stage of crack initiation, however, is itself dominated by
the growth of short fatigue cracks. Based on the assumption that a fatigue life to initiate a micro-
structurally short crack of dimensions in the order of 10Pm may be neglected Dankert et al. [1]
have proposed a so-called unified elastic-plastic model for fatigue crack growth evaluation which
describes the whole fatigue life (technical crack initiation and stable crack growth) of notched and
unnotched components by integrating an appropriate crack growth law. In the meantime this model
has entered a guideline for the proof of the strength of components [2].
The present paper reports on the results of an investigation on this model which led to an
improvement of the model’s assessment of effective ranges of the crack driving force (i.e. the J-
Integral). Furthermore, a generalization enabling its application for arbitrary notch situations is
presented. Additionally, new results are shown of comparisons of experimentally determined and
calculated crack growth curves as well as life curves for both the initiation of a technical crack
(crack depth 1mm) and the final failure .
The basic modules of the model are the following:
The range of the cyclic J-integral is taken as the relevant crack driving force. This is essential
because generally short fatigue cracks grow in a notch field where elastic-plastic conditions prevail
in the uncracked situation. Therefore, the limits for the application of the linear fracture mechanics
are violated. Nevertheless, the special case of the linear fracture mechanics based calculation
(negligible cyclic plastic deformation) is incorporated. Approximation formulas for the J-integral
of semi-elliptical and quarter-elliptical surface cracks as well as through-the-thickness cracks in
components with elliptical notches are presented.
A procedure for assessing cracks in arbitrarily notched components is outlined such that this
problem can be reduced to the solved case with elliptical notches.
Effective ranges are considered exclusively. This means that only the range between the crack
closure level and the peak level of a cycle is taken to calculate the J-integral range.
The approximation formulas for the crack opening level are taken as proposed in the literature
by Dankert et al. [1] and Savaidis and Dankert [3].
however, the reference load for cyclic loading is taken to arrive at appropriate effective J-
integral ranges instead of its monotonic counterpart.
A Paris-type crack growth equation is formulated in terms of this effective J-integral range.
Integration of this equation leads to the crack growth curves and fatigue lives.
246 J. Bruening et al.

The integration starts at a suitably chosen initial crack length. The relevant criterion is that the
integration starting at this very initial crack length will result in the material’s strain-life curve for
the special case of an unnotched material specimen.
Variable amplitude loading is taken into account by identifying cycles according to the
rainflow algorithm.
Saal [4] has performed well documented experiments on the initiation (technical crack size
1mm) and growth of fatigue cracks in notched plates of low alloyed steel. The comparison of the
life curves with the corresponding curves calculated by applying the improved unified model
reveals the reasonable accuracy of the model. The differences of the new results compared to those
of the original model are exemplified.

References
1. M. Dankert, S. Greuling, T. Seeger. A Unified Elastic-Plastic Model for Fatigue Crack
Growth at Notches Including Crack Closure Effects.Symposium on Advances in Fatigue
Crack Closure Measurement and Analysis, San Diego, Calif., 1997, ASTM, 1999, STP 1343,
S.480
2. FKM-Richtlinie ”Bruchmechanischer Festigkeitsnachweis”, 2. Auflage, VDMA Verlag
GmbH, Frankfurt am Main, to be published 2005
3. G. Savaidis, M. Dankert, T. Seeger, An analytical procedure for predicting opening loads of
cracks at notches. Fatigue and Fracture of Engineering Materials and Structures, 1995,
18(4), S. 425-442
4. Saal, H., Einfluß von Formzahl und Spannungsverhältnis auf die Zeit- und Dauerfestigkeit
und Rißfortschreitungen bei Flachstäben aus St52, Heft 17, Institut für Stahlbau und
Werkstoffmechanik, Technische Universität Darmstadt, 1971
 2T5. Fatigue and fracture 247

INFLUENCE OF MICROSTRUCTURE ON FATIGUE PROPERTIES OF NI-

BASE SUPERALLOY AT ELEVATED TEMPERATURE

QY. Wang , Y. Matsuyama1, N. Kawagoishi1, M. Goto2 and K. Morino3

Sichuan University, Chengdu 610065, China
1Kagoshima University, 1-21-40 Korimoto, Kagoshima, Japan
2Oita University, 700 Tannoharu Oita, Japan
3Tokuyama College of Technology, 3538 Kume Takajyo Shunan, Japan

wang@imr.ac.cn, hiro@mech.kagoshima-u.ac.jp, masagoto@cc.oita-u.ac.jp,

morino@tokuyama.ac.jp

Ni-base superalloys are used under sever conditions like a high temperature or sever corrosive
environment because of their superior properties on corrosion, creep, static strength at high
temperatures and so on. Therefore, many studies on creep and fatigue properties of the alloy have
been carried out, e.g. Brown and Hicks [1]. Recently, it is very important to know the fatigue
properties in long life region, because machines and structures are used for long-term from the
point of view of environmental and economical demands. However, most of the studies on fatigue
were very limited within the short life up to107cycles. It is reported that fracture occurs from an
internal inclusion in long life region in many high strength steels and surface treated steels, while
surface fracture occurs in short life region, e.g. Murakami et al. [2]. Moreover, the fatigue
properties may be influenced by microstructure. However, the mechanism and the evaluation
method for fatigue life of internal fracture are not clarified.
In the present study, fatigue properties of Inconel 718 and the influence of grain size on the
properties were investigated in the wide region of fatigue life until 108 cycles at room temperature
and 500 under rotating bending. Fatigue tests were carried out using materials with two kinds of
grain sizes, about 20m and 100m. Table 1 shows the mechanical properties of the materials. Prior
to fatigue tests, all the specimens were electro-polished about 20m from the surface layer.
Observations of the change in the surface state of a specimen due to stress repetitions and fracture
surface were carried out under an optical microscope using plastic replication technique or under a
scanning electron microstructure (SEM) directly.
Fig. 1 shows S-N curves. Fatigue strength is high in the fine grain at both temperatures,
especially at 500. Although, all the fracture originated from the surface except for the fine grain at
500 in long life region, resulting duplex S-N curve.
Fig. 2 shows crack growth curves. There is no or little differences in the crack initiation life
248 QY. Wang et al.

and the growth rate of a larger crack between the fine grain and the coarse grain at both
temperatures. However, the growth rate of a crack smaller than about 200m is higher in the coarse
grain than in the fine grain at room temperature. This is caused that the crack length initiated is
larger in the coarse grain than in the fine grain. On the other hand, a definite suppression of a crack
growth at 500, when a crack is small. The suppression is caused by oxide induced crack closure at
500. These are main reasons for the difference in the influence of grain size on the fatigue strength
at both temperatures.

References
1. Brown, C.W. and Hicks, M.A., Int. J. Fatigue, vol. 4, 73-81, 1982
2. Murakami, Y., Yokoyama, N. N. and Nagata, J., Fatigue Fract. Engng. Mater. Struct., vol.
25, 735-746, 2002
 2T5. Fatigue and fracture 249

MODDELLING FATIGUE CRACK CLOSURE USING DISLOCATION

DIPOLES

P. F. P. de Matos and D. Nowell

University of Oxford, Oxford, UK
paulo.dematos@eng.ox.ac.uk, david.nowell@eng.ox.ac.uk

Over the past 30 years the study of fatigue crack closure has been one of the most challenging
research topics in fatigue crack growth. This phenomenon was first noticed by Elber [1] and then
has been documented by several researchers, both experimentalists and analysts. Different closure
mechanisms have been observed, although the primary mechanism of closure seems to be
plasticity-induced. One of the reasons why fatigue crack closure has been so extensively studied is
because of the need of predicting fatigue lives of structural components, e.g. aircraft industry. It
has been proved that fatigue crack closure plays an important role in crack growth rates together
with load ratio and load history.
To quantify the crack closure effect, different techniques can be used, e.g. experimental,
numerical and analytical. Usually experimental techniques are not used for practical purposes as
they are time consuming and because of the high cost of the experimental work. Numerical
techniques such as FEM modelling have been successfully used to model crack closure. However
they are time consuming since elasto-plastic models are required to accurately model the plastic
deformation ahead the crack tip and the plastic wake. Usually FEM results are dependent on the
level of mesh refinement, crack-tip node release scheme etc. Concerning the analytical techniques
different approaches can be used. A novel technique was developed by Newman [2], his model is
based on the Dugdale model but was modified to leave deformed material on the wake of the
advancing crack. This model is dependent on an empirical constraint factor which is used to
establish the link between the plane stress and plane strain condition. Furthermore, yield strip
models are only suitable to model cracks under plane stress; therefore the use of this constraint
factor for modelling plane strain is only an approximation since the deformation mechanism is
completely different.
In the present paper a quadratic programming technique is used to model fatigue crack closure.
The model herein used was developed by Nowell [3] for modelling fatigue cracks under plane
stress. In the present paper we are going to show how this model can be changed to model crack
closure under plane strain conditions. In the original model, Nowell [3] used displacement
discontinuity opening dipoles (see FIGURE 1) to model both crack and plastic region ahead of the
crack tip. For modelling crack closure under plane strain this model was changed in order to take
into account the different deformation mechanism. Sliding dislocation dipoles were collocated
along the planes of maximum shear stress to quantify the plastic deformation along these planes
and at the tip of the crack. Initial results compare well with those given by Kanninen and
Atkinson’s [4] superdislocation model for the case of a static crack, FIGURE 2. The aim of the
present paper is to investigate the deformation of the crack tip in both deformation modes plane
stress and plane strain. The results are compared with different methods, e.g. FEM and Newman
model [2]. The case of more general loading conditions under plane strain will be discussed.
250 P. F. P. de Matos and D. Nowell

FIGURE 1. a) Opening and sliding dislocation dipoles; b) Dipole displacement field.

FIGURE 2. Crack tip opening displacement. Present model vs Kanninen and Atkinson [4].

References
1. Elber W., Engineering Fracture Mechanics, vol. 2(1), 37-44, 1970
2. Newman Jr. J.C., ASTM STP, vol. 748, 53-84, 1981.
3. Nowell D., Fatigue & Fracture of Engineering Materials & Structures, vol. 21(7), 857-871,
1998.
4. Kanninen M.F. and Atkinson C., International Journal of Fracture, vol. 16(1), 53-69, 1980.
 2T5. Fatigue and fracture 251

COMPARISON BETWEEN FATIGUE CRACK GROWTH MODELLED BY

CONTINOUS DISLOCATION DISTRIBUTIONS AND DISCRETE
DISLOCATIONS

P. Hansson, S. Melin and C. Persson1

Division Mechanics, 1Division of Materials Engineering
Lund Institute of Technology, Box 118, SE-22100 LUND
Per.Hansson@mek.lth.se

Short fatigue cracks are known to have a growth behaviour different from that of long cracks, the
latter well predicted by linear elastic fracture mechanics. Short cracks can grow at high rates at
load levels well below the threshold value for long cracks, before entering into the long crack
region, or arrest and become nonpropagating cracks. The growth behaviour of short cracks is
strongly influenced by the microstructure of the material, such as grain boundaries and direction of
slip planes within the grains, as well as of local plasticity around the crack tip. Microstructurally
short cracks, typically shorter than a few grains, grow in a single shear mechanism along specific
slip planes within the grains, cf. Suresh [1], leading to a zigzag crack path, cf. Fig.1.

FIGURE 1. Zigzag shaped edge crack.

For low growth rates, in the order of a few burgers vectors per cycle, it is important to account
for the discrete dislocations within the material. Studies taking the nucleation, movement and
annihilation of discrete dislocations along specific slip planes into account have been performed by
Riemelmoser et al. [2] to study the cyclic crack tip plasticity for a long mode I crack. A similar
model have also been developed by Bjerken and Melin [3] to study the influence of grain
boundaries on a short propagating mode I crack, subjected to fatigue loading. Another approach
was used by Krupp et al. [4] to study the propagation of short cracks in a duplex steel. In this
model, the plastic zone was described by dislocation elements distributed along the slip planes, and
the crack growth rate was assumed to be determined by the crack tip shear displacement.
In this study, the propagation of a short edge crack, situated within one grain in a bcc material
subjected to fatigue loading is modelled using a dislocation formulation. The external boundary,
defined as the free edge and the crack itself, is modelled by dislocation dipole elements, consisting
two glide dislocations and two climb dislocations, allowing the crack surfaces to both open and
shear. The plasticity is modelled by a continuous distribution of dislocation elements along the slip
planes, allowing the slip planes only to shear. The grain boundary is treated as an impenetrable
hinder for the dislocations, and the crack is assumed to be the only dislocation source in the
initially dislocation free material. A schematic description of the model is seen in Fig. 2.
252 P. Hansson et al.

As the applied load exceeds a certain critical value, cf. [5], dislocations will nucleate at the
crack tip or a corner point of the crack, forming plasticity along the slip planes. As the load is
increased, the plastic zone is increased until maximum load is reached and the load is reversed.
During unloading dislocations will move back towards the crack and annihilate, resulting in crack
growth in the corresponding direction under the assumption that no healing of the crack surfaces is
allowed. To describe the plasticity with continuous dislocation distributions instead of with
discrete dislocations as used in earlier studies by Hansson and Melin [5], creates a more efficient
model as regards to computational time, which, in turn allows for simulations at higher growth
rates and growth through additional grains. Though, the approach results in decreasing accuracy in
describing the plastic zone and loss of physically correct dislocation distribution as compared to
the approach in [5].

FIGURE 2. Schematic description of the model.

References
1. Suresh, S. Fatigue of Materials, second edition. University Press, Cambridge, 1998.
2. Riemelmoser, F.O, Pippan, R. and Kolednik, O. Cyclic crack growth in elastic plastic solids:
a description in terms of dislocation theory. Computational Mechanics, 20:139-144, 1997.
3. Bjerkén, C. and Melin, S. A study of the influence of grain boundaries on short crack growth
during varying load using a dislocation technique. Engineering Fracture Mechanics,
71(15):2215-2227, 2004.
4. Krupp, U., Düber, O., Christ, H.-J. and Künkler, B. Application of the EBSD technique to
describe the initiation and growth behavior of microstructurally short fatigue cracks in a
duplex steel. Journal of Microscopy., 213(3):313-320, 2003.
5. Hansson, P. and Melin, S. Dislocation-based modeling of the growth of a microstructurally
short crack by single shear due to fatigue loading. Int Jnl of Fatigue, 27:347-356, 2005.
 2T5. Fatigue and fracture 253

FATIGUE EVALUATION CONSIDERING THE ENVIRONMENTAL

INFLUENCE USING A MONITORING SYSTEM

R. Cicero, I. Gorrochategui and J. A. Alvarez

University of Cantabria
E.T.S. Ingenieros de Caminos, Av/Los Castros s/n, 39005, Santander, Spain
ciceror@unican.es

The Nuclear Power Plants were designed for an initial life of 40 years, however nowadays, is under
study the possibility of its life extension. This process requires, among other issues, the analysis of
all the time limited aging calculations to the life period of the original design.
A typical case of this kind of calculations are the fatigue evaluations. In early designs fatigue
damage was not considered in the Nuclear Power Plants. Once it was introduced, the fatigue
damage evaluation was carried out based uniquely on stress considerations, being necessary to
demonstrate that the fatigue damage was lower than the unity.
Along the years, new considerations have been introduced like the variation of the material
properties with the temperature, new computer applications and programs to calculate the stresses
in components with higher accuracy, as well as the possibility of implementing monitoring
systems to register all the real transients that occur in the Nuclear Power Plants, and therefore,
allowing a more realistic evaluation of the fatigue damage. With all these new elements, it has
been being demonstrated that the fatigue damage is always lower than one even considering a life
of 60 years.
In the last years, it has been observed the influence of the environment where the components
are submerged on the fatigue damage [1] and methods have been developed to estimate this effect.
A first analysis of fatigue, considering the environmental effect [2], extrapolating the number of
cycles linearly and applying correction factors for the environmental effect, can lead to the result
that the fatigue damage in some components for 60 years of life is higher than one, and then they
are in a critical situation of fatigue failure.
For these reasons, nowadays it becomes necessary to focus the resources on the development
of new more precise evaluation methods of the environmental influence. These new methods have
to take into account the changes that occur during the transients (temperature, O2 concentration...),
the composition of the affected materials, the loading process...
One possible solution for the fatigue damage evaluation could be the implementation of a
monitoring system that registers the environmental real conditions and the type of loading in each
moment [3, 4]. In this way, it would be possible to establish an accurate approach with the
possibility of a life extension till 60 years, as for this degradation mechanism is referred.
This work illustrates this methodology through its application to a real case.

References
1. NUREG/CR-6260, Application of NUREG/CR-5999 Interim Fatigue Curves to Selected
Nuclear Power Plant Components.
2. Mehta, H.S. and Gosselin, R.S., Environmental factor approach to account for water effects in
pressure vessel and piping fatigue evaluations. Nuclear Engineering and Design 181, 1998
254 R. Cicero et al.

3. Stevens G.L., Deardorff F.A. and Gerber. A.D., Fatigue Monitoring for Demonstrating
Fatigue Design Basis Compliance.
4. D. Pando, J. A. Alvarez and I. Gorrochategui, On the use of a monitoring system for fatigue
usage calculations, Engineering Failure Analysis. 2004. Vol. 11, 765-776.
 2T5. Fatigue and fracture 255

THERMAL FATIGUE CRACK INITIATION AND PROPAGATION BEHAVIOR

OF STEELS FOR BOILER

S. Aoi, T. Marumiya, R. Ebara, T. Nishimura and Y. Tokunaga

Department of Advanced Materials Science, Kagawa University
2217-20Hayashi-cho, Takamatsu-city,Kagawa,761-0396,Japan
ebara@eng.kagawa-u.ac.jp
Miura Institute of Research & Development,
7 Horie-cho,Matsuyama-city,Ehime,799-2696, Japan
Miura Co.,Ltd., 7 Horie-cho,Matsuyama-city,Ehime,799-2696, Japan

Thermal fatigue tests were conducted for SB410 steel and SUS310S steel for boiler. A laboratory
made thermal fatigue testing apparatus was used. This apparatus consisted principally of a heating
device using oxygen and LPG gas, a temperature control device for the heated zone and a rapid
cooling device for the specimen. During thermal fatigue tests the heating and cooling cycles were
repeatedly loaded on the specimen placed on the specimen holder. City water was used as a
cooling medium. The heating temperatures were 673K, 773k and 873K,and cooling water was
sprayed on to the specimen surface through a nozzle. As it was difficult to measure the surface
temperature of the specimen, small holes for a thermocouple were prepared to measure the
temperature of the notch during the thermal fatigue tests. Thus the measured temperatures were
used as the testing temperatures. The plate specimens with an electric discharged notch at

Figure 1.Thermal fatigue test specimen

the bottom of the mechanical U notch. The thicknesses of the specimens were 24mm and
12mm.The thermal fatigue crack initiation tests and thermal fatigue crack propagation tests were
conducted up to 100 cycles for specimens with 24mm thick and 300 cycles for specimens with 12
mm thick. Figure 1 shows the figure and the size of the thermal fatigue test specimen. The thermal
fatigue crack length was measured at every ten cycles by use of a viewing microscope with
magnification of 20 after interrupting thermal fatigue tests. The thermal fatigue cracks were
examined by an optical microscope and thermal fatigue fracture surfaces were examined by a
JEOL scanning electron microscope (JSM5500S). Figure2 shows crack propagation curves for
SB410 steel with 12mm thick. The higher the testing temperature the smaller the number of heat
cycles is. The higher the testing temperature the faster the crack propagation rate is. The same
phenomenon was observed for SUS310S. The thermal fatigue crack propagation rate of SB410
steel was faster than that of SUS310S steel. Plural thermal fatigue cracks were observed on the
bottom of the electric discharged notch for SB410 steel. The crack branching was observed on the
256 S. Aoi et al.

specimen tested at 873K. Striation was observed on fracture surfaces of SB410 steel. Striation was
predominant on fracture surfaces of SUS310S. The striation spacing per cycle ,S obtained from the
measured striation spacing S versus K curve was well coincident with the da/dN K curve in the
high crack propagation rate. It can be concluded that thermal fatigue crack of boiler steels
propagate in association with striation.

Figure2. Thermal fatigue crack propagation curves of SB410 steel with 12mm thick
 2T5. Fatigue and fracture 257

RECENT DEVELOPMENTS IN FATIGUE CRACK GROWTH

R. Jones , S. Pitt1, and E. Siores2

Cooperative Research Centre for Railway Engineering, Department of Mechanical Engineering,
P.O. Box 31, Monash University, Victoria, 3800, Australia. Rhys.jones@eng.monash.edu.au
1Air Vehicles Division, Defence Science and Technology Organisation, 506 Lorimer Street,

Fishermans Bend 3207, Australia, Susan.pitt@dsto.defence.gov.au

2Centre for Materials Research and Innovation, Bolton University,Deane Road, Bolton , BL3 5AB,

England, Esiores@bolton.ac.uk

Recent research [1 - 3] has shown that the Frost Dugdale law [4] can be applied to a wide class of
engineering problems, and that for complex load spectra the crack growth rate per loading block
da/dB is essentially a linear function of a, see Figure and [1] for more details. These findings
together with the realisation that “in the threshold regime, there is something missing either in the
model”, see [5], led to the conjecture [1-3] that in Region I the crack growth rate can be expressed
in the form:

da/dN = C ( a/a*)(1-m/2) ('K) m (1)

which for small cracks integrates to give the Frost-Dugdale relationship, viz:

a = ao e EN (2)

Figure 1: Crack growth rates for A7 wing, from [1, 3].

where N is the number of cycles, E is a parameter that is geometry, material and load dependent, a
is the crack depth and, a0 is the initial size of the defect. This paper presents a large number of
examples to show that this relationship is also true for both over and underloads, see Figs 2 and 3
which present the experimental results presented by Liu and Dittmer [6] and Schijve and Broek [7]
respectively.
This methodology has been linked with NASTRAN and is used, in conjunction with a finite
element model of the F-111 wing, to predict crack growth in the 1969 F111 wing test, which was
performed in the US as part of the F-111 certification program.
258 R. Jones et al.

References
1. Barter S., Molent L., Goldsmith N. and Jones R., 2004, "An Experimental Evaluation Of
Fatigue Crack Growth", Journal Engineering Fracture Mechanics, 2004.
2. R. Jones, S. Barter, L. Molent and S. Pitt, "Crack growth at low K's and the Frost Dugdale
law", Journal Chinese Institute of Engineers, Special Issue in Honour of Professor G. C. Sih,
27, 6, 869-875, 2004.
3. L. Molent, R. Jones, S. Barter, and S. Pitt, “Recent Australian Developments In Fatigue Life
Assessment”, Submitted International Journal of Fatigue, April 2005.
4. N. E. Frost and D. S. Dugdale, “The propagation of fatigue cracks in test specimens”, Journal
Mechanics and Physics of Solids, 6, pp 92-110, 1958.
5. J.C. Newman Jr., A. Brot, C. Matias, “Crack-growth calculations in 7075-T7351 aluminum
alloy under various load spectra using an improved crack-closure model”, Engineering
Fracture Mechanics 71 (2004) 2347–2363, 2004.
6. F. Liu and D. F. Dittmer, “Effect of multi-axial loading on crack growth”, AirForce Flight
Dynamics Laboratory, Air Force Systems Command, AFFDL-TR-78-175, Volumes I – III,
September 1978.
7. J. Schijve and D. Broek, “Crack Propagation Based on a Gust Spectrum with Variable-
Amplitude Loading”, Aircraft Engineering, 34, pp. 314-316, 1962.
 2T5. Fatigue and fracture 259

CRACK CLOSURE EFFECTS IN A CRACKED CYLINDER UNDER

PRESSURE

J. Zhao, R. Liu, T. Zhang1 and X. J. Wu2

Department of Mechanical and Aerospace Engineering, Carleton University, 1125 Colonel By
Drive, Ottawa, Ontario, Canada K1S 5B6, rliu@mae.carleton.ca
1Department of Mathematics, Northeastern University, Shenyang, Liaoning, P.R. China 110004,

ztmath@163.com
2Institute for Aerospace Research, National Research Council Canada, 1200 Montreal Road,

Ottawa, Ontario, Canada K1A 0R6, Xijia.Wu@nrc-cnrc.gc.ca

The circumferential stress varies through the wall thickness when a cylinder is subjected to internal
or external pressure. Internal pressure will induce tensile circumferential stress while external
pressure compressive stress, in terms of the thick-walled cylinder theory [1]. The resultant stress
due to the synergetic contribution of internal and external pressure may be tensile at the inner
surface and compressive at the outer surface or vice versa in some cases, depending on the load
levels, which would lead to crack-face closure at the compressive edges when the cylinder contains
an axial crack. Historically, crack problems in shells were formulated in terms of either the
classical theory [2] or the transverse shear theory [3], which were all based on the linearized
shallow shell theory [4-6]. However, one deficiency of these solutions is that the crack face
interpenetration or overlap was allowed at the compressive edge when a bending load was
involved, which is physically unrealistic. In reality, crack-face closure on the compressive edge
may occur when a shell or plate containing a through-the-thickness crack is subjected to bending
load. The present research is aimed to develop a formulation for the determination of stress
intensity factor for a cylinder containing an axial crack, which incorporates the effect of the crack-
face closure.
According to the formulation developed by Delale and Erdogan [7-9], the problem of a cracked
shell subjected to membrane force and bending moment can be reduced to a pair of coupled
singular integral equations,

1GW
1  dW  1>k K,W G W   k K,W G W @dW 2SF1K,
³ ³ 11 1 12 2
1W K 1 (1a)

1Q 2 G2W 
1 1
h
4 ³
dW  ³>k21K,W G1W   k22K,W G2W @dW 2S F2 K,
O0 1 W K 1
a
(1b)

³ G W dW
i 0,
1 (i = 1, 2) (1c)

where K y / a , -a < y < a, and G1 and G2 denote the derivatives of the crack-face normal
displacement, u(0+, K), and the crack face rotation, E(0+, K), respectively,
260 J. Zhao et al.

G1 K 
w
wK
 
u 0 ,K G2 K 
w
wK
 
E 0 ,K .
and (2)
The simulation of the crack-face closure is achieved by introducing a contact force at the
compressive edge so that in the closure region the normal displacement of the crack face at the
compressive edge must remain equal or greater than zero, that is,

h

u 0  ,K r  2a

E 0  ,K t 0 
(3)
Due to the curvature effect of shell crack-face closure may not always occur on the entire
length of the crack, depending on the geometry of the shell and the nature (direction) of bending
load. The closure regions will be determined as a mixed-boundary value problem through an
iterative process such that either the normal displacement at the compressive edge is equal to zero
or the contact pressure is equal to zero. The stress intensity factor at the crack tip is due to the
synergetic contribution of membrane and bending effects, which is evaluated by Delale and
Erdogan [8],

E a ª z º
k I z   «¬ g 1 1  a g 2 1»¼
2 (4)
which shows that the stress intensity factor varies through the shell wall thickness. The results
demonstrate that due to the curvature effect the crack closure behavior in shells differs from that in
flat plates. Full-length closure always occurs in flat plates, but it is not a case in shells. Partial
closure may occur in shells, depending on the shell geometry and the bending load direction. It is
found that crack-face closure greatly reduces the bending component but increases the membrane
component of the stress intensity factor. The crack-face closure prevents the penetration of the
crack faces at the compressive edges physically when a cracked shell is subjected to bending load,
which leads to a zero value of the stress intensity factor at the closure edge.

References
1. Benham, P.P., Crawford, R.J. and Armstrong, C.G., Mechanics of Engineering Materials,
Longman Group Limited, Singapore, 1996.
2. Reissner, E., in Proceedings of the First Symposium on Naval Structural Mechanics, 1958,
74-113.
3. Sih, G.C. and Hagendorf, H.C., Plates and Shells with Cracks, Sih, Noordhoff Int Pub, 1977.
4. Reissner, E., J. Math. Phys., vol. 25, 80-85, 1946.
5. Reissner, E., J. Math. Phys., vol. 25, 279-300, 1946.
6. Nagdi, P.M., Quart. Appl. Math., vol. 14, 331-333, 1956.
7. Delale, F. and Erdogan, F., Quart. Appl. Math., vol. 37, 37:239-258, 1979.
8. Delale, F. and Erdogan, F., Int. J. Solids Struct., vol. 15, 907-926, 1979.
9. Delale. F. and Erdogan, F., Eng. Fract. Mech., vol. 18, 529-544, 1983.
 2T5. Fatigue and fracture 261

AN EXPERIMENTAL STUDY OF TEARING-FATIGUE INTERACTION

Peter Birkett, Michael Lynch and Peter Budden1

Serco Assurance, Birchwood Park, Warrington, Cheshire, WA3 6AT, UK
1British Energy Generation Ltd., Barnett Way, Barnwood, Gloucestershire, GL4 3RS, UK

peter.birkett@serco.com

Structural integrity assessments of defects in components under cyclic loading typically add the
individual contributions to crack extension from tearing and fatigue, calculated using the J-
resistance (JR) curve and fatigue crack propagation data, respectively. This approach, with the
limiting crack size determined by an instability analysis, is employed in the widely used R6 defect
assessment procedure [1]. However, it is likely that under conditions of variable or severe cyclic
loading, the fatigue crack growth rate can be retarded or enhanced by tearing occurring within the
fatigue cycles. Similarly, crack extension due to tearing can be affected by prior fatigue crack
growth. An experimental and analytical programme is in progress designed to explore tearing-
fatigue interaction, and to refine and validate R6 advice, using load-history transients typical of
those considered in nuclear reactor pressure vessel (RPV) assessments.

FIGURE 1. A typical load-crack mouth opening displacement trace for a tearing fatigue test.

In the programme, transients with tearing and fatigue components were applied to standard
fracture toughness test specimens and the resulting crack extension behaviour was compared with
predictions made using the R6 methodology. A ferritic RPV steel was used, but an austenitic steel
was also used for some tests. Baseline fatigue crack growth rate data were determined at relevant
R-ratios and temperatures. The baseline JR curve was defined at ambient temperature and 288ºC
using standard test procedures. Fifty-eight tearing-fatigue (TF) tests were performed, consisting of
a series of pre-defined fatigue stages separated by monotonic loading stages. The parameters
varied included: R-ratio (-1.1 to +0.8); temperature (ambient, 288ºC,); ‘’ and ‘’ factors (ratios of
the maximum and minimum cyclic loads to the most recent tearing load); specimen thickness
(15mm to 40mm); and the numbers of cycles and stages. Fig. 1 shows a typical load-displacement
curve for a test loaded to P1 followed by fatigue cycling between DP1 and EP1, a further tear to P2,
and so on. The tests were analysed to estimate the crack length at each stage of the test. Fig. 2
shows a plot of the estimated and measured crack growth values at the end of each test. The
majority of points fall above the 1:1 reference line, indicating that estimated values are
conservative. However, a significant number of points fall below the line and were found to relate
mainly to thinner specimens. The analyses were refined by (i) increasing the estimated tearing
crack growth incrementally, rather than assuming that tearing recommences at the start of the JR
262 P. Birkett et al.

curve, and (ii) using an enhanced fatigue crack growth law at high K values. These refinements
decreased conservatism. Sensitivity studies were also carried out in order to examine how the
estimated crack growth was affected by changes in the limit load solution, fatigue crack growth
law and JR curve. Further analysis used crack driving force

FIGURE 2. Comparison of estimated versus measured crack growth for tearing-fatigue tests.

diagrams to carry out ductile tearing analyses, from which maximum allowable crack sizes at
instability were obtained. Three alternative J-estimation schemes were used, R6 Options 1 and 2
and the GE-EPRI approach. The results indicated that: (i) the R6 guidance on tearing-fatigue
generally gave conservative estimates of tolerable crack size compared with the measured crack
sizes from the TF tests, with those based upon the GE-EPRI scheme being the least conservative;
(ii) simple addition of tearing and fatigue crack growth components did not always give
conservative estimates of crack growth, compared with the crack sizes measured; (iii) the use of
the tolerable crack size from EPRI as a limiting crack size value for the simple addition of tearing
and fatigue crack growth components was conservative for the thicker specimens, but may not be
for the thinner specimens and (iv) the simple addition results were sensitive to relatively small
changes in the definitions of the JR curve and Paris Law.

References
1. R6 – Revision 4, Assessment of the Integrity of Structures Containing Defects, British Energy
Generation Ltd, Amendment 3, 2004.
 2T5. Fatigue and fracture 263

SIF SOLUTIONS FOR CRACKS IN RAILWAY AXLES UNDER ROTATING

BENDING

S. Beretta, M. Madia, M. Schodel and U. Zerbst1

Politecnico di Milano, Dipartimento di Meccanica,
Via La Masa 34, 20156 Milano
1GKSS Research Center Geesthacht GmbH

Max Planck Straße, 21502 Geesthacht, Germany

Railway axles are designed for infinite life. However, in order to correctly manage the few failures
detected in service [1] for this safety component, there has been an increasing attention to damage
tolerance analysis and determination of inspection intervals for railway axles [2]. Among the
different input parameters of a damage tolerance analysis, a wide knowledge of SIF solutions for
railway axles is not available because of the presence of an effect induced by rotating bending
upon the SIF at the surface crack tip of a growing crack (Fig. 1.a and 1.b): K is higher in rotating
bending than in plane bending and in particular there exists an angle T0 such that
K(T0)=max{K(T)}. The only solutions available for this kind of problem are the ones by Carpinteri
et al. [3,4] who obtained combined SIF solutions obtained under plane bending at T = 0° and T =
90° for calculating SIF in smooth bars.

Figure 1. Schemes adopted for the analyses: a) scheme of the crack; b) scheme of the rotating
bending and explanation of the superposition of the effects (M’X and M’Y).

In order to analyze this problem, a series of SIF solutions for some typical notches (named as
T-notch, U-notch and S-transition) [5] of railway axles under plane and rotating bending were
carried out. In particular the effect of rotating bending upon cracks in axle body and near the
press-fittings is investigated.
Considering the three different notches, K values have been calculated for different crack
depths and shapes, taking into account different rotation angles for the axle. In particular analyses
for T = 0° and T = 90° resulted to be useful to determine the behavior of K for different rotation
angles. By this point of view, the results show that SIF of surface cracks onto axle body, under
rotating bending, can be obtained as a simple superposition of the effects of the two bending
moments (M’X and M’Y) acting on the crack (Fig. 1.b).
This simple approximation works well also for cracks near press-fittings (Fig. 2.a). It is very
interesting to compare SIF for cracks at S-transition obtained with press-fitting and with a tied
connection (Fig. 2.b). While there is no significant effect for tip A, SIF at point B with press-fitting
is approx. 25% higher than the one obtained with the tied connection: this fact seems to be due to a
264 S. Beretta et al.

different stress distribution at the notch root. It is of some importance to remark that this result
shows that the evaluation of SIF near (or at) the press-fittings has to be carried out considering the
effective load transfer and SIF solutions for axle body cannot be directly applied.

Figure 2. Results in terms of SIF at S-transition for a crack with a/D = 0.1267 and a/b = 0.736
(Snom = 107 MPa): a) superposition of the effects vs. FEM in case of press-fitting; b) comparison
of FEM results for tied-contact and press-fitting.

References
1. Smith, R. A. and Hillmansen, S., “A brief historical overview of the fatigue of railway axles”,
Proc. Instn Mech. Engrs, 218 (2004).
2. Zerbst, U. Vormwald, M. Andersch, C. Mädler, K. and Pfuff, M., “The development of a
damage tolerance concept for railway components and its demonstration for a railway axle”,
Engng Fract. Mech., 72, 209-239 (2005).
3. Carpinteri, A. and Brighenti, R., “Part-through cracks in round bars under cyclic combined
axial and bending loading”, Int. J. Fatigue, 18, 33-39 (1996).
4. Carpinteri, A. Brighenti, R. and Spagnoli, A., “Surface flaws in cylindrical shafts under rotary
bending”, Fatigue Fract. Engng Mater. Struct., 21, 1027-1035 (1998).
5. EN 13261, “Railway applications – Wheelsets and bogies – Axles – Product requirements”,
CEN, September 2003.
 2T5. Fatigue and fracture 265

MECHANICAL CHARACTERIZATION OF SINGLE CRYSTAL BARS WITH

CAPACITOR DISCHARGE WELDING AND LASER CLADDING

Samanta Chiozzi, Vito Dattoma and Francesco W. Panella

Dipartimento di Ingegneria dell’Innovazione, Università degli Studi di Lecce
Via per Arnesano, 73100 Lecce, Italy
samanta.chiozzi@unile.it, vito.dattoma@unile.it, francesco.panella@unile.it

The efficiency of gas turbine systems used for energy production and for aeronautical engines
construction can be improved through the elevated thermo-mechanical capabilities of advanced
materials employed for blades production. It is well known in fact, the importance to increase the
turbine inlet temperature; this temperature is limited by the highest working temperature of steels
and super-alloys of which blades and rotor are made. Therefore it is necessary to study the
employment of new materials more resistant to high temperatures.
Actually two different ways are being employed to reach the above results: the application of
stators and rotors cooling systems or the substitution of ordinary steels with more resistant high
temperature superalloys. Both methods are valid, but the first one is less preferred by aeronautical
industries because it needs the blade profile modification to facilitate the inside cooler circulation,
affecting the blade efficiency and increasing the total weight [1].
Among the innovative materials currently under study, multicrystal and Single Crystal Nickel-
base superalloys allow the theoretical achievement of 1300°C without damage and with the ability
to withstand high loads if compared with the ordinary materials and they are being strongly
considered for the future aeronautical engines.
Single Crystal superalloys are of particular interest since they have the advantage to avoid the
presence of microstructural discontinuities originated by the grain edges and the sites of
precipitates concentration, which in multicrystal superalloys generally represent the zone of crack
initiation and propagation [2,3].
In the present work the static behavior of Single Crystal Nickel-base superalloy CMSX-4 has
been analyzed, performing deep mechanical and microstructural characterization, necessary to
establish in detail the superalloy behaviour at the expected service stresses. Tests have been
performed both at room and working temperature (about 800°C), comparing base material, CD
welded and Laser Clad cylindrical bars.
The microstructure has been successively analyzed through investigations performed with the
optic microscope and the scanning electron microscope (SEM) observations. The micrographic
analysis concerned base material and as-welded specimens; this way the microstructural
modifications and the Heat Affected Zones have been studied as well as the interaction between
the weld material layers and parent metal. The mechanical and microstructural characterization for
“as welded” material is needed because of the requirement to develop advanced technologies for
the repair of damaged blades, particularly in the case of blisks, constituted of blading and rotor in
one piece.
The Capacitor Discharge Welding (CDW) technique is particularly suitable because it allows
the removal of the damaged part and the subsequent substitution of a new blade without the
replacement of the whole block and additional machining; it allows to achieve extremely thin
welded joints with narrow Heat Affected Zones, in order to avoid stress concentration effects at the
weld toe and to reach good material continuity [4-7]. Actually the CDW process for aeronautical
applications is under analysis, since it is not yet employed in production lines because of the lack
266 S. Chiozzi et al.

of enough experimental tests on advanced materials with the involved welding parameters. Laser
cladding is an highly flexible repair technique which is already applied for the aero engine industry
[8].
In this work the first welding trials on Single Crystal specimens have been analyzed in order to
establish the mechanical properties decay of a single crystal welded component, but also the
microstructural modifications of the parent material after the welding process in relation to the
used technological parameters had to be investigated.
Static test results reveal acceptable structural abilities of the welded joints, but it is not possible
to compare them with other data since in literature very few works about CMSX-4 can be found.
Laser clad specimens show excellent tensile strength and higher yield behavior than those obtained
with CD welded joints; the welding parameters and the heat treatments have not yet optimized and
the welded material distribution and microstructure shell improove in the next experiments.

References
1. Gell, M., Duhl, D.N. and Giamei, A.F., The development of Single Crystal superalloy turbine
blades, Pratt & Whitney Aircraft Group, Superalloys, 1980, 205-214.
2. Hemmersmeier, U. and Feller-Kniepmeier, M., Element distribution in the macro- and
microstructure of nickel base superalloy CMSX-4, Material Science and Engineering A374,
2004.
3. Wagner, A., Shollock, B.A. and McLean, M., Grain structure development in directional
solidification of nickel-base superalloys, Material Science and Engineering A256, 1998.
4. Chiozzi, S., Dattoma, V. and Panella, F.W., First results on the mechanical behaviour of
CDW welded superalloys, 5th International Conference on Fatigue and Fracture, Bari-Italy,
9-10 May 2005.
5. Casalino, G., Dattoma, V., Ludovico, A. and Panella, F.W., Numerical model for Capacitor
Discharge Welding, 13th DAAAM International Symposium, Vienna-Austria, October 2002.
6. Alley, R.L., ASM Handbook, Vol. 6: welding brazing and soldering Capacitor Discharge
Study Welding”, American Welding Society, 1991.
7. Dattoma, V. and Panella, F.W., Studio della resistenza su saldature CDW di barre in acciaio
AISI 304, XXXII Convegno AIAS, Salerno-Italy, 3-6 September 2003.
8. Richter, K.H., Orban, S. and Nowotny, S., Laser Cladding of the Titanium Alloy Ti6242 to
restore damaged blades, XXIII International Congress on Applications of Lasers and Electro-
Optics, 2004.
 2T5. Fatigue and fracture 267

FRACTAL DIMENSION ANALYSIS OF FRACTURE TOUGHNESS USED

HIGH STRENGTH CAST IRON

S. Doi and M. Yasuoka

700 Dannoharu Oita, Japan
doi@cc.oita-u.ac.jp

The rotating bending fatigue tests were done using ADI and PDI with ion carburizing. The plain
specimen with ion carburizing about 30MPa than the virgin PDI, and the fatigue strength of the
carburized ion notched specimen was given an equivalent to the virgin PDI.
Especially, the ion carburized notched specimen is effective material they have a better effect
than lower that the notched PDI. In the recapitulation, on the element parts demanded a shallow
notch, the clear distinction appeared whether ion charge or not, respectively. An important point is
that carburized PDI presented two bending S-N curve not clear in high cycle fatigue. If the
hardness is constantly holed to an internal point, the shape of S-N curve is unclearly appeared. But,
its phenomena appeared from ion boundary layer by observation of SEM photograph. The ion
Carburizing can control the charged depth by the sputtering time. Consequently, this method
warrants consideration as a non-polluting process and economical use of energy.
1 Pcn S-N curve by ion-carburized effect was controlled to affect on the crack propagating in
our researched meso-scopic structure.
2 Typical of quasi-fish eye in the surface vicinity can be decided by a/b. Where, a: equivalent
small defect radius, b: equivalent cleavage radius (yield area). Accordingly, it is expressed
by life cycle of S-N curve.
3 The plastic zone of ADI is shorter than that of PDI at the fracture. The fracture toughness
of PDI is larger than that ADI. In other words, there is not relation toughness and
elongation too much.

Fig.1 Profile and dimensions Fig.2 Fracture surface ADI,PDI

268 S. Doi and M. Yasuoka

Fig.3 Fracture toughness and plastic zone

 2T5. Fatigue and fracture 269

INVESTIGATING GAP EFFECTS IN FATIGUE LIFE OF SPOT WELDED

JOINTS

M. Zehsaz and S. Hasanifard

Faculty of Mechanical Engineering, University of Tabriz, Tabriz, Iran
zehsaz@tabrizu.ac.ir, hasanifard@tabrizu.ac.ir

The use of aluminium alloys in vehicle bodies are ever increasing especially when the main
manufacturing process of these components is spot welding.
The gap effects in 3-D finite element analysis are generally neglected. This effect in spot
welded joints is important in terms of the mechanical behaviour of them. In this research, the gap
effects on the fatigue life of the joints has been studied using strain-based approach to obtain the
fatigue life.
The numbers of spots are two in a row, joining two sheets of 5182-0 aluminium alloy. To do
so, the problem is physically modeled using ANSYS FEM based software. The three-dimensional
mesh models, the gap effects and a non-linear analysis have been used for the joints of sheets with
the gap of 0.06mm, 0.12mm, 0.18mm, as shown in Fig. 1. Because of symmetry, only one of the
hot spot-welded joints is modelled.

FIGURE 1. 3-D meshes and the boundary conditions in two different views of a hot spot weld

The elastic-plastic stresses at edges of spot welds were considered and Manson-Coffin formula
has been applied to obtain fatigue life.
The comparison between the experimental results and numerical approach shows that the
existence of gap between sheets increases the amount of the stresses near the roots of nuggets.
Therefore, the fatigue life of spot welded sheets decreases with increasing the gap distance.
This is illustrated in Fig. 2.
270 M. Zehsaz and S. Hasanifard

FIGURE 2. L-N curve for different sheet gaps

References
1. Ni, K., Mahadevan, S., Strain-based probabilistic fatigue life prediction of spot-welded joints,
International Journal of Fatigue 26(2004) 763-772.
2. Adib, H., Gilgert, J., Pluvinage, G., Fatigue life duration prediction for welded spots by
volumetric method, International Journal of Fatigue 26(2004) 81-94.
3. Radaj, D., Sonsino, C.M., Fatigue Assessment of Welded Joints by Local Approaches,
woodhead publishing Ltd.1998.
4. SHARP, M.L., NORDMARK, G.E., MENZEMER, C.C., Fatigue design of aluminium
components & structures, McGraw-Hill, 1996.
5. GEAN, A., WESTGATE, S.A., KUSZA, J.C., EHRSTROM, J.C., Static and Fatigue
Behaviour of Spot-Welded 5182-0 Aluminium Alloy Sheet, 1999.
 2T5. Fatigue and fracture 271

FATIGUE OF PMMA BONE CEMENT

S. L. Evans
School of Engineering, Cardiff University
The Parade, Cardiff CF24 3AA, UK
EvansSL6@cardiff.ac.uk

The use of PMMA bone cement is the most successful way of fixing implants such as hip and knee
replacements, and has many other uses in orthopaedic surgery. However, in the long term
mechanical failure of the cement is common and may lead to clinical failure. Despite a large
volume of published research, the fatigue of bone cements is not well understood and much more
research is needed before fatigue failure can be reliably predicted and avoided.
Most of the published studies of cement fatigue have used simple S-N tests on unnotched
specimens. The presence of stress concentrations that act as sites for crack initiation has a critical
effect on the fatigue life and so many of these studies have identified porosity as an important
factor. However, it is not clear whether porosity is as important in the clinical setting where there
are many other stress concentrations and perhaps cracks due to curing shrinkage. Here crack
propagation may be more important, and there are many complications such as variable amplitude
loading, slow crack growth and microstructural effects.
PMMA has been widely used as a model for studies of fatigue and fracture in polymers, but
there is still a limited understanding of the effects of factors such as variable amplitude loading.
Bone cements have a more complicated structure, with previously polymerised beads in a softer
matrix which cures on implantation, and other components such as particles of barium sulphate or
zirconia to make the cement visible in radiographs. This microstructural complexity means that
the cement may behave very differently from pure PMMA.
In smooth specimens, cracks initiate from pores, at the stress concentrations formed between
previously polymerised beads, and there is a correlation between porosity and fatigue life. The
stress concentration Kt at a spherical pore depends only on the Poisson’s ratio and for cement it is
about 2.06. The size of the pores does not affect Kt, unless the pores are large enough to
significantly reduce the cross- section. This reduction in cross- sectional area can account for the
observed reduction in fatigue life for specimens with greater porosity. Arguably the effect of
porosity is therefore an artefact of the specimen design, rather than a change in the properties of the
material. Apart from this effect, porosity may only be important when it increases the likelihood of
interaction between pores, which can cause higher stress concentrations. There is evidence for
crack initiation due to shrinkage stresses developed during curing, although this does not always
occur. In the clinical setting, this may be more relevant than initiation from pores by fatigue
processes.
Crack propagation has not been widely studied, and there are several complications such as
slow crack growth and microstructural differences in the crack path. Physiological loading varies
widely in amplitude and frequency content and the effects of these variables have not been fully
investigated. Cracks typically propagate through the beads, but in some cases it is found that the
beads pull out of the matrix. It is not clear why this occurs. There does not appear to be poor
adhesion between the beads and the matrix, and it may be that the presence of residual compressive
stresses in the beads due to curing shrinkage of the surrounding matrix causes the crack to deviate.
Since the residual stresses relax over periods of weeks or months, this may explain some of the
inconsistencies between different studies, and the change at different 'K levels that has been
observed. The crack deviation that is caused by this phenomenon leads to considerable
272 S. L. Evans

disturbance of the crack front, with some shielding by intact ligaments that bridge the crack. This
may account for much of the increased resistance to crack propagation in cements as compared to
pure PMMA. Even when the crack propagates through the beads, the fracture surface is much
rougher than in pure PMMA.
Clinical failure of the cement occurs over long time periods, and this implies that the crack
growth rate is very low, perhaps as low as 10-12 m/cycle. Many studies have assumed a threshold
at a much higher growth rate, but when load shedding is correctly carried out it appears that there
may not be a threshold and the Paris Law is obeyed to 10-11 m/cycle or below. However, crack
arrest can occur at higher 'K, perhaps for microstructural reasons.
Preliminary measurements of the effects of variable amplitude loading raise more questions
than they answer. In pure PMMA, considerable retardation occurs, particularly at lower 'K. 30%
overloads every 100 cycles resulted in crack growth rates as much as two orders of magnitude
lower than constant amplitude loading. At higher 'K the effect was much less pronounced. In
cement, however, little or no retardation occurred as a result of individual overloads. Single
overloads applied at a crack growth rate of around 10-9 m/cycle produced an acceleration of
subsequent crack growth by up to a factor of five, and in some cases the acceleration persisted for
25,000 cycles or more. Single overloads at this level produced no retardation whatsoever. The
mechanisms by which these effects occur are unknown at present, but it is likely that plasticity
induced crack closure may play a role. Other mechanisms such as extended crazing ahead of the
crack tip or breakage of ligaments that bridge the crack may also be relevant; in pure PMMA in
particular there can be extensive bridging at the sides by shear- yielded material in the plane stress
region and this may be affected by overloading.
Despite extensive research the fatigue behaviour of bone cements is not well understood. The
effect of porosity has been highlighted in many studies, but it is not clear how relevant it is to
clinical cement failure. Crack propagation behaviour is very different to that of pure PMMA due
to the complex microstructure of cements. The effects of slow crack growth under constant load
may be important, and there is evidence that variable amplitude loading may produce dramatic
effects including both acceleration and retardation of subsequent crack growth. As yet the
mechanisms involved are unknown and much further research is needed before it is possible to
predict fatigue failure with any accuracy.
 2T5. Fatigue and fracture 273

INFLUENCE ON THERMAL BARRIER COATING DELAMINATION

BEHAVIOUR OF EDGE GEOMETRY

H. Brodin, X. H. Li1 and S. Sjoestroem1

Department of Mechanical Engieering, Linköping University,
SE-581 83 LINKÖPING, Sweden
1SIEMENS Industrial Turbomachinery AB, SE-612 83 FINSPÅNG, Sweden

hakan.brodin@siemens.com, soren.sjostrom@siemens.com

Ceramic thermal barrier coatings are commonly used in gas turbine hot components (e.g.,
combustor liners/buckets and guide vane platforms). In components that are only partially coated
or have cooling-air outlets, coating-end stress singularities may lead to the spallation of the
coating.
Depending on the geometry of the transition from coated to uncoated material, the severity of
the stress singularity will vary. Basic references for the analysis of such stress singularities are, for
instance, Bogy [1], Bogy [2], Bogy [3], and Dundurs [4], where it is shown that the severity of the
stress singularity depends on the chamfer angle \ and that making this angle \ > 0 decreases the
singularity order at the coating end, see Fig. 1.

FIGURE 1. Schematic drawing of a TBC system. A chamfer angle \ is defined in the figure.

In the present study, a thin thermal barrier coating system has been studied. Bond- and top
coats have been sprayed to a thickness of 150µm and 350µm, respectively. Vacuum-plasma-
spraying technology was used, and the test specimens were rectangular (30x50x5mm) coupons of
a nickel-based superalloy, Haynes 230. A NiCrAlSiY bond coat and an Y2O3 stabilised ZrO2 top
coat were used.
In order to achieve well-defined chamfers, sprayed components were ground on the edges with
SiC grinding paper to desired geometry. By inspections of cross-sections that had not undergone
thermal fatigue cycling, it was ensured that no damage was introduced into the system.
Mechanical testing was done in a thermal cyclic test rig where specimens are heated in a
furnace and cooled with compressed air. The thermal cycle data were:
Tmin=100°C, Tmax=1100°C, dwell time at max temperature Tdwell=40 min and cycle time
Tcyc=60 min.
FE modelling of the system has been done aiming at supporting the findings from thermal
fatigue tests. A parametric study including variation of the chamfer angle \has been done and the
stress state near the chamfer evaluated.
274 H. Brodin et al.

Evaluation of fatigue damage can be done visually for observation of coating failure
(macroscopic observation on coating surface). 20% area with complete spallation was considered
as thermal barrier coating failure. For evaluation of damage development, additional light
microscopy investigations of cross-sections have been done.
Results show that the fatigue life benefits from introduction of a chamfer angle at the coating
end during thermal fatigue cycling. This is seen in Fig. 2, where fatigue life is plotted against
chamfer angle \

FIGURE 2. Influence on fatigue life on chamfer angle \.

Reference
1. Bogy D.B., J. Appl. Mech., vol. 35, 460-466, 1968
2. Bogy D.B., Int. J. Solids Struct., vol. 6, 1287-1313, 1970
3. Bogy D.B., J. Appl. Mech., vol. 38, 377-386, 1971
4. Dundurs J., J. Appl. Mech., vol. 36, 650-652, 1969
 2T5. Fatigue and fracture 275

LOW CYCLE FATIGUE AND FRACTURE OF A COATED SUPERALLOY

CMSX-4

S. Stekovic
Division of Engineering Materials, Linköping University
58183 Linköping, Sweden
svjst@ikp.liu.se

The single crystal CMSX-4 is the second-generation rhenium containing nickel-base superalloy
developed by Cannon-Muskegon Corporation, Davis [1]. The alloy is widely used because of its
good mechanical properties such as long-time strength and toughness at high temperatures, Simms
et al. [2]. Alone, the alloy has limited oxidation and corrosion resistance and to solve this problem
it is protected with coatings.
The objective of this study is to examine and establish fatigue and fracture behaviour under
low cycle fatigue test conditions of the coated single crystal superalloy CMSX-4. For this purpose
three different coatings have been chosen, an overlay coating AMDRY997, an aluminide diffusion
coating RT22 and an innovative coating called IC1. Cylindrical solid specimens were cyclically
deformed with fully reversed tension-compression loading with total strain amplitude control at
two temperatures, 500oC and 900oC. The tests were also done at the uncoated specimens taken
from the same batch. The empirical relationship between plastic strain amplitude and the number
of reversals to fatigue failure was determined by Coffin-Manson, Klesnil and Lukas [3], se eq. (1).
The results will be given in the full paper.

'H p
H '(2 N f )c
2 (1)

At 500oC the coatings had detrimental effects on the fatigue life of CMSX-4 relative to the
uncoated specimens while at 900oC the coated specimens exhibited longer life than the uncoated
specimens at the same temperature. AMDRY997 gave the longest life compared to RT22 and IC1.
The coated superalloy exhibited hardening and higher stress level at higher applied strains and
lower temperature. At 900oC softening occurred together with lower stress response level. The
coatings lowered the stress response level from about 12% to 31% compared to the uncoated
specimens under the same test conditions. Most of the observed cracks initiated at the coating
surface, one example is shown in Fig. 1. Majority of the cracks was arrested at the interface
between the coating and the superalloy. Surface roughness or rumpling was found in AMDRY997
with some cracks initiated from the rumples.

FIGURE 1. SEM micrograph showing surface crack initiation observed at the surface of
AMDRY997 at 500oC
276 S. Stekovic

The fatigue behaviour of CMSX-4 can be affected by precipitation of topologically close

packed (TCP) phases such as round shaped P phase and acicular shaped V phase by two damage
mechanisms, embrittlement damage mechanism related to the brittle nature of those phases and
softening of J matrix by depletion of strengthening elements to form TCP phases, Simonetti and
Caron [4]. The phases were observed under and in the interdiffusion zone, although no
microcracks initiated from them were found.
The cracks found in the coatings grew more or less perpendicular to the load axis. The initial
crack path is flat along the plane with high tensile stress, normal to the stress axis. The cracks
began to fluctuate then with visible fatigue striations.

References
1. Davis, J.R., Heat-resistant materials, ASM specialty handbook, Materials Park, ASM
International, The USA, 1997.
2. Simms, C.T., Stoloff, N.F. and Hagel, W.C., Superalloys, Wiley, New York, 1987.
3. Klesnil M. and Lukas P., Fatigue of Metallic Materials, Elsevier, Amsterdam, The
Netherlands, 1992.
4. Simonetti, M. and Caron, P., Materials Science and Engineering A, vol. A254, 1-12, 1998.
 2T5. Fatigue and fracture 277

THERMOMECHANICAL FATIGUE OF OPEN-CELL ALUMINIUM SPONGE

T. Guillen, A. Ohrndorf, U. Krupp, H. J. Christ, S. Derimay1, J Hohe2 and W Becker1

Institut für Werkstofftechnik, Universität Siegen
Paul-Bonatz-Str. 9-11, 57068 Siegen, Germany
Phone +49 271 2184
Fax +49 271 2545
ohrndorf@ifwt.mb.uni-siegen.de
1Institut für Mechanik, TU Darmstadt
2Fraunhofer-Institut für Werkstoffmechanik, Freiburg

While closed-cell metal foams have been intensively studied during the last ten years, only little
attention has been put to the group of open-cell metal sponges. Due to their low density and good
homogeneity metal sponges are promising candidates for structural and at the same time functional
applications. For the successful design of new components using metal sponges a sound
understanding of the behaviour under monotonic and cyclic loading conditions is required. The
aim of the present study is to investigate the behaviour of the open-cell aluminium foam under this
kind of loading conditions placing special attention to thermomechanical fatigue and creep
characteristics.
The studied metal sponge was precision-cast AlSi7Mg obtained by m-pore GmbH, Dresden,
having a porosity of 8 ppi and a relative density of 5.1%. Figure 1 shows the macroscopic cell
structure of the studied material which has similar mechanical properties than the PORMET
material (AlSi9Cu sponge) that was studied in an earlier work [1].

FIGURE 1. Macrograph of the AlSi7Mg sponge.

Rectangular specimens (25mmx25mmx70mm) were glued to a special specimen mount and

subjected to monotonic loading and fatigue at room temperature. The tests were carried out by
means of a servohydraulic testing machine (MTS 810).
Since an induction heating system could not be used to heat the inhomogeneous sponge
specimens, thermomechanical fatigue tests were carried out in a self-designed temperature
chamber, where the specimens can be heated by a hot air fan and cooled by compressed air (see
Fig. 2). For this tests the specimen’s ends were cast-infiltrated with Zn serving as specimen mounts
in order to a allow the performance of tests at temperatures higher than 200°C.
278 T. Guillen et al.

Results of monotonic tests revealed that the plateau stress in compression lies in a range of
0,7- 0,75 MPa.
With respect to the tensile behaviour it was observed that the amount of plastic deformation at
fracture is strongly influenced by the ductility of the cell strut material. This could be confirmed
within the present study resulting in strain of fracture levels of 1,5%.

FIGURE 2. Temperature chamber for the realisation of TMF tests on open-cell aluminium sponge.

Thermomechanical fatigue behaviour of the aluminium sponge has been compared to that
under isothermal fatigue of room temperature and at elevated temperatures. The fatigue damage
mechanisms for each loading condition was analysed by optical and scanning electron microscopy.
The experimental results were compared with a micromechanical model that is based on the
idealization of the Kelvin model foam and subsequent examination of the effective mechanical
properties to be used within an energetic homogenisation scheme [2].

References
1. A. Ohrndorf, U. Krupp, H.-J. Christ, Proc,. Fatigue 2002, A.F. Blohm (ed) pp.3093-3098,
2. J. Hohe, W. Becker , Int. J. Mech. Sci., 45, 2003, 891-913.
 2T5. Fatigue and fracture 279

THE INFLUENCE OF ALTERNATE BLOCK LOADING ON THE FATIGUE

LIFETIME

M. Kohut and T. Lagoda

Technical University of Opole
ul. Mikoajczyka 5, 45-271 Opole, Poland
tlag@po.opole.pl

The paper presents the results of fatigue tests of cylinder specimens made of duralumin PA6 under
alternate block loading for bending and torsion. Each block contains n and n cycles with sinusoidal
course. Based on the fatigue curves amplitudes suitable for 105, 3˜105 and 106 number of cycles to
failure were determined. The loading was applied in alternate two-amplitudes blocks by 104, 3˜104
or 105 each one (10% of the fatigue life for the given loading level) until failure of the specimen
(fig.1).

Fig.1. Part of the stress course

The obtained experimental results were described with the criterion of normal and shear
energy density parameter in the critical plane [1]

W eq (t) ȕW Șs (t)  țW Ș (t)

(1)

where ȕ, ț – constant dependent on S-N description for bending and torsion of

suitable material,

1
W Șs
2
IJ Șs (t)İ Șs >
(t)sgn IJ Șs (t), İ Șs (t) @
(2)

1
WȘ
2
>
ı Ș (t)İ Ș (t)sgn ı Ș (t), İ Ș (t) @
(3)
The critical plane was determined with the fatigue damage accumulation method and the
criterion of shear strain energy density parameter.
The fatigue damages were accumulated in the critical plane according to the Palmgren-Miner
hypothesis for strain energy density parameter
280 M. Kohut and T. Lagoda

§ ·
¨ ¸
j ¨ 1 ¸
S (N block ) ¦ ¨ m ' ¸
i 1 ¨ § W af · ¸
¨ N o ¨¨ W ¸¸ ¸
© © ai ¹ ¹ for W ai t aW af
(2)
where Wai are the adequate amplitudes of strain energy density parameter
according to (1).
After determination of the damage degree, the fatigue life (number of blocks) was calculated
using the equation
N block
N cal
S ( N block ) (3)
where N block is a single block described by equation

N blok n IJ  nı
(4)
The obtained calculation results are similar to the experimental results (fig.2)

Fig.2. Comparison of the calculated and experimental results for duralumin AlCu4Mg

The most results are included in scatter bands with the factor of three.

REFERENCES
1. AGODA T., MACHA E., BDKOWSKI W.: A critical plane approach based on energy
concepts: Application to biaxial random tension-compression high-cycle fatigue regime.
Int.J.Fatigue, 1999, Vol.21, pp.431-443.
 2T5. Fatigue and fracture 281

FATIGUE DESIGN AND INSPECTION PLANNING OF WELDED JOINTS

BASED ON REFINED PHYSICAL MODELLING

T. Lassen and N. Recho

Agder University College, Faculty of Engineering, Grimstad, Norway
LaMI, Université Blaise Pascal - Clermont II – France
Tom.Lassen@hia.no

The fatigue behaviour of welded joints subjected to typical in-service stress levels is far more
complex than conventional bi-linear S-N curves and pure fracture mechanics models predict.
These models are based on tests carried out in accelerated laboratory conditions. When the stress
range approaches the fatigue limit, the crack initiation period becomes a very important part of
fatigue life. Based on this fact, a Two Phase Model (TPM) for the fatigue process is developed and
calibrated. The model includes both the crack initiation and the crack propagation phase. The
physical model is suggested as complementary tool to statistical based S-N curves. These curves
are only valid for a given joint geometry and loading conditions. It is dangerous to extrapolate
them outside the range of the data. The large amount of data pertain to accelerated laboratory
condition whereas in service stresses usually are close to the fatigue limit stress region where very
few tests actually are carried out. At this stress regime a reliable physical model is useful both for
fatigue life predictions and to determine the damage process. The latter gives necessary
information to scheduled inspection planning; we must know when and what to look for. The S-N
curves which do not work with the notion of a crack cannot be of any help on this matter.
Furthermore, the present model provides physical insight that statistical methods do not provide.
The practical consequences of the model when it comes to design and inspection planning are
illustrated and discussed.
It is assumed that the total fatigue life consists of two major phases: N=NI + NP.
The number of cycles to crack initiation NI is modeled by a local strain approach using the Manson
Coffin equation, whereas the propagation phase is modeled by fracture mechanics adopting the
simple version of the Paris law. The model is calibrated to fit the fatigue behaviour of fillet welded
joints where cracks emanate from the weld toe, Fig 1. The model is fitted to both experimental
crack growth histories at high stress ranges and to fatigue lives at any stress level. Emphasis is laid
on how to determine the variable weld toe notch factor, the transition crack depth between the two
phases and the material parameters. The model is valid for high quality joints where a thorough
post fabrication inspection is carried out. The model is capable of taking into account the effect of
the global geometry of the joint, the local weld toe geometry, applied stress ratio and the residual
stress condition. The S-N curve obtained from the model coincides with the F-class S-N curve
(British Standard 5400) at high stress levels, but becomes non-linear as the stress range approaches
the fatigue limit. In this stress region the present model fits the experimental S-N data far better
than the conventional bi-linear S-N curve.
282 T. Lassen and N. Recho

FIGURE1 – Typical joint configuration and definition of crack shape

We have constructed an S-N curve which is nonlinear for a log-log scale and that predicts
substantially longer lives at stress ranges below 100 MPa (14.5 ksi) than does the F-class linear
curve. Furthermore, at these long fatigue lives (more than 5×106 cycles) the initiation life is at least
70% of the entire fatigue life. We will show by a simple example the practical consequences of
these results with respect to:
• Predicting fatigue life for selecting dimensions of a joint
• Predicting crack growth evolution for inspection planning
We compare our results with the results obtained by the median F-class curve and a pure
Fracture Mechanics Model (FMM) calibrated to make life predictions made by F-class and by the
FMM coincide.
The fatigue behaviour of high quality welded joints is far more complex than conventional bi-
linear S-N curves and pure fracture mechanics model predict.
At typically low in-service stress ranges the crack initiation period, defined as time to reach a
crack depth of 0.1 mm, represents the dominating part of the fatigue life. To describe this
behaviour a TPM is required.
The non-linear S-N curves obtained from the TPM coincides with the conventional S-N curve
predictions at high stress levels, but fits the experimental data near the fatigue limit far better. The
abrupt knee-point of the bi-linear curve does not fit the facts in this stress region.
The theoretical consequence of the model is that the fatigue limit is explained by an extremely
long initiation period and not, as for the fracture mechanics model, by a threshold value in the
stress intensity factor. The shallow surface breaking crack near the weld toe do not obey the
threshold law, they tend to grow considerably faster.
The first practical consequence of the model is that it predicts considerably longer fatigue lives
at low stress levels than the conventional bi-linear curves in rules and regulations. This will results
in higher permissible stresses and smaller dimensions of the joints. The second practical
consequence is that a scheduled in-service inspection program should be more progressive with
service time compared with programs based on a pure fracture mechanics model. This will result in
a long initial inspection interval and short intervals at the end of service life.
 2T5. Fatigue and fracture 283

A MIXED MODE FATIGUE CRACK GROWTH MODEL INCLUDING THE

RESIDUAL STRESS EFFECT DUE TO WELD

S. Ma, X. B. Zhang, N. Recho and J. Li1

Blaise Pascal University of Clermont II, LaMI, France
1University of Paris XIII, Institute Galilée, LPMTM, France

zang@moniut.univ-bpclermont.fr

When a crack is subjected to cyclic loading, the traditional studies of fatigue crack were
concentrated on the mode I crack growth mechanism. A number of crack propagation laws were
developed to evaluate the fatigue crack growth rate. However the effect of the loading angle on the
propagation was not considered in these laws. In the other hand, when a crack exists in a metallic
welded structure, the residual stress due to the weld may influence the crack growth.
In this work, the experiments of a fatigue crack under mixed-mode loading are performed with
CTS (Compact Tension Shear) specimen associated to a mixed mode loading device. The effect of
loading angle on the crack growth rate and on crack bifurcation angle is analyzed. Also, the welded
specimens are introduced in the experiments in order to investigate the influence of the filled weld.
Furthermore, on the basis of the experimental results, a crack propagation model is proposed in
order to evaluate numerically a fatigue crack growth rate, in which the effects of the loading mode
and of the residual stresses due to weld are considered.
Two different types of CTS specimens are used in the work, i.e. the non welded specimen and
the welded specimen. In the welded specimen, the initial crack is parallel to the weld. Each type of
specimens is made from two materials, aluminum alloy 7020 and steel S460. The fatigue tests are
performed at room temperature. The CTS specimens are tested with two loading levels and three
loading angles with respect to the crack axis. For each specimen, the relationship between the
crack length a and the number of cycles N is obtained by marking method during the test. The
stress intensity factor K is calculated by FEM. For a mixed-mode loading, the equivalent stress
intensity factor Keq is used, which is the combination of KI (stress intensity factor in mode I) and
KII (stress intensity factor in mode II). The crack growth rate da/dN is issued from the values of N
and a. And then da/dN is presented as function of the equivalent stress intensity factor Keq.
Effect of the loading angle on the crack propagation:
The experimental results obtained on aluminum specimens and steel specimens show the same
tendency. It can be found that for the same initial value of Keq, the crack grows faster in the case
of 30° loading than in the other loading cases. This is for the reason that to obtain the same Keq,
the loading level increases as the loading mode tends to pure mode II. Therefore, because of the
lowest load, the crack growth rate is the lowest when the crack is subjected to pure mode I loading
whatever the material is, whatever with or without weld.
Effect of the weld on the crack propagation:
The welded and non welded specimens are subjected to the same loading angle and to the same
loading level. The experimental results show that for the same loading level, the crack growth rate
is greater in non welded specimen than that in welded specimen. It means that the weld introduces
compressive residual stresses near the weld toe. These residual stresses decrease the crack growth
rate. When the specimens are subjected to pure mode I loading condition, this influence is more
obvious than in the case of 60° and 30° loading because of the difference of the crack growth path.
When loading angle is 60° and 30°, the crack grows obviously far from the weld. Therefore, there
is only a little influence on the crack propagation.
284 S. Ma et al.

The same effect of weld is obtained from the steel specimen. Because the steel specimens are
less thick than the aluminum ones, so the influence of the welded residual stresses is less important
than that of in aluminum specimens.
Numerical Model
According to the experimental results, we have proposed a numerical model to study the crack
growth. In this model, two parameters are considered: the mixity parameter and the residual stress
due to weld. The Paris’ law is used with the equivalent stress intensity factor Keq.
Effect of loading mode on the crack growth rate:
The experimental results show that the exponent m which is the slope of the Paris’ law keeps
constant when the crack grows under mixed mode loading. But the coefficient C changes as
function of loading mode. Therefore, the influence of mixed mode on the crack growth rate can be
introduced in the coefficient C. So we use a coefficient C* which is function of the mixity
parameter Mp and an experimental coefficient which is equal to 3 for all the CTS specimens in
this work. Mp is equal to elastic mixity parameter in this work because of the low stress level and
can be written as the ratio between the stress intensity factors of mode I and mode II. In this model,
the constants m and C may be measured in a pure mode I fatigue test. KI and KII are calculated by
FEM. So Keq and Mp can be determined from the values of KI and KII.
Effect of the welded residual stresses on the crack growth rate:
The same experimental observation is obtained. Only the coefficient C changes as function of
the residual stress level. So the coefficient C becomes CR which is written as function of different
stress intensity factors and which is also equal to 3 for all the CTS specimens in this work.
The verification of the proposed model on steel specimens has showed that the numerical
results are in good agreement with the experimental results. More studies must be carried out in
future to verify this model with different materials and different types of specimens.

Conclusions
From this work, we can draw out the following conclusions:
1 The crack growth rate is related to the loading mode. For the same initial Keq, the closer to
pure mode II, the faster the crack growth rate is.
2 When a crack is parallel to the filled weld in the case of pure mode I loading, for the same
loading level, the compressive residual stresses due to the weld decrease the crack growth
rate. In the case of mixed mode loading condition, this effect is weak as the crack grows far
from the weld.
3 The filled weld has low influence on the crack growth direction.
4 A numerical model is proposed in order to consider the effects of the loading mode and of
the residual stress on the propagation and bifurcation of a crack under cyclic loading.
 2T5. Fatigue and fracture 285

EFFECTS OF SHOT PEENING ON FATIGUE PROPERTY IN SICP/AL-MMC

Yasuo Ochi, Kiyotaka Masaki , Takashi Matsumura and Tatsuhiko Hamaguchi1

Department of mechanical Engineering & Intelligent Systems,
University of Electro- Communications, Tokyo, Chofu, Tokyo 182-8585, Japan
ochi@mce.uec.ac.jp
1Aluminum Development Division, Aisin Takaoka Co., Toyoda, Aichi, 473-0934, Japan

SiC particle reinforced aluminum alloy matrix composite (SiCp/Al-MMC) has been expected for
applications to components of aircrafts and automobiles because of high wear resistance, high
temperature strength and light weight characteristics. Most of the components are subjected to
fatigue loading. However, there have been little study for fatigue property of SiCp/Al-MMC. And
also, it has been well known that shot peening is one of useful treatments for improvement of
fatigue strength of machine components.
In this study, two kinds of materials ; Al-Si-Mg cast aluminum alloy (JIS AC4CH) and SiC
particle reinforced cast aluminum alloy (SiCp/AC4CH-MMC), were prepared. And also shot
peening treatments were performed on two materials. Figure 1 shows microstructures of two
materials, (a) is AC4CH and (b) is SiCp/AC4CH-MMC. In AC4CH, it shows typical dendrite
structures with aluminum matrix and eutectoid silicon particles. In MMC, the volume fraction of
SiC particle is about 10%, the average size of SiC particle is 9~14Pm, and SiC particles are

(a)AC4CH (b) SiCp/AC4CH-MMC

FIGURE 1. Microstructures.

TABLE 1. Mechanical properties.

distributed on boundaries of primary D(Al) crystals. Table 1 shows mechanical property of two
materials. There were no clear difference of tensile strength and Young’s Modulus between two
materials, but reduction of area of AC4CH is larger than that of SiCp/AC4CH-MMC. The shot
peening were treated with steel shot (HRC53) with 0.6mm in diameter. The almen intensity was
0.135mmA and the coverage was 300%.
286 Y. Ochi et al.

High cycle fatigue tests of rotating bending loading were carried out with frequency of
2760rpm in air at room temperature. Effects of SiC particle reinforcement and shot peening
treatment on fatigue properties were investigated. Observations of surface crack behaviors were
using replication technique. And fracture surface were observed after fatigue tests by a field-
emission type scanning electron microscope (FE-SEM) in order to specify the crack initiation sites.
Figure 2 shows S-N diagrams of all materials. Comparing with results of AC4CH and SiCp/
AC4CH-MMC without shot peening, the fatigue strength at 107 cycles of AC4CH is 100MPa and
that of MMC is 110MPa. So, the improvement of fatigue strength by SiC particle reinforcement
was small, and also, the fatigue lives before 106 cycles of MMC were less than those of AC4CH.
As investigating the effects of shot peening treatment on the fatigue strength, the fatigue strength
of entire regime of both materials were almost larger than those of non-peening materials, but the
scatter of the fatigue life of peening materials were larger than those of the non-peening materials.
And then, the fatigue strength at 107 cycles of AC4CH-SP was about 130MPa. On the other hands,
the strength of SiCp/AC4CH-MMC-SP was about 110MPa. And also, from these results, it was
shown that the fatigue lives before 107 cycles in SiCp/AC4CH-MMC were improved by shot
peening treatment in spite of some scatter, but the effects for the fatigue strength at 107 cycles was
not clear. From the observation of fracture surface, the crack initiation sites of all materials were
casting defects, and there are no clear difference in these materials.

FIGURE 2. S-N diagrams.

 2T5. Fatigue and fracture 287

FATIGUE BEHAVIOUR OF FRICTION STIR WELDED 6061-T6 ALUMINIUM

ALLOY

Yoshihiko Uematsu, Keiro Tokaji, Yasunari Tozaki1 and Hideaki Shibata1

Gifu University, 1-1 Yanagido, Gifu City 501-1193, Japan,
1Gifu Prefectural Research Institute, 1288 Oze, Seki, Gifu 501-3265, Japan

yuematsu@cc.gifu-u.ac.jp

Friction stir welding (FSW) is a recently developed solid state welding method, and now being
used increasingly for joining aluminium alloys. However, the fatigue behaviour of FSW joints with
relatively complicated microstructure is still unclear. In this study, fatigue behaviour of FSW joints
of 6061-T6 aluminium alloy was investigated. The 6061-T6 plates were joined with welding
speeds of 100 and 200 mm/min and the rotation speeds of tool of 1200 and 1800 rpm. Fatigue tests
were conducted at stress ratio R = -1 under axial loading.
The microstructure of the weld zone is usually classified into three regions, namely stir zone
(SZ), thermo-mechanically affected zone (TMAZ) and heat affected zone (HAZ), Lomolino et al.
[1]. Fig. 1 shows the typical microstructure of the longitudinal section around the pin of the FSW
tool, where the microstructure is asymmetric between the top and bottom sides. In the SZ, fine
equiaxed grains can be seen, resulting from dynamic recrystallization. It is clear that aluminium
grains are severely deformed in the TMAZ that is the microstructural transition zone between SZ
and HAZ. Vickers hardness profiles are shown in Fig. 2, revealing softening inside the weld zone.
The low hardness plateau extends in the fine grained area, SZ, and hardness minima are located at
the TMAZ. The softening can be attributed to the dissolution of precipitates due to the elevated
temperature during FSW process, Su et al. [2]. It was found that the microstructures and hardness
profiles were hardly affected by the welding condition.

FIGURE 1. Optical micrograph of longitudinal section in weld zone (1200rpm-200mm/min).

The tensile strength, B, of the parent material was 308 MPa, while that of FSW joints was
around 200 MPa regardless of the welding condition. All FSW joints fractured at the softened
TMAZ.
Fig. 3 shows the S-N diagram of FSW joints. The fatigue strength of FSW joints is lower than
that of the parent material, especially in the high stress region. The welding condition has almost
no influence on the fatigue strength as well as microstructure, hardness profile and tensile strength.
It should be noted that the location of fatigue fracture is dependent on stress level as shown in Fig.
3, where the solid and open symbols indicate that the fracture at TMAZ and HAZ, respectively. In
the high stress region, fracture occurred at the TMAZ, while in the low stress region, at the HAZ.
Macroscopic observation using an optical microscope revealed that localized plastic deformation
took place at the TMAZ due to the softening and asymmetric microstructure when high stress was
288 Y. Uematsu et al.

applied. The stress concentration at the locally deformed TMAZ was responsible for the fatigue
fracture at the TMAZ and the much lower fatigue strength than the parent material in the high
stress region. On the other hand, such deformation at the TMAZ was not recognized when the
applied stress was relatively low and fracture occurred at the HAZ. It implies that if the stress
concentration at the TMAZ in not introduced, then the FSW joints of 6061-T6 aluminium alloy
would fracture at the HAZ under fatigue condition. From the hardness measurements before and
after fatigue tests, it is concluded that the fatigue fracture at the HAZ in FSW joints was attributed
to both the grain refinement at the SZ and the dynamic aging at the SZ and TMAZ by cyclic
loading.

FIGURE 2. Hardness profiles in friction stir welded joints.

FIGURE 3. S-N diagram.

References
1. Lomolino, S., Tovo, R. and dos Santos, J., Int. J. of Fatigue, vol. 27, 305-316, 2005
2. Su, J.-Q., Nelson, T.W., Mishra, R. and Mahoney, M., Acta Mat., vol. 51, 713-729, 2003
 2T5. Fatigue and fracture 289

TRANSFORMATION OF A NONPROPORTIONAL MULTIAXIAL LOADING TO

AN EQUIVALENT PROPORTIONAL MULTIAXIAL LOADING

A. Chamat, Z. Azari1, M. Abbadi and F. Cocheteux2

Laboratoire de Fiabilité Mécanique-Ecole National d’Ingénieur de Metz, Université de Metz, Ile
du Saulcy, F-57045 Metz cedex 01, France
1Université de Picardie, 80039 Amiens, France
2Agence d’Essai Ferroviaire AEF-SI Vitry Sur Seine Paris, France

azari@sciences.univ-metz.fr

Fatigue fracture is a complex domain that manufacturers have always been intended to control in
order to predict the life duration of structures. For this purpose, many criteria saw the day and each
of them is based on a specific formulation. Palin Luc proposed an energetic criterion that takes into
account the stress gradient, nature of applied load and triaxiality during a complete cycle of
loading and over the whole cross section. Nevertheless, this approach is only valid for alternate
proportional multiaxial loading, which is not compatible with the signal imposed in our
investigation. Indeed, the applied loading (combined traction-torsion) corresponding to that
undergone by a railway wheel, is a nonproportional multiaxial loading.
To transform a nonproportional multiaxial loading to an equivalent proportional multiaxial
loading, using Palin Luc’s criterion, four methods are possible:
• Case 1: the equivalent signal is sinusoidal (Figs. 1-2) and stress average values are chosen
as:

V trac  tor
k 6 . 1954
W trac  tor

FIGURE 1. Applied signal FIGURE 2. Equivalent signal

• Case 2: k factor is taken as the maximum stress ratio. Fig. 3 depicts the applied
nonproportional multiaxial loading while Fig. 4 illustrates the equivalent proportional
multiaxial loading whose stress ratio is:
max V trac  tor
k 3 . 18
max W trac  tor

• Case 3: If all the points of the signal are taken into account (Fig. 6), k ratio varies during
the cycle and its average is computed by integration. To apply Palin Luc’s approach, one
290 A. Chamat et al.

compares the average strain energy density and the endurance limit of the strain energy
density

.
FIGURE. 3: Applied load FIGURE. 4: Equivalent load

• Case 4: one compares the maximum strain energy density and the endurance limit of the
maximum strain energy density.

FIGURE. 5: Applied signal FIGURE. 6: Evolution of k

To preserve the effect of the whole points of the nonproportional multiaxial loading on the life
duration, one took a load factor as a parameter to define the signal. Its value is equal to one for the
real signal. Experimental results predict a load factor of 2.56 in the endurance domain.
Table 1 summarizes Palin Luc’s model predictions for the different cases using the first
(ellipse quarter) and second (energetic) criteria.

TABLE 1: Predictions of Palin Luc’s model

Application of the nonproportional multiaxial loading to Palin Luc’s approach revealed that
only propositions established in cases 2 and 4 predict a load factor close to the experimental value.
In view of the previous results, the transformation of any nonproportional multiaxial loading to a
proportional multiaxial loading, by implementing its extreme values in Palin Luc’s formulation,
gives satisfactory results.
 2T8. Polymers and composites 291

ACOUSTIC EMISSION MONITORING OF DELAMINATION GROWTH IN

FIBER-REINFORCED POLYMER-MATRIX COMPOSITES

A. J. Brunner and M. Barbezat

Laboratory of Materials and Engineering, Empa,
Swiss Federal Laboratories for Materials Testing and Research
Überlandstrasse 129, CH-8600 Dübendorf, Switzerland
andreas.brunner@empa.ch, michel.barbezat@empa.ch

Acoustic Emission (AE) analysis can be applied to investigations of damage initiation and damage
growth in fiber-reinforced polymer-matrix (FRP) composite materials and structures. The potential
of AE as an aid in interpreting fracture mechanics tests of polymer-matrix composites has been
shown, e.g., by Bohse et al. [1]. Contrary to other non-destructive techniques such as, e.g.,
ultrasonic C-scan, AE monitoring can be applied on-line during the tests. Results of the AE
analysis are available virtually in real-time, even though a full analysis may require post-test data
processing. Fig 1 shows an example of AE data obtained from a standardized Mode I opening load
test on a unidrectionally carbon fiber-reinforced epoxy beam specimen. Delamination initiation,
for example, can easily be identified by the steep increase in AE activity. More recently, Cartié et
al. [2] have applied AE analysis to fracture mechanics tests on complex, three-dimensionally
reinforced FRP specimens.

FIGURE 1. Acoustic Emission activity (number of hits per second, left) from a standard double
cantilever beam test in Mode I (opening) loading of a carbon fiber reinforced epoxy beam
specimen, the applied load is shown in green (in kN, right hand scale)

Analysis of AE activity and AE intensity versus time or load can complement and even replace
standard visual monitoring of delamination initiation and growth [1]. This eliminates the
subjective, operator-dependent determination of crack or delamination length needed for fracture
toughness calculations. Initiation is determined both, more consistently and conservatively with
the use of AE analysis.
Some limitations of the AE analysis of fracture mechanics tests will be discussed mainly by
taking AE data from fracture mechanics tests in Mode I opening and Mode II shearing load. AE
signal attenuation with distance from the source location is quite substantial in FRP composites.
AE signal source location accuracy is limited by the anisotropy in FRP properties and also affected
by large signal attenuation. This constitutes limiting factors in fracture mechanics testing of larger
292 A. J. Brunner and M. Barbezat

elements or structures. Separation of extraneous noise signals from those originating from damage
accumulation is the main problem in fatigue tests.
On the other hand, AE analysis can help in identifying whether a single mechanism (e.g.,
delamination growth) is dominating damage accumulation or whether several mechanisms are
active at the same time or sequentially [3]. Combined with visual observation during and after
testing, complex damage accumulation (initiation) and fracture mechanics processes can be
investigated.

References
1. Bohse, J., et al., In Proceedings of the Second ESIS TC4 Conference on Fracture of Polymers,
Composites and Adhesives, edited by J.G. Williams, A. Pavan, ESIS Publication No. 27,
Elsevier, Oxford, 2000, 15-26.
2. Cartié D.D.R. et al., In Proceedings of the Third ESIS TC4 Conference on Fracture of
Polymers, Composites and Adhesives, edited by B.R.K. Blackman, J.G. Williams, A. Pavan,
ESIS Publication No. 23, Elsevier, Oxford, 2003, 503-514.
3. A.J. Brunner et al., In Proceedings of the European Conference on Macromolecular Physics:
Surfaces and Interfaces in Polymers and Composites, edited by R. Pick, Volume 21B,
European Physical Society, 1997, 83-84.
 2T8. Polymers and composites 293

FRACTURE MECHANICS VERSUS STRENGTH CONCEPTS FOR

EVALUATION OF ADHESION QUALITY

B. Lauke
Leibniz-Institut für Polymerforschung Dresden e.V.
Hohe Str. 6, 01069 Dresden, Germany
laukeb@ipfdd.de

The quality of adhesion between reinforcing components and matrix in composites, bonding
between a substrate and adherent or the strength of a welding line in two component injection
moulding is important for the mechanical properties of these materials. Consequently the
determination of micromechanical adhesion parameters is decisive for the understanding and
improvement of macromechanical properties of composites.
The characterisation of adhesion between different materials generally follows two concepts:
determination of adhesion strength or determination of fracture mechanics parameters. The
application of the strength concept at the interface between two materials involves major principle
problems. Because of the inhomogeneous stress distribution in most of the applied test methods the
normalisation of the applied critical force with the cross section of the sample provides only a
rough approximative measure of the composite quality but not a material property describing
adhesion strength between the components. Another way to characterise adhesion is given by the
consideration of a composite with a pre-crack, that leads to the fracture mechanical approach of
bimaterial composites.
In the following presentation some aspects of these two approaches are discussed for special
experiments proposed in literature (Single fibre pull-out test) and own developed tests (Curved
specimen particle test and Curved bimaterial test, see Fig. 1)

(a) (b)
Figure 1. Determination of adhesion strength between particle and matrix (a), determination of
adhesion strength between two polymers (b).

Single fibre pull-out and microdroplet test

These test have been the subject of numerous fundamental experimental and theoretical
investigations during the last years. In these tests one of the most obvious problems is the very
inhomogeneous stress state induced at the interface. Normal stresses at the entrance point and
shear stresses more inside the matrix droplet are responsible for a changing stress state for a
moving crack at the interface. Both test set-ups have been modelled by the finite element method /
1/ and the calculated stress state has been used to evaluate fracture mechanical parameters as:
294 B. Lauke

stress intensity factors and energy release rate. A very essential result of the analysis is the
disclosure of the strong influence of the mixed-mode stress state on the debonding process.

Single-particle in curved specimen test

As a consequence of the disadvantages of the above test geometries for fibre/matix adhesion
testing we have developed a test set-up that induces a rather uniform stress state normal to the fibre
interface /2/. This basic idea was also applied to the particle-matrix bonding problem as shown in
Fig. 1a, cf. /3/. Tensile axial loading of such a curved specimen concentrates stresses in the
smallest cross section and causes a multiaxial loading situation in the centre, i. e. in the vicinity of
the enclosed particle. By variation of sample curvature the distribution of normal tensile stresses at
the interface between particle and coating can be changed. This enables the variation of the
interface area which is under tensile stress. A finite-element analysis provides the stress field
within the whole specimen and especially in the vicinity of the coated particle. The calculations
provide the maximum radial stress at the particle surface as a function of applied load. Assuming
that normal stresses at the interface are responsible for debonding, the adhesion strength can be
obtained from the experimentally determined critical load at debonding initiation.

Curved bimaterial test

The determination of adhesion strength between compact materials is still an unsolved

problem. There are various tests available to determine shear strength values, for example, from
single and double shear lap tests or tensile adhesion strength from butt like tensile tests. However,
the stress analysis of these testing arrangements reveals that the stress state is uniform only within
the central regions of the sample whereas it is singular at the edges. Therefore the calculation of an
adhesion strength by normalising the critical applied load by the whole interface area is
questionable. The curved bimaterial test /4/ proposes an experimental arrangement that provides a
uniform transverse tensile stress at the interface which does not have concentrations or
singularities at the edges. The stress state at the interface is purely tensile and does not contain
shear components. A special optimised sample geometry generates this stress state.

References
1. Lauke, B., Schüller, T., Beckert, W., In The Application of Fracture mechanics to Polymers,
Adhesives and Composites, edited by D.R. Moore, Elsevier Ltd. and ESIS, 2004, pp. 227-
232.
2. Schüller, T., Beckert, W., Lauke, B., Friedrich, K., Composites Sci. and Technology, vol. 60,
2077-2082, 2000.
3. Lauke, B, submitted to Composites Sci. and Technolog, 2005.
4. Schüller, T., Lauke, B., Intern. J. of Adhesion & Adhesives, vol. 22, 169-174, 2002.
 2T8. Polymers and composites 295

ALTERNATIVE APPROACHES FOR THE EVALUATION OF THE SLOW

CRACK GROWTH RESISTANCE OF POLYETHYLENE RESINS USED IN
THE PRODUCTION OF EXTRUDED WATER PIPES

F. M. Peres and C. G. Schon

Dept. of Metall. and Mater. Engng., Escola Politecnica da Universidade de Sao Paulo
Av. Prof. Mello Moraes 2463, CEP 05508-900 São Paulo-SP, Brazil
fabiano.peres@poli.usp.br, schoen@usp.br

Polyethylene (PE) pipes have been increasingly used in the water distribution industry in the last
decades. Major advantages of this kind of material over its competitors (e.g. cast iron or PVC
pipes) include flexibility, low costs and ease of installation. Despite of its popularity, PE pipes,
which are expected to present service lives of about 50 years after installation [1,2], suffer from
premature failures by fracture due to time-dependent deformation (creep). These failures lead to
both environmental and economic (water loss and maintenance) costs.
The search for improved materials leads the petrochemical industry to the continuous
development of new resins. Strategies like the increase of mean molecular weight,
copolimerization (chain branching) and molecular weight distribution engineering have been
suggested to increase resistance to failure [2], which is nowadays accepted to occurs via slow crack
growth (SCG) [3]. This resistance to in-service failure is usually measured in Long-Term
Hydrostatic Strength (LTHS) tests, in which the tube is subject to different inner hydrostatic
pressures, P, and different temperatures, T. The output of such tests is a bi-log graph of the highest
principal stress (since the stress state of the tube is not uniaxial, this corresponds to the
circumferential, or “hoop”, stress, Vh) versus time-to-fracture (tf) , known in the industrial praxis as
the “regression curve” of the given resin. In a typical PE resin the Vh u tf curve is linear with
negative slopes in such graphs (i.e. it is typical of a power-law behavior, with negative exponents),
but two regimes can be identified: A high stress regime (Vh t Vc), in which the fracture is followed
by extensive macroscopic plastic deformation (with the bulging or “ballooning” of the tube wall),
denoted the “ductile”-mode, and a low stress regime (Vh < Vc), characterized by negligible
macroscopic plastic deformation and a small through longitudinal fracture (slit fracture), denoted
the “brittle-like” mode [1,2]. The “ductile-to-brittle” transition stress, Vc, is characteristic of a
given formulation (i.e. resin + additives) and is not easily determined in tests conducted at ambient
temperatures.
In spite of its widespread acceptance in the industry, the LTHS test must be criticized. Due to
the long duration of the tests (up to ~104 hours) LTHS tests are considerably expensive (which
makes them an impracticable quality control tool) with most of the data being collected at high
inner P (i.e. in the “ductile” region) to save time, while the tube is expected to operate in the
“brittle-like” region. The LTHS philosophy, as used in the industry, also implies that tf is a
material property, while extrinsic factors (e.g. damage introduced during the installation) may
shorten the fracture nucleation time, leading to a premature failure (as observed in the practice).
The aim of the present work is to present and to test two alternative approaches, which may be
applied to the comparative evaluation of two or more resins without the need to proceed with a full
scale LTHS test. For this purpose five different resins designed for pipe extrusion (for which the
“regression curves” are known) and two resins designed for other purposes, obtained from four
traditional PE resin suppliers, have been compared. The two approaches are the “ramp test”,
proposed by Zhou et al.[4], and the Essential Work of Fracture (EWF) method [5].
296 F. M. Peres and C. G. Schon

The “ramp test” method consists in the evaluation of the yield stress (Vy) and the drawing
stress (Vd) of standard tensile test samples as a function of the strain rate. Both quantities, when
plotted against the logarithm of the strain rate, follow straight lines with different slopes. The
intersection of both lines, according to the original proposition, corresponds to the critical stress
(Vc) for the ductile-to-brittle transition in the “regression curve” [4]. In spite of its minimalistic
simplicity, this method is based on solid hypotheses about the micromechanics of fracture process
(craze nucleation-and-growth and fibril rupture) in PE [6]. The procedure here proposed
corresponds to combining a limited number of shorter LTHS tests at higher P, to determine the
ductile wing of the “regression curve”, with the “ramp test”, to determine the lower limit of stress
(Vc) at which an extrapolation of this wing would be safe.
The EWF, on the other hand, allows a direct estimation of the essential part (we) of the specific
work of fracture in standard (e.g. Double-edge notched tensile, DENT) samples. Since this
quantity, in principle, is a material property [5], it would lead to better results in the evaluation of
the resistance to slow crack growth of the resin, in contrast with the standard “regression curve”
approach.
The results show that the values of Vc determined via the “ramp test” are compatible with the
ones estimated from the available “regression curves” for the resins and that the EWF
measurements allow to order the resins consistently with their expected behaviors. There are also
indications that the results obtained by the EWF method are richer in information compared with
the standard LTHS approach.

References
1. Janson, L.-E., Plastic pipes for water supply and sewage disposal, VBB/SWECO
International, Stockholm, Sweden, 2003.
2. Mills, M. J., Plastics: microstructure and engineering applications, 2nd ed., Edward Arnold,
London, UK, 1993.
3. Hamouda, H. B. H., Simoes-betbeder M., Grillon, F., Blouet, P., Billon, N., Piques, R.,
Polymer, vol. 42, 5425, 2001.
4. Zhou, W., Chen, D., Shulkin, Y. Chudnowsky, A., Liraj, N., Sehanobish, K. and Wu, S., In
Plastic failure – Analysis and Prevention, edited by Moalli, J., William Andrew Publishing/
Plastic Design Library, 2003.
5. Clutton E. In Fracture Mechanics Testing Methods for Polymers, Adhesives and Composites,
edited by Moore, D. G., Pavan, A. and Williams, J. G., Elsevier Science, Amsterdam,
Holland, 2001.
6. Chudnowsky, A. and Shulkin, Y. Int. J. Fracture, vol. 97, 83, 1999.
 2T8. Polymers and composites 297

A STEREOSCOPIC METHOD FOR FRACTOGRAPHIC INVESTIGATIONS OF

ORDINARY CERAMICS

C. Manhart and H. Harmuth

Christian Doppler Laboratory for Building Materials with Optimised Properties, University of
Leoben, 8700 Leoben, Austria
Christian.Manhart@unileoben.ac.at, Harald.Harmuth@unileoben.ac.at

Quantitative evaluation of fractographic investigations of ordinary ceramic materials with a

maximum grain size of e.g. 4 mm is less common compared with investigations of fine ceramics
and metals. For the investigations presented here fracture surfaces of a wedge splitting test
described in [1] have been applied. The wedge splitting test already enabled a fracture mechanical
characterization, and fracture mechanical parameters have been correlated with the investigations.
Two of these fracture mechanical parameters where the notch tensile strength NT and a
characteristic length lch (Eq. 1):

G f ˜E
l ch 2
V NT
(1)
In (1) Gf is the specific fracture energy and E Young’s modulus. The characteristic length is
inverse proportional to a brittleness number, a decreasing brittleness of the material is shown by an
increasing characteristic length.
For stereoscopic investigations two images of each fracture surface with different tilt angle
have been used. From these a digital surface profile was calculated on the basis of a three
dimensional surface model, see Fig. 1.

FIGURE 1. Schematic drawing of a 3D-digital surface model

This profile was evaluated in dfferent ways. One the hand, surface texture parameters
according to [2] and the fractal dimension have been calculated. On the other hand, an
autocorrelation function A(k), according to Eq. (2) has been calculated:

1 N k
N
> 
˜ ¦ xi  x ˜ xi k  x @ 1 N
A (k ) i 1
1 N

˜ ¦ xi  x
2

with x
N
¦
i 1
xi (2)
N i 1

In (2) xi is the ordinate of point i of the fracture surface pependieular to its base line, and N is the
number of points evaluated.
298 C. Manhart and H. Harmuth

FIGURE 2. Dependence of the characteristic length lch and the nominal notch tensile strength
ıNT on the average of the lateral correlation length of refractory specimen

A lateral correlation length  was determined according to [4]. As an example results for burnt
refactory specimens with different brittleness are shown in Fig. 2. It is obvious that the
characteristic length increases with the lateral correlation length , and the nominal notch tensile
strength NT decreases with . Contrary to this no correlation of the fracture mechanical parameters
determined in this investigation with the fractal dimension was observed.
The results will be applied to establish stucture/property relationships and to investigate
mechanisms of brittleness reduction for ordinary ceramic materials.

References
1. Tschegg, E.K., New equipments for fracture tests on concrete, Materials Testing, 338-342,
1991.
2. EN ISO 4287, Geometrical Product Specifications (GPS) - Surface texture: Profile method --
Terms, definitions and surface texture parameters , Deutsches Institut für Normung, 1997.
3. Ervin, E., Metals Handbook, Ninth Edition, Vol. 12, Fractography, ASM, Metals Park Ohio
44073, 193-210, 1987.
4. Yang, H.-N., Wang, G.-C, Lu, T.-M. Diffraction from Rough Surfaces and Dynamic Growth
Fronts, World Scientific, Singapore, p. 64, 1993.
 2T8. Polymers and composites 299

MODELINGS OF FIBER DEFORMATION DURING MACHINING ARAMID-FRP

Eitoku Nakanishi, Masao Fukumori1, Yutaka Sawaki and Kiyoshi Isogimi

Department of Mechanical Engineering, Faculty of Engineering, Mie University
1Department of Mechanical Engineering, Graduate school of Engineering, Mie University

Kurima-machiya 1577, Tsu, Mie 514-8507, JAPAN

nakanishi@mach.mie-u.ac.jp, fuku@sks.mach.mie-u.ac.jp, sawaki@mach.mie-u.ac.jp,
isogimi@mach.mie-u.ac.jp

Recently, composite materials are used in very widely. But the machining the composite materials
is very difficult due to the difference of machinability and of mechanical properties between the
fiber and matrix materials. Especially, machining the Aramid Fiber Reinforced Plastics (A-FRP)
causes the poor machined surface by large fluffs of Aramid fiber[1] as shown in Fig.1(A). In this
paper we tried to clarify the fracture phenomena of Aramid fiber during machining to suggest an
optimum machining conditions. We observed the deformation of Aramid fiber bundle inside of the
matrix in transparently during machining by using optical microscope. By this method, we can
observe the deformation of Aramid fiber and delamination between fiber and matrix clearly as
shown in Fig.1(B).

FIGURE 1. SEM and optical microscope observations.

The machined surface appearances and delamination between fiber and matrix are much
affected by fiber orientations of specimen. We expand the single fiber beam theory[2] to
addressing the fiber bundle deformation[3]. In this theory, Aramid fiber is regarded as beams and
the matrix is regarded as elastic foundation. And we can express the deformation of Aramid fiber
bundle during machining as shown in Fig.2. Further, we consider that the fracture point of fiber
may much affect the fluffs length and surface appearances. And we try to make a simple model
that is based on elastic theory to evaluate the fracture point or higher stress filed of Aramid fiber
bundle as shown in Fig.3.
300 E. Nakanishi et al.

FIGURE 2. Deformation of Aramid fiber bundle.

FIGURE 3. Stress distribution.

References
1. E. NAKANISHI, J. SUZUKI and K. ISOGIMI, Effect of Vibrations on Fracture Phenomena
in Machining Aramid-Glass Hybrid Composites, Engng., Trans., 47, 2, 145-154, 1999.
2. X. WANG, K. NAKAYAMA and M. ARAI, Investigation on the Cutting of Fiber Reinforced
Composite Materials (2nd Report) -Mechanism of Surface Generation in Cutting FRP-(in
Japanese), Journal of Japan Society of Precision Engineering, Vol.57, No.8, 1437-1442, 1991.
3. E. NAKANISHI, Y. SAWAKI and K. ISOGIMI, Modeling of Deformation of Aramid Fiber
Bundle During Machining Aramid FRP (in Japanese), Journal of Japan Society of Advanced
Production Technology, Vol.22, No.1, 67-74, 2004.
 2T8. Polymers and composites 301

QUALITY CONTROL AND THE STRENGTH OF GLASS

Fred Veer, Christiaan Louter and Ton Romein1

Faculty of Architecture , Delft University of Technology
1van Noordenne groep

Po Box 5043, 2600 GA Delft

f.a.veer@bk.tudelft.nl

To determine the strength of glass is important to determine the dimensions of glass structures.
Conventional materials have singular strength values that are quite reproducible and independent
of size. The strength of glass is strongly dependent on the size of the glass and the quality of the
edges.
In earlier work, Veer [1], it was shown that the strength of small specimens of glass cannot be
fully described by a simple Weibull function. Later research, Veer, [2], has shown that in larger
pieces of annealed and also tempered float glass the data can still not be described by a single
function. In addition significant differences are found when glass is tested flat or standing with the
full edge stressed in 4P bending. Again these results can not be described adequatly with a Weibull
function as is shown in figure 1.

Figure 1 : Weibull plot of fully tempered float glass 1000´125´10 mm

One theory that explains the deviation was proposed. Namely that the glass specimens in the
series were processed differently resulting in different flaw shapes and distributions. To investigate
this several series of glass specimens were ordered with the following specifications.
• Final size of specimens should be 1000´125´10 mm.
• All specimens in one serie should be processed on one grinding line with all specimens
being done one after the other without interruption
• All machines should be cleaned and the grinding heads checked and if necessary replaced
before starting the processing
• As many different grinding lines as possible should provide a series
All specimens are then tested under identical conditions. Each series is processed separately
with Weibull coefficents being determined. The results of the series are then mixed and a seperate
302 F. Veer et al.

statistical analysis of the whole is made. The result show that each individual group has behaviour
that can be described better than the collective results of all series.
It is thus concluded that the differences between the individual processing lines contributes to
the scatter. The strength of glass is thus also dependent on the quality of the machinery and the
quality control in processing.

References
1. F.A.Veer, J.Zuidema, The strength of glass, effect of edge quality, Proceedings Glass
processing days conference 2003, Tampere, Finland, Tamglass ltd.Oy.
2. F.A.Veer , F.P. Bos, J. Zuidema, T. Romein, Strength and fracture behaviour of annealed and
tempered float glass, Proceedings ICF 11 conference 2005 , Turin, Italy, edited by A.
Carpinteri
 2T8. Polymers and composites 303

EXPERIMENTAL STUDY OF CRACKED LAMINATE PLATES BY CAUSTICS

G.A. Papadopoulos and E. Sideridis

Section of Mechanics, The National Technical University of Athens, 5 Heroes of Polytechnion
Avenue, GR-157 73, Zografou, Athens, Greece.
gpad@central.ntua.gr

The strength of a composite material in form of laminate is obtained from the properties of the
constituent laminae. The interface between different laminae is an important factor since it
influences the stresses which are developed in the laminate and hence the strengths.
In this study an experimental investigation of the strength of symmetrical and unsymmetrical
cracked laminates made of isotropic layers [Lexan (PCBA) and Plexiglas (PMMA)] was
attempted. An analysis based on the lamination theory was performed in order to determine the
elastic constants of these laminates.
Experimental measurements of the strength and the stress intensity factor KI in cracked
specimens made of laminates with different stacking sequences were carried out. The material we
deal is a laminate made of 2, 3 and 4 layers of isotropic material.
In plane stress conditions in xyz axis system the stress-strain relationships are given as [9]

­V x ½ ª Q 11 Q 12 0 º ­ H x ½
° ° «Q Q 0 » °H °
®V y ¾ « 12 22 » ® y ¾
° ° ° °
¯W xy ¿ ¬« 00 Q 66 ¼» ¯ J xy ¿ (1)
The elements Qij of the stiffness matrix are related with the material properties as follows

E QE E
Q 11 2
Q 22 , Q 12 2
, Q 66 G
1 Q 1 Q 2 (1  Q ) (2)

Eq.(1) can be thought of a stress-strain relationship for the kth layer of a multi-layered
laminate. Thus, it can be written as

^V `k >Q @k ^H `k (3)
On the other hand the elastic constants in the laminate plane Ex, Ey, xy, Gxy can be determined
by the classical theory of laminated plates by the aid of models which assume uniform stress
through the thickness of the laminate. Thus:

1 1  A12c 1
A11c , c ,
A 22 Q xy , c
A 66
Ex Ey A11c G xy
(4)

1ª N º
where A ijc A ij  1 and A ij « ¦
t ¬ k 1
( Q ij ) k t k » .
¼
Here tk is the thickness of kth layer and t the thickness of the whole laminate.
304 G.A. Papadopoulos and E. Sideridis

Specimens were made of 2-4 layers in symmetrical or unsymmetrical combination of Lexan

(PCBA) and Plexiglas (PMMA) layers having 1u103 m thickness each, thus providing the form
of a symmetrical laminate [see Fig. 1a]. For the bonding of the layers special glue
(trichloroethylene-dicloromethane 2/1) was used. In order to measure the strains longitudinal and
transversal strain gauges (KYOWA type with gauge length 2 mm) and Huggenberger
extensometer were used.
Tensile experiments were carried out, to measure mechanical properties, using dogbone
3
specimens with total thickness varying from 2 u103 m to 4u10 m according to ASTM D638.
2 3
The width of the specimens was 30 u 10 m near the grips and 13 u 10 m at the mid – length
whereas the total length vas 180 u 10  3 m (see Fig. 1b). An edge crack of 5x10-3 m was cut in the
specimens.
The experimental obtained values of the Elastic modulus, E, Poisson ratio, Q , and ultimate
stress V ult for PCBA and PMMA are given in Table 1. The elastic constants E, of the sandwich
were used in caustics experiments caustic for the calculation of the stress intensity factor KI. The
stress intensity factor was compared with those of the layers of the sandwich.

(a) (b)
Figure 1. (a) The sandwich laminate, (b) Geometry of the dogbone spesimens.

TABLE 1: Elastic modulus, Poisson ratio and ultimate stress of Lexan and Plexiglass
 2T8. Polymers and composites 305

FRACTURE OF COMPOSITES IN MILITARY AIRCRAFT

Robert Pell, Nick Athiniotis and Graham Clark

Defence Science & Technology Organisation
506 Lorimer St. Fishermans Bend, Victoria, Australia, 3207
graham.clark@dsto.defence.gov.au

The Defence Science and Technology Organisation (DSTO) provides failure analysis and accident
investigation support to the Australian Defence Forces (ADF) and over 60 years has developed a
strong capability as an impartial adviser on aviation failures.
This paper provides a brief overview of military aviation accident investigation and failure
analysis in Australia, an activity which draws upon a wide range of scientific and technical
capability, including engineering analysis.
In some specialised areas, notably composite materials, DSTO anticipated a requirement for a
significant capability in failure analysis, on the basis that failure investigations in which the failure
mode of the composite structure in an aircraft is not understood will always be open to doubt. As
the use of composite materials in aircraft manufacture continues to increase the requirement to be
able to analyse failures in composite materials also increases.
Over some years, DSTO conducted research and participated in an international collaboration
program to develop a functional capability in the failure analysis of composite materials. This
effort was built upon a strong science foundation that already existed at DSTO in the design and
manufacture of composite components, especially with respect to the repair and refurbishment of
aging aircraft structures.
This approach has resulted in a small, but expert capability in DSTO. In most instances
composites failure analysis, by methodical examination of the fracture surfaces of broken fibres,
can be used to determine whether fibre fracture has occurred under tension, shear or compression
loading. Similarly, by detailed examination of the matrix material on the surface of a
delamination, composites failure analysis can be used to determine whether the delamination was
introduced as a defect during manufacture or if it occurred, in service, as a result of shear or
transverse tension loading.
The ability to identify failure modes in carbon-fibre, boron-fibre and glass-reinforced
composites under a variety of loading conditions, and at a variety of scales, has been applied on an
opportunity basis to failures experienced in the aircraft flown by the ADF, as well as a number of
non-aviation cases.
Citing one example, the ability to identify the features generated in carbon fibre fractures
produced by compression loading (Fig. 1) made a significant contribution to an accident
investigation conducted for the ADF. Part of the accident damage included the tail rotor blades of
an ADF helicopter all of which were broken in an identical manner. This could have been caused
by ground impact, however, it was possible that the blades may have been severed in flight. It was
essential to determine the cause of failure and the identification of compression features in each of
the tail rotor blade fractures confirmed that the failure had occurred under bending loads that
would be consistent with ground impact.
306 R. Pell et al.

Figure 1. A schematic of the fracture process that occurs under compression loading and examples
of the micro features used to identify compression fracture,

The paper will present an overview of the variety of composite materials and structures,
ranging from carbon fibre yacht masts to fibreglass reinforced automotive timing belts that have
been successfully analysed at DSTO using composites failure analysis techniques.
In the context of DSTO’s failure analysis and accident investigation activity as applied to
military aircraft, the paper addresses, in some detail, a number of examples of failures in both
carbon fibre and fibreglass composite components.
These examples serve to illustrate the complex nature of the fracture/failure process in
composite materials, and the contribution made by analysis of the fractures surfaces generated
during failure. The paper also highlights the need to analyse fractures on both the macro and
micro-scales with the need to reconcile these observations to achieve a valid conclusion.
 2T8. Polymers and composites 307

ANALYSIS OF 7005/AL2O3/10P MMC SHEETS JOINED BY FSW BY

THERMOELASTICITY

P. Cavaliere, G. L. Rossi1, R. Di Sante and M. Moretti1

Dept. of Ingegneria dell’Innovazione, Engineering Faculty, Univ. of Lecce,
I-73100-Lecce, Italy
1Dept. of Industrial Engineering, Univ. of Perugia

Friction Stir Welding represents a very attractive technology in reducing residual stresses and
distortion in joints compared to conventional fusion welding ones. In particular, FSW of
Aluminium based metal matrix composites is less affected by the inconveniences deriving from
debonding and segregation of reinforcing particles. In this study, the metal matrix composite under
investigation was a 7005 aluminium alloy reinforced with 10% of alumina particles Friction Stir
Welded (T6 condition) by employing a threaded tool rotating speed of 600 RPM and a welding
speed of 250 mm/min. The optical and scanning electron microscopy observations performed on
the different zones of FSW joints cross section revealed the different structures of the nugget, the
thermo-mechanical affected zone and the heat affected zones thanks to the difference in
reinforcing particles dimensions as a consequence of friction process. Such phenomenon is
accompanied with a strong grain refinement due to a dynamic recrystallization acting during the
severe plastic deformation to which the material is subjected during the welding process.
Only a few data are available in the literature on the fatigue behaviour of these materials. The
aim of this work is also to apply thermoelastic stress analysis to the study of crack formation and
propagation of friction stir welded MMC sheets, during cyclic fatigue tests. The room temperature
tensile properties, previously measured in transverse direction respect to the welding one gave a
response on the yield strength of 290 MPa, on UTS of 310 MPa with a measured strain to fracture
of 1.5 %. Fatigue tests were carried out under the axial total stress-amplitude control mode with
R=Vmin/Vmax = 0.1 using a resonant electro-mechanical testing machine (TESTRONICTM 50r 25
KN by RUMUL (SUI)). All the mechanical tests were performed up to failure which occurred at
the interface with the welded area. The TSA measurement system allowed the crack evolution to
be observed in real-time during fatigue cycles and stress fields to be derived on the specimens from
the temperature variation measured.
In the last stage of this work, the microstructure resulting from the FSW process was studied
by employing optical and scanning electron microscopy, in order to correlate the microstructural
evolution produced by the welded process with the fatigue properties previously determined.

Keywords: FSW, MMC, TSA

 2T8. Polymers and composites 309

SURFACE MODIFICATION OF LIGHTWEIGHT AGGREGATE AND

PROPERTIES OF THE LIGHTWEIGHT AGGREGATE CONCRETE

T. Y. Lo and H. Z. Cui
Department of Building & Construction
City University of Hong Kong
Hong Kong, P. R. China
bctommyl@cityu.edu.hk, 50006504@student.cityu.edu.hk

Structural lightweight aggregate concrete (LWAC) offers design flexibility and substantial cost
savings by providing less dead load [1]. Because of higher water absorption of lightweight
aggregate (LWA), process of mixing LWAC is more complex than that of normal weight concrete
(NWC). Generally 1-hour water absorption of structural LWA is 6~15%, but that of normal
weight aggregate (NWA) is lower than 1% [2]. Therefore pre-wetting LWA is typical method for
LWA treatment for producing qualified LWAC, though the process of pre-wetting LWA is
inconvenient [3]. In this paper, three aspects of LWA study have been carried out, i.e. a) LWA
surface modification, b) mechanical properties and durability of LWAC made with the modified
LWAs and, c) the study of interfacial zone of the LWACs.
The effects of different concentration of surface modifier on water absorption of modified
LWA were studied. Figure 1 shows the water absorption of unmodified and modified LWAs and
displays the surface modification for LWA can reduce the water absorption obviously. For LWA
modified with 1:5 concentration modifier, the water absorption only was 10.5% of unmodified
LWA.
Studies of mechanical properties and durability of LWACs [4] made with modified and
unmodified LWA were carried out. The mechanical and durability properties of different LWAC
series were determined by conducting cubic compressive, static modulus of elasticity, shrinkage
and water permeability tests. 28 days LWAC compressive strength could be up to 46.1 MPa
corresponding to the modified LWA used 1:20 concentration modifier. Based on the test results, it
can be known that the surface modification can reduce water absorption of LWA effectively but
without obvious disadvantageous effect on mechanical and durability of the LWAC.

FIGURE 1: Water absorption of unmodified and modified LWAs

Scanning electron microscope (SEM) was used to study the interfacial zone of the LWACs.
From the study of the microstructure of the interfacial zone it can be known that, for LWA
310 T. Y. Lo and H. Z. Cui

modified with a lower concentration modifier, bonding strength between the hardened cement
paste and modified LWA was better.

References
1. Satish Chandra, Leif Berntsson, Lightweight aggregate concrete: Science, Technology and
Applications, William Andrew Publishing, USA, 2002
2. FIP Manual of Lightweight Aggregate Concrete, 2nd ed., Halsted Press, London, 1993.
3. Y. Lo, et al, Microstructure of pre-wetted aggregate on lightweight concrete, Building and
Environment, Volume 34, Issue 6, November 1999, Pages 759-764
4. Joao A. Rossignolo and Marcos V.C. Agnesini, Durability of polymer-modified lightweight
aggregate concrete, Cement & Concrete Composites 26 (2004) 375–380
 2T8. Polymers and composites 311

FINITE ELEMENT BASED PREDICTION OF FAILURE IN LAMINATED

COMPOSITE PLATES

H. Hosseini-Toudeshky, B. Hamidi1, B. Mohammadi2 and H. R. Ovesi3

Associate Professor, 1,2Post Graduate Student, 3Assistant Professor
Aerospace Engineering Department, Amirkabir University of Technology, No. 424, Hafez Ave.,
Tehran, Iran
hosseini@aut.ac.ir

Damage accumulation in laminated composite is a major concern in the design and use of
composite structures. Accumulated damage can affect laminated response and ultimate strength,
which are critical to design of load-carrying structural components. For the optimal design of
composite structures, laminate response beyond the point of initial matrix cracking must be known,
and subsequent damage and failure modes induced by accumulated matrix crack must be
understood.
Matrix cracking is often part of the failure process in fiber composites. It typically manifests
itself as cracking in the off-axis plies of a laminate, for example, in the 90° plies of a laminate
loaded in the 0° direction. The matrix cracks affect the stress distribution within the plies of the
laminate by lowering the transverse stiffness of the plies. Matrix cracks also can serve as flaws for
delamination initiation. Experimental results showed that matrix cracking commonly leads to
subsequent forms of damage including micro-cracking and delamination [1, 2]. Tensile strength
and response of composite laminates are commonly characterized through uniaxial tensile testing
of finite-width laminate specimens.
In order to predict accurately the tensile response of laminate, these damage mechanisms and
their dependence on laminate configuration must be understood and accounted for through
analysis. Certainly one tool to use in this design procedure is to calculate the interlaminar stresses
and compare these with interlaminar allowable stress. Another point of view is to use the approach
of fracture mechanics principles.
O’Brien [3] developed a simple analytical method for evaluating the strain energy release rate
associated with delamination growth from a matrix crack. McCartney considered the mechanics
for predicting stress transfer in a general symmetric laminate, having a uniform distribution of ply
cracks in single orientation, to combined general in-plane and through the thickness loading [4]. A
two-part investigation was performed to study the stiffness response and strength of composite
laminates subjected to uniaxial tension by Johnson et al.[1]. They performed extensive tests on
specimens with specially tailored ply orientations to characterize damage initiation and
propagation induced by matrix cracks. In second part of their investigation analytical models were
developed to account for the damage modes observed in experimental investigation [2]. The focus
of the experiments was to characterize the matrix crack-induced damage progression and failure
modes in composite laminates under uniaxial tension loads.
In this paper, initiation and propagation of matrix cracking and induced delamination between
layers up to the final failure of the laminated composite plates are investigated using finite element
method. The obtained results are compared with the available experimental results in the literatures
[1]. Three dimensional finite element analyses of the laminated plates (graphite/epoxy, T300/508)
are performed for various lay up configurations and thicknesses. A macro program using the
capability of macro programming in ANSYS was developed to handle the full procedure of
modeling including the mesh generation, matrix crack initiation and propagation, and induced
delamination up to the final failure.
312 H. Hosseini-Toudeshky et al.

The major steps in the developed macro program are; 1- Geometry definitions, mesh
generation, constraints and material property definitions, 2- Applying an initial load, 3- Linear
elastic analysis to find the stress and strain fields, 4- Calculation of matrix (eM) and delamination
(edelam) criteria, 5- If eM>1 and edelam>1 then reduce the magnitude of the initial load and go to
step 3. 6- If eM>1 and edelam<1 then generate a matrix crack at the critical position and continue
the fracture analysis of plate with intralaminate crack, 7- If eM<1 and edelam>1 then generate a
delamination crack at the critical position and continue the analysis with interlaminar crack
propagation, 8- If eM<1 and edelam<1 then increase the loading magnitude and go to step 3. In the
case of interlaminar crack propagation the gap elements are used between the crack surfaces to
avoid interference of the boundaries. The above procedure continues to satisfy the fiber breakage
and/or unstable interlaminar crack growth criterion.
A list of typical laminates and description of the obtained numerical and existed experimental
results are presented in Table 1. More results will be presented and discussed in the full paper.

TABLE 1. Comparison between numerical and experimental results

References
1. Johnson P. and Chang F-K. "Characterization of matrix crack induced laminated failure-Part
I: Experiments", J. Comp. Mate. 2001; Vol. 35; 2009-2035.
2. Johnson P. and Chang F-K. "Characterization of matrix crack induced laminated failure-Part
II: Analysis & Verification", J. Comp. Mate. 2001; Vol. 35; 2037-2074.
3. O’Brien T.K, "Local Delamination in laminates with Angle-Ply Matrix Cracks, Part II:
Delamination Fracture Analysis and Fatigue Characterization", Composite Materials: 1993,
507-538.
4. McCartney L N., "Model to predict effects of triaxial loads on ply cracking in general
symmetric laminates", Comp. Scie. & Tech, 2000; Vol. 60; 2255-2279.
 2T8. Polymers and composites 313

AN EMBEDDED CYLINDRICAL PZT WITH ELECTRODED IMPERFECT

INTERFACE

H. M. Shodja and S. M. Tabatabaei

Sharif University of Technology
Department of Civil Engineering, P.O. Box 11365-9313, Tehran, Iran
shodja@sharif.edu

Recently Shodja and Kamali [1], and Kamali and Shodja [2], introduced a 3D semi-analytical
approach for determination of the electro-mechanical fields of piezoelectric solids with material
singular surfaces. The proposed methodology is particularly effective for problems involving
external and internal boundaries with complex geometries, which have closed form expressions,
and not necessarily in the form of polynomials. The previous formulations are devoted to material
singular surfaces with perfect bonding. Due to the promising features of the formulation, the
authors have continued to study various capabilities as well as convergence rate and accuracy of
the approach. The present work extends the formulation to solids in which the material
discontinuity surfaces may have one of the following six conditions: (1) perfect bonding; (2) pure
debonding; (3) in-plane pure sliding; (4) out-of-plane pure sliding; (5) full debonding; or (6) partial
debonding. Moreover, the interface is electroded in the sense that the interface is subjected to an
arbitrary electric potential function, ĭ f x, y  . One of the advantages of the proposed
method is that all of the above-mentioned cases are treated in a unified manner.
In this presentation the formulations are given in the context of the following physical
problem. Consider an infinitely extended cylindrical Lithium tantalate, : embedded in an
unbounded anisotropic elastic material, < . The configuration of the set problem and Cartesian
coordinate system are shown in Fig.1. Lithium tantalate crystal has higher piezoelectric coupling
than quartz. Lithium tantalate and quartz are different types of trigonal crystals. In the present
problem, the cartesian coordinate system is such that, z is the trigonal axes, and x is normal to a
mirror plane for the lithium tantalate crystal. It is intersting to note that, although the applied
electrical loading is a function of x and y only, but due to the intrinsic electro-mechanical
properties of the Lithium tantalate the generalized plane strain conditions will prevail. Various
types of imperfections along the interface * s between : and < are modeled by different spring
layers of zero thickness on * s .
For imperfect interface * s , the discontinuity of displacements and continuity of tractions
require that

>u i @ z 0 on ī s , i=1,2,3 (1)

>ı @n
ij j 0 on ī s , i,j=1,2,3 (2)

ĭ f x, y  on * s (3)

respectively, where [.] denotes the jump across * s in the quantity under consideration, and n j is
the outward unit normal to *s . The total electric enthalpy of the system consists of the external
314 H. M. Shodja and S. M. Tabatabaei

work Fī 0 , electric enthalpy in the piezoelectric material H ȍ  , electric enthalpy in the elastic
material H Ȍ  , and the stored energy in the spring layer of vanishing thickness F ī s

ȍ  Ȍ 
F ³ ȍ
H dv  ³ <
H dv  Fī s  Fī 0
(5)

FIGURE 1. A cylindrical Lithium tantalate in an infinite anisotropic elastic body with electroded
imperfect interface

Appropriate choice of the functions for the components of the generalized displacement field
followed by minimization of the total electric enthalpy leads to determination of the electro-
mechanical fields. For demonstration the distribution of the radial displacement for pure
debonding is shown in Fig. 2.

FIGURE 2. Distribution of the radial displacement for pure debonding

References
1. Shodja, H.M., Kamali, M.T., International Journal of Solids and Structures, Vol. 40, 4837-
4858, 2003
2. Kamali, M.T., Shodja, H.M., Computer methods in applied mechanics and engineering, Vol.
194, 4588-4604, 2005
 2T8. Polymers and composites 315

CHARACTERIZATION OF COMPOSITES FOR THE MAESLANT STORM

SURGE BARRIER

J. Degrieck, W. Van Paepegem, L. Van Schepdael1, P. Samyn, P. De Baets, E. Suister2 and J. S.

Leendertzc
Ghent University, Dept. of Mechanical Construction and Production, Sint-Pietersnieuwstraat 41,
9000 Gent, Belgium
Joris.Degrieck@UGent.be, Wim.VanPaepegem@UGent.be, Pieter.Samyn@UGent.be,
Patrick.DeBaets@UGent.be
1SOLICO BV, Solutions in Composites, The Netherlands, Ludo@solico.nl
2Ministry of Transport, Public Works and Water Management, The Netherlands

j.s.leendertz@bwd.rws.minvenw.nl

The south-west part of the Netherlands is located in a low-lying delta, where the Rhine and IJssel
rivers run into the North Sea. In 1953, the fatal combination of a north-western storm and spring
tide resulted in the inundation of large parts of the provinces of Zeeland and South Holland. In
order to prevent a repetition of the disaster, a set of measures were laid down in the Delta Act. As
the seaports of Rotterdam and Antwerp had to remain accessible, no dams could be built in the
New Waterway and the Western Scheldt. The elevation of the dikes was not an option as well.
Finally, in the 1980s, a movable storm surge barrier appeared to be the most attractive in terms of
cost, environmental effects and safety [1].
This Maeslant storm surge barrier consists of two hollow semi-circular gates attached by
means of steel arms to a pivotal point on both banks (Figure 1). In the event of a storm tide, the
“parking” docks are filled with water, so that the hollow gates start to float and can be turned into
the New Waterway. Once the gates meet, the cavities are filled with water and the gates sink to the
bottom, thus sealing off the 360 metres wide opening. The two ball-joints are designed to ensure
that the gates can move in all directions, both horizontally (when the gates are turned into the
waterway) and vertically (upon submersion). With a diameter of ten metres and a weight of 680
tons, the ball-joints are more than three times as big as the largest ball-joint ever used.

FIGURE 1. View on the Maeslant storm surge barrier.

When the barrier was commissioned in 1997, the convex and concave surfaces of the ball joint
were covered with one layer, a mixture of MoS2 and Poly Tetra Fluor Ethylene (PTFE), in order to
obtain a low static and dynamic design value for friction resistance. An additional layer of PTFE
had been applied for "run in" purposes, to overcome the damaging effect of the initial friction. In
316 J. Degrieck et al.

1999 after a test closing, wear marks were observed at the bottom concave bearings and the ring
elements. In the following years similar damage was observed. This was unacceptable for the
owner, Rijkswaterstaat Directie Zuid Holland that initiated a project for improvement of the ball
bearing.
This paper discusses the new design, where the concave surface of the ball joint has been
replaced by 500 discrete circular composite discs, consisting of an Ultra High Molecular Weight
PolyEthylene (UHMWPE) pad, reinforced with filament wound carbon/epoxy rings. Figure 2
shows the design (left) and a detail of the placement of one of the 500 composite discs in the
concave part of the ball joint.

FIGURE 2. Top view of the ball-joint (left) and placement of a composite disc in the pedestals
(right) (courtesy of Rijkswaterstaat).

Although the yield stress of the UHMWPE is about 20 MPa, the normal pressure on the
composite material discs can reach values of about 150 MPa in normal working conditions, and
laboratory tests have been done up to a load carrying capacity of 400 MPa. As the polyethylene has
been “locked up” by the carbon/epoxy ring, these high pressures can be attained. The carbon/epoxy
rings were manufactured by filament winding and only unidirectional hoop winding was applied.
Finite element simulations of the polyethylene disc with the reinforcing carbon/epoxy rings
have been combined with extensive experimental testing. Both full-scale and coupon testing was
done to determine the bearing capacity of the composite discs and to guarantee low friction sliding
contacts at high normal pressures (up to 150 MPa).
In September 2004, a successful test closing of the storm surge barrier was performed and
inspections afterwards revealed not any occurrence of damage.
 2T8. Polymers and composites 317

WEIGHT FUNCTION, J-INTEGRAL AND MATERIAL FORCES APPROACH

TO CERAMIC MULTILAYERS

Javier Pascual, C. R. Chen1, O.Kolednik2, F. D. Fischer1,3, R. Danzer and T. Lube

Institut Struktur- und Funktionskeramik Montanuniversität Leoben
1Materials Center Leoben, 2Erich Schmid Institute of Material Science, Austrian Academy of

Sciences, 3Mechanik Institut Montanuniversitat Leoben.

Peter Tunner Strasse 5, 8700, Leoben, Austria
javier.pascual@unileoben.ac.at

The aim of this contribution is to compare three different methodologies that can be apply to solve
an engineering problem as is the assessment of apparent toughness in a residual stresses-containing
material.
In the attempt of developing tougher and stronger ceramics different types of laminar ceramics
have been designed. Examples are layered composites with weak interfaces that display graceful
fracture among the layers, or layered composites with strong interfaces that are reinforced by the
presence of residual stresses in the different layers used.
This last strategy was used to produce laminar composites by stacking alternatively Al2O3 (A)
and Al2O3/3Y-TZP (AZ) layers. Thin ceramic layers were produced by tape casting. Dense
symmetric hybrid structures were obtained by warm pressing of the green sheets and pressureless
sintering. Different sintering rates during heating and thermal expansion mismatch during cooling
down from the sintering temperature induce biaxial compressive stresses in the A layers and tensile
stresses in the AZ. As it is well known these residual stresses influence properties as apparent
toughness and strength.
The residual stress field is dependent on the thermal expansion mismatch and on the elastic
mismatch. FE simulations were carried out to evaluate the residual stress field, strong edge effects
result in evidence after the analysis.
A model was used to evaluate the apparent toughness through different methodologies as the
weight function method [3], the J-Integral and material forces [4].
Apparent R-curves were calculated by means of the weight function method that also applies
to non-homogeneous materials [5]. A typical apparent or effective R-curve is presented in Fig. 1
where the influence of the architecture and, therefore, the residual stresses is revealed.

FIGURE 1. Apparent R-curve for two different ceramic laminates as calculated by means of the
weight function method.
318 J. Pascual et al.

Comparisons are established between the results from the different theoretical approaches in
relation to the experimental results. The weight function method, the J-Integral method, and the
material forces procedure were the different theoretical approaches utilized. The experimental part
was carried out by 4-point bending tests.

References
1. Clegg, W. J., Kendall, K., Alford, N. M., Button, T. W., Birchall, J.D., Nature, vol. 347, 455-
457, 1990
2. Rao, M., Sanchez-Herencia, J., Beltz, G., McMeeking, R. M., Lange, F., Science, vol. 286,
102-105, 1999.
3. Fett, T., Munz, D.: Stress Intensity Factor and Weight Functions, Computational Mechanics
Publications, 1997.
4. Simha, N. K., Fisher, F. D., Kolednik, O., Chen, C. R., J. Mechanics and Physics of Solids,
vol. 51 209-240, 2003.
5. Fett, T., Munz, D., Yang, Y. Y., Fatigue and fracture of engineering material structures, vol.
23, 191-198, 2000.
 2T8. Polymers and composites 319

ASSESSMENT OF MATRIX FATIGUE DAMAGE IN CFRP

K. J. Cain and A. Plumtree

Department of Mechanical Engineering, University of Waterloo
200 University Ave. W., Waterloo, Ontario, Canada
kcain@conestogac.on.ca, plumtree@engmail.uwaterloo.ca

An experimental study was performed on a 45o off-axis HTA/6376 carbon fibre-epoxy composite
with a 60% nominal fibre volume fraction by testing under cyclic three-point bending loading
conditions at a stress ratio of 0.1. This allowed a fundamental study to be carried out on the
continuous damage evolution in the matrix without the influence of delamination.
Damage was assessed using three techniques: measuring crack density and permanent bending
as well as monitoring the fatigue modulus degradation of the composite. The general trends of the
results of these damage measurements were similar, as seen in Fig. 1.
The damage can be divided into main two stages, with rapid growth at the onset of loading
(stage I) and a leveling off after about 10-20% of life (stage II). The transition from stage I to stage
II appears to indicate that there is a change in the damage mechanism. During this second stage,
each monitoring method shows an approximately equal increase in the amount of damage,
accumulating at a similar rate. Since stage II relates to bulk damage, the effect of the fatal crack is
not apparent.
Considering three-point bending, the amount of permanent deflection is related to the crack
density which increases with increasing cyclic stress [2]. However, fatigue modulus degradation
monitors another form of damage that includes a shift in the hysteresis curve, known as strain
ratcheting [3]. This shows the effect viscoelastic behaviour of the matrix; i.e. the effect of chemical
bonding as opposed to mechanical bonding. In the present case, the fatigue modulus damage is
dominated by this viscoelastic effect and is cycle dependent. Obviously, it is a very significant
factor in polymer-based composites that undergo long lives.
Crack density measurements are an indication of the bulk damage. They may become less
important as fracture approaches. The initial crack density appears to be stress dependent and as
number of cycles increase, the number and length of longitudinal cracks increase giving the overall
effect that the crack densities at different stress levels tend to become similar [4]. Further work on
the amount of cracked area is required to give a more details. However, it would appear that the
significant damage evolution factor at the end of the second stage is the distribution of the cracks.
Only one region of favourable crack coalescence is required to result in the formation of the fatal
fatigue crack.
320 K. J. Cain and A. Plumtree

FIGURE 1: Types of damage in CFRP tested in cyclic 3-pt bending at R=0.1. (a) Damage based on
the number of transverse cracks in 16 000 Pm2 on the tensile side. (b) Damage based on the
amount of permanent bending in the composite. (c) Damage based on the degradation of fatigue
modulus.

References
1. Plumtree A. and Shen G., Polymers and Polymer Composites, vol. 2, 83-90, 1994
2. Plumtree A. and Ostagathe M., In Proceedings of the Eleventh International Conference on
Fracture, edited by A. Carpinteri, Politecnico di Torino, Italy, 2005, Paper 3342, 1-5
3. Cain, K.J., Damage Characterization of A Unidirectional Off-Axis Graphite-Bismaleimide
Laminate, Ph.D. Thesis, University of Waterloo, 2002
4. Sorensen, L. and Plumtree A., Composite Materials: Testing and Design Fourteenth Volume,
ASTM STP 1436, edited by C.E. Bakis, ASTM International, 336-348, 2003
 2T8. Polymers and composites 321

PROGRESSIVE FAILURE OF COMPOSITE MATERIALS UNDER DYNAMIC

LOADING

Liqun Xing, Xinyu Huang and Ken Reifsnider

Department of Mechanical Engineering
University of Connecticut, Storrs, CT 06269
Reifsnider@engr.uconn.edu

Polymer-based composite materials are widely used in applications subjected to a variety of

loading types, including fluid shock loading in the range of hundreds of strain per second. The
behavior of such composite laminates loaded at those rates is typically nonlinear, and may involve
rather large strains to failure. For example, Figure 1 shows the tensile stress strain behavior of a
plain-weave vinyl ester material for different angles of loading relative to the principal material
directions, at a strain rate of 0.002/s.

FIGURE 1. Tensile behavior of woven glass reinforced vinyl ester laminates.

The reduction of such data to a single constitutive equation that includes large-deformation
“plasticity” effects has been discussed by several authors, including the present ones.[1-4] The
influence of high strain rate is less well understood; representations and understandings of those
effects out to strains of the order of 20 percent have not yet been achieved. However, strain rate
dependence is evident in such materials, even for strain rates that differ by only a few orders of
magnitude. Figure 2 shows an example of such dependence. We will discuss the concepts and
process involved in reducing such data to constitutive equations that can be used in commercial
computational codes such as ABAQUSTM to enable structural analysis in the presence of large-
strain progressive damage under dynamic loading.
322 L. Xing et al.

FIGURE 2. Tensile behavior of woven glass reinforced vinyl ester laminates pulled at different
strain rates, at 150 to the principal material direction.

For the present work, we are interested in not only in the constitutive equations for elastic-
plastic deformation, but we are also interested in the fracture behavior of the laminates, as a
function of strain rate and multi-axial stress / strain. Indeed, we have found that the fracture
patterns of the laminates we have tested over about five orders of magnitude of strain rate do
depend greatly on the angle of unaxial loading, and the strain rate. We have begun the process of
understanding those fracture mode changes, and have postulated failure concepts and metrics that
may provide a failure criterion that is essential for progressive failure analysis in the presence of
nonuniform stress / strain fields. An initial indication of the success of that effort is the possibility
that concepts such as variations on the classical Monkman-Grant idea that strain rate times the time
to failure may be a material constant will be discussed. Data for a wide range of loading rates and
loading angles will be presented. Discussion of secondary effects such as the dependence of
Poisson’s ratio on strain rate, and the possibility that Poisson’s ratio (and its rate dependence)
depends on the level of (nonlinear) strain will also be discussed.

References
1. Tamuzs, V., Dzelzitis, K. and Reifsnider, K.L., Applied Composite Materials, Vol. 11 No. 5,
259-279, 2004
2. Tamuzs, V., Dzelzitis, K. and Reifsnider, K.L., Applied Composite Materials, Vol. 11, No. 5,
281-293, 2004
3. Ogihara, S. and Reifsnider, K.L., Applied Composite Materials, Vol. 9, 249-263, 2002
4. Sun, C.T. and Chen, J.L., J. Composite Materials, 23, 1009-1020, 1989
 2T8. Polymers and composites 323

AGING AIRCRAFT TRANSPARENCIES: A CASE HISTORY FROM ITALIAN

AIR FORCE FLEET

Lt.Col. M. Bernabei, Lt. D. Caucci and Capt. L. Aiello

Italian Air Force Flight Test Center
Airport M. de Bernardi
rc.gms@tiscali.it

Aircraft acrylic transparencies are structural components that must withstand flight and ground
loads; like metal airframes, they are subject to aging with attendant reductions in structural
capabilities, generating warnings about aircraft safety as well as operational readiness [1].
Nevertheless residual stresses have been avoid during manufacturing, and maintenance operators
have been tasked to use only washing products which are not aggressive at all, one of the most
common causes of substitution of aircraft transparencies is still crazing appearance. This form of
aging, defined Environmental Stress Cracking (ESC) is a predominantly physical phenomenon
where an active liquid can cause craze formation at a much lower stress that would be required in
air [2]; crazing stress values are temperature dependant, correlated to both acrylics’ void and
entanglement density [3].
In this paper, we investigated an extensive phenomenon of network ESC occurred on
transparencies of many trainee ultrasonic aircraft operating in the same fleet, causing their
replacement well before the required service life.
All damages were placed only on cockpit double transparencies, saving the windshield;
furthermore, they were localized on central/right area, Fig. 1.

FIGURE 1. Environmental Stress Cracking localization

During on service maintenance, the required washing procedures, supposed to be not

aggressive at all, have been confirmed to have been applied. While long term no fly periods,
because flight instruments and controls may be severely affected either by heat coming from sun
irradiation and by water infiltration dew to condensed moisture, OEM required that the whole
cockpit should be protected by means of the application of a cover.
Acrylics’ conformance to MIL-PRF-5425 specification has been confirmed, by means of FTIR
spectrometry and GC-Py. Aged transparency tensile strength, executed out of crazing area, even
lower than standard specimens, has been confirmed to be above the required minimum value;
tensile strength inside the crazing area has been measured to be decreased (about 50%), confirming
safety concerns as reasons to promote parts replacement.
324 M. Bernabei et al.

Crazing appearance on aged transparencies has been tested by means of ASTM F 483; results
confirmed that phenomena are temperature dependant, Table 1, and showed that high temperatures
may be sufficient even if not associated to aggressive environments.

TABLE 1. Stress crazing stress on aged transparencies.

Cover application has been individuated to be the critical reason in crazing appearance.
Extracted solution by hot water did not show crazing promoting on standard acrylics. Thus cover
has been supposed to act as physical barrier for condensed water, not allowing night moisture
evaporation. Moreover, sun irradiation causes cover heating and subsequent underlocated acrylics
heating; thermal conductivity of acrylics and silicon-PVC/nylon (the multilayer material the cover
consist of) do not allow that the whole part share the same temperature, figure 3-4.

Figure3: thermography of internal side of transparency, cover applied

Between irradiated area and the surround, temperature variation may get over 30 °C; by
simulation tools such as FEM linear elastic analysis, it has been calculated that this range is enough
to promote surface stress in the range of 10 MPa, more or less the 20 % of the original tensile
strength value.
Parking area latitude (N 40°) and aircraft magnetic heading while parking (055° NW) is
coherent with the phenomena localization.

References
1. Bouchard M.P., Bowman D.R. and Whitney T.J., Proceedings of the Second Joint Conference
on Aging Aircraft, 1999, 91-100.
2. Arnold J.C., J. Mat. Sciences, vol. 33, 5193-5204, 1998.
3. Yamamoto T. and Furukawa H., Polymer, vol. 36, 2393-2396, 1995.
 2T8. Polymers and composites 325

FATIGUE CRACK GROWTH IN QUENCHED AMORPHOUS POLYMERS PC

AND PET

M. Kitagawa and D. Nishi

Dept. of Human & Mechanical Engineering Systems, Kanazawa University
2-40-20 Kodatsuno, Kanazawa 920-8667, Japan
masayosi@t.kanazawa-u.ac.jp

When tensile or shear test is performed in as-received(AR) or annealed samples of amorphous

polymers, yielding occurs after the initiation of numerous slip lines near the peak stress and the
subsequent formation and growth of a pair of visible shear bands. But in the case of ‘quenched’ (Q)
sample subjected to simple shear, yielding starts at the sudden initiation of vague shear bands
without visible slip lines. The stress softening just after yielding is steeper in AR sample than in Q
one. The plastic deformation mechanism seems to be very different between AR and Q samples.
To know this difference may lead to the intrinsic understanding of the yield mechanism of
amorphous polymers. In order to know the effect of quenching, mechanical and physical properties
such as strain rate and temperature dependence of yield stress, dynamic properties, density, DSC
and diffusion of kerosene have been measured. In this paper, effect of quenching on fatigue
crack growth behavior involved with plastic zone in front of fatigue crack tip, crack growth rates
and fracture surface was investigated using amorphous polymers PC and PET.
Materials and Experimental Procedure
The materials used were commercial transparent polycarbonate (PC; Takiron Co., Japan) and
poly(ethylene terephthalate)(PET; Takiron Co., Japan) plates 0.5mm in thickness. Rectangular
plates 100mmx50mm cut from these plates were preheated at 70 for 1 hour to evaporate moisture
and then were maintained in an oven slightly above the glass transition temperature Tg for 1 hour,
and were quickly immersed into a salt water kept at –10(quenching). Fatigue crack was introduced
in front of the notch cut with a filet saw at a sufficiently low stress amplitude. The specimens made
in this way were subjected to fatigue loading using a homemade machine and the crack growth
behavior was observed through a probe type optical microscope. The tests were performed at about
1Hz in room temperature(20±3). The quenched samples were tested within 12hours, during which
the effect of physical aging may be small.
Results and Discussions
The plastic zones ahead of the crack tip in AR and Q samples were observed through an optical
microscope. The plastic region in AR sample is associated with numerous long slip lines, while it
is covered by small craze like flaws in Q sample. This may suggest that the plastic deformation
mode in front of the crack tip is divided into two types, a type controlled by slip lines and a type
governed by craze like flaws. The length ratios of plastic zone (s) to crack (c) s/c were applied well
to Dugdale model in both AR and Q samples.
Fatigue crack growth rates (dc/dN) as function of stress intensity factor rangeK in AR and Q
samples in PC are shown in Fig.1 where many experimental data are excluded due to the limit of
the abstract length. The rates of Q samples are slightly smaller than that of AR samples at the
sameK. In the results of PET where the experimental data are not shown, the rates for both samples
nearly falls on the same curve.
Fatigue striations in both AR and Q samples of PC and PET were compared by SEM
observation. The observation showed that the striation shape was very different between AR and Q
samples. In AR samples, there were many tear lines running radially along the crack growth
326 M. Kitagawa and D. Nishi

direction (perpendicularly to the striations). But the tear lines were not observed in the Q sample.
For both PC and PET, the striation of AR was observed to rise gently to the fracture surface like a
hill, while it seemed to sink down sharply from the surface like a cliff.
These differences of fatigue crack growth behavior between AR and Q states in amorphous
polymers may be attributed to the difference in the plastic deformation mode ahead of the crack
tip.
 2T8. Polymers and composites 327

THERMO-MECHANICAL STATE OF BIMATERIAL WITH AN INTERFACE

CRACK

R. Martynyak, M. Matczynski1 and K. Honchar

Pidstryhach Institute for Applied Problems of Mechanics and Mathematics, NASU,
3–b Naukova St., Lviv, 79060, Ukraine
labmtd@iapmm.lviv.ua)
1Kielce University of Technology, Faculty of Management and Computer Modeling,

7 Al. Tysiaclecia Panstwa Polskiego, 27 314 Kielce, Poland

mmatcz@ippt.gov.pl

Temperature and stresses distributions around the crack in composite materials depend on the
properties of the heat-conducting crack. The heat conduction around the crack surfaces strongly
depends, besides other factors, on the properties of the thin layers covering the crack surfaces,
which are formed due to the different physical and chemical processes, and the properties of the
material filling the crack.
Keeping in mind these observations, we present the model of the interface crack, which allows
effectively studying the distribution of temperature and stresses around the cracks
We assume that the thickness h 0 of the layers along the crack surfaces and the crack opening
displacement h ( x ) are small in the comparison with the half-length of the crack . In

addition, it is assumed that the thermal conductivity coefficients O 0 of the layers along the crack
surfaces and the thermal conductivity coefficients of the material filling the crack are less than
thermal conductivity coefficients O1 and O2 of the components of the bimaterial

( O 0 , O c  O 1 an d O 0 , O c  O 2 ) .
Making use of these assumptions, we define the thermal resistance of the crack as

h(x) 2 h0
R ( x) r0  where r0 (1)
Oc O0
Thus, the thermal conditions along the crack surfaces will take the following form:

q y q y , T 
T 
q y R ( x )
,
   
where T , q y and T , q stay for the temperature and normal components of the heat flux
y

along the crack surfaces, respectively.

Taking into account Eq. (1) the thermoelastic problem of the bimaterial consisting of two
materials of different thermal properties is considered. The mechanical properties of the materials
are identical. The composite is weakened by an interface crack of length 2a. The body is subjected
to the external, uniformly distributed and perpendicular to the interface, tensile load p and heat flux
of a constant intensity q.
Making use of the Muschelishvili method of complex potentials, the problem is reduced to the
solution of singular integro–differential equations in terms of jumps of temperature
 
J (x) T (x)  T ( x ) and opening displacements h ( x ) along the crack surfaces
328 R. Martynyak et al.

J x O a
J c t  d t
R x

2S ³
a
tx
q, x  a,

1
a
h c t  d t O K  Kp
³  J x  , x  a,
S a
tx 2 2

where O 2 O1 O 2  O1  O2  ,K

K 2  K1 , K n D n 1  Q  O n ,  n 1, 2  ,

K 4 1  Q  G ,

while Q , G a n d D 1 , D 2 stay for Poisson's ratio, shear modulus and coefficients of linear
thermal expansion, respectively.
The following problems were considered:
1 The layers covering the crack surfaces do not exist or the thermal resistance of the layers is
much less than the thermal resistance of the material filling the crack ( r0 h ( x ) / O c )
what leads to the assumption that the thermal resistance of the crack is equal to the thermal
resistance of the crack filler,R ( x ) h( x) / Oc .
2 Thermal resistance of the crack filler is much less than the thermal resistance of layers
covering the crack surfaces ( h ( x ) / O c r 0 ) and R ( x ) r0 .

3 The layers covering the crack surfaces do not conduct the heat ( r0 f) what means
that the crack surfaces are thermally insulated and R ( x ) f .
The second and third case corresponds to the thermoelastic problem of the partially closed
crack. The numerical analysis shows that the increase of the heat flux leads to the increase of the
contact region of the crack surfaces. This region reaches its maximal length in the case of the
insulated crack surfaces and decreases with the increase of the thermal resistance of the crack. The
contact pressure of the crack surfaces reaches its maximal value in the case of the completely
insulated crack surfaces. In the case of the partially closed crack, the stress intensity factors are
nonlinear functions of the heat flux.
 2T8. Polymers and composites 329

THERMO-MECHANICAL STATE OF BIMATERIAL WITH AN INTERFACE

CRACK

M.Tarfaoui, S. Choukri1, A. Neme and M. Mliha-Touati1

ENSIETA, Laboratoire MSN, 2 rue F. Verny, 29806 Brest Cedex - France
tarfaomo@ensieta.fr, nemeal@ensieta.fr
1EMI, Département Génie Mécanique, Rabat - Maroc

choukri@emi.ac.ma, mliha-touati@emi.ac.ma

The composite materials with organic matrix found important applications in the shipbuilding
industries. A new applications of the composites are identified, including their current and
potential use in the superstructures, the platforms, the watertight bulkheads, the engines
(propellers), the axes of propulsion, the pipes, the pumps, the valves, the machines and any other
equipment on large men-of-war such as frigates, destroyers and aircraft carriers. Potential
applications of the composites to the submarines are also identified, like their possible use in the
engines, the control surfaces, the machines and the trimmings. Composite materials exhibit
excellent mechanical properties over metallic materials and hence are increasingly considered for
high technology applications.
In many practical situations, the structures are subjected to loading at very high strain rates like
slamming, impact or underwater explosions (table 1). Material and structural response vary
significantly under such loading as compared to static loading.
The purpose of this work is the use of Split Hopkinson Pressure Bar (SPHB) for the dynamic
characterization of the high strain rates effects in particular on the mechanical behavior, the
damage and the strength of a glass/epoxy composite, very much used in the naval domain. It is also
a question of understanding the microscopic mechanisms leading to the damage and the fracture of
material and of quantifying their evolution with the strain rate.
In the current investigation, the response of glass/epoxy laminated composites under high
strain rate compression loading is considered using SHPB setup at different strain rates of 100 s-1
to 4000 s-1. The laminates were fabricated using 40 plies of unidirectional glass/epoxy. The
samples were tested in the thickness as well as in-plan direction (figure 1). The fiber orientations of
simples are 0°, r45° and 90°. Dynamic stress-strain plot was obtained for each sample and
compared with the static compression test result.
330 M. Tarfaoui et al.

TABLE 1. Material properties.

Load Strain rates (s-1) Times to reach 100 MPa, (s)

Slamming 2.10-3 … 6.10-1 250.10-3

Collision 2 250.10-6

Impact 20 25.10-6

underwater explosions 50 … 100> 5.10-6

The tests show a strong sensitivity of material to the dynamic loading and the load direction. A
threshold effect of the dynamic rigidity of the laminated plates is observed for tests in the plan. The
failure and damage mechanisms are implicitly related to the rise in temperature during dynamic
compression. We will see in details in this report the various experimental results of the tests of
characterization of material followed by tests of dynamic compression with the system of Split
Hopkinson Pressure Bar. The last part of this work treats the numerical aspect modelling and
correlation with the experimental results. A good reproducibility is noted.

(a) (b)
FIGURE 1. Compression tests: (a) In-plane, (b) Out of plane.
 2T8. Polymers and composites 331

INDENTATION RESPONSE OF FIBRE REINFORCED COMPOSITE

LAMINATES

P. Bourke and I. Horsfall

Cranfield University [RMCS]
Shrivenam, Oxon, SN6 8LA, England
P.D.Bourke@Cranfield.ac.uk

Indentation is widely used as method of characterising materials. Although a significant body of

work exists on the use of indentation test on polymers, ceramics and particularly metals this
technique has not generally been used on long fibre or laminated composites.
In some applications such as ballistic armour the indentation response is directly linked to the
service conditions as the initial stages of projectile arrest will take the form of indentation of the
composite armour surface. Typically a high velocity (300-800ms-1) projectile will have two stages
of penetration into composite, in the initial stage a hole is produced of approximately the projectile
diameter with compression and shear failure taking place ahead of the projectile and only limited
damage occurring remote from the impact point. This is followed by a second stage in which the
projectile is caught in the laminate, gross delamination takes place around and in front of the
projectile and the stresses in the laminate ahead of the projectile are primarily tensile along the
reinforcement fibres.

FIGURE 1. Penetration Mechanisms

The initial stage of penetration has significant similarities with conventional indentation
testing as the stress system is mainly compressive around the indenter and properties such as
compressive yield strength become dominant. In order to more accurately model the impact of a
penetrator a relatively slim indenter could be used in order to promote radial flow or a blunt
indenter could be used to promote shear failure
In the current work a range of thick glass fibre reinforced polymer composites were subjected
to quasi static and dynamic indentation by ogival, conical and blunt indentors. The results of these
tests are used to derive basic mechanical properties and the response of the laminates under these
conditions is compared to ballistic impacts using similar shaped projectiles.
The equipment used will consist of an Instrumented Falling Weight (IFV) drop tower which
will be used to force the chosen indentors through the material to be tested at a range of velocities
and energies. This will be achieved by varying the drop mass and the height at which this mass is
released from.
Ballistic testing will be conducted in an indoor 25mr test range facility using a fixed test gun
mount and optical velocity measurement.
332 P. Bourke and I. Horsfall

FIGURE 2. IFW drop-tower indentor (Left) and Universal test gun mount (Right)
 2T8. Polymers and composites 333

ANALYSIS OF TUBULAR COMPOSITE CYLINDRICAL SHELLS

Reza Mohammadzadeh Gheshlaghi, Mohammad Hassan Hojjati1

and Hamid Reza Mohammadi Daniali2
Engineering Research Center, Tabriz, Iran
1,2Department of Mechanical Engineering, Mazandaran University, Babol, Iran

4th Floor, Ommat Building, Amin Cross, Tabriz, Iran

mohammadzadeh@noavar.com

Composite materials due to their small weight to strength ratio, have found extended use,
particularly in aerospace industries. Trident II space shuttle rocket casing and many other new
rocket systems are carbon fiber filament winding structures which have wonderful specific
strength.
Composites characteristics greatly depend on measurement direction. On of the best
advantages of composite materials is that they can be changed to achieve special design
requirements. This happens by aliening fibers in expected load directions. So extremely high
strength of fibers can carry load directly in appropriate direction.
In this paper, tubular composite cylindrical shells with different geometries are analyzed. The
aim of this research is to investigate the behavior of the previously mentioned cylindrical shells,
under different load aspect ratio and different winding angles. The problem was solved using the
classical theory of composite laminated shells and finite element methods.
A fiber glass filament wounded thin composite cylindrical shell is used for study. Vessel is
investigated under various meridians to axial load ratio. Also in each load condition filament
winding angle is being varied. Finite element model of composite cylindrical shell is presented in
Fig. 1.
Based on classic theory of composite laminated shells, on-axis and off-axis stresses and strains
of each ply is determined. A Tsai-Wu failure criterion is used as design limit. It is assumed that
layers are completely bounded to each other and there is no sliding between tow conjunction
layers.
Comparing the results of analytical method and finite element method shows the reliability of
procedure. Based on the results, by increasing the aspect ratio of (k), minimum of Tsai-Wu failure
criterion leads to the small winding angles. So the fibers align according to maximum stress and
make structures more efficient.
334 R. M. Gheshlaghi et al.

FIGURE 1. FE model of composite filament wounded cylindrical shell

References
1. ASM International Hand book Committee (1987), "Engineered Materials Hand Book.",
Volume 1 composites.
2. Eckold G.C (1985), "A Design Method for Filament Wound Grp Vessels and Pipework."
Composites, Volume 16, No. 1, January 1985.
3. Wilson J. F. (1986), "Linear Analysis of Uniformly Stressed Orthotropic Cylindrical Shells."
Journal of Applied Mechanics, Vol. 53, PP. 249-256.
4. Tew B. W. (1995), "Preliminary Design of Tubular Composite Structures Using Netting
Theory and Composite Degradation Factors." Journal of Pressure Vessel Technology, Vol.
117, PP. 390-394.
5. Xia M., Takayanagi H., Kemmochi k. (2001), "Analysis of Multi-Layered Filament-Wound
Composite Pipes under Internal Pressure." Composite Structures, 53, PP. 483-491.
6. Be'akou A. , Mohamed A. (2001) , "Influence Of Variable Scattering On The Optimum
Winding Angle Of Cylindrical Laminated Composites." , Composite Structures , 53 , PP. 287-
293.
7. Hoa S.V (1991),"Analysis for Design of Fiber Reinforced Plastic Vessels and Piping." U.S.A,
Technomic Publishing Co.
8. Sherrer R. E. (1967), "Filament Wound Cylinders with Axisymetric Loads.", Journal of
Composite Materials, Vol. 1, P. 344.
 2T8. Polymers and composites 335

ANALYSIS OF COMPOSITE PRESSURE VESSELS

Reza Mohammadzadeh Gheshlaghi, Mohammad Hassan Hojjati1 and Hamid Reza Mohammadi
Daniali2
1Engineering Research Centre, Tabriz, Iran
1,2Department of Mechanical Engineering, Mazandaran University, Babol, Iran

4th Floor, Ommat Building, Amin Cross, Tabriz, Iran

mohammadzadeh@noavar.com

Composite pressure vessels have found unique place in industrial, commercial, research and
defence applications due to their higher efficiency and good resistance under hard working
conditions. The composite gas cylinder is about 70% lighter than steel and about 30 to 50 percent
lighter than aluminium lined tanks. According to unique capabilities of composite pressure vessels,
especially significant weight saving and special application various research and studies were done
on this kind of pressure vessels.
Geodesic dome contours were generated using netting analysis procedures. The geodesic shape
is considered by some analysts to be the optimum deformed shape of the dome. Large deflection
finite element analysis is used to identify the original shape, which will deform to the geodesic
shape under internal pressure and nozzle thrust loads. In this paper, a simple iterative technique for
generating the geodesic dome contour for internal pressure loading is provided.
In this paper the mechanical behaviour of three kinds of composite pressure vessels are
investigated. These vessels include the pressure vessel with spherical end cap, geodesic with
constant thickness and geodesic with variable thickness end cap.
In this research, netting theory is utilized to develop the model at preliminary design step. The
profile of geodesic end cap was calculated and then modelled in ANSYS FEM software
environment as is presented in Fig. 1. This procedure also repeated for two other vessels. To
investigate the mechanical behaviour, three different analytical and numerical methods including
preliminary design with netting theory, analysis with classic theory of laminated shells and
analysis with finite element method are employed.
According the results of this study for cylindrical section of vessels finite element method and
classical theory of laminated shells shows a close compatibility. Evaluation using Tsai-Wu and
maximum stress failure criteria shows that netting theory is not a reliable design method for
composite pressure vessels. Using maximum stress design criterion trends to over design
structures. According to design criteria and results of analysis, the pressure vessel with geodesic
and variable end-cap shows better mechanical behaviour in comparison whit other two kinds of
pressure vessels.
336 R. M. Gheshlaghi et al.

FIGURE 1. FE model of composite pressure vessel with geodesic end-cap.

References
1. Prasad R.C, Ramakrishnam P. (2000), "Composites Science and Technology.” New Delhi,
New Age International (P) Ltd.
2. Tew B. W. (1995), "Preliminary Design of Tubular Composite Structures Using Netting
Theory and Composite Degradation Factors.” Journal of Pressure Vessel Technology, Vol.
117, PP. 390-394.
3. Z.Kabir Mohammad (2000),"Finite Element Analysis of Composite Pressure Vessels with a
Load Sharing Metallic Liner.” Composite Structures, 49, PP. 247-255.
4. Park Jae-Sung, Hong Chang-Sun (2002), "Analysis Of Filament Wound Composite Structures
Considering The Change Of Winding Angles Through The Thickness Direction.” Composite
Structures, 55, PP. 63-71.
5. Peters S. T., Hamphrey W. D., Ford R. F. (1990), "Filament Winding Composite Structures
Fabrication.” 2nd Edition, SAMPE Publishers.
6. Puck A. , Schneider W. (1969) , "On Failure Mechanisms And Failure Criteria Of Filament-
Wound Glass-Fiber/Resin Composites." , Plastic 7 Polymers , February 1969 , PP. 33-34.
7. Withney J. M. (1971), "On the Use of Shell Theory for Determining Stress in Composite
Cylinders.” Composite Materials, Vol. 5, P. 340.
 2T8. Polymers and composites 337

GRADIENTS INFLUENCE ON DAMAGE AND CRACKING IN CRYSTALLINE

POLYMERS

S. Castagnet and J.-C. Grandidier

Laboratoire de Mecanique et Physique des Materiaux (UMR CNRS 6617)
ENSMA , 1 avenue Clement Ader, BP 40109, 86961 Futuroscope cedex
castagnet@lmpm.ensma.fr, grandidier@lmpm.ensma.fr

Many polymers used in structural applications (vessels, pipes, etc…) are crystalline thermoplastics
like polyethylene (PE), polypropylene (PP), polyvinylidene fluoride (PVDF). These biphasic
materials are made of crystalline lamellae bonded to an amorphous phase. When highmy
crystalline polymers like mentioned above are strained at a temperature higher than the glassy
transition temperature (Tg), the amorphous phase, softer than crystals, is very sensitive to
cavitation. Depending on thermal and mechanical loading conditions, cavities nucleation occurs in
a generalized or localized manner and lead to macroscopic plastic strain and ductile fracture or to a
slow crack growth regime. The latter regime is crucial since industrial installations must be
certified over decades. Mechanisms of damage and fracture in this slow crack growth (SCG)
regime are expected to depend both on micro-scale gradients (resulting from processing and
further thermal annealing) and on macroscopic gradients induced by structure shape and boundary
conditions.
The aim of the paper is to separate the influence of mechanical gradients at macro and micro
scales on damage. The main part of the work involves a commercial PVDF. Its glassy transition
temperature is –40°C and the melting temperature 170°C. The cristallinity ratio is 55%. The
microstructure effect will result from the evolution induced by a long-term high temperature
annealing (thirty degrees below the melting temperature).
The first part of the study is performed in a low triaxiality framework considering dumbbell
specimens in tension and thin pipes under constant pressure. Post-mortem observations of damage
were made in a pipe tested under constant low pressure at high temperature. As shown by
Gacougnolle and Castagnet [1] and depicted in Fig. 1, a very localized damage nucleation was
observed below the inner surface of the pipe further leading to a small crack slowly propagating in
the pipe wall. Nevertheless, a cavitation onset in the pipe wall is contradictory to the lack of
cavitation measured by Castagnet et al [2] in dumbbell specimens of the same material tested at the
same temperature and exhibiting a very close triaxiality ratio.
Finite elements calculations were performed on perfectly axi symmetrical homogeneous pipes
to simulate the long-term creep behavior of homogeneous pipes and to focus on the structure effect
in the nucleation process of cavities. In the stress range of the SCG regime, stress and strain
distributions are evaluated just after loading and their evolution with time is considered. Whereas
the macroscopic structural effect does not seem determinant to explain the onset of cavitation in a
perfectly axi-symmetrical pipe, thickness defects due to extrusion appear very detrimental. The
long lasting localization of the hydrostatic stress and the hoop stress below the inner surface are
expected to favor cavity nucleation and further crack propagation. The microstructure evolution
resulting from thermal annealing has been characterized from several techniques. After annealing,
the material seems to exhibit a more homogeneous, less defective and / or thicker crystalline
lamellae population. It means that the local triaxiality ratio in the amorphous phase would be
raised. Measurements of volume strain during tensile tests in dumbbell specimens indicate a
stronger cavitation in annealed specimens. Such a result confirms the major part of micro-scale
gradients in damage nucleation in case of a limited macroscopic triaxiality
338 S. Castagnet and J.-C. Grandidier

FIGURE 1. Post-mortem observation of the damage zone ahead of a crack tip formed in a pipe
tested in the SCG regime

The consequences of microstructure changes on the propagation stage are also studied. Creep
tests were performed on three-notched samples. Such tests, very close to that developed for PE by
Brown et al. [3] are classically used to investigate the slow crack growth resistance of
thermoplastics. The curvature radius of the crack tip and the morphology of the damaged zone
ahead of the crack tip were examined. It is observed that the crack opening is reduced and that
cavities coalescence is easier after thermal annealing.
A second part of the study is currently performed in a higher triaxility framework. Samples
notched with different curvature radii are tested in tension. As-received and annealed materials are
tested to compare the respective part of microstructure affect and macroscopic gradients on
cavitation. The latter is expected to be predominant here.

References
1. Gacougnolle, J.L., Castagnet, S. and Werth, M., to be published in Engineering Failure
Analysis.
2. Castagnet, S., Girault, S., Gacougnolle, J.L and Dang, P.,.Polymer, vol. 41, 7523-7530, 2000.
3. Lu, X. and Brown, N., Polymer Testing, vol. 11, 309-319, 1992.
 2T8. Polymers and composites 339

AN ELASTO-PLASTIC SHEAR-LAG MODEL FOR SINGLE FIBER

COMPOSITE

Souta Kimura, Jun Koyanagi1 and Hiroyuki Kawada1

Graduate School of Waseda University
1Department of Mechanical and Engineering, Waseda University

3-4-1, Okubo, Shinjuku, Tokyo, 169-8555, Japan

imagine-128@fuji.waseda.jp

One of the most principal problems in the fibre-reinforced plastics (FRP) lies on the control of the
adhesion between the matrix and fibre. Recently, an energy release rate to initiate an interfacial
debonding in a fragmentation test (interfacial energy) is suggested to be valid for the evaluation of
the interfacial adhesion (Wagner et al. [1]). However, the value of the interfacial energy varies
along with a stress distribution obtained from a shear-lag model. Therefore, the accuracy of the
shear-lag model is necessary for the quantitative evaluation of the interfacial adhesion. As
presented in many researches, the plastic deformation around the interfacial debonding is not
negligible, so it is necessary to apply an elasto-plastic shear-lag approach (Landis and Mcmeeking
[2], Morais [3], Okabe and Takeda [4]).
The technique of laser Raman spectroscopy (LRS) is effective to measure a distribution of the
fibre axial stress or strain (Galiotis [5]). In this research, the distributions of the fibre axial strain
derived from some different shear-lag models are compared with the experimental results obtained
from LRS, and an effect of the difference of the shear-lag model on the value of the interfacial
energy is discussed. The fragmentation tests with a model of carbon fibre-reinforced epoxy matrix
composite were performed. The geometry of a test specimen is shown in Fig.1, and the mechanical
properties of the fibre and matrix are listed in Table1.

FIGURE 1. Size and dimension of specimen

TABLE 1. Material properties.

340 S. Kimura et al.

Fragmentation tests were performed with a compact-tensile machine, monitored by Micro

Raman Spectroscopy (MRS). We measured a fibre Raman frequency in the vicinity of a fibre-
break point when a fibre break occurred, and the Raman frequency was converted into a
distribution of the fibre axial strain by means of the following equation.

O  20 . 43 ˜ H  2694
f
(1)

where O [cm-1]and Hf[%] are fiber Raman frequency and fiber axial strain respectively.
In our previous research (KIMURA et al. [6]), we derived a distribution of the fibre axial strain
assuming that shear stress prevails in the matrix and the interfacial shear stress in the plastic zone
is equal to the matrix shear yielding stress W ym (matrix perfect plasticity). The comparison of the
theoretical model with the experimental result against a distribution of the fibre axial strain is
presented in Fig.2.

FIGURE 2. Comparison of a distribution of the fibre axial strain between Theory and MRS

According to Fig.2, the theoretical distribution shows a good agreement with the experimental.
Therefore, our shear-lag model gives a sufficiently accurate stress distribution, and the value of the
interfacial energy was calculated as 61.8[J/m2].

References
1. Zhou X. F., Nairn J. A. and Wagner H. D., Composites: Part A, vol. 30, 1387-1400 1999
2. Landis C. M. and Mcmeeking R. M., Comp. Sci. Tech., vol. 59, 447-457 1999
3. Morais A. B., Comp. Sci. Tech., vol. 61, 1571-1580 2001
4. Okabe T. and Takeda N., Comp. Sci. Tech., vol. 61, 1789-1800 2001
5. Galiotis C., Comp. Sci. Tech., vol. 48, 15-28 1993
6. Kimura S., Koyanagi J. and KAWADA H., Int. J. JSME, accepted
 2T8. Polymers and composites 341

PROGRESSIVE FAILURE OF COMPOSITE LAMINATES; ANALYSIS VS

EXPERIMENTS

Veera Skytta, Olli Saarela and Markus Wallin

Helsinki University of Technology, Laboratory of Lightweight Structures
P.O. Box 4300, 02015 TKK
veera.skytta@tkk.fi

The failure of composite structures divides into three stages; damage initiation, damage
progression and final failure. A typical mechanism of damage initiation is matrix cracking in some
layer of the laminate. Consequently, damage initiation is often referred to as the first ply failure
(FPF). Damage progresses with several mechanisms that often occur interactively. Most important
of these are additional matrix cracking and crack propagation, debonding of fibres and the matrix,
and delamination of layers. The final failure typically takes place when the ultimate load carrying
capability of the fibres is reached in some layer.
The first ply failure (FPF) of composite laminates is commonly predicted with layerwise
failure criteria. In a simple loading condition the failure criteria based on maximum stress and
maximum strain may be adequate. Typically, however, the stress state in an embedded ply is
multidirectional and the so-called interactive failure criteria perform better. Even these criteria
work well for only particular loading cases. Therefore, it is often necessary to use different failure
criteria to detect different failure modes as was done e.g. by Shokrieh & Lessard [1].
Since multidirectional laminates can carry considerable loads after the FPF, failure models
extending the analysis beyond the initial failure are needed. A common approach is to model a
degrading laminate with layerwise failure criteria and degradation rules. Such models are called
progressive failure models. According to Fig. 1, the FPF stress is first computed by applying the
classical lamination theory (CLT) and an applicable failure criterion. The material properties of the
damaged layers are then revised according to the degradation rules. The failure stress and failure
mode of the modified laminate are next determined. The process is continued to the final failure of
the laminate. A progressive failure model can thus estimate the ultimate failure load of a laminate.
It can also be used with finite element analysis to estimate the damage progression and strength of
a structure.

FIGURE 1. Progressive failure analysis of a laminate

342 V. Skytta et al.

The degradation rules are often based on knockdown factors with which the affected material
properties are decreased when the failure is detected in a layer. The simplest degradation model is
called the ply discount model. It uses a knockdown factor close to zero which means that the
affected stiffnesses of a failed ply are reduced to zero and the load is transferred accordingly to the
remaining undamaged plies.
The paper discusses different failure criteria and degradation rules and their use in the
progressive failure model. Numerical results are given for both glass fibre and carbon fibre
laminates. They have been calculated with the commercial finite element software ABAQUS. The
progressive failure has been added to the model with the use of a user subroutine and material
dependencies. The numerical results are compared with the experimental results.
The experiments were performed for glass fibre and carbon fibre prepreg laminates made of
unidirectional plies. The resin for the glass fibre laminates was chosen such that the manufactured
laminates were transparent. Thus, the damages in the laminate could be detected visually. The
carbon fibre laminates were of the type AS4/3501-6. Damage accumulation was measured with
acoustic emission.
Tests were first performed for unidirectional laminates to determine the ply properties. The
results were in good agreement with the data found from the literature. These material properties
were used in the progressive failure analyses of laminates.
A reasonably large test series was performed to investigate how the damage progresses in
laminates with different lay-ups and in different loading conditions. Some notched laminates were
also tested. The laminates were analysed with the progressive failure model by applying different
failure criteria and different degradation rules. The analysis and experimental results were in a
reasonable agreement with each other though the effects of the applied failure criterion and
degradation rules on analysis results were evident.
Worth of noting is that the progressive analysis approach used cannot take into account the
effect of the stacking sequence. In other words, the analysis gives only one result for laminates
composed of the same layers with the same layer orientations but with different stacking
sequences. This is a serious limitation of the model. Test results showed, for example, that
changing the stacking sequence for carbon fibre laminates may decrease the laminate strength by
over forty percent. The corresponding figure measured for the glass fibre laminates was
approximately 25%. The reduction is due to the well-known fact that the laminates with thick
stacks of unidirectional layers are prone to cracking.

References
1. Shokrieh M., Lessard L., “Progressive fatigue damage modeling of composite materials, Part
I: Modeling” J. Comp. Mat., vol. 34, no. 13, 2000
 2T8. Polymers and composites 343

A TEMPERATURE DEPENDENT VISCOELASTIC-DAMAGE MODEL FOR

CERAMICS FAILURE

Vassilis P. Panoskaltsis, Lynn M. Powers and Dario A. Gasparini

Department of Civil Engineering, CASE University, USA
vpp@nestor.cwru.edu

Ceramics have many important applications one of them being in gas turbine engines, where they
operate under high temperatures (about 1300 ºC). The mechanical behavior of ceramic materials
under high temperature is complex with the following main characteristics: nonlinear rate-
dependent behavior and asymmetric behavior in tension and compression.
The authors have recently developed [1] a nonlinear, multi-dimensional, viscoelastic model for
the description and prediction of ceramics behavior. In the present work, this model is further
extended in several ways: The new model is fully nonlinear in the sense that nonlinearity (which
has been earlier introduced in the evolution equation for the volumetric strain component only) is,
in addition, now introduced in the evolution equations for the deviatoric strains.
The introduction of nonlinearity into the evolution equation for the volumetric strain is
motivated by Eyring’s [2] rate process theory, according to which the inelastic strain is
proportional to the hyperbolic sine of the applied stress. The model is given a rheological
representation and is also expressed in an internal variable formulation.
Next, the viscoelastic model is further enhanced in order to be able to describe the material’s
damage characteristics. Damage is assumed to be isotropic, and therefore is described by a scalar
variable, which is treated as an additional internal variable. One way by which damage is
manifested is through the reduction of the material’s elastic stiffness. Taking this into account, the
introduction of damage in the model is accomplished by modifying the stiffness of the external
spring in the standard linear solid and by providing a suitable evolution equation for the damage
internal variable.
In order to include the temperature dependence of the material behavior the creep function is
modified according to the so-called time-temperature superposition principle. In an alternative
way, the temperature dependence is introduced in the internal variable formulation of the model in
a novel way recently introduced by the first author of this work.
The nonlinear evolution equations were numerically integrated by using the unconditionally
stable backward Euler method and the resulting nonlinear equations were solved by Newton-
Raphson method. The model is implemented into the ABAQUS [3] finite element environment.
The user material law (UMAT) subroutine was used to input the constitutive model. The
subroutine computes the algorithmic (consistent) moduli of the constitutive model as well as
evaluates the stresses and internal variables at the end of each time increment. The algorithmic
(consistent) moduli are used, instead of the continuous, in order that the quadratic rate of
convergence of Newton-Raphson’s method for the solution of the global equilibrium equations is
preserved [4], [5]. Moreover, in the case of viscoelasticity the continuous moduli are not defined
[6]. The algorithmic moduli are defined as the Gateaux derivative of the stress tensor with respect
to the strain tensor, at each time step, after convergence has been obtained.
A detailed procedure for parameter estimation is presented. Parameter estimation is
accomplished by using a Levenberg-Marquardt optimization algorithm implemented in IMSL
libraries. The model parameters are determined from uniaxial tensile and compressive creep
experiments on silicon nitride.
344 V. P. Panoskaltsis et al.

The derived model is tested against four-point bending and ball-on-ring experimental results.
The simulations were carried out by using the values for the material parameters already obtained
from the uniaxial tension and expression experiments. If an asymmetric material model is used for
the beam under four-point bending the neutral axis of the beam shifts; the neutral axis shift has
been observed experimentally [7]. A comparison between the maximum deflection predicted from
the finite element analysis and the measured deflection is favorable. The ball-on-ring specimen is
a disk and the load is applied to a ball at the center on top of the disk. The deflection is measured
at the center of the disk. Again, the agreement between the predicted and the measured
displacements is good.

References
1. Powers, L.M., Panoskaltsis, V.P. and Gasparini, D.A., International Journal of Non-Linear
Mechanics, to appear in 2006, available on line at www.sciencedirect.com.
2. Eyring, H., Journal of Chemical Physics, Vol. 4, 283-290, 1936.
3. ABAQUS. User’s Manual, Hibbitt, Karlsson and Sorensen, Inc., USA, 2003.
4. Nagtegaal, J.C., Computer Methods in Applied Mechanics and Engineering, Vol. 33 [1-3]
469-484, 1982.
5. Simo, J.C., and Taylor, R.L., Computer Methods in Applied Mechanics and Engineering, Vol.
48 [1], 101-118, 1985.
6. Panoskaltsis, V.P., and Lubliner, J., in Anisotropy and Localization of Plastic
7. Deformation, A. Khan and J.-P. Boehler eds., Elsevier, 1991, 651-654.
8. Fields, B.A., and Wiederhorn, S.M., J. Amer. Society, Vol. 79 [4], 977-986, 1996.
 2T11. Fracture mechanics analysis 345

AVALANCHE MECHANICS: LEFM VS. GRADIENT MODEL

A. Konstantinidis, N. Pugno1, P. Cornetti12 and E.C. Aifantis2

Laboratory of Mechanics and Materials, Polytechnic School,
Aristotle University of Thessaloniki, 54124, Thessaloniki, Greece
akonsta@gen.auth.gr
1Department of Structural Engineering, Politecnico di Torino,

Corso Duca Degli Abruzzi 24, 10129 Torino, Italy

2Laboratory of Mechanics and Materials, Polytechnic School,

Aristotle University of Thessaloniki, 54124, Thessaloniki, Greece

and
Center for the Mechanics of Material Instabilities and Manufacturing Processes,
Michigan Technological University, Houghton, MI 49931, USA

In this study the initiation of avalanches by failure of the interface between the show slab and the
underlying bedrock is considered. A pre-existing shear crack is assumed at the interface, in
analogy with the classical work of Palmer and Rice [1] on shear bands in over-consolidated clay. A
criterion for failure by shear band propagation was obtained and it was applied to snow slab
avalanches by McClung [2]. According to this criterion, for a given load due to the weight of the
slope above the plane of shear, the slope starts to slide when the length of the shear band exceeds a
critical value. Analogous criteria have been obtained in the work of Fyffe et al. [3] and Zaiser et al.
[4], where, in addition, variations in interface toughness due to the presence of small-scale
heterogeneities were considered. Similar expressions are also derived in the present study. A
Linear Elastic Fracture Mechanics (LEFM) approach (Pugno and Carpinteri [5]; Chiaia et al. [6])
is first used in order for a critical value of the energy release rate G for avalanche initiation to be
calculated; this provides a critical value for the height of the fallen snow. A gradient model, similar
to the ones that have been used successfully in the past for problems of shear banding in metal
plasticity (Aifantis [7], Zbib and Aifantis [8]), is then utilized providing expressions for the energy
release rate and the height of the fallen snow. The results of both models are compared. Although
these simple one-dimensional models have a completely different origin, it is shown that their
predictions in terms of the critical height for the fallen snow and the critical slope angle are very
similar.
The problem considered herein is the initiation of an avalanche in a snow slab of height h,
length l and width w adhering with shear stresses IJ to an oblique substrate forming an angle ș
with respect to the horizontal plane, as shown in Fig. 1. It is assumed that the snow has a density
ȡ in a gravity field of acceleration g, so that an axial force N ( x ) is present at the generic cross-
section x.

FIGURE 1. Geometry of the problem

346 A. Konstantinidis et al.

Using the Linear Elastic Fracture Mechanics model, as well as the gradient one, the critical
nominal tangential stress generalized by including a Coulomb-like friction term is calculated as

2EG Ch 4G ChE c
IJ CL E F M  µ cȡ g h c o s ș IJ gCra d  µ cȡ g h c o s ș
a , a (1)

where a is the initial crack length, G C is the critical value of the energy release rate, µ c is the

friction coefficient at the interface, E is the Young’s modulus and E c is the compression modulus.
The aforementioned models lead to similar expressions for the height of the fallen snow for
avalanche triggering, i.e.

2G C E
H gCrad | H CL E F M 2
ȡ g c o s ș  s in ș  µ c c o s ș  a 2
2 2
(2)
For both models, when friction is taken into account, the critical height of the fallen snow

H C has a minimum value for ș  

arctan 0.5 3µ c  9 µ c2  8  (critical slope) and two
vertical asymptotes for ș a r c ta n µ c and ș 9 0 $ , both reasonable. Without friction, the
critical slope (i.e. the slope for which avalanche triggering is easier) is about 54 $ ; otherwise it is a
bit higher and depends on the value of the friction coefficient µ c .

References
1. Palmer, A.C. and Rice, J.R., Proc. R. Soc. London, Ser. A 332, 527-548, 1973.
2. McClung, D.M., J. Geophys. Res., vol. 84, 3519-3526, 1979.
3. Fyffe, B., Zaiser, M. and Aifantis, E.C., J. Mech. Behav. Materials, vol. 15, 185-202, 2004.
4. Zaiser, M., Fyffe, B., Moretti, P., Konstantinidis, A. and Aifantis, E.C., In Proceedings of the
2nd Int. Symp. on Continuous and Discontinuous Modelling of Cohesive Frictional Materials
(CDM2004), edited by P.A. Vermeer et al., A.A. Balkema Publishers, 2004, 435-446.
5. Pugno, N. and Carpinteri, A., J. Applied Mechanics, vol. 70, 832-839, 2003.
6. Chiaia, B., Cornetti, P., Frigo, B. and Luisi, A., In Proceedings of the XVII AIMETA meeting
(Italian Association of Theoretical and Applied Mechanics), 11-15 September 2005.
7. Aifantis, E.C., Int. J. Plasticity, vol. 3, 211-247, 1987.
8. Zbib, H.M. and Aifantis, E.C., Scripta Metall. Mater., vol. 22, 703-708, 1988.
 2T11. Fracture mechanics analysis 347

INFLUENCE OF AUSTEMPERING ON FRACTURE MECHANICS

PARAMETERS OF 65 SI 7 STEEL

Dragan Pustai, Franjo Cajner and Martina Lovreni

Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb,
Ivana Luia 5, 10000 Zagreb, Croatia
dragan.pustaic@fsb.hr, franjo.cajner@fsb.hr and martina.lovrenic@fsb.hr

The aim of the performed investigations was to determine the influence of two different heat
treatment procedures, i. e. hardening and tempering vs. austempering, on some mechanical and
fracture properties of a particular type of steel. An advantage of austempering over hardening and
tempering is in obtaining the bainite microstructure which has greater toughness, strain,
contraction, fatigue strength and a better fracture toughness than a tempered martensite of the same
type of steel [2,3]. All these advantages of bainite refer to untempered state. By tempering of
bainite microstructure steel its toughness decreases and in tempered bainite it is lower than the
toughness of temepered martensite [3].
All investigations were performed on 65Si7 steel (chemical composition: 0.69% C, 1.56% Si
and 0.90% Mn). The specimens were first submitted to heat treatment by the parameters previously
determined by our own investigations >2,3]. After heat treatment, photographs of microstructures
of the starting conditions (prior to tempering) were taken. From the photos it is clearly seen that the
specimens in the starting conditions had two different microstructures:
• by cooling down in oil the obtained microstructure was a hardened microstructure with
visible plates of martensite common to the steel with this share of carbon, and its hardness
was 850 HV1,
• by austempering the obtained microstructure was a microstructure of lower bainite metal
base, with probably lower share of residual austenite and with hardness of 440 HV1.
These investigations comprise two groups of experiments, namely the static tensile tests (2
specimens) and tests to investigate the fracture toughness of the material (3 specimens). A
laboratory testing of fracture toughness of steel was carried out and diagrams of force
relationships - CMOD, CTOD-'a and J-integral-'a were recorded. A standard three points
bending test specimen (SENB) was investigated. All investigations of the fracture mechanics
parameters were performed according to the British standard BS 7448: 1991 – Fracture mechanics
toughness tests: Part 1. Method for determination of KIc , critical CTOD and critical J values of
metallic materials. Crack mouth opening displacement (CMOD) in dependence of the load on the
specimen (force F) is measured. The measured values are recorded in the form of a diagram F-
CMOD, Fig.1. Also, the diagrams CTOD-'a and J-integral-'a were recorded. After the critical
value of notch opening displacement Vc has been determined, it is recalculated into the critical
value of the CTOD parameter in the crack tip, i.e. in Gc.
Elastic-plastic analysis of FE models is performed using the commercial FE package
ABAQUS. For non-linear analysis, the Ramberg-Osgood type stress-strain relationship is used

n
H V § V ·
D ˜ ¨¨ ¸¸
V 0 E V0 ©V0 ¹ (1)

where D and n are material constant and strain hardening exponent, respectively.
348 D. Pustai et al.

FIGURE 1. Dependence of crack mouth opening displacement CMOD on the force F, a) for
hardened and tempered specimens, b) for austempered specimens, untempered and tempered

Austempering without tempering gives significantly higher ductility in comparison to the most
favourable case of hardened and tempered state. Austempering with a subsequent tempering at the
temperature of -t = 480oC contributes to a more significant decrease in tensile strength Vm, for
about 20%. The analysis of the critical value of the stress intensity factor KIc shows that with
higher temperature of tempering -t, higher values of the parameter KIc in hardened and tempered
specimens are obtained. For example, parameter KIc is approximately twice as high at the
temperature of tempering -t = 480oC than the one at the temperature of tempering -t = 300oC. The
best critical values of the stress intensity factor KIc are obtained by austempering without
tempering and they are approximately 50% higher than the ones obtained by tempering at the
temperature -t = 480oC, or even three times higher than the values obtained by hardening and
tempering at the temperature -t = 300oC. By hardening and tempering at the temperature -t =
480oC, the obtained critical values of crack tip opening displacement Gc are approximately 3.5 to
even eight times higher than those obtained by hardening and tempering at the temperature -t =
300oC. By austempering without tempering, the obtained critical value of crack tip opening
displacement Gc is about seven times higher than the value obtained by hardening and tempering at
the temperature -t = 300oC.

References
1. Pustai, D. and Cajner, F., Inžynieria materialowa, vol. 5 (124), 733-736, 2001.
2. Cajner, F., Strojarstvo, J. Theo. Apl. Mech. Engng., vol. 33 (5/6), 289-296, 1991.
3. Cajner, F., Kovine, zlitine, tehnologije, vol. 28 (6), 533-538, 1994.
 2T11. Fracture mechanics analysis 349

MODELLING THE EVOLUTION OF ELASTIC SYMMETRIES OF GROWING

MIXED-MODE CRACKS

H. Schutte and K. M. Abbasi

Institute of Mechanics, Ruhr University of Bochum
D-44780 Bochum, Germany
henning.schuette@rub.de
kian@tm.bi.rub.de

A numerical study of growing elliptical cracks in a unit cube is undertaken with the help of an
FEM simulation.

FIGURE 1. Growth of a crack in a unit cell.

The propagation of the microcrack is governed by the principle of maximum driving force [1],
which is a direct consequence of the variational principle of a body containing a crack. According
to this criterion, a crack grows in the direction of maximal driving force
2
1 Q 1
G* m ax di [( K I ) 2  ( K I I ) 2  ( K I I I ) 2 ]
I E 1Q
where

K D FDE (I ) K E

and FDE is a matrix whose elements are universal functions of the kinking angle f [4] and K E
are the stress intensity factors (SIFs) prior to the crack growth. The SIFs are computed with the
help of an stress extrapolation method without the use of singular elements.
For each propagation step the tensor of elasticity is calculated and analyzed with the help of its
spectral decomposition and its eigenvalues [3]. For the original elliptical crack its is known that it
results in a orthogonal symmetry with respect to the crack plane an the axes of the ellipse. After a
considerable amount of growth the crack approaches a penny shape perpendicular to the axis of
tension, which leads to transversally symmetry. For the other propagation steps the best orthogonal
approximation of the elasticity is determined.
Many damage mechanics models for quasi-brittle materials are based on the reduction of
stiffness due to elliptical crack or penny-shaped microcracks in the material. To incorporate the
results of the numerical analysis in a continuum damage model it is possible to avoid the
computational expansive FEM simulation of crack growth in a unit cell, representing a continuum
point, by using an replacement crack approach. For an elliptical crack there are analytical solutions
350 H. Schutte and K. M. Abbasi

available for the SIFs and so for the direction of growth and change of shape of the crack. Also the
evolution of the additional (damaged) compliance can directly be computed from the dissipation of
the crack as

w2 *
³ G a d s
S
d
*
wı … wı

For the next growing steps there exists no analytical solution for the SIFs at multiply kinked
crack fronts, so that in each step the grown crack is replaced by an elliptical crack, which has
certain properties it in common. For thermodynamic consistency it is necessary that the dissipation
of the kinking crack and the change of the potential of the growing crack are identical

<
crack
³ G * a d s ( a , b , D ,\ , T )
*

where the potential depends on the two half-axes of the ellipse and a set of three orientation angles.
To keep the physical meaning of the local crack growth rate, the rate of area increase has to be the
same for both cracks. As these postulates do not define uniquely a replacement ellipse, from this
set the one nearest to the real propagated crack front is taken.
A comparison between the evolution of the elastic modulus in the direction of tension of an
inclined crack from a FEM simulation and the replacement crack model is shown below

References:
1. Le, K. C., Schütte H. and Stumpf H., Archive of Applied Mechanics, vol. 69(1), 337-344,
1999
2. Mehrabadi, M. M. and Cowin, S. C., Mech. Appl. Math., vol. 43(1), 15-41, 1990
3. Leblond, J. B., Int. J. Solids Structures, vol. 25(11), 1311-1325, 1989
4. Amestoy, M, Leblond, J. B., Int. J. Solids Structures, vol. 29(4), 465-501, 1992
 2T11. Fracture mechanics analysis 351

EFFECT OF AGING ON THE MICROSTRUCTURE AND FRACTURE OF

ALUMINUM-LITHIUM

J. M. Fragomeni
Engineering and Science Consulting Services,
PO Box 1446, Royal Oak, Michigan 48068-1446, USA
jamesfrag@yahoo.com

The effect of microstructure as a consequence of precipitation aging on the fracture behavior,

deformation mechanisms, mechanical properties, and microstructures of aluminum-lithium was
studied. The alloy studied was an aluminum-lithium-zirconium alloy. The precipitation response
with aging time and temperature was studies in order to correlate the deformation response to the
alloys to the heat treating, microstructure, and fracture surface characteristics and features. The
primary focus of this study was to relate the variation in ductility with aging to the microstructural
parameters and fracture mechanisms. An aluminum alloy containing 2.6wt.%Li and 0.09wt.% Zr
exhibited very low tensile ductility consistently prior to the peak-aged strength independent of
thermal treatment. A transition was characterized by very low ductility in the slightly underaged
condition up to the near peak-aged condition, then followed by a substantial increase in ductility
with aging after the peak-aged treatment. Based on the quantitative microscopy of the size of the
precipitates, it was proposed that the increase in the ductility of the alloy after aging was a
consequence of particle coarsening with aging and resulting in Orowan looping due to the
transition from dislocation particle shearing to dislocation particle bypassing with increasing
precipitate size. As the interparticle spacing increased with overaging, and the dislocations were
impeded by and thus bypassed the larger particles, the amount of plastic deformation increased as
was reflected by the strength and ductility experimental data.
As seen from Fig. 1, the ductility reaches a minimum before the maximum peak-aged strength
condition is obtained for the 185oC heat treatment, and occurs around the peak-aged condition for
the 193oC aging practice This minimum in the ductility can be attributed to a large extent on the
planar deformation and strain localization of this alloy. The ductility of the alloy becomes very
brittle, to approximately 0.1% elongation, and thus the influence of planar deformation in
controlling the ductility is reflected in this elongation response to heat-treatment. The scanning
electron micrographs of the tensile fracture surfaces revealed predominately transgranular fracture
mechanism below and around peak-aging associated with the low ductility. However, after this
brittle behavior, the ductility improved past the peak-aged heat treatment into the overaged
condition. After the peak aged condition and longer aging times the ductility continued to increase
as the strength continued to decrease. A transition was characterized by very low ductility in the
slightly underaged condition up to the near peak-aged condition, then followed by a substantial
increase in ductility with aging after the peak-aged treatment. The ductility decreased to less than
one percent elongation in the slightly below peak-strength, and then substantially increased with
continued precipitation after the peak-strength. Microstructural variables that influence the
ductility, strength and toughness include the grain size, the size and distribution of the precipitate
particles, grain boundaries, and dislocations. The strengthening response of the alloy was a
consequence of ordered coherent Al3Li precipitates in the microstructure which impeded the
dislocation glide motion during plastic deformation. Based on a detailed quantitative microscopy
study, it was proposed that the increase in the ductility of the alloy after peak-aging was a
consequence of particle coarsening with aging resulting in the Orowan process during plastic
deformation due to the transition from dislocation particle shearing to dislocation particle
bypassing. Thus, from this study it was found that the ductility was directly related to the size,
352 J. M. Fragomeni

distribution, and spacing of the intermetallic precipitate particles. The Gc particle size, spacing,
volume fraction, and distribution are a direct consequence of the aging practice and composition,
and control the extent of plastic deformation.

FIGURE 1. Comparison Between the Ductility, Elongation Percent, of the 185oC and the 193oC
Aging Tempers for an Al-2.6wt.%Li-0.09wt.%Zr Alloy.
 2T11. Fracture mechanics analysis 353

BUCKLING OF MULTICRACKED COLUMNS

Christian Carloni, Cristina Gentilini and Lucio Nobile

Department DISTART, Alma Mater Studiorum-University of Bologna
Viale Risorgimento, 2 – 40136 Bologna, Italy
lucio.nobile@unibo.it

The analysis of buckling of elastic columns is one of the first problems in structural engineering
that was historically solved. Critical loads of perfect columns with various end restrains have been
derived. Nevertheless, the perfect column is an idealized model. In reality, inevitable imperfections
should be considered. A frequent imperfection to be taken into account is the weakness at an
interior location due to a partial crack.
A crack on a structural member introduces a local flexibility that affects its static and dynamic
response. The local flexibility of the cracked region of the structural element has been related to
the crack stress intensity factors (SIFs) (Irwin [1], Bueckner [2], Westmann and Yang [3]). The
stress intensity factors were obtained in many cases, and a well-known relationship was discovered
among the energy release rate, the stress intensity factors and the compliance of the cracked
member. The above relationship was then used to study the stability of the cracked columns
(Liebowitz et al,[4], Liebowitz and Claus [5], Okamura et al.[6]). A general method was
considered by Okamura et al. [7] for extending fracture mechanics through the compliance concept
to the analysis of a structure containing cracked members.
Stress intensity factors for many configurations are available. In most cases the results were
obtained by means of analytical and numerical methods. In many cases the results were obtained
by finite element methods and boundary element methods. Experimental methods have been
applied to simple cases in order to determine the fracture toughness KIC of engineering materials.
Solutions for many structural configurations are not available in the handbooks (see Tada et al.
[8]). Simple engineering methods which allow a fast and approximate determination of the stress
intensity factors are highly valued by design engineers. Remarkably simple methods for close
approximation of stress intensity factors in cracked or notched beams were proposed by Kienzler
and Herrmann [9] and by Nobile [10]. The former has been based on elementary beam theory
estimation of strain energy release rate as the crack is widened into a fracture band, the latter has
been based on elementary beam theory equilibrium condition for internal forces evaluated in the
cross section passing through the crack, taking in account the stress singularity at the crack tip.
The influence of such a kind of imperfection on the critical load has been analyzed by Wang et
al. [11]. The local flexibility due to the crack is modeled as a massless rotational spring. The
spring constant is determined on the basis of the energy released due to the crack and Castigliano’s
theorem. This method has been employed by Nobile and Carloni [12] to compute exact critical
loads for a cracked T-column with various end conditions and crack locations.
In this paper the method [11] is extended to compute exact critical loads for a multicracked T-
column with various end conditions and crack locations. The case of a T-column containing two
cracks is analized (Fig. 1).
354 C. Carloni et al.

FIGURE 1. Multicracked T-column.

References
1. Irwin, G.R., J. Appl. Mech., vol. 79, 361-364, 1975.
2. Bueckner, H.F., Trans. ASME, vol. 80, 1225-1229, 1958.
3. Westmann, R.A. and Yang, W.H., Theor. Appl. Fract. Mech., vol. 32, 693-701, 1967.
4. Liebowitz, H., Vanderveldt, H. and Harris, D.W., Int. J. Solids Struct., vol. 3, 489-500, 1967.
5. Liebowitz, H. and Claus, W.D., Eng. Fract. Mech., vol. 1, 379-383, 1968.
6. Okamura, H., Liu, H.W., Chu, C.S. and Liebowitz, H., Eng. Fract. Mech., vol. 1, 547-564,
1969.
7. Okamura, H., Watanahe, K. and Takano, H., ASTM Spec. Tech. Publ., vol. 536, 1973.
8. Tada, H., Paris, P.C. and Irwin, G.R., The stress analysis of cracks handbook, Del. Research
Coorporation, Hellerton, Pa, 1985.
9. Kienzerl, R. and Herrmann, G., J. Appl. Mech., vol. 53, 561-564, 1986.
10. Nobile, L., Theor. Appl. Fract. Mech., vol. 33, 107-116, 2000.
11. Wang, C.Y., Wang, C.M. and Tun, M. A., J. Eng. Mech., vol. 130, 1373-1376, 2004.
12. Nobile, L. and Carloni, C., In Proceedings of the 4th International Conference on Fracture
and Damage Mechanics, edited by M.H. Aliabadi et al., 2005, 161-165.
 2T11. Fracture mechanics analysis 355

EXPERIMENTAL AND NUMERICAL ANALYSIS OF INTERACTIONS

BETWEEN STRESS CORROSION CRACKS

M.Lamazouade, M.Touzet and M.Puiggali

Laboratoire de Mécanique Physique, UMR 5469 CNRS, Université Bordeaux 1,
351 Cours de la Libération, 33405 Talence Cedex, France
m.lamazouade@lmp.u-bordeaux1.fr

Surface crack colonies are commonly observed during stress corrosion phenomena. In this case,
the propagation rate of interacting cracks vary depending on their lengths and relative positions, Le
Poulain et al [1]. The objective of the study is to link mechanical and kinetic parameters. Surface
observations show that plasticity remains always localized around crack tips. This very confined
plasticity allows the use of linear fracture mechanics to describe the stress field near the crack
front.
In this present work, we study numerically the effects of the relative position of two merging
cracks on the variations of the Stress Intensity Factor (SIF) along their front and hence to provide a
better understanding of the experiment observations.

FIGURE 1. Two semi-elliptical surface cracks

We model a finite thickness plate containing two coplanar or non-coplanar surface cracks with
a semi-elliptical shape (Fig. 1).
Several ratios are introduced, a/c= flaw aspect ratio, a/t= relative crack depth, d/c= relative
distance between the two cracks and c/h the crack separation ratio.
The SIF is calculated from the J-integral along the crack front, with the finite element code
SAMCEF, using the equivalent domain integral (EDI) defined for a linear elastic material by Raju
and Shivakumar [2]. The J-integral values are obtained at different points along the crack front. A
common way to express the SIF changes due to the presence of multiple cracks is to use the
interaction factor J defined as

J
K d 
K d o f  (1)

K(d) and K d o f  are respectively the SIFs with and without interaction influence.
For interacting cracks, we show that the SIF distribution along the crack front is asymmetric
and depends on the relative position between the cracks. Experimental and numerical
investigations allow to correlate the crack growth with the mechanical parameter K and to
356 M. Lamazouade et al.

understand the influence of the size, the shape and the proximity of the interacting cracks on their
growth

References
1. 1. Le Poulain, F., Touzet, M. and Puiggali, M., Journal of material science, vol. 40, 1-11,
2005.
2. 2. Raju, I.S., Shivakumar, K.N., Engineering fracture mechanics, vol.42(6), 935-959, 1992.
 2T11. Fracture mechanics analysis 357

AN IMPROVED UPPER BOUND LIMIT LOAD SOLUTION FOR WELD

STRENGTH ANISOTROPIC OVERMATCHED CRACKED PLATES IN PURE
BENDING

N. Kontchakova and S. Alexandrov

Voronezh State University, Department of Applied Mathematics, Informatics and Mechanics,
University Sq., 1, 394006 Voronezh, Russia
kontchakova@yandex.ru
Institute for Problems in Mechanics, Russian Academy of Sciences
101-1 Prospect Vernadskogo, 119526 Moscow, Russia
sergei_alexandrov@hotmail.com

Metallic sheets after forming processes such as rolling are usually anisotropic (orthotropic), and
the anisotropy of plastic properties may have a significant effect of the limit load. For several
typical cracked specimens, it has been demonstrated in [1 – 3]. In all of these papers, welded
overmatched specimens have been considered assuming that the base material is orthotropic and
the weld material is isotropic. Also, the kinematically admissible velocity fields assumed consisted
of velocity discontinuity lines with no plastic zone of finite size. In particular, an upper bound
solution for welded cracked specimen subject to pure bending has been obtained in [3]. That
solution generalizes the solution for isotropic specimen proposed in [4]. The present solution
improves the solution given in [3].
The geometry of the specimen, the system of loading and the kinematically admissible velocity
field proposed are shown in Fig.1. The main difference from the kinematically admissible velocity
field proposed in [3, 4] is that there is a plastic zone of finite size below the velocity discontinuity
line in the form of a circular arc. The new velocity field is similar to the exact characteristic field
available for the homogeneous specimen. Therefore, the new solution provides a better limit load
than the solution given in [3]. Also, it is expected that the new solution does not differ much of the
exact limit load.
It is assumed that the weld material is isotropic and obeys Mises yield criterion, whereas the
base material is orthotropic and obeys Hill orthotropic yield criterion [5]. The solution depends on
the parameter c defined by
358 N. Kontchakova and S. Alexandrov

F G
c 1
4T 2
FG GH  HF  (1)

1 1 1 1 1 1 1 1 1
2F   , 2G  2  2 , 2H  
Y2 Z2 X 2 Z2 X Y X2 Y2 Z2 (2)
where X, Y, and Z are the tensile yield stresses in the principal directions of anisotropy, x, y and the
thickness direction, respectively, and T is the shear yield stress in the xy-plane (see Fig.1). The
isotropic specimen is obtained at c 0 . Using the solution at c 0 it is possible to evaluate an
effect of anisotropy of the limit load. Since this effect appears to be quite significant, it is important
to account for anisotropic plastic properties in structural defect assessment procedures such as
SINTAP [6].

References
1. Alexandrov, S., Gracio, J., J. Fatigue & Fracture of Engng Mater. Struct., vol. 26, 399-403,
2003
2. Alexandrov, S., Konchakova, N., J. Materials Science& Engineering A, vol.A 387-389, 395-
398, 2004.
3. Alexandrov, S., Konchakova, N. J. Engineering Fracture Mechanics, vol. 72, 151-157, 2005.
4. Joch, J., Ainsworth, R.A., Hyde, T.H., J. Fatigue Fract. Engng Mater. Struct. vol.16, 1061-
1079, 1993.
5. Hill, R., The mathematical theory of plasticity, Clarendon Press, Oxford, 1950.
6. SINTAP: Structural integrity assessment procedure for European industry. Final procedure,
Brite Euram Project No. BE95-1426, British Steel, 1999.
 2T11. Fracture mechanics analysis 359

FRACTURE PARAMETER ESTIMATION OF ALLOY STEEL REINFORCED

WITH MARAGING STEEL

Sunil Bhat, Vijay.G Ukadgaonker1, Mahendra Jha2 and S. M Nirgude3

Scientist “D”, ARDE, Defence Research and Development Organisation,
Pune - 411021, India
1Professor, Mech. Engg. Dept., Indian Institute of Technology, Bombay,

Mumbai - 400076, India

2Scientist “F”, ARDE, Defence Research and Development Organisation,

Pune - 411021, India

3Scientist “F”, ARDE, Defence Research and Development Organisation,

Pune - 411021, India

sunildrdo@rediffmail.com

Implementation of bimetallic configurations in critical technologies such as the liners for liquid
rocket combustion engines, gas turbine disks and bimetallic tubes and liners confirms their
viability and potential for futuristic use in many more engineering applications. High strength and
high toughness maraging steel inserts may be used in ordinary low strength alloy steel components
to withstand high localised stresses without compromising on the fracture toughness. This study
theoritically examines fracture aspects of such a bimetallic body. The crack is considered to be in
the parent body of low strength alloy steel and is assumed to grow across the interface into high
strength maraging steel.
As the load is applied to the bimetallic cracked body, a plastic zone develops at the crack tip.
Interaction of this plastic zone with maraging steel results in change in fracture parameter of the
approaching crack in alloy steel. A subcrack develops in maraging steel as the crack tip reaches
near the interface. This interaction is modelled by complex variable technique. Explicit analysis for
the propagation of the main crack and the subcrack is presented in the paper under plane stress
condition. Magnitude of applied load is varied to simulate the conditions of linear elastic (small
scale yielding) and elastic plastic (large scale yielding) regimes under the validity of the concept of
small deformations.

STAGE 1 STAGE 2 STAGE 3

(Plastic zone in alloy steel) (Plastic zone entering into maraging steel) (Subcrack in maraging steel)
FIGURE 1. Different stages of crack advancement in a bimetallic body

Crack growth across the bimetallic interface has been shown in Fig. 1. STAGE 1 depicts entire
plastic zone contained in alloy steel. Till the plastic zone touches the interface, the maraging steel
has no effect on the fracture parameter of the advancing crack as both parent and interface metals
have same elastic properties.
360 S. Bhat et al.

STAGE 2 marks growth of the crack with penetration of plastic zone into maraging steel.
Relation for crack opening displacement in the plastic zone in the form, G = f(Kapplied , VA,VB,
a,l,x) [1] is used to obtain the value of crack opening displacement at the crack tip and at any
location in the plastic zone. Dugdale model [2] postulating constant cohesive stresses equal to the
material yield stress in plane stress condition is employed to obtain the value of applied stress
intensity parameter. In linear elastic regime, the following expression is used :

A 2 (a  l) A 2l 2l
K a p p lied 2V  2V  2V B
S S S
(1)

In elastic plastic regime, the expression for applied stress intensity parameter at fracture has
been derived from the basic principles of quasi-linear fracture mechanics and is obtained as :

8z z + a + l 8 z + a  z + a + l 8 z + a  z + a + l
K a p p lie d ( a t fr a c tu r e ) V A ln - V A ln + V B ln
S z S (z  a) S (z  a)
(2)

where Kapplied (at fracture) = plin Sz for an infinite body and p lin and pf at fracture are
1 2
ª§ Spf · º 2
S § p lin ·
ln « ¨ c o s ¸ » = ¨ ¸
related by the expression «¬ © 2V A ¹ »¼ 8 © V A ¹

STAGE 3 highlights incipient subcrack (Y) in maraging steel as the main crack approaches the
interface. Effect of the subcrack is considered by modifying the relation for crack opening
displacement to the form, G = f(Kapplied ,VA,VB,a,l,x,Y) in order to compute the growth of main as
well as the subcrack. Equations (1) and (2) which hold good in STAGE 2 are suitably modified by
incorporating the effect of the subcrack to make them valid in STAGE 3.
The concerned equations are solved simultaneously by numerical iteration for different
unknowns till convergence is achieved. Effect on parameters of interest like crack tip stress
intensity, crack tip opening displacement, penetration of plastic zone in maraging steel and fracture
toughness of the bimetallic body with advancement of the crack towards the interface is presented
in detail.

References
1. Wappling, D., Gunnars, J. and Stahle, P., International Journal of Fracture, vol. 89, 223-
243, 1998.
2. Hellan, Kare., Introduction to fracture mechanics, Mcgraw-Hill Book Company, 17-21,1985.
 2T11. Fracture mechanics analysis 361

INCORPORATION OF LENGTH SCALES IN PLANE STRESS FRACTURE

ANALYSIS

V. P. Naumenko
Leading scientist, G.S. Pisarenko Institute for Problems of Strength of NAS of Ukraine
2 Tymiriazievska str., Kyiv 01014, Ukraine
v.p.naumenko@ipp.kiev.ua

This paper deals with a new engineering methodology that was developed by Naumenko et al. [1,
2] for assessing the Steady State Tearing (SST) in thin-wall components. We seek to correlate the
SST with the mechanical characteristics of a material that suffered preset levels of structural
damage. Attention is focused on quantifying the geometrical parameters of a Naturally-Forming
tear Crack (NFC). The NFC denotes a through-the-thickness crack that freely nucleates in thin-
wall plates or shells without initial stress raisers. The NFC length 2ci and the related value of the
crack mouth opening spacing 2si are treated as critical length scales needed for proper
characterization of the SST crack growth in a uniformly damaged material.
The uniaxial tensile tests were conducted on specimens made from an aluminium alloy in the
form of 1.47 mm thick sheets. For the first set of specimens, the material was thought of as
notionally homogeneous, having no initial stress, work hardening, and structural damage (Fig. 1a).
The specimens of the second set contained a well-defined zone of Reference Structural Damage
(RSD). It is shown in Fig. 1b as a strip with the height 2Ht and the reduced thickness Bt. Within
this strip, the same level of the RSD was introduced in each specimen by loading it to a given
plastic strain value. We suggest that due to the RSD existence the specimens of the width 2W0
120 mm should fracture under the conditions close to the transverse plane strain, when the
horizontal displacement of the points N is zero (Wi | W0 in Fig. 1b).
The length scales of interest (ci, si) are incorporated in fracture analysis by correlating the
parameters of two types of Active Damage Zones (ADZ). The first one is formed in specimens
having no initial stress raisers (Figs. 1a and 1b). The dimensions li and 2hi of this ADZ, called a
stationary one, were determined at the instant “i” of the NFC nucleation. They are unaffected by
the presence of crack-wake regions. Similar parameters of an ADZ located in front of a moving
crack tip are associated with the instant “s” of the attainment of the SST crack growth stage. The
characteristics of both ADZ types were expressed in terms of the applied load, displacements, and
crack extensions. The displacements vk and up of the extreme points k and p (Fig. 1) located on the
boundary of the Fracture Nucleation Region (FNR) play a key role in the quantification of the
stationary ADZ. The FNR is represented by a 2R u 2R square area of the specimen as shown in
Fig. 1. As for the moving ADZ, of key importance are the displacements and distances between
two pairs of the points “m” and “n” on the crack profile (Fig. 1c).
It should be emphasized that there was no switching from one mode of failure for the crack
nucleation event to another for the SST crack growth. Slant fracture by the mixed Mode I and
Mode III cracking occurs all the time from the NFC nucleation instant “i” until the complete
failure at the instant “f” (see Fig. 1c). This experimental fact enables easy SST analysis in the
framework of the engineering methodology [1, 2] under development.
362 V. P. Naumenko

(a)

(b)

(c)
FIGURE 1. Schemes of the near ligament geometry of two specimens without (a) and with (b) the
RSD. They are shown together with the completely cracked ligament (c).

References
1. Naumenko, V.P. and Volkov, G.S., Proceedings of the Second ASTM-ESIS Conference on
Fatigue and Fracture Mechanics, ASTM STP1461, West Conshohocken, PA, 2003.
2. Naumenko, V.P. and Atkins, A.G., Proceedings of the Seventh International Conference on
Biaxial / Multiaxial Fatigue and Fracture, Berlin, 2004, 6 pages.
 2T11. Fracture mechanics analysis 363

MODE III CRACK IN A FUNCTIONALLY GRADED PIEZOELECTRIC/

PIEZOMAGNETIC HALF PLANE

W.-H. Hsu and C.-H. Chue

Department of Mechanical Engineering, National Cheng Kung University
No. 1, Dasyue Road, 70101 Tainan, Taiwan, Republic of China
chchue@mail.ncku.edu.tw (C.-H. Chue), n1893152@ccmail.ncku.edu.tw (W.-H. Hsu)

The problem for an internal crack located within a functionally graded piezoelectric/piezo-
magnetic half plane is studied. It is subjected to the mechanical, electrical and magnetic loads. The
crack is normal to the free boundary of a functionally graded piezoelectric/piezomagnetic
materials. The elastic modulus, piezoelectric constant, dielectric constant, piezomagnetic
coefficient, magnetoelectric coefficient and magnetic permeability of the functionally graded
piezoelectric/piezomagnetic materials are assumed in exponential forms and vary along the crack
direction. By using the Fourier Transform theorem, the problem can be reduced to a system of
singular integral equations and solved by using the Gauss-Chebyshev integration technique. The
stress, electric displacement and magnetic induction intensity factors will be obtained.
Assume that the poling direction is along z-axis, the antiplane mechanical field, inplane
electrical and magnetic fields are coupled. The constitutive equations can be written as:

ww wI w\
W xz c0e Ex  e0e Ex  q 0e Ex
wx wx wx (1)

ww wI w\
W yz c0e Ex  e0e Ex  q0e Ex
wy wy wy (2)

ww wI w\
Dx e0e Ex  H 0e Ex  d 0e Ex
wx wx wx (3)

ww wI w\
D y e0e Ex  H 0e Ex  d 0e Ex
wy wy wy (4)
ww wI w\
Bx q 0e Ex  d 0e Ex  P 0e Ex
wx wx wx (5)

ww wI w\
By q 0e Ex  d 0e Ex  P 0e Ex
wy wy wy (6)

where W ij , w , Di , I , Bi , and \ are the shear stress, antiplane displacement, inplane

electrical displacement, electric potential, magnetic induction and magnetic potential. The
constants c0 , e0 , H 0 , q0 , d 0 and P0 are the material properties at free boundary. The constant
E is called nonhomogeneous parameter.
The stress, electric displacement and magnetic induction intensity factors are expressed as
following:
364 W.-H. Hsu and C.-H. Chue

G 1* ( b ) G * (b ) G * (b )
k 3 (b ) lim 2 ( x  b )W yz ( x , 0 )  c0e Eb  e0e Eb 2  q 0e Eb 3
xo b a0 a0 a0
(7)

G 1* ( a ) G * (a ) G * (a )
k3 (a ) lim 2 ( a  x )W yz ( x , 0 ) c0e Ea  e0 e Ea 2  q0e Ea 3
xo a  a0 a0 a0
(8)

G 1* ( b ) G * (b ) G * (b )
k 3D ( b ) lim 2 ( x  b ) D y ( x ,0 )  e0 e Eb  H 0e Eb 2  d 0e Eb 3
xo b a0 a0 a0
(9)

G 1* ( a ) G * (a ) G * (a )
k 3D ( a ) lim 2 ( a  x ) D y ( x ,0 ) e0 e Ea  H 0e Ea 2  d 0e Ea 3
xo a a0 a0 a0
(10)

G 1* ( b ) G * (b ) G * (b )
k 3B ( b ) lim 2 ( x  b ) B y ( x ,0 )  q 0e Eb  d 0e Eb 2  P 0e Eb 3
xo b a0 a0 a0
(11)

G 1* ( a ) G * (a ) G * (a )
k 3B ( a ) lim 2 ( a  x ) B y ( x ,0 ) q0e Ea  d 0e Ea 2  P 0e Ea 3
xo a a0 a0 a0
(12)
*
where G i ( t ) are bounded functions. The constant a0 is a half length of crack which aู xู b
lies in

FIGURE 1. Geometry of this problem

References
1. Delale, F., Erdogan, F., ASME J. Appl. Mechanics, vol. 50, 609-614, 1983
2. Erdogan, F., ASME J. Appl. Mechanics, vol. 52, 823-828, 1985
3. Huang, J.H., Kuo, W.S., J. Appl. Physics, vol. 81(3), 1378-1386, 1997
4. Rivlin, T. J., The Chebyshev Polynomials, Wiley, New York, 1974.
5. Sih, G.C., Song, Z.F., Theor. Appl. Fract. Mechanics, vol. 39, 209-227, 2003
6. Song, Z.F., Sih, G.C., Theor. Appl. Fract. Mechanics, vol. 39, 189-207, 2003
7. Van Suchtelen, J., Philips Res. Repts, Report No. 27, 1972.
8. Wang, B.L., Mech. Res. Communications, vol. 30, 151-159, 2003
9. Weng, G.J., Li, C.Y., J. Appl. Mechanics, vol. 69(4), 481-488, 2002
 2T11. Fracture mechanics analysis 365

ELECTRO-MECHANICAL FIELD OF A PIEZOELECTRIC FINITE WEDGE

UNDER ANTIPLANE LOADING

W.-J. Liu and C.-H. Chue

Department of Mechanical Engineering, National Cheng Kung University
No.1, Ta-Hsueh Road, Tainan 701, Taiwan, R.O.C.
chchue@mail.ncku.edu.tw (Chue, C.H.), n1893151@ccmail.ncku.edu.tw (Liu, W.J.)

This paper presents the general solutions of antiplane electro-mechanical field solutions for a
piezoelectric finite wedge subjected to a pair of concentrated forces and free charges as shown in
Fig.1. The boundary conditions on the circular segment are considered as fixed and grounded.
Employing the finite Mellin transform method, the stress and electrical displacement at all fields of
piezoelectric finite wedge are derived analytically. In addition, the singularity orders and intensity
factors of stress and electrical displacement can also be obtained too. These parameters can be
applied to examine the fracture behavior of the wedge structure. After being reduced to the
problem of an antiplane edge crack or an infinite wedge in a piezoelectric medium, the results
compare well with those of previous studies.
From Fig.1, the radial conditions of the wedge (T rD) are as follows:
where h1 d h 2 . The boundary conditions of the circular segment of the wedge (r = a) are
assumed as clamped and electrically closed:

WTz r,D  FG r  h1,WTz r,D  FG r  h2 , DT r,D  QG r  h1, DT r,D  QG r  h2 

(1)

w (a ,T ) 0 , I (a,T ) 0
(2)
Because the wedge has an infinite length along the z-axis, this problem becomes a generalized
plane deformation problem. Assuming the piezoelectric material is polarized along the z-axis, the
constitutive equation in cylindrical coordinate system can be written as:

ªW Tz º ªC 44 0 0  e15 º ªJ Tz º
«W » « 0 C e 0 » « »
« rz » « 44 15 » «J rz »
« Dr » « 0 e15 H 11 0 » « E r »
« » « »« »
¬ DT ¼ ¬ e15 0 0 H 11 ¼ ¬ ET ¼
(3)
where Wij are the shear stress, Jij are the shear strain, Diare the electric displacements and Ei are
the electric field vectors. The material properties C44, e15 and H11 are the elastic stiffness constant,
the piezoelectric constant and the dielectric constant, respectively. Substituting Eq. (3) into the
static equilibrium equations and Maxwell’s equation, the governing equations for antiplane
displacement w and inplane electric potential I are obtained as:
2
’ 2w 0, ’ I 0 (4)
2
where ’ is the Laplacian in (r, T). Applying the finite Mellin transforms of the first kind and
it’s inverse fourmula on (4) with boundary conditions, we can obtain the displacement and
electrical potential at whole fields. Thus, the stress, electrical displacement, singularity orders and
intensity factors of stress and electrical displacement can be obtained by using the eq. (1).
366 W.-J. Liu and C.-H. Chue

FIGURE 1. A piezoelectric wedge with a wedge angle 2D and a finite radius a subjected to a pair
of concentrated forces F and free charges Q.

References
1. Chue, C. H., Wei, W. B. and Liu, J. C., J. Chin. Inst. of Eng., vol. 26, 575-583, 2003.

2. Chue, C.H. and Liu, W.J., J. Solids Struct., vol. 41, 5023-5024, 2004.
3. Chue, C.H. and Liu, W.J., J. Chin. Inst. of Eng., vol. 27, 841-850, 2004.
4. Kargarnovin, M.H., Shahani, A.R. and Fariborz, S.J., J. Solids Struct., vol. 34, 113-128, 1997.
5. Xu, X. L. and Rajapakse, R. K. N. D., J. Solids Struct., vol. 37, 3253-3275, 2000.
 2T11. Fracture mechanics analysis 367

SENSITIVITY OF CRACK NUCLEATION PARAMETERS TO THE

GEOMETRIC IMPERFECTION

V.P. Naumenko and Yu.D. Skrypnyk

Leading scientist and senior scientist, G.S. Pisarenko Institute for Problems of Strength,
2 Tymiriazievska str., Kyiv 01014, Ukraine
v.p.naumenko@ipp.kiev.ua

In this experimental study, we compared the plane stress crack growth in identical specimens with
and without a geometric imperfection. The initial imperfection is intended to trigger a progressive
process of single-site necking followed by single-site cracking in a predefined location and
direction. This has an important bearing on the development of a new engineering methodology for
assessing the Steady State Tearing (SST) in brittle and ductile, metallic and non-metallic materials
by Naumenko at al. [1-3].
According to this methodology the input data on the SST are to be collected on specimens with
original cracks or starting slots of different lengths. In such a way, we attempt to formalize the
characterization of the whole crack profile using only the interrelations between loads,
displacements, and crack extensions. This is a purely mechanistic approach not taking into
consideration the physical damage mechanisms of the fracture process. As to the effects of
particular material properties, special-purpose tensile tests are to be performed in parallel. They
provided baseline data on the mechanical behaviour of the material within an Active Damage Zone
(ADZ).
It is assumed that an unnotched, nondeformed, and unstressed specimen contains a centrally
placed geometrical imperfection of sufficiently small size. The sensitivity of test records to this
imperfection is thought to be negligible. In this case, one can correlate the material tensile
properties in a freely nucleating ADZ with the parameters of the subsequent SST crack growth.
Freely nucleating ADZ denotes a certain volume of severely transformed material in which a
specific damage morphology develops in the course of tearing. At the instant “i” of nucleation of a
naturally forming tear crack, the ADZ is considered to be fully developed. The peculiarity of the
accepted damage characterization is that the stress-based, strain-based, and energy-based
parameters of the ADZ are unaffected by the presence of crack-wake regions.
The uniaxial tensile tests under monotonic loading were conducted on notched and unnotched
specimens at different in-plane constraint states. The low-constrained specimens had the width 12
mm and the high-constrained ones were of width 120 mm. Each notched specimen contained a
centrally placed hole whose diameter varied in the range from 0.1 mm to 5 mm. The test material
was a high-strength low-hardening aluminium alloy in the form of 1.47 mm thick sheets. Its
composition and mechanical properties are nearly identical to those of AL 2024-T3.
We assumed again that two adjacent ADZs are located symmetrically about the central point of
the specimen (Fig.1), which coincides with the centre of a small imperfection. For a specimen of
sufficiently large in-plane size, fracture nucleation is a local phenomenon. Micromechanisms of
this physical event by their nature must be independent or only slightly dependent on the specimen
geometry. It is very important to express the local damage accumulated within the ADZ in terms of
directly measurable quantities. With this in mind, we incorporated in the fracture analysis the
notion Fracture Nucleation Region (FNR). It is represented by a centrally located square of size 2R
u 2R embedded in the specimen as shown in Fig.1. Here, the basic idea is to quantify the
interaction between the FNR profile and the outer boundaries of the specimen in terms of
geometric variables, which can be readily measured.
368 V.P. Naumenko and Yu.D. Skrypnyk

FIGURE 1. Scheme of the near ligament geometry of a specimen with an Active Damage Zone
that is located in the centre of a Fracture Nucleation Region.

In our tests, the displacements vM and vk of the characteristic points M and k (Fig. 1) along the
y-axis and the displacements uN and up of the characteristic points N and p along the x-axis were
recorded in parallel. It allowed determining the averaged mechanical parameters of fully-
developed ADZs in low-constrained and high-constrained specimens. The same set of test data
was used to quantify in-plane constraint in terms of the FNR distortion ratio dr = up·Rki / vk·Rpi and
the specimen distortion ratio db = uN·HMf / vM·WNf. Here, Rki, Rpi are the FNR dimensions at the
instant “i” and 2HMf, 2WNf are, respectively, the specimen height and width at the instant of
complete fracture denoted by the index “f”. These constraint parameters, as well as the crack
nucleation parameters, are highly sensitive to the size variation of the initial geometrical
imperfection represented in our study by an open hole. Our experimental results demonstrate that
there is an intimate relation between the material damage within a fully-developed ADZ in an
unnotched specimen and the resistance to subsequent crack growth in this very specimen.

References
1. Naumenko, V.P., Fracture Processes in Concrete, Rock and Ceramics (RILEM/ESIS), J.G.M.
van Mier, J.G. Rots, and A.. Bakker, Eds., London, Vol.8., 1991, 183-192.
2. Naumenko, V.P. and Volkov, G.S., Proceedings of the Second ASTM-ESIS Conference on
Fatigue and Fracture Mechanics, ASTM STP1461, West Conshohocken, PA, 2003.
3. Naumenko, V.P. and Atkins, A.G., Proceedings of the Seventh International Conference on
Biaxial / Multiaxial Fatigue and Fracture, Berlin, 2004, 6 pages.
 2T13. Probabilistic approaches to fracture mechanics 369

AN EXPERIMENTAL EVALUATION OF A LOCAL APPROACH MODEL FOR

GRADED MATERIALS

B. Bezensek, J. Flasker and J.W. Hancock1

University of Maribor, Faculty of Mechanical Engineering,
Smetanova 17, SI-2000 Maribor, SLO
1University of Glasgow, Mechanical Engineering Department, Glasgow G12 8QQ, UK

b.bezensek@eng.gla.ac.uk, joze.flasker@uni-mb.si, j.hancock@eng.gla.ac.uk

A new local approach model has been developed for functionally graded materials (Banerjee [1],
Bezensek et al [2]). The model stems from the Beremin’s [3] model for describing the statistics of
crack propagation and allows yield strength and toughness to vary spatially in a systematic
manner. The model associates the curve-fitting constant Vu in the two-parameter Weibull
cumulative distribution function with the mean strength of a material. In the case of cracks where
crack tip fields exhibit self-similarity, the mean strength can be associated with a non-dimensional
Weibull stress (Lei et al [4]). At failure the applied J is associated with the mean fracture
toughness, allowing the expression for cumulative failure probability to be rewritten with spatially
varying strength and toughness.
In the paper the model is experimentally verified using standard and customised fracture
mechanics experiments. For the later, the spatial gradients in yield strength and fracture toughness
have been obtained by establishing a temperature difference along the length of a bend bar of a
ferritic steel. The advantage of these tests is that intrinsic material properties can be determined by
standard testing procedures across the required temperature range on a material with defined
micro-mechanical properties. The bar was free to contract such that the secondary stresses do not
feature significantly in the test. Tests have been performed using a large temperature gradient near
the crack tip. Apparent fracture toughness was measured for the configuration using the J-integral,
associated with the area under the load-crack mouth opening curve. Preliminary results are shown
in Fig. 1 as histograms of failure probability plotted against apparent fracture toughness. These are
compared with the values for a homogeneous configuration tested at the same temperature.
Using the local approach model predictions of failure probability and crack extension direction
have been obtained for the configurations with uniform and spatially varying properties. A
pragmatic approach is used for quantifying the remaining free variable in the Weibull distribution
function: the modulus m. In normally distributed large populations the curve-fitting constant Vu
attains the mean value, while the Weibull modulus measures the spread of the data. It is shown that
Weibull modulus can be taken as the ratio of the mean value to the standard deviation of the
variable. For typical fracture toughness distributions (Joyce and Tregoning [5]) this gives m values
in the range of 2 to 5. This is close to the theoretical predictions of Wallin [6] based on the size
effect arguments. The predictions of the new local approach model are superimposed on Fig. 1.
Good agreement with the experimental data can be observed.
In practice fracture tests are frequently performed on welded joints using impact Charpy
testing due to simplicity. It would be advantageous for the engineering practice to correlate the data
obtained from such testing with the values from fracture mechanics tests. Such correlations however
must take into account the local micro-mechanics associated with the local material properties. Using
the new local approach model, a procedure can be developed where fracture resistance from Charpy
tests is correlated with that from fracture mechanics tests. Examples are presented using test data
obtained on laser welded joints, discussed by Bezensek and Hancock [7].
370 B. Bezensek et al.

References
1. Banerjee A., The Fracture Mechanics of Bi-Material Systems, Ph.D. thesis, University of
Glasgow, September 2003
2. Bezensek B., Banerjee A. and Hancock J.W., In: Proceedings of the ASME Pressure Vessels
and Piping Conference, Ed. P-S Lam, Denver, Co, USA, July 2005
3. Beremin F.M., Metallurgical Transactions, vol. 14A, 2277-2287, 1983
4. Lei Y., O’Dowd N.P., Busso E.P. and Webster G.A., Int. J. Fracture, vol. 89, p. 245, 1998
5. Joyce J.A. and Tregoning R.L., Eng.Fract.Mechanics, vol. 68, 861-894, 2001
6. Wallin K., Eng.Fract.Mechanics, vol. 19, p.1085, 1984
7. Bezensek B. and Hancock J.W., In: Proceedings of the ASME Pressure Vessels and Piping
Conference, Ed. P-S Lam, PVP-474, San Diego, Ca, USA, July 2004
 2T13. Probabilistic approaches to fracture mechanics 371

A STOCHASTIC MODEL FOR CRACK GROWTH

C.-R. Chiang
Department of Power Mechanical Engineering, National Tsing Hua University
Hsin Chu 30013, Taiwan
crchiang@pme.nthu.edu.tw

Considering the nucleation sites around the crack tip as the events of a Poisson’s process with
intensity O, a micromechanical model for the prediction of the fracture toughness of materials is
developed. Stress and strain ahead of a crack under mode-III loading are reviwed[1] and it is
shown that the distribution of the critical stress intensity factors (SIF) for stress controlled failure
can be expressed by

§ K4 ·
G (K ) 1  exp ¨¨  O ¸¸
© 4 SW c4 ¹ (1)

where W c is the critical stress for decohesion or cracking of the second phase particles. On the
other hand the distribution of critical K for strain controlled failure is

§ K 4 ·
G (K ) 1  exp ¨  O ¸
¨ 4 SP 2 W y2 J 2 ¸
© c ¹ (2)

where P is the elastic shear modulus, W y is the yield stress in shear and J c is the critical strain
for crack growth. The results are then modified and extended for the mode-I case. The distribution
of K is shown to be

§ § ( K  K min ) 4 · ·
G (K ) 1  exp ¨¨  O ¨¨ ¸¸ ¸
¸
© © M ¹¹ (3)

where K min is a material parameter reflecting the finite size of the inclusion, M is the counterpart
of Mode-I as in the Eq(1) or (2). .Interpreting Eq(3) as failure probability of a specimen under a
given K, the theoretical models are then compared to other statistical models[2,3].
The weakness of Eq(3) is that the critical event is assumed to be always catastrophic which
does not reflect the actual behavior of the material. To be more realistic, it is assumed that there are
two parent distributions (say similar to Eq(3) with different material parameters). One is associated
with low energy barriers and the other is associated with high energy barriers for crack growth.
The low energy barriers mean that when the critical event occurs the crack grows in a stable
manner. On the other hand when the critical event is associated with high energy barriers the crack
grows catastrophically. Based on the arguments in[4], it is shown that for a given growth distance
'a , the distribution for the crack maintaining stable is given by
D
'a § K  K ·
G1(K ; 'a) exp  ¨¨ i
¸¸
A © Si ¹ (4)
While the distribution for crack starts to grow catastrophically is given by
372 C.-R. Chiang

E
'a § K  Ku ·
G2 (K ; 'a) 1  exp  ¨¨ ¸¸
A © Su ¹ (5)

where K i , Si , D , K u , Su and E are material parameters; A is a length scale. Crack growth

behavior based on Eq(4) and (5) is shown in Fig.1

References
1. Rice, J. R. J. Appl. Mech vol.34, 287-298,1967
2. Curry, D.A. and Knott, J. F. Metal Sci. vol. 13, 341-345, 1979
3. Wallin, K. Eng.Frac. Mech. vol.19, 1085-1093, 1984.
4. Chiang, C. R. Int. J. Frac. Vol.53, 337-342, 1992
 2T13. Probabilistic approaches to fracture mechanics 373

STOCHASTIC EVALUATION OF FATIGUE CRACK INITIATION AND

PROPAGATION

G. S. Wang
System technology division, The Swedish Defence Research Agency
Gullfossgatan 6, Kista, 164 90 Stockholm, Sweden
wgs@foi.se

Stochastic fatigue crack growth analyses have been performed using the experimental fatigue
crack growth data obtained for the cracks initiated from metallurgical inclusions at the notch of the
specimen, the AGARD short crack growth experimental data [1]. This investigation is not intended
to analyse the feature of short crack growth behaviour, but to analyse the effect of the inclusion
distribution on the time that the crack is initiated, and the consequent stochastic fatigue crack
growth behaviour. The incremental stochastic crack growth analysis based on FFT [2] (fast Fourier
Transformation) and Monte Carlo simulation has been made according to a two step differential
stochastic fatigue crack growth model which assumes that the fatigue crack growth consists of a
rapid and slow pulse process. The fatigue crack growth is simulated from an initial flaw size,
which is the inclusion size, to a large crack size.
The distribution of inclusions in the crack initiation direction used in the crack growth analysis
is based on the electron scanning microscope inspection results. It represents the physical initial
flaw in the analysis when cracks are considered to be initiated from inclusions in the sizes ranging
from 1 Pm to 20 Pm. The crack is assumed to be initiated from the location where a particle has its
largest size on the surface of the notch. This initial flaw is then grown into the final size although
multiple cracks are often observed in the experimental results. A criterion has been applied when
collecting the experimental data to prevent crack interaction.
Along with the incremental FFT and Monte Carlo simulation, different closed form models
have been investigated. Among them are Brownian closed form solutions with or without the
initial flaw distribution, and a model based on the deterministic crack growth law and initial flaw
distribution (similar to the USAF model [3] but with different initial crack definition). These
models represent different methods in state of the art of the probabilistic crack initiation and
growth analysis methodology when the initial crack distribution is involved. The investigation is to
evaluate consequence of simplifications in the analysis of the crack that is initiated from a
distribution of initial flaws in the material matrix.
Both the incremental simulation and the closed form solution agree reasonably well with the
experimental results although parameters used in the stochastic model are directly derived from the
experimental data of large cracks. The investigation indicates that scatter in fatigue life is affected
mostly by the stochastic process of crack growth instead of scatter in initial flaws so that the
stochastic solution is important when the crack size is considered to be significantly larger than the
initial flaw size. The stochastic solution combined with scatter in initial flaws gives a good
description of the fatigue damage although the stochastic solution with an average initial flaw size
may provide exceptional good results for the reliability analysis for the case when the size of initial
flaw is small.
A comparison of different analytical methods is shown in Fig.1b). The model with a
deterministic crack growth and scatter in initial flaws seems to be unacceptable since deterministic
crack growth analyses will give predictions of fatigue life with much smaller scatter. The models
based on deterministic crack growth methods should be therefore avoided when the physical initial
flaws are used as source of crack initiation since the analysis results may be unacceptable in the
374 G. S. Wang

reliability analysis of crack growths. The model with deterministic crack growth relation should
not be used unless an equivalent initial flaw size distribution is considered instead of the physical
initial flaw size distribution, while the equivalent initial flaw distribution will be determined by
values extrapolated from fatigue tests according to the same deterministic crack growth law.

(a) (b)
FIGURE 1. a) Experimental crack growth against load cycles for a single notch specimen. b)
comparison between different models and experimental data for the load cycles when crack grows
to a size of 1 mm.

References
1. Newman, J. C., Jr., and Edwards, P.R., AGARD-R-732, Dec, 1988.
2. Shinozuka, M., Deodatis, G., Appl. Mech. Rev. vol. 44, no. 4, 191-203, 1991.
3. Manning, S.D. and Yang, J.N., Wright-Patternson Air Force Base, Ohio, AFWAL-TR-88-
3119, Feb. 1989.
 2T13. Probabilistic approaches to fracture mechanics 375

A WEIBULL-BASED METHOD TO PREDICT THE STRENGTH OF

ADHESIVELY BONDED JOINTS OF PULTRUDED FRPS

Till Vallée, João R. Correia1 and Thomas Keller

Composite Construction Laboratory, Swiss Federal Institute of Technology, Switzerland
1Instituto Superior Técnico, Technical University of Lisbon, Portugal

till.vallee@epfl.ch, jcorreia@civil.ist.utl.pt, thomas.keller@epfl.ch

Fibre reinforced polymer (FRP) pultruded composites are more and more used as load bearing
structural elements for civil engineering applications. Besides the inherent advantages of
composites, they are relatively cost-effective. Like for all types of composites, the topic of
connections is one of the key issues for which still no satisfactory solution exists. The actual
practice of bolting cannot be considered as being material adapted and adhesively bonded
connections have already proven, by far, its better suitability to the anisotropic and brittle
behaviour of the pultruded material.
The failure of bonded joints is also usually brittle, as shown in Fig. 1, which illustrates a
typical failure sequence filmed with a fast speed camera (at 2’000 fps). The determination of the
strength of adhesively bonded connections of pultruded FRPs is a relatively complex task. This is
on one hand due to the fact that pultrusion leads to fibre architecture not comparable to layered up
composites, and on the other hand due to a relatively complex multi-axial stress state in the
overlapped zone. Another challenge is represented by the relatively sharp stress profiles making
simple continuum mechanics methods difficult and tributary of calibration factors [1].

FIGURE 1. A typical brittle failure of adhesively bonded joints, filmed at 2’000 fps.

To overcome the limitations of the continuum mechanics approach, the Authors have
developed a method to predict the quasi-static strength of adhesively bonded joints of pultruded
FRPs based on the experimentally gathered material resistance towards the multi-axial stresses
acting inside the joint, which takes into account the sharp stress profiles and the associated size
effects. To gather the material resistance towards the acting stresses, a Shear-Tensile Device was
specifically designed (see Fig. 2).
376 T. Vallee et al.

FIGURE 2. The CCLab Shear-Tensile Device.

The method relies on a Weibull-based description of the material strength [2]. Despite their
advantages, such methods are almost not used for civil engineering applications where issues
related to scale effects are usually simply taken into account by experimentally gathered stress
concentration factors needed to fit experimental to theoretical results.
By linking the material resistance, including the corresponding statistics, and the numerically
gathered stress state inside the joint, it was possible to formulate a method to predict the strength of
adhesively bonded joints in an almost closed-form manner, without having to make any
supplementary assumptions.
The prediction method was then compared with own experimental results where several
geometrical parameters (adhesive layer thickness, size of the adhesive layer fillet and overlap
length) were varied, and a good agreement between theoretical and experimental results was
found.

References
1. Keller, T. and Vallée, T., Composites Part B: Engineering, vol. 36, no. 4, 331-340, 2005.
2. Weibull, W., Journal of Applied Mechanics, vol. 18, 293-287, 1951.
 2T14. Computational fracture mechanics 377

THE LATERAL CONSTRAINT INDEX AS A NEW FACTOR TO ASSESS THE

INFLUENCE OF THE SPECIMEN THICKNESS

Alfonso Fernandez-Canteli, Daniel Fernandez-Zuniga and Enrique Castillo1

EPSIG, University Oviedo, Campus de Viesques, 33203 Gijon, Spain
1ETSICCP, University Cantabria, Avda. de los Castros s/n, 39005 Santander, Spain

castie@unican.es, afc@uniovi.es

In this work, the stress intensity factors KIxx, KIyy and KIzz are formally introduced for the in-plane
stresses, xx and yy, as well as for the out-of-plane stress zz, as an extension of the conventional stress
intensity factor for mode-I of loading, KI in the LEFM:

K Ixx lim 2S r V xx - 0º
ro 0

K Iyy lim 2S r V yy - 0º
ro 0

K Izz lim 2S r V zz - 0º .
ro 0
(1)
*
This facilitates the definition of a lateral constraint index, , as the ratio K Izz / K Izz at the
*
crack front, or, as the more general expressión, V zz 2S r / K Izz , outside the crack front,
*
where K Izz is referred to plane strain conditions. This index differs from the degree of plane
strain, given by the relation V zz / Q V xx V yy  , as used by Nakamura-Parks and other authors,
[1-2], which rests on the unacceptable asumption that the condition V zz / Q V xx V yy  is
sufficient to guarantee a state of plane strain at the crack front.
The index is calculated by means of 3-D FEM numerical simulations (see Figs. 1 and 2) and
proves to be a valuable factor to establish a general relation between the stress intensity factor KI
and the J-integral under mode-I of loading for whatever specimen thicknesses. The index also
facilitates the study of the influence of the lateral constraint on the off-plane stress field ızz and
strain field İzz [see 3,4], thus opening new perspectives for the consideration of the constraint
effect on the stress distribution at the crack front and on the crack tip opening displacements at
different z/B positions as a function of the specimen thickness.
This will have practical consequences, particularly to the conversion used in finite element
codes and to establish a more adequate effective elastic modulus for the ASTM E1921 standard, as
pointed out by Scibetta [5].
The approach can be extended straightforwardly to mode-II or mixed-mode with the aim to
investigate the lateral constraint effects on the 3-D stress distributions and the fracture toughness in
more general cases of loading.
378 A. Fernandez-Canteli et al.

FIGURE 1. Lateral constraint index, J z / B  , along the crack front, r=0, for a specimen
thickness B=25 mm.

FIGURE 2. Lateral constraint index, J B  , in the middle of the crack front, r=0 and z/B=0, for
different specimen thicknesses, B.

References
1. Nakamura, T., Parks, D.M., J. of Applied Mechanics, Vol. 55, 805-813, 1988.
2. Kwon, S.W., Sun, C.T., Int. J. of Fracture, Vol. 104, 291-315, 2000.
3. Leung, A.Y.T., Su, R.K.L., Finite Elements in Analysis and Design, Vol. 18, 389-401, 1995.
4. Su, X.M., Sun, C.T., Int. J. of Fracture, Vol. 82, 237-252, 1996.
5. Scibetta, M., Int. J. of Fracture, Vol. 117, L3-L8, 2002.
 2T14. Computational fracture mechanics 379

ANALYSIS OF CRACK PROPAGATION IN ALUMINA-GLASS

FUNCTIONALLY GRADED MATERIALS

V. Cannillo, L. Lusvarghi, T. Manfredini, M. Montorsi, C. Siligardi and A. Sola1

Dipartimento di Ingegneria dei Materiali e dell’Ambiente -
University of Modena and Reggio Emilia
Via Vignolese 905 – 41100 Modena (ITALY)
1sola.antonella@unimore.it

Functionally Graded Materials (FGMs) are heterogeneous systems whose composition and
microstructure vary in space, thus creating a resulting gradient in properties and functions which
can be tailored to the prescribed thermo-mechanical loading conditions [1]. Moreover, FGMs
allow to fully exploit the peculiar properties of the ingredient materials, whilst the smooth change
in composition and microstructure minimizes the stresses which can arise when heterogeneous
phases are directly coupled [2, 3]. Though the study of fracture mechanics in FGMs is complicated
by their intrinsic microstructural complexity, a good understanding of crack initiation and
propagation is required in order to ensure a higher reliability of functionally graded structural
components.

FIGURE 1. Graphics of (a) the glass percolation into the alumina substrate (natural transport based
technique) and (b) the deposition of a multi-layered coating (constructive technique)

In particular, the present work was focused on the evaluation of glass (CaO-ZrO2-SiO2) -
alumina FGMs obtained by means of two different methods, i.e. molten glass percolation and
thermal spraying. In the former technique, a high temperature thermal treatment (4 hours at
1600°C) was performed in order to melt the glass and induce its percolation into a polycrystalline
alumina substrate. Such fabrication route, which is based on a natural transport phenomenon [4],
gives rise to a macroscopically continuous gradient, as schematically shown in Fig. 1a. In the latter
case, multilayered thick coatings (up to around 700 µm) were deposited via a DC plasma torch on
the same ceramic substrates. The coating mean composition changed from a 100% vol. glass top
layer to a 80 or 100% vol. alumina layer at the interface with the substrate. This method represents
an example of a constructive technique [4], since the gradient is arbitrarily built up layer by layer,
as described in Fig. 1b.
The microstructural and micromechanical properties of the two kinds of FGMs were tested by
means of several techniques, such as SEM-EDS investigation, XRD analysis and, most of all,
depth-sensing Vickers microindentation, which provided for a trustworthy measurement of local
properties [5]. Great attention was paid to the spatial variation of mechanical properties, such as
hardness, caused by the compositional gradient; moreover the crack propagation eventually
380 V. Cannillo et al.

induced by Vickers indentation was carefully evaluated. To conclude, a comparison between the
fracture mechanics in the two kinds of FGMs was proposed, focusing on the effect of
microstructural features which are governed by the different fabrication techniques.

References
1. Miyamoto, Y., Kaysser, W.A., Rabin, B.H., Kawasaki, A. and Ford, R.G., Functionally
Graded Materials. Design, Processing and Applications, Kluwer Academic Publishers,
Dordrecht/Boston/London, 1999.
2. Koizumi, M. and Niino, M., MRS Bulletin, Vol. XX, No. 1, 19-21, 1995.
3. Cannillo, V., Manfredini, T., Montorsi, M., Siligardi, C., Sola, A., Journal of the European
Ceramic Society, in press.
4. Mortensen, A. and Suresh, S., International Materials Review, Vol. 40, No. 6, 239-265, 1995.
5. Oliver, W.C. and Pharr, G.M., Journal of Materials Research, Vol. 7, No. 6, 1564-1583, 1992.
6. Jones, G. A., Applied Fracture Mechanics, Soringer, Berlin, Germany, 1978.
 2T14. Computational fracture mechanics 381

NUMERICAL SOLUTION OF INTEGRO-DIFFERENTIAL EQUATIONS FOR

FRACTURE MECHANICS PROBLEMS

A. V. Andreev
Federal State Unitary Enterprise “Electrogorsk Research and Engineering Center on Nuclear
Power Plants Safety”
Bezymyannaya St. 6, 142530, Moscow region, Electrogorsk, Russia
andreev@erec.ru

The paper is dedicated to the development of methods for solving one-dimensional singular and
hypersingular integro-differential equations with generalized Cauchy kernels. Using the means of
the theory of special functions, constructive methods of direct (without regularization) numerical
solution of such equations are suggested. One-dimensional singular integro-differential equation
(SIDE) with standardized integration interval in sufficiently general case is considered:
1 1 1
I c ([ ) d [ I ([ ) d [
AI c(K )  B I (K )  C ³ D³  ³ K ( [ , K )I c ( [ ) d [ 
1
[  K 1
[ K 1

1
dI ( x)
 ³ L ([ , K )I ([ ) d [ p (K ) I c( x )
1  K  1
1 , , dx (1)

Here I ( [ ) is unknown function, p (K ) is bounded continuous function known on the interval

[–1, 1], and kernels K ( [ , K ) and L ( [ , K ) can have “fixed” singularities on the end points of
integration interval, i.e. becomes unbounded only if [ and K approach simultaneously to one of
end point of the interval [  1, 1] ( [ K o r 1 ). In the latter it is assumed, that K ( [ , K )
and L ([ , K ) are generalized Cauchy kernels, i.e. their singularities have form U  1 ( U o 0 )
(Erdogan et al. [1]). A, B, C, and D are constants, which may be functions of K under certain
assumptions (see Muskhelishvili [2]). Constants and functions K ( [ , K ) , L ( [ , K ) and p (K )
are real or complex.
The developed methods provide for availability of derivative of the unknown function only out
of the integral terms, so at C , K ( [ , K ) z 0, SIDE (1) is reduced to equivalent (at
I c ( [ )  H ) hypersingular integro-differential equation (HSIDE) by means of integration by
parts of corresponding integrals (1). Note, that hypersingular integral equations of elasticity theory
have advantage on singular (and Fredholm) ones, since HSIDE are formulated in terms of the
values characterizing contact interaction of deformable solids – displacements (or its discontinuity)
and stresses (Linkov [3]). This allows the avoiding additional integration of derivative of
displacements (or its discontinuity), which contain the singular equation. Thus, solution of a mixed
and contact elasticity problems, in particular, problems on crack with contacting surfaces (Andreev
et al. [4]), makes it possible for significant simplifying.
It is assumed, that solution of SIDE (1) and the corresponding HSIDE has the power-type
asymptotic on the end points of integration interval, which is determined both “movable” Cauchy-
type singularities (first and second integrals (1)) and “fixed” singularities of the kernels K ( [ , K )
and L ([ , K ) if exist. Note, that availability of “fixed” singularities in kernels K ( [ , K ) and
382 A. V. Andreev

L ([ , K ) effects significantly on solution asymptotic in the vicinity of the end points of

integration interval. In such situation, generally, solution has no usual square-root asymptotic, but
looks like r O , 11// 22 z
z RRee OO // 00 , r o 0 [1, 2].
The evaluation of asymptotic exponent perform before full solution of SIDE (1), which
describes a boundary value problem, by means of asymptotic analysis of the dominant part of
SIDE [1, 2], or based directly on asymptotic analysis boundary problem [3]. Explicit analytical
account of asymptotic on numerical solution allows the estimation dominant terms of asymptotic
expansion of the solution in vicinity of the end points of integration interval with high accuracy. It
is important, particularly, in elasticity problems for bodies with crack, since a dominant term of
asymptotic expansion of the solution in vicinity of crack tips control the stress intensity factor,
which has prime importance in brittle fracture.
Approaches to solution HSIDE with complex or real asymptotic are distinguished
qualitatively. The former is based on expansion of the solution on the finite system of orthogonal
polynomials (with explicit account for the end points asymptotic), analytical calculation of
singular and hypersingular integrals, and replacement generalized kernel by confluent one with
further analytical integration of this term (or direct numerical calculation of the latter). Then
obtained functional equation by means of the collocation method is reduced to linear algebraic
system, which used for evaluation of the expansion coefficients. The direct approach is used for
real asymptotic, which is based on Lagrangian approximation of the unknown function (with
explicit account for the end points asymptotic), interpolation-type quadrature formulae, and the
collocation method for reduction HSIDE to linear algebraic system relative to values of unknown
function in the discrete points.
The numerical results and their comparison to analytical solution of some crack problems are
presented in the paper. The developed methods allow to solving SIDE arising in different applied
problems of physics and mechanics, particularly, in mixed elasticity problems, contact problems of
solid mechanics, and two-dimensional problems of fracture mechanics.

References
1. Erdogan, F. E., Gupta, G. D., Cook, T. S., The numerical solutions of singular integral
equations, In Mechanics of Fracture. V. 1. Methods of Analysis and Solutions of Crack
Problems, Noordhoff Intern. Publ., Leyden, 1973
2. Muskhelishvili, N. I., Singular Integral Equations, P. Noordhof, Groningen, The Netherlands,
1953.
3. Linkov, A.M., Complex boundary integral equations method in elasticity, Nauka, Sankt-
Peterburg, Russia, 1999.
4. ndreev, .V., Goldstein, R.V., Zhitnikov, Yu.V., Izv. RAN echanics of Solids, No. 4, 96-112,
2002
 2T14. Computational fracture mechanics 383

ANALYTICAL METHOD OF GENERATING DA/DN CURVE FOR

AEROSPACE ALLOYS

B. Farahmand
Boeing Integrated Defence System
5301 Bolsa Avenue, Huntington Beach, California 92647
bob.farahmand@boeing.com

.Life assessment of aircraft or aerospace fracture critical parts, through the linear elastic fracture
mechanics approach, needs to have fatigue crack growth rate (FCGR) data for the material under
consideration. These tests are costly and time consuming. In many cases, the time required to
complete the standard FCGR test exceeds the deadline set forth by the customers. In other cases,
due to the budget constraint, the analyst is forced to select a similar material, which in many cases
life estimation results can be too conservative or incorrect. The proposed method is capable of
generating the regions I, II, and III of the FCGR curve without conducting the ASTM–E647 testing
standards. The ASTM FCGR tests include machining the C(T) or M(T) specimens, pre-fatiguing,
measuring crack growth, gathering data, and interpreting of the data. Moreover, in region I of the
FCGR, the load shedding technique is commonly used to capture the threshold stress intensity
factor, which can be tedious and time consuming. The proposed analytical method can
independently estimate each region of the curve. These regions will also be connected to establish
the total FCGR curve through the well known fatigue crack growth rate equation established by
Forman and Newman [1]. The final failure in the accelerated region (region III), where 'K|Kc,
can be estimated through the Griffith energy balance equation by using the full stress-strain curve,
established from a reliable source, for the material [2]. The Paris region can be established by
selecting two points on the curve (the two points are both situated before the threshold and
accelerated regions, where 'Kth<'K<Kc). For these two points, the ratio of 'K/'Kth and 'K/Kc
were found to be a constant for many metallic alloys. The threshold values for Aluminium and
Titanium alloys were found to be related to the material plane strain fracture toughness, KIc
(Figure 1), and were falling between the KIc/4S and KIc/3S as shown in Figures 2a & b for the
2000 series aluminums, respectively.

FIGURE 1. Trend between KIc and Kth for R=0 were observed (dashed circle)
384 B. Farahmand

FIGURE 2. Estimated Kth for R=0 is falling between KIc/4S (a) and KIc/3S (b)

Based on this approach, a computer code was established under the NASA contract called
Fatigue Crack Growth (FCG) that can generate the FCGR curve. Several Aluminum and Titanium
alloys were selected and the da/dN versus 'K curves were plotted by the FCG and compared with
the test data extracted from the NASGRO data base [1]. Figure 3 shows the FCGR curve
established through analysis versus the test data for 2219-T87, 7075-T73, Ti-6Al-4V (STA
condition) and Ti-6Al-4V (Mill condition) alloys. Excellent correlation can be found between the
analysis and tests. Work is in progress to extend this method to other metallic materials used for
the aerospace vehicles.

FIGURE 3. Analysis versus test data for several aerospace alloys

References
1. Forman, R. G., Shivakumar, V, Newman, J. C., Fatigue Crack Growth Computer Program
“NASA/FLAGRO, JSC-22267A, January 1993
2. Farahmand, B., Fatigue and Fracture Mechanics of High Risk Parts, Chapman & Hall, 1997,
Chapter 5
 2T14. Computational fracture mechanics 385

THERMO-ELASTIC FRACTURE OF EDGE CRACKED PLATE UNDER

SURFACE ‘SHOCK’ LOADING

B. P. Fillery, X. Hu and G. Fisher1

Department of Mechanical Engineering, University of Western Australia
35 Stirling Hwy, Crawley, Western Australia 6009
1Woodside Energy Ltd

Woodside Plaza, 240 St Georges Terrace, Perth, Western Australia 6000

bfillery@mech.uwa.edu.au

Early attempts to characterise the fracture response under thermo-elastic loads tried to draw a
correlation between the maximum tensile stress at the surface and the probability of failure; Cheng

[1] and Kingery [2]. Such formulations ignored the 1 r stress singularity present for thermo-
elastic fracture problems; Sih [3]; hence, future fracture mechanics investigations evaluated the
temperature distributions and stress intensity factors (SIF’s) associated with the geometric crack
discontinuity within simplified domains; Sekine [4], Sekine [5], Sumi [6].
More recently, increasing attention has been devoted to the severe thermo-elastic loads
associated with a thermal shock incident (TSI). Nied [7] discussed the thermal shock problem of an
edge cracked plate, Neid and Erdogan [8] a transient thermal stress problem for a circumferentially
cracked hollow cylinder, and Noda et al. [9], the thermal shock problem of a hollow circular
cylinder with a crack. In these examples, the method of superposition has been applied in
conjunction with an integral equation solution under equivalent crack face loads. Such a
formulation enables calculation of the transient SIF’s, though extends only to certain simplified
domains, and to the author’s knowledge, the applicability of the superposition technique under
these severe thermo-elastic loads is not known. It is clear that an additional technique is required to
analyse more complicated domains. Such a technique may be found within the finite element
methods.
In this paper, the accuracy of the superposition principle under severe thermo-elastic ‘shock’
loads will be discussed with reference to a robust finite element (FE) method. In recent years, the
finite element technique has been used extensively to formulate local weight functions for surface
cracks in plate and cylindrical domains; Shen and Glinka [10], Wang and Lambert [11]; though
little analysis has been applied to thermo-elastic conditions with the exception of Kokini [12].
Kokini explored a concept similar to that presented here, though control parameters were adjusted
to match the finite element results to those presented in [7], hence did not assess the validity and/or
accuracy of the assumptions disclosed within [7].
The results of the current finite element model will be extended to construct a quantitative
relation between the maximum SIF, crack length, and thermo-elastic material constants. The aim,
to present an easy to use ‘engineering’ relation that may be used in practical fatigue life assessment
without the need to solve integral based equations, for example weight function techniques. A
similar quantitative relation concept was presented in [9], though only explored a circumferential
crack in a hollow cylinder. In this instance, a more general solution form is sought to cover a range
of crack domains.
Future work will extend to include more extensive crack domains (i.e. 3D semi-elliptical) and
explore the influence of elastic-plastic material behaviour on the thermo-elastic fracture properties.
386 B. P. Fillery et al.

References
1. Cheng, C.M., J Am. Rocket Soc, vol. 21, 142, 1951
2. Kingery, W.D., J Am. Ceram Soc, vol. 38, 3, 1955.
3. Sih, G.H., J Appl. Mech, vol. 29, 587-589 1962.
4. Sekine, H., Engrg Fract. Mech, vol. 7, 713-729, 1975.
5. Sekine, H., Engrg Fract. Mech, vol. 9, 499-507, 1977.
6. Sumi, N., Katayama, T., Nucl. Engrg. Des, vol. 60, 389-394, 1980.
7. Neid, H.F., J Therm. Str., vol. 6, 217-229 1983.
8. Neid, H.F., Erdogan, F., J Therm. Str., vol. 6, 1-14, 1983.
9. Noda, N., Matsunaga, Y., Nyuko, H., Int. J. Pres. Ves. & Piping, vol. 42, 247-257, 1990.
10. Shen, G., Glinka, G., Theor Appl. Fract. Mech, vol. 20, 247-255, 1994.
11. Wang, X., Lambert, S.B., Theor Appl. Fract. Mech, vol. 23, 199-208, 1995.
12. Kokini, K., Engrg Fract. Mech, vol. 24, 843-850, 1986.
 2T14. Computational fracture mechanics 387

FAILURE PREDICTION OF IC INTERCONNECT STRUCTURES USING

COHESIVE ZONE MODELLING

B. A. E. van Hal, R. H. J. Peerlings, M. G. D. Geers and G. Q.Zhang1

Eindhoven University of Technology
Den Dolech 2, WH 4.115, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
1Philips Semiconductors

High Tech Campus, HTC60 / 3.20, P.O. Box 80021, 5600 JZ, Eindhoven, The Netherlands
B.A.E.v.Hal@tue.nl, g.q.zhang@philips.com

A major trend in the semiconductor industry is to reduce the feature sizes in the interconnect
structure of integrated circuits (ICs) from currently 120 nm to 90 nm and even to 65 nm. The
electro-thermal requirements on such microelectronic interconnects necessitate the introduction of
new materials with altered thermo-mechanical properties, such as copper for the wiring and low-k
dielectrics for insulation. Due to the combination of reduced feature sizes and new materials, the
thermo-mechanical reliability of ICs becomes an important design issue (see Mercado et al. [1]).
An important failure mode of the interconnect structure is delamination of the wiring from the
surrounding insulating material. The simulation of this failure mode forms a challenge for
currently available computational methods based on the finite element (FE) method. Several
interfacial cracks can occur simultaneously and their initial location is unknown. This limits the
use of standard linear elastic fracture mechanics.
We believe that interface damage mechanics is more appropriate. The inelastic material
response in the interface region is lumped into a plane in three-dimensional (3D) problems or into
a line in 2D problems. A cohesive zone (CZ) model, e.g. the one of Ortiz and Pandolfi [2],
describes the non-linear relation between the relative displacement of the two faces of the plane/
curve and the tractions between them. The interfaces with the considered thin low-k films show a
brittle behaviour. Chaboche, Feyel and Monerie [3] illustrate that the interface brittleness causes
bifurcations in quasi-static analyses, which are difficult to control with standard incremental-
iterative solution procedures. For insufficiently refined discretisations, these bifurcations are mesh
dependent. However, in case of low-k films, the high brittleness of the interfaces requires an
element size, which is so small that the FE mesh becomes too large for practical use, and therefore
it prohibits the use of CZ modelling in a straightforward manner.
The objective of this research is to develop CZ methods which allow us to predict delamination
in the interconnect structure despite the high brittleness of the interfaces. We believe that the
advantage of CZ modelling, that crack initiation follows from the analysis, outweighs the
disadvantage of the required fine FE mesh. The assumption is made that only a limited number of
interfacial cracks grow at the same time, which means that only the regions near a small number of
crack tips require a fine FE mesh. Therefore, the use of adaptive mesh refinement is proposed here
combined with adaptive mesh coarsening. The latter is inevitable for limiting the FE model size.
The mesh adaptivity procedures require the definition of two criteria, which control them. These
criteria follow from the damage levels of the CZ elements on the interfaces. Furthermore, a
strategy has been developed to map the state variables of the CZs from one instance of the mesh to
the next.
Several numerical simulations have been performed. The considered FE models are 2D plane
strain models of a representative part of the microstructure of the interconnect structure underneath
the bond pad. The deformations are assumed to be small. The material behaviour outside the
interfaces is linear elastic and the CZ model of Ortiz and Pandolfi [2] is used. An adaptive arc-
388 B.A.E. van Hal et al.

length control method allows the recovery of the snap-through and the snap-back behaviour of the
quasi-static solution and ensures that the traction–separation response of each CZ is sampled with a
sufficient number of evaluations. The outcome of the numerical simulations provides insight in the
failure mechanisms in the IC’s interconnect structure under thermo-mechanical loading. E.g., Fig.
1 illustrates that several interfacial cracks develop under loading and Fig. 2 illustrates the snap-
back behaviour of the corresponding loading path. This kind of simulations shows that the
explored procedures can assists engineers in the future to improve the thermo-mechanical
reliability of ICs.

FIGURE 1. Several cracks develop in representative part of interconnect structure.

FIGURE 2. Corresponding loading path with snap-back behaviour.

References
1. Mercado, L.L., Goldberg, C., Kuo, S.-M., Lee, T.-Y.T. and Pozder, S., In Proceedings of
2003 Electronic Components and Technology Conference, 2003, 1784-1790.
2. Ortiz, M. and Pandolfi, A., Int. J. Num. Meth. Engrg., vol. 44, 1267-1282, 1999.
3. Chaboche, J.L., Feyel, F. and Monerie, Y., Int. J. Sol. Structures, vol. 38, 3127-3160, 2001.
 2T14. Computational fracture mechanics 389

NON-LOCAL DAMAGE SIMULATION IN COMPOSITES USING CRACK

PROPAGATION AND MESH ADAPTIVITY

Frederik Reusch, Christian Hortig and Bob Svendsen

Department of Mechanical Engineering, University of Dortmund
University of Dortmund, D-44221 Dortmund, Germany
frederik.reusch@uni-dortmund.de

The numerical analysis of ductile damage and failure in engineering materials and metal matrix
composites is often based on a micromechanical description of the damage and failure process
(Gurson [1], Needleman and Tveergard [2], Tveergard and Needleman [3]). In heterogeneous
metal matrix composites, ductile crack extension occurs only in the ductile metallic phase, whereas
cracks of rigid inclusions and decohesion is not necessarily experimentally observed.
The failure process in the metal phase consists of the nucleation of voids resulting from rigid
second-phase inclusions, the growth and coalescence of voids up to final failure and
macroscopically observable crack growth.
Numerical studies in the context of Finite Element simulations of the damage and failure
behaviour of heterogenous materials demonstrates that the application of local damage models to
such problems and the numerical simulation of the initiation and propagation of localized damage
zones is not generally reliable and strongly mesh-dependent (e.g. Klingbeil and Brocks [4],
Mishnaevsky [5]). The numerical problems concern the global load-displacement response as well
as the onset, size and orientation of localized damage zones and thus to the reliability of the
obtained results.
One possible way to overcome these problems with is the application of so-called non-local
damage models. In particular, these are based on the introduction of a gradient type evolution
equation of the damage variable regarding the spatial distribution of damage and thus the
incorporation of a material length scale (Feucht [6], Ramaswamy and Aravas [7], Reusch et al.
[8,9]).
This work will present a recently developed non-local formulation of a damage model for
ductile damage at locally large deformations (Reusch et al. [8,9]) and its successful application to
the Finite Element simulation of crack initiation and propagation in two- and three-dimensional
heterogeneous microstructures. Such a non-local formulation of a damage model exhibits a
multifield problem, which needs a closer look on the formulation of possible criteria for the
preservation of the well-posedness of the underlying constitutive equations and thus the stability of
the deformation process (Benallal and Tvergaard [10], Liebe et al. [11], Reusch [12]). The
development and application of a criterion for loss of ellipticity (Reusch [12]) is presented and
accounts for the regularization of the solution obtained by the non-local damage model.
To this purpose, a recently developed strategy for mesh adaptivity is used, to prevent excessive
element distortions and to optimize the element distribution through problem oriented mesh
adaptivity with help of different remeshing criteria (Comi and Perego [13]).
Furthermore the regularizing effects of the non-locality of the damage evolution are
investigated and its effect on the stability of the numerical solution is presented compared to
results obtained with standard damage modeling techniques.
The results document quantitatively the effect of the delocalization of the model damage
process and the minimization of mesh-dependence on the characteristic dimension of the damage
process zone and the global structural response with help of the non-local damage model.
390 F. Reusch et al.

References
1. Gurson, A. L., J. Engng. Mat. and Tech., vol. 99, 2-15, 1977
2. Needleman A., Tvergaard, V., J. Mech. Phys. Solids, vol. 32, 461–490, 1984
3. Tvergaard, V., Needleman, A., Int. J. Sol. Structures, vol. 32, 1063–1077, 1995
4. Klingbeil, D. and Brocks, W., Verifikation von Schädigungsmodellen zur Vorhersage von
Risswiderstandskurven für verschiedene Probengeometrien und Materialien, Report Nr.
Br521/5, BAM, Germany, 1998.
5. L. Mishnaevsky Jr., et. al., Comp. Materials Science, vol. 16,133-143, 1999
6. Feucht, M., Ein gradientenabhängiges Gursonmodell zur Beschreibung duktiler Schädigung
mit Entfestigung, Institut für Mechanik, Technische Universität Darmstadt, Germany, 1998.
7. Ramaswamy, S., Avaras, N., Com.Met.Appl.Mech.Eng., vol. 163, 33–53, 1998
8. Reusch, F., Klingbeil, D. and Svendsen, B., Computational Materials Science, vol. 26, 219-
229, 2003
9. Reusch, F., Klingbeil, D., Svendsen, B., European J. Mechanics A/Solids, vol. 22, 779-792,
2003
10. Benallal, A., Tvergaard, V., J. Mech. Phys. Solids, vol. 5, 741-770, 1995
11. Liebe, T., Steinmann, P., Benallal, A., Comput. Methods Appl. Mech. Engrg., vol. 190, 6555-
6576, 2001
12. Reusch, F., Entwicklung und Anwendung eines nicht-lokalen Materialmodells
zur Simulation duktiler Schädigung, Dissertation, Fakultät Maschinenbau, University
Dortmund, Germany, 2003.
13. Comi, C., Perego, U., European Journal of Mechanics A/Solids, vol. 23, 615–632, 2004
 2T14. Computational fracture mechanics 391

ELASTIC WAVE MOTION IN A CRACKED, MULTI-LAYERED GEOLOGICAL

REGION UNDER TRANSIENT CONDITIONS

P. S. Dineva, T. V. Rangelov1 and G. D. Manolis2

Institute of Mechanics & 1Institute of Mathematics and Informatics
Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria
2Department of Civil Engineering,

Aristotle University, Thessaloniki GR-54124, Greece

gdm@civil.auth.gr, petia@imbm.bas.bg

A hybrid boundary element method (BEM) for transient problems in elastodynamics is developed
here [1] as a means for investigating ground motion phenomena in geological regions with
complex geometry, variable material properties and in the presence of both interface and internal
cracks. Two different aspects of the problem are considered, namely computation of (a) ground
motions in the form of synthetic seismograms that are manifested at the free surface of the
geological region as it is swept by a seismically-induced pressure wave and (b) evaluation of the
near crack-tip stress concentration field (SCF) that develops around cracks buried within the
deposit for the same type of loading. The present method combines both displacement and
regularized traction BEM in the Laplace transformed domain [2] for the crack-free and cracked
states, respectively, while the transient nature of the wave scattering phenomenon is reconstructed
through use of the numerical inverse Laplace transformation. Furthermore, plane strain conditions
are assumed to hold and the response of the geological region remains within the linear elastic
range. The basic strategy, whereby the aforementioned two states are superimposed, has been
successfully used in the past for problems in fracture mechanics. Following numerical
implementation of the hybrid BEM, validation-type examples serve to calibrate the methodology.
Finally, the method is used for solving the seismic response of a complex geological region so as to
reach some conclusions regarding the relative influence of various key parameters of the problem
(such as layering, surface canyon, crack interaction, etc.) on the scattered displacement field and
on the SCF. Further extension of the method to cover mildly inhomogeneous continua is also
planned [3].
A finite, cracked geological region with non-parallel layers and surface topography is shown in

Figure 1, where region : N contains a semi-circular canyon with radius r = A. The deposit
consists of a number of soil layers : i , i=1,2,…,N, and two types of cracks are present, namely

interface cracks such as FE of length a FE and internal cracks such as MN of length a MN . A

state of plane strain is assumed. The excitation is a displacement field produced by the passage of
transient P waves in : with two components along the principal directions as
& &
(u x , u y ) u 0 (sin Į i ,  cos Į j )f(t  (x sin Į  y cos Į) C p ) , where D is the

angle of wave incidence, u 0 is the wave amplitude and C P is the wave speed.

Preliminary results produced show that both scattered wave displacement field on the free
surface and stress concentration field near the crack-tips are sensitive to site conditions such as the
existence of surface relief, the presence of layers and cracks, and most importantly on interaction
effects due to the presence of multiple cracks. In sum, this type of work is useful in the field of
392 P. S. Dineva et al.

earthquake engineering, where a detailed image of local ground motions is of paramount

importance [4,5].

FIGURE 1. Layered geological region with surface irregularities and cracks.

Acknowledgment: The authors acknowledge financial support through NATO grant

EST.CLG. 980303.

References
1. Dineva P.S., Manolis G.D. and Rangelov, T.V., Transient seismic wave propagation in a
multilayered cracked geological region, Journal of Sound and Vibration, Vol. 273(1-2), 1-32,
2004.
2. Dineva P.S., P. Borejko P., Hadjikov L. and Ziegler F., Transient elastic waves in a half-
space: Comparison of the DBIE-method with the method of generalized ray, Acta Mechanica,
Vol. 115, 203-211, 1996.
3. Manolis G.D., Dineva P.S. and Rangelov T.V., Wave scattering by cracks in inhomogeneous
continua using BIEM. International Journal of Solids and Structures, Vol. 41 (14), 3905-
3927, 2004.
4. Zhang C. and Gross D. On Wave Propagation in Elastic Solids with Cracks. CMP,
Southampton, 1998.
5. Aliabadi M. and Rooke D. Numerical Fracture Mechanics. CMP, Southampton, 1991.
 2T14. Computational fracture mechanics 393

WOOD BEAM STRENGTHENED WITH GLASS/EPOXY COMPOSITE

SHEETS

G. E. Papakaliatakis, G-S. P. Diamantopoulos, P.A. Kalaitzidis and E. M. Marinakis

Department of Civil Engineering, Democritus University of Thrace
Vassilisis Sofias 12, Xanthi, 67100
gpapakal@civil.duth.gr, stefanosdiamantopoulos@yahoo.gr, pkalait@civil.duth.gr,
marinakisv@yahoo.gr

The technology of wood applications has a significant role in many engineering constructions.
Wood is an adaptable material with good proportion of strenght to weight ratio and has wide use in
applications of engineering constructions [Winandy J.E., Encyclopedia of Agricultural Science.
Orlando, FL: Academic Press: Vol. 4, 549-561,October 1994]. It is a construction material for
rafters, struts and other parts of the buildings [2]. It is important to increase the strength of beams
with big length or strengthen struts in old buildings. The strengthening can be achieved adhering
wood in his upper and lower surface with composite material (sandwich beams with core from
wood) or only in his lower surface. Typically, two strong and stiff skins that carry most of the in-
plane and bending loads are adhesively bonded to wood [3], [4], [5], [6].
In the present work a study of the mechanical behavior of a wood beam strengthened with
glass/epoxy composite sheets at the lower or upper and lower facing is presented. The composite
sandwich beam consists of a wood core with dimensions 20x25x200mm and glass/epoxy faces
with thickness 2mm stuck with polyurethane adhesive. The new values of mechanical properties of
strengthened beams that were observed, were compared with the beam without faces.
All the above specimens were tested with three-point bending loading and the experimental
load-deflection curves were presented. The finite elements programme ABAQUS was used for the
arithmetical calculation of stress and diplacement fields . Wood and glass / epoxy composite
material were inserted in the program ABAQUS as orthotropic materials which the properties are
set out in the TABLE 1.

TABLE 1. Material properties

The stress and displacement fields are presented graphically with contour curves. The
experimental load-deflection curves for all the above described cases are compared both with the
analytical solution of the deflection theory and with arithmetical solution of the programme
ABAQUS (FIGURE 1).
394 G. E. Papakaliatakis et al.

FIGURE 1. Load versus deflection of the different cases of the beams in the cross-middle span.

The convergence of experimental results with the corresponding arithmetical and analytical
solutions was satisfing. The developing normal stresses either in the core or faces, in the cross
middle span, are given in the TABLE 2.

TABLE 2. For load P=3400 N, normal stress for the core and facings at the middle cross section
(units in N/mm2)

From this table we observe reduction of maximum stresses at the core (wood) of the
strengthened beam compared to the wood beam.

References
1. Winandy J.E., Encyclopedia of Agricultural Science. Orlando, FL: Academic Press: Vol. 4,
549-561,October 1994
2. Rainforest Action Network, Rainforest Action Network ,San Fransisco, USA, June 2000
3. Bekisli B., Grenestedt J. L. ,International Journal of Mechanical Science, Vol.45,Issue
8,1327-1346,August 2003
4. Konsta-Gdoutos M. S. , Gdoutos E. E., Applied Composite Materials, Vol.12, 165-176, May
2005.
5. Triantafillou T. C., J. Mat. in Civ. Engrg., Vol. 9, Issue 2, 65-69, May 1997
6. Allen, H. G., Analysis and Design of Structural Sandwich Panels, Pergamon Press, London,
1969.
 2T14. Computational fracture mechanics 395

COMPUTATION OF DYNAMIC STRESS INTENSITY FACTORS USING

ENRICHED FINITE ELEMENTS

M. Saribay and H. F. Nied

Mechanical Engineering and Mechanics
Lehigh University
Bethlehem, PA 18015-3085
hfn2@lehigh.edu

Elastodynamic analysis of cracks is of critical importance for determining the ultimate strength of
structures subjected to rapid loading. In the case of sudden impact loading, dynamic stress intensity
factors are significantly higher than stress intensity factors obtained for structures subjected to
static loading. Thus, an efficient approach for computing transient stress intensity factors, e.g., an
approach that does not require creation of a special crack tip mesh, is of considerable value for
accessing the influence of complex time dependent loading. The enriched finite element approach
is a very effective technique for obtaining dynamic stress intensity factors for three-dimensional
fracture problems. This approach utilizes the asymptotic crack tip fields for direct computation of
stress intensity factors. Example three-dimensional solutions for dynamic fracture problems with
stationary crack fronts are presented in this study, demonstrating the spatial and temporal variation
of the stress intensity factors due to crack surface interactions with propagating elastic waves.
Details concerning the formulation of 3-D enriched crack tip elements for elastodynamic
problems are given by Saribay in [1]. The FEM formulation essentially combines the explicit/
implicit time integration algorithms given by Belytschko et al. [2] and the crack tip enrichment
procedure described by Ayhan and Nied [3]. The enriched elements contain additional unknown
degrees of freedom that correspond to the stress intensity factors for nodes that define the crack
front. Thus, the stress intensity factors are directly computed, along with the nodal displacements.
The element mass and damping matrices for the enriched elements need to be modified to account
for these additional degrees of freedom. The velocities and the accelerations for the enriched
elements, can be expressed in terms of the time rates of change of the stress intensity factors
interpolated along the crack front, permitting a consistent formulation of the element mass matrix
>M @ and damping matrix >C @ for the enriched crack tip elements.
A finite element mesh for a plate, with an internal crack, subjected to step loading is depicted
in Fig. 1. The geometry of this plate corresponds to a 2-D problem given by Chen [4] and solved
using a finite difference algorithm. With the appropriate out-of-plane constraints, the results from
the finite element formulation are essentially identical to those given in [4]. Figure 2 contains plots
of the normalized mode I stress intensity factors and compares the 3-D result with the plane strain
solution. The differences in the solutions are due to differences in the arrival time of the S, R and P
waves along the crack front. The transient stress intensity factors also display a spatial variation
along the crack front as shown in Fig. 2b. Interestingly, the location of the maximum stress
intensity factor along the crack front does not always lie on the plane of symmetry, but changes as
a function of time.
396 M. Saribay and H. F. Nied

(a) (b)
FIGURE 1. a) Rectangular bar containing an internal crack, subjected to suddenly applied axial
loads. b) Cubic crack tip element (32-noded hexahedron) showing orientation of local crack tip
coordinate system with respect to global coordinates [3].

FIGURE 2. a) Normalized mode I stress intensity factors (K ) for the central node on the crack
I dyn
front as a function of time for the internally cracked rectangular plate. b) Stress intensity factors
*
along crack front at t = 0.685e-05 s. K 1 K 1 d ym V 0 Sa.

References
1. Saribay, M., Dynamic Stress Intensity Factors for Cracks Using The Enriched Finite Element
Method, MS Thesis, Lehigh University, May 2005.
2. Belytschko, T., Wing, K.L., Moran, B., Nonlinear Finite Elements for Continua and
Structures, John Wiley & Sons, Ltd, England, 2000.
3. Ayhan, A.O., and Nied, H.F., Int. J. Num. Meth. Eng., Vol. 54, Issue 6, 899–921, 2002.
4. Chen Y.M, Eng. Fracture Mech., 7, 653-660, 1975.
 2T14. Computational fracture mechanics 397

PARTLY CRACKED XFEM INTERFACE FOR INTERSECTING CRACKS

J. L. Asferg, Ted Belytschko1, P. N. Poulsen and Leif Otto Nielsen

Department of Civil Engineering, Technical University of Denmark,
DK-2800 Kgs. Lyngby, Denmark
1Department of Mechanical Engineering, Northwestern University, Evanston, IL, US.

jla@byg.dtu.dk, tedbelytschko@northwestern.edu, pnp@byg.dtu.dk, lon@byg.dtu.dk

Throughout the last century intense research has been carried out regarding methods to determine
the ultimate strength of reinforced concrete structures. Today well documented methods are
available for estimating the ultimate strength of most reinforced concrete structures, however, most
of these methods requires the use of empirical factors and do not consider phenomena such as size
effects and reinforcement arrangement in detail.
Regarding reinforced concrete structures in the serviceability limit state the predictive
capability of existing methods of analysis is limited. Complex models dealing with the
serviceability limit state requires prediction of the complex cracking which takes places in the
concrete during loading.
Modelling of cracks in plain concrete has been a focus area in the research community since
the mid seventies where Hillerborg et al. [1] presented their fictitious crack model and Bažant [2]
proposed the concept of a crack band. Today several FEM codes have interface elements suitable
for discrete cracking and elements for smeared cracking. The use of interface elements however
requires the crack path to be known beforehand, while crack modelling applying the smeared
approach is not well-suited for modelling of localized crack growth.
Among the methods that allow modelling of discrete crack growth without knowing the crack
path beforehand is the extended finite element method - XFEM, Belytschko et. al. [4], Moes et. al.
[3]. XFEM has been applied to a number of different problems within the area of fracture
mechanics among which are cohesive cracking, Wells and Sluys [5], Zi and Belytschko [6].
Modelling of crack growth in plain concrete for benchmark tests such as three point bending and
four point shear bending was also considered in Asferg et. al. [7], applying a simplified concept for
the enrichment of the displacement field.
Modelling of reinforced concrete beams applying the FEM was first carried out by Ngo and
Scordelis [8]. Since then several approaches for modelling of the interaction between
reinforcement and concrete has emerged. Today the most widely used concept for modelling the
interaction between steel and concrete is the application of interface elements for the bond zone
e.g. Lundgren [9] .
The use of standard interface elements however poses some difficulties. Generation of more
complex models with multidirectional reinforcement is cumbersome and special attention is
required when ever two reinforcement bars are crossing each another. When the aim is to model
crack growth without knowing the crack path beforehand applying e.g. the XFEM concept for the
bulk concrete traditional interface elements are not applicable for the bond zone between
reinforcement and concrete.
The goal of the present research project is to develop a more feasible method for modelling of
reinforced concrete structures and a super element may be the final goal of the project. However
before such an element can be formulated an XFEM interface element for the bond zone has to be
formulated. One of the requirements to the XFEM interface element is that it shall be able to
handle intersecting cracks – longitudinal cracking along the reinforcement is initiated by cracks
398 J. L. Asferg et al.

crossing the reinforcement. Furthermore the interface element shall be formulated as an element
that is able to partly crack. Bond between concrete and reinforcement is in nature a 3D problem
and the confining pressure is one of the key effects for the stress transfer between concrete and
reinforcement. However initially the XFEM interface is considered in a plane version. c.f. Fig 1.

FIGURE 1: Development of crack in interface between concrete and reinforcement.

As initial test case for the XFEM interface a plane model of a concrete beam holding one
reinforcement bar is considered. A constitutive bond model is chosen from literature and the
results compared with results from the literature.

References
1. Hillerborg, A, Modéer, M, Petterson P-E. Analysis of crack formation and crack growth in
concrete by means of fracture mechanics and finite elements. Cem. Concr. Res., 1976; 6:773-
782.
2. Bažant, Z P. Instabillity, ductility and size effect in strain-softening concrete, ASCE Journal
of Engineering Mechanics; 1976 102: 331-344.
3. Belytschko T, Black T. Elastic crack growth in finite elements with minimal remeshing. Int.
J. Numer. Meth. Engng. 1999; 45(5):601-620.
4. Moes, J. Dolbow and Belytschko, T., “A finite Element Method for Crack Growth without
Remeshing”, Int. J. Numer. Meth. Engng., 1999; 46(1): 131-150
5. Wells GN, Sluys LJ. A new method for modeling of cohesive cracks using finite elements.
Int. J. Numer. Meth. Engng. 2001; 50(12):2667-2682.
6. Zi G, Belytschko T, New crack-tip elements for XFEM and applications to cohesive cracks.
Int. J. Numer. Meth. Engng. 2003; 57:221-2240.
7. Asferg, J.L., Poulsen, P.N., Nielsen; L.O. “ Cohesive crack modeling applying XFEM – fully
cracked element vs. partly cracked elements” Manuscript under preparation.
8. Ngo. D., Scordelis C., “ Finite element analysis of reinforced concrete beams”, 1967, ACI J.
Proc. 64, pp. 152-163.
9. Lundgren, K, “Three-Dimensional Modelling of Bond in Reinforced Concrete” 1999, Ph.D.
thesis, Publication 99:1, Div. of concrete, Chalmers University of Technology.
 2T14. Computational fracture mechanics 399

ON THE EVALUATION OF ELASTIC COMPLIANCE TENSOR DUE TO

GROWING MIXED-MODE MICROCRACKS

K. M. Abbasi and H. Schutte

Institute of Mechanics, Ruhr University of Bochum
D-44780 Bochum, Germany
kian@tm.bi.rub.de, henning.schuette@rub.de

Our aim is to determine the evolution of the elastic compliance tensor of a unit cell with a growing
mixed-mode microcrack in it. To be able to validate certain damage evolution laws, we determine
this so far missing evolution of the elastic compliance due to growing mixed-mode microcracks.
The matrix of the unit cell is considered to be linearly elastic homogenous and isotropic and
the crack is small compared to the unit cell, so the overall behaviour of the unit cell is linearly
elastic. This enables us to use linear elastic fracture mechanics (LEFM) concepts for the
propagation of the microcrack.
Finite element simulations have been performed to simulate the quasi-static, elastic growth of
microcracks. The models provide a general framework for mixed-mode linear elastic fracture
mechanics analysis under small strain assumptions.

FIGURE 1. Growth of a microcrack in a unit cell.

As the crack propagates, remeshing algorithms delete the mesh, extend the crack and mesh the
cube according to the new crack configuration.
The propagation of the microcrack is governed by the principle of maximum driving force [1],
which is a direct consequence of the variational principle of a body containing a crack. According
to this criterion, a crack grows in the direction of maximal driving force, and it reads
2
1 Q 1
di [( K I ) 2  ( K II ) 2 
( K III )2 ]
E 1 Q . (1)
Where

K D FDE ( I ) K E
(2)

and FDE is a matrix whose elements are universal functions of f [4], f being the kinking angle, and
K E are the so-called stress intensity factors (SIFs) prior to the crack growth[4].
400 K. M. Abbasi and H. Schutte

The initial crack is taken as an inclined elliptical or penny-shaped crack, for which analytical
expressions for SIFs and the elastic compliance tensor are available. Rotation of the cracks around
their centre points enables the simulation of possible modes of crack propagation.
Many damage mechanics models for quasi-brittle materials are based on the reduction of
stiffness due to elliptical crack or penny-shaped microcracks in the material. With our approach it
becomes possible to validate the evaluation of damage and so the evolution of the compliance
tensor resulting from different models.

FIGURE 2. Reduction of stiffness in the loading direction.

Verification and validation studies are undertaken to compare the resulting SIFs and the
components of the compliance tensor from the simulation with the available analytical results.

References:
1. Le, K. C., Schütte H. and Stumpf H., Archive of Applied Mechanics, vol. 69(1), 337-344,
1999
2. Mehrabadi, M. M. and Cowin, S. C., Mech. Appl. Math., vol. 43(1), 15-41, 1990
3. Leblond, J. B., Int. J. Solids Structures, vol. 25(11), 1311-1325, 1989
4. Amestoy, M, Leblond, J. B., Int. J. Solids Structures, vol. 29(4), 465-501, 1992
5. Wu, C. H., Journal of Elasticity, vol. 8, 183-205, 1978
6. Nemat-Nasser, S., Hori, M., Micromechanics: overall properties of heterogeneous materials,
North-Holland, Amsterdam, Holland, 1993.
 2T14. Computational fracture mechanics 401

ON THE PROBLEM OF DETERMINATION OF SAFETY FACTORS FOR

MACHINE-BUILDING PARTS USING THE FINITE ELEMENT
COMPUTATIONS

L. B. Getsov, B. Z. Margolin and D. G. Fedorchenko

SPbSPU (St. Petersburg), CRISM “Prometei” (St. Petersburg), SNTK (Samara)
guetsov@online.ru
The main principles of normalizing the safety factors on local static and low-cycle fatigue strength
considered in this paper were substantiated experimentally (the report in preparation).
The process of rupture at static loading may generally be of three types: a) caused by
exhausting short-time plasticity of material; b) caused by creep; c) brittle.
It is evident that the differentiation of safety factors depending on the type of rupture should be
implemented in the normalization of local stresses. It is clear that the greatest safety factor value
should be taken in the case of brittle fracture, with its strength values characterized by the
maximum scatter of material parameters values.
Static strength of plastic materials: The following approaches are expedient to apply for
evaluation of safety factors on local stresses: 1. The application of a widespread model of
kinematic hardening is completely justified for solving many practical problems. However, the
optimum is SSS computation providing for the choice of a plasticity model depending on the
material analyzed and ways of loading, in accordance with the multimodel approach concept. It
should be also noted that in using deformation curves, strains will be related to a current full strain.
2. Static strength of plastic materials should be evaluated by exhaustion of the ultimate material
plasticity *, which in its turn depends on the loading rate or time. Incidentally, one should discern
the ultimate states under the conditions of intragranular and intergranular rupture. The
intragranular rupture is characterized by the absence of the dependence of ultimate strains on
loading rate; at the same time under the condition of intergranular rupture, ultimate strains
diminish with the decrease of loading rate. 3. If local strength is evaluated under the condition of
short-time plastic strain, the safety factor on strains */p (p – plastic strain) shall be not less than 2.0,
* being defined with account of the stress state rigidity by the following formulas:
İ * = İ p u l t 1 . 7 e x p ( - 1 . 5 V / V i ) (1a), İ * = İ p u lt K eV i
2
/ 3 ( V i V ) (1b),

which give a conservative estimation of plasticity. Here pult is the ultimate strain (deformability) at
short time tension, and Ke, the characteristic of material state (Ke=1at the brittle state; Ke=1.2 at
plastic state); V is the mean stress value. The value of p is defined using elasto-plastic
computation, an appropriate plasticity model, and the lower strain envelop curve. In this case, the
safety factor on stresses should not be lesser than 1.2-1.4. The evaluation of rupture situation of the
parts operating in the creep condition is implemented with the use of the ultimate strain value,
which depends on temperature, time and rigidity of stress state. Therefore, just as in the case of
normalizing the safety factors on static strength of parts from plastic materials, the introduction of
FEM computations and more exact knowledge of SSS in stress concentration locations does not
provide the basis for correcting the values of safety factors in the condition of creep. There is no
need to use the modern methods of stress computation in this case. The safety factors in the
condition of creep should be defined by using the crack initiation criteria. Accounting for
nonstationary situation by the application of computations with the formulas of linear summation
of damages (in deformation or time interpretation), as has been shown by numerous researchers,
allows us to obtain the conservative estimation, on the condition that the sum of damages is taken
equal to 0.87.
402 L. B. Getsov et al.

At present, for evaluating the cyclic strength safety factors, various methods were suggested
for determining the local strength safety factors that may be conditionally divided into five groups:
computational for a rigid cycle; computational for a general situation; computational-experimental;
based on the theory of adaptability; based on deformation criteria. For cyclic loading in general
case, when unilateral accumulation of strains (characteristic for a mild cycle and generally called
“ratchetting”) and stress variation (characterictic for a rigid cycle) take place, different approaches
to the evaluation of cyclic strength safety factors may be applied. Among all known strength
characteristics of material, life time under cyclic loading mostly depends on the constructional,
technological, metallurgical and operational factors. Therefore, it is expedient to carry out the
evaluation of the life time under cyclic loading for constructions basing on the results of testing
specimens and construction components, with account of all abovementioned factors. The main
operational factors affecting the life time of a part under cyclic loading are temperature, holding
duration at maximum loads and temperatures, cyclic asymmetry, superposition of the high-
frequency component upon the low-frequency variation of loads. The conducting of tests within
the whole range of operational loads is rather laborious task. Therefore is most urgent to develop
the methods based on conventional tests of specimens and allowing the evaluation of life time of
the constructions subjected to complex loading in operation. In the general case of low-cyclic
loading, the material damages may be computed with the use of deformation or energy criteria of
rupture. Here, for computing the kinetics of stress-strain state both for complex noncyclic loading
and for cyclic loading with altering loading parameters, instead of the number of cycles n (or the
number of semicycles k), it is expedient to use the relations of Odquist’s type as parameters of
state. These relations are expressed by the following formulas:

Ȝ1= ³ dH p  H p dH p (2 / 3d H p ij dH p ij ) 0 .5 H (2 / 3d H dH ) 0 .5
; p p ij p ij
(2)

Ȝ2 ³ dp  p dp (2 / 3 dp ij dp ij ) 0.5 p ( 2 / 3 p ij p ij ) 0 .5
; (3)

' Ȝ 1 = Ȝ (k ) - Ȝ (k -1 ) t 0 , where k – the ordinal number of a semicycle.

The increment of nonelastic strains (dne) and the value of nonelastic strain intensity (ne) are defined by
ne
formulas : dijne=dpij+ dpij; H ( 2 / 3 d H inj e d H ne
) 0 .5 (4)
In the case of creep, when stresses are known, the accumulated creep strains are defined by their separation
from nonelastic strains.
The methods of adaptability computation determine the cyclic strength safety factors for the general case:
in the conditions of sign-variable flow and progressing deforming. Ultimate material characteristics for a sign-
variable flow are the following:
The methods of adaptability computation determine the cyclic strength safety factors for the general case:
in the conditions of sign-variable flow and progressing deforming. Ultimate material characteristics for a sign-
variable flow are the following: ı s – half-value of cyclic yield strength S0.4 in a stabilized cycle with the
tolerance on plastic strain amplitude within a cycle equal to 0.4%. In the case of the presence of stress
concentrators: ı s = E H ( N )V H (N ) where H ( 1 ) – semiamplitude of full strain corresponding to the
appearance of low-cycle fatigue macrocrack during N cycles; V H (N )
– complying with H ( 1 ) on the

isochronous cyclic strain curve. In the case of creep in one of the semicycles – ı s =S0.4 c - 0.5S0.4 , where
S0.4 c – cyclic yield strength on condition of the creep presence. For progressing deforming, as ultimate

characteristics are taken ı s = ı Ǻ – for transitional modes and ı s = ı LTS ( t , ¦ ' W ) for stationary

modes. Here ı LT S – long-time strength complying with the total duration of the mode, during all life time.
 2T14. Computational fracture mechanics 403

DYNAMIC EXPLICIT CELL MODEL SIMULATIONS IN POROUS DUCTILE

METALS

L. Siad and M. O. Ouali

Groupe de Mécanique, Matériaux et Structures (E.A. No 2617)
Université de Reims Champagne Ardenne
PHT du Moulin Le Blanc
7 boulevard Jean Delautre
F-08000 Charleville-Mézières, France.
larbi.siad@univ-reims.fr

The problem of ductile fracture is a complex phenomenon due to various factors such as inelastic
behaviour, large deformations, stress gradient … and has received considerable attention in recent
years. For example, the occurrence of ductile fracture by hole growth and coalescence is the
limiting factor in many metal-forming operations and consequently the ability to predict ductile
fracture would allow modification of a production process and so to reduce the risk of failure
occurring. Detailed micro-mechanical analysis for characteristic cell unit models with known
porosity have been an important tool in the understanding of the influence of porosity in ductile
materials. Indeed, cell models are usually considered as a structure in microscale within the
material and the solutions obtained for the cell model can be used to rationally link microscopic
properties and mesoscopic quantities. This approach allows to set up a general macroscale
constitutive law for a damaged material in the frame of continuum mechanics. In this context,
detailed cell model studies available in literature have ascertained that dilatant plasticity model
particularly suitable to model porous ductile metals are the well-known micromechanically based
model proposed by Gurson [1] and phenomenologically extended by Tvergaard and Needleman
[2] [3].
On the other hand, in the last years, applications of dynamic explicit time integration schemes
to problems involving high nonlinearity has increased, notably to those for which material failure
occurs. This paper deals with the study of a quasi-static axymmetric cell behaviour by the finite
element method using ABAQUS/Explicit code taking full account of finite deformation effects.
The analyses are based on an axisymmetric unit cell model with special boundary conditions
(Tvergaard [2] , Koplik and Needleman [4], Brocks et al. [5]), which allow for a relatively simple
investigation of a full three dimensional array of voids, without having to solve the full 3D
numerical problem. The influences of the initial porosity, the stress triaxiality ratio, and two void
shapes (spherical and spheroidal) on the cell behaviour are investigated. The numerical quasi-static
solutions are obtained by a full transient analysis of the equations of motion, in which the loading
is applied so slowly that the quasi-static solution is well approximated. The aim of the analysis is to
find the overall stress-strain behaviour of the cell model on the 'mesoscopic' level that represents
the behaviour of a volume element on the macrospcopic level. The calculations have been carried
out at constant overall stress triaxialities and a special technique has ben used to maintain the
triaxiality constant. The deformations of the cell are monitored by means of the displacement of
one of its corner to which a linear elastic truss is tied. The loading is strain controlled in such a way
that the stress triaxiality ratio retains a constant prescribed value This requirement is ensured by
adapting the displacement of one of the truss edges during the deformation process of the cell
(Leblond and Siad [6]). For comparison purposes, the material parameters adopted are taken from
literature (Koplik and Needleman [4], Pardoen and Hutchinson, [7] [8], Benzerga et al., [9]).
To validate and check the accuracy of the method of analysis, a comparison of the obtained
results using both quasi-static explicit and implicit finite element analysis along with available
404 L. Siad and M. O. Ouali

solutions in literature has been carried out. Results regarding the fracture initiation site, the
evolution of the damage variable field are illustrated by contour plots from which it can be seen
that damage starts developing either inside the cell or on its external surface, both on the ligament
depending on the initial porosity and the stress triaxiality ratio. Particular attention is also devoted
to the role of localized flow and to evolution of the void shape.
Extension of the numerical model to cope with important situations where the coalescence
parameters (more particularly the situation for which the matrix material is porous) and/or the
nucleation of new voids have to be accounted for are possible but require considerable additional
work. Furhermore, research towards the modelisation of the void shape effects based on the
Gologanu-Leblond-Devaux model (Gologanu et al., [10] [11]) including dynamic effects is under
progress.

References
1. Gurson, A., J. Engng. Mater. Technol., vol. 99, 2-15, 1977.
2. Tvergaard, V.., Int. J. Fracture., vol. 18, 237--252, 1982.
3. Tvergaard, V. and Needleman, A., Acta Metall., vol. 32, 157-169, 1984
4. Koplik, J. and Needleman, A., Int. J. Solids and Struct., vol. 24, 835-853, 1988
5. Brocks, W. , Sun, D.-Z. and Hönig, A., Int. J. Plasticity, vol. 11, 971-989, 1995.
6. Leblond, J.-B., and Siad, L., unpublished work, 2003.
7. Pardoen, T. and Hutchinson, J.W., J. Mech. Phys. Solids, vol. 48, 2467-2512, 2000.
8. Pardoen, T. and Hutchinson, J.W., Acta Materialia, vol. 51, 133-148, 2003.
9. Benzerga, A., Besson, J. and Pineau, A., Acta Materialia vol. 52, Part I, 4623-4628, Part II,
4639-4650, 2004.
10. Gologanu, M. and Leblond, J.-B., J. Mech. Phys. Solids, vol. 48, 1723-1754, 1993.
11. Gologanu, M., Leblond, J.-B., and Devaux, J., J. Engng. Mater. Technol., vol. 116, 290-297,
1994.
 2T14. Computational fracture mechanics 405

NUMERICAL EVALUATION OF ENERGY RELEASE RATES FOR

BIMATERIALS INTERFACE CRACKS

M.Belhouari, B.Bachir Bouiadjra, B.Boutabout and K.Kaddouri

Department of Mechanical Engineering, University of Sidi Bel Abbes
BP 89, Cité Ben M’hidi, Sidi Bel Abbes, 22000, Algeria
belhouari@yahoo.com

Bimaterials are extensively used in many engineered material systems, such as composite
structures, electronic packaging, and thin film construction. Accurate stress intensity factor and
strain energy release calculations are essential in the prediction of failure and the calculation of
crack growth rates in these structures.
The crack problems with interfaces in dissimilar materials are of paramount importance for
many micromechanics and numerical fracture mechanics. The complexity of interfacial failure
mechanism has caused researchers to devote a great deal of attention to study the selection of crack
growth path. Generally, fracture along and adjacent to bimaterial interfaces has several
morphological manifestations. In some cases the fracture defect into the interface, while in others
fracture penetrate the interface into another material [1-6].
In present work the finite element methods is used to calculate the strain energy release rates at
the interface crack. The effects of the crack length and the orientation of the interface were
highlighted as well as the effects of the elastic properties of two bonded materials.
In figure 1 are represented the variations of the energy release rate according to the Young
moduli ratio E 1 / E 2 for different crack length. This figure shows that whatever the length of the
crack, the fracture energy at crack tip decreases brutally from E 1 / E 2 < 10; beyond, this value,
and the energy does not vary practically with the ratio. The increase of the crack length involves an
increase in its fracture energy.
Figure 2 illustrates the variation of the normalised energy release rate G / G 0 according to the
of dephasing angle \ for a/w = 0,1. One can observe that the fracture energy of the interfacial
crack strongly depends on the value of the parameter \ The values of this parameter lie between
0 and 90° showing that the crack propagation is in mixed mode (opening + shearing). For \ = 45°
the energy release rate grows exponentially with the parameter \ whereas for values lower than
45°, this variation is weak.
Figure 3 presents the variation of the fracture energy at the interfacial crack according to the
ratio E 1 / E 2 for various angles of inclination of the interface. The fracture energy decreases
gradually when the angle of inclination of the interface of the bimaterial increases.
406 M. Belhouari et al.

FIGURE 1. Variation of G / G0 vs E1 / E2. FIGURE 2. Variation of G / G0 vs \.

FIGURE 3. Variation of G / G0 vs E1 / E2.

References
1. He, M.Y. and Hutchinson, J.W., Int J Solids Struc., vol. 25 ,1053-67, 1989
2. He, M.Y. and Hutchinson, J.W., Journal of Applied Mechanics, vol. 56, 270-278, 1989
3. Erdogan, F. and Biricikoglu, V., Int J Engrg Sci., vol. 11, 745-66, 1973
4. Evans,A.G., Dalgleish,B.J., He,M.Y.and Hutchinson, J.W.,Acta Metall, vol.37,3249-54, 1989
5. Evans, A.G. and Hutchinson, J.W., Acta, Mat., vol.37, 909, 1989
6. Rice, J.R. and Sih, G.C. Journal of Applied Mechanics, vol.32, 418-423, 1965.
 2T14. Computational fracture mechanics 407

INCLUSION EFFECT ON THE PLASTIC ZONE SIZE IN CONFINED

PLASTICITY

M. Benguediab, M. Elmegueni, M. Nait-Abdelaziz1 and A.Imad1

Department of Mechanical Engineering, University of Sidi Bel Abbes
BP 89, Cité Ben M’hidi, Sidi Bel Abbes, 22000, Algeria
1Ecole Polytechnique Universitaire de Lille (France)

benguediab_m@yahoo.fr

In understanding fatigue properties of metallic alloys, of outmost importance is the precise

characterization of the plastic deformation of the material in the vicinity of the crack tip. The
plastic behaviour near the tip of stationary growing fatigue crack in engineering materials has been
intensively studied using classical plasticity theory based on the Von-mises yield criterion and the
associative flow rules [1-2]. It is known that micro defects such inclusions are the possible zone of
crack initiation. Several researches have been made in order to analyse the interaction effect
between a crack and inclusion using the LEFM concepts [3-5]. The studies related to this effect of
interaction for a elastic plastic behaviour are not very current in the literature.
The aim of the study is to analyse the effect of presence of inclusions in metallic cracked plate
on the plastic zone size at the crack tip using the finite element method. The study relates to the
case of small scale plasticity. This case was selected because we estimate that an inclusion does not
affect the plastic zone in large scale plasticity because of its low dimension.
The model used consists of a plate in aluminium alloy 2024 subjected to tensile load. The
plate contains an edge crack with length 30 mm and near the crack one supposes the existence of
an inclusion in oxide of aluminium with circular shape having a radius of 200 µm. The material
constituting the matrix has an elastic-plastic behaviour and that of inclusion is elastic linear. The
ratio between the Young modulus of inclusion and the matrix is about 5. The geometrical model is
idealised by quadrilateral elements with eight nodes. Special quarter node elements are implanted
around the crack tips.
As examples of the obtained results, figures 1, 2 and 3 presents the computed plastic zone
contour for d/a = 1.25, 8 and 12,5 % compared with the case without presence of inclusion. The
parameter d is the distance between the crack tip and the interface inclusion-matrix and a is the
crack length. It can be seen that when the inclusion is very close to the crack tip (d/a =1.25 %), the
radius of the plastic zone decreases. There is thus an effect of thinning of this zone. This behaviour
can be explained by the fact that the presence of the inclusion, of which the materials is harder,
causes the increase of the stresses in the vicinity of the crack, which will involve a reduction of
the size of the plasticized zone.
When inclusion is not in the plastic zone but it is located at a distance very close to this zone
(d/a=8%). One can note that the radius of the plastic zone increases compared with the case
without presence of inclusion (figure 2). That is due to the fact that inclusion being the seat of
raised stress fields, the stresses in the part located upstream of inclusion cause plastic deformations
in metal. This effect involves This effect involves the increase in the radius of the plastic zone.
When inclusion is far from the plastic zone d/a=12.5%, the effect of interaction disappears and
the plastic zone has a classical form which corresponds to the case without presence of micro
defect.
408 M. Benguediab et al.

FIGURE 1. Contour of plastic FIGURE 2. Contour of plastic

zone for d/a = 0,0125 zone for d/a = 0,05

FIGURE 3. Contour of plastic zone for d/a = 0,125

References
1. Benguediab,M., Belhouari, M. and Ranganathan,N., Jour of Mater. Science and Techn., vol
9, 134-141., 2001
2. Kuang , J. and Chen, Y., Engineering Fracture Mechanics, vol 55, 869–881,1996
3. Lambert, D.J. and , Parks, D.M., Mech of Meterials, vol32,43-55, 2000
4. Xiao, Z.M., Mech of Materials, vol 25, 263-272, 1997
5. Murakami, Y., Proceeding of ICF 11, Editor J.Petit, 31-52, 1996
 2T14. Computational fracture mechanics 409

MODIFIED KEY-CURVE-METHOD FOR DETERMINATION OF DYNAMIC

CRACK RESISTANCE CURVES

U. Muhlich, A. Emrich and M. Kuna

TU Bergakademie Freiberg
Institute of Mechanics and Fluid Dynamics
Lampadiusstraße 4, Freiberg, 09599, Germany
Uwe.Muehlich@imfd.tu-freiberg.de

Dynamic J-'a curves might be constructed by means of the results of various low-blow tests,
where the experimental settings of the instrumented Charpy-V test are chosen in order to assure
significant crack growth due to impact loading on the one hand and in order to avoid complete
rupture on the other hand. While the final crack advance 'af can be measured for every low-blow
test, the corresponding values for J have to be estimated in general from the load-deflection curve
using standard concepts of conventional elastic plastic fracture mechanics. The experimental
effort could be significantly reduced if a single specimen test method were used instead of this
multi-specimen testing. The aim of the present work is to qualify the Key-Curve-Method (KCM),
originally proposed by Ernst et al. [1] for quasi-static fracture tests, for this purpose.
The idea of the KCM is to estimate the instantaneous crack length a as a function of the
deflection s directly from the load-deflection record. Eberle et al. [2] have shown by means of
numerical simulations that the corresponding value for J can be determined according to the
estimation given in the ESIS-P2 standard [3].
Experimental data obtained from dynamic fracture tests with side-grooved specimens made of
nodular cast iron GJS-400 are taken as reference. Finite element simulations of the fracture tests
have been performed, where a continuum damage model (CDM) was employed in order to
simulate crack growth numerically. The continuum damage model considered here was first
proposed by Gurson [4] and modified among others by Tvergaard [5], Tvergaard and Needleman
[6]. Here, an extension of this model was used which incorporates rate dependent yield and
adiabatic softening. The numerical calculations were carried out with the aim to possess an
equivalent numerical simulation for every considered experiment which offers the advantage that
in contrast to the experiment the instantaneous crack length can be monitored. Furthermore, the
corresponding value of J can be calculated directly according to its theoretical definition.
Plane strain simulations were used here because they predict sufficiently well the behaviour of
side-grooved Charpy specimens as shown in [2]. Furthermore, fully explicit analysis confirms that
inertia effects can be neglected because the deflection rates considered here lie between 0.6 m/s
and 1.2 m/s.
410 U. Muhlich et al.

FIGURE 1. Numerically obtained crack resistance curve versus estimation based on KCM

Most of the material parameters of the CDM have been determined from static and dynamic
tensile tests, whereas the critical void volume fraction and the height of the finite elements within
the ligament were chosen in order to fit the load-deflection curve of one Charpy-V test with
complete rupture. Then the low-blow tests have been simulated in order to validate the numerical
calculations by means of the following criteria: the load-deflection curve and the final crack length
predicted by the numerical simulation must fit the corresponding experimental results using a
unique set of material parameters for all low-blow tests.
Finally, the J-'a curves obtained directly by finite element simulations have been compared
with their corresponding estimates based on KCM. One example is given in Figure 1. which
indicates that for the considered material and the applied testing conditions the estimation of the
crack resistance curve based on KCM is in good agreement with the direct computation.

References
1. Ernst, H., Paris P.C. and Landes, J.D., Fracture Mechanics, ASTM STP 743, edited by R.
Roberts, American Society for Testing and Materials, 1981
2. Eberle, A., Klingbeil D. and Schicker, J., Nuc. Eng. Design, vol. 198, 75-87, 2000
3. ESIS-P2, ESIS procedure for determining fracture behaviour of materials, 1992
4. Gurson, A.L., J. Eng. Mat. Tech., Trans. ASME, vol. 99, 2-15, 1977
5. Tvergaard , V., Int. J. Frac., vol. 17, 389-407, 1981
6. Tvergaard, V., Needleman, A., Acta Metall, vol. 32, 157-169, 1984
 2T14. Computational fracture mechanics 411

A COUPLED COMPUTATIONAL FRAMEWORK FOR DUCTILE DAMAGE

AND FRACTURE

Ron H. J. Peerlings, Jesus Mediavilla1 and Marc G. D. Geers

Department of Mechanical Engineering, Eindhoven University of Technology
PO Box 513, 5600 MB Eindhoven, The Netherlands
1Netherlands Institute for Metals Research

PO Box 5008, 2600 GA Delft, The Netherlands

R.H.J.Peerlings@tue.nl, J.Mediavilla@tue.nl, M.G.D.Geers@tue.nl

Metal forming processes generally introduce a certain amount of damage in the material being
formed. Predictions of the damage formation and growth in a series of forming steps may assist in
optimising the individual operations and their order. This is particularly true for operations such as
cutting and blanking, which rely on the nucleation of damage and cracks in order to separate
material. The precise moment and location of crack initiation and the trajectory followed by the
crack(s) have an important influence on the quality of the products resulting from these processes.
Since the nucleation and growth of cracks may be influenced by damage induced in previous
forming steps, simulations require an integral approach towards damage and fracture. For this
purpose we have developed a coupled damage–fracture framework, which uses a nonlocal
continuum damage approach to model the evolution of material damage and full remeshing to trace
the crack growth.
In the continuum damage modelling, a scalar damage variable is used which models the
influence of micro-defect evolution by degrading the elastic as well as the plastic response of the
material in an isotropic fashion. The evolution of this damage variable depends on the plastic strain
rate and the stress triaxiality; the latter influence models the well-known asymmetry of ductile
damage growth in terms of tensile/compressive hydrostatic stress. As a result of the damage
evolution, the local load-bearing capacity of the material at some stage starts to gradually decrease
with plastic strain. In order to avoid pathological localisation and mesh-dependence issues due to
this strain-softening, a nonlocal formulation of the damage growth is employed which is based on
that proposed Geers [1]. This formulation uses an additional partial differential equation – next to
the equilibrium equation – to guarantee that the damage process takes place in a finite volume and
it thus dissipates a finite amount of energy. It can be shown to introduce a strong nonlocality,
similar to the integral formulation used e.g. by Leblond et al. [2], but can be implemented in a
more efficient manner using a two-field discretisation [1]. The length scale associated with the
nonlocality must be related to the microstructure, e.g. to the average void spacing.
Crack initiation is predicted when the continuum damage variable becomes critical somewhere
in the domain and the local load-bearing capacity thus is completely exhausted. A transition to a
discontinuous description is then made by inserting the crack in the geometry of the problem, fully
remeshing the updated problem domain, transferring the relevant history data from the previous
mesh, and continuing the analysis on the new mesh. These operations are repeated each time when
the damage field becomes critical in front of the crack tip or elsewhere on the domain. The latter
situation implies the initiation of a new crack, whereas the former results in the propagation of an
existing crack. Crack propagation is thus driven by the growth of damage in a process zone in front
of the crack tip and no separate fracture criteria are required (cf. the so-called local approach to
fracture). A finite crack growth rate is ensured by the use of a nonlocal damage formulation as
outlined above.
412 R. H. J. Peerlings et al.

Computationally, (locally) making the transition from a continuous damage description to a

discontinuous, discrete crack is quite demanding. Migrating from one discretisation to the other is
facilitated by an Updated Lagrange implementation of the coupled damage–plasticity modelling,
cf. [1]. When crack initiation or propagation is predicted, the current problem geometry is
extracted from the (updated) finite element discretisation and the new crack segment is inserted in
it. A standard remesher is then used to discretise the new problem geometry. Remeshing is also
used – even before crack initiation – in order to avoid element distortion and to accurately capture
steep damage gradients; a damage-rate error indicator is used for this purpose. Crucial for the
numerical stability of the simulations is the proper transfer of state variables from one
discretisation to the next. If this transfer is not done carefully, inconsistencies between the different
fields may arise, resulting in poor convergence or even divergence of subsequent loading
increments. We have found this to be particularly true in the presence of damage and cracks. A
dedicated transfer–crack opening algorithm has therefore been developed to ensure robustness of
the simulations.

FIGURE 1. Evolution of the ductile damage variable Zp and the resulting crack propagation in a
blanking simulation; shown is the zone between punch (top left) and die (bottom right).

Fig. 1 shows an application of the coupled ductile damage–fracture framework to blanking of a

sheet metal. The diagram shows the distribution of damage, Zp, in the shear zone between punch
and die at four subsequent stages of the punch displacement. As the damage evolves, cracks are
initiated at the punch and at the die, which subsequently propagate towards each other. Effects
which are observed in practice, such as a smaller crack length for smaller clearances between
punch and die, are properly captured by the modelling.

References
1. Geers, M.G.D., Comp. Meth. Appl. Mech. Engng., vol. 193, 3377–3401, 2004.
2. Leblond, J.-B., Perrin, G., Devaux, J., J. Appl. Mech, vol. 61, 236–242, 1994.
 2T14. Computational fracture mechanics 413

MARBLE DISCS UNDER DISTRIBUTED LOADING: THEORETICAL,

NUMERICAL AND EXPERIMENTAL STUDY

Ch. Markides, E. Sarris1, D.N. Pazis, Z. Agioutantis1 and S. K. Kourkoulis

Department of Mechanics, National Technical University of Athens,
5, Heroes of Polytechnion Avenue, 157 73 Zografou Campus, Athens, Greece
1Department of Mineral Resources Engineering,

Technical University of Crete, 731 00, Hania, Crete

stakkour@central.ntua.gr, zach@mred.tuc.gr

The stress field developed in a cylindrical disc under the influence of a compressive load
distributed uniformly along a predefined portion of its periphery, corresponding to an angle 20
(Fig.1), is studied in the present paper analytically, numerically and experimentally.
For the theoretical solution of the problem it is assumed that the material of the disc behaves
isotropically and linearly elastic. The stresses are calculated taking advantage of the solution
introduced by Muskhelishvili for the 1st fundamental plane problem of linear elasticity. For this
problem the complex potentials 1 and 1 are obtained through the following equations:

Z (s ) "
f s  M 1 (s )  M 1c (s )  \ 1 (s ), F "  i ³  X n  Y n d " (1)
Z c (s ) 0

where (Xn+iYn) is the force applied per unit length and unit depth on the loaded portion of the
cyclic boundary. For simplification the conformal mapping of the actual disc on the unit radius
disc was used (z=()=R), as it is seen in Fig1. This disc represents a simply connected subset of the
complex plane within which the complex potentials are holomorphic functions. Using the Cauchy
type integrals with holomorphic weight functions together with the well known relations:

FIGURE 1. The geometry of the problem and the definition of symbols

V rr  V -- 4 Re ) (z), V --  V rr  2i V r- 2 ª¬ z ) c  z   <  z  º¼ e 2i- (2)

one obtains the analytic expression for the stress tensor components in polar coordinates.
As a second step the problem was solved numerically using the Finite Element Method and the
MSC.Marc-Mentat code [2]. In order to simulate the tests in a more accurate manner both the
specimen and the device used for the application of the load (Fig.2a) were modelled and contact-
type elements were introduced for the description of the interface between the marble specimen
414 Ch. Markides et al.

and the metallic grips of the device. Coulomb type friction was assumed with a coefficient of
friction which represents a friction angle of about 22 deg (no separation was included).

FIGURE 2: The model of the numerical analysis (a) and the experimental device (b).

The experimental procedure included a series of experiments with cylindrical specimens made
from Dionysos marble, which were subjected to compression with the aid of the device shown in
Fig.2b. Dionysos marble is a white color stone, composed of 98% of calcite and 2% of quartz,
muscovite, sericite and chlorite. During the tests the strain components were recorded with the aid
of triple strain rosettes positioned at various strategic points of the specimens and using these
values the respective stresses were obtained according to the generalized Hooke’s law.
Comparing the results of the three approaches it was concluded that both the theoretically and
numerically predicted stresses (which are found in excellent mutual agreement) are close enough
to the ones of the experimental study. Some discrepancies detected are attributed to the slight non-
linearity which characterizes Dionysos marble.

References
1. Muskhelishvili, N.I., Some Basic Problems of the Mathematical Theory of Elasticity,
Noordhoff, Groningen, The Netherlands, 1963.
2. Sarris, E., Agioutantis, Z., Kaklis, K. and Kourkoulis, S.K., In Proceedings of the 7th
International Workshop on Bifurcation, Instabilities and Degradation in Geomechanics,
edited by G.E. Exadaktylos and I. Vardoulakis, Chania, Crete, Greece, 2005, p.84.
 2T14. Computational fracture mechanics 415

SIMULATION OF THE MECHANICAL BEHAVIOUR OF THE LUMBAR

INTERVERTEBRAL DISC

M. Satraki, E. A. Magnissalis1, G. Ferentinos2 and S. K. Kourkoulis

Department of Mechanics, National Technical University of Athens
5, Heroes of Polytechnion Avenue, 157 73 Zografou Campus, Athens, Greece
1First Orthopaedic Department of the University of Athens, Greece
2Kostas Liontos and associates-channel partner of Ansys-Greece

stakkour@central.ntua.gr

The aim of the present paper was the simulation of the mechanical behaviour of the human lumbar
intervertebral disc under various types of static and time varying loads. The study is carried out
with the aid of the finite element method. The model is built using the commercially distributed
finite element program ANSYS 9.0, a powerful software package with great capabilities for
simulating nonlinear materials and performing various types of analysis.
Concerning the geometry and the mechanical properties of the disc, numerical values from the
literature [1] were adopted after suitable elaboration [2,3]. The geometry and the structure of the
disc according to the present simulation are shown in Fig.1. The endplates were assumed to behave
as linear elastic bodies, the annulus was modeled as laminated composite material consisting of
successive transversely isotropic layers while the nucleus was considered as viscoelastic material
described by the curve-fitting model.

FIGURE 1. The discretization of the regions of the disc: (a) The cartilaginous vertebral endplates,
(b) The annulus fibrosus and (c) The nucleus pulposus.

All degrees of freedom were constrained at the lowest surface of the disc (the outer side of the
lower endplate) whereas all the nodes of the uppermost surface were enslaved with respect to its
central node (where the loading was applied) for the translational degrees of freedom. The final
model consisted of 154386 elements, 172662 nodes and 697641 degrees of freedom.
The model was subjected to static axial compression and bending as well as to a time varying
loading program in an effort to simulate the daily activities of the human disc. Characteristic
results are shown in Fig.2, where the deformed shape of the disc is shown in case of bending
(Fig.2a) together with the variation of the stress components along a characteristic line (Fig.2b).
416 M. Satraki et al.

FIGURE 2. The deformed shape of the disc under bending (a) and the distribution of the stress
components along the central horizontal line at the mid-height(b).

The time-dependent loading programme started with a linearly increasing compressive force
for the first hour up to a maximum level of 850 N. The load was kept constant for 16 hours, it was
removed gradually (linearly) during the 17th hour and the model was kept load-free for 7 hours.
The results of the analysis are shown in Fig.3 together with the respective ones given by Natarajan
[4]. The qualitative agreement is good, even for the present relatively simplified model.
Preliminary results of sophisticated models, which are under development, support the present
analysis.

FIGURE 3. Results of the experimental and numerical study by Natarajan et al. [4] (a) and the
respective predictions of the present model for the vertical displacement vs. time (b).

References
1. Chen C.-S., Cheng C.-K., Liu C.-L., Lo W.-H., Medical Engineering and Physics, vol 23,
483-491, 2001.
2. Lekhnitskii S. G., Theory of Elasticity of an Anisotropic Elastic Body, Holden-Day Inc, San
Francisco, 1963.
3. Whitney J. M., Riley M. B., “Elastic Properties of Fiber Reinforced Composite Materials,
AIAA Journal, vol 4, No 9, 1966.
4. Natarajan R.N., Williams J.R., Anderson G.J., Computers and structures, vol 81, 835-842,
2003.
 2T14. Computational fracture mechanics 417

THE PULL-OUT STRENGTH OF TRANSPEDICULAR SCREWS IN

POSTERIOR SPINAL FUSION

P. Chazistergos, G. Ferentinos2, E. A. Magnissalis1 and S. K. Kourkoulis

Department of Mathematics, National Technical University of Athens
5, Heroes of Polytechnion Avenue, 157 73 Zografou Campus, Athens, Greece
1First Orthopaedic Department of university of Athens-Greece.
2Kostas Liantos and associates-channel partner of Ansys-Greece.

stakkour@central.ntua.gr

Finite element methods are widely used for the analysis of the failure mechanisms of spinal
fixation systems and the stress distribution patterns within vertebral bodies as an important part in
clinical evaluation of spinal injuries. For the purpose of fixation and stabilization of the spine using
pedicle screws the pull-out strength of the screw is one of the most important factors to be
considered. In this study the influence of various factors, related to the contact interface of the
bone-screw system, and the geometry of the screw, on the pull-out strength of the fixation screws
are explored. Towards this direction a finite element model (FEM) of the human lumbar vertebral
bone and transpedicular fixation screw was designed simulating the main characteristics of
commercially available fixation pedicle screws. The material of the screw was assumed to be a Ti
alloy, which was modeled as linear elastic material with modulus of elasticity E=117 GPa and
Poison’s ratio Q =0.3. The respective values for the cancelous bone were E=100 MPa, Q =0.2
while for the cancelous shell were E=780 MPa, Q =0.3 [1-3]. Both the bone and the shell were
assumed isotropic and linearly elastic. The analysis was performed using the commercially
available software ANSYS8 and the finite model is shown in Fig1.

Fig. 1: Detailed view of the FEM indicating the screw (green) the cancelous bone (purple) and the
cortical shell (red).
Special attention was paid to the optimum simulation of the bone-screw interface, which
defines the load transfer mechanism between the bone and the screw [1]. This FE study included
two different scenarios, the first one for the bone-screw interface of the first postoperative weeks,
where screw and bone are in simple contact (contact model) and the second one for the fully
bonded bone-screw interface, which simulates the long term conditions (bonded model).
418 P. Chazistergos et al.

Fig.2: The equivalent stress into the bone plotted along a line that passes from every thread edge
for the two cases.
During the first part of the study both the contact and the bonded model were tested under
constant displacements of 0.005mm at the pull-out direction and the stress distribution patterns on
the screw and bone were studied. For the displacement applied of the stresses developed into the
bone were mainly below the yield stress, on the contrary, for displacement 0.006mm the yield
stress was exceeded.
The analysis revealed the different stress distribution mechanisms of the contact and the
bonded model. At the bonded model the stress into the bone was distributed along the threads in
such a way that no of strong stress concentration appeared. On the contrary for the contact model
areas of strong stress concentration into the bone were developed at the edges of the threads. For
both models the results of the FE analysis indicated that major role in the process of load bearing is
played by the cortical shell. The maximum stresses developed into the vertebral bone are on the
interface between the cortical shell and the screw. As one moves deeper the stresses tend to be
uniformly distributed in the areas between the threads of the screw. Such an observation leads to
the conclusion that the fracture of the bone will first appear on the cortex-screw interface and the
deeper surfaces will follow. The pull-out strength required to produce the same displacement to the
screws of the two models is of course significantly higher for the bonded model.
The second part of the analysis included the parametric study of the dependence of the pull-out
force on the geometric characteristics of the screw. The parameters studied were the depth and the
angle of the thread. The analysis revealed increase of the pull-out force with the depth and the
inclination angle of the threads. The pattern of the stress distribution was considered also,
revealing that, higher thread angles result to a more uniform distribution of the stresses relaxing the
strain field developed in the bone.

References:
1. Chen S.-I,. Lin R-M and Chang C-H., Medical engineering & physics, vol (25) 275-282.
2003.
2. Zhang Q.H, Tan S.H.,. Chou S.M, Journal of biomechanics vol (37),479-485 2004.
3. Michael A. K. Liebschner,Phd,D.L. kopperdahl,Phd,W.S. Rosenberg,MD,T M. Keaveny,
Finite Element Modeling of the Human Thoracolumbar Spine Spine vol(28),number 6,pp
559-565. 2003.
 2T14. Computational fracture mechanics 419

MECHANICAL BEHAVIOR SIMULATION OF HIP PROSTHESES STRESS

DISTRIBUTIONS ANALYSIS

M. Kadi, R. Boulahia, K. Azouaoui, N. Ouali, A. Ahmed-benyahia and T. Boukharouba

Laboratoire de Mécanique Avancée (LMA), Faculté Génie Mécanique & Génie des Procédés
USTHB, Bab-Ezzouar BP. 32, El-Alia, 16111 Alger (Algeria)
atboukha@yahoo.fr

Numerical simulation by finite elements, of in-vitro mechanical behavior of human femur and hip
total prosthesis is carried out. Three models are considered: right human femur, femur + cement +
prosthesis (pre-clinical model) and prosthesis + cement (laboratory model). The obtained results
show that in presence of an implant, the orthopedic cement thickness and the distally from femoral
head of the constrained stem in an embedding medium, notably modify the stress fields and the
extent of the compressed zones and those in tension in cement.
In this work we are interested to the mechanical behavior of a femoral implant. To describe this
type of behavior, we choose a simulation based on finite element method. The analysis of the in-
vitro behavior of this type of implant requires:
• determination of the stress distributions generated along the femoral stem, in the cement
and along the femur,
• localization of the hot-point and evaluation of the tensile stress amplitude on the femoral
stem; stress responsible of the failure and even sometimes of the rupture,
• dentification of the compressed zones and those in tension in the orthopedic cement.
The results obtained enabled us to analyze the stress distributions by comparing the
localization of the maximum tensile stress designated hereafter by hot-point and the stress
amplitudes for the various analyzed cases. Figure (1) gives the stress distributions along the femur
(models: a and b) and along the femoral stem (model: b). We observe that the hot-point
localization moves from 37mm (Figure 1.b1) to 30mm (Figure 1.b2) along the (z) axis. Therefore,
we note a change of the magnitude and distribution of the stresses according to the two models.
This can be explained by the fact that for right human femur the load is applied and directly
transferred from the head towards the remainder of the femur. On the other hand, in presence of an
artificial implant, the load transfer towards the femur is effectuated by deformation of the implant
stem, which is in direct contact with the femur; it is the case of an implant cemented in the femur.

Figure 1: Von Mises equivalent stress(b1): right human femur and (b2): femur + cement + implant
420 M. Kadi et al.

From the stress distributions relating to models (b and c), we noticed that laboratory model
(model c) does not reflect reality at different levels (Figure 2):
- the stress peaks present very high values,
- the stress value decrease rapidly from the constrained section located at 80mm from the
loading point, until reaching zero at mid-height of constrained stem. On the other hand, in the case
of the pre-clinical model (model b) the stress decreases gradually from a distance equivalent to that
of constrain (model c) to change sign from 150mm of load application point (model b),
- the laboratory model (model c) over-estimates the real loading.

Figure 2: Variation of tensile stress along the femoral stem for 5 cement thickness (model (c):
cement + implant)

We can conclude that the presence of a standard cemented hip total prosthesis (model (b):
femur + cement + implant) influences notably the distribution of the stresses and the localization of
the hot-point especially. The latter moves towards the outside of at least 7mm accompanied by a
sensible growth of maximum stress value. This is mainly due to the mode of load transfer.
Moreover, the distribution of the stresses in the case of a right human femur (model a) is much
more uniform than that generated in the femoral stem, case of the model (b): femur + cement +
implant. The latter increases rapidly to reach the peak (maximum value).
We note also that the orthopedic cement thickness used to constrain the stem in an embedding
medium influences:
• less the stress distribution and the amplitude of these stresses, generated in the femoral
stem,
• the extent of the compressed zones and those in tension at the cross-sections of orthopedic
cement.
We can conclude that ISO7206 standard concerns a very large range of hip total prostheses
(various forms, various cross sections, various sizes, etc.), which explains the differences in the
found results for the models (b and c) and this, although the results given by model (c) which over-
estimates the real loading. This requires to develop new laboratory models which takes into
account the different aspects of pre-clinical loading.
 2T14. Computational fracture mechanics 421

DBEM ANALYSIS OF AXISYMMETRIC CRACK GROWTH IN A PISTON

CROWN

T. Lucht
Department of Mechanical Engineering, Solid Mechanics
Technical University of Denmark
2800 Lyngby, Denmark
trl@mek.dtu.dk

An engineering problem is considered in which an axisymmetric crack initiates from the inside of
a piston crown due to both thermal and mechanical load from the combustion as shown in Fig.1.
To study the crack propagation the boundary element method is well suited because discretization
only occurs at the boundary.
Portela et al. [1] developed an analysis for mixed mode crack growth in two dimensions using
a single region boundary element method. This method also known as the Dual Boundary Element
Method (DBEM) avoids the singularity in the final system of equations by using displacement
equations on one surface and traction equations on the other. Two dimensional problems, which
include thermal load besides the mechanical load, have been analyzed by Prasad et al. [2]. The
DBEM uses displacement and temperature equations on one crack surface and traction and flux
equations on the other crack surface. The thermal load is treated as a body force term which
normally requires domain integrals. By a Galerkin technique the domain integrals are transformed
to boundary integrals, so that all the advantages of the DBEM remain.
The stress intensity factors (SIFs) for [1], [2] are computed by the use of the J-integral.
Another approach for the evaluation of SIFs for thermoelastic problems by BEM is presented by
Mukhopadhyay et al. [3], who use a formulation based on the modified crack closure integral
(MCCI).
Lacerda L.A. and Wrobel L.C [4] present a DBEM for axisymmetric crack analysis. The SIFs
are evaluated by employing the MCCI and very accurate results were obtained, as compared to
numerical and analytical results. The method for predicting the direction of the crack growth is
adopted from [1], as direction of the crack growth is assumed to be perpendicular to the maximum
principal stress.

FIGURE 1. Piston crown loaded with the maximum combustion pressure and steady state
temperature field
422 T. Lucht

References
1. Portela A., Aliabadi M.H. and Rooke D.P., International Journal for Numerical Methods in
Engineering, vol. 33, 1269 -1287, 1992
2. Prasad N.N.V., Aliabadi M.H. and Rooke D.P., International Journal of Fracture, vol. 66,
255-272, 1994
3. Mukhopadhyay N.K., Maiti S.K., Kakodkar A., Nuclear Engineering and Design, vol. 187,
277–290, 1999
4. De Lacerda L.A. and Wrobel L.C., International Journal of Fracture, vol. 113, 267-284, 2002
 2T14. Computational fracture mechanics 423

RESIDUAL SHEAR STRESSES AND KII COMPUTATION

W. Cheng and I. Finnie

BEAR, Inc. 2216 5th Street Berkeley, CA 74710-2217, USA
Department of Mechanical Engineering
University of California, Berkeley, CA 94720 USA

When calculating stress intensity factors due to residual stresses, the stresses must satisfy
equilibrium conditions and boundary conditions. In this case Legendre polynomials of orders 2 and
higher are often used for residual normal stresses. For mode II loading it is desirable to have a
similar polynomial series for residual shear stresses. Consider a plate of thickness t subjected to
shear stress Wxy along the x-direction with origin located at the centerline of the plate and distance x
normalized by t/2. Since the shear stress must vanish at free surface, x = r1, Wxy(x) may be
expressed as

IJ xy (x) (1  x) (1  x) J(x)
(1)
where J(x) is a function to be determined, and it must satisfy the force equilibrium
1 1
³ 1
IJ xy (x) dx ³1
(1  x) (1  x) J(x) dx 0
(2)
Clearly, J(x) can be constructed by a set of orthogonal functions with a weight function equal
to (1-x)(1+x). In fact, J(x) belongs to the family of Jacobi polynomials, which also include the
Legendre polynomials,

(  1) n wn
J n (x) n
2 n!
(1  x 2 )  1
wx
>
(1  x 2 )1  n @
(3)

In practice, the nth order polynomial Jn is more conveniently computed by using the recurrence
relation

J n (2  n)n (2n  1)(n  1) x J n  1  n(n  1) J n  2 with J0 1, J 1 (x) 2x

(4)
It is seen that for J0 = 1, the shear stress corresponds to a parabolic distribution. For Ji with i >
0, the resultant force over the thickness is always zero. It is noticed that Ji consist of a complete set
of polynomials. Thus, an arbitrary shear stress may be expressed as
n
IJ xy (x) (1  x)(1  x) ¦ ȕ i J i (x)
i 0 (5)
where Ei are amplitude coefficients. From eqn (5) a residual shear stress can be constructed when
the first term (i = 0) is omitted.
Next, a method based on FEM for KI [1] is extended to compute KII. The displacement, u(a,S),
due to a crack of size a in the x-direction at a location S is obtained by introducing a virtual force Q
at point S in the direction of u(a,S). Following the approach for KI, we find
424 W. Cheng and I. Finnie

q
ª w u(a, S) º ª w u (a, S) º
K II (a) « wa »/« »
¬ ¼ ¬ wa ¼ (6)

in which uq(a,S) is the displacement in x-direction produced by the virtual force Q. The
formulation of eqn (6) not only simplifies the computation of stress intensity factor but also
eliminates the need of a crack element near the crack tip. To validate the solution given by eqn (6),
KII is obtained for a uniform shear stress and a parabolic shear stress acting on the faces of an edge-
cracked plate using 108 elements along the crack plane. Very good agreement is found between the
computed results and those given in Refs. [2] and [3]. Using the same element mesh, KII for shear
stresses i = 1, 2 and 3 given in eqn (5) are obtained and plotted in Fig. 1.

FIGURE 1. Computed KII for shear stresses.

References
1. Cheng, W. and Finnie, I., Computation of Stress Intensity Factors for a 2-D body from
Displacements at an Arbitrary Location, Int. J. of Fracture, vol. 81, 259-267, 1996.
2. Tada, H., Paris, P. and Irwin, G., The Stress Analysis of Cracks Handbook. Del Research
Corp., Hellertown, PA, 1972.
3. M. Ichikawa and T. Takamatsu, Fracture toughness test for the thin plate under mode II
loading. Trans. Japan Soc. Mech. Engrs, vol. 51, 1115-1121, 1985.
 2T15. Experimental fracture mechanics 425

QUANTITATIVE INTERPRETATION OF CRACK TIP STRAIN FIELD

MEASUREMENTS

A. M. Korsunsky
Department of Engineering Science
University of Oxford, Parks Road, Oxford OX1 3PJ, United Kingdom
alexander.korsunsky@eng.ox.ac.uk

Recent years have witnessed greatly improved availability of experimental measurements of two-
and three-dimensional strain and stress distributions, notably through the development of
diffraction techniques utilising penetrating radiation, such as high energy X-ray or neutrons
(Korsunsky and James, [1]). These methods now provide unprecedented level of detail that brings
with it challenges for interpretation. Of particular interest is the question whether conventional
fracture mechanics parameters, such as the stress intensity factor, can be extracted from such
measurements with confidence. The present paper describes a rational interpretation procedure that
can be applied to a set of measurements in order to answer the question posed above.

FIGURE 1. (a) Crack opening strain measured from the Al matrix 311 peak in an Al-SiC cracked
bar. (b) LEFM K-field predicition, K=28 MPam.

The experimental configuration used to collect strain maps from engineering samples is
described, with particular attention being paid to the efficiency of data collection in the energy-
dispersive mode, and the evaluation of macroscopic engineering strain values from the pointwise
data. Using these techniques, the characteristic strain variation in the vicinity of the crack front was
studied in a nominally single-phase aluminium alloy and in a composite reinforced with SiC
particles.
The transition was studied between the surface and the bulk of a specimen containing a crack,
demonstrating the capability of the technique to extract plane strain information often required for
LEFM analysis.
The experimental map of matrix strain in a composite was matched to the LEFM prediction to
determine the value of the stress intensity factor. The methodology used is related to the variational
eigenstrain procedure introduced in the context of residual elastic strain interpretation (Korsunsky
et al. [2]). A variational problem was formulated about minimising the sum-of-squares mismatch
between the prediction and the measurements, and determining the unknown parameter (SIF) from
the resulting linear system of equations. The implications and applications of the newly proposed
technique are discussed.
426 A. M. Korsunsky

References
1. Korsunsky, A.M. and James K.E., In Proceedings of the Third International Conference on
Experimental Mechanics, edited by C. Quan, F.S. Chau, A. Asundi, B.S. Wong, C.T. Lim,
SPIE Publications, Singapore, 2005, 487-493.
2. Korsunsky, A.M., Regino G. and Nowell D., In Proceedings of the Sixth European
Conference on Residual Stresses, edited by A.M. Dias, J. Pina, A.C. Batista, E. Diogo,
TransTech Publications, Coimbra, 2002, 329-334.
 2T15. Experimental fracture mechanics 427

MIXED MODE (I+II) STRESS INTENSITY FACTOR MEASUREMENT USING

IMAGE CORRELATION

A. Shterenlikht, P. López-Crespo1, P. J. Withers, J. R. Yates1 and E. A. Patterson2

Materials Science Centre, Manchester University, Grosvenor Street, Manchester, M1 7HS, UK
1Mechanical Engineering Department, Sheffield University,

Mappin Street, Sheffield, S1 3JD, UK

2Mechanical Engineering, Michigan State University, 2555 Engineering Building,

East Lansing, MI 48824-1226, USA

a.shterenlikht@manchester.ac.uk

Image correlation is becoming a popular full field optical experimental technique in wide range of
applications because it is cheap, fast and easy. It performs equally well at the micro and the macro
scales and requires minimal surface preparation. In this work the 2D image correlation technique
was used to measure the in-plane displacement fields near the crack tip. Commercial image
correlation software DaVis 6.0 by LaVision was employed.
This work is a logical continuation of Shterenlikht et al. [1] where the stress intensity factors
were calculated from the strain fields, or more precisely, from the first strain invariant. However,
the strains calculated using numerical differentiation of the noisy experimental displacement data
proved to be of such poor quality, that reliable stress intensity values were not achievable.
Accordingly, in this work the stress intensity factors are calculated directly from the
experimentally obtained displacement fields without further differentiation.
Muskhelishvili’s [2] approach was used to represent the displacements and the stress intensity
factors as complex Fourier series of the two analytical functions. The expression for the stress
intensity factors was taken from Sih et al. [3]. This leads to a system of equations for the unknown
Fourier coefficients. If the number of terms in the series expansion for the analytical functions is
smaller than the number of experimental points, then the system is over-determined. If, in addition,
the crack tip location is added as another two unknowns, then the system becomes non-linear and
can be solved using a genetic algorithm.
This method was applied to a centre cracked sample proposed by Otsuka et al. [4]. This sample
can be loaded under 7 combinations of mixed mode (I+II) from pure mode I to pure mode II. Fig. 7
shows the sample loaded at 45o to the crack plane. Several samples were machined from 7010 Al
alloy with the gauge section thickness varied from 2 to 4 mm and the crack length varied from 20
to 40 mm. The samples were subjected to several combinations of mixed mode (I+II) loading. A
combination of the CCD resolution and the image correlation parameters resulted in the
displacement resolution of 3.7 Pm/pixel.
Results to date indicate that the solution is very sensitive to the crack tip location. If the crack
tip is is assumed to be where it appears in the initial image of the specimen surface, then the least
squares solution produces the stress intensity factors within 50% of the nominal applied values. If,
however, the best fit crack tip position is sought as a part of the solution, the accuracy of the
calculated stress intensity factors can be increased significantly.
The best results are obtained when the number of the data points is 5-10 times greater than the
number of terms in the Fourier series.
428 A. Shterenlikht et al.

At present the data points for the analysis are chosen manually. The efforts are made to
automate this process as this would greatly speed up the analysis and make it possible to move to
the real time crack monitoring.

FIGURE 1. Experimental setup showing: the specimen clamped by two pairs of loading plates
pulled through a pair of loading holes such that the notch is at 45o with the loading axis; the testing
machine; and the optical system consisting of a ring light, lenses and a CCD camera.

References
1. Shterenlikht, A., Días Garrido, F.A., López-Crespo, P., Withers, P.J. and Patterson, E.A.,
Applied Mechanics and Materials, vol. 1-2, 107-112, 2004.
2. Muskhelishvili, N.I., Some Basic Problems of the Mathematical Theory of Elasticity, 4 edn,
Noordhoff International, Leyden, The Netherlands, 1977.
3. Sih, G.C., Paris, P.C. and Erdogan, F., Journal of Applied Mechanics, vol. 29, 306-312, 1962.
4. Otsuka, A., Tohgo, K. and Matsuyama, H., Engineering Fracture Mechanics, vol. 28, 721-
732, 1987.
 2T15. Experimental fracture mechanics 429

FRACTURE OF TURBINE BLADES UNDER SELF-EXCITING MODES

C.A. Sciammarella, C. Casavola, L. Lamberti and C. Pappalettere

Politecnico di Bari, Dipartimento di Ingegneria Meccanica e Gestionale
Viale Japigia 182 – Bari, ITALY
csciamma@poliba.it; casavola@poliba.it; lamberti@poliba.it; carpa@poliba.it

Failure of blades of a turbo-compressor occurred during operation. Three different typical failure
geometries were observed repeatedly. One mode was clearly identifiable as the first bending mode.
The other two (Figs. 1-2) could not be directly identified.
Experimental mechanics was utilized to identify the frequencies causing the blade fracture.
Stroboscopic holographic interferometry analysis of the blades was performed. A quantitative
evaluation of the recorded vibration amplitudes was made. An initial image of the blade at rest
was recorded. The blade was excited and the image was subtracted continuously from the initial
image and the result of the subtraction was observed on a monitor. The illumination was
synchronized with the vibrations and four images taken with phase changes of 0o, 90o , 180o, 270o,
were recorded. The recorded images were processed with the Holo Moiré Strain AnalyzerTM. The
strains were computed and transformed to the local coordinates of the curved blade surface. Using
Hooke’s law, stresses were determined. The stress trajectories of the blade surface were utilized to
obtain the stress-trajectories (isostatics). The stress trajectories were found to match with the initial
crack surface trajectory. With this process it was possible to identify the excitations that caused the
blade fracture.
This analysis reveals that the initial crack trajectories follow the isostatics thus failing in a
similar mode to the maximum-tensile-strain law. This law states that when the larger principal
strain reaches a critical value, then cracks perpendicular to the direction of the principal strain will
appear. This experimental observation has very important implications from the point of view of
multi-axial fatigue behavior. One possible way to understand this behavior is to consider the
damage theory of fracture. During the alternating loading caused by the self-induced vibrations the
material experiences an accumulative damage patterned by the prevailing stress field. As the
cycles accumulate the material initial homogeneity and isotropy cease to exist and weaker regions
where defects accumulate are created thus providing a natural path to the crack propagation at a
distance long enough before the dynamics of the problem changes due to the crack propagation.
The authors believe that any attempt to explain this behavior by arguments based on
continuum mechanics of homogeneous and isotropic bodies will not be able to provide an
explanation for the documented behavior of the blade. Therefore, this paper presents an analysis of
failure mechanisms justifying the experimental evidence. The observed failure appears to be a
form of the principal normal stress fracture criterion. In the present case, it provides not the actual
applied stress causing failure but gives the geometry of the minimum resistance of the crack path.
Hence, an argument based on damage accumulation must be introduced to explain the observed
behavior.
430 C.A. Sciammarella et al.

FIGURE 1. Picture of the broken blade in a compound mode. a) Picture of the blade; b) Isostatics
showing the actual crack trajectory as measured from broken blade.

FIGURE 2. Picture of the broken blade in a compound mode. a) Picture of the blade; b) Isostatics
showing the actual crack trajectory as measured from broken blade.
 2T15. Experimental fracture mechanics 431

PREDICTING CRACK ARREST BEHAVIOUR OF STRUCTURAL STEELS

USING NEW PROCEDURES

C. Gallo, J. A. Alvarez, F. Gutierrez-Solana and J. A. Polanco

Department of Materials Science and Engineering. University of Cantabria
E. T. S. de Ingenieros de Caminos, C. y P. Av/ Los Castros s/n, 39005, Santander, Spain
galloc@unican.es, alvareja@unican.es

The crack arrest philosophy, based on the concept that a brittle crack can be arrested when it
emerges from a critical region, is especially useful for such situations where additional safety
against unforeseen circumstances is needed. This criterion is included in the R6 procedure [1] as a
complementary method against fracture toughness criterion for the assessment of the structural
integrity of components containing defects.
It is well known that structurally representative crack arrest tests, such as double tension test
(DTT) or compact crack arrest (CCA) test, incorporate important limitations that motivates the use
of alternative methods [2]. Different types of equations and estimations from small – scale tests
(Pellini, Charpy, etc.) are continuously used in order to predict the main properties of crack arrest,
which are the crack arrest temperature (CAT) and the crack arrest fracture toughness (KIa).
However, not always these results have the required precision or the needed conservative
estimation [3].
Various steel grades covering a wide range of strength, toughness levels and industrial
applications [4], have been used to investigate alternative methods for obtaining better results and
predictions of crack arrest properties.
Two different methods have been developed. On the one hand, an intermediate scale crack
arrest (ISCA) test has been performed as an additional testing method based upon the existing
double tension test [5]. The proposed method provides an alternative approach avoiding the
difficult testing procedure and associated high cost.
On the other hand, a modified instrumented Charpy test has been developed to obtain a better
reproduction of the propagation process using a specimen that presents a modification in the area
of initiation by means of a brittle weld metal and electro-discharged machined notch instead of the
conventional one, thus minimising the energy required for crack initiation.
CAT and KIa results obtained from ISCA, DTT and CCA tests are in agreement, as can be seen
in Fig. 1 that represents KIa results obtained as a function of working temperatures for AH32
structural steel [5]. In addition, crack arrest predictions based on Pellini or conventional Charpy
tests, sometimes non-conservative, are improved when results from modified Charpy tests are
considered, showing closer values to the direct experimental results and maintaining always
conservative characteristics, as can be also seen in Fig. 1.
432 C. Gallo et al.

FIGURE 1. KIa results and predictions for AH32 structural steel.

References
1. “Assessment of the Integrity of Structures Containing Defects”. R6, Revision 4, British
Energy. 2001.
2. Wiesner C. S. and Hayes B., “A Review of Crack Arrest Tests, Models and Applications”,
Crack Arrest Concepts for Failure Prevention and Life Extension. Abington publishing. 1996.
3. Gallo C., Álvarez J. A., Gutiérrez-Solana F. and Polanco J. A., “Predicting Crack Arrest
Behaviour of Structural Steels Using Small-Scale Material Characterisation Tests”, From
Charpy to Present Impact Testing, ESIS publication 30, Elsevier, pp. 271-278. 2002.
4. “An Energy Balance Approach For Crack Arrest”. ECSC Sponsored Research Project. Final
Report. Contract No. 7210/PR/182. 2003.
5. Slater S., Gallo C. Álvarez J. A. and Gutiérrez-Solana F., “Development of a New
Intermediate Scale Crack Arrest (ISCA) Test”. Engineering Fracture Mechanics, to be
published. 2005.
 2T15. Experimental fracture mechanics 433

MECHANICAL PROPERTIES OF LARGE PLASTIC-MOLD STEEL BLOOMS.

M. Chiarbonello, D. Firrao, R. Gerosa1, A. Ghidini2, M.G. Ienco3, P. Matteis, G. Mortarino,

A.Parodi3, M.R. Pinasco3, B. Rivolta1, G. Scavino, G. Silva1, E. Stagno3 and G. Ubertalli
Dip. SMIC, Politecnico di Torino, Corso duca degli Abruzzi 24, 10129, Torino, Italy
1Dip. Meccanica, Politecnico di Milano
2Lucchini Sidermeccanica S.p.A.
3Dip. DCCI, Università di Genova

Tel. +390115644663, Fax. +390115644699

donato.firrao@polito.it.

Molds for plastic automotive components such as bumpers and dashboards are usually machined
from large pre-hardened steel blocks. Due to the large size, the blooms undergo a slack quench, so
that mixed microstructures occur throughout, both after quench and after the tempering stages.
Mechanical properties that are both not homogeneous in the section and everywhere lower
(particularly in fracture toughness) than those of correctly quenched and tempered alloy steel
specimens are obtained. Successive machining to form molds may be so deep that any of the
microstructure occurring at different positions in the original bloom can be found at the mold face,
where notch effects are commonly present. Welding, for local shape alterations, may yield further
defects.
The examined and most commonly used steel grade is 1.2738 (or 40CrMnNiMo8-6-4, ISO
4957 standard), a heat-treatable, 0.4% C, high-hardenability, low-alloy steel.
Previous studies have assessed the deleterious influence of the mixed microstructures due to a
slack quench upon the toughness of quenched and tempered low alloy steels (Sachs, Sangdahl, and
Brown [Sachs, G., Sangdahl, G.S., Brown, W.F., Iron Age, Nov.23, p 59 and Nov. 30, p 76,
1950.], Zhang and Knott [Zhang, X.Z., Knott, J.F., Acta Mater., Vol. 47, No. 12, p 3483, 1999.]);
nevertheless, whereas the fracture toughness properties of other tool steels have been known since
many years (Baus et. al. [Baus, A., Charbonnier, J.C., Lieurade, H.P., Marandet, B., Roesch, L.,
Sanz, G., Rev. Met, 72, p 891, 1975.], Mosca, Partendo, and Zocchi [Mosca, E., Partendo, R.,
Zocchi, R., Met. It., 67, 1975, p 562.], Okorafor [Okorafor, O.E., Mat. Sci. Tecn., 1987, vol. 3, p
118.]), the same property has been analyzed in the present mold steel only recently by Firrao et. al.
[Firrao, D., Matteis, P., Scavino, G., Ubertalli, G., Ienco, M.G., Parodi, A., Pinasco, M.R., Stagno,
E., Gerosa, R., Rivolta, B., Silva G., Ghidini A., In 2nd Int. Conf. Heat treatment and surface
engineering in automotive applications, Riva del Garda (Italy), 2005, paper n. 64, to be
published.], who considered the microstructure as obtained in large blooms, and by Firrao, Matteis
and Vassallo [D. Firrao, P.Matteis, M. Vassallo, In Proc. 11th Int. Conf. on Fracture, Turin, Italy,
2005, paper n. 5590.], who presented some preliminary results upon re-heat treated specimens.
In the present work, fracture toughness and resilience tests have been performed upon 1.2738
steel specimens that were cut from different positions inside a commercial bloom and individually
re-heat-treated. The specimens were re-austenitized, gas-quenched at a cooling rate sufficient to
achieve a full quench and tempered in two stages at the usual temperatures. Results have been
compared with a previously reported [Firrao, D., Matteis, P., Scavino, G., Ubertalli, G., Ienco,
M.G., Parodi, A., Pinasco, M.R., Stagno, E., Gerosa, R., Rivolta, B., Silva G., Ghidini A., In 2nd
Int. Conf. Heat treatment and surface engineering in automotive applications, Riva del Garda
(Italy), 2005, paper n. 64, to be published.] pointwise survey of the as-received microstructures and
mechanical properties of the same bloom and of another one (both blooms were actually used to
434 M. Chiarbonello et al.

machine bumper molds); the results of such a survey have been completed in the present work by
further Charpy KV and tensile measures performed upon the as-received material.
The re-heat treated specimens, despite achieving a higher strength than specimens cut from as-
received blocks, resulted in a significantly higher fracture toughness, albeit with a large dispersion,
and in a proportionally improved, albeit not yet satisfying, RT impact energy.
The relationship between the mechanical properties, the morphology of the fracture surfaces
and the microstructure is discussed, particularly by comparing the as-received and the re-heat-
treated specimens.
The following technological conclusions are proposed:
• the commonly employed production cycle, by performing the heat treatment before the
gross machining and upon blooms whose section exceeds the steel hardenability, does not
yield the steel’s maximum capabilities in terms of toughness and (to a minor extent)
strength;
• the fracture toughness of the large blooms should be considered a critical characteristic and
fracture mechanics verifications, already usual in other fields of industry, should dutifully
be applied to the molds' design;
• alternative classes of steels, whose properties may be less affected by the molds’
dimensions, should be considered for future research and application.

References
1. Sachs, G., Sangdahl, G.S., Brown, W.F., Iron Age, Nov.23, p 59 and Nov. 30, p 76, 1950.
2. Zhang, X.Z., Knott, J.F., Acta Mater., Vol. 47, No. 12, p 3483, 1999.
3. Baus, A., Charbonnier, J.C., Lieurade, H.P., Marandet, B., Roesch, L., Sanz, G., Rev. Met,
72, p 891, 1975.
4. Mosca, E., Partendo, R., Zocchi, R., Met. It., 67, 1975, p 562.
5. Okorafor, O.E., Mat. Sci. Tecn., 1987, vol. 3, p 118.
6. Firrao, D., Matteis, P., Scavino, G., Ubertalli, G., Ienco, M.G., Parodi, A., Pinasco, M.R.,
Stagno, E., Gerosa, R., Rivolta, B., Silva G., Ghidini A., In 2nd Int. Conf. Heat treatment and
surface engineering in automotive applications, Riva del Garda (Italy), 2005, paper n. 64, to
be published.
7. D. Firrao, P.Matteis, M. Vassallo, In Proc. 11th Int. Conf. on Fracture, Turin, Italy, 2005,
paper n. 5590.
 2T15. Experimental fracture mechanics 435

NON-LINEAR PHOTOELASTIC METHOD FOR STUDY FRACTURE

PROBLEMS

G. Albaut
Novosibirsk State University of Architecture and Civil Engineering (Sibstrin)
113 Leningradskaya st., t. Novosibirsk, 630008, Russia
rustam@alemar.nsk.su

In this report the problems of fracture mechanics with geometrical and physical nonlinearity were
investigated by non-linear photoelastic methods [1, 2, 3]. The strain changed from -50% till
+250% of relative lengthening. Changes in geometry and in thickness of the specimens were taken
into consideration. Non-compressible optical-sensitive polyurethane rubber was used for model
producing.

Main directions of studies

1 Developing of some theoretical aspects of non-linear photoelastic methods.
2 Analysis of rubber models with crack or side notches under tension.
3 The investigation of plastic problems in fracture mechanics by photoelastic coating
method.
4 Determination of stresses in plate steel necking under tension by different means.

• Nonlinear photoelasticity method has been developed to study large elastic and plastic
strains changing from -50% till +300% of relative lengthening. Optical-mechanics
dependences of the nonlinear photoelasticity were obtained and methods of stress-strain
division were elaborated.
• The investigation of rubber plates and strips with cracks and cuts was executed. Under
loading the cracks transformed into ellipse or circle. The stress fields and their
concentration coefficients near cracks were obtained.
• Progressive method of photoelastic coating was developed for the study of problems of
technological plasticity under strains till 100% of relative lengthening and more.
• Some problems of plastic strains were solved for steel and aluminum specimens with
catches under large strains. These were the problems of determination of plastic strain
kinetics, stress fields near sharp notch tip and other.
• Important plastic problem of fracture was solved experimental. This was stress-strain
distribution in localized necking in thin tensile bar. For example, in Figure 1 and 2, the
pictures of interference fringe patterns in coating and epures of principal stresses 1 and 2 in
cross section of necking are given. They were obtained different means.
It is noted necessary that non-linear photoelastic method is a basic experimental method for
creation of non-linear fracture mechanics.
436 G. Albaut

FIGURE 1. The fringe pictures in the samples with the different kinds of necking.

FIGURE 2 Epures ı1 and ı2 in cross section of necking obtained by the five different
methods.

Reference
1. M. Akhmezyanov, G. Albaut, Internatinal Journal of Fracture. vol. 128 (1), 223-231, 2004.
2. Albaut G.N., In Proceedings of 21st Symposium on experimental mechanics of solids.
Jachranca, Poland, 2004, 23-128.
3. Albaut G.N., Nonlinear photoelasticity in application to problems of fracture mechanics,
Novosibirsk: NGASU, Russia, 2002 (in Russian).
 2T15. Experimental fracture mechanics 437

FATIGUE CRACK LENGTH MEASUREMENT METHOD WITH AN ION

SPUTTERED FILM

Gang Deng, Koutarou Nasu1, Tilahun Daniel Redda1 and Tsutomu Nakanishi
Faculty of Engineering, University of Miyazaki, Miyazaki, 889-2192, JAPAN
1Graduated School, University of Miyazaki, Miyazaki, 889-2192, JAPAN

t0d114u@cc.miyazaki-u.ac.jp

The purpose of this research is to show a simple and high precision method to measure the length
of a crack as well as a micro crack using an extremely thin ion sputtered film. An ion sputtered grid
film was proposed to measure the crack length for a bending test specimen. Based on the
comparisons between the measurement results and that by an optical microscope, the ion sputtered
gird film has a very high measurement precision. This method is considered convenient and
practicable for the crack length measurement of insulating materials such as ceramics. Meanwhile,
an one-piece ion sputtered film was used on a metal bending specimen, the possibility to apply the
ion sputtered film in the crack length measurement of a metal machine element was confirmed by
the change of the electric resistance of the film with crack growth.
Crack length measurement method with a thin grid film
One of the simple and practical methods measuring a fatigue crack length is the crack gauge
similar to a strain gauge. The electric resistance of the crack gauge will change when the crack
grows and snaps the grids of the crack gauge, The crack length can be known by counting the
number of the snapped grids. However, the precision of the crack gauge is not so satisfied because
of the existence of a plastic film between the grids and crack surface and the thickness of the grid.
The snapping position is always behind the crack tip, the measurement error is about 0.1-0.2mm.
The effective method to increase the precision of the crack gauge method is removing the plastic
film and reducing the thickness of the film and grids. An ion sputtered metal film, the thickness of
which is only several tens of nanometer, was formed directly on an acryl surface. And the film was
scribed as a grid pattern. Since the length l of a grid is exceedingly larger than its width w, the
electric resistance R of a cracked grid can be expressed as following,

­ cons tan t (a z w)
R ®
¯ f (a w) (1)
where a is the length of crack in the grid. So, it can be understood that the resistance of a cracked
grid changes only at the instant when the crack snaps off the grid.
Crack length measurement results
A very simple system was used to measure the resistance of the ion sputtered grid film in this
research. Figure 1 shows the output voltage corresponding to the resistance of the film during a
three-point bend fatigue experiment, in which the snapping instant of grids are identified very
clearly. The precision was checked by an optical microscope. Table 1 gives crack length
comparisons measured by the film and a optical microscope. From table 1 it can be concluded that
the ion sputtered grid film has a very high measurement precision.
438 G. Deng et al.

FIGURE 1. Voltage of a grid pattern film

FIGURE 2. Electric resistance of an one-piece ion sputtered film

For the applications of the ion sputtered film on a metal surface, an insulating film between the ion
sputtered film and metal surface is necessary. And making the grids without damaging the
insulating film is so difficult that an one-piece (no grids) ion sputtered film is desirable. An
investigation for the possibility to use an ion sputtered film on a metal surface was performed on a
steel three-point bend specimen. Silicon was used as an insulating material. The silicon film was
formed by vacuum evaporation, the thickness of which was only one micrometer. The ion
sputtered film was made on the silicon film. Figure 2 shows the change of the resistance of the one-
piece ion sputtered film during crack propagation in a fatigue test. Although the relationship
between the resistance of the film and crack length is waiting for clarified and the researches are
planed further into the techniques to make insulating film as well as the details of the measurement
method, the possibility to use the ion sputtered film on a metal surface is confirmed.
 2T15. Experimental fracture mechanics 439

INDIVIDUAL FRACTURE EVENTS IN CELLULAR FOODS

Hannemieke Luyten, Eva M. Castro-Prada, Eefjan Timmerman, Wim Lichtendonk

and Ton van Vliet
WCFS
c/o A&F, PO Box 17, 6700 AA Wageningen, the Netherlands
Hannemieke.Luyten@wur.nl

Crispy foods like biscuits and breads crusts are cellular solids with relatively large pores of a wide
size distribution. Important for the crispy character of such products are both the fracture
behaviour and the acoustic emission (Luyten et al. [1]). With aging under deteriorating conditions,
these products become less crispy. This is often ascribed to an increase in water activity of the
crusts resulting in a change in the mobility of the different molecules, and thus a change in the
solid material properties. However, it is difficult to relate directly the change in molecular
properties to changes in crispy behaviour. One of the reasons for this is the product morphology,
dry crispy foods are irregular built cellular solids. An example is given in Fig. 1.

FIGURE 1. Macro-photography picture of the cellular structure of a Cracotte. Picture by H.

Tromp, WCFS.

We will present a method to measure the fracture properties and accompanying acoustic
emission of the individual fracture events in a crispy cellular solid food. This method allows
measuring simultaneously mechanical and acoustical behaviour on deformation at a sample speed
of 65 kHz (Luyten et al. 2004 [2]). The number of fracture events was found to be related to the
morphology of the crispy food, or more precisely to the number of beams and struts in the cellular
solid that fractures. This method enables us to study the solid material behaviour of the crispy
foods independently from their morphology. Also it is now possible to relate the molecular
behaviour and changes therein to the mechanical behaviour of the whole crust and the perception
of crispness.
The method makes it also possible to study if the behaviour of crispy foods as perceived with
biting and eating is just the sum of a series of single fracture events, or if interference between
single fracture events is important for the fracture properties of the whole material. This applies
especially to the acoustic emission. Results for the effect of deformation speed and the morphology
of the cellular solid food will be presented and discussed. Not only the amount of events (an
example of results is given in Fig. 2) is important for the perception of crispness, but also the size
440 H. Luyten et al.

of the peaks, and possible effects of interference on the frequency of the emitted sound in the
audible region.

FIGURE 2. The effect of the cutting speed on the number of fracture events in rusk rolls. Results
are given for the sound emission (red), force (blue) and the number of cell walls per unit length
(green line) that could be distinguished in a picture like Fig. 1.

References
1. H. Luyten, H., Plijter, J.J. and van Vliet, T., J. Texture Studies, vol. 35, 445-492, 2004.
2. Luyten, H., Lichtendonk, W., Castro, E.M., Visser, J., and van Vliet, T., In Food Colloids
Interactions, Microstructure and Processing, edited by E. Dickinson, Royal Society of
Chemistry, Cambridge, 2004, 380-392.
 2T15. Experimental fracture mechanics 441

EXFOLIATION FRACTURE MODE IN HEAVILY DRAWN PEARLITIC STEELS

J. Toribio and F. J. Ayaso

Department of Materials Engineering, University of Salamanca
E.P.S., Campus Viriato, Avda. Requejo 33, 49022 Zamora, Spain
Tel: (34-980) 54 50 00, Fax: (34-980) 54 50 02
toribio@usal.es

Cold drawing is used to produce prestressing steel wires for prestressed concrete employed in civil
engineering construction. Such a manufacturing technique affects the steel microstructure [1-3],
thus leading to crack deflection and anisotropic fracture behavior in air atmosphere and aggressive
environments [4-6].
This report deals with the strength anisotropy of cold drawn steels and studies, by means of
computer-assisted image analysis techniques, a special mode of fracture (called exfoliation
fracture throughout this paper) in notched samples of heavily drawn steels supplied from
commercial stock. Samples from a real industrial process were used. The manufacturing chain was
stopped, and samples of five intermediate stages were extracted, apart from the original material or
base product (hot rolled bar: not cold drawn at all) and the final commercial product (prestressing
steel wire: heavily cold drawn). Fracture tests under tension loading were performed on
axisymmetric notched specimens with a circumferentially-shaped notch. Four notch geometries
(cf. Fig. 1) were used with each material, in order to achieve very different stress states in the
vicinity of the notch tip and thus very distinct constraint situations.

FIGURE 1. Notched specimens used in the experiments.

A significant result related to the notched samples of heavily drawn steels was the anisotropic
fracture behavior exhibiting a propagation step oriented quasi-parallel to the wire axis or cold
drawing direction (exfoliation fracture) whose aspect is given in Fig. 2 for specimen 5B. The
fractographic appearance of the step shown in Fig. 2 resembles cleavage-like fracture. However, it
is not conventional cleavage, but a sort of oriented and enlarged cleavage, its orientation being
parallel to the wire axis or cold drawing direction, and with river patterns which are detectable in
such a direction.
In the case of conventional cleavage taking place in pearlitic eutectoid steel, the cleavage facet
size is a strong function of the prior austenite grain size, although it is always somewhat less [7].
This size is the zone in which the adjoining pearlite colonies of the grain share a common
crystallographic orientation of ferrite. It represents the critical fracture unit and determines the
intrinsic toughness in an isotropic pearlitic material. For anisotropic materials such as the cold
drawn steels analyzed in this paper, there is a orientation of all microstructural units (and
particularly of the pearlitic colonies) in the direction of cold drawing and therefore ferrite lamellae
442 J. Toribio and F. J. Ayaso

change their orientation during cold drawing (as a part of a colony and a set of ferrite/cementite
plates which do orientate in the cold drawing direction). Then the new critical fracture unit in the
drawn material would be the pearlite colony more than the prior austenite grain, because different
pearlite colonies in the same grain follow distinct orientations paths along the manufacturing route.
Thus the slender pearlitic colonies become the new microstructural fracture units and determine
the size of the enlarged cleavage facets characteristic of the exfoliation fracture in notched samples
of heavily drawn steels.

FIGURE 2. Fractographic appearance in specimen 5B.

References
1. Langford, G., Metall. Trans., vol. 1, 465, 1970.
2. Embury, J.D. and Fisher, R.M., Acta. Metall., vol. 14, 147, 1966.
3. Toribio, J. and Ovejero, E., J. Mater. Engng. Perform., vol. 9, 272, 2000.
4. Toribio, J. and Lancha, A.M., J. Mater. Sci., vol. 31, 6015, 1996.
5. Cherry, B.W. and Price, S.M., Corros. Sci., vol. 20, 1163, 1980.
6. Sarafianos, N., J. Mater. Sci. Lett., vol. 8, 1486, 1989.
7. Park, Y.J. and Bernstein, I.M., Metall. Trans., vol. 10A, 1653, 1979.
 2T15. Experimental fracture mechanics 443

INVESTIGATION OF CRACK CLOSURE BY USING THERMOEASTIC

STRESS ANALYSIS

L. Marsavina, R. A. Tomlinson1, E. A. Patterson1 and J. R. Yates1

University POLITEHNICA Timisoara, Blvd. M. Viteazul, Nr. 1, Timisoara 300222, RO
1University of Sheffield, Department of Mechanical Engineering, Mappin Street, Sheffield S1

3JD, UK
msvina@linux1.mec.utt.ro

A comparison between the effective stress intensity factors obtained using the replicas method and
the experimental value of stress intensity factors determined by thermoelastic stress analysis is
presented.
The investigation was performed on mixed-mode propagated cracks in a biaxial fatigue testing
machine using cruciform specimens with 45q cracks. A successive load cycle was used to grow the
cracks in order to prevent crack branching during propagation. Thermoelastic data was recorded
for a crack half length of 9 mm with an applied mixed-mode ratio 'KI/'KII = 0.45, and the stress
intensity factor and closure ratio determined for a range of R ratios. The thermoelastic
measurements were performed at lower applied cyclic loads than those used to propagate the crack
in order to prevent crack growth and branching during the data acquisition. Surface replicas were
used to measure the opening displacements, but not the sliding displacements due to the very small
amount of sliding displacement produced by the predominantly mode I applied loads. Three R
ratios of R = 0, 0.2 and 0.5 were considered. The measured maximum and minimum opening
displacements from the replicas were found to be higher than the theoretical ones. This can be
explained by the plastic deformation left by the successive loading cycle used for crack
propagation.
The comparison between the theoretical, the experimental (obtained by thermoelasticity) and
the effective values (obtained from replicas method) of the stress intensity factor 'KI highlights
that the experimental results obtained by thermoelasticity represents the driving force for crack
propagation, as shown in Fig.1. However, the experimental values of the stress intensity factors
obtained by thermoelasticity are below the theoretical ones but above the effective values obtained
taking into account the closure using the replicas.

FIGURE 1. Theoretical, experimental and effective variation of the stress intensity factor 'KI
against R – ratio, for an applied 'KII/'KI=0.45
444 L. Marsavina et al.

References
1. Tomlinson R.A., Yates J.R. – An investigation of crack closure uing thermoelasticity, In
Proceedings of the 2000 SEM IX International Congress on Experimental Mechanics,
Orlando, USA, 2000.
2. Dulieu – Barton J.M., Fulton M.C., Stanley P. – The analysis of thermoelastic isopahic data
from crack tip stress fields, Fatigue Fract. Engng. Mater. Struct., 23, 301 – 313, 2000.
3. Tomlinson R.A., Marsavina L. - Thermoelastic investigation for fatgue life assessment,
Experimental Mechanics, vol. 44, 487 - 494, 2004.
4. Diaz F.A., Patterson E.A., Tomlinson R.A., Yates J.R. - Measuring stress intensity fctors
during fatigue crack growth using thermoelasticity, Fatigue Fract. Engng. Mater. Struct., 27,
571 – 583, 2004.
 2T15. Experimental fracture mechanics 445

FRACTURE TOUGHNESS INVESTIGATIONS OF SEVERE PLASTIC

DEFORMED TUNGSTEN ALLOYS

M. Faleschini, W. Knabl1 and R. Pippan

Erich Schmid Institute for Materials Science, Leoben, Austria
1Plansee AG., Reutte, Austria

mario.faleschini@unileoben.ac.at

Tungsten and several W-alloys are widely used in high temperature applications. They offer high
melting points, low vapour pressures, good thermal conductivity, high erosion resistance and good
thermoshock properties. These properties are necessary and essential in future applications like
fusion reactors. Like other bcc materials tungsten shows a DBTT (ductile-to-brittle transition
temperature) which is very dependent on its processing and impurities. This complicates the
machining (turning etc.) of complex parts. Thus a low brittleness is essential and desired. In
contrast to usual widely used deforming techniques (rolling, drawing etc.) Severe Plastic
Deformation (SPD) is not only capable of increasing the tensile strength and hardness of a
material, it also can increase its ductility.
HPT is a well-known method to produce fine-grained structures with different degrees of
deformation within a single specimen [1, 2]. This procedure was therefore used to deform three
different rolled tungsten alloys, pure tungsten (W), a lanthanum-oxide dispersion strengthened
tungsten alloy ('WL10', W 1La2O3) and a potassium doped tungsten alloy ('WVM', W 0.005K).
These deformed specimens were subjected to fracture toughness tests at room temperature. The
HPT specimens were also heat treated in a vacuum furnace to determine its influence on
microstructural properties.
The High Pressure Torsion straining is performed at 400°C with an inductive heating unit. The
pill-shaped specimen (6mm diameter) is inserted in between two hardmetal-anvils which are
pressed together. One of the anvils is rotated under a high hydrostatic pressure of several GPa. This
leads to a continuous straining of the specimen. The microstructure of the material is strongly
influenced by this process, resulting in a high dislocation density and grain sizes in the region of
200-400 nm. EBSD-scans reveal information of orientation relations of grains, proving that the
misorientation is increased by HPT.
The thermal stability of high pressure strained tungsten alloys is also investigated. This is done
by a heat treatment in a vacuum furnace (p~10-6 mbar) at 800, 1000 and 1200°C for one hour.
Pure tungsten is nearly fully recrystallized after one hour at 1200°C, also losing its increased
material properties after SPD. In contrast to that, the potassium doped WVM still has a grain size
of approximately 500nm and remarkable microhardness. This is due to the fact that the potassium
bubbles impede grain boundary movement at high temperatures [3] and thus increase thermal/
creep stability. The recrystallization behaviour of severe plastic deformed materials thus is another
important property which has been investigated, especially because the generated high angle grain
boundaries have high mobilities and low activation energies [4].
Fracture toughness tests have been performed at room temperature. Therefor small SENB-
specimens were prepared from HPT-deformed tungsten alloys. Fig.1 shows the measured values
vs. applied true strain.
446 M. Faleschini et al.

Fig. 1: Facture toughness vs. applied true strain for three different tungsten alloys.

References
1. R. Z. Valiev, R. K. Islamgaliev, I. V. Alexandrov, Progress in Mat. Science 45, 103-189, 2000
2. A. Vorhauer, R. Pippan, Scripta Materialia 51, 921-925, 2004
3. E. Pink, L. Bartha, The metallurgy of doped non-sag tungsten, Elsevier Appl. Sci. (1989)
4. F. J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phenomena,
Pergamon 2002.
 2T15. Experimental fracture mechanics 447

PHOTOELASTIC ANALYSIS OF MODE I STRESS INTENSITY FACTOR IN

BEAMS WITH ANGULAR NOTCHES

M. Tabanyukhova and V. Pangaev

Novosibirsk State University of Architecture and Civil Engineering (Sibstrin)
630008, h. 113, st. Leningradskaja, t. Novosibirsk, Russia
rustam@alemar.nsk.su

The values of Mode I of stress intensity factors (SIF) were analyzed in the case of normal crack
opening depending on angle parameter in the elements having angle notches. The data were
obtained both from numerical and polarization-optical experiments with the use of the simplified
procedure of the determination of Mode I SIF for angle notches considering the symmetry of
investigated models. Result compared with work [1].
1. Basic dependences
The generalized Mode I SIF KI in that case may be determined from a dependence obtained on
the basis of Williams’ asymptotic solution [2].

V x (0, y ) 2S
K I ,
y O1  1 (1)
where x(0,y) - stress on a symmetry axis, y - coordinate of a considered point, (1-1) - the degree of
a singularity. Procedure of its determination was explained in work [3].
Too Mode I SIF was determined by formular of classical fracture mechanics (constant value of
a singularity degree -0.5) as

KI V 2S r . (2)
2. Beams with the angle notch
The determination of the generalized Mode I SIF was executed under three-dot bending of the
beams loosened by an angle notch in the tension zone with considering of the modified singularity
degree (1-1) (1) and with halp of dependence for classical crack (2). Angle parameters varied from
00 up to 1500. The model geometry and the loading scheme were given in Figure 1. At first the
field of dimensionless stresses was found by the finite element method. Then two complete sets of
beam models were studied by polarization-optical method by step-by-step loading. The depth of
the notch angle in the first batch was made l=0,5b (as in case of numerical experiment), in the
second it was l=0,25b.
Photos of interference fringe patterns were obtained. The fragments from some of them were
shown in Figure 1. In result the stress values were obtained near notches’ tops by two
means - numeerical and experimental. Mode I SIF were calculated on the basis of these
stresses with halp of dependence (1) and (2).
It is possible to note that all values Mode I SIF practically coincided at the notch angles from
00 up to 600, but further they were different.
448 M. Tabanyukhova and V. Pangaev

FIGURE 1. Fragments of interference fringe pattens in beam under thee-got bending

Reference
1. Srinivasa Murthy N., Ragnavendra Rao P, Photoelastic parametric studies of mode I stress
intensity factors, Experimental Fracture Mechanics,Pergamon Press Ltd. V.22, 3, 527-532,
1985.
2. Williams M. L., Stress singularities resulting from various boundary conditions in angular
corners of plates in extension, J. Appl. Mech. V.74, 526-528, 1952.

3. Albaut G.N., Kurguzov V.D., Kurbanov . B., Tabanyukhova M.V., Kharinova N.V.,
J.Higher Educational establishments. Building, Novosibirsk, Russia, 9, 92-98, 2004.
 2T15. Experimental fracture mechanics 449

FULL-SCALE EXPERIMENTAL INVESTIGATIONS ON PRESSURE TUBES

RUPTURE OF RBMK

Natalya Yu. Medvedeva, Andrey V. Andreev, Sergey V. Timkin, Igor A. Peshkov, Vladislav N.
Zhilko, Dmitri Ye. Martsiniouk and Olga A. Poshtovaya
Federal State Unitary Enterprise “Electrogorsk Research and Engineering Center on Nuclear
Power Plants Safety”
Bezymyannaya St. 6, 142530, Moscow region, Electrogorsk, Russia
natalya_medvedeva@erec.ru, andreev@erec.ru, timkin@erec.ru

The investigation is concerned with the processes occurring in the RBMK reactor space in the
course of a fuel channel (FC) rupture. Experimental approach to evaluate the risk of multiple FCs
rupture as a result of escaping coolant effect is presented.
Now three incidents related to fuel channel rupture occurred at different NPP units. The most
severe scenario of this accident is a sequential breaking of several FCs (i.e. multiple pressure tube
rupture, or MPTR) as a chain reaction following a single channel rupture.
Since the problem is complex and interdisciplinary (mechanical, thermo-hydraulic, and fluid-
structure interaction processes), the analysis of MPTR requires performing of a series of
theoretical and experimental studies of separate physical processes running in the RBMK reactor.
The experimental rigs concerned the MPTR have been designed and constructed at Electrogorsk
Research & Engineering Center, Russia. Full-scale TKR test facility was designed and constructed
as a main part of the investigation project.
The TKR tested core – a module of the reactor stack (MRS), is a full-scale fragment of the
RBMK-100 reactor space from the support plate of upper metal structure to the safety plate of the
bottom metal structure with a graphite stack with FC tubes of 5u5 cells (25 graphite columns with
FC). The fragment is surrounded with a series of graphite columns with reflector cooling channels
(RCC). General view of the TKR test facility is represented in Fig. 1.
The TKR test facility is intended for the study of an emergency single fuel channel rupture and
its consequences, in particular:
- measurement of pressure and temperatures in the graphite stack;
- measurement of bending of the graphite columns;
- measurement of strains of the channel tubes;
- assessment of parameters of fracture of the graphite blocks and channel tube (see Fig. 2).
Since the TKR test facility contains only 45 of 2488 channels of the RBMK reactor,
experimental solution of the MPTR was based on experimental simulation of the worst case
scenario as the most dangerous accident progression from point of view of sequential FC rupture.
The wide range of experimental tests performed in the small-scale test facilities, theoretical
investigations, and the preparatory tests in the TKR test facility made it possible to define and
realize worst case scenario in the TKR test facility.
450 N. Y. Medvedeva et al.

FIGURE 1. Module of the reactor stack of the full-scale TKR test facility.

FIGURE 2. View of the fracture center after the full-scale test at the TKR test facility.
 2T15. Experimental fracture mechanics 451

STUDY OF FRACTURE MECHANISM OF COMPOSITE MATERIAL

BUILDINGS BY PHOTOELASTICITY AND PHOTOELASITC COATING
METHODS

Olga Ivanova, Galina Albaut, Valerij Mitasov, Vladimir Nikiforovskij

and Marina Tabanyukhova
Novosibirsk State University of Architecture and Civil Engineering (Sibstrin)
113 Leningradskaya st., t. Novosibirsk, 630008, Russia
rustam@alemar.nsk.su

Two types of problems have been solved in the present investigations. The influence of cracks on
the stress under loading in tensile zone of reinforced beam models has been investigated by
photoelasticity. The stress fields and their concentration coefficients have been obtained in this
case. Another of solved problems is a study of fracture mechanisms in composite material
elements by photoelastic coating method.
First problem
Beam with one or two naturally developing cracks were investigated to their destruction. Some
results of the problems solved were given in figure 1 here. The models of beams were made from
optical sensitive materials. This was an epoxy resin or plexiglass. The models were reinforced by
the wire with the reinforcement coefficient equal to 0.01. During the process of the model
manufacture thin notches were filled with epoxy glue. The beam models were loaded by pure bend
and the cracks in the places of a glued notch appeared.
In figure 1 the isochromatic fringe patterns were given. They represent the curves with
principal stress differences of equal level. Epures of the normal stress x, y and the shear stress xy
were obtained by means of numerical integrating of the equilibrium equation for the plane
problem. While analyzing these distributions one may note that the engineering hypotheses of
bending theory are not correct near the cracks.
Second problem
The fracture mechanism of elements from natural materials was studied with the help of
photoelastic coating method. Several results of the investigation were given in figure 2. Here
fracture schemes were calculated by the finite element method numerically. The first scheme was
obtained from the fracture by the shear stress and the second one was obtained by the compression
stresses. The calculation was checked by the photoelastic coating method experimentally. Several
photos were given in figure 2. The comparison of results showed that the fracture of the studied
element was executed by the shear stress but not the compression stress.
452 O. Ivanova et al.

FIGURE 1. Isochromatic fringe patterns and stress epures in reinforced beam models with one or
two cracks

FIGURE 2. Fracture schemes of concrete cube obtained by numerical method and isochromatic
fringe patterns in photoelastic coating
 2T15. Experimental fracture mechanics 453

FRACTURE ENERGY IN MODE I AND MODE II OF

TEXTILE REINFORCED WOOD

R. Putzger and P. Haller

Institute for Steel- and Wood Construction, Technische Universität Dresden
Eisenstuckstraße 33 NB, 01069 Dresden, Germany
Robert.Putzger@mailbox.tu-dresden.de

Textile reinforcement in timber constructions enables substantial improvement of load bearing

behaviour. Failures in wood due to insufficient strength perpendicular to grain caused by
anisotropy are reduced. Durability of especially unprotected outdoor applications is increased, too.
The laminate used for reinforcement consists of a textile which is embedded in a matrix and
connected to the wood by an interface according to figure 1. These constituent parts are exposed to
several physical and chemical factors. Only a good compound allow to efficiently use high
strengths of synthetical fibres or heavy textiles.
We aim at guaranteeing high bond strength for a long period of time. Two methods are used for
assessment of wood-textile compound, see figure 1 – right. First, peel force is determined
according to ASTM D 3167 and serves for comparison of the influence of the environmental
factors mainly among each other. Secondly, fracture mechanics deliver a characteristic bond value
independent of specimen geometry. The fracture energy Gf as a measure for the quality of bond
strength can be used in analysis and design later.
Two different specimens were tested in the laboratory by means of fracture mechanics. The
DCB-cleavage specimen is used in a tensile test and fails in mode I because of delamination.
Unlike this, the TENF-specimen is examined in a three point bending test and fails in mode II -
shear failure.
Results of all fracture tests published in this paper follow the recommendation of RILEM TC-
133. The tests confirmed stable crack growth also in glued joints for intermediate textile layers.

FIGURE 1. Factors of influence on bond strength and test methods used.

454 R. Putzger and P. Haller

TABLE 1. Experimental results of fracture tests (COV in %) and load vs. deflection curve.

The results of fracture mechanics tests are summarised in table 1. By means of textile
reinforcement it was possible to increase fracture energy by approximately 40 % in mode I and by
25 % in mode II respectively. An exception is the laminate of epoxy resin and aramid fabric. In this
case fracture energy in mode II was lower in comparison to specimens made of solid wood. Tests
series with polyester resin matrix resulted in higher fracture energy than the one where an epoxy
matrix was used. This general tendency was confirmed also by evaluation of peel tests, Haller [1].
An influence of textile with respect to textile weight or structure could not be observed.
Further on, table 1 contains mean values of ultimate load of each test series. The fracture
energies were based on a small average sample number with a partially high coefficient of
variation. This has to be considered in the interpretation of the results.
The fracture mechanics investigations have confirmed a stable crack growth as well for
intermediate textile layers. But for successful testing a suitable wood selection is of great
importance.
The tests provided a basis for using fracture energy Gf as a measure for the quality of
compound. Based on peel tests, the expectations of good bond strength of textile reinforced wood
were met and confirmed by high values of fracture energy. As result of strengthening the fracture
energy could significantly be improved compared as to solid wood.

References
1. Haller, P.; Putzger, R.; Curbach, M. (Editor). In: Proceedings of the 2nd Colloquium on
Textile Reinforced Structures (CTRS2), Technische Universität Dresden: Eigenverlag, 2003,
S. 247-258
2. Aicher, S.; Boström, L.; Gierl, M.; Kretschmann, D.; Valentin, G. In: RILEM TC 133 Report,
SP Swedish National Testing and Research Institute, Building Technology, SPReport 1997:
13. Boras, 1997
3. Aicher, S.; Gustafsson, P. J. (Editor); Haller, P.; Petersson, H. In: A report of RILEM TC-133.
Structural Mechanics, Lund University, Sweden, 2002
 2T15. Experimental fracture mechanics 455

MEASUREMENT BASED PERFORMANCE PREDICTION OF THE

EUROPABRUCKE AGAINST TRAFFIC LOADING

R. Veit and H. Wenzel

Vienna Consulting Engineers
Hadikgasse 60, A-1140 Vienna, Austria
veit@vce.at, wenzel@vce.at

Bridges are ageing and traffic is growing, which creates a demand for accurate fatigue life
assessment. The Europabrücke – a well known Austrian steel bridge near Innsbruck, opened in
1963 - is one of the main alpine north-south routes for urban and freight traffic. A long-term
preoccupation of VCE with BRIMOS£ (BRIdge MOnitoring System) on the Europabrücke (since
1997) and the assessed prevailing vibration intensities with regard to fatigue problems and possible
damage led to the installation of a permanent measuring system in 2003. Today’s monitoring
abilities enable us to measure performance precisely. High-precision sensor data of accelerations
and displacements in dependence of separately registered wind and temperature data and their
implementation into analytical calculation provide the possibility to realize lifetime considerations,
which are of eminent importance for bridge operators.
The superior goal, which is to be shown in this paper, is to determine the relation between the
randomly induced traffic loading (vehicles per day) and the fatigue-relevant, dynamic response of
the structure, exclusively caused by freight traffic. As life-time predictions in modern standards
depend on lots of assumptions, the emphasis is to replace those premises – referring to loading - by
measurements. In that context the present work is focused on three levels:
• Global behaviour in dependence of all relevant loading cases (Based on laser-displacement
measurement)
• Cross-sectional behaviour under special consideration of the cantilever regions (Based on
laser-calibrated acceleration measurement – Fig.1)
• Local systems analysing the interaction between tyres and the beam-slab connections
(Based on inductive displacement transducers)
In each of these levels of analysis the consumption of the structure’s overall-capacity per year
is to be determined. An indispensable requirement is to reduce the permanent monitoring system’s
data by Rainflow-Counting, describing the remaining fatigue-relevant recurring response-cycles in
different categories of intensity and occurrence. As the present lifetime calculations are performed
in terms of stresses by means of damage-accumulation, global and local Finite Element Analysis is
necessary for the transition of measurment data (Nominal & Structural Stresses). The detailed
knowledge about the progression of the prevailing traffic from the very beginning up to these days
and the implementation of published future trend studies can be used for an extrapolation of the
measured impact for the whole lifetime. As this research work tries to encourage in-situ
measurements instead of “design situations”, it is also aspired to analyse the consequence of
statistical scatter in each level of impact as well as on fatigue resistance.
456 R. Veit and H. Wenzel

FIGURE 1. Reproduced cantilever displacements (blue) vs. directly measured ones (red).

References
1. Veit R., Wenzel H., Fink J.: Measurement data based lifetime-estimation of the Europabrücke
due to traffic loading - a three level approach, In Proceedings of the 58th International
Conference of International Institute of Welding, Prague, July 2005
2. Haibach E.: Betriebsfestigkeit – Verfahren und Daten zur Bauteilberechnung, VDI-Verlag,
Düsseldorf, 2002.
3. Niemi E.: Structural Stress Approach to Fatigue Analysis of Welded Components -
Designer’s Guide, IIW doc. XIII-1819-00/XV1090-01
4. ESDEP - European Steel Design Education Program: WG12 Fatigue, Lecture Notes,
Katholieke Universiteit Leuven
5. Wenzel H., Pichler D.: Ambient Vibration Monitoring, J. Wiley and Sons Ltd, Chichester -
England, 2005, ISBN 0470024305
6. Spaethe G.: Die Sicherheit tragender Baukonstruktionen, 2nd Edition, Springer-Verlag, Wien,
New York, 1992
7. Bronstein I., Semendjajew K.A.: Taschenbuch der Mathematik, 5th Edition, Harri Deutsch-
Verlag, Frankfurt am Main, 2001
8. Dreßler K., Gründer B., Hack M., Köttgen V.B.: Extrapolation of Rainflow Matrices, SAE
Technical Paper No. 960569 (1996)
 2T15. Experimental fracture mechanics 457

THE EFFECT OF THE LABORATORY SPECIMEN ON FATIGUE CRACK

GROWTH RATE

S. C. Forth, W. M. Johnston and B. R. Seshadri

NASA Langley Research Center
Lockheed Martin Corporation, West Reid Street, MS 188E, Hampton, VA 23681 USA
National Institute of Aerospace
Scott.C.Forth@nasa.gov

One of the responses of a material to extreme forces, such as stress, temperature, etc., is to crack.
A crack appears when the material reaches a limit in its capability to absorb damage and fails.
Sometimes, a crack will grow under a periodically applied condition, such as cyclic loading, that
are well below the stresses required to fail the material, denoted fatigue crack growth. Over the
past thirty years, laboratory experiments have been devised to develop fatigue crack growth rate
data that is representative of the material response. The crack growth rate data generated in the
laboratory is then used to predict the safe operating envelope of a structure. The ability to
interrelate laboratory data and structural response is called similitude. In essence, a
nondimensional term, called the stress intensity factor, was developed that includes the applied
stresses, crack size and geometric configuration. The stress intensity factor is then directly related
to the rate at which cracks propagate in a material, resulting in the material property of fatigue
crack growth response.
Standardized specimen configurations have been developed for laboratory testing to generate
crack growth rate data that supports similitude of the stress intensity factor solution. Recent
research into the fatigue crack growth threshold has exposed some limitations in the standards [1,
2]. The typical approach to generate the threshold at a specific stress-ratio is accomplished by
reducing both the maximum and minimum applied load at a specific rate until the crack arrests.
Pippan, et al. [3], Smith and Piascik [4 ] and Forth, et al. [5] have all postulated that this method
may produce results that are not representative of the material behaviour. Furthermore, Liknes and
Stephens [6] and Garr and Hresko [7] have suggested that specimen configurations contained
within the standards can have an effect on threshold. The middle crack tension, compact tension
and eccentrically loaded edge-crack tension specimen configurations are widely used for
generating fatigue crack growth rate data [8]. In this proposed paper, the authors present: (1)
laboratory fatigue crack growth rate test data of a high strength steel alloy that shows specimen
configuration, width and thickness effects (Fig. 1); (2) three-dimensional finite element analyses of
each of the specimens identifying the differences on stress intensity solution; and (3) propose
modifications to the laboratory testing standards to support similitude of the stress intensity factor
solution.
458 S. C. Forth et al.

Figure 1. Crack growth rate versus stress intensity factor range data for different specimen
configurations.

References
1. Tabernig, B., Powell, P. and Pippan, R., Fatigue Crack Growth Thresholds, Endurance
Limits, and Design, ASTM STP 1372, ASTM, 96-108, 2000.
2. Newman, J.C., Jr., Fatigue Crack Growth Thresholds, Endurance Limits, and Design, ASTM
STP 1372, ASTM, 227-251, 2000.
3. Pippan, R., Stuwe, H.P. and Golos, K., International Journal of Fatigue, 16, 579-582, 1994.
4. Smith, S.W. and R.S. Piascik, Fatigue Crack Growth Thresholds, Endurance Limits, and
Design, ASTM STP 1372, ASTM, 109-122, 2000.
5. Forth, S.C., Newman, Jr., J.C. and Forman, R.G., International Journal of Fatigue, 25, 1, 9-15,
2003.
6. Liknes, H.O. and Stephens, R.R., Fatigue Crack Growth Thresholds, Endurance Limits, and
Design, ASTM STP 1372, ASTM, 175-191, 2000.
7. Garr, K.R. and Hresko, G.C., Fatigue Crack Growth Thresholds, Endurance Limits, and
Design, ASTM STP 1372, ASTM, 155-174, 2000.
8. Liaw, P.K., Peck, M.G. and Rudd, G.E., Engineering Fracture Mechanics, 43, 379-400, 1992.
 2T15. Experimental fracture mechanics 459

VALIDITY OF THE CAUSTICS METHOD FOR PLATES WITH CIRCULAR

HOLE

P. Tsirigas, G. Kontos1, D. N. Pazis, S. K. Kourkoulis and Z. Agioutantis1

Department of Mechanics, National Technical University of Athens
5, Heroes of Polytechnion Avenue, 157 73 Zografou Campus, Athens, Greece
1Department of Mineral Resources Engineering,

Technical University of Crete, 731 00, Hania, Crete

stakkour@central.ntua.gr, zach@mred.tuc.gr

The method of caustics introduced by Manogg [1] and developed further by Theocaris [2] is
nowadays considered as a very useful and flexible tool for the investigation of the stress
concentration and the stress intensity factors for the case of structures with geometrical
discontinuities. However, a number of questions regarding the range of validity of the method
remain unanswered, in spite of the intensive research carried out. The most crucial points are
related to the influence of the plastic zones developed in the immediate vicinity of the singular
points [3], to the triaxiality [4,5] and to the size of the discontinuities especially in the case of holes
[6].
In the present paper the range of validity of the method is studied for the case of a plate of
finite dimensions with a central circular hole. The study was carried out experimentally. In
addition the Finite Elements Method was employed for the determination of the stress and strain
distributions within the plate. The specimens used for the experimental part of the study were made
from PMMA. They were rectangular plates with dimensions Height x Width x Thickness equal to
300 x 120 x 2 mm3. A central hole was drilled adopting a novel procedure, in order to avoid the
development of residual stresses. The diameter of the hole varied between 2mm and 12mm. The
specimens were subjected to monotonic, quasi-static tensile load up to their final failure.
The typical optical arrangement described by Kalthoff [7] was adopted for the realization of
caustics. During the loading process the caustics developed from the light beams reflected at the
front and rear sides of the specimens as well as those developed from the light beams transmitted
through the specimens were photographed at predefined load steps. Then measuring the maximum
diameter Dmax of the caustics (using suitable software for the minimization of the experimental
errors) it was possible to determine the stress applied by the well-known formula:
4
1 1 § D m ax ·
p 2 ¨ ¸¸
12 O 3m H z 0 c r ,t ,f R d ¨© 2, 67 ¹ (1)
where p is the stress applied at infinity, is a constant equal to 1 for the front reflection and equal to
2 for the rear reflection and the transmission case, R is the diameter of the hole, d is the thickness
of the plate and cr,t,f denote the stress-optical constants for rear reflection, transmission and front
reflection, respectively. In the above equation m is the magnification factor of the optical
arrangement equal to (z0-zi)/zi, where zo is the distance between the specimen and the screen and zi
is the distance between the focal point of the convergent light beam and the specimen.
Using the above equation and the load recorded from the loading frame it was possible to
determine for each load step the ratio e ( V c a u s tic / V a c tu a l ) , i.e. the ratio of the stress
predicted by the caustics over the actually applied stress. The variation of e versus the load level is
460 P. Tsirigas et al.

plotted in Fig.1 for a characteristic test. It is clearly seen that the results obtained from the method
of caustics approximate the value of the actual load in a satisfactory manner for a well-defined
interval of the load range, marked with two red vertical lines. The agreement becomes poor outside
this interval and for loads approaching the fracture load the discrepancies increase dramatically.

FIGURE 1. The ratio of the theoretical over the experimental stress versus the applied load

Taking advantage of the experimentally determined yield stress of the material used and the
stress variation in the specimens as obtained from the numerical analysis it was concluded that the
upper limit is dictated by the development of the plastic zones around the hole. Concerning the
lower limit of the validity interval it was concluded that it is dictated by the size of the initial curve
of the caustic. Indeed for specific combinations of the diameter of the hole and the load level it is
possible that the initial curve of the caustic falls within the hole of the specimen rendering the
results obtained from the study of the respective caustics erroneous.

References
1. Manogg, P., Anwendung der Schattenoptik zur Untersuchungs des Zerreissvongangs von
Platten, Dissertation, Freiburg, Germany, 1964.
2. Theocaris, P.S., J. Appl. Mechanics, vol. 37, 409-415, 1970.
3. Konsta-Gdoutos, M. and Gdoutos, E.E., Engng. Fract. Mechanics, vol. 42, 251-263, 1992.
4. Meletis E.I., Huang W. and Gdoutos E.E., Engng. Fract. Mechanics, vol. 39, 875-885, 1991.
5. Rosakis, A.J. and Ravi-Chandar, K., Int. J. Solids Structures, vol. 22, 121-134, 1986.
6. Tsirigas, P., Stress concentration around a hole in a plate of finite dimensions: Numerical and
experimental study, Master Thesis, Nat. Techn. Univ. Athens, Greece, 2005 (In Greek).
7. Kalthoff, J.F., Handbook on Experimental Mechanics, edited by A.S. Kobayashi, VCH, 1993,
407-476.
 2T15. Experimental fracture mechanics 461

AN ENHANCED NORMALIZATION METHOD FOR DYNAMIC FRACTURE

TOUGHNESS TESTING

S. M. Graham and D. J. Stiles

Assistant Professor
United States Naval Academy, 590 Holloway Rd, Annapolis, MD 21402
Naval Surface Warfare Center, Carderock Division, 9500 MacArthur Blvd,
West Bethesda, MD 20817
smgraham@usna.edu, daniel.j.stiles@navy.mil

Fracture toughness testing of ductile materials can be difficult in situations where it is not possible
to measure crack extension during the test, such as under high rate loading or in aggressive
environments. In these situations, an alternative method of inferring crack extension must be used
to generate the tearing resistance curve, and thereby determine ductile crack initiation. ASTM test
method E1820-01 uses the Normalization method to generate the plasticity function for the
specimen, which can then be used to calculate crack extension. This method relies heavily on
accurate measurement of the load, displacement and crack length at the end of test. These
measurements are used to generate a point on the normalized load-displacement plot, known as the
“anchor point”. If a test ends with unstable crack extension, the anchor point cannot be determined
and the method cannot be applied. If there is a large amount of crack extension in a test, the
uncertainty in the derived plasticity function increases, which can lead to non-conservative J-R
curves. An alternative approach that can be used to reduce the dependency on the anchor point is
the Compliance Ratio (CR) method. This method uses a quasi-static test with unloading
compliance to generate a load-displacement record with no crack extension, or “key curve”. Once
this key curve is obtained, it can be scaled to account for the effect of loading rate. The specimen
compliance at any point in a dynamic test record is determined by comparing the key curve with
the actual load at the same displacement, and from this the crack extension can be determined.
There are problems with this approach when the quasi-static key curve does not match the dynamic
test record early in the test where blunting transitions to ductile tearing.
Fortunately, there is a way to overcome the problems associated with each of these methods
and draw from the advantages of each. The basic premise behind the proposed Enhanced
Normalization (EN) method is that the CR method key curve is reasonably accurate beyond the
“knee” of the curve where the specimen is undergoing stable ductile crack extension, which is
precisely where the Normalization method provides the least information about the plasticity
function. By using a quasi-static test to generate the plasticity function beyond initiation, and then
scaling it to account for rate effects, the region beyond the tangency point can be filled in and the
dependence on the anchor point can be removed.
The proposed Enhanced Normalization method is used to analyze a series of dynamic fracture
toughness tests of ASTM A336, Grade F22, forged steel. This test program included 3 quasi-static
tests and 6 dynamic tests, all conducted at -2 °C (28 °F). The dynamic tests were conducted in a
drop tower with an impact velocity of 3.5 m/s and impact energy of 2175 J. The measured
compliances from the quasi-static test of specimen S1 were used to calculate crack lengths and to
generate the normalized load-displacement curve. Next, the dynamic data from specimen D4 was
normalized using the initial crack length with estimated blunting. The ratio of the maximum loads
in the dynamic and quasi-static tests was used as an initial estimate of the scaling ratio for the
quasi-static plasticity curve. The actual scale value was varied to obtain a good visual match
between the dynamic and quasi-static curves in the vicinity of the “knee”. The two normalized
curves are plotted together in Figure 1. The anchor point is shown for comparison with the quasi-
462 S. M. Graham and D. J. Stiles

static plasticity curve, however, it is not used in the subsequent analysis. The point of divergence
between the two curves was estimated and used to establish the dividing line. All points on the
dynamic plasticity curve to the right of this line, and all quasi-static points to the left were
eliminated, thereby leaving the combined plasticity curve. This curve was then used to calculate
crack lengths for the dynamic test by determining the crack length that would bring a data point
onto an interpolated spline curve. The resulting J-R curve is shown in Figure 2. There is very little
scatter in crack extensions along the blunting line and the initiation of ductile tearing is clearly
visible. This J-R curve is a big improvement over what was obtained using the CR method, and is
more conservative than what was obtained with straight Normalization.

Figure 1. Dynamic and Quasi-static plasticity curves.

Figure 2. J-R curve obtained from Enhanced Normalization Method

 2T15. Experimental fracture mechanics 463

THE POTENTIAL DROP TECHNIQUE FOR MEASURING CRACK GROWTH

IN SHEAR

V. Spitas and P. Michelis

Laboratory of Applied Mechanics, Technical University of Crete
University Campus, 74100 Chania, GREECE
Institute of Mechanics of Materials and Geostructures S. A.
22 Askiton Str., 15236 Penteli, GREECE
immg@otenet.gr

The classical potential drop technique [1] is used to measure a growing crack in a conductive
material and has been a standard methodology for measuring “open” cracks, i.e. those propagating
in predominantly mode I (tension). In tension the electrical modeling is rather straightforward
since a known electrical current passes through the body of the tensile specimen and the developed
voltage due to the constantly changing resistance as the crack propagates is measured at two
electrodes fixed at both sides of the developing crack. In mode II (in-plane shear) however there is
little data available in the existing literature as the issue of generating a pure and uniform in-plane
shear stress field on a specimen (i.e. without the existence of bending or any other normal stresses)
is still open and under discussion.
A solution to this problem comes from a specially designed shear specimen which can develop
a uniform and pure in-plane shear stress field at its central region (Spitas et al. [2]) if loaded as
illustrated in Fig. 1. The geometry of the specimen is patented [3] and resembles the Iosipescu
specimen [4] but its main difference lies both in the shape of the grooves (U grooves instead of V
grooves) for minimizing stress concentration and in the way of exerting the loading onto the
specimen. For this latter reason a specially developed servo-hydraulic shear-testing machine has
been developed in the last 15 years through EU funded industrial research projects.

FIGURE 1. The shear specimen and its loading conditions.

The modified potential drop technique on the shear specimen uses the same concept of the
change of the electrical resistance with the advancement of the crack tip but differs from the
classical one since the electrical current density varies widely within the specimen therefore the
potential drop it is not only indicative of the length but of the direction of the crack as well. The
current passes through two thin electrodes welded on the surface of the specimen and the resulting
potential difference between two other similar thin electrodes is measured as the crack tip
advances. In this way the length of the crack can be estimated from the measured change in
electrical potential drop.
464 V. Spitas and P. Michelis

A rigorous F.E. Analysis using FRANC2D and MAXWELL for the mechanical and electrical
field respectively was performed and it illustrated that the stress intensity factor in Mode II (KII)
can be linked with the length of the crack and hence with the potential drop between the measuring
electrodes welded onto the specimen. From this analysis a relationship between the change in
relative electrical resistance and crack length can be established and therefore any set of
experimental results which include change in electrical resistance versus the number of fatigue
cycles can be translated into da / dN versus ' K II plots.
The modified potential drop method for the shear specimen was successfully applied on single
crystal nickel based superalloy CMSX4 tested in fatigue at 950qC. The accuracy of the numerical
predictions of the F.E. Analysis for the electrical field was experimentally verified both by
measuring artificial cuts of known length and by optical crack length measurements of the actual
cracks using a robotic travelling long-distance optical microscope. For small crack lengths the
numerical predictions coincided with the actual measurements within 2%, which corresponds to an
error of roughly 20m.

Reference
1. ASM Handbook Vol. 8, Mechanical Testing and Evaluation, ASM, Ohio, U.S.A., 2000.
2. Spitas, V., Besterci, M., Michelis, P., Spitas, C., Sulleiova, K. and Balokova, B., Powder
Metallurgy Progress, vol. 4 (4), 233-242, 2005.
3. Michelis, P., European Patent EP 0687899.
4. Iosipescu, N., J. of Materials, vol. 2 (3), 537-566, 1967.
 2T15. Experimental fracture mechanics 465

A MODIFIED DCB GEOMETRY FOR CTOA MEASUREMENT IN THIN SHEET

2024-T3 ALUMINIUM ALLOY

Y. H. Tai, S. H. Hashemi, R. Gay, I. C. Howard and J. R. Yates

The University of Sheffield, Department of Mechanical Engineering
Mappin Street, Sheffield, S1 3JD, UK
mep04yht@shef.ac.uk

Failure analysis carried out by various researchers has shown that fracture models based on the
Crack Tip Opening Angle (CTOA) criterion which have been calibrated based on data from large
C(T) and M(T) specimens can be transferred successfully to cracked aircraft fuselage structures
and other low constraint structures for residual strength predictions [1,2,3]. One major difficulty
with this method is the experimental measurement of CTOA [4] either in real structures or in a
laboratory scale test. This could be the limiting factor that prevents the more extensive use of this
promising fracture parameter. There are a number of ways to obtain CTOA data but the most
appropriate and reliable method has yet to be agreed.
The paper describes the use of a modified double cantilever beam specimen (Fig. 1) for the
direct measurement of the critical CTOA data for thin sheet 2024-T3 aluminium alloy commonly
used in aerospace applications. It highlights the features of ductile growth in low constraint
configurations and the experimental technique used to reproduce these features in a laboratory
scale experiment. This includes the specimen design geometry, loading configurations used,
specimen preparation, and CTOA measurement technique. The experimental technique used is a
development from the previous work of Shterenlikht et al [5].

FIGURE 1. Modified double cantilever beam specimen geometry.

Using the developed technique, CTOA resistance curves for test samples of 2.3mm ligament
thickness were generated for fractures with the crack propagating in the axial (TL) and in the
transverse (LT) direction of the rolled aluminium alloy sheet. The specimen geometry and loading
configuration allowed for extensive crack growth and managed to achieve a fairly straight crack
path throughout the test (Fig. 2) which is desirable. Results showed that the technique was capable
of producing large amounts of highly consistent CTOA data even from one single specimen. The
tests produced a steady state mean value of CTOA of 4.2° with a standard deviation of 0.5°.
This technique is promising as it provides a precise and easy experimental technique for direct
CTOA measurement using relatively small scale laboratory specimens. A comparison of the test
results from the current work with similar data from literature will be included in the paper.
466 Y. H. Tai et al.

FIGURE 2. Fractured specimen with scribed mesh.

References
1. Dawicke, D. S., and Newman Jr, J. C., Residual Strength Predictions for Multiple Site
Damage Cracking Using a Three Dimensional FEA and a CTOA Criterion, Fatigue and
Fracture Mechanics, Vol 29 ASTM STP 1332, 1998.
2. Hsu, C., Tom, J. J., and Anderson, B. L., Residual Strength Analysis Using CTOA Criteria for
Fuselage Structures Containing Multiple Site Damage, In Proceedings of ICF 16, Turin,
March 2005.
3. Newman Jr, J. C., James, M. A., and Zerbst, U., Engineering Fracture Mechanics, Vol 70,
371-385, 2003.
4. Schwalbe, K. H., Newman Jr, J. C., and Shannon Jr, Engineering Fracture Mechanics, Vol
72, 557-576, 2005.
5. Shterenlikht, A., Hashemi, S. H., Howard, I. C., Yates, J. R., and Andrews, R. M.,
Engineering Fracture Mechanics, Vol 71, 1997-2013, 2004.
 2T15. Experimental fracture mechanics 467

COULD COD SERVE AS FRACTURE CRITERION IN CASE OF MARBLE?

A. Marinelli, S. K. Kourkoulis1 and I. Vayas

Faculty of Civil Engineering, National Technical University of Athens
9, Heroes of Polytechnion Avenue, 157 80 Zografou Campus, Athens, Greece
amarin@central.ntua.gr
1Department of Mechanics, National Technical University of Athens

5, Heroes of Polytechnion Avenue, 157 73 Zografou Campus, Athens, Greece

stakkour@central.ntua.gr

An experimental study is presented here related to the behaviour of U-notched prismatic Dionysos
marble specimens subjected to either three- or four-point bending or to direct tension, in an effort
to investigate the variation of the Crack Opening Displacement (COD) versus the externally
applied load. The purpose of the study is the determination of an-easy-to-use and reliable tool that
could serve as fracture criterion in the hands of engineers working for the restoration projects of
various ancient monuments as well as in every day’s practical applications.

The specimens were either of the form of Single Edge Notched (SEN) prismatic beams of
rectangular cross section (bending tests) or Double Edge Notched Tensile (DENT) dog-bone
shaped plates (tension tests). Series of experiments were carried out for each loading type with
specimens of various sizes and various crack lengths in an effort to quantify the dependence of the
mechanical constants and COD on the size and the shape of the specimens. Such an investigation is
necessary since it is known that the results of laboratory tests with Dionysos marble (like most
geomaterials) are shadowed by the extremely pronounced size and shape effects, as it is seen in
Figure 1, where the tensile strength is plotted versus the diameter of the specimen [1].
The tests were carried out by controlling the (vertically) induced displacements. The strain
components at some strategic points were measured using a system of triple strain gauge rosettes.
The COD was measured as Crack Mouth Opening Displacement (CMOD) with the aid of suitable
clip gauges mounted by pairs of machined knife edges.
468 A. Marinelli et al.

FIGURE 2. CMOD vs. the applied load for characteristic tests with DENT (a) and 3PB specimens
(b).

Characteristic results concerning the variation of the CMOD versus the applied load are plotted
in Fig.2 for both types of tests. It is interesting to observe the non-linearity exhibited by the CMOD
especially in the case of DENT specimens at the notch from which the fracture started. Such
behaviour should be expected since the constitutive behaviour of Dionysos marble is relatively
non-linear [2] for loads approaching the fracture load.
The results of the experiments indicated that the values of the critical COD vary depending
strongly on the orientation of the specimens with respect to the bedding planes of the marble. It is
thus indicated that COD could be used as a fracture criterion only in the form of a surface in a three
dimensional space (, , COD) where and are the inclination angles of the crack and the axis of the
load with respect to the principal anisotropy directions of the material.

Acknowledgements
The present study is part of the scientific project “PROHITECH: Earthquake Protection of
Historical Buildings by Reversible Mixed Technologies”, Contract number: INCO-CT-2004-
509119 supported financially by EU. The support is gratefully acknowledged.

References
1. Vardoulakis, I. and Kourkoulis, S. K., Monuments under seismic action, Final Report EV5V-
CT93-0300, Nat. Techn. Univ. of Athens, 1997.
2. Kourkoulis, S.K., Exadaktylos, G.E., Vardoulakis, I. In Proc. 14th European Conference on
Fracture, Vol. II, pp. 243-250, Krakow, Poland, A. Neimitz, I.V. Rokach, D. Kocanda, K.
Golos (Eds.), EMAS Publishing, United Kingdom, (2002).
 2T16. Creep fracture 469

CREEP RUPTURE OF A LEAD-FREE Sn-Ag-Cu SOLDER

C.-K. Lin and D.-Y. Chu

Department of Mechanical Engineering, National Central University
Chung-Li 32054, TAIWAN
t330014@cc.ncu.edu.tw

Due to environmental and healthy concern, the allowable usage of lead will be decreased. The
search of new lead-free solders with equivalent mechanical properties and microstructural stability
to eutectic tin-lead (Sn-Pb) solder is an ongoing task. Solder joints in electronic products play not
only a role to interconnect the electronic components but also to ensure the structural reliability of
the electronic packages. As service temperatures for electronic assemblies are usually located in
the high homologous temperature ranges (> 0.5Tm) of solder alloys, creep damage may take place.
Therefore, a thorough understanding of the creep behavior of the newly developed lead-free
solders is of extreme importance to the designers of electronic assemblies.
Although there have been a few creep studies on Sn-Ag-Cu alloys in the literature [1-5],
studies related to temperature effects on the creep behavior and lifetime prediction methodology in
such lead-free solders were quite limited. Hence, the purpose of this study was to investigate and
quantify the creep properties of a promising lead-free Sn-3.5Ag-0.5Cu solder at various
environmental temperatures so as to provide important and helpful information for design of
reliable electronic packages. The creep tests were performed under constant load at room
temperature (RT), 60oC, and 90oC under a tensile stress range of V/E = 10-4 to 10-3. These testing
temperatures correspond to a range of about 0.6-0.74 Tm for the given lead-free solder.
Experimental results indicated that at a given temperature, the given Sn-3.5Ag-0.5Cu alloy
showed much better tensile and creep strength than the conventional Sn-37Pb solder due to a
dispersion-strengthening mechanism. The ultimate tensile strength (UTS) and creep resistance
were found to be decreased with increasing temperature for the given lead-free solder. The stress
exponents (n) of minimum strain rate (H min) were decreased from 9 at RT to 6 at 60oC and 90oC,
suggesting the controlling creep mechanism for the given conditions was a lattice-controlled
dislocation climb coupled with a dispersion-strengthening mechanism and the part of dispersion
strengthening was more effective at a lower temperature.
As Monkman-Grant expressions for each set of data at a given temperature are quite
comparable, all the data from various temperatures can be fitted very well by a single Monkman-
Grant relationship, as shown in Fig. 1. A model, using a term of applied stress normalized by
Young’s modulus, was proposed in this study to correlate the rupture times at various temperatures
and could explain the rupture time data reasonably well for the given lead-free solder, as shown in
Fig. 2. Apparently, Monkman-Grant relationship and the proposed model each individually
showed very good results in estimating the creep rupture time of the given ternary Sn-Ag-Cu
solder subjected to various combinations of stress and temperature by a unified, single expression.
470 C.-K. Lin and D.-Y. Chu

FIGURE 1. Monkman-Grant plot of all data for Sn-3.5Ag-0.5Cu alloy.

FIGURE 2. Normalized stress vs. rupture time for Sn-3.5Ag-0.5Cu alloy.

References
1. Plumbridge, W.J., Gagg, C.R. and Peters, S., J. Electron. Mater., vol. 30, 1178-1183, 2001.
2. Guo, F., Lucas, J.P. and Subramanian, K.N., J. Mater. Sci., vol. 12, 27-35, 2001.
3. Yu, J., Joo, D.K. and Shin, S.W., Acta Mater., vol. 50, 4315-4324, 2002.
4. Joo, D.K., Yu, J. and Shin, S.W., J. Electron. Mater., vol. 32, 541-547, 2003.
5. Guo, F., Choi, S., Subramanian, K.N., Bieler, T.R., Lucas, J.P., Achari, A. and Paruchuri, M.,
Mater. Sci. Eng. A, vol. 351, 190-199, 2003.
 2T16. Creep fracture 471

QUANTITATIVE EVALUATION OF ACCELERATION CREEP IN

MAGNESIUM-ALUMINUM ALLOYS AT 0.65TM

H. Sato
Department of Machines and System Engineering,
Faculty of Science and Technology, Hirosaki University,
Bunkyo-3, Hirosaki, Aomori, 036-8561 JAPAN
g4sato@mech.hirosaki-u.ac.jp
In creep failure, ternary stage is important portion of creep deformation. Although accelaretion
creep is important to understand whole creep behavior, creep characteristics are evaluated mainly
based on minimum or steady state creep rate and time to failure. In this report, creep
characteristics in accerelation creep of magnesium-aluminum alloys are quantatively evaluated.
It is well recognized that the steady state or minimum creep rate of solution-strengthened
alloys can be reasonably described by following Dorn-type equation (1) in power-law stress range.
n
Gb m §V · § Q ·
H A© N ¨ ¸ exp ¨  c ¸
kT ©G ¹ © RT ¹ (1)
Here, A’, G, b, k and R are the constants, and others show regular meaning. The values T, N and ?
are the absolute temperature, the solute concentration and the applied stress. The value m, n and
Qc characterize the creep behavior and are the concentration exponent, the stress exponent and the
apparent activation energy of creep, respectively. In magnesium-aluminum solid solutions, creep
characteristics depend on deformation conditions in complex manners. At the temperatures below
0.7Tm, the power-law creep characteristics are attributed to Alloy-type and Metal-type at lower
and higher stresses, respectively (1,2). In Mg-3mol%Al alloys, the change of creep characteristics
appears at 30MPa at 600K.
In compression creep of magnesium-aluminum binary alloys, simple relationship between
creep rate and creep strain appears in ternary creep regime(1). The relationship is similar to that of
Omega-method(3), but in magnesium-aluminum alloys, the primary stage is not igoreble. In this
report, stress dependence of acceralation creep characteristics of magnesium-aluminum solution
strengthened alloys are quantitatively evaluated based on similar relationship of the Omega
method.
In ternary stage of constant stress creep of the alloys, logarithmic of creep rates are
proportional to creep strain, and show relationship described by equation (2).

ln H k H  ln H 0 (2)

Here, k and H 0 are constants determined experimentally, which shows magnitude of acceleration
of strain rate in ternary stage and initial creep rate, respectivily. This equation is similar to that
proposed in the Omega method(3), but in magnesium-aluminum alloys, the relation is satisfiled in
ternary stage even though true stresses are kept constant, and primary creep is observed. Equation
(2) derives strain-time relationship as equation (3).

1 § t ·
H H init  ln ¨ 1  ¸¸
k ¨© tr ¹ (3)
472 H. Sato

Here, H init and t r are constants that shows magnitude of primary creep and time to infinite strain,
respectively. In equation (3), three parameters detarmine shape of creep curve under paticular
stress and temperature.
Figure 1 shows example of strain rate-strain relationship of magnesium-3mol% aluminum
solution strengthened alloys. The relations described by equation (1) are shown strain above 0.1 in
various stress levels. Figure 2 shows an expample of stress dependence of the k-value, one of three
paramters which determin creep curve, i.e. equation (3). The stress dependence of the k changes at
around 30MPa that correspounds to the change of creep characteristics which is detarmined by
minimum creep rates(1). These stress dependences are also observed in other parameters.
As conclusion, creep characteristics of magnesium-aluminum solid solution alloys in
compression are reasonably described Omega like concept. Stress dependence of creep parameters
changes at the stress that corresponds to the change of deformation mechanisms. Quantitative
evaluation of ternary acceleration creep suggests possibility of reasonable prediction of time to
failure. Change of microstructure during acceleration creep will also be presented.

Figure1.Example of strain rate-strain curve Figure2.Stress dependance of the k-value

References
1. Hiroyuki Sato, et al., In The Fourth Pacific Rim International Conference on Advanced
Materials and Processing (PRICM4), Edited by S. Hanada, et al., The Japan Institute of
Metals, 2001, 1155-1158.
2. H. Oikawa, In Hot deformation of Aluminum Alloys, edited By T.G. Langdon, et al., TMS,
Warrendale, (1991), 153
3. M. Prager, In Strength of Materials, edited by H. Oikawa, et al., The Japan Institute of Metals,
1994, 571-574
 2T16. Creep fracture 473

LONG-TERM CREEP RUPTURE PREDICTION IN UNIDIRECTIONAL

COMPOSITES

Jun Koyanagi, Fumio Ogawa1 and Hiroyuki Kawada

Department of Mechanical Engineering, Waseda University
3-4-1 Okubo, Shinjuku-ku, Tokyo, Japan
1Graduate of Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo, Japan

jun@aoni.waseda.jp

In many articles, analyses about creep behavior and creep rupture in unidirectional composites
have been done in the past two decades. McLean [1] formulates the creep behavior in
unidirectional composite with a simple constitutive equation in matrix. Curtin establish the GLS
theory that can determine a rupture strain in unidirectional composites. By combining these two
models, Du and McMeeking et al. [3,4] and Ohno and Miyake [5,6] predict the creep rupture in
unidirectional composites. However, several problems are pointed out: an interfacial debonding is
not considered, McLean model can not express the exact creep behavior. On the other hands, the
previous author’s works [7] and Okabe et al. [8] show that micro damages such as an interfacial
debonding causes the composite rupture strain to decrease. That is to say, the rupture strain in
unidirectional composite has time-dependency on an assumption that the interfacial debonding
propagates or the stress recovery length increases with time. However, the decrease of the rupture
strain has not been formulated on a consideration of history dependent accumulation of micro
damages. The fact results in that the predicted creep rupture time is overestimated. This is
because despite of there is an interaction between stress recovery length and fiber breaks
probability, the formulation has not been done.
In this study, the matrix constitutive equation is applied the compliance component as a
function of time, power low, and its deformation behavior can be expressed by a convolution
integral equation. Hence, to formulate the specimen strain as a function of time, Laplace
transformation is used on way to the equation derivation, and the Shapery approximation is applied
as the inverse Laplace transformation. (Eq. 1)

J t V
H t 
1  V f  J  t V f E f
(1)
Also, for the long-term creep, the time-dependnet rupture strain as a funciton of the average
stress recovery length is formulated in Eq. 2.
1

V0 § L0 ·m
H rup ¨ ¸¸
E f ¨©  m  1  Lr ¹ (2)
Fiber breaks probability that dominates rupture strain and that is dominated the stresss
recovery lengh is formulated on a consideration of that each stress recovery length depends on
when the fiber breaks as shown in Eq. 3.

2 § E fH · 1 t d q t 
q t  ¨ ¸ ³ L r  t  t cc  d t cc
L0 © V 0 ¹ q t  0 d t cc (3)
474 J. Koyanagi et al.

By solving this equation, the decrease of the rupture strain in unidirectional composite can be
formulated on a consideration of the history dependent micro damage growth for the long-term
creep rupture. That is to say, if the stress recovery length increase function regarding one fiber
break is obtained, the long-term time-dependednt rupture strain can be calculated.
By using glass fiber reinforced polymer unidirectional composite specimen, the creep rupture
tests were performed. The results are compared with the analytical prediction involving follows
assumptions.

L r t  N ˜ t  Lr  0 
(4)
Parameters used in this analysis are as follows: Ef=72000MPa, Vf=0.02, J0=0.000294 1/MPa,
D=0.27, Tc=2.0E+05 sec, Js=0.577, V0=2500 MPa, L0=25 mm, and Lr(0)=500 Pm. The rate of
stress recovery length increase, N, is assumed 5.0E-09, 5.0E-10 and 5.0E-11. The results of the
creep rupture prediction involving the above assumption is shown in Figure

Figure Prediction of creep rupture time and experimental results

References
1. McLean M. Creep deformation of metal-matrix composites. Compos Sci Technol 1985; 23:
37-52.
2. Curtin WA. Theory of mechanical properties of ceramic-matrix composites. J Am Ceram Soc
1991; 74: 2837-2845.
3. Du ZZ, McMeeking RM. Creep Models for Metal Matrix Composites with Long Brittle
Fibers. J Mech Phys Solids 1995; 43: 701—726.
4. Sofronis P, McMeeking RM. The effect of interface diffusion and slip on the creep resistance
of particulate composite materials. Mech Mater 1994; 18: 55-68.
5. Ohno N, et al. A model for shear stress relaxation around fiber break in unidirectional
composites and creep rupture analysis. J Soc Mater Sci Japan 1998; 47: 184-191.
6. Ohno N, Miyake T. Stress relaxation in broken fibers in unidirectional composites: modeling
and application to creep rupture analysis. I J Plasticity 1999; 15: 167-189.
7. Koyanagi J, et al. Prediction of Creep Rupture in Unidirectional Composite :Creep Rupture
Model with Interfacial Debonding and its Propagation. Adv Compos Mater 2004; Vol 13 No.
3-4: 199-213.
8. Okabe T, et al. A 3D shear-lag model considering micro-damage and statistical strength
prediction of unidirectional fiber-reinforced composites. Compos Sci Technol 2001; 61: 1773-
1787.
 2T16. Creep fracture 475

A COMPUTATIONAL MODEL FOR CARDBOARD CREEP FRACTURE

Julia Schonwalder, Gideon P. A. G. van Zijl1 and Jan G. Rots

Delft University of Technology, 1University of Stellenbosch
Faculty of Architecture, Berlageweg 1, 2628 CR Delft, The Netherlands
j.schoenwaelder@bk.tudelft.nl

Recently paper and cardboard is emerging as a construction material. This involves further
research into the mechanical response of cardboard and cardboard structures concerning structural
safety and long-term stability. The behaviour of paper and cardboard is highly time- and rate-
dependent. The time-dependent crack growth can cause delayed structural collapse. In a
computational model, which aims to predict structural response of a material, the time-dependent
crack growth can be considered by the incorporation of the time scale.
In this paper an interactive experimental and a computational research effort is reported,
through which the tensile creep fracture phenomenon is characterised and a computational
prediction tool developed. The focus is on tensile creep in machine direction of a solid board. In an
extensive test series the response of the board under sustained and strain-rate loading was
determined. A finite element model which was developed for creep fracture of masonry and
cementitious materials e.g. by De Borst [1], Van Zijl [2] and Lourenço et al. [3] was modified. The
model is based on linear visco-elasticity, combined with computational plasticity to capture the
limited tensile resistance.
With the modified parameters the tensile responses of cardboard at the same rates employed in
the experiments were obtained by FE analyses. The results are summarised in Table 1. It is seen
that the maximum tensile strength of the experiments and the modelling match very well.

TABLE 1. Experimental and numerical results of tensile peak strength of different constant strain
rate loading
Peak strength (N/mm2)
Rate (mm/min) Experimental (nr Numerical
tests, CoV)
0.001 16.0 (1, -) 16.5
0.5 25.0 (10, 10%) 24.0
5 26.0 (10, 10%) 26.0
50 28.2 (10, 10%) 28.0

Also the creep tests were modelled with the FE analyses, at the same load levels where failure
occurred in the experiments. Fig. 1 shows that the analysed time to failure coincides with the
experimental values fairly well.
476 J. Schonwalder et al.

FIGURE 1. Experimental and numerical time to failure under sustained load.

Even though cardboard is known as a non-linear viscoelastic material, see e.g. [4], the model
based on a linear viscoelastic assumption, showed evidence of predicting the time to failure under
creep loading considerably well. Also the creep model is confined to primary and secondary creep.
An integration of tertiary creep would further improve the results.

References
1. De Borst, R., Smeared cracking, plasticity, creep and thermal loading – a unified approach.
Computer Methods in Applied Mechanics and Engineering, vol. 62, 89-110, 1986.
2. Van Zijl, G.P.A.G., Computational Modelling of masonry creep and shrinkage, Dissertation,
Delft University of Technology, The Netherlands, 1999.
3. Lourenço, P.B., Rots, J.G. and Blaauwendraad, J., Continuum model for Masonry: Parameter
estimation and validation, ASCE Journal of Structural Engineering, vol. 124, 642-652, 1998.
4. Haslach, H.W. JR., The moisture and rate-dependent mechanical properties of paper: a
review, Mechanics of Time-Dependent Materials, vol. 4, 169-210, 2000.
 2T16. Creep fracture 477

CREEP FRACTURE OF BINARY AND TERNARY COMMERCIAL ALUMINUM

ALLOYS

K. Ishikawa
Department of Mechanical Engineering, Toyo University
2100 Kujirai, Kawagoe, Saitama, Japan 350-8585
ishikawa@toyonet.toyo.ac.jp

Binary and ternary commercial aluminum alloys were crept to the rupture under a condition of
constant applied stress. The applied stress was guaranteed within 0.5% of the expected value. The
commercial aluminum alloys have medium and high strength at the ambient temperatures. The
binary alloy was A5083, the main alloy element of which was ca. 5% magnesium. Two kinds of
ternary alloys were A6061 and A7075. The former contains 3% magnesium and 1% silicon. The
latter consists of 3% magnesium and 7% zinc. The test temperatures were kept in the region of the
solid solution of the alloys. The temperatures were guaranteed within 2K. The whole creep tests
were carried out until the rupture of the specimens at the constant applied stress and temperature.
The elongation was measured with the laser extensometer with the accuracy of 50 Pm. The creep
curves were converted into the true strain of the plastic elongation. The steady state creep was not
observed for the whole alloys. The minimum strain rate was described by the following equation
[1].

H min v V m
exp  Q min kT  (1)
Where m depends upon the test temperature, and Qmin depends upon the applied stress. The
experimental values of m were showed in Table 1.

TABLE 1 Stress exponent, m of minimum strain rate for the respective alloys

The creep life was also expressed by the following equation [2].

t life f V n
exp Q life kT  (2)
Where n depends upon the test temperature, and Qlife depends upon the applied stress. The
experimental values of n were showed in Table 2.
478 K. Ishikawa

TABLE 2 Stress exponent, n of creep life for the respective alloys

For A5083, Q min | Q life 210  0 . 13 V kJ mol ,

for A6061, Q min | Q life 352  1 . 08 V kJ mol 

and for A7075, Q min | Q life 439  3 . 52 V kJ mol  . V is the applied stress. The activation
energy for creep fracture was determined by the atomic interaction between the defects at high
temperatures. Furthermore, there was found the simple relationship between the minimum strain
rate and creep life given by the following equation [3].

H min u t life K
(2)
The constant, K does not depend on the test temperature and the applied stress. For A5083, K =
0.11, for A6061, K = 0.12 and for A7075, K = 0.1. The experimental results would suggest that the
creep behaviour and fracture was intensively associated with the interaction between the defects in
the alloys. The strength could be due to the difference in the interaction between the defects. The
diffusion process was dominant, since the final fracture pattern was dimple [4].

References
1. Skrzypek, J. and Ganczarski, A., Modeling of Materials Damage and Failure of Structures,
Springer, Berlin, Germany, 1999
2. Altenbach, H. and Skrzypek, J. J., Creep and Damage in Materials and Structures, Springer-
Verlag, Wien, Austria, 1999
3. Barsoum, M., Fundamentals of Ceramics, McGraw-Hill, New York, USA, 1997
4. 4. Weertman, J. and Weertman, J. R. Elementary Dislocation Theory, Macmillan, New York.
USA, 1964
 2T16. Creep fracture 479

ANALYSIS OF CREEP CRACK INITIATION AND GROWTH IN

LABORATORY SPECIMENS

K. Wasmer
Swiss Federal Laboratories for Materials Testing and Research (EMPA)
Feuerwerkerstrasse 39, 3602 Thun, Switzerland
kilian.wasmer@empa.ch

In the power generation, petro-chemical and nuclear industries, predicting the life of aging
equipment has become both a safety issue and an economic necessity. In order to perform such
assessments, a number of in-house procedures, as well as national standards (e.g. BS7910 and R5)
have been developed. These procedures use a deterministic approach based on certain assumptions
and material properties. Depending on the procedures, methods and material properties used,
significant variability in predictions can be observed. In such circumstances there is a need to
develop statistical and probabilistic approaches in order to increase confidence in the application of
the methods. However, it has been observed that statistical and probabilistic methods may over or
under predict by factor over 100 creep crack initiation and growth (CCI and CCG, respectively).
This is due to two main facts. Firstly, often data from laboratory specimens such as Compact
Tension (CT) or C-Shape (C) is used to predict CCI and CCG in order to reduce the costs.
Secondly, although the procedures adopt the same basic principles, often different formulae are
employed to make an assessment. The main parameters that are used are reference stress, ıref ,
stress intensity factor, K, and the creep strain rate, İ˙ , which are required to estimate the creep
fracture mechanics term C*. Hence, first, it is necessary to select the best parameters in order to
characterise as well as possible the creep crack initiation and growth behaviour in laboratory
specimens as well as real components. This has already been done for some real components such
as pipes and bent pipes subjected to internal pressure as well as plate subjected to bending stress
for different materials [1-3]. In contrast, very little studies has been carried out to find the origin
and the reasons as well as the parameters which makes that one set of parameters predicts well, for
example, the CCI and/or CCG for one specific CT test and under or over estimates significantly
another identical test apart from the applied load and this is the focus of this study.
In this paper, sensitivity analysis of CCI and CCG in pre-cracked CT and C specimens of a
2¼Cr1Mo ferritic steel and a 9%Cr-steel designated as P22 and P91, respectively, is performed.
The test temperature was at 565 and 650ºC for P22 and P91, respectively. The results have been
interpreted in terms of the creep fracture mechanics parameter C*. The experimental C* for CT
and C specimens were obtained according to the recommendations of ASTM E 1457-00 [4]. In
contrast, in making the calculation, the reference stress method based on Eq. (1) has been used to
determine C* for the laboratory specimens [5]. Several formulae can be employed for calculating
reference stress depending on whether it is based on a plane strain (P) or plane stress (P).
Additional parameters such as the yield criterion (Von Mises and Tresca) and the choice of strain
rate (secondary or average, İ˙s and İ˙Ave , respectively) also influence the calculation of C*.
2
§ K ·
C *
V ref ˜ H ref ˜¨ ¸ n 1
A ˜ V ref ˜ K 2 , assuming the Norton law (1)
ref ¨V ¸
© ref ¹
It has been found that, based on the different assumption mentioned above, the discrepancy
between the extreme solutions is greater than 600 in calculating C* from Eq. (1). The parameter
responsible for a most this disparity is the reference stress. In fact, depending on ref chosen, a factor
480 K. Wasmer

ranging between 200 and 300 is obtained. Much less influential is the choice of the strain rate
which is found to vary between 3 and 7 depending on the material, material condition and
temperature.

*
By plotting C Pred * for all CCG tests from the same
(calculated from one solution) vs. C Exp
batch, a ratio between 600 and 2000 is measured between the upper and lower bounds. This
demonstrates that, for laboratory specimens, the selection of the best set of parameters for one
specific test is by far not enough to be certain that this solution will work for all other tests
performed from the same batch of material under the same condition except the applied load. To
establish the origin of this problem, a comparison between Eqs (1) and (2) has been carried out.
The results show that using the Norton low to characterise the strain rate in Eq. (1) is the main
problem. In fact, the difference computed varies from 50 to 700 times.

P ˜ ' c n
C * Exp ˜ ˜ K pl , from ASTM E – 1457 [4] (2)
B ˜ W  a  n  1

References
1. Wasmer, K., Nikbin, K. M. and Webster, G. A., in Proceeding of the ASME Pressure Vessels
and Piping, Vol. 438, Nº 1191: New and Emerging Computational Methods: Applications to
Fracture, Damage, and Reliability, Vancouver, pp: 17-24, 2002.
2. Wasmer, K., Nikbin, K. M. and Webster, G. A., Int. J. of Pressure Vessels and Piping, vol.
80, Issues 7-8, pp: 489-498, 2003.
3. Nikbin, K. M., Yatomi, M., Wasmer, K. and Webster, G. A., Int. J. of Pressure Vessels and
Piping, vol. 80, Issues 7-8, pp: 585-595, 2003.
4. ASTM E 1457-00, Annual Book of ASTM Standards, vol. 3, Issue 1, pp: 936-950, 2001.
5. Ainsworth, R.A. Some Observations on Creep Crack Growth, Int Jnl of Fracture, vol. 20, pp:
147-159, 1982.
 2T16. Creep fracture 481

TEMPERATURE GRADIENT EFFECTS ON THE CREEP BEHAVIOUR OF

STRUCTURES

F. Vakili-Tahami and S. Hasanifard

Faculty of Mechanical Engineering, University of Tabriz, Tabriz, Iran
af.vakili@tabrizu.ac.ir, hasanifard@tabrizu.ac.ir

$
Increasing the operating temperature level beyond 400 C required systematic investigations on
the influence of temperature upon the behaviour of highly stressed components.
Limited creep deformation and longer creep lifetime are the major criteria for the evolving
science of high-temperature component design.
To satisfy the safety and economic requirements for competitive design of engineering
components, two major research fields have been identified. In one field, metallurgists try to
understand the microstructural mechanisms of creep in order to develop new creep resistant alloys.
Parallel to this, in the other field, scientists and engineers try to describe the evolution of creep
deformation and damage for different materials using the empirical or physics-based constitutive
models. The latter field is the frame work of this research study.
To provide a ‘right first time’ design, it is essential to predict the creep behaviour of
components at the design stage using accurate computer modelling techniques. The modelling
procedures can also help to determine the inspection procedures and intervals.
In this paper the effect of temperature gradient on the thermal stress distribution and creep
stress-strain redistribution has been studied over selected structures using uncoupled Finite
Element based numerical analysis.
For this purpose, the steady state temperature gradient has been obtained over a super-heater
pipe-line support using the FE based numerical method.
The temperature profiles for the pipe and the support have been used as input data to obtain
thermal stress distribution and creep stress-strain redistribution over the plate.
In contrast to the low temperature creep, the minimum creep strain rate, H min, at high-
temperatures is a function of stress and temperature:

H min f (V , T )
(1)

In this paper the stress dependency of the minimum creep strain rate,H min, has been expressed
using a power-law equation, whereas the temperature dependency has been expressed using two
different models.
It has been shown that the structural behaviour is severely affected by the existing temperature
gradient. This is due to the exponential relationship between temperature and creep strain rate.
It has also been shown that the creep lifetime can be estimated more accurately using this
method.
482 F. Vakili-Tahami and S. Hasanifard

References
1. Vakili-Tahami, F., Hayhurst, D. R., growth of damage and subsequent creep crack growth
due to Reheat Cracking of A 316H stainless steel welded pressure vessel at 550oC, Internal
Research Report DMM.01.05, Third issue, 19th October 2001, Mech. Aerospace and
Manufacturing Engineering, UMIST.
2. Hall, F.R., Hayhurst, D.R., Continuum damage mechanics modelling of high-temperature
deformation and failure in a pipe weldment, 1991.
3. Perrin, I. J., Hayhurst, D. R., Ainsworth, R. A., Approximate creep rupture lifetimes for butt
and welded ferritic steel pressurised pipes, European Journal of Mechanics, 2000, A/Solids,
19, 223-258.
4. British Energy Generation Ltd., An assessment procedure for the high-temperature response
of structures, R5, Issue 2, Revision 2, 2001.
 2T17. Environment assisted fracture 483

A SURGICAL IMPLANT CREVICE-ASSISTED CORROSION FATIGUE IN-

BODY FAILURE

Hossein Amel-Farzad, Mohammad-Taghi Peivandi1 and S. Mohammad-Reza Yosof-Sani

Department of Materials and Metallurgy Engineering, Engineering Faculty,
Ferdowsi University of Mashhad, Mashhad, Iran
1Department of Orthopaedics, Medical Faculty, Mashhad University of Medical Sciences,

Mashhad, Iran
hh_amel@yahoo.com, drpeivandy@yahoo.com, sani@ferdowsi.um.ac.ir

An orthopaedic stainless steel implant, called “DCS Barrel Plate”, was received which had
fractured inside the patient's body. It had been in his leg for almost two years and it is guessed that
it has been fractured during the first few months, when the bone has been recovering itself, but had
not completed this step; and it had has been forbidden for the patient to walk freely, not to deform
the redonstructing bone. Different investigations containing visual assessments, stereoscopy, SEM
fractography, EDX chemical analysis, metallography, hardness testing and quantometry was
performed to analyze the mechanism of failure and its cause(s). These plates are under cyclic
tensile loading and relaxing during the patient’s walking. They have a bended zone which lets
them fit with the bone better. It is obvious that the plate is mostly stressed in the part in-contact
with the fractured zone of the bone, which is exactly this bended zone in our case, because there is
no bone structure to divide the applied stress, there. It is also necessary to note that, the implant has
a distinct contact with the bone all over its length, containing this bended and frequently tensile
stressed zone. The fracture had occurred exactly from this frequently tensile stressed region. It
must be noted that the alloy was not the standard 316L, but an attentively different austenitic steel
containing 0.115%wt C, and 5.45%wt Mn, 9.995wt Cr and 1.44%wt Ni; which, as follows, has
enough corrosion resistance even in the presence of crevices, but low when static or dynamic stress
applied. A great number of “Crevice Corrosion pits” was observed exactly in this zone and side,
which were sometimes a few millimeters deep. But, none of these dangerous pits had led to the
final fracture. The main fracture has a mechanism of crack growth, initiated from the frequently
tensile stressed region and some other severe crevice corrosion pits formed inside the hole.
Although this place had not been in-contact with bone, but the best sealed location for the
concentration of the Cl- ions, and had been in contact with the screw too. SEM fractography and
EDX chemical analysis was performed from the main fracture surface, some of secondary cracks’
and also pits’ surfaces, which were both opened carefully. The EDX analysis showed the presence
of Cl atoms in the deposits formed inside both cracks and pits; and so supported the guess of the
effect of Cl- ions on both the cracks growth and pits formation. The SEM fractography showed the
fatigue mechanisms, even though there were lots of corrosion deposits all over the fracture surface,
even after severe ultrasonic cleaning. It also showed fretting in some points of the fracture surface,
caused by the frequent loading after the definite fracture. No sign of general classic grain boundary
sensitization was observed. But, severe intergranular cracks were observed in the in-contact zone
and frequently stressed zone, but no other place, during the SEM fractography and also
metallography. Also, the hardness values were about 25 HRC and had no significant change along
the sample. Finally, the main failure mechanism was determined as “crevice effect assisted
corrosion fatigue”. Although the formation of these crevice corrosion pits were assisted by the
applied stress themselves.
484 H. Amel-Farzad et al.

Fig 1. Microstructure of the plate along the plate near the fracture point, containing its in-contact
side, showing crevice corrosion pits, and a continuous net of intergranular cracks near the in-
contact surface leading to deeper crack similar to IGSCC cracks too (a). The plate and its fracture
zone are shown from two different views (b and c). Crevice corrosion pits and the cracks initiated
from these pits are evident in (c). No crevice corrosion pit or crack nor independent or initiated
from the pits, can be seen far from the fracture point. Pay attention that after the nearest holes to
the fracture point, the applied stress is divided between the bone and the plate, and so the applied
stress is much lower, even though both the crevice and the Cl- exist.

Fig 2. The severely rusted fracture surface of the plate, in which the top side is the in-contact one is
evident. As it can be seen the fracture surface is typical of fatigue fractures. White arrows indicate
the initiation points, from which fatigue cracks have initiated inside the hole, in contact with the
screw (a). Photograph of the surface of the biggest crevice corrosion pit in the plate, in which the
rusted brown surface, which contains a long crack too, is the inner side of the hole; and the rusted
and cracked surface in right hand side is the in–contact side whose pre-existing crack has been
opened more during the overloading. The extremely rusted and cracked surface in the center
belongs to the pit, which continues some millimeters till it passes even all over the thickness of the
sample. This extremely deep pit can be seen in the figure 1-c in the vicinity of the fracture surface
of the left hand side part of the plate, at the bottom of the hole. The blue-like fully dimpled surface
is made during the ductile fracture by overloading intentionally in the lab.
 2T17. Environment assisted fracture 485

ASYMPTOTICALLY STABLE GROWTH OF DELAMINATIONS UNDER

HYDROGEN EMBRITTLEMENT CONDITIONS

A. V. Balueva
Spelman College, Department of Mathematics
350 Spelman Lane SW, Atlanta, GA 30314, USA
abaloueva@spelman.edu

Delamination, defined here as separation of the surface layer from the solid body, is observed in
many engineering processes and natural phenomena, ranging from surface buckling of layered
composites to surface fracturing caused by hydrogen embrittlement in metals. For example, in the
case of pipelines for hydrocarbon transport, anti-corrosion, polymer coating sometimes results in
more frequent appearance of small scale delaminations. In time, these delaminations spread under
molecular hydrogen accumulated in a cavity, damaging the coating and allowing the moisture
access to the metal; the result being the external corrosive fracture of the pipeline and its premature
replacement. Understanding the mechanism of protective coating delamination may improve
estimates of the pipeline longevity, thereby creating an essential industrial potential.
We consider delamination crack growth controlled by gas diffusion into the crack. If the gas is
accumulated inside the delamination, after some incubation period, it starts growing under the
pressure of the accumulated gas. An important example is given by hydrogen induced
delamination. Hydrogen absorbed by a metal is typically dissolved in the proton form within the
lattice. Some of the protons reach the surface of pre-existing or freshly created cracks or
delaminations where they recombine with electrons and form molecular hydrogen in the crack
cavity. Because usually the molecular form of hydrogen is thermodynamically more stable, this
process leads to accumulation of hydrogen gas inside the delamination crack. Then, the fracture
often takes place even in the absence of any external loading, that is, only under the excessive
hydrogen pressure.

FIGURE 1. Kinetic dependence of the fracture toughness, KIc, on crack velocity, v.

Since the fracture toughness, KIc, and, therefore, fracture energy, J, in most cases cannot be
assumed crack velocity independent in conditions of hydrogen embrittlement, we use the actual
kinetic dependencies, KIc(v) (e.g. in Fig. 1). As numerically shown in the author’s earlier work for
the internal crack growth, the crack velocity first increases in accordance with the kinetic
486 A. V. Balueva

dependence until reaching some value of vs and remains at the same level afterwards. Since, as has
been obtained in the previous paper by the author, kinetic equations for internal and delamination
cracks are essentially identical, the same conclusion can be derived for the latter. In this paper the
stability and asymptotic approach to the constant velocity of delamination growth is proved
analytically. If vs is known, the corresponding value of fracture toughness, Ks, is obtained by
substituting vs into the full kinetic function, KIc(v), resulting in K s= KIc(vs).
 2T17. Environment assisted fracture 487

CORROSION AND MECHANICAL STRENGTH OF RUSSIAN LIGHT WATER

REACTORS

B. T. Timofeev
CRISM “Prometey”, 49 Shpalernaya street, 193015, St. Petersburg, Russia
mail@prometey.nw.ru

The Russian light water reactors of types WWER-440 and WWER-1000 made of heat irradiation
resistant steels is usually protected from coolant effect by anticorrosive austenitic cladding. Only
ten reactors type WWER-440 of the first generation were manufacturing without protection of
inner surface. Now only four such reactors are operated in Novo-Voronezh and Kola NPPs. Effect
of boric acid on corrosion rate of CrMoV steel was presented in Fig.1. The other type of operating
reactor pressure vessels has anticorrosive protection of inner surface. The thickness of austenitic
clad metal in inner surface of reactor is not more than 9 mm. Not long ago, by the selection of main
reactor dimensions and strength analysis the thickness of this austenitic clad metal was not taken
into account and cladding cracking was not considered at service life assessment. Later this
problem was paid greater attention and some investigations in our country and abroad were carried
out. In these works were studied low-cycle fatigue, fatigue crack growth rate and brittle fracture
resistance of 15Cr2MoVA steel and austenitic-ferritic clad metal.
The coolant influence on low cycle fatigue was investigated for base metal - 15Cr2MoVA
steel. The tests were carried out by biaxial stress state (tension-reduction) in water medium
(pressure – 6 MPa, temperature – 270°C) by various loading frequencies. By this it was established
that the medium effect on fatigue strength of materials is absent even by the frequency of loading
by 20 times from 0,0085 to 0,167 Hz (Fig.2). This may be explained by the formation of protective
magnetic film at the surface of investigated specimens. By small amplitudes of strain this film is
practically not damaged by cyclic loading.

Fig.1 Effect of boric acid on corrosion rate of 15Cr2MoVA steel ( ) and weld metal carried
out by SAW using Sv-10CrMoVTi wire ( ) at 300°
488 B. T. Timofeev

The report presents summarized results on corrosion and mechanical strength of materials for
Russian light water reactors including experience of long-term operation. The comparison of initial
properties of these materials with the analogous characteristics after 100000 and 200000 hours of
operation showed that in most cases the variations were insignificant and the values specified by
the requirements of normative documentation according to PNAE G-7-002-86 for the used types of
steel and according to PNAE G-7-010-89 – for weld metal.
 2T17. Environment assisted fracture 489

CORROSION FATIGUE CHARACTERISTICS OF CF8A STEEL DEGRADED

AT HIGH TEMPERATURE

Seong-Cheol Jang, Dong-Ho Bae1, Gyou-Young Lee, and Seong-Yeb Baek

Student, Graduate School, Mechanical Engineering Department, Sungkyunkwan University, 300
Chunchun-dong, Jangan-gu, Suwon, Kyunggi-do 440-746, Korea.
rydls98@skku.edu
1Professor, School of Mechanical Engineering, Sungkyunkwan University, 300 Chunchun-dong,

Jangan-gu, Suwon, Kyunggi-do 440-746, Korea.

bae@yurim.skku.ac.kr

Even though nuclear power showed remarkable increase as an industrial energy source, recently,
its demand has been slowed down by conservatism of nuclear power industry and market
stagnation. Therefore, many researchers have so far investigated on improving and developing
technologies to maximize the economical efficiency as well as safety maintenance of nuclear
power plant using the destructive and nondestructive approaches. In particular, evaluation of
mechanical characteristics of materials by aging degradation due to corrosion, creep, and fatigue is
actively investigated.
In this paper, as a fundamental study to evaluate fracture characteristics and material
degradation by corrosion, evaluated electrochemical corrosion and corrosion fatigue
characteristics of CF8A steel using as a material of the piping system in nuclear power plant.
CF8A steel was artificially aged at 400qC for 3 months. The environmental test condition is
3.5wt.% NaCl solution of room temperature.
Summarized conclusions are as follows,
1 Corrosion rate of CF8A steel in NaCl solution of room temperature increases with
concentration of NaCl solution increase. However, concentration of NaCl solution will be
more than 4.0wt.%, it showed decreasing tendency. Corrosion rate in 3.5wt.% NaCl
solution was assessed as 4.927mpy that is 2.7 times of one in distilled water (1.846 mpy).
2 Corrosion rate of aged CF8A steel in NaCl solution of room temperature increases with
concentration of NaCl solution increase. However, concentration of NaCl solution will be
more than 4.0wt.%, it shows decreasing tendency. Corrosion rate in 3.5wt.% NaCl solution
was assessed as 11.889 mpy that is 4 times of one in distilled water (2.930 mpy) and 2.4
times of unaged material in the same environmental condition.
3 Crack growth rates of aged and unaged CF8A steel in air condition do not show
remarkable difference. However, in 3.5wt.% NaCl solution, crack growth rates of them
showed higher than ones in air. Particularly, crack growth rate of aged material remarkably
increases compare to one of unaged material.

TABLE 1 Chemical compositions of CF8A (wt.%)

C Mn Si Cr Ni S
0.05 0.626 1.18 19.88 8.74 0.014
490 S.-C. Jang et al.

Figure 1. Comparison of corrosion rates between aged and unaged CF8A steel in NaCl solution

Figure 2. Comparison of da/dN-'K curves of CF8A steel

References
1. ASTM, G5, 73-76, 1987
2. Bandy, R. and Jones, D. A., J. Corrosion, vol. 32, 126, 1976
3. ASTM, E647-00, 578-614, 2000.
4. Marsh, K. J., Smith, R. A., and Ritchie, R. O., EMAS, 11-37, 1990.
5. Wei, R. P. and Brazill, R. L., ASTM, STP 738, 103-119, 1979.
 2T17. Environment assisted fracture 491

MODELING ENVIRONMENT-ASSISTED FATIGUE CRACK PROPAGATION

J.-A. Ruiz-Sabariego and S. Pommier

Lab. of Mechanics and Technology (LMT), 61 av. du Président Wilson, 94235 Cachan, France
ruiz@lmt.ens-cachan.fr
Predicting fatigue crack growth in metals under random loadings remains difficult since fatigue
crack growth is very sensitive to load history. Load history effect in fatigue is known for decades
to stem from plastic deformation in the vicinity of the crack tip. Moreover, history effects in
fatigue are closely related to the elastic-plastic behaviour of the material [1].
The thermodynamics of dissipative processes made possible to establish a model of crack
propagation in mode I, where the variables are namely, the position of the crack tip “a”, the stress
intensity factor “KIf” and the intensity of crack tip blunting “U”. This model contents:
• one elastic-plastic constitutive equation for the cracked structure, established through the
FEM and using an elastic-plastic constitutive behavior of the material

dU § J ·
g ¨I ,... ¸
dt © 2 ¹ (1)
• one cracking law with one fitting parameter D

da dU
D
dt dt
(2)
In this model, the crack growth rate is a time derivative equation (da/dt) instead of a cycle-
derivative equation (da/dN). It has the advantage to avoid explicitly any cycle counting in random
fatigue. The model has already been implemented and tested to predict fatigue crack growth under
random load.
This latter would also be a good way to model environment-assisted fatigue crack propagation,
as diffusion of chemical species, creep or oxidation are time dependent phenomena. The idea is to
transcribe some of the known mechanisms that are responsible of the deterioration of the resistance
against crack propagation of most of the metallic alloys ( “blunting model”, hydrogen-assisted
cracking … [2] ).
For example, in the case of a “blunting model”, adsorbed gas molecules coming from gaseous
atmosphere are responsible for slip irreversibility in the crack tip and thus localize the plastic strain
by favoring the activation of new slip planes. This irreversibility can be model considering D
parameter as a “S” curve, which represents the saturation of the mechanism, function of the
blunting rate. This parameter rather represents oxidation on the crack tip surface [3].

FIGURE 1. Frequency effect with D function of blunting rate.

492 J.-A. Ruiz-Sabariego and S. Pommier

Diffusion mechanism involved in the hydrogen-assisted cracking needs to introduce in the

model a new kinematics variable that would represent the flux of chemical species through the
crack tip. It’s then possible to enrich the model with a E parameter, function of the stress intensity
factor and/or others of the variables representing the load.

da dU
D  E K 
dt dt
(3)

FIGURE 2. Effect of a E constant parameter modeling diffusion mechanism.

References
1. Pommier S et Risbet M, “Partial-derivative equations for fatigue crack growth in metal”, Int
Jal of Fracture. Vol 131. Num 1. Pages 79-106. 2005.
2. J. Petit, G. Hénaff, C. Sarrazin-Baudoux, 1997, Gaseous atmosphere influence on fatigue
crack propagation. In: R.A. Smith Editor, Reliability assessment of cyclically loaded
engineering structures Kluwer Academic Publishers, Dordrecht, pp. 301–342
3. V. V. Bolotin, A. A. Shipkov, J. Appl. Maths Mechs, Vol 65, No. 6, pp. 1001-1010, 2001
 2T18. Dynamic, high strain rate, or impact fracture 493

MEASURING THE FRACTURE RESISTANCE OF COMPOSITES AND

ADHESIVELY BONDED JOINTS AT HIGH TEST RATES.

B. R. K. Blackman, D. D. R. Cartie1, A. J. Kinloch, F. S. Rodriquez-Sanchez and W. S. Teo

Department of Mechanical Engineering, Imperial College London, South Kensington Campus,
London SW7 2AZ, UK.
1Advanced Materials Department, Cranfield University, SIMS bldg 42A, Cranfield,

MK43 0AL, UK.

b.blackman@imperial.ac.uk

Research into the mode I delamination resistance of uni-directionally reinforced composites and
the fracture resistance of adhesively bonded joints led to the publication of two test standards in
2001 [1, 2]. The ESIS TC4 technical committee contributed significantly to the development of
both of these test standards via pre-standardisation round-robin testing, and protocol development.
Since then, the committee has focussed its attention on other important technical issues including
delamination in multi-directional and cross-ply laminates, mode II fracture in both laminates and
adhesive joints, and most recently delamination fatigue and high rate loading. In the present work,
we will describe the strategy and progress made towards developing a protocol for high rate
loading at 1m/s.
Various workers have reported results from performing mode I DCB tests at high rates [3-6]
and at least one review has been published [7]. However, no standards exist, and a number of
technical issues remain to be resolved. Perhaps the first of these is the issue of whether the DCB
test specimen is suitable for high speed testing. The recommended dimensions for the DCB in [1]
specify a minimum beam thickness of 3mm for carbon-fibre reinforced composites and 5mm for
glass-fibre reinforced composites. This permits the manufacture of rather compliant specimens
which tend to suffer from severe flexural wave effects at the higher test rates. These in turn lead to
problems with measuring load, and even high frequency response load-cells can resonate at
loading rates of 1m/s. A second issue is the potential loss of symmetry at high rates, as the
standard quasi-static test applies displacement to one half of the specimen only. This has been
noted to be quite a pronounced effect at loading rates of 10m/s [3]. A third issue is crack stability,
as frequently a transition from stable to unstable, stick-slip crack growth is observed with
increasing test rate.
From an experimental perspective, performing high rate DCB tests requires specialist
equipment. High-speed servo-hydraulic test machines have been available now since the late
1980s and these represent an efficient way to load the samples. Also, various workers have used
drop weight towers to impact and load one half of the specimen. An additional requirement is the
need for high speed recording of the test parameters, e.g. crack length, crack opening displacement
and/or applied load. From an analytical perspective, determining the resistance to crack growth at
high rates requires consideration of the kinetic energy effects in the fracture mechanics energy
balance and also of the transient phenomena that are observed. It is against this background that
the ESIS TC4 group has commenced work to develop a high rate DCB test protocol for composite
laminates and bonded joints. Details of the protocol and of the first experimental results will be
presented.
494 B. R. K. Blackman et al.

References
1. ISO, Standard test method for mode I interlaminar fracture toughness, GIC, of unidirectional
fibre-reinforced polymer matrix composites. 2001. ISO 15024.
2. BSI, Determination of the mode I adhesive fracture energy, GIC, of structural adhesives using
the double cantilever beam (DCB) and tapered double cantilever beam (TDCB) specimens.
2001. BS 7991.
3. Blackman, B.R.K., Dear, J.P., Kinloch, A.J., MacGillivray, H., Wang, Y., Williams, J.G. and
Yayla, P., The failure of fibre-composites and adhesively-bonded fibre-composites under high
rates of test. Part I: Mode I Loading-Experimental studies. Journal of Materials Science, 1995.
30: p. 5885-5900.
4. Blackman, B.R.K., Kinloch, A.J., Wang, Y.and Williams, J.G., The failure of fibre-
composites and adhesively bonded fibre-composites under high rates of test. Part II: Mode I
Loading-dynamic effects. Journal of Materials Science, 1996. 31: p. 4451-4466.
5. Gillespie Jr, J.W., L.A. Carlsson, and A.J. Smiley, Rate-dependent mode I interlaminar crack
growth mechanisms in graphite/epoxy and graphite/peek. Composites Science and
Technology, 1987. 28: p. 1-15.
6. Smiley, A.J. and R.B. Pipes, Rate effects on mode I interlaminar fracture toughness in
composite materials. Journal of Composite Materials, 1987. 21(7): p. 670-687.
7. Cantwell, W.J. and M. Blyton, Influence of loading rate on the interlaminar fracture
properties of high performance composites- A review. Appl Mech Rev, 1999. 52(6): p. 199-
212.
 2T18. Dynamic, high strain rate, or impact fracture 495

QUASISTATIC AND DYNAMIC FRACTURE

OF PEARLITIC STEEL

B. Strnadel, P. Hausild1 and M. Karlik1

Technical University of Ostrava, Department of Materials Engineering,
17. listopadu 15, 708 33 Ostrava, Czech Republic
Bohumir.Strnadel@vsb.cz
1Czech Technical University, Faculty of Nuclear Sciences and Physical Engineering,

Department of Materials, Trojanova 13, 120 00 Praha 2, Czech Republic

Petr.Hausild@fjfi.cvut.cz, Miroslav.Karlik@fjfi.cvut.cz,

The processes of cleavage crack initiation and characteristic features of fracture surfaces in
pearlitic hypo-eutectoid steel (the chemical composition is given in Table 1) have been
investigated using Charpy impact specimens and CT specimens for fracture toughness KIc
assessment. The results demonstrated that initial cracking in both dynamically and quasistatically
loaded specimens was mainly shear cracking of pearlite, e.g. localized slip bands in ferrite
promoted cracking of the cementite plates, which was then followed by tearing of the adjacent
ferrite laths (Fig. 1). Inclusion triggering cleavage exhibiting as facet where cleavage river lines
extended from the particle in its centre was observed as a second fracture micro-mechanism. In CT
specimens cleavage occurred by multi-initiation due to the sharp stress peak ahead of the crack tip
throughout the whole specimen thickness. In Charpy specimens, the cleavage initiation has been
preceded by the ductile tearing changing the stress distribution at the notch root. In consequence of
the ductile crack tunneling, the single cleavage triggering site situated in the centerline of Charpy
specimens was found.

FIGURE 1. Cleavage initiation by shear cracking of pearlite (CT specimen).

TABLE 1. Chemical composition (wt. %)

The results of fracture toughness tests were statistically treated using the following formula
(1):
496 B. Strnadel et al.

4
ª B §K K · º
PF 1  exp «  ¨¨ Ic m
¸¸ ln 2 »
«¬ B o © Ku ¹ »¼
(1)
where KIc is the fracture toughness, PF is the probability of fracture, B is the specimen thickness,
and Bo, Km, and Ku are the parameters.

FIGURE 2. Distribution of the brittle fracture probability.

Graphically the distribution function of fracture toughness is given in Fig. 2. The similar
statistical analysis was carried out on the set of Charpy notch toughness results. The relationship
between both statistical distributions has been established using the results of the fractographic
analysis.
 2T18. Dynamic, high strain rate, or impact fracture 497

FRAGMENTATION IN THE EXPANDING RING EXPERIMENT

H. Zhang and K. Ravi-Chandar

The University of Texas at Austin
1 University Station, C0600
Austin, TX 78712-0235
kravi@mail.utexas.edu

Ever since the inspiring work of Mott [1] on the explosive fragmentation of shells, the expanding
ring test has been used as an effective tool for evaluating the dynamic constitutive and failure
behavior of materials at strain rates of about 104 s-1. Due to the complexity of the experimental
configuration used in the expanding ring test, very few experimental results are available in the
literature. A large database collected by Grady and Benson [2], which shows that the ring fails by
generating a large number of localized necking deformation and a number of fractures at the neck
points, has served as the main benchmark for analytical and numerical investigators. Two different
points of view are prevalent in the literature, one is based on a statistical approach following on
Mott and the other based on the growth of perturbations that localize deformation into necks (see
for example, Guduru and Freund, [3]). Both approaches provide realistic predictions of the
statistics of the fragmentation and its dependence on the rate of loading; hence additional
experimental investigations that are instrumented to identify the appropriate necking and
fragmentation mechanisms at these strain rates to are essential.
In this paper, we report on electromagnetically driven high-strain-rate expanding ring tests
conducted on aluminum rings; this experiment is a recreation of the pioneering work of Niordsen
[4]. A large capacitor is discharged through a copper coil. The coil is held firmly in place to
prevent expansion; the current passing through the coil – on the order of 10 to 20 kA – is measured
with a Rogowski coil. The ring specimen is placed outside this coil with a very small gap filled
with an insulator. A large current is induced in the ring specimen which then interacts with the
field in the coil and propels the ring specimen radially outward at high speeds. With our
experimental setup, copper and aluminum rings can be accelerated to 200 m/s expanding velocity
within about 20 microseconds, corresponding to a strain rate of 104 s-1. Under this loading
condition, the specimen initially develops large plastic strains uniformly around the ring, but at
strains above 35 percent, multiple necks are initiated and growen further into fractures. The main
innovation in our experiments is the use of high-speed and high-resolution photography to study
the dynamic necking and failure behaviors of the rings. A high speed camera capapble of obtaining
pictures at roughly 10 microsecond intervals with spatial resolution of better than 1000 lines per
inch is used to image the process of expansion, necking and final fracture. A selected frame from
the expansion experiment is shown in Figure 1. The effects of expanding velocity, ring radius, and
on the fragmentation of the ring were examined based on image analysis of high speed
photographs and statistical analysis of ring fragments. Our results show that for the aluminum
rings, almost all necks occur nearly simultaneously without much communication between each
other during the ring expansion; the fragmentation process – from the onset of the first fracture to
the total fragmentation of the ring – is complete within about 10 Ps. Since the complete time
sequence of the strain evolution in the ring, the sequence of appearance of necks and fracture
points is identified in the experiment, numerical simulations of any particular experiment, with the
appropriate comparison to the measured kinematic quantities, can be performed to infer the
constitutive and failure properties of the ring material. A comparison of the experimental results
with numerical examinations based on an assumed elastic-plastic constitutive response for the
material will be presented.
498 H. Zhang and K. Ravi-Chandar

Figure 1. Fragmentation inan expanding ring experiment; annealed Al 6061 alloy with an initial
ring diameter of 30 mm.

References
1. N.F. Mott, Fragmentation of shell cases, Proc of the Royal Soc London, Series A,
Mathematical and Physical Sciences, 189, 300-308, 1947
2. D.E. Grady and D.A.Benson, Fragmentation of metal rings by electromagnetic loading, Exp
Mech, 12, 393-400, 1983.
3. P.R. Gududru and L.B. Freund, The dynamics of multiple neck formation and fragmentation
in high rate extension of ductile materials, Int J of Solids and Struct, 39, 5615-5632, 2002.
4. F.I. Niordson, A unit for testing materials at high strain rates, Exp Mech, 1965, 5, 23-32.
 2T18. Dynamic, high strain rate, or impact fracture 499

INFLUENCE OF FRICTION ON RESULTS OF AN INSTRUMENTED IMPACT

TEST

I. V. Rokach
Kielce University of Technology
Al. 1000-lecia PP 7, 25-314 Kielce, Poland
rokach@tu.kielce.pl

Contrary to the quasi-static case, where adjustable rollers used as supports reduce effect of friction
significantly, the fixed supports are usually used in dynamic tests. Thus, there is always some
friction in the specimen/support contact zones during an impact test. In the literature related to the
numerical modelling of an impact fracture test, the problem of friction between a specimen and
anvil has not attracted much attention. From the very beginning of the impact test modelling it was
clear that friction affects results of a test considerably (see Saxon et al. [1]). However, detailed
quantitative analysis of friction caused changes in time variation of contact forces and dynamic
stress intensity factor (DSIF) is still unavailable. In this work, some results of the finite element
(FE) modelling of an impact test for a wide range of specimen configurations and values of
coefficient of friction are presented.

FIGURE. 1. Impact specimen model.

Let us consider an impact specimen loaded by the striker force F(t) and normal and tangential
anvil forces Rn(t) and Rt(t), respectively (Fig. 1). In this study, the simplest form of the friction law
(Coulomb law with constant sliding coefficient of friction (CoF) f) has been considered. This law
assumes that |Rt(t)| = f Rn(t) if there is sliding in the specimen/support contact zone and |Rt(t)|< f
Rn(t) otherwise.
In quasi-static case, both elementary beam theory analysis and FE computations show that
monotonic growth of the bending force (i) always causes sliding of the specimen in the specimen/
support contact zone in the outside direction with respect to crack tip and (ii) corresponding
friction forces act as a compressive load for the crack tip zone. The opposite is true for the
decreasing load case. Thus, quasi-static stress intensity factor (SIF) KIqs(t,f) can be calculated as

K Iqs ( t , f ) K Iqs ( t , 0 )(1  sign( F ) fg ( O )) (1)

where KIqs(t,0) is the SIF for the same loading in the frictionless case, =a/W, the dot means
derivative with respect to time. Using the results of Bakker [2] and Kaya and Erdogan [3], it was
500 I. V. Rokach

found that for = 0.01-0.8 function g() can be approximated by 0.157 + 0.353 with accuracy of
about 1%.
For the preliminary stage of specimen deformation during an impact test, high oscillations of
both tup and anvil forces are observed. During this stage, each friction force can either change its
sign along with the sliding direction of the specimen with respect to the support or its absolute
value can be lower than |fRn(t)| when the specimen is stick with the support. Due to these reasons,
predictions of quasi-static theory and results of dynamic analysis could be different.
Dynamic calculations for the plane stress model of the impact specimen have been performed
using commercial FE program ADINA 8.0.2 for the following sets of parameters: L/
W=4.5(0.5)6.0, =0.3(0.1)0.6, f=0(0.1)0.8, rt/W=0.2, ra/W=0.25. All calculations have been
performed for:
• Poisson’s ratio of the specimen material equal to 0.3

• Constant impact velocity vi = 7 .5 u 1 0  4 (E/U)1/2 (that corresponds to about 4 m/s for

steels), where E and U are the Young’s modulus and density of the specimen material,
respectively
• Perfectly stiff striker and supports.
The following conclusions have been made from the results of calculations:
1 Friction between the specimen and the supports considerably affects contact forces and
DSIF during a dynamic fracture test and should be taken into account especially for high
CoF values.
2 Contrary to the quasi-static case, the relative movements of the specimen in the specimen/
support contact zone are quite complex and include movement in the reverse direction and
temporal stick conditions. Due to this reason friction force is not always equal to the
normal contact force multiplied by sliding CoF.
3 In general, friction causes increasing of the amplitude of DSIF oscillation and reduction of
the mean value of the DSIF. A simple formula for prediction of amount of this reduction
has been proposed.

References
1. Saxon, H.J., Jones, A.T., West, A.J. and Mamaros T.C. In Instrumented Impact Testing,
ASTM STP 563, American Society for Testing and Materials, Philadelphia, 1974, 30-49.
2. Bakker, A., Fat. & Fract. Engng Mat. & Struct., vol. 13, 145÷154, 1990.
3. Kaya, A.C., Erdogan, F.: Int. J. Fract., vol. 16, 171÷190, 1980.
 2T18. Dynamic, high strain rate, or impact fracture 501

INFLUENCE OF MOISTURE CONTENT ON THE DYNAMIC BEHAVIOUR OF

CONCRETE

I. Vegt and J. Weerheijm

Delft University of Technology, Faculty of Civil Engineering and Geosciences
P.O. Box 5048, 2600 GA Delft, The Netherlands
I.Vegt@citg.tudelft.nl

Terrorist attacks, explosion scenarios in tunnels and the potential hazards from storage of high
energetic materials have become important safety issues. Knowledge about the response of
concrete structures to impact and explosive loading is required for reliable safety assessment and
the design of protective structures. A complicating factor is the fact that concrete is a rate-
dependent material, which means that the mechanical properties of concrete depend on the applied
loading rate. The mechanical response of structures exposed to explosive loading can only be
predicted properly with material models that include this rate effect.

FIGURE 1. Experimental results on rate effect in concrete

The rate dependency of concrete can be divided into two regimes; the regime with moderate
rate effects for loading rates in the range from 10-9 s-1 (static loading rate) up to 1 s-1, and the
regime with extensive rate effects for loading rates beyond 1 s-1, see Fig. 1. The moisture content,
Rossi [1], and the change in geometry of the fracture planes, Körmeling et al [2], are the main
causes of the increase of the material parameters of concrete for the moderate regime. It is assumed
that the free water in the micro-pores exhibit the so-called “Stéfan effect” causing a strengthening
effect in concrete with increasing loading rate. This is the phenomenon that occurs when a viscous
liquid is trapped between two plates that are separated quickly, causing a reaction force on the
plates that is proportional to the velocity of separation. The second explanation of the rate-
dependency is based on the alteration of the fracture planes when the loading rate increases. The
cracks are forced to develop along a shorter path of higher resistance, which results in a higher
strength. The micro-inertia effects in the fracture process zone mainly cause the rate dependency of
concrete in the high regime, Weerheijm [3]. The contribution of moisture content in this regime is
still unknown.
This study aims to combine experimental research on the material properties of concrete,
numerical models and microscopic research to describe and understand the basic mechanisms that
cause the rate effect in concrete at moderate and high loading rates.
To determine the dynamic tensile properties of concrete under moderate strain rates (1 s-1) the
gravity driven Split Hopkinson Bar (SHB) at the Stevin laboratory of the Delft University of
502 I. Vegt and J. Weerheijm

Technology is used. The specimens have a length of 100 mm and a diameter of 74 mm. The
moisture content of the specimens is varied (normal, wet and dry). The tensile strength as well as
the fracture energy are determined. As a reference deformation controlled static tests with a strain
rate of 10-9 s-1 have been conducted. After the impact tests, the specimens are impregnated with
epoxy and studied with an optical microscope to study the width of the fracture zone and the crack
patterns at the different moisture levels. The material properties of concrete at moderate loading
rate and the dynamic/static ratio of the strength and fracture energy are determined. Preliminary
results show an increase in tensile strength for all three moisture contents. The high moisture level
(wet) shows a larger dynamic/static strength ratio than the normal and low moisture level (dry). A
similar trend is observed in the results of the fracture energy. The results of the SHB-tests, the
static tests and the post-test examination of the fracture planes will be presented and discussed in
this paper. The influence of moisture on the rate effect and the changes in the fracture zone are also
discussed.
The rate dependent tensile material properties emerging from the experimental research are not
the true material properties since inevitably the combined material and structural response is
recorded. To eliminate the effect of structural behaviour from the results inverse computational
modelling is applied. The strength, fracture energy, time to failure and the characteristics of the
fracture zone resulting from the experiments are used as input parameters for the numerical model.
In literature various rate dependent material models are proposed. Most of these models reproduce
the strength rate dependency quite well. Data and knowledge of the rate effect on the failure
process, the softening branch, is very scarce and consequently the modelling has been quite
speculative hitherto. In the computational study of the SHB tests a visco-elastic visco-plastic
damage model [4] was used. An attempt to deduct the true material behaviour is covered in this
paper.

References
1. Rossi, P., Materials and Structures, Supplement March 1997, 54-62
2. Körmeling, H.A., Zielinsky, A.J. and Reinhardt, H.W., Experiments on concrete under single
and repeated uniaxial impact tensile loading, Delft University of Technology, Report No. 5-
80-3, 1980
3. Weerheijm, J., Concrete under impact tensile loading and lateral compression, Dissertation,
Delft University of Technology, 1992
4. Pedersen, R.R., Sluys, L.J., Vegt, I., Weerheijm, J. and Simone, A., Modelling of impact
behaviour of concrete – a computational approach, Submitted to ECF16
 2T18. Dynamic, high strain rate, or impact fracture 503

STRENGTH AND TOUGHNESS PROPERTIES OF STEELS UNDER

DYNAMIC LOADING

J. Fang
Technical Center, Baoshan Iron & Steel Co., Ltd.
Kedong Road, Fujin Road, Shanghai 201900, China.
fangjian@baosteel.com

It is feasible and convenient to prepare miniature round bar and Charpy V notch specimens
simultaneously from structural steels for conventional tensile, Charpy impact and impact tensile
testing, which provides the relative testing rate high up to 102 /s for uniaxial loading and 5×105
MPa·m1/2·s-1 for impact in this study, achieving the possibility of acquiring the dynamic response
properties of HSLA combined with its quasi static performance all together.

FIGURE 1. Illustration of the impact equipment and specimens.

As a result, the strength property of X80 pipeline steel is proved to be dependent of the loading
rate. The four order of magnitude does rate increase by, its yield stress and tensile strength rise by
100MPa. As to remove the noise signals due to the vibration and inertia effect on the impact tensile
stress-strain curves, dynamic flow yield stress is nominally calculated as the average of first two
peak and valley stress values on the curve, see Fig.2.
The validation of the Server equation (1) for the estimation of dynamic yield stress by means
of general yield force, i.e. Fgy by instrumented impact was also studied by performing correlated
testing between impact tensile and Charpy impact. The linear relationship of Vyd with Fgy is
authenticated satisfactorily, whereas the coincidence of dynamic tensile strength, i.e. Vbd with Fm
is void. Moreover, crack extension resistance curve J-'a was established for monitoring the energy
absorbed by crack initiation and propagation, which implicates the toughness property of HSLA
under dynamic loadings, as Fig. 3.

S
V yd 2 . 99 F gy (1)
4 B (W  a 0 ) 2
504 J. Fang

FIGURE 2. Stress-Strain curves obtained by tensile under dynamic and static loading.

FIGURE 3. Dynamic crack extension resistance curves.

As above, quasi static tensile, impact tensile and instrumented Charpy impact testing are
integrated into the unique experimental platform, on which the all-round assessment of material
strength, plasticity and toughness properties could be accomplished with the strain loading rate
from 10-2-102 /s and various simulated loading conditions such as tensile or impact.

References
1. Server W.L., J. Engng. Mat. Technol., Trans. ASME, vol. 100, 183-188, 1978.
2. Kobayashi T., Yamamoto I., Niinomi M., J. of Testing and Evaluation, JTEVA, vol. 21, 145-
153, 1993.
3. Bohme W. In Evaluating Material Properties by Dynamic Testing, edited by E. van. Walle,
Mechanical Engineering Publication, London, 1996, 1-23.
4. Fang J., Ding F.L., Wang C.Z., J. Phys. IV France, vol. 110, 551-557, 2003.
 2T18. Dynamic, high strain rate, or impact fracture 505

RUBBER PARTICLE SIZE EFFECT ON IMPACT CHARACTERISTICS OF PC/

ABS (50/50) BLENDS

M. Nizar Machmud, Masaki Omiya, Hirotsugu Inoue and Kikuo Kishimoto

Department of Mechanical and Control Engineering, Tokyo Institute of Technology,
2-12-1 O-okayama, Meguro-ku, 152-8552, Tokyo, Japan
nizar@mep.titech.ac.jp, oomiyam@mep.titech.ac.jp, inoueh@mep.titech.ac.jp,
kkishimo@mep.titech.ac.jp

Impact charateristics of simply-supported circular thin plates made of PC/ABS (50/50) blends
having 10 wt% content of rubber with a rubber particle diameter of 270 nm called PA-1 and of
150-170 nm called PA-2 tested at room temperature by use of a drop weight impact apparatus
under different speeds: 2, 3, and 4 m/sec has been studied. Features of the target were viewed to
describe definite alteration of the plates induced by a hemispherical tip-ended cylindrical striker
due to effect of rubber particle size distributed in the blends.
It was found that the blends with a rubber particle diameter of 150-170 nm were not in
shattering and exhibited a multiaxial crack shape at the speed of 3 m/sec. Equivalent ABS
materials of the blends were then examined at the similar speeds. The results of this study showed
that brittle ABS materials produced tough PC/ABS blends. It was apparent that distribution of
bimodal type with smaller rubber particle toughened the PC/ABS blends at room temperature.
Examination by use of scanning electron microscope (SEM) on fracture surface of the failed
specimens was also carried out. The brittleness and toughness of the PC/ABS blends at the speed
of 3 m/sec were then analyzed using an optical microscope and a transmission electron microscope
(TEM) for observation on global and local deformation, respectively.
Drop weight impact test
The drop weight impact test has been carried out on our laboratory apparatus at room
temperature under a range of impact speeds: 2, 3 and 4 m/sec. Behavior of the blends was induced
by a hemispherical tip-ended cylindrical striker made of stainless steel (SS55) with diameter of 20
mm that struck perpendicularly the center of the simply-supported circular thin plate with an initial
thickness of 2 mm and diameter of 99.3 mm. The body of the impactor attached to guiding rods
was designed to give a total impactor mass of 2.4 kg. Description of the test is presented in Fig. 1.
The range of impact speeds v was calculated from

v 2gh d
(1)
where g is the acceleration due to gravity, and hd is drop height measured during the impact tests.
Although the impactor speed is presumed to be accurately predicted by this formula, the actual
speed may be lower than the calculated speed due to friction caused by the guiding rods. However,
other researchers have verified that the friction has a negligible effect on the accuracy of speed
prediction using the formula [9].
506 M. N. Machmud et al.

a) b) c)
Fig. 1 Drop weight impact test: (a) test apparatus, (b) specimen, (c) striker

Materials

Table 1. Properties of PC/ABS blends with monomodal and bimodal types

References
1. R.A. Kudva, H. Keskkula and D.R. Paul, Polymer, vol. 41, 225-237, 2000
2. Th. Seelig and E. van der Giessen, Int. J. Solids and Structures, vol. 39, 3505-3522, 2002
3. J.P.F. Inberg and R.J. Gaymans, Polymer, vol. 43, 2425- 2434, 2002
4. J.P.F. Inberg, A. Takens and R.J. Gaymans, Polymer, vol. 43, 2795-2802, 2002
5. J.P.F. Inberg and R.J. Gaymans, Polymer, vol. 43, 3767-3777, 2002
6. W. Jiang, H. Liang, J. Zhang, D. He and B. Jiang, J. Appl. Polym. Sci., vol. 58, 537-539, 1995
7. C.M. Tai, R.K.Y. Li and C.N. Ng, Polymer Testing, vol. 19, 143-154, 2000
8. W.G. Perkins, Polymer Engineering and Science, vol. 39, 2445-2460, 1999
9. L.S. Kistler and A.M. Waas, Int. J. Impact Engng, vol. 21, 711-736, 1998
10. B.C. Simonsen and L.P. Lauridsen, Int. J. Impact Engng, vol. 24, 1017-1039, 2000
 2T18. Dynamic, high strain rate, or impact fracture 507

EFFECT OF STRAIN RATE ON MECHANICAL PROPERTIES OF

REINFORCED POLYOLEFINS

Marcus Schossig, Christian Bieroegel, Wolfgang Grellmann, Reinhard Bardenheier1 and Thomas
Mecklenburg2
Institute of Materials Science, Martin-Luther University Halle-Wittenberg,
D-06099 Halle/Saale, Germany
1Instron Limited, HP12 3SY High Wycombe, UK
2Basell Polyolefine GmbH, D-65926 Frankfurt, Germany

marcus.schossig@iw.uni-halle.de

This paper deals with results of high-speed tensile tests on glass-fibre reinforced thermoplastic
materials. Starting point of the investigations is the fact, that requirements for a modern material
have been steadily increased. One reason here is the increasing use of computer-aided design over
the past years. Especially for the automotive industry, not only the mechanical properties as results
of quasi-static tests are of interest. According to EuroNCAP crash tests, strain rates up to 200 s-1
can be observed in the region of crash boxes at 64 km/h initial speed (Fig. 1a) [1, 2]. Therefore, it
is necessary to perform dynamic tests with strain rates up to maximum values of 104–106 s-1 [3, 4].
For example, with commercially available servo-hydraulic testing machines strain rates between
10-3 to 103 s-1 can be realized (Fig. 1b).

a) b)
FIGURE 1. a) Schema of strain rates in crash tests [2] and b) spectrum of strain rates [3]

At the moment, there is no standard for the execution of high-speed tensile tests. A tentative
draft of the High Strain Rate Experts Group from the International Iron and Steel Institute for
metals was presented in March 2004 [5]. However, a corresponding draft for polymeric materials
does not exist. Furthermore, in the literature high-speed tensile tests of polymers are hardly
described. This is due to experimental difficulties during performance of high-speed tensile tests
and high costs. The results shown in this paper will contribute to a better understanding of
deformation behaviour of polymeric materials under high strain rates.
Aim of the investigations was to assess the influence of strain rate on stress–strain behaviour of
glass-fibre reinforced materials. For this reason, polypropylene and polybutene materials with
glass-fibre contents of 0, 20, 30 and 40 wt.-% were examined. Polybutene has a wide application
potential. Presently it is used mainly as unreinforced pipe material and new application fields can
be opened by the addition of glass fibres.
The high-speed tensile tests were performed with specimen type 1A according to ISO 527-2; a
clamping of 115 mm was used for a better comparability with the results from quasi-static tests.
508 M. Schossig et al.

The tests were carried out with an Instron servo-hydraulic testing machine. The load was measured
with the help of a piezoquarz load cell and the displacement was measured by Linear Variable
Differential Transducer (LVDT). The following test speeds were realized: 1/120, 1/2, 1, 2, 5, 10
and 20 m/s. The corresponding strain rates, at a clamping length of 115 mm, are in the range of
0.0725 to 174 s-1.
With increasing the strain rate, an increase of the strength and a decrease of the deformation
ability can be detected for the investigated materials (Fig. 2a). This behaviour is described in the
literature as positive strain rate dependence [3, 6]. A change of the fracture behaviour with
increasing content of glass fibres at a constant strain rate can be observed. The fracture behaviour
of the polypropylene materials changes from single break over multiple breaks to multiple splinter
break (Fig. 2b).The polybutene material system shows a transition from single break to multiple
break at higher glass fibre contents.

a) b)
FIGURE 2: a) Schematic stress–strain curves and b) fracture appearance for various glass-fibre
reinforced polypropylene materials at a strain rate of 87s-1

References
1. Frontal Impact Testing Protocol, European New Car Assessment Programme (EuroNCAP),
March 2004.
2. Werner, H. and Gese, H., Importance of strain rate dependent material paramters for crash
simulations, DVM-Tagung "Werkstoffprüfung 2002 – Kennwertermittlung für die Praxis",
Bad Nauheim, December 5–6, 2002.
3. Bardenheier, R. and Rogers, G., Dynamic Impact Testing, Instron Ltd., 2003.
4. Bleck, W. and Larour, P., Effect of Strain Rate and Temperature on the Mechanical Properties
of LC and IF Steels, RWTH Aachen,
5. Borsutzki, M., et al., Recommendations for Dynamic Tensile Testing of Sheet Steels, High
Strain Rate Experts Group, International Iron and Steel Institute, March 2004.
6. Guden, M. and Hall, I. W., High strain-rate compression testing of a short-fiber reinforced
aluminum composite, vol. A232, 1–10, 1997
 2T18. Dynamic, high strain rate, or impact fracture 509

FRACTURE RELATED MECHANICAL PROPERTIES OF AIRCRAFT CAST

ALUMINUM ALLOY A357

N. D. Alexopoulos
Laboratory of Technology and Strength of Materials
Department of Mechanical and Aeronautical Engineering
University of Patras
Panepistimioupolis Rion, 26500 Patras, Greece
nalex@mech.upatras.gr

Precision casting is currently attracting considerable attention as a reliable manufacturing process

for producing aeronautical and automotive aluminum components of complex shape geometries
cost efficiently. Inferior mechanical properties, specifically in terms of ductility and toughness,
and increased scatter compared to the respective wrought aluminum alloys, represent serious
drawbacks for their increased exploitation in aeronautical applications. The tighter controls
currently applied during the casting process, as well as the advancements on the casting processes
and the better understanding of the physical metallurgy background of the age-hardened aluminum
alloys led to an improvement of the material quality and, hence, to an appreciable increase of the
competitiveness of aluminum casting products. The most widely used alloy for the above
applications is the precision-hardened Al-Si-Mg A357 cast alloy. Alloy A357 with minor
modifications in chemical composition has been extensively investigated in recent research
projects, e.g. ADVACAST [Anon., Advanced Aluminum Precision Casting For Integrally
Stiffened Net - Shape Components (ADVACAST). Final Technical Report of the BRITE Project
4084, Brussels, Belgium, 1996.], YPER [Alexopoulos, N.D., Development of new near net shape
cast aluminum alloys with improved mechanical properties. Final Technical Report of the Project
YPER 94/274, Greek General Secreriat of Research and Technology, Patra, Greece, (in Greek)
2002.] and PABE [Anon., Production and development of cast aluminum components for
aeronautical applications, Final Technical Report of the Project PABE 96BE/219, Greek General
Secreriat of Research and Technology, Bolos, Greece, (in Greek) 2000.].
The characterization of the quality of a cast alloy involves non-destructive inspection,
quantitative metallography and mechanical testing, e.g. Alexopoulos [Alexopoulos, N.D., Int.
Journal of Cast Metals Research, submitted for publication, 2005.]. Hardness, tensile and impact
tests are the tests currently used to characterize the quality of a cast aluminum alloy in terms of
mechanical performance. The demands for increased damage tolerance abilities of cast aluminum
alloys have made fracture and tensile toughness important properties. Notice that in aircraft and
automotive industry, certain minimum values in tensile strength, ductility and fracture toughness
are prerequisite for considering a material as a candidate for structural applications. e.g. [Anon.,
Advanced Aluminum Precision Casting For Integrally Stiffened Net - Shape Components
(ADVACAST). Final Technical Report of the BRITE Project 4084, Brussels, Belgium, 1996.].
Hardness measurements and tensile mechanical properties of the aircraft cast aluminum alloy
A357 have been evaluated in Alexopoulos and Pantelakis [Alexopoulos, N.D. and Pantelakis,
Sp.G., Materials & Design, Vol. 25, 419-430, 2004.] for 25 different artificial aging heat treatment
conditions. Yield strength, tensile strength, elongation to fracture and strain energy density W were
evaluated from all tensile tests. Strain energy density W was evaluated as the area below the stress-
strain curve, accounts for both, ductility and toughness and is actually the energy given per
material volume for the material fracture. In Alexopoulos and Pantelakis [Alexopoulos, N.D. and
Pantelakis, Sp.G., Metallurgical and Materials Transactions A, Vol. 35A, 3079-3089, 2004.],
impact Charpy V-notch tests were performed for the same heat treatment conditions of the A357
510 N. D. Alexopoulos

alloy. Impact resistance RCVN was evaluated from the impact tests as the fraction of the material’s
energy to fracture to the specimen’s cross-section.
In the early 70’s the empirical equation (1) was proposed by Barson and Rolfe [Barson, J. and
Rolfe, S., In Impact Testing of Metals, ASTM STP 466, Ohaio, 1970, 281-302.], for both, low-
alloyed and austenitic steels to calculate the alloy’s plane strain fracture toughness KIc for known
Charpy V-notch impact energy WCVN and yield strength values Rp. The m and n coefficients are
both, empirical and material dependant constants. However, the correlation of the material’s
fracture toughness and impact energy has not been reported on the open literature for aluminum
alloys.

2 2
§ K Ic · §W ·
¨ ¸ m ˜ ¨ CVN ¸ n
¨ R ¸ ¨ R ¸
© p ¹ © p ¹ (1)
In the present work, the material property impact resistance RCVN has been empirically related
to the material’s strain energy density W. To this end, the strain energy density W has been
involved in equation (1) instead of the plane strain fracture toughness KIc, for the case of cast
aluminum alloys. According to Yeong et al. [Yeong, D., Orringen, O. and Sih, G., Journal of
Theoretical and Applied Fracture Mechanics, Vol. 22, 127-137, 1995.], strain energy density W is
proportional to KIc2, thus providing a background for this assessment. Different empirical
correlations are proposed to establish useful relationship between RCVN and W. Performed
fractographic analyses by using a scanning electron microscope are supporting the physically
arbitrary correlation of tensile strain energy density and impact resistance. The above correlations
can be used to estimate the ductility and toughness of the under-consideration A357 material, from
a Charpy impact test.

References
1. Anon., Advanced Aluminum Precision Casting For Integrally Stiffened Net - Shape
Components (ADVACAST). Final Technical Report of the BRITE Project 4084, Brussels,
Belgium, 1996.
2. Alexopoulos, N.D., Development of new near net shape cast aluminum alloys with improved
mechanical properties. Final Technical Report of the Project YPER 94/274, Greek General
Secreriat of Research and Technology, Patra, Greece, (in Greek) 2002.
3. Anon., Production and development of cast aluminum components for aeronautical
applications, Final Technical Report of the Project PABE 96BE/219, Greek General Secreriat
of Research and Technology, Bolos, Greece, (in Greek) 2000.
4. Alexopoulos, N.D., Int. Journal of Cast Metals Research, submitted for publication, 2005.
5. Alexopoulos, N.D. and Pantelakis, Sp.G., Materials & Design, Vol. 25, 419-430, 2004.
6. Alexopoulos, N.D. and Pantelakis, Sp.G., Metallurgical and Materials Transactions A, Vol.
35A, 3079-3089, 2004.
7. Barson, J. and Rolfe, S., In Impact Testing of Metals, ASTM STP 466, Ohaio, 1970, 281-302.
8. Yeong, D., Orringen, O. and Sih, G., Journal of Theoretical and Applied Fracture Mechanics,
Vol. 22, 127-137, 1995.
 2T18. Dynamic, high strain rate, or impact fracture 511

SHEAR FAILURE OF TI-6AL-4V BY DIRECT IMPACT AND ANALYSE OF

THE PROCESS OF ELASTIC AND PLASTIC WAVE PROPAGATION

P. Chwalik, A. Rusinek and J. R. Klepaczko

Laboratory of Physics and Mechanics of Materials, UMR 7554
University Paul Verlaine, Ile du Saulcy, 57045 Metz Cedex 01, France
chwalik@lpmm.univ-metz.fr, rusinek@lpmm.univ-metz.fr, klepaczko@lpmm.univ-metz.fr

An Adiabatic Shear Banding (ASB) in Ti–6Al–4V alloy is studied in this work by application of
the finite element technique. Two approaches of impact shearing have been considered for a wide
range of impact velocities, that is an infinite layer and a finite band. The Critical Impact Velocity
(CIV) in shear was estimated.
A special experimental technique developed at LPMM-Metz has permitted shear tests at the
strain rates up to 105 s-1. After analysis of experimental data available in the open literature for Ti-
6Al-4V alloy and the results obtained with the MDS technique, the following explicit form of the
constitutive relation has been implemented in LPMM-Metz [1-3]
Q 1/ m
P (T ) ª § T · § T * · º
W «B¨ ¸¸ *  *   n T 
 ¨¨ 1  log 0 ¸¸ »
P 0 « ¨© T 0 * ¹ »
p
¹ © D
¬ ¼ (1)

P (T ) P 0 1  AT  CT 2
 (2)

§ T ·
n (T ) n 0 ¨¨ 1  ¸¸
© T m ¹
(3)
B, µ0, Q, n, m are respectively, the modulus of plasticity, the shear modulus at T=0K, the
temperature index, the strain hardening exponent and the logarithmic rate sensitivity, T0, *p, *0
and D are normalization constants, also T, Tm, A, C respectively, the absolute temperature, melting
point temperature, material constants. The constitutive relation accounts for strain hardening, rate
sensitivity and thermal coupling. In this way a complete dynamic approach with elastic-plastic
wave propagation, ASB development and failure could be considered.
After identification of all material constants for Ti-6Al-4V, the relations (1)-(3) have been
introduced into the finite element code ABAQUS Explicit. Many aspects in FE analysis were
optimized in [1], for example the meshes density, boundary and initial conditions, contacts etc. All
2D simulation was assumed with plane stress elements CPS4R (four nodes, reduced integration).
An innite-length layer with 1% geometrical imperfection defined in the layer thickness was
submitted to shear [1]. The apparition of CIV was observed for impact velocity 94 m/s. The CIV
occurs due to trapping of plastic deformation triggered by plastic waves. Different risetimes were
chosen and different constitutive relations (for example Johnson-Cook [4]) were also studied. In
this way the results obtained for shear by application of constitutive relation (1-3) are very close to
the localisation phenomena observed by Klepaczko and Rezaig [5], and by Rusinek et al. [6] in the
case of tension. The wave reflection leads to different strain localisation.
512 P. Chwalik et al.

FIGURE 1. Impact shearing: layer at 20m/s (a), layer at 100m/s (b), ASB formation in shearing
specimen at 20m/s (c), and 100m/s (d).

As the second task, the development and evolution of ASB and failure in double shear
specimens used in LPMM was analysed. In order to calculate numerically an evolution of the
plastic zones and failure, it is necessary to introduce a local failure criterion defined by (5). Value
of the critical strain *f can be determined as proportional to the instability strain defined by (wW/
w*)A=0 in the adiabatic conditions of deformation with the proportionality coefficient D.

* f ( * ) D * i ( * )
(5)
At lower impact velocities, for example 20m/s, Fig 1a and c, the localization of shear band is
situated in the middle of the sheared zone. When the impact velocity increases approximately to
100 m/s, Fig 1b and d, the shift of the band is observed in the direction of the impact side of
specimen. The value of CIV is similar to these estimated for an infinite layer.
The constitutive relation developed in LPMM offers a good possibility of implantation in finite
element codes like ABAQUS Explicit. It has permitted to study the ASB formation, elastic and
plastic wave propagation and to estimate the CIV for Ti-6Al-4V.

References
1. Klepaczko, J.R., Klosak, M., Eur. J. Mech. A/ Solids, vol. 18, 93-113, 1999.
2. Rusinek, A., Klepaczko, J.R., Int. J. of Plasticity, vol. 17, 87-115, 2001.
3. Chwalik, P., Ph.D Thesis, University of Metz, 2005
4. Seo, S., Min, O., Yang, B., Journal of Impact Engineering, vol. 31, 735–754, 2005.
5. Klepaczko, J.R., Rezaig, B., Mechanics of Materials, vol. 24, 125-139, 1996
6. Rusinek, A., Zaera R., Klepaczko J.R., Cheriguene R., Acta Materialia, submitted 2005.
 2T18. Dynamic, high strain rate, or impact fracture 513

EVALUATING OF FRACTURE MECHANICS PROPERTIES AT

INTERMEDIATE STRAIN RATES, TRANSFERABLE TO COMPONENTS

Peter Trubitz, Annette Ludwig, Gerhard Pusch and Hans-Peter Winkler1

TU Bergakademie Freiberg
Institute of Materials Engineering
Gustav-Zeuner-Str. 5, Freiberg, 09599, Germany
trubitz@ww.tu-freiberg.de
1GNS Gesellschaft für Nuklear-Service mbH

Hollestraße 7 A, 45127, Essen, Germany

The evaluation of fracture mechanics properties at intermediate strain rates can be carried out on
the basis of different standards. The starting point of the lecture is a comparative evaluation of the
relevant standards and test procedures [1–4]. The aim of the fracture mechanics test at intermediate
strain rates are not only generally materials characterizations but often the portability of the
determined parameters on bigger sizes in case of evaluation of real components. In this situation
the definition of component transferable material parameters is an important problem. In this
context the crack initiation point gets an all-dominant importance [5, 6]. In case of multiple
specimens testing typically the definition of the crack initiation point based on the results of
stretch-zone measurements.
With the scope of a bigger research program for the evaluation of the behaviour of bulk
containers made of spheroidal graphite cast iron (ductile cast iron DCI) grad GJS-400 [7] one task
is the determination of component transferable fracture mechanic values at intermediate strain
rates at different temperatures. This group of materials shows damage behaviour at intermediate
strain rates without typical stretching, located at the crack tip. Therefore the crack initiation value
can not by determined by stretch-zone measurements. By static loading crack growth was detected
as a process of dissociation of the graphite particles and crack growth between the particles
(“graphite-particle-blunting”). Based on this scanning electron fractography of the DCI a
characteristic structure parameter – the mean distance of the graphite particles O Fig. 1 – was
submitted for the definition of the crack initiation point.
Starting from a generally materials characterization including determination of the most
important structural parameters the first results of the instrumented pre-cracked Charpy impact
tests (multiple-specimen technique applied as low-blow-procedure) at room temperature and at 40
°C will be presented. The estimated dynamical crack-resistance curves (dynamic R-curves) and the
dynamical crack-initiation-values will be discussed taking into account the different testing
standards and test procedures as well as our own proposal for the definition of the crack
initialization point (Fig. 2).
514 P. Trubitz et al.

FIGURE 1. Definition of OFIGURE 2. dynamic R-curve of DCI (schematic)

References
1. ISO 12135:2002: Metallic material – Unified method of test fort he determination of
quasistatic fracture toughness. First Edition. 2002-12-01
2. ASTM E 1820-01: Standard Test Method for Measurement of Fracture Toughnes. ASTM
International, 2001
3. ESIS P2-92: ESIS Procedure for Determination the Fracture Behaviour of Materials.
European Structural Integrity Society – ESIS, Technical Committee I: Elastic-Plastic
Fracture. Subcommittee I.4: Fracture Mechanics Testing Standards. ESIS Office, c/o
Materials Laboratory, Delft University of Technology, Delft, The Netherlands. January 1992
4. Charpy fracture test methods. Proposed standard methods for instrumented pre-cracked
Charpy impact testing of steels. Draft 19. European Structural Integrity Society [ESIS],
Technical Sub-Committee on Dynamic Testing at Intermediate Strain Rates. April 2005
5. Roos, E.: Grundlagen und notwendige Voraussetzungen zur Anwendung der
Risswiderstandskurve in der Sicherheitsanalyse angerissener Bauteile. Habilitationsschrift
Universität Stuttgart 1992. VDI-Fortschr.-Bericht Reihe 18, Nr. 122. VDI-Verlag GmbH,
Düsseldorf 1993
6. Holland, D.: Einfluß des Spannungszustandes auf die Vorgänge beim Gleitbruch von
Baustählen. Dissertation RWTH Aachen, 1992
7. Winkler, H.-P.; Trubitz, P.; Pusch, G.; Warnke, E.P.; Beute, K. and Novotny, V.: Dynamic
Fracture Toughness Data for CASTOR Casks. Vortrag auf der Tagung Patram 2004. 20. - 24.
September 2004, Berlin
 2T18. Dynamic, high strain rate, or impact fracture 515

CRACK RESISTANCE DETERMINATION FROM THE CHARPY IMPACT

TEST

R. Chaouadi
SCK-CEN
Boeretang 200, 2400 Mol, Belgium
rchaouad@sckcen.be

Many engineers and scientists investigated the possibility to correlate Charpy impact energy with
the fracture toughness. As a result, many empirical correlations can be found in literature [1-2].
However, most of these correlations had a limited application range due primarily to their
empirical basis. Recently, a simple procedure was provided to determine crack length from the
load-displacement test record [3]. The basic idea is that crack length is proportional to the absorbed
energy, namely:
2
§ Ei  E0 ·
' ai v ¨ ¸
¨E ¸
© final  E 0 ¹ (1)
where "Ei" is the absorbed energy at time "i", E0 is a threshold energy corresponding to onset of
crack extension and Efinal is the final energy corresponding to a measured crack extension 'afinal.
This procedure was validated on a large number of materials using various cracked geometries.
The main objective of this paper is to investigate the possibility to apply a similar procedure to a
V-notched geometry, namely the Charpy specimen. Such an evaluation would lead to determine
the crack resistance from the Charpy-V impact test.
Two low alloyed steels used in the reactor pressure vessel industry were selected. Tests were
performed on Charpy-V notched specimen loaded statically as well as dynamically. Precracked
Charpy specimens with the crack length-to-width ratio of 0.5 and 0.2 were also tested at both
loading rates. It is found that this procedure is applicable to the notched geometry (see Figure 1).
The experimental results obtained on the notched and cracked geometries at both loading rates can
be rationalized by taking into account the notch and loading rate effects.
516 R. Chaouadi

FIGURE 1. Crack resistance behavior from the Charpy-V notched geometry in static loading:
multiple specimen versus single specimen method.

References
1. Rolfe S.T. and Barsom, J.M., Fracture and Fatigue Control in Structure – Application of
Fracture Mechanics, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1977.
2. Schindler, H.J., In Pendulum Impact Testing: A Century of Progress, edited by T.A. Sievert
and M.P. Manahan, ASTM STP 1380, 2000, 337-353.
3. Chaouadi, R., J. Testing and Evaluation, Vol. 32, No. 6, 469-475, 2004.
 2T18. Dynamic, high strain rate, or impact fracture 517

A STOCHASTIC INTERFACE MODEL FOR THE FRACTURE OF BARS

S. Nagy and F. Kun

Department of Theoretical Physics, University of Debrecen, Debrecen, Hungary
H-4032 Debrecen, Poroszlay u. 6/c, Hungary
nagys@dtp.atomki.hu

The fracture of heterogeneous materials is an important scientific and technological problem which
has attracted an intensive research during the past years. The interaction of the heterogeneous
stress field with the disordered microstructure of the material makes the fracture process difficult
to handle theoretically, hence, most of the theoretical approaches rely on computer simulations of
discrete models.
Recently, it was found experimentally that the cracking of a bar under three point bending
proceeds in bursts which are characterized by power law distributions. Motivated by these
experimental findings, in this work we study the damage process of an elastic bar under three point
bending focusing on the bursts of microscopic breaking events and on the spatial distribution of the
damage. We model the bars as two rigid blocks which are glued together by an elastic interface. It
is assumed that the interface region can deform while the two rigid blocks remain intact under the
deflection of the specimen. The rigidity implies that the deformation of the specimen can be
characterized simply by the deflection of the middle of the bar d. The interface region is discretized
by elastic fibers of number N which are placed equidistantly between the two blocks. If the local
deformation ei of a fiber exceeds a threshold value eic it breaks and a microcrack nucleates in the
interface. The disordered properties of the material are represented by the randomness of the
breaking thresholds eic which are independent identically distributed random variables. During the
deformation process those fibers which exceed their threshold value break, i.e. they are removed
from the interface.
The amount of disorder of the failure thresholds eic has a substantial effect on the macroscopic
response of the specimen. In case of zero disorder the failure of the interface is completely brittle,
i.e. fiber breaking starts at the bottom and continues upwards as d is increased and the consitutive
curve is sharply peaked. As the strength of the disorder is increased the constitutive curve gets
more and more rounded and develops into a quadratic maximum. On the microlevel it means that
that the neighbouring fibers do not simply break one after the other.
To characterize the damage process of the loaded bar under stress controlled conditions we
determined the distribution of the burst sizes s of fiber breaks for the disorder distribution varying
the width of the distribution d. Simulations revealed that the avalanche size distribution D(s) shows
a power law behaviour and the exponent t is universal, i.e. it is insensitive to the strength of the
disorder, see FIG. 1. It turned out that t coincides with the exponent of the classical parallel bundle
of fibers with equal load sharing t»2.5. Furthermore, we showed that ahead the crack tip a fracture
process zone is formed where the local failure mechanism is percolation in a gradient field. The
process zone proved to shrink with increasing deformation making the crack tip sharper as the
crack advances The largest burst has a power law dependence on the strength of the disorder.
518 S. Nagy and F. Kun

FIG. 1

Based on our simple model we also studied the energetics of the impact process during three
point bending induced dynamically by impact. Depending on the energy E of the hit on the
specimen the impact process can be classified into two states. Calculations revealed that there
exists a critical value of the impact energy below which partial failure of the interface occurs
(damaged phase), while above it the crack passes through the entire specimen resulting in global
failure (fractured phase). The critical energy separates the two phases of the system and a
continous phase transition occurs between them. Identifying the number of intact fibers Nint by the
order parameter it was analytically proven that Nint has a power law dependence on the distance
from Ec with exponent b=1/2 in case of zero disorder. Numerical results showed that the exponent
grows with growing strength of disorder up to the value b=2/3.

References
1. F. Kun et. al Phys. Rev. Letters 93, 227204 (2004)
 2T18. Dynamic, high strain rate, or impact fracture 519

THE ANTI-PENETRATION PROPERTIES OF SPACE ARMOR

Tso-Liang Teng, Cho-Chung Liang and Cheng-Chung Lu

Department of Mechanical and Automation Engineering, Da-Yeh University
112. Shan-Jeau Rd., Dah-Tsuen, Chang-Hua, Taiwan, R. O. C.
tlteng@mail.dyu.edu.tw, ccliang@mail.edu.tw

New types of armor, including space armor, multiple-layered armor, composite armor and modular
armor have been successfully developed and installed on the armored vehicles of several nations.
The protective capability of armor against penetration is established. Of developed composite
armor, space armor has a simple structure and is easy to produce and can be produced at low cost.
This study uses the finite element package DYTRAN and the pre and post processor PNTRAN to
elucidate the ballistic resistance and penetration of space armor. Factors such as armor thickness,
space of armor and projectile profile are considered. A tool for simulating the protection afforded
by armor and supporting the design of space armor is developed.
Figure 1 presents the deformation plot of the steel ball as it penetrates the space armor. The
total thickness of the space armor is 6.35 mm. Each layer is assumed to be 3.175 mm in thick and
62.8 mm in long and wide. The armor incline at 60 degree to the vertical, and the space between
the two layers is 10 mm.

(t=0sec) (t=24sec) (t=64sec)

(t=96sec) (t=144sec) (t=240sec)

Figure 1 the deformation plot of the steel ball and plate( t1 = t2 = 3.175 mm, d=10 mm )

The first layer is completely penetrated and the steel ball is deformed. The steel ball rebounded
from the second layer. The surface of the second layer has only a bowl-shaped contour, indicating
that the steel ball was stopped. The figure thus shows the anti-penetration capacity of space armor.
The improved capability of space armor is inferred to be related to the following causes.(1) The
penetration of the first layer will deteriorate the projectile and thus reduces the energy with which
it collides with the second layer. (2) The fall in speed of the projectile upon penetration into the
first layer also reduces the energy of collision with the second layer.(3) The deviation of the
520 T.-L. Teng et al.

direction of projectile from its original direction also increases the angle of incidence of the
projectile, causing it to be rebounded after it has penetrated through the first layer.
Figure 2 plots the residual velocity of the steel ball after it has penetrated the space armor. The
plot implies that the ballistic resistance increase with the gap width. Figure 3 plots the relationships
between the residual velocity of projectile and the ratio of thicknesses of the two layers (t1/t2). The
magnitude of the residual velocity is proportional to the ratio of thicknesses. Figure 4 plots the
change in residual velocity versus the L/D ratio. In only the L/D=1 case, the residual velocity is
less than that of a steel ball projectile of the same mass, given the same gap width. In the other
cases, the residual velocity exceeded that of the ball projectile, increasing with the L/D ratio. The
space armor has a poor ballistic resistance to slender projectiles.

Fig.2 the residual velocity of the steel ball with various distance of space armor

Fig.3 the relationships between the residual velocity of projectileand the ratio of thicknesses

Figure 4 the residual velocity of a cylindrical projectile versus the L/D ratio
 2T18. Dynamic, high strain rate, or impact fracture 521

KEY CURVE METHODS FOR DYNAMIC FRACTURE MECHANICS OF CAST

IRON

W. Baer
Federal Institute for Materials Research and Testing (BAM)
Unter den Eichen 87, D-12200 Berlin, Germany
wolfram.baer@bam.de

Today, ductile cast iron materials (DCI) are used for safety relevant components such as for
instance transport and storage casks for radioactive materials. The corresponding methods and
codes for safety assessment take into account fracture mechanics concepts more and more.
Nevertheless, some necessary prerequisites for the practical application of these codes like the
quantification and availability of both the loading as well as the materials resistance in terms of
fracture mechanics quantities have not been fulfilled sufficiently up to date. The focus of the
present paper is laid on the material side only.
The experimental fracture mechanical material characterization of DCI at quasi-static loading
conditions is practised by means of crack resistance curves. These can be determined with respect
to accepted test standards using efficient single specimen methods for the measurement of crack
growth.
Nevertheless, it is of special importance to assess the safety at accidental conditions too where
fast changes of the stress and strain state in the material and the component prevail due to dynamic
(high rate) loading. At present, there are only multiple specimen techniques available for the
determination of crack resistance curves of DCI at these loading conditions, like the low blow
method. However, significant disadvantages of these methods are their high demand of material
and time what makes them expensive. Furthermore, they integrate over a larger area of the material
why it is not possible to identify and map local gradients of microstructure and properties.
That is the reason why the further development of the fracture mechanics test methods for DCI
materials at dynamic loading is absolutely necessary.
Therefore, BAM has started a research project recently that is financed by the German Science
Foundation and whose goal is to solve the above mentioned task by development and qualification
of an appropriate single specimen method for the determination of dynamic crack resistance curves
and characteristics of DCI.
From the experimental as well as the economical point of view major advantages of this
approach are a low demand of material and high efficiency of time. Furthermore, there is also the
possibility to integrate a single specimen method into the manufacturing quality control system
where only small amounts of material can be taken out of the component without destroying it.
The well-known efficient single specimen methods for crack growth measurement at quasi-
static loading, like the compliance method or the DC potential drop technique, can not be used in
dynamic testing successfully. Whereas the key curve method basically offers the possibility for
crack growth measurement even at high loading rates, higher temperatures or corrosive
environment.
Therefore, it is first of all reported in the present paper on the review and systematization of
published key curve methods that have nearly exclusively been applied to steels in the last 25 years
with different success. Based on that, an assessment of their applicability to dynamic testing of
DCI is made. It is basically distinguished between the experimental, analytical and numerical
determination of key curves. An analytical approach and a numerical method were then chosen as
522 W. Baer

advanced key curve methods promising success also with dynamic DCI testing. Further research
work will now be done to investigate these methods in comparison and with respect to
qualification of one method and application to DCI.
The final goal of the project will be to provide an efficient single specimen method which is as
simple as possible and allows determining dynamic crack resistance curves of DCI as well as
deduced material characteristics for component safety analysis reproducibly and reliably.
 2T18. Dynamic, high strain rate, or impact fracture 523

DYNAMIC TENSILE BEHAVIOR OF ARAMID FRP USING SPLIT

HOPKINSON BAR METHOD

Yutaka Sawaki, Jun Watanabe1, Eitoku Nakanishi2 and Kiyoshi Isogimi3

Department of Mechanical Engineering, Faculty of Engineering, Mie University
Kurima-machiyacho 1577, Tsu 514-8507, Japan
sawaki@mach.mie-u.ac.jp
1Department of Mechanical Engineering, Graduate School of Engineering, Mie University

Kurima-machiyacho 1577, Tsu 514-8507, Japan

Jun-w@sks.mach.mie-u.ac.jp
2Department of Mechanical Engineering, Faculty of Engineering, Mie University

Kurima-machiyacho 1577, Tsu 514-8507, Japan

nakanisi@mach.mie-u.ac.jp
3Department of Mechanical Engineering, Faculty of Engineering, Mie University

Kurima-machiyacho 1577, Tsu 514-8507, Japan

isogimi@mach.mie-u.ac.jp

It is well known that fiber reinforced composite materials are new industrial materials replaced
with metal, wood or concrete, etc. and are equipped with high intensity and high elasticity. An
aramid fiber has relatively high impact intensity compared with carbon fiber or glass fiber. When
an aramid fiber was made into a composite material (Aramid Fiber Reinforced Plastic: AFRP), it
has the feature which shows the metallic ductility which was similar to aluminium. From the
above-mentioned reason, AFRP attracts attention as reinforced composite polymer material.
Several recent works on various FRP(s) subject to tensile impact loading are performed [1], [2],
[3].
In this paper, dynamic tensile behaviour of Aramid Fiber Reinforced Plastic (AFRP) at
relatively high strain rates of approximately 70 to 230 s-1 is investigated by using one bar method,
which is a variation of sprit Hopkinson bar method. In this experiment, the specimen is created
using unsaturated polyester resin as a matrix material and contains an Aramid fiber bundle (Kevlar
49: Du Pont-Toray Co., LTD.) in an axial direction in order to obtain fundamental properties of
fiber reinforcement as shown in Fig. 1.

FIGURE 1. Dimensions of specimen.

Experimental results on the stress-strain relationships of AFRP at several strain rate are
compared with those of GFRP and matrix material (Polyester). Static tensile test is also carried out
for comparison of dynamic test results. An example of the stress-strain curve obtained by dynamic
524 Y. Sawaki et al.

tensile test at strain-rate 230 s-1 is shown in Fig. 2. From this figure, it turns out that the value of
maximum stress of the present AFRP specimen is about 9 times compared with one of only
unsaturated polyester specimen and the value of maximum strain is also about 2.5times. It is
thought that there is reinforcement of the matrix material by the aramid fiber.

FIGURE 2. Stress-strain curve at strain rate of 230 s-1.

As the result, it is recognized that the value of maximum stress and strain of AFRP increased
with increasing strain rate.

References
1. Gilat, A., Goldberg R.K. and Roberts, G.D., Composites Sci. Tech., vol. 62, 1469-1476,
2002.
2. Rio, T.G., Barbero E., Zaera R. and Novarro, C., Composites Sci. Tech., vol. 65, 61-71,
2005.
3. Sun, B., Liu F. and Gu, B., Composites :Part A , in press, 2005.
 2T19. Damage mechanics 525

DETECTION OF LOW-VELOCITY IMPACT DAMAGE IN CARBON-EPOXY

PLATES USING NDT

A. M. Amaro, M. F. M. S. de Moura1 and P. N. B. Reis2

CEMUC – Centro de Engenharia Mecânica da
Universidade de Coimbra, Polo II, 3030 Coimbra, Portugal
1Faculdade de Engenharia da Universidade do Porto,

Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

2Departamento de Engenharia Electromecânica, Universidade

da Beira Interior - UBI, 6200 Covilhã, Portugal

ana.amaro@dem.uc.pt, mfmoura@fe.up.pt, preis@ubi.pt

Composite materials are excellent for aeronautical applications due to their high specific strength
and stiffness. However, these materials may absorb limited amounts of energy through localized
damage mechanisms without extensive plastic deformation. Different types of damage may be
encountered in the impacted region, including matrix cracking, delaminations and broken fibbers.
The presence of this damage is detrimental to the material performance and structural integrity of
the composite structure. For example, impact damage is considered as the primary cause of in
service delamination in composites giving rise to reductions of the compressive residual strength
up to 60%, de Moura and Marques [1]. Therefore considerable research have been developed on
impact response of composite materials, particularly to study the relationships between the
composite constituent properties, the damage mechanisms and the degradation of mechanical
properties due to impact, Cantwell and Morton [2].
A major problem with composite materials is the detection of low-velocity impact damage. In
addition to the conventional destructive, nondestructive evaluation techniques have been receiving
much attention for detection and monitoring of damage evolution within the composite structure.
In order to realize the full potential of fibber reinforced composites in critical load bearing
applications, reliable and cost effective non-destructive test (NDT) methods must be developed,
Adams and Cawley [3]. However, many of the NDT techniques used for metallic structures are
unsuitable for composite structures.
The objective of this work is to evaluate features and capabilities of four different methods
used to detect and quantify impact damages on carbon-fibber-reinforced epoxy composite.
Composite laminates were prepared from high strength unidirectional carbon pre-preg
“TEXIPREG HS 160 REM” and processed in agreement with recommendations of the
manufacturer. Two stacking sequences were considered for carbon fibber laminates: [0,90,0,90]2s
and [904,04]S. The overall dimensions of the plates were 300x300x2.4 mm. Quality control of the
plates was performed by C-Scan, to evaluate the eventual presence of defects resulting from
manufacturing process. The impact energy used in the tests, was 3 J, which correspond to a load
around of 3000 N. Finally, the damages in the laminates were analysed by Electronic Speckle
Pattern Interferometry (ESPI), Shearography, Ultrasonic Testing and X – rays radiography.
It was concluded that the four techniques studied are able to detect almost of the defects.
However, the interferometric methods showed some limitations. For example, the ESPI can
require an experient operator capable of recognizing and interpreting the presence of defects inside
all the fringe patterns provide by the method. The application of this technique can be more
difficult when small damages are produced. On the other hand Shearography has an easier
interpretation. However, there are also difficulties on the visualization and quantification of defects
526 A. M. Amaro et al.

since damage evaluation is based on anomalies in the displacement gradient fields. X – rays
radiography method is a good alternative technique, but it does not allow to define the position of
defects along thickness. The X – rays radiography method was compared with the Deply
Technique (destructive technique), which confirmed that superposition of different delaminations
correspond to x-rays image. The Ultrasonic methods, A-Scan and C-Scan, showed to be the best
solution to inspect the laminates. According to the experimental results, these techniques are able
to detect and measure the defect extension with great precision. Another important conclusion is
that these techniques are complementary in the detection of position of defects and quantification
of their variation along thickness.

References
1. de Moura, M. F. S. F. and Marques, A. T., Composites: Part A, vol. 33, 361-368, 2002.
2. Cantwell, W. J. and Morton, J., Composites, vol. 22, 347-362, 1991.
3. Adams, R.D. and Cawley, P.D., NDT International, vol. 21, 208-222, 1998.
 2T19. Damage mechanics 527

DAMAGE ACCUMULATION AT HIGH TEMPERATURE CREEP OF A

SINGLE-CRYSTAL SUPERALLOY

A. Staroselsky and B. Cassenti

Pratt and Whitney
400 Main Street MS 165-16
East Hartford, CT 06108, USA
Alexander.Staroselsky@pw.utc.com
Brice.Cassenti@pw.utc.com

The micromechanics of the high temperature creep and damage acumulation in single crystal
nickel base superalloy is the subject of this study. These alloys are used in turbine blade and vane
applications in advanced commercial and military gas turbines and in turbopumps for the space
shuttle main engines. The objective of this program was to develop a robust predictive tool to
relate single crystal structure macroscopic behavior and fracture crack initiation to
micromechanical events. We have developed the crystallographic–based model for non-isothermal
high temperature cyclic deformation and coupled it with damage kinetics.
The approach used in simulations of deformation process of L12 single crystal alloys gives us
a tool to obtain the deformation behavior of the structure. Although the quantitative description of
plastic flow by crystallographic slip may be traced back to early works of Taylor, Elam, and Sachs
(1923, 1925, 1928), there have been considerable recent advances in the understanding of
anisotropic elasto-viscoplastic deformation behavior, damage kinetics and thermal mechanical
fatigue of single-crystal materials, allowing improvements the prediction of the deformation and
failure behavior of the high temperature turbine components.
We have developed a unified materials constitutive model including thermally-dependent
creep activation mechanisms for different crystallographic orientations. The model extends
existing approaches [1, 2] to increase the accuracy of material deformation response predictions on
cyclic and thermal-cyclic loading.
Historically, only secondary creep effects were considered (e.g., Larson-Miller, etc.) in
engineering calculations. However, during the thermal-mechanical loading of high temperature
single crystal turbine parts, all three creep stages: primary, secondary and tertiary, manifest
themselves and none of them can be neglected. Account must be taken of all creep mechanisms,
and is especially important in the case of non-homogeneous thermal loading of components with
extensive stress redistribution and relaxation. We have experimentally observed and simulated,
with our unified microstructure-based model, loading cases with only primary and tertiary creep
regimes as well as creep without any noticeable primary creep.
Having developed the constitutive model and implemented it in the comercial finite element
software, ANSYS, as a material user routine, we predicted yield/creep anisotropy and yield –
thermal dependence. The general form of the governing relation for the plastic shearing rate may
be written as follows:
D D D D D
s (W ,W ,W , T ), Z
pe cb se
, d D ,T ) , octahedral slip systems
D D
, s (T ), Z D , d D , T ) , cube slip systems
(1)
528 A. Staroselsky and B. Cassenti

D D
where is the deformation resistance and Z is a kinematic hardening of D-th slip system; W is
D D D D
resolved shear stress, and W pe ,W cb ,W se - “non-Schmidt” shear stresses, and d is the amount of
damage assosiated with plastic shear. These constitutive equations are fully coupled with evolution
equations for the state variables listed above. The process for damage nucleation/accommodation
is a Poissonian stochastic process, with the evolution (probability rate) described by a Boltzmann
formula. Damage accomodation causes tertiary creep and shear localization around local
concentrators. Shear bands are one of the major causes of single crystal cracking and the model
prediction of the initial shear localization was used as a criteria for crack initiation.
Small deformation analysis usually does not provide enough information to distinguish and
calibrate different mechanisms of inelastic deformation and damage accumulation. We extended
the analysis to finite deformations (Asaro [3]), and coupled it with crystallographic texture analysis
(Bronkhorst et al [4]). The change of the local crystal lattice is strongly dependent on each slip
system activity and hardening mechanisms. We calibrated the model strain-stress and
crystallographic texture predictions with test data for small and moderate deformations. This
allows us to numerically predict the position and orientation of shear localization.
The developed non-isothermal, crystal –viscoplastic, damage mechanics model is used for
cyclic ratcheting and thermomechanical fatigue (TMF) analysis. The model results will be used for
developing reduced order models efficiently reflecting long term thermal cyclic behavior and crack
nucleation in high temperature single crystal components of gas turbines.

References
1. Nissley, D., Meyer, T., and Walker, K. Life Predictions and Constitutive Models for Engine
Hot Section Anisotropic Materials, Pratt & Whitney, Report NAS3-23939, 1991
2. Hesebeck, O., Int. J. Damage Mech., vol. 10, 325-346, 2001
3. Asaro, R.J. Advanced Applied Mechanics, vol. 23, 1983.
4. Bronkhorst, C.A., Kalidindi, S.R., and Anand, L., Phil. Trans. R. Soc. Lond. A, vol. 341, 443-
477, 1992
 2T19. Damage mechanics 529

ASYMPTOTIC HOMOGENISATION FOR HETEROGENEOUS MEDIA WITH

EVOLVING MICROCRACKS

E. K. Agiasofitou, C. Dascalu and J. L. Auriault

University Joseph Fourier, Laboratoire Sols Solides Structures
Domaine Universitaire B.P. 53, 38041 Grenoble Cedex 9 FRANCE
eleni.agiasofitou@hmg.inpg.fr, Cristian.Dascalu@hmg.inpg.fr,
Jean-Louis.Auriault@hmg.inpg.fr

In this work, an asymptotic expansion homogenization is used to study the overall behaviour of a
damaged elastic body with a locally periodic distribution of growing microcracks. The
microstructure evolution is represented, at the macroscopic level, by a local internal variable
related to the microcracks lengths. An evolution damage law is deduced, through asymptotic
homogenisation, by assuming a microscopic fracture criterion of Griffith type. Finite element
solutions, in elasticity, are presented in order to illustrate this new approach.
Many articles have been devoted to the overall behaviour of micro-fractured solids (see for
instance Nemat-Nasser and Hori [1] for a review). Almost all these works are confined to the case
of stationary cracks. Exceptions are Prat and Bazant [2] and Caiazzo and Constanzo [3], which
take into account the fracture evolution. Their approach of scale change is not the asymptotic
homogenization. On the other hand, asymptotic homogenization techniques have been used for
stationary microcracks in Leguillon and Sanchez-Palencia [4] and Telega [5]. Our aim is to use this
rigorous link between microstructure and the overall behaviour in the case of propagating
microcracks.

The elastic medium is locally periodic with the period : chosen so as to contain a set of
microcracks * of effective length l l(x , t). In the solid part : S the equation of motion is
wV S ij w 2 u Si
U , V S ij a ijk h e x k h ( u S ) in : S ,
wx j wt2
ª¬ V S ij
º¼ N j 0 on *,
(1)

where ı S and u S are the stress and the displacement fields, U is the mass density and a is the
elastic tensor, which verifies the classical symmetries and the ellipticity condition. We assume that
the crack faces are traction free.
Following the method of asymptotic homogenisation (e.g. Bensousssan et al [6], Bakhvalov et
al [7], Auriault [8]), we assume that the period size of the elastic medium is characterized by a
small parameter H and we look for u S in the form

uS u ( 0 ) ( x , y , t )  H u (1) ( x , y , t )  H 2 u ( 2 ) ( x , y , t )  
(2)
(i )
where the u are y-periodic functions with y x H , y is the microscopic (fast) variable and x
is the macroscopic (slow) variable. The macroscopic equation of motion, at the first order of
approximation, is obtained as
530 E. K. Agiasofitou et al.

w w 2 u i( 0 )

wx j
 C ijkh e xkh ( u ( 0 ) )   U
wt 2
0,
(3)

kh kh
where C ( l ) ¢ a ijk h  a ijlm e ylm ( ȟ ( l ))² is the effective elastic tensor and ȟ are the
characteristic functions [6], [7] which depend on the crack evolution.
A propagation criterion of Griffith type (see Freund [9]) is assumed for microcracks. The
equation of energy written in a cell concludes in the following result after applying the asymptotic
expansion

w § : ·
¨ C ijk h ( l ) e x kh ( u ( 0 ) ) e x ij ( u ( 0 ) ) ¸  2 G c r 0
wt © 2 ¹ (4)

where Gcr is the critical value of the energy release rate and it is a material constant and : is
the measure of : .
Combining the macroscopic equation of motion (3) and the equation (4), we obtain an equation
for the damage variable l l ( x , t ) in the form

1 w C i jkh  w C ijkh w l
le x k h ( u ( 0 ) ) e x ij ( u ( 0 ) )  e x kh ( u ( 0 ) ) u i( 0 ) 
2 wl wl wx j
w
 C ijk h
wx j
 e xk h ( u ( 0 ) ) u i( 0 )   2 G: c r  wwt §¨© 12 U ( u i( 0 ) ) 2 ·¸¹ 0.
(5)
Finite element solutions for the coupled problem (3) and (5) are obtained for particular, two-
dimensional, geometries of the micro-fractured cells. The influence of the cracks evolution on the
homogenized mechanical response is analyzed through the obtained numerical solutions.

References
1. Nemat_Nasser, S. and Hori M., Micromechanics: overall properties of heterogeneous
materials, Elsevier, Amsterdam-Lausann-New York, 1999.
2. Prat, P.C. and Bazant, Z.P., J. Mech. Phys. Solids, vol. 45, 611-636, 1997.
3. Caiazzo, A.A . and Constanzo, F., Int. J. Solids Struct., vol. 37, 3375-3398, 2000.
4. Leguillon, D. and Sanchez-Palencia E., J. Méc. Théor. et Appl., vol. 1, 195-209, 1982.
5. Telega, J.L., Comput. Mech., vol. 6, 109-127, 1990.
6. Bensounssan, A., Lions, J.L. and Papanicolaou G., Asymptotic analysis for periodic
Structures, North_Holland, Amsterdam, 1978.
7. Bakhvalov, N. and Panasenko, G. Homogenisation: averaging processes in periodic media,
Kluwer Academic Publishers Group, Dordrecht, 1989.
8. Auriault, J-L., Int. J. Engng. Sci., vol. 29, 785-795, 1991.
9. Freund, L.B., Dynamic Fracture Mechanics, Cambridge University Press, 1998.
 2T19. Damage mechanics 531

ON THE ANALYSIS OF DAMAGE LOCALIZATION AS PRECURSOR OF

MACRO-CRACKS

H. Stumpf and K. Hackl

Lehrstuhl für Allgemeine Mechanik, Ruhr-Universität Bochum
D-44780 Bochum, Germany
stumpf@ruhr-uni-bochum.de

In damage mechanics it is well-known that localization zones are often precursors of macro-
cracks. Correspondingly, in fracture mechanics extended experimental investigations (e.g. Klemm
and Kalthoff, 1984; Bertram and Kalthoff, 2003) have shown that at the tip of a crack there is no
stress singularity, instead, in front of the crack a small process zone can be observed caused by
intensive dislocation motion in the case of ductile materials as metals or by intensive micro-
cracking in the case of brittle materials as limestone or granite. At the same time high temperatures
develop in the process zone. The size of the process zone depends strongly on the material
properties. For glass the process zone is very small, but a very high temperature can be measured
there. Another essential feature is the fact that the process zone evolves with high speed, if the
crack moves with high speed.
In this paper we present a framework for the analysis of dissipative processes with narrow
zones of plastification and/or intensive micro-cracking and heat production taking into account
also the inertia effects of moving micro-defects (for nonlocal brittle damage see Stumpf and Hackl,
2003 and for ductile damage Stumpf et al. 2004).
To describe fields of dislocations and micro-cracks we choose as independent kinematical
variables of physical and material space F, Fd , Fp and their gradients,

e ^F , F d , F p , ’ F d , ’ F p ` , (1)

where F is the physical deformation gradient, Fd a damage tensor and Fp the plastic
deformation tensor.
Power-conjugate to (1) is a set of physical and material forces and stresses,

s ^T , T d , T p , H d , H p ` , (2)
where T is the physical first Piola-Kirchhoff stress tensor and the others are material stress tensors
representing forces and stresses acting on defects. As additional variables we include temperature
ș and ’ș .
Formulating the classical dynamical balance laws of physical forces and nonlocal dynamical
balance laws of material forces, furthermore first and second law of thermodynamics, we have as
many equations as unknowns.

From the second law of thermodynamics it follows that the free energy ȥ and the stresses (2)
are of the form

ȥ ȥˆ ( e , ș ) , s w e ȥˆ ( e , ș )  sˆ * ( e , ș , e , ’ ș ) , (2)
532 H. Stumpf and K. Hackl

where the dissipative driving forces sˆ * ( e , ș , e , ’ ș ) are given in terms of a dissipation potential
ij ijˆ ( e , ș , e , ș ) by

sˆ * ( e , ș, e , ’ ș) w e ijˆ ( e , ș, e , ’ ș ) . (3)
The dissipative driving forces determine also the evolution of the macro-crack.

References
1. Stumpf, H. and Hackl, K., Int. J. Solids Structures, vol. 40, 1567-1584, 2003.
2. Stumpf, H., Makowski, J., Gorski, J. and Hackl, K., Mech. Res. Comm., vol. 31, 355-363,
2004.
3. Klemm, W. and Kalthoff, J.F., Entwicklung eines thermomechanischen Messverfahrens zur
Charakterisierung des Werkstoffverhaltens bei schneller duktiler Rissausbreitung in
Rohrfernleitungsstählen, DVM, Berlin, 1984.
4. Bertram, A. and Kalthoff, J.F., Crack propagation toughness of rock for the range from low to
very high crack speeds, Key Engineering materials, vols 251-252, 23-430.
 2T19. Damage mechanics 533

FATIGUE ASSESSMENT BASED ON STATISTICAL ANALYSIS OF

THEORETICAL PARAMETERS

J. Cacko
Slovak Academy of Sciences
Institute of Materials and Machine Mechanics, Raianska 75, SK-831 02 Bratislava
ummscac@savba.sk

The basic problem in material research is to predict lifetime behaviour of structures [1]. That is not
only a material problem, but also loading conditions, technology design and environmental
properties are very significant [2]. A service loading has usually a random character. Traditional
experimental investigation methods are very expensive and time-consuming in this case.
Therefore, a computer-simulation method seems to be a very suitable way to this purpose. In order
to evaluate fatigue damage under dynamic loading, the cumulative representation of a damage
increase over a closed cycle is accepted [3]. Under a harmonic loading with constant both
amplitudes and mean values of cycles, stress-life (V-N) and/or strain-life (H-N) curves can be
experimentally obtained. Then it is possible to specify a relative damage over a closed loading
cycle according to some hypothesis (e.g. Miner, Corten-Dolan, etc.) [4]. The difficulties arise
under varying mean value of cycles because the relative damage accumulated over individual
cycles cannot be simply added. The complete loading history must be respected, and it is a big
problem especially under random loading.
The serious difficulty is that we must know the complete time loading history until the actual
loading cycle. Therefore, a new approach to suitably interpret the loading history has been created.
Knowledge of a fatigue damaging process is a necessary condition for an optimum structure design
and for effective production with respect to material and energy saving. Especially, CAD, CAM
and CAE technologies are very successful because they are by far quicker and less expensive in
comparison with traditional theoretical and experimental methods. Moreover, an analytical
solution is usually not possible regarding to a stochastic nature of a service loading. For the
optimum structure design, it is very important to estimate an elapsed time and/or number of cycles
to failure. Because there is no deterministic principle to describe a cumulative damage mechanics
in a material unambiguously, especially under stochastic loading, we must use some of cumulative
hypotheses and failure criteria. The hypotheses should follow a physical nature of the cumulative
mechanics in a material but they are usually originated according to experimental experience and
practice. Nevertheless, non-homogeneity of material properties makes some risk of estimation [5].
Therefore, the calculation must be completed by guaranteed reliability of estimated characteristics.

The number of closed hysteresis cycles to fracture Nf can be estimated as

R 1
ª VC
m
º
« »
N f N 0 «k ³[
*q
 
f [* d[ *» ,
« 0 »
¬ ¼ (1)
where
Rm – ultimate strength
534 J. Cacko

VC – endurance limit
N0 – corresponding number of cycles to fracture for VC

q
q – slope exponent of Wöhler´s curve ( V a N f const )
k – parameter (for Miner´s rule k=1)

V a*
[* *
V C , where V a is equivalent cycle amplitude (equivalent cycle = cycle with zero
mean value).
Because the parameters in (1), especially VC and q cannot be considered as deterministic
values, they must be determined using some probabilistic method, e.g. Monte Carlo one. Then we
can use a Weibull´s model

g N 
m
m 1 d N
f md N f e f
; N f t 0, (2)
where m and d are parameters of the model. According to this model, the most reliable number of
cycles to failure can be assessed.
In the paper, a theoretical basis of the procedure is described, as well as a practical experience
is presented, too.

References
1. Cacko, J. Structural Safety, vol. 12, No 2, 151-158, 1993.
2. Cacko, J. Int. J. Fatigue, 14 No 3, 183-188, 1992.
3. Cacko, J. Engineering Against Fatigue, edited by J.H. Beynon, Balkema, Rotterdam, 1999,
357-364.
4. Cacko, J. Fracture From Defects, edited by K.J. Miller, EMAS, Sheffield, 1998, 235-240.
5. Cacko, J. ECF 13, edited by M. Fuentes, Elsevier, Amsterdam, 2000, CD-R.
 2T19. Damage mechanics 535

DETERMINATION OF DUCTILE DAMAGE PARAMETERS BY LOCAL

DEFORMATION FIELDS

M. Kuna and M. Springmann

Technische Universität Bergakademie Freiberg, Institute of Mechanics and Fluid Dynamics,
Lampadiusstr. 4, 09596 Freiberg, Germany
Meinhard.Kuna@imfd.tu-freiberg.de

Modern continuum damage models have been successfully developed to describe the non-linear
deformation process, damage and fracture of ductile metals. Most prominent models are due to
Gurson-Tvergaard-Needleman (GTN) [1] and Rousselier (R) [2]. However, the determination of
the unknown material parameters involved in such models is still a challenging issue. Usually,
smooth or notched tensile specimens are used to determine the damage parameters from global test
data as forces and displacements, whereby numerical simulations are somehow fitted to the
experimental point of specimen failure, see the round robin [3]. In fact, the large number of
unknown material parameters and the intrinsically coupling between matrix hardening and damage
require more comprehensive information from the experiment, which can be gained from
measured inhomogeneous deformation fields.
The present work comprises the development and application of methods for the parameter
identification of damage laws by means of locally measured displacement fields and measured
force-displacement curves. Both damage models (GTN) and (R) have been implemented into an
own finite element system SPC-PMHP [5, 6]. The classical Rousselier model was complemented
by accelerated void growth and void nucleation according to [4]. Thus, both models cover the same
features: power law hardening of matrix material, strain controlled void nucleation, void growth

and void coalescence (failure). Each model contains 12 material parameters p ^ pi `

(i=1,2,…,12), including 3 for hardening and 9 for damage [5, 6].
Experiments are performed at notched flat bar tensile specimens made of steel StE690, see Fig.
1. The spatial displacement fields are recorded by means of optical field measuring technique in
the region of the notch by the object grating method [8], in order to observe directly the local
damage development. The corresponding finite element mesh is shown in Fig. 1. The deviation
between measured displacements values u k and the numerical results u k (p ) from a finite
element analysis is quantified by the objective function ) ( p ) , summed up over all measuring
points nM and load steps nL. The material parameters p must be iteratively improved until the
objective function achieves the global minimum. For this purpose, a gradient based method is
used, whereby the gradients with respect to p are computed by a sensitivity analysis.
nL nM 3 2
ª º
) (p ) 1
2 ¦ ¦ ¦ «  u k ( p ) i j -  u k i j »¼
i 1 j 1 k 1 ¬
o m in
(1)

A successful strategy to identify the material parameters was found by careful numerical
studies [7]. The efficiency of the proposed solution method for the inverse problem was
demonstrated by various applications for different notch geometries. The obtained parameters are
in agreement with data from the literature.
536 M. Kuna and M. Springmann

FIGURE 1. Notched tensile specimen and FEM-discretization of one eighths.

References
1. Tvergaard, V., Needleman, A., Acta Metallurgica vol. 32, 157–169, 1984.
2. Rousselier, G., Nuclear Engineering and Design vol. 105, 97-111, 1987.
3. Bernauer, G., Brocks, W., Numerical round robin on micromechanical models, Phase II, Task
B1, Technical report, ESIS European Structural Integrity Society, 1999.
4. Chu, C.C., Needleman, A., J. Engineering Materials in Technology, vol. 102, 249-256, 1980.
5. Springmann, M., Kuna, M., Computational Materials Science vol. 26, 202-209, 2003
6. Springmann, M., Kuna, M., Computational Materials Science, vol. 32, 544-552, 2005
7. Springmann, M., Identifikation von Materialparametern schädigungsmechanischer Gesetze
unter Einbeziehung der Dehnungslokalisierung. Dissertation, TU Freiberg, 2005.
8. Benutzerhandbuch ARAMIS: Verformungsmessung nach dem Rasterverfahren. Gesellschaft
für optische Messtechnik mbH, Braunschweig, 2000.
 2T19. Damage mechanics 537

FRACTURE OF CONCRETE DUE TO CORROSION

N. Thanh, A. Millard1, Y. Berthaud, S. Care and V. L’Hostis2

LMT Cachan ; 61 Av Pt Wilson F-94235 Cachan ; thanh@lmt.ens-cachan.fr
1CEA/DEN CEN Saclay 2CEA/DEN/DPC/SCCME/LECBA, 91191 Gif sur Yvette Cedex, France

Cracking of concrete due to corrosion is a very common pathology that can be observed on various
structures. Corrosion is initiated by the penetration of chlorides or by the reactive transport of
carbon dioxide in most cases. The consequences of corrosions are (i) the reduction of the resistive
section of reinforcements (ii) the creation of expansive products (commonly denoted rust) (iii) the
fragilization of steel and finally (iv) the cracking of concrete [1-4].
Three stages are distinguished: (1) the incubation that corresponds to diffusion processes of
chemical species (chloride or carbon dioxide) inside concrete from the free surface to the steel
concrete interface; (2) the initiation stage starts when the passivation film of rebars is broken ; (3)
the cracking stage corresponds to the expansive growth of corrosion products leading to the lack of
carrying capacity of the structure. We have decided to study this stage in Laboratory. The first
stage takes decades for real cases. In our study it has been accelerated by the use of accelerated
corrosion under the effect of electrical current or potential. We decide to impose the electrical
current density that allow to predict (using the Faraday’s law) the production of iron oxides. The
imposed values range between 1 and 5 A/m2. Two types of specimen have been designed for plane
strain and plane stress analyses: a 100*150*300 mm3 one and a 100*150*20 one mm3. Figure 1
shows the reinforcement either at the corner (P1) or in the middle (P2) and the typical cracking
pattern for case P2.

FIGURE 1. Specimen for accelerated corrosion on the plates (20 mm thick) and typical cracking
patter for case P2.

FIGURE 2. Evolution of the strains on the lateral face (gauge B) for two different specimens (case
P1).
538 N. Thanh et al.

The evolution of the strain given by gauges glued at the upper surface (case P1) for two
different specimens (not shown in this abstract) and on the lateral surface (case P1) show that in a
first regime no deformation is measured due to corrosion: the oxide fill up the porosity. Then
pressure rises up to the rupture of concrete which explains the diminution of the lateral strain (Fig.
2)
Figure 1 shows the typical cracking pattern obtained within few days under a low intensity
current (1A / m2). For case P2 it is interesting to notice that the first crack to appear is the one
above the steel bar (it starts from the free surface). The inclined cracks do not appear at the free
surface. Using a CCD camera it has been possible to record the overall evolution of cracking.
A FEM modeling [7] has been developed based on a classical damage model for concrete, on
the identified properties of iron oxides (very low stiffness). It gives for case P2 a typical result
(Fig. 3) for the damage field, which corresponds to the experimental data.

FIGURE 3: Damage field for case P2

References
1. Castel A., François R., Arliguie G., Mat. & Struct., vol. 33, 539-544, 2000
2. Almusallam A. A., Al-Gahtani A. S., Aziz A. R., and Rasheeduzzafart, Const. and Build.
Mat., vol. 10, 123-129, 1995
3. Andrade C., Alonso C., Molina F. J., Mat. and Struc., vol. 26, 453-464, 1993
4. Petre – Lazar I., Evaluation du comportement en service des ouvrages en béton armé soumis à
la corrosion des aciers, PhD thesis, Université de Laval, Québec, 2000
5. Rodriguez J., Ortega L. M., Casal J., Cons. and Build. Mater., 239-248, 1997
6. Samsonov G.V., The oxide handbook », IFI/PLENUP, 1973
7. A. Millard, V. L’Hostis, K. Beddiar, Y. Berthaud, S. CARE, Modelling the cracking of a
reinforced concrete structure submitted to corrosion of steels – first validation of a damage
model based on experimental tests, Proceedings of the OECD/NEA/CSNI-RILEM Workshop
on use and performance of concrete in NPP fuel cycle facilities, Madrid, Espagne (15-16 mars
2004).
 2T21. Concrete and rock 539

EXPERIMENTAL STUDY OF SPRAYED CONCRETE STRENGTH USING

MARBLE AGGREGATES

A. Sotiropoulou and Z. G. Pandermarakis

Department of Civil and Structural Engineering Technology Teachers
School of Pedagogical and Technological Education
“Irene” Train Station, Athens - Greece
dmitsop@tee.gr, Sotiropoulou@aspete.gr

Sprayed concrete provides today the combined features of high productivity, increased quality and
low cost in work which concerns repairs, preservations, and general work which requires support.
Cement usually mixed with fine aggregate (up to 8mm) is forced with pressure to walls with
appropriate and controlled addition of water into the pump at mixture stage (wet process) or just
before the exit by nozzle (dry method). Even if with enough advantages –included the increased
bonding strength– this particular process and hence whole the final product (sprayed concrete) is
infected by many uncontrolled factors [1]. Multivaried parameters as substrate morphology,
aggregate character and also of cement type and aggregates, the non-constant flow of water and
material, the unavoidable rebound of aggregates and mainly the human factor through pump and
nozzle handling leads to a varied water to cement ratio (w/c) and also of bulk properties of sprayed
concrete. In such a situation the final product will be hardly controlled and investigated [2]. So
experimental work in this specific area is internationally limited. In present study we try to
describe the influence of aggregate type, modifying the traditional composition replacing the
limestone aggregates with marble particles. Marble grains are a sub product of marble extraction
industry and efforts for satisfactory uses has also an environmental meaning besides the special
modifications of final product.
The cement was of type I55 and as aggregate was used a group with common limestone grains
and a second group by marble grains. They were prepared specific panels with dimensions
100x100x14cm into which we sprayed by dry process and suitable pump the concrete. From these
panels we extracted sprayed cement cores and shaped cylinder specimens with dimensions of
length to diameter: 10cm/5cm. From each group we manufactured also as cast specimens by
analogous dimensions in order to use them as reference index. For these specimens we checked
their strength to unidirectional compression stresses and to an indirect tension stress condition
through splitting test at 3, 7, 14, 28 and 56 days after their construction by sprayed process or
casting [3,4].Part of results they are presented at figs.1-4. The loading rate was approximated at
1kN/s. From the corresponding diagrams we could directly see that specimens with limestone
aggregates gave in general higher strengths than these with marble aggragates, mainly at as cast
specimens. This could be attributed to lower mechanical characteristics of marble which compose
from remains of its extraction process, or even to an increased portion of clay’s constituents.
Important differentiations in the morphology of aggregates though analysis of corresponding
micrographs were not observed. Besides, we identify an inter-granular fracture of aggregates and
an perfect bonding to cement paste. This leads us to say that the appearing coincidence of sprayed
and as cast specimens could attribute to special effects that they introduce to concrete and
modifying it highly eliminating other special features. Perhaps, here, the developing
inhomogeneousness and the intrinsic special appearance due to way of manufacture and the
significant losses of aggregate through rebound effect, could depress the initial differences.
540 A. Sotiropoulou and Z. G. Pandermarakis

The present work –which constitutes one of the first research programs supported financially
by government in Technological Institutions in our country- is continued at this moment trying to
improve the properties of sprayed concrete with marble aggregates modifying its composition and
strengthening it using suitable reinforcing fibers

References
1. “Shotcrete: Engineering Developments” ed. E.S. Bernard, A.A.Balkema publ., Lisse, 2001
2. “Sprayed Concrete Conference” by TEE and EBEA, Athens - Greece, Sept. 2001
3. European Standard: EN 12390-6: “Testing hardened concrete-Part6: Tensile splitting
strength of test specimens”, Oct. 2000
4. European Standard: EN 12390.03: “Testing hardened concrete-Part3: Compressive strength
of test specimens”, Dec. 2001
5. Austin S.A. and Robins P.J. “Sprayed Concrete: Properties, Design and Application”,
Whittles Publ. 1995
 2T21. Concrete and rock 541

ANALYSIS OF THE BEHAVIOUR OF INTERFACE CRACKS IN GRAVITY

DAM

B.Bachir Bouiadjra, A.Bachir Bouiadjra, M.Belhouari and B.Serier

Department of Mechanical Engineering, University of Sidi Bel Abbes
BP 89, Cité Ben M’hidi, Sidi Bel Abbes, 22000, Algeria
bachirbou@yahoo.fr

The interface between concrete dam and rock foundation is one of the most important regions
governing the strength and stability of gravity dams [1]. The fundamental understanding of the
fracture behaviour at the Rock-Concrete interface requires evaluation of fracture criteria such as
stress intensity factor, energy release rate etc. Fracture mechanics concepts and theories have been
successfully applied to study the cracking phenomena in dams [2-6]. However, not much work is
reported regarding the crack at the interface of concrete dam/rock foundation, witch is one of
potential sites of cracks initiation and propagation. The determination of the stress intensity factor
at the crack tip is one the possible means to analyse the behaviour of interface cracks in gravity
dam. It is known that the finite element method gives, with a great accuracy, the stress intensity
factors at the crack tip.
In this study, the finite element method is used to analyse the behaviour of interface cracks
between concrete and rock in gravity dam by the plot of the mode I and II stresses intensity factors
as a function of the crack length. The effect of crack inclination according to the interface is
highlighted. The effects of hydrostatic pressure in all studied cases were highlighted.
The dam used for the analysis is a gravity dam type. It is 80 m high and 60 m width. The crest
of the dam is 5 m width. The geometrical model is idealised by a mesh having 2125 eight nodded
elements. The crack tip is modelled with the standard quarter-point elements. A linear elastic
material model coupled to a linear elastic discrete fracture model was used in the analysis.
As an example of the obtained results, figure 1 presents the variation of the mode I stress
intensity factor (KI) as a function of the normalised crack length a/w (where w is the width of dam)
for different applied pressures ( h= 40, 60 and 80 m).
It can be noted that whatever the size of the crack, an increase of the high h leads to an
elevation of the stresses intensity factor at the crack tip. The opening of the crack is thus highly
affected by the applied hydrostatic pressure. One can see that the Mode I stress intensity factor
increases slightly according to the normalised crack length when this last is lower than 40%.
Beyond this value, the rate of increase of KI is more accentuated, what allow us to confirm that the
opening of the crack becomes dangerous when its length exceeds 40% of the dam width.
Figure 2 illustrates the variation of the mode II stress intensity factor (KII) according to the
normalised crack length a/w. It is shown that KII practically does not vary with the crack length
when this last is lower than 70% of the dam width. From this length, the absolute value of KII,
increases with a great rate. According to figure 2, It can be seen that the effect of the water pressure
on the KII variation is negligible when the ratio a/w is lower than 0.7.
542 B. Bachir Bouiadjra et al.

FIGURE 1. Variation of KI according to the crack length FGURE 2. Variation of KII according to
the crack length

References
1. Bruhwiler, E. and Wittmann, F.H., Engng Fract Mech, vol 35, 565-571, 1990
2. Ingraffea, A.R, Linsbauer, H.N. and Rossamanith, H.P., In Shah SP, Swartz SE, editors. SEM/
RILEM International Conference on Fracture of Concrete Rock. Houston, Texas, 1987,311-
33.
3. Saouma,V.E, Ayari, M.L. and Boggs, H., In Shah SP, Swartz SE, editors. SEM/RILEM
International Conference on Fracture of Concrete Rock. Houston, Texas, 1987, 334
4. Widmann, R., Engng Fract Mech, vol. 35, 1-3, 1990
5. Chandra JM, Singh KD, , Engng Fract Mech 2001;68;201-19.
6. Ingraffea, A. R., Engng Fract Mech, vol. 35, 553-564, 1990
.
 2T21. Concrete and rock 543

APPLICATION OF COMPOSITE MECHANICS TO COMPOSITES ENHANCED

CONCRETE STRUCTURES

C. C. Chamis and P. K. Gotsis

NASA Glenn Research Center, Technological Education Institute
Cleveland, OH 44135, 62125 Serres, Greece
Christos.C.Chamis@nasa.gov, pkgotsis@email.teiser.gr

Reinforced concrete is widely used in the construction industry. Concrete tends to crack, chip and
be damaged as a result of inadvertent loads or overloads which may not have been accounted for in
the initial design. The damage in concrete structures may extend to a state where the safety of that
structure becomes a major concern. Recently, a considerable effort is being expended on repairing
damaged or upgrading concrete structures by using fiber reinforced composites. The use of
composites is natural since the repairing composites tend to be thin laminates which are easily
bonded to damaged concrete structures which are of cylindrical and flat surfaces in general.
Composite jacketed concrete is the most effective repair and upgrade method for concrete
structural members such as columns that are accessible for all sides. The method is commercially
available. Concrete surfaces need be cleaned and prepared for bonding. Unidirectional fiber
composite prepreg tape is wrapped around columns helically with sufficient number of layers to
provide the necessary confinement of concrete. The composite tape is usually glass/epoxy for
small to medium size columns. More expensive graphite/epoxy tapes may be needed for
upgrading larger columns that require greater stiffness composite reinforcement. Graphite/epoxy
composites that have greater stiffness and are more expensive are also not as ductile as glass/epoxy
composites. After sufficiently wrapping the structural member with the required type of prepreg
tape, it needs to be heated to the necessary curing temperature to cure the thermosetting epoxy
binder. The curing temperature is generally in the range of 121-177°C (250-350°F). Commercial
companies provide the necessary tools that automate the fiber composite tape wrapping and curing
processes. This method is very economical and effective for upgrading columns of existing older
structures that, according to present code, do not have sufficient lateral reinforcement for concrete
confinement.
Composite protective covers are for environmentally exposed concrete infrastructure subjected
to chronic degradation due to exposure to hygral, thermal, and mechanical cycling, deicing agents,
and loading effects. The maintenance of structures at a good service condition requires constant
vigilance for inspection and repair of degraded components. Inspection and repair of
transportation structures is costly and requires the disruption of traffic for significant internals. In
many cases transportation structures are neglected until they degrade beyond repair and a very
costly removal and replacement process becomes necessary. When degraded concrete structural
components such as bridge girders and decks are repaired or replaced, their future durability may
be significantly enhanced by protecting the concrete from environmental factors by providing a
resilient fiber composite cover. The most significant protection is from the surface water and
deicing salts. Similarly, when pavements on bridges are replaced, a fiber composite membrane
layer may be added above the structural deck to protect the deck and girders.
Transportation structures are subject to adverse environmental factors as well as mechanical
loading. The selection of a particular design must take into account all significant factors affecting
durability. Infrastructure elements can be designed with more general and adaptable fiber and mat
reinforcement architectures than currently available. The potential to design structures with
braided/woven composites to satisfy long term durability and safety requirements is excellent. For
544 C. C. Chamis and P. K. Gotsis

the very same multiplicity of design options that promise great potential, the realization of that
potential will depend on the availability of evaluation tools that take into account all aspects of
structural durability. The fundamental premise in the implementation of a structural durability
model is that interaction among the various factors affecting durability is of utmost importance and
must always be taken into account. The methodology used to enable hybrid structural durability
assessment could be a generalization of the proven and demonstrated computational capability for
the progressive damage and fracture evaluation of laminated composite structures. Extensive
experience gained in the development of computational capabilities on laminated composites will
be highly beneficial for the success of the assessment of fiber composite infrastructure
applications.
Different methods for designing and analyzing thin laminates have been developed and are
available in many computer codes. Though this area of research is receiving considerable attention
lately, the application of composite mechanics, in order to simultaneously simulate the reinforced
concrete section with the thin composite layer, has not been recognized as yet. Recent research at
Glenn Research Center demonstrated that damaged concrete structures and their repairing
composites can be simulated simultaneously by composite laminate analogy which is available in
some of those computer codes. By using composite laminate analogy, we can represent any
concrete structural section by assuming that it consists of several layers through its thickness. By
so doing, we take advantage of all the features available in computer codes for composite
mechanics – for example ICAN (Integrated Composite Analyzer). The code that is used in the
simulation described herein is the ICAN [1]. The objective of the proposed paper is to describe
those composite mechanics and attendant computer codes, and illustrate their application to select
reinforced concrete structural sections and structures. The appropriate composite mechanics is
described briefly. Then, it is applied to select structural sections and to select structures (special
arch and a dome). Note that results presented herein are computational. ICAN is designed to carry
out a comprehensive analysis including the hygral, thermal and mechanical properties/response of
multilayered continuous fiber reinforced polymer matrix composites.

Reference:
1. Murthy, P.L.N. and Chamis, C.C., Integrated Composite Analyzer (ICAN), Users and
Programmers Manual. NASA TP 2515, March 1986.
 2T21. Concrete and rock 545

INITIATION AND COALESCENCE OF LOCALS DAMAGES ON BLANCO DE

MACAEL MARBLE

Kais Mehiri, Pascal Vieville1, Paul Lipinski2, Albert Tidu3 and Valentin Tijeras4
Ecole Nationale d’Ingénieurs de Metz (ENIM)
LFM1 / LPMM2 / LETAM3
Centro Tecnico de la Piedra, Spain4
k.mehiri@enim.fr, vieville@enim.fr, lipinski@enim.fr, tidu@letam.univ-metz.fr, vtijeras@ctap.es

In the context of a project of characterisation of Spain marble and its integration in the construction
domain like a building support material, an analysis of the different Macael’s marbles is in
progress. The final purpose is the conception of a simulation tool able to transform the material
from a safe status to a ruin one. The damage description software tool is based on the Incremental
Self Consistent method which permits to realise the transition between the mesoscopic and the
macroscopic scales [1]. This method takes into account the void inclusions (pores), which
represent one of the principal sources of the local-failure mechanisms initiation.
A first study concerning observations at various scales of three Macael marbles (Blanco de
Macael, Travertino Rojo and Crema Real) has been done [2]. This study permitted us to define the
internal architecture of the three studied marbles, and to determine the various chemical
compositions and the different structural heterogeneities.
To succeed in defining the design of the simulation tool, the second step is the identification
and the characterisation of the local failure mechanisms. In fact, considering the various
heterogeneities, the analysis of their propagation and their coalescence makes possible the
anticipation of the critical status of the material for a given morphology and a typical load [3]. That
is why we concentrate in this article on the study of various local damage mechanisms of Blanco
de Macael.
The precedent observation’s study permits us to conclude that the Blanco de Macael is a
marble with a polycrystalline structure and a fine grain (Fig. 1a) and such that the average grain
size is about 100µm (1425µm for largest one and 56µm for smallest observed). This size is
comparable to the building white marbles like Carrara marble 188µm and Dionysos 30µm [4].

FIGURE 1 : a-Optical (Polorasid Light) microscope picture of the Blanco de Macael, b-SEM
(BSEI) pictures of mineral inclusion on the Blanco de Macael

This marble presents a very marked homogeneous composition (Table. 1). In fact, the
observation under the SEM, with the BSEI to highlight the contrast of density, reveals that the
mineral inclusions are more dense then the calcite and the average size are about some
micrometers (Fig. 1b).
546 K. Mehiri et al.

TABLE 1. Some physical and chemical propreties of the Blanco de Macael marble.

According to the imperfection observations at various scales, we note that the defect
coalescence in the white marble is intergranular (Fig . 2a) and the damage induces the crush of the
grain into a microcrystals of a distinguishable rhomboedral crystal structure recognisable thanks to
its three edges (Fig. 2c) .
This paper aims to characterise and understand the various damage mechanisms on the Blanco
de Macael marble. The principal results deduced from the compressions tests and the observations
on samples at various damage levels are exposed. We also study the acoustics activities during the
compression tests in order to establish a correlation between the saved acoustic activities and the
damage level.

FIGURE 2 : a-Optical (PL) microscope picture of damaged Blanco de Macael sample b,c- SEM
(BSEI) pictures

References
1. Berveiller M., Zaoui A., Modelisation du comportement mécanique des solides
microhétérogènes. Learning documentations CNRS, 1998.
2. K .Mehiri, P.Viéville, P.Lipinski, A.Tidu , V.Tijeras, Fifth international conference on
fatigue and fracture, Politecnico di Bari, Italy, paper 15, 2005.
3. Broohm A.,Lipinski P., Zattarin P., Prediction of mechanical behaviour of inhomogeneous
and anisotropic materials using an incremental scheme, Arch. Mech., 52, 6, pp 947-967, 2000.
4. Cardani.G., Meda.A., Flexural strength and notch senitivity in natural building stones:
Carrara and Dionysos marble. Construction and Building Materials 13, 393-403 ,1999.
 2T21. Concrete and rock 547

INFLUENCE OF CONCRETE´S MINERALOGICAL COMPONENTS ON

FRACTURE COMPRESSIVE AND TRACTIVE

M.P. Morales Alfaro and F.A.I. Darwish

Universidade Federal Fluminense RJ,Rua Passo da Pátria 156-3 andar – Sala 365 - bloco “D” São
Domingos – Niterói - RJ Brazil – CEP 24210-020
pattiwasi@hotmail.com or fadarwish@civil.uff.br
Intoduction
Diverse specialties as Geology, Chemistry, and Civil Engineering, are responsible for the selection
of materials utilized to produce and prepare construction materials. The isolated knowledge
between who manufactures and who constructs, leads, in general, to control its quality for its
mechanical behavior without knowledge of external properties of the mineralogical components of
the matrix that conforms them.
History of the problem
Few researches about the mineralogical influence of concrete components on the mechanical
properties, could not explain many mechanical behaviors. Said researches have generated a
controversy since 1964 in relation to the reason for the effect of water/cement in the tenacity of the
fracture of concrete. According to Dos Santos (1998, p.16) [1]:
• "As well as Petersson (1980), one concludes that the value of the relation water/cement
influences inversely in proportional form the tenacity of the concrete’s fracture. When the
relation water/cement is increased, the tenacity of the fracture diminishes. This conclusion
opposes the position of LOTT and KESLER (1964)".
Objective
The objective of this research is to determine if the mineralogy of the components of concrete
influences significantly in the formation of the surfaces of fracture when submitting it in
compression, traction for compression, and specially, to the direct traction for the item above
displayed.
Materials and experimental techniques
This Study used Cement Portland V-ARI-RS for high initial resistance, course aggregate had
been triturated and were from three geologic origins with Tnom. max. ½ inch, m.f 7.6 and the
same granulometry. Fine aggregate had m.f 2.43. The tests were done in cylindrical concretes of
6x12 inches except the high strength concrete, which was done in 4x8 inches. The test in tenacity
used “short rod” cylindrical specimen.
The study was conducted in:

TABLE 1: Characteristic Compressive Strength’s Concretes and slump

Results and analise

548 M.P. Morales Alfaro and F.A.I. Darwish

TABLE 2 – Summarized Properties mechanics of the concretes

The results in this study of the reason for the influence water-cement in the tenacity of the
fracture in the concrete standard revealed essentially the independent type of course aggregate and
the reason water-cement in virtue of the predominance of the effect of pores, humidity, and the low
cement content of the mixture. For the concrete of medium resistance, the tenacity increases with
the increase of the reason water-cement. Being kept invariable the cement contents increase the
tenacity of the concrete. The increase of the reason water-cement can be attributed to the
performance of the pores as energy spendthrifts during the process of consistent fracture observed
in the concrete standard. The superiority of the tenacity of the concrete of medium resistance in
relation to the concrete standard is attributed to the biggest cement content. In relation to the
tenacity of the concrete of high performance a value of around 2 MPa—m resulted as minimum
limit with a relation water-cement of 0,36. Comparing with the results of the tenacity for concrete,
and the standard of average resistance, this high value resulted from a reduction of the reason
water-cement; even with similar cement content of this last one, a comparison is not possible. The
study with high performance concrete, used microsilica and additive superfluidificante; the effect
of these additives tends to eliminate the porosity and transition zones are filled in almost its totality
so make the interface aggregate-mortar very resistant, and the rupture mechanism in general goes
to be given by breaking of aggregates different of the mechanism of rupture of the concrete
standard and medium resistance. It is a probably explication for the high value of the high strength
concrete.

References
1. Santos,A.C; Sousa, J.L.A.O.; Bittencourt, T.N. “Determinação Experimental da Tenacidade
ao Fraturamento do Concreto com Corpos de Prova do tipo “short rod”, BT/PEF/9807,
Boletim Técnico da Escola Politécnica da Universidade da São Paulo, (1998, p16).
 2T21. Concrete and rock 549

CONSTITUTIVE MODEL FOR DESCRIPTION OF HIGH-STRAIN RATE

BEHAVIOR OF CONCRETE

I. R. Ionescu and O. Cazacu

Laboaratoire de Mathematiques, Universite de Savoie, Le Bourget-du-Lac, France
ioan.ionescu@univ-savoie.fr
Department of Mechanical and Aerospace Engineering, GERC, University of Florida, Shalimar,
FL 32579, USA
cazacu@gerc.eng.ufl.edu

Kinetic energy penetration phenomena are of interest in a variety of applications ranging from
terminal ballistics to protection of spacecraft due to meteoroid impact, containment of high mass or
high velocity debris due to accidents or high rate energy release, design of hardened protective
facilities, erosion and fracture of solids due to impact, etc. The occurrence of multiple phenomena
in the target such as localization, plasticity, anisotropic damage, fragmentation pushes the limits of
existing modeling and computational capabilities for description of the target response. Since the
deformation rates are very large an eulerian description coupled with a fluid-like model are
suitable to describe the target. Fluid-type constitutive equations have been used to describe the
high-strain rate behavior of metallic materials (e.g. Batra [1]). Implicit in these models is the
hypothesis of incompressibility, hence those models cannot describe the irreversible volumetric
changes observed in porous materials (e.g. geologic or cementitious materials).
In this paper, an extension of the Bingham fluid model that captures the combined effects of
plasticity, damage and viscous effects on the behavior of porous media when subjected to large
deformations and high strain rates is presented. In contrast with a classical fluid, which cannot
sustain a shear stress, we suppose that at rest the Cauchy stress tensor V must belong to an
admissible convex set K. Conversely, if the stress is in K, then the rate of deformation tensor
D D (v) ( ’ v  ’ T v ) / 2 (v denotes the Eulerian velocity field) vanishes. If the stress
tensor is not in K then there is flow, the rate of deformation tensor D being subject to certain
kinamtic conditions, i.e.

­° V = f ( D , h ) if D (v ) z 0 ,
D (v)  C , ®
°̄ V  K if D (v ) 0.
(1)
where h is a set of internal variables. At difference with a classic fluid, the function f involved in
(1) is not defined and cannot be prolonged by continuity in D =0. The convex K is defined by a
continuous scalar function that describes the flow-no flow condition. To capture the combined
effects of plasticity and damage, we propose the following specific form of (1):

­ ª k(P , U ) º
° V c = « 2K + » D c if D c z 0 ,
® ¬ Dc ¼
°
¯ V c d k ( P , U ) if D c 0.
(2)

­ tr (V ) / 3 =  p c ( P )  ( O  2K / 3) div ( v ) if div (v )  0,
®
d iv ( v ) d 0 , ¯ tr (V ) / 3 t  p c ( P ) if div (v ) 0.
550 I. R. Ionescu and O. Cazacu

with two internal variables: the volumetric strain P and a damage parameter G associated to
the loss of cohesive strength due to air pore collapse. The model is further applied to the
description of the steady-state flow of a cementitious material over a penetrator. A mixed finite-
element and finite-volume strategy was developed. Fig. 1 shows the distribution of D . Note that
a zone of intense inelastic deformation develops around the penetrator and it extends outward to
about 3 projectile radii from the centerline. Outside this zone, the material is rigid. To capture
sharply the boundary between the rigid domain and the domain of inelastic, a mesh adaptation
strategy and an anisotropic mesh generator was used.

FIGURE 1. The distribution of D for a concrete material.

We have found that damage occurs ahead of the projectile and along planes which are not
symmetric with respect to the penetrator centreline. This damage induced anisotropy may be
responsible for observed trajectory instabilities.

References
1. Batra, R.C., Int. J. Engng.Sci, vol. 25, 1131-1141, 1987.
 2T21. Concrete and rock 551

HYDRAYLIC FRACTURING IN WEAK ROCKS

P. Papanastasiou
Department of Civil and Environmental Engineering, University of Cyprus,
1678 Nicosia, Cyprus
panospap@ucy.ac.cy

Hydraulic fracturing (HF) is a technique used to stimulate oil and gas reservoirs by inducing
fractures in the formation and then propagating them by the injection of a high viscosity fluid [1].
HF is also used in Environmental engineering for waste disposal in shallow formations, for
cleaning up contaminated sites and in Geotechnical projects such as injection of grout,
permeability testing, deep well injection and dam construction.
A correct prediction of hydraulic pressure needed to initiate and propagate the fractures and the
created fracture geometry is vital for safely designing such a process. The fracturing fluid-pressure
is the only parameter measured in the field that is available for controlling and evaluating the HF
treatment. The currently-used hydraulic fracture propagation simulators are based on linear
elasticity and often underestimate the down-hole pressure that is measured in the field operations
[2]. A particular task of this research was to investigate the effect on non-linear rock behaviour on
the pressure needed for propagation and on the dimensions of the created fractures.
The physical and mathematical modelling of hydraulic fracturing leads to strong non-linearity
characterized by full coupling between the viscous flow of the fracturing fluid, the rock
deformation and the fracture propagation processes. Fluid-flow in the fracture is modelled by
lubrication theory. Rock deformation is modelled by the Mohr-Coulomb flow theory of plasticity.
The propagation criterion was based on the softening behaviour of rocks and it was incorporated in
a non-linear cohesive type model with interface elements. Such a criterion is often used in the
modelling of the cracking process in concrete. We solved the problem numerically by developing a
fully coupled elastoplastic HF FEM code. In the FE code we incorporated a meshing/ remeshing
scheme and a special continuation method (arc-length type) based on the volume of the injected
fluid for controlling the solution during fracture propagation and closure [3].
We found that plastic yielding provides a shielding mechanism near the tip resulting in an
increase of the effective fracture toughness which was determined, independently from the
propagation criterion, using the J-integral [4]. Higher pressure is needed to propagate an
elastoplastic fracture and the created fracture is shorter and wider than an elastic fracture [5]. We
demonstrated that the standard HF simulators, which are all based on elasticity, will yield better
results if the unloading modulus is used as the Young's modulus. This is explained by the fact that
in along fracture propagation the material unloading behind the advancing fracture dominates in
the rock deformation process.
We have also modelled fracture closure in an elastoplastic formation. In practice, the
measured pressure vs time is analyzed during closure of a mini hydraulic fracture for determining
the insitu stresses and permeability, parameters that are subsequently used for calibrating the long
propagation modelling [6]. We found that the closure pattern of a fracture which has first been
propagated is completely different from the closure pattern of a pressurized stationary fracture. A
pressurized stationary elastoplastic fracture closes uniformly bur remains open after the applied
load is released. An elastoplastic fracture which has first been propagated makes surface contact
initially near the fracture tip and subsequently towards the mouth of the fracture [7]. Therefore, the
assumption made that the fracture closes completely once the fluid-pressure in the fracture drops to
the value of the far-field stress, is not valid. These results are in agreement with the observations in
552 P. Papanastasiou

experiments carried out on fracture propagation and closure on large blocks of soft rocks [8]. In
addition we showed that the formation will be more stable after it is fractured due to the
redistribution of the stresses [9]. Therefore, rock failure causing sanding during production of
hydrocarbon is minimized after hydraulic fracturing.

Acknowledgement
This research was supported by Schlumberger Cambridge Research.

References
1. Economides M.J. and Nolte, K.G., Reservoir Stimulation, Prentice Hall, Englewood, N.J.,
USA, 1989.
2. Palmer, I.D. and Veatch Jr R.W.. In Proceedings of the 62nd Annual Technical Conference
and exhibition of the Society of Petroleum Engineers, Dallas, Texas, 1987, SPE 16902,
3. Papanastasiou, P. , J. Comp. Mech.s, Vol. 24, 258-267, 1999.
4. Rice J.R., J. Appl. Mech., Vol. 35, 379-386, 1968
5. Papanastasiou, P., Int. J. Fracture, Vol. 96, 127-147, 1999
6. Nolte, K.G., In proceedings of the SPE Annual Technical Conference and Exhibition, New
Orleans, LA, 1990, SPE 20704
7. Papanastasiou, P. Int. J. Fracture, Vol. 103, 149-161, 2000.
8. Dam, D.B.: "Influence of inelastic rock deformation on hydraulic fracture geometry", Ph.D.
Thesis, Delft University of Technology (1999).
9. Papanastasiou, P. Int. J. Num. Anal. Meth. Geomech., Vol. 23, 1927-1944,1999.
 2T21. Concrete and rock 553

APPLICATION OF FRACTURE MECHANICS ON UNREINFORCED

CONCRETE WALLS

Thomas Eck, Bong-Gu Kang and Wolfgang Brameshuber

Institute of Building Materials Research, RWTH Aachen University
Schinkelstr. 3, 52064 Aachen, Germany
eck@ibac.rwth-aachen.de

In this contribution investigations on the size effect of unreinforced concrete walls which are
funded by the Deutsche Forschungsgemeinschaft (German Research Foundation) are described.
Inducement of this project was the subordinated meaning of unreinforced concrete walls for
residential buildings caused by not available design models, although these walls have build-
practically and economically substantial advantages. The major reason for the absence of relations
according to plain concrete, which could be consulted for design, are the missing experimental
investigations on the load configuration of vertical force and bending acting together.
In order to cover all possible load cases at altogether 5 different concrete mixtures,
investigations on the bending and simultaneous acting longitudinal force demand as well as pure
bending (to determine the mechanical and fracture-mechanical properties of the concrete) without
longitudinal force and pure excentric longitudinal force (with an excentricity up to d/3) at three
specimen sizes were carried out.
With the results of accompanying investigations determining concrete-technological parameter
the experiments could be reconstructed on the basis of numerical simulations and material laws for
the respective concrete mixtures have been evaluated. After the final evaluation of the bending test
the determined scale influences are arranged and on the basis of these results FE-models are set up
for the simulation of the one-dimensional experiments up to first computation of a wall including
the boundary conditions. In the next step the boundary conditions of the wall are varied in the
context of the FE simulation, in order to activate the system load-carrying capacity apart from the
concrete tensile strength.
Vibrated concrete C25/30 with a maximum particle size of 8 and 32 mm, C50/60 with a
maximum particle size of 32 mm and C80/95 with a maximum particle size of 16 mm were
examined also like self-consolidating concrete C50/60 with likewise 16 mm maximum particle
size. Due to the investigation of notched and not notched samples, the variation of excentricity and
loading level as well as the execution of at least three similar tests, a total number of approx. 300
test results.
Beams of three different sizes have been tested, the smallest with a height of d = 100 mm and a
span of l = 500 mm, the medium with dimensions d/l = 200/1000 mm and the largest with d/l =
400/2000 mm. The specimen thickness of all sizes was kept constant 100 mm.
In the next figures some results of the different experiments and simulations are shown. In
figure 1 the FE-Model used for the simulation of the three point bending test (only one half of the
beam was simulated) and the material law of vibrated concrete C25/30 is given.
554 T. Eck et al.

FIGURE 1. FE-Model of the three point bending test and resulting material law.

The next figure shows the experimental setup for the excentric longitudinal force tests by the
example of a small beam and the fracture pattern on the right hand site.

FIGURE 2. Excentric longitudinal force (setup and fracture pattern).

With the results of the experiments based on the three load cases it is possible to simulate the
fracture behaviour of a wall including the effects of different boundary conditions. The last step to
verify the fracture process and examine the results of the FE-analysis is the investigation of
unreinforced concrete walls in full scale.
 2T21. Concrete and rock 555

SUBCRITICAL CRACK GROWTH IN ROCKS UNDER WATER

ENVIRONMENT

Yoshitaka Nara, Hirofumi Kurata and Katsuhiko Kaneko

Graduate School of Engineering, Hokkaido University
Kita 13 Nishi 8, Kita-ku, Sapporo 060-8628, JAPAN
nara@eng.hokudai.ac.jp

Classical fracture mechanics postulates that the crack propagates dynamically when the stress
intensity factor reaches a critical level, that is, fracture toughness [1]. However, the crack can
propagate slowly even when the stress intensity factor is less than the critical level. This
phenomenon is called subcritical crack growth, and the main mechanism of subcritical crack
growth is stress corrosion [2]. Subcritical crack growth is one of the main causes of time-
dependent behavior in rocks. It has been shown that subcritical crack growth in rocks under air
condition is facilitated by water vapour pressure [3]. In this study, subcritical crack growth in rocks
under water environment is investigated.

In this study, the Double Torsion (DT) test [4] was used as a testing method. In Fig. 1, a
schematic illustration of a DT specimen and the loading configuration are shown. In Fig. 1, P is the
applied load. In this testing method, the stress intensity factor is independent of the crack length.
The experimental result obtained for rock is shown in Fig. 2. Rock studied was Kumamoto
andesite [5]. It is shown that subcritical crack growth is facilitated when the temperature is high
from Fig. 2. In Fig. 3, the experimental result obtained in air [2] and in water at the same
temperature (284 K) is shown. It is clear that subcritical crack growth is facilitated in water. It can
be concluded that subcritial crack growth in rock is facilitated by water and temperature.
556 Y. Nara et al.

References
1. Griffith, A.A., Phil. Trans. Roy. Soc., vol. 221, 163-198, 1920.
2. Atkinson, B.K. and Meredith, P.G., in Fracture Mechanics of Rock, Academic Press, London,
U.K., 111-166, 1987.
3. Nara, Y. and Kaneko, K., Int. J. Rock Mech. Min. Sci., 2005 (in press).
4. Evans, A.G., J. Mater. Sci., vol.7, 1137-1146, 1972.
5. Jeong, H.S., Nara, Y., Obara, Y. and Kaneko, K., In Proceedings of the International
Symposium on the Fusion Technology of Geosystem Engineering, Rock Engineering and
Geophysical Exploration, Seoul, Korea, 2003, 221-228.
 2T22. Sandwich structures 557

STRESS ANALYSIS AND PREDICTION OF FAILURE IN STRUCTURALLY

GRADED SANDWICH PANELS

Anders Lyckegaard, Elena Bozhevolnaya and Ole Thybo Thomsen

Institute of Mechanical Engineering, Aalborg University
Pontoppidanstræde 101, 9220 Aalborg East, Denmark
al@ime.aau.dk

Structural sandwich panels can be considered as a special type of composite laminate where thick,
lightweight and compliant core material separates two thin, stiff and strong face sheets. Sandwich
panels transfer bending and shearing load very effectively, but at the same time they offer limited
resistance towards application of transverse loads, see e.g. Zenkert [1] or Allen [2]. A commonly
used method of overcoming the problems associated with concentrated transverse loading is to
introduce core inserts of higher strength and stiffness. This usually solves the problem of local
reinforcement, but it also introduces new interfaces between materials of very different elastic
properties.
Bozhevolnaya and Thomsen [3-4] investigated the static strength and fatigue strength of the
core junctions experimentally for a number of different designs that were originally proposed by
Bozhevolnaya et al. [5], see Fig 1. The beams tested in [3-4] were comprised of aluminium face
sheets and core materials made from cross-linked PVC foam. It was found that the static strength
and the fatigue strength in particular could be improved by redesign in the form of a change of
shape of the core junction (referred to as structural grading of the core).
In the present investigation the sandwich structures that were evaluated experimentally by
Bozhevolnaya and Thomsen [3-4] will be the subject of a detailed stress analysis. The stress
distribution near the tri-material wedge at the core junction is known to exhibit stress singularities
within the frame work of linear elasticity theory, see e.g. Pagaeu [6]. Hence, it is assumed that
crack initiation will start here, which the aforementioned experimental results generally supported.
For the purpose of comparing various designs of core junctions, the stress fields at the
junctions will be evaluated based on a point stress criterion for crack initiation, which was
previously used successfully for cross linked PVC foams by Ribeiry-Ayeh and Hallström [7]. If
the application of a point stress criterion proves successful, then it may be used as a future basis for
optimisation of core junctions in sandwich structures.
558 A. Lyckegaard et al.

FIGURE 1. The geometry and loading of the studied sandwich beams, a) conventional butt
junction, b) scarf junctions with obtuse angles of the main core at the lower side; c) scarf junction
with sharp angles of the softer core at the lower beam side; d) butt junctions with reinforcing
patches.

References
1. Zenkert, D., An Introduction to Sandwich Construction, EMACS, 1995
2. Allen, H. G. Analysis and Design of Structural Sandwich Panels, Franklin Book Co, 1969
3. Bozhevolnaya, E and Thomsen, O.T., Composite Structures, Vol. 70, pp 517-527, 2005.
4. Bozhevolnaya, E and Thomsen, O.T., Composite Structures, Vol. 70, pp 528-533, 2005.
5. Bozhevolnaya, E., Lyckegaard, A., Jakobsen, L. and Thomsen, O.T. “Sandwich Panel and a
Method of Producing a Sandwich Panel” Application No./Patent No. 03027020.1- (Danish
Patent P14477DK00), 2003
6. Pageau, S. S., Joseph, P. F., Biggers, Jr., S. B., International Journal Solids and Structures,
Vol. 31, No. 21, pp. 2979-2997, 1994
7. Ribeiro-Ayeh, S, Hallström , S., Engineering Fracture Mechanics, Vol. 70, pp 1491-1507,
2003.
 2T22. Sandwich structures 559

DEBONDING AND KINKING IN FOAM-CORE SANDWICH BEAMS

D. A. Zacharopoulos, V. D. Balopoulos, Z. S. Metaxa, P. A. Kalaitzidis and E. E. Gdoutos

Department of Civil Engineering, Democritus University of Thrace
12 Vassilissis Sofias St., Xanthi, 67100, Greece
dzachar@civil.duth.gr, vbalop@civil.duth.gr, zoi_metaxa@yahoo.gr, pkalait@civil.duth.gr,
egdoutos@civil.duth.gr

In this work we consider the effects of debonding in a double cantilever beam (DCB) specimen of
aluminum faces and PVC-foam core (Divinycel H, see [DIAB International AB, Divinycell Grade
H Technical Manual, Sweden, 2003.]), as shown in Figure 1 below. The configuration follows the
one proposed by Prasad and Carlsson [5], which is similar to the standard ASTM D5528 94a peel
test [ASTM Standard D5528-94a. “Standard Test Method for Mode-I Interlaminar Fracture
Toughness of Unidirectional Fiber-Reinforced Polymer Matrix Composites,” ASTM Annual Book
of ASTM Standard 15.03, 283–292, American Society for Testing and Materials, 1999.]. All
materials are linear elastic, and four cases of core density are considered (H 60, H 80, H 100, and H
250), which have been the subject of other investigations [e.g., Shivakumar, K., Chen, H., and
Smith, A. S., Journal of Sandwich Structures and Materials vol. 7, 77–90, 2005.] too. All cases are
typical of foam-core sandwich structures, where the core is much more flexible than the adhesive
and faces. In each case debonding is introduced between the core and adhesive at the loaded end.

FIGURE 1. Sandwich peel-test specimen. (a) DCB specimen. (b) DCB specimen geometry.

Debonding is a common manufacturing defect of sandwich beams and panels that may severely
hamper their load bearing capacity. It is akin to pre-cracking between either the face and adhesive
or the adhesive and core. Face debonding is usually ignored, since face and adhesive materials are
commonly much tougher than core materials. Pre-cracking between adhesive and core may grow
either interfacially or into the core (kinking), as decided by energy release criteria. Under load-
control conditions, initiation of crack growth of either kind may leads to immediate (brittle)
structural failure. Stability considerations aside, the rational analysis and design of sandwich
sections requires at least some qualitative understanding of how the energy release rates and stress
intensity factors (SIFs) depend on the loading conditions, the length of debonding, and the
properties and dimensions of the surrounding materials.
Cracks in sandwich constructions have characteristic lengths (e.g., initial debonding length)
that are very large compared either to material characteristic lengths (e.g., bubble size in PVC
foam cores) or to possible process zones. Therefore, the assumptions and machinery of linear
elastic fracture mechanics (LEFM) may be employed in the study of sandwich debonding, and all
analytical and numerical results presented forthwith depend explicitly or implicitly on LEFM.
560 D. A. Zacharopoulos et al.

Thus, we concentrate our investigation on obtaining the stress intensity factors (SIFs) and total
energy release rates under unit loadings. Failure loads and type of crack propagation (interfacial or
kinked) follow immediately by comparison with the corresponding toughness measures for the
interface and the core.
Step-by-step crack propagation under plane-strain conditions is simulated in FRANC 2D (see
[Ingraffea, A. and Wawrzynek, P., User’s guide for FRANC 2D Version 3.1, www.cfg.cornell.edu,
1993.] for an introduction), and we obtain the total energy release rates ( G ) and the stress
intensity factors ( K I , K II ) for increasing lengths of interfacial and volume cracks. The
computational results in all cases considered are characterized by immediate kinking, followed by
rapid curving of the sub-interfacial crack and eventual propagation parallel to the interface. These
results are discussed with respect to their qualitative and quantitative features, with emphasis on
their theoretical and experimental ramifications.

References
1. Daniel, I, M., Gdoutos, E. E., Wang, K.-A., and Abot, J. L., International Journal of Damage
Mechanics, vol. 11, 309-334, 2002.
2. Zenkert, D., An Introduction to Sandwich Construction, Engineering Materials Advisory
Service, Sheffield, UK, 1995.
3. Prasad, S. and Carlsson, A.L., Engineering Fracture Mechanics, vol. 47, 813–824, 1994.
4. Prasad, S. and Carlsson, A.L., Engineering Fracture Mechanics, vol. 47, 825–841, 1994.
5. Sukumar, N., Huang, Y.Z., Prevost, H.J., and Suo, Z, International Journal for Numerical
Methods in Engineering, vol. 59, 1075–1102, 2004.
6. Oh, Hae-Soo, Journal of Computational Physics, vol. 193, 86–114, 2003.
7. Shivakumar, K., Chen, H., and Smith, A. S., Journal of Sandwich Structures and Materials
vol. 7, 77–90, 2005.
8. Ingraffea, A. and Wawrzynek, P., User’s guide for FRANC 2D Version 3.1,
www.cfg.cornell.edu, 1993.
9. ASTM Standard D5528-94a. “Standard Test Method for Mode-I Interlaminar Fracture
Toughness of Unidirectional Fiber-Reinforced Polymer Matrix Composites,” ASTM Annual
Book of ASTM Standard 15.03, 283–292, American Society for Testing and Materials, 1999.
10. DIAB International AB, Divinycell Grade H Technical Manual, Sweden, 2003.
11. Araldite Technical manual for “Adhesive AV 138M with Hardener HV 998”.
12. Pechiney Renalu Companny, Certification for Aluminium 2024-T3.
 2T22. Sandwich structures 561

MODELING CORE FAILURE BY THE TSAI–WU CRITERION IN THE DESIGN

OF FOAM-CORE SANDWICH BEAMS

E. E. Gdoutos, V. D. Balopoulos, P. A. Kalaitzidis and M. Konsta

Department of Civil Engineering, Democritus University of Thrace
12 Vassilissis Sofias St., Xanthi, 67100, Greece
egdoutos@civil.duth.gr, vbalop@civil.duth.gr, pkalait@civil.duth.gr, mkonsta@civil.duth.gr

This work is part of an extensive program at the Laboratory of Applied Mechanics of D.U.Th.
involving experimental and numerical research, to expand our knowledge of composite materials
and propose enhanced techniques for rational design of sandwich beams and shells. In particular, it
is a study of the importance of describing core failure in elastic sandwich beams by the Tsai–Wu
criterion (not by simple shear), for a variety of support conditions and loading patterns.
The Tsai–Wu criterion can account for combined normal and shear stresses in anisotropic
materials. In describing sandwich core failure, it represents a conservative elaboration of its
simple-shear counterpart and is used in combination with failure criteria for all other limit states.
For the plane stress conditions of a beam section, the Tsai–Wu criterion takes the simple form:

Fc  Ft V2 W2
V 1  1  62 1
Fc Ft Fc Ft F6 (1)

where the second-order stress V 3 has been ignored and the material strength in compression (Fc),
tension (Ft), and shear (F6) has been introduced. As a first approximation, we apply both criteria in
the context of Bernoulli kinematics and engineering beam theory.
We consider the foam core of a sandwich beam that satisfies common assumptions (almost
constant shear stress, M F c F t  1 , and ) without axial forces. The critical
fiber in any section is the most compressed one and its failure, according to the Tsai–Wu criterion
(for length L, width b, core thickness h c , and face thickness h f ), is given by:

(2)

The parameters and introduce scale

(typical action over typical resistance), is the load intensity, and

P  [ and J  [ describe the dependence of section moment and shear on the non-dimensional
position along the beam (i.e., the influence of the support conditions and the loading
pattern). For each given case of support and loading and each section location , equation (2) can be
solved for , so as to obtain the position of the critical section and [ c r i t  L  the critical load
intensity .
The process described above is implemented in MatLab and applied to a large number of cases
(presented in the full paper). In this summary we include two indicative diagrams obtained for a
sandwich beam with glass/epoxy faces and core made of PVC foam (Divinycell H100). In
particular, Figure 1 presents the critical load ( ) as a function of beam length for a tip-
562 E. E. Gdoutos et al.

loaded cantilever, superimposing new results for core failure according to the Tsa–Wu criterion on
earlier results (see [2]) for core failure in simple shear and face failure by wrinkling. Figure 2
presents the same information for a simply supported beam under parabolic loading.

FIGURE 1: Critical load vs length for tip-loaded cantilever sandwich beam

FIGURE 2: Critical load vs length for simply-supported sandwich beam under parabolic load

References
1. Allen, H.G., Analysis and Design of Structural Sandwich Panels, Pergamon Press, London,
UK, 1969.
2. Gdoutos, E.E and Daniel, I.M., in Proceedings of the 17th National Conference of Italian
Group of Fracture, edited by , Bologna, 2004, –.
3. Daniel, I.M., Gdoutos, E.E., Wang, K. A. and Abot, J.L., International Journal of Damage
Mechanics, vol. 11, 309–334, 2002.
 2T22. Sandwich structures 563

NUMERICAL INVESTIGATION OF CRACK PROPAGATION IN SANDWICH

STRUCTURES

E. E. Theotokoglou
Faculty of Applied Sciences, Dept. of Mechanics-Lab. of Strength Materials, The
National Technical University of Athens, Zographou Campus,
Theocaris Bld., GR-0157 73, Athens, Greece
stathis@central.ntua.gr

Low-density foam core sandwich composite find an increasing use as structural materials. They
are often used for high performance structural applications such as in the hull forms of racing
yachts, patrol crafts, lifeboats and catamaran ferries. There is a growing emphasis on selecting
different materials and their behavior under flexural loading. In addition, fracture mechanics
analyses of low-density foam core have been performed by several authors [1-3] considering linear
fracture mechanics on a macroscopic level. However only a limited amount of data are evaluated
about the fatigue crack growth in the core of sandwich beams [3-5].

FIGURE 1. Crack propagation in the core of a sandwich beam.

The purpose of the present work is to numerically investigate the fatigue crack propagation in
the core of sandwich beams under flexural loading. At first a core-skin debond parallel to the beam
axis is considered (Fig. 1), in accordance with experimental data [5]. When subjected to flexural
loading, this debond will propagate slowly along the top interface and eventually kinked into the
core as shear crack. Stress intensity factors are calculated using Finite Element Method, and
assuming linear fracture mechanics and plane strain conditions. Results from the finite element
analysis combined with the experimental data predict the crack growth behavior under flexural
loading. The simplicity of the proposed procedure and the numerical model developed, make
possible the prediction of the crack propagation for various types of sandwich beams under
flexural loading.

References
1. Gibson, LH. and Ashby, M.F., Cellular solid – structure and properties, 2nd ed. Cambridge
University Press, 1997.
564 E. E. Theotokoglou

2. Zenkert, D. and Backlund, J., Compos Sci Technol, vol. 34, 225-42, 1989.
3. Harte, A.M., Fleck, N.A., Int. J. Fatigue, vol. 23, 499-507, 2001.
4. Burman, M. and Zenkert, D., Fatigue crack initiation and propagation in sandwich
structures, Doctoral thesis, Report 98-29, Department of Aeronautics, Royal Institute of
Technology, Sweden, 1998.
5. Kulkarni, N., Mahfuz, H., Jeelani, S. and Carlsson, L.A., Composite Structures, vol. 59, 499-
505, 2003.
 2T22. Sandwich structures 565

LOCAL EFFECTS IN SANDWICH BEAMS: MODELLING AND

EXPERIMENTAL INVESTIGATION

M. Johannes, J. Jakobsen, V. Skvortsov1, E. Bozhevolnaya and O. T. Thomsen

Institute of Mechanical Engineering, Aalborg University
Pontoppidanstræde 101, 9220 Aalborg East, Denmark
1State Marine University of St. Petersburg, Russia

maj@ime.aau.dk

It is known that the joining of different cores in sandwich beams and panels leads to local
effects at these core junctions. The local effects arise due to mismatch of the elastic properties of
the adjoining core materials, and they manifest themselves by a rise of the in-plane stresses in the
sandwich faces as well as of the shear and through-the-thickness stresses in the adjacent cores.
These stress concentrations may cause local fracture of the weak core materials, but severe face
damage is also a highly possible scenario. This may jeopardize the structural integrity of the
sandwich element and cause its global failure.
Skvortsov and Thomsen [1] were the first to shed light on the physics of the local effects at
core junctions in sandwich beams subjected to transverse loading. Closed-form analytical
estimates of the stress concentrations are given in this work, and the accuracy of these estimates
was subsequently verified experimentally by Bozhevolnaya et al. [2]. These studies have
demonstrated that the locally induced stresses are of considerable magnitude under transverse
external loading. Thus the question arises what kind of local effects are induced in sandwich
beams/panels subjected to axial in-plane loading, which also occurs commonly in practice.
An analytical modelling of such a situation is performed on the basis of the exact 2-D theory of
elasticity solution. This solution considers the mismatch of the global stress-strain state of the
compliant and stiff cores far from the considered core junction. Here an axially applied force N
will cause in-plane normal stresses in the face and cores, which can be approximated by

N N G 1( 2 ) E 1( 2 ) h c
V f1(2) f , V c1(2) f , G 1( 2 )
2 h f (1  G 1( 2 ) ) h c (1  G 1( 2 ) ) 2E f h f (1)
The locally induced stresses that arise at the junction area are described with the help of an
extended Winkler foundation model, which couples the behaviour of the faces with the
longitudinal in-plane stresses in the supporting core materials.
The maximum local bending stress in the face and core shear stress are estimated to be

f 6h f ª 6Q c (1  Q c ) 1 º
V f
bend
k fV f1 c (1  c ) , W c1 V c1 f c « k 1  k2O »
hc ¬ 1 Q c O ¼ (2)

3
E c1 6 E c1 h c
c 4 , O 4
3
E c2 E fhf
(3)
where kf, k1 and k2 are known coefficients close to unity, c quantifies the mismatch of the elastic
core properties and describes the rate of decay of the local effects in the compliant core.
566 M. Johannes et al.

An example of a distribution of the face stresses of a sandwich beam subjected to uni-axial

tension is shown in Fig. 1. It is seen that face stresses, which are shown along the outer face surface
and the face-core interface, display drastic variations close to the junction (which is situated at
x=100 mm).

FIGURE 1. Stress distribution in the face of a sandwich beam subjected to uni-axial tension

An experimental investigation will be pursued for two different realistic combinations of

joined core materials; one example corresponding to a junction between polymeric foam cores of
different densities, and another example corresponding to the joining of a low density polymeric
foam core and a stiff wooden reinforcing patch, as used for rigging purposes in shipbuilding.
The nature of the local effects for the tensile in-plane loading condition will be studied by
comparing experimentally determined strains of the sandwich beam faces with results from finite
element modelling and estimations given by the analytical model.
In addition to data acquired in the elastic regime of deformation, the sandwich beams will be
loaded quasi-statically until the occurrence of fracture/failure, and the face strains, the overall
deformation patterns, the loads where the on-set of fracture occurs as well as the modes/patterns of
fracture will be monitored.
The analytical and finite element model results will be compared with the experimental
measurements, and various failure criteria will be evaluated in relation to their capacity to
accurately predict failure of the core junction and thereby of the sandwich structures considered.

References
1. Skvortsov, V. and Thomsen O.T., ‘Analytical Estimates for the Stresses in Face Sheets of
Sandwich Panels at the Junctions between Different Core Materials’, In Proceedings of the
6th International Conference on Sandwich Structures (ICSS-6), Ft. Lauderdale, Florida,
March 31-April 2, 2003, CRC Press, New York, 2003, 501-509.
2. Bozhevolnaya, E., Thomsen, O.T., Kildegaard, A. and Skvortsov, V., Composites. Part B:
Engineering, Vol. 34, pp 509-107, 2003.
 2T22. Sandwich structures 567

TYPICAL IN-PLANE RESPONSE SURFACES FORPRISMATIC FOAM-CORE

SANDWICH BEAMS

V. D. Balopoulos, P. A. Kalaitzidis, D. A. Zacharopoulos and E. E. Gdoutos

Department of Civil Engineering, Democritus University of Thrace
12 Vassilissis Sofias St., Xanthi, 67100, Greece
vbalop@civil.duth.gr, pkalait@civil.duth.gr, dzachar@civil.duth.gr, egdoutos@civil.duth.gr

This work employs the technique of response surfaces [1] to the 2-D statics of prismatic foam-core
sandwich beams subject to typical support conditions and transverse loading. The goal is to obtain
approximate functional forms for all responses that characterize the limit states of such beams
(e.g., support reactions, extremes of deformation and stress) and, hence, influence their optimal
design through the inequality constraints. This is part of a program of experimental and numerical
research at the Laboratory of Applied Mechanics, to expand our knowledge of composite materials
and propose enhanced techniques for rational design of sandwich structures.
A response surface (or curve, or hyper-surface) is an interpolation of a quantity of interest
(called the “response”) which is expensive to obtain either experimentally or numerically and
depends on one or more a priori known parameters. First, the response is measured or computed at
all points of a regular grid in the parameter space and, then, the resulting values are fitted with
inexpensive functions. However, the ease of application of this technique and the information
content of the results depend greatly on the choice of parameters and interpolating functions. An
obvious requirement is to choose parameters that are mutually independent and have a clear
physical meaning. An equally desirable, if less obvious, objective is that the dependence of the
response on the parameters may be decoupled, at least approximately. To achieve these goals, one
must have significant understanding of the response under study.
The use of response surfaces is essential in the context of robust global optimization, for
reasons of feasibility and efficiency explained at length in our separate work on the subject.
Nevertheless, the results presented are useful in the design and analysis of prismatic beams in
general, provided their shear flexibility is considerable and, hence, their response does not admit
closed-form solutions. All investigations carried out in this work regard linear behavior, both
geometrically (as ensured by the strict limits of serviceability) and materially. Furthermore, all
orthotropic materials are considered aligned with the principal sectional axes, so that the state of
pure bending is possible. Thus, the strain energy is proportional to the load amplitude, consists of
distinct shear and bending contributions, and involves no cross-terms and no contribution from
uniform stretch (geometric linearity).
Sandwich beams with foam core are highly deformable in shear. Therefore, Bernoulli
kinematics is inaccurate for describing their stress and strain fields and their sectional stiffness.
Even Timoshenko kinematics may occasionally prove insufficient, especially in the recovery of
intra-sectional fields from sectional strain measures and stress resultants. In such circumstances,
the engineering split of 3-D behavior into 2-D (sectional) and 1-D (beam) behavior is salvaged by
means of Variational Asymptotic Beam Sectional Analysis [2]. This method provides enhanced
accuracy in the evaluation of sectional stiffness and in the recovery of sectional fields not only for
prismatic rods but for pre-curved and pre-twisted ones as well. A piece of software, that relies on
this method and goes by its initials (VABS), is available free for academic use from [3].
Due to the decoupling of the strain energy, the sectional stiffness is diagonal and beam
behavior is governed by k ss (shear stiffness, units of force), k b b (bending stiffness, units of
568 V. D. Balopoulos et al.

moment u distance), and (slenderness), where is the approximate static depth

of the section. One may also scale the sectional strain measure to produce
and make commensurate with k s s . Both k s s and k b b are multiples
of , with non-dimensional coefficients of proportionality that depend strongly on
(geometry) and (material), and only weakly on E c G c (i.e., on Q c ).
Finally, the intra-sectional stress field depends on the section characteristics, as quantified by Og ,
Om , and Q c , and on the scaled measures of strain, İ , and stress, .
The product of this work is three types of response surfaces, which interpolate:

1 the stiffness coefficients and (produced by VABS and compared to

those of “engineering” beam theory) as functions of O g , O m , Q c  ,

2 the support reactions, w m a x L , and ı , İ at the critical sections (obtained by solving
separately for each combination of support and loading conditions and for “unit”
amplitude, i.e., for total load equal to k ref ) as functions of k ref , and
3 the stresses at critical fibers of a section (produced by VABS and compared to those of
“engineering” beam theory) as functions of  ı , İ , O g , O m , Q c  .
The first type is simple and requires little work. The second type is easy, in principle, but
tedious in practice, because of the numerous combinations that must be considered. Conversely,
the third type is, in principle, the hardest to construct, because it depends on many parameters.
However, only response surfaces for the corrections to the “engineering” stresses (that VABS
introduces) are necessary and, since these corrections are small, rough fitting of the data is
sufficient.

References
1. Myers, H. M., Montgomery, C. D, Response Surface Methodology : Process and Product
Optimization Using Designed Experiments, Wiley-Interscience, 2002
2. Yu, W., Hodges, H. D., Volovoi, V., Cesnik, E.S. C., ‘On Timoshenko-like modeling of
initially curved and twisted composite beams’, International Journal of Solids and Structures
39, 2002, 5101–5121
3. Utah State University Department Mechanical and Aerospace Engineering, Dr Webin Yu Site
http://www.mae.usu.edu/faculty/wenbin/
 2T23. Novel testing and evaluation techniques 569

NON-DESTRUCTIVE EVALUATION OF YIELD STRENGTH USING A NOVEL

MINIATURE DUMB-BELL SPECIMEN-AN EMPIRICAL APPROACH

G. Partheepan, D. K. Sehgal1 and R. K. Pandey1

Research Scholar, 1Professor
Department of Applied Mechanics, Indian Institute of Technology, Delhi, INDIA-110 016.
partheepan@gmail.com, profsehgal@yahoo.com, rkp@am.iitd.ac.in

This paper describes a method used to extract yield strength of the material using a newly
developed miniature specimen. An experimental setup is also developed to perform the miniature
test. Miniature tests were conducted on the proposed dumb-bell specimen fabricated from different
materials using the developed test setup. The miniature test output i.e. load-elongation diagrams
are obtained for the materials chosen for the present study. Standard test methods for predicting
mechanical properties require the removal of large volume of material samples from the in-service
component, which is generally impractical where there is a restriction on the available specimen
volume. The proposed specimen can be taken from in-service component without disturbing the
integrity of the structure. A linear empirical equation is proposed for the determination of yield
strength of the material. The yield strength of the material is obtained based on the miniature test
result i.e. yield load (load at breakaway) and the miniature specimen geometrical parameter. The
predicted yield strength value corroborates well with the experimental results.
Researchers in the past used either thick specimen or thin specimen having number of
machining operations. For the present study, a new dumb-bell shaped miniature specimen is
designed as shown in Fig. 1. The procedure for making this specimen is easier as compared to the
conventional sub size tension test specimens, which need a number of machining operations. More
number of machining operations on the specimen may change the properties as well.

Figure 1. Design of miniature tensile specimen.

The present work involves design of miniature specimen holder as the specimen used in the
present case being miniature in nature could not be directly gripped in the testing machine, fixture
for attaching the miniature specimen in the test machine along with the specimen holders. The
present study is carried out in Zwick machine. The miniature tests were conducted on the
specimens prepared from the materials selected for the present study. The tests are conducted at a
speed of 0.05 mm/min and at room temperature. At least five specimens were tested from each
material. Fig. 2 shows the miniature test load-elongation for the specimen from AR66 (Aluminum
alloy) material.
570 G. Partheepan et al.

Figure 2. Miniature test load-elongation curve for AR66.

The yield load (load at breakaway) is obtained from the miniature load-elongation curve for
AR66 and also for other materials. Before testing the miniature specimens were measured for its
geometrical parameters. Statistical analysis was performed on some input and output variables of
the miniature test on different materials. It is to check the repeatability of test so that the test results
can be used in predicting the material properties. It is observed that the statistical analysis results
are very satisfactory and consent to use the test results in predicting the yield strength of the
material. A new linear empirical relation for the prediction of yield strength based on yield load
and geometry of the miniature specimen is developed as follows:

§ Py ·
V y 1 .7 6 3 9 ¨ ¸  6 5 .9 7 3
©tw ¹ (1)
The results obtained from the empirical equation for different materials are in good agreement
with the experimental results.

Conclusions
A new empirical correlation is proposed for the determination of yield strength of the unknown
material using the miniature test results and its geometry. This miniature specimen can be prepared
from material taken from any in-service components in a non-destructive manner. The result from
the empirical correlation is found to be in good agreement with the experimental results.
 2T23. Novel testing and evaluation techniques 571

3D MEASUREMENT OF THE STRAIN FIELD SURROUNDING CRACK TIP

Daniel Vavrik, Jan Bryscejn, Jan Jakubek1 and Jaroslav Valach

Institute of Theoretical and Applied Mechanics of the Czech Academy of Sciences,
Prosecka 76, Prague 9, Czech Republic
1Institute of Experimental and Applied Physics of the Czech Technical University in Prague,

Horska 3a/22, 128 00, Prague 2, Czech Republic, E-mail: vavrik@itam.cas.cz

vavrik@itam.cas.cz

Finite elements method is widely used for numerical simulations of solid object behaviour in many
applications questioning in some cases the necessity of physical experiments. Available databases
of material behaviour are quite sufficient for simulations of linear systems. However for the
simulations of nonlinear systems, it is still necessary to have experimental information’s.
Unfortunately, the feedback based on strain measurement in several points only can lead to
accurate but imaginary numerical solution. Full field strain measurement is routinely performed
today. Let’s reduce the class of the analyzed object onto flat specimens only. Dependency on
loading type and studied feature, either the in-plain or out-of plain experimental method is usually
chosen. One of the highly non-linear problems is the failure of ductile specimen with stress
concentrator especially when failure is accompanied by the developing of damage zone in the
vicinity of the stress concentrator. Corresponding damage models have a number of parameters
which is necessary to determine. The feedback based on in-plane strain measurement only can lead
to a non-unique solution as well as without prior knowledge of material damage type. From this
reason both in-plane and out-of plane displacement measurement is required. On the other hand
monitoring the out-of plane displacement can detect the onset of intensive damage zone
development [1].
It is possible to employ the combination of two different optical methods using only one
camera to achieve 3D displacement/strain measurement. The proposed Coded Photometric Stereo
method for the measurement of the out-of plane displacement and the optical grid method for the
measurement of the in-plane strain field can be applied simultaneously.
The Photometric stereo (PS) method allows topographic reconstruction of the object surface
using its optical images [2]. PS utilizes the fact of close relation between the relative lightness of
uniformly illuminated surface and the angle to the direction of light. It means that knowing the
illumination geometry we can determine the slope (normal) of the surface at every point. Surface
topography is obtained as an integration of this normal field. This fact can be used for the purpose
of out-of plane measurements of loaded object deformations.
Three images required by PS using three lights illuminating surface from different directions
are sufficient for determination of both x and y slopes at any studied point of the surface regardless
on its colour variations. Knowledge of both slopes is advantageous for the reconstruction
integration especially in the case of locally irregular surfaces. For this reason three lights have been
used in our experiment.
Reconstructed topography is dimensionless, so dimension calibration must be done. It can be
done by one point mechanical measurement of the elevation differences between two points of the
studied surface or it is necessary to have some calibration object of known height/depth connected
with the surface.
Coded Photometric Stereo (CPS) is the enhanced PS method utilizing Red-Green-Blue (RGB)
lights coding. Monochromatic RGB lights positioned around the observed area produce directional
572 D. Vavrik et al.

illumination. Three different scenes coded in one compose image can be separated using standard
RGB colour channels of the digital camera we use. Out-of plane displacement is measured by CPS
using these scenes.
The compose image serves for the measurement of the in-plane displacement field by the grid
method. The grid method analyzes the deformation of the surface measuring grid. The method of
Interpolated Ellipses (MIE) has been selected for this purpose [3]. The MIE is a technique based on
the optical monitoring of deformations during loading processes of hexagonal grids of dots
deposited on the surface of the monitored specimen. Loading the specimen deforms a circle on the
surface into an ellipse. Each ellipse is interpolated by six neighbouring dots of the hexagonal grid.
Knowledge of the ellipse parameters directly yields the magnitude and the direction of principal
strains on the specimen surface.
Proposed combination of above described optical methods is suitable also for dynamical
experiments as the 3D displacement/strain field can be determined from one image.
Developing of the 3D displacement/strain field during loading of a flat high ductile specimen
with crack will be presented. Calculated plastic strain intensity developing will be presented as
well. Specimen will be loaded until failure will occur. Characteristic behaviour of the strain/
displacement fields regarding to damage zone evolution will be explored. As it is known from the
experiments performed in the past [1,3] intensive damage zone developing is accompanied by the
intensive developing of the plastic strain intensity [3] as well as by the characteristic delay in a
contraction evolution [1].

References
1. Vavrik, D.; Jakubek, J.; Pospisil, S.; Visschers, J.: In Proceedins of the 9th International
Conference on the Mechanical Behaviour of Materials, Switzerland, Geneva, May 25-29,
2003
2. Woodham, R. J.: Optical Engineering, 19, 139-144, (1980).
3. Vavrik, D.; Zemankova, J.: Experimental Mechanics, Vol. 44, August 2004, pp. 327-335,
 2T23. Novel testing and evaluation techniques 573

RADIOGRAPHIC OBSERVATION OF DAMAGE ZONE EVOLUTION IN HIGH

DUCTILE SPECIMEN

D. Vavrik1, T. Holy, J. Jakubek, M. Jakubek and Z. Vykydal

Institute of Experimental and Applied Physics of the Czech Technical University in Prague,
Horska 3a/22, 128 00, Prague 2, Czech Republic, E-mail: vavrik@itam.cas.cz
1Institute of Theoretical and Applied Mechanics of the Czech Academy of Sciences,

Prosecka 76, Prague 9, Czech Republic

This work reports on new results of the experimental observation of material damage evolution in
high ductile flat specimens manufactured from aluminium alloy. Failures in ductile materials
precede intensive internal material damage evolution. Not only damage existence but also its
quantification has to be determined for a numerical simulations purpose. An experimental method
called “X-Ray Dynamic Defectoscopy (XRDD)” was developed from this reason [1]. The test
sample is illuminated by X-rays during the loading process. Measured changes in transmission
represent alterations of effective thickness of the specimen. The effective thickness changes are
understood as weakening of the material due to damage volume fraction and transversal thickness
reduction (contraction) resulting from loading stress. Alterations in the sample thickness due to
void volume fraction are separated from the total thickness reduction using independent optical
measurement of the out-of plane displacement field by the photometric stereo method. We observe
time evolution of the damage zone shape and proportional volume fraction of voids (damage
intensity) by XRDD. In-plane strain field and consequent plastic strain intensity at a surface is
investigated by the optical Method of Interpolated Ellipses [2].
Failure of a high ductile specimen usually precedes a decrease of the loading force (material
softening) under the condition of displacement control in the monotonic loading. This
phenomenon is usually induced by a large plastic deformation and eventually by a related dynamic
development of voids and micro-cracks inside the specimen (damage process). The experimental
investigation of this highly interesting stage can be made accessible only under very well
controlled loading. Unfortunately, the elastic energy stored in the loading equipment can result in
sudden failure within very short time corresponding to the loading instability. We can follow this
time interval with real fast readout system or, alternatively, we can prolong this time interval by
reducing the elastic energy in the loading equipment. New designed loading frame fulfils the
indicated requirements and enables extraordinary precise displacement loading control even
beyond the maximal loading force. This transferable and highly stiff loading device is equipped by
four stepper engines ensuring symmetrical loading of the specimen with stable position of the
observed area in the X-ray beam. The loading force is recorded by a load cell and the prolongation
of the specimen is measured by an extensometer. Loading force capacity of the device is 100 kN
and weight only 25 kg, with dimensions 377x343x190 mm. These parameters allow to fix loading
equipment onto PC controlled motorized stage during experiment; see Fig. 1 for whole
experimental setup. It is possible to put loading frame together with the motorized stage into fully
shielded box. Therefore X-ray radiographic transmission and tomographic observation is easy
accessible.
574 D. Vavrik et al.

Fig. 1: Experimental setup. X-ray tube on the left, loading frame fixed on the loading stage in the
middle, X-ray detector Medipix on the right.

There are several principal requirements which are necessary to achieve the high resolution
X-ray imaging: high quality of the X-ray beam, precision positioning of the observed object and
high quality of the X-ray detector. As the X-ray sensor we are using the digital Medipix-2 [3]. The
X-ray generator Hamamatsu has X-ray emission spot of the diameter 5 Pm and divergent cone
beam. It allows magnification of factor 10 allowing resolution of the micrometer scale. Regarding
the precision positioning of the observed object, stepper engines are employed in the motorized
stage. This stage has two linear axes in a plane perpendicular to the beam direction and one
rotation around vertical axis. Linear axes allow scanning objects bigger than X-ray detector area.
Rotation allows computer tomography reconstruction.
Each active part of the experimental setup is operated using universal serial bus USB. Fully
operational and controllable are the motorized stage, the parameters and the exposition of the
X-ray source, the acquisition of the X-ray images, optional acquisition of the optical images, the
driving of the stepper motors of the loading frame, the data acquisition from the load cell and of the
extensometers. It is possible to operate and control all these processes using one notebook thanks
to this unification. In addition all these functions have automatically the same time base.

References
1. Vavrik, D.; Jakubek, J.; Pospisil, S.; Visschers, J. (2005), Materials Science Forum, Vol. 480,
pp. 231-236
2. Vavrik, D.; Zemankova, J.: Experimental Mechanics, Vol. 44, August 2004, pp. 327-335
3. Medipix collaboration: http://www.cern.ch/MEDIPIX/
 2T23. Novel testing and evaluation techniques 575

CALIBRATION OF FRACTURE PARAMETERS BY INSTRUMENTED

INDENTATION AND TEST SIMULATION

Massimiliano Bocciarelli, Gabriella Bolzon and Giulio Maier

Department of Structural Engineering, Politecnico di Milano
piazza Leonardo da Vinci 32, 20133 Milano, Italy
bocciarelli@stru.polimi.it, bolzon@stru.polimi.it, maier@stru.polimi.it

Surface engineering is an emerging discipline in technologies leading to products such as

electronic packages, magnetic recording media, optical devices and tribological protection of
mechanical components. The fruitful use of thin coatings in these applications require an accurate
characterisation of mechanical properties of these this layers after deposition, a process which can
alter their characteristics with respect to the original bulk state. Elastic moduli, yield strength,
fracture energy and interfacial adhesion play a crucial role on the correct functioning and long-
term performance of coatings, avoiding cracking or delamination from the substrate.
A spreading methodology for material characterisation is based on indentation tests. Fracture
induced in brittle materials by indenters with sharp corners and delamination between film and
substrate caused by indentation tests have been investigated by several researchers, e.g. Xiaodong
et al. [Xiaodong, L., Dongfeng, D. and Bhushan, B., Acta Mater., vol. 45, 4453-4461, 1997.],
Abdul-Baqi and Van der Giessen [Abdul-Baqi, A. and Van der Giessen, E., Int. J. Solids Struct.,
vol. 39, 1427-1442, 2002.], Li and Siegmund [Li, W. and Siegmund, T., Acta Mater., vol. 52,
2989-2999, 2004].
The model calibration technique proposed in Bolzon et al. [Bolzon, G., Maier, G., and Panico,
M., Int. J. Solids Struct., vol. 41, 2957-2975, 2004.] and Bocciarelli et al. [Bocciarelli, M., Bolzon,
G. and Maier, G., Mech. Mat., vol. 37, 855-868, 2005.] is based on indentation test, imprint
mapping and inverse analysis. This method can be applied to the identification of fracture material
parameters both for delamination in film-substrate systems and for the fracture characterisation of
ceramic materials at the micro-scale. The novelty is represented by the mapping of the residual
imprint (through atomic force microscope or laser profilometer, depending on the imprint size),
and by the use of these deformation measures, besides the traditional indentation curves, for the
parameter estimation through the minimisation of a function which represents the discrepancy
between the experimental measurements and the same measurable quantities obtained by the
numerical simulation of the experiment. Details can be found in Maier et al. [Maier, G.,
Bocciarelli, M., Bolzon, G. and Fedele, R., Int. J. Fracture, submitted, 2005.].
The test simulation has been performed by the finite element commercial code ABAQUS, using
the large plastic strain capability provided therein Fracture and delamination processes have been
simulated by means of interface elements, implemented in the code by the authors, endowed with
the cohesive-crack model originally proposed by Rose et al. [Rose, J.H., Ferrante, J. and Smith,
J.R., Physical Review, vol. 9, 675-678, 1981. ] and Xu and Needleman [Xu, X. P. and Needleman,
A., J. Mech. Phys. Solids, vol. 42, 1397-1434, 1994.].
The purpose of this communication is to present some recent or current developments and the
main results recently achieved by the authors in the growing area of fracture mechanics centred on
the calibration of cohesive fracture models for quasi-brittle materials, by approaches which
combine experimentation, experiment simulation and minimisation of the discrepancy between
measured and computed quantities.
576 M. Bocciarelli et al.

References
1. Xiaodong, L., Dongfeng, D. and Bhushan, B., Acta Mater., vol. 45, 4453-4461, 1997.
2. Abdul-Baqi, A. and Van der Giessen, E., Int. J. Solids Struct., vol. 39, 1427-1442, 2002.
3. Li, W. and Siegmund, T., Acta Mater., vol. 52, 2989-2999, 2004
4. Bolzon, G., Maier, G., and Panico, M., Int. J. Solids Struct., vol. 41, 2957-2975, 2004.
5. Bocciarelli, M., Bolzon, G. and Maier, G., Mech. Mat., vol. 37, 855-868, 2005.
6. Maier, G., Bocciarelli, M., Bolzon, G. and Fedele, R., Int. J. Fracture, submitted, 2005.
7. Rose, J.H., Ferrante, J. and Smith, J.R., Physical Review, vol. 9, 675-678, 1981.
8. Xu, X. P. and Needleman, A., J. Mech. Phys. Solids, vol. 42, 1397-1434, 1994.
 2T23. Novel testing and evaluation techniques 577

INTERNAL CRACK DETECTION AND ANALYSIS USING THERMOELASTIC

STRESS ANALYSIS

N. Sathon and J.M. Dulieu-Barton

School of Engineering Sciences, University of Southampton
Highfield, SO17 1JB, UK
moss@ship.soton.ac.uk, janice@ship.soton.ac.uk

The detection of damage at early stage is of prime important to initiate timely and cost effective
repair and maintenance. In a shell structures such as pipes or pressure vessels, cracks may initiate
at the internal surface and propagate through the thickness. The cracks are usually semi-elliptical
or semi-circular shape. These type of defects can be observed using traditional inspection
techniques such as ultrasound. The aim of the current research is to devise an approach of using
thermoelastic stress analysis (TSA) to detect, analyse and quantify the severity of these sub-surface
cracks in terms of the stress field surrounding the crack.
TSA is a non-contact technique concerned with the measurement of the surface temperature
change of a solid elastic body under dynamic loading. The technique is based on the thermoelastic
effect. The temperature change is measured using a highly sensitive infra-red detector. The
relationship between the temperature change and the stress change under adiabatic condition can
be expressed as [1]:

DT 3
'T 
Uc
¦ 'V
1 1
i (1)

where T is the local temperature change, is the coefficient of thermal expansion, T is the surface
temperature, is the density, c is the specific heat and delta sigma is the sum of the principal stress.
Primarily TSA is a surface technique and has been used extensively to evaluate the SIFs from
through cracks. In the late 1980s it was recognised that the technique can also be used to obtain
sub-surface stresses provided that phase of temperature oscillation is available [2]. Some initial
validations of this idea have been carried out experimental and by using numerical solutions [3].
The approach is based on the effect of heat conduction caused by the stress gradient and hence
thermal gradient (see equation 1). At a typical loading frequency for TSA the assumption of an
adiabatic state may not be valid around the region of damage such as a crack due to the very high
stress gradient which results in an inconsistency of thermal amplitude and phase. This non-
adiabatic behaviour is exploited in the present work to evaluate the severity of the sub-surface
crack.
This paper describes the behaviour of thermal response from the surface of solid body with
embedded multiple sub-surface flaws at different severity in terms of thermal amplitude and phase.
Finite Element modelling of the thermal conditions is carried out and the results from the
numerical simulation are then compared with the TSA to confirm the feasibility of using the
technique for internal crack severity evaluation.
578 N. Sathon and J.M. Dulieu-Barton

References
1. Dulieu-Barton, J.M. and P. Stanley, Development and applications of thermoelastic stress
analysis. Journal of strain analysis for engineering design, 1998. 33: p. 93--104.
2. McKelvie, J. Consideration of the surface temperature response to cyclic thermoelastic heat
generation. in Proc. 2nd Conf. on Stress Analysis by Thermoelastic Technique. 1987: SPIE.
3. Lesniak, J.R. Internal Stress Measurement. in Proc. 6th Congress on Experimental
Mechanics. 1988. Portland: SEM, BETHEL CT.
 2T23. Novel testing and evaluation techniques 579

ULTRAHIGH-RESOLUTION TRANSVERSAL POLARIZATION-SENSITIVE

OPTICAL COHERENCE TOMOGRAPHY: STRUCTURAL ANALYSIS AND
STRAIN-MAPPING

Karin Wiesauer, Michael Pircher1, Rainer Engelke2, Gisela Ahrens2, Gabi Grutzner2,
Reinhold Oster3, Christoph K. Hitzenberger1 and David Stifter
Upper Austrian Research GmbH, Hafenstr. 47 – 51, 4020 Linz, Austria
Karin.Wiesauer@uar.at
1Centre of Biomedical Engineering and Physics, Medical University, Währingerstr.13, 1090 Wien,

Austria
2micro resist technology GmbH, Koepenicker Str. 325, 12555 Berlin, Germany
3Eurocopter Deutschland GmbH, 81663 München, Germany

Optical coherence tomography (OCT), originally developed and so far nearly exclusively used for
biomedical applications (e.g., [1,2]), is a contact-free, non-destructive technique based on low-
coherence interferometry to image structures within translucent and turbid materials. Commonly,
cross-sectional reflectivity images with a depth-resolution determined by the coherence length of
the near-infrared light source are obtained. When OCT is performed in a polarization sensitive way
(PS-OCT), additional information about birefringence within a material is obtained by mapping
the retardation between ordinary and extraordinary rays [3]. Because birefringence is induced
when strain occurs, PS-OCT provides depth resolved information about the internal stress within a
sample.
In this work, we apply PS-OCT for material analysis. The materials investigated, mainly
plastics, polymers and compound materials, show feature sizes often in the range of only a few
microns (e.g., diameter of glass-fibres, thickness of layers). The depth-resolution of 10 – 20µm,
provided by superluminescence diodes commonly used as light sources, is too low for these
applications. Using a femto-second (fs-)Ti:sapphire laser, we obtain ultra-high depth-resolution of
around 2µm for typical plastic materials [4]. For the first time, ultra-high resolution imaging is
combined with transversal en-face) scanning PS-OCT, where images at a defined depth parallel to
the sample surface are obtained. With our setup, we are able to scan areas as large as 3x3mm2
within seconds, and we can switch immediately from transversal to cross-sectional scanning [4].
We demonstrate the advantages of transversal scanning for two different types of samples: a
helicopter rotor blade showing cracks due to loading tests, and high-aspect ratio moulds in thick
photoresist layers for micro electromechanical parts (MEMS) where the resist-wafer interface was
investigated. Figure 1a) shows a cross-sectional reflectivity image of the rotor blade consisting of a
glass-fibre – epoxy resin laminate. Beneath a fibre-free top layer, the first fibre bundles are
observed a depth of around 700µm. However, the crack running perpendicularly to the scanning
plane (arrow) is hardly visible. In contrast, the en-face reflectivity image (Fig. 1b)) taken at a depth
as marked in a), clearly reveals the crack running horizontally across the crossing fibre bundles.
580 K. Wiesauer et al.

FIGURE 1. a) Cross-sectional and b) en-face reflectivity scans of a helicopter rotor blade with a
crack, indicated by the arrows. Dotted lines: Positions of cross-sectional and en-face scans.

In Fig. 2a), a cross-sectional reflectivity image of the photoresist mould on a gold coated wafer
is displayed. Besides the resist layer thickness of 1.3mm, nearly no information about the structure
itself is obtained. In contrast, the en-face reflectivity scan taken at the optical depth of the interface
reveals the geometrical wheel structure (Fig. 2b)). In that case, a faulty wafer was investigated. The
defect structures (ridges at the rear resist surface, indicated by the arrows) were detected contact-
less and non-destructively with the moulds still on the wafer. On the other hand, retardation images
taken at the interface level of a faultless mould show the interfacial strain distribution (Fig. 2c)).
The areas around the teeth of the wheels are clearly highly strained, and increased strain occurs in
between adjacent wheels. Images like these add valuable information for the design and quality
control of MEMS moulds, because areas with excessive stress where defects may occur are
revealed.

FIGURE 2. Photoresist moulds on a gold coated wafer. a) Cross-sectional reflectivity scan with
schematic drawing showing the different optical levels. b) En-face reflectivity scan at the interface
of a faulty mould. The arrows indicate some of the defect structures. c) Interfacial en-face
retardation scan of a good mould, showing the strain distribution (detail shown in the inset).

References
1. Huang, D., Swanson E. A., Lin C. P., Schuhman J. S., Stinson W. G., Chang W., Hee M. R.,
Flotte T., Gregory K., Puliafito C. A. and Fujimoto, J. G., Science 254, 1178, 1991.
2. Fercher, A. F., Hitzenberger C. K., Drexler W., Kamp G. and Sattmann, H., Am. J.
Ophthalmol. 116, 113, 1993.
3. Hitzenberger, C. K., Götzinger E., Sticker M., Pircher M. and Fercher, A. F., Opt. Express 9,
780, 2001.
4. Wiesauer, K., Stifter D., Pircher M., Götzinger E., Hitzenberger C. K., Bauer S., Engelke R.,
Ahrens G. and Grützner, G., Opt. Express 13, 1015, 2005.
 2T23. Novel testing and evaluation techniques 581

APPLICATION OF DIGITAL SHEAROGRAPHY IN DETERMINING OPENING

MODE SIF IN EDGE CRACKS

M. Ghassemieh, A. Ghazavizadeh1 and N. Soltani1

Civil Engineering Department
1Intelligence Based Experimental Mechanics, Mechanical Engineering Department

The University of Tehran, Iran

mghassem@ut.ac.ir, akbar_ut@yahoo.com, nsoltani@ut.ac.ir

In this paper the non-destructive technique of digital shearography is utilized in order to obtain
stress intensity factor (opening mode) in edge cracks.
Being a relatively new non-destructive evaluation technique, digital shearography was
developed to eliminate some of the shortcomings of holography and has proven to be a powerful
method to measure in-plane and out-of-plane strains. Digital shearography has several advantages
over other optical methods such as simple setup, insensitivity to environmental stability, not
requiring any special device for vibration isolation and yielding out-of-plane strain components
directly. Its advantages have made it suitable for industrial environments and research centers (e.g.
Steinchen et al. [1] and Hung [2]).
In digital shearography, by digital subtraction of before and after loading speckle patterns
recorded by a digital CCD-camera, shearograms are obtained which dark fringes represent
contours of equal in-plane or out-of-plane deformation partial derivatives. Through combination of
linear elastic fracture mechanics and optical (digital shearography) relations, first mode stress
intensity factor of cracks and first partial derivatives of out-of-plane deformation are correlated. In
order To minimize random experimental errors, information for several points are registered and
least square method is used to obtain stress intensity factor.
Fig.1 presents the proposed scheme of the test specimens. Cracks normal to the length are
formed in the middle of the specimen length and placed at the edge. The geometric specifications
L
of specimens are detailed in Table (1). Making 4 creates a uniform tensile stress
W

FIGURE 1. Proposed scheme of the test specimens

distribution in the specimen. The cracks were blunt, not sharp which may be a source of error in
comparison with empirical values.
582 M. Ghassemieh et al.

Table 1.Geometric specifications of specimens

According to handbooks of crack stress analysis,(e.g. Tada [3]), the stress intensity factor in
the opening mode for edge cracks is as follows

K Yı ʌa
I (1)
where
2 3 4
§ a · § a · § a · § a ·
Y 1.122  0.231¨ ¸  10.55 ¨ ¸  21.71 ¨ ¸  30.382 ¨ ¸
©W¹ ©W¹ ©W¹ ©W¹ (2)

The above equations are used in order to compare and verify the results obtained from
shearography tests. From Table (2) it can be concluded that the experimental results are in good
agreement with empirical values, thus proving the potentials of this non-destructive technique in
measurement applications as well as qualitative evaluations.

Table 2. Empirical and experimental SIFs for different specimens and loadings

References:
1. Steinchen, W., Yang, L.X., Kupfer, G., Mäckel, P., and Vössing, F., J. Aerospace Eng., vol.
212, 21-30, 1998.
2. Hung, Y.Y., J. Nondestructive Testing, vol. 8, 55-67, 1992.
3. Tada, H., Paris, P.C., and Irwin, G.R., The stress analysis of cracks handbook, 3rd edition,
ASME Press, 2000.
 2T23. Novel testing and evaluation techniques 583

FINITE ELEMENT MODELING OF PULSE TRANSIENT IR THERMOGRAPHY

M. Krishnapillai, R. Jones, I. H. Marshall, M. Bannister1 and N. Rajic2

Mechanical Engineering, Monash University, Clayton, VIC, Australia
1Cooperative Research Centre – Advanced Composite Structures, 506 Lorimer Street, Fishermans

Bend, VIC, Australia.

2DSTO-Platforms Sciences Laboratory, 506 Lorimer Street, Fishermans Bend, VIC, Australia

mukunthan.krishnapillai@eng.monash.edu.au

Given the life extension program being currently implemented in both the military and commercial
aircraft regime the increased need to operate aircraft well beyond the original design life is
becoming more prevalent with each life cycle that is passed. Nondestructive testing is critical for a
safe aircraft life extension program. Infrared thermography is a non-contact, non-destructive
examination technique that utilizes radiated electromagnetic energy to acquire subsurface defect
information[1]. The material within which these defects exist, and upon which IRT can be applied,
range from concrete to metals to the focal material of this research, carbon fibre reinforced plastics
(CFRP). As a nondestructive evaluation technique thermography is gaining acceptance due
primarily to the recent advancements in the quality of thermal data acquisition systems as well as
the interest in broad area scanning procedures coupled with the potential for a truly non contact
NDE system that can handle inspections of intricately complex structures, Maldague [1].
To date there has been considerable work done concerning the use of finite difference models,
both 2D, Perez [2] and 3D, Plotnikov [3], to simulate the thermographic process. But the limits of
these methods are such that to create a simulated structure, differential equations are required to be
derived and manipulated to suit [2]. And in the case of a non-axisymmetric problem of complex
geometry this method becomes unfeasibly unrealistic in shortening the time of analysis and
inspection.
The use then, of IRT as a nondestructive evaluation technique is becoming increasingly
attractive in the detection of sub-surface defects in composite structures due primarily to its
potential ability to predict subsurface information given a suitable calibration template. To date,
extensive in-field damage data acquisition studies have been performed, the results of which
provide extensive qualitative results. Yet given this establishment a restriction to damage
prediction applies when considering a more complex geometry upon which experimental data has
not been collected or in collection would lead prediction redundant. This study reports on the use
of numerical FE models as a flexible tool to create and simulate the thermographic process. It
looks at the response of an FE model to predict damage parameters with the intention of
determining numerical relationships and modeling methods that may be used upon more complex
non-axisymmetric geometry.
As a result of this study a suitably calibrated finite element simulation technique has been
generated and validated (an example of a typical FE model with thermal response can be seen in
Fig. 1). This validation technique has also yielded its application towards modelling upon what can
be considered the epitome of a complex component, the curved structure. What makes a part
complex, in the thermal sense, is the presence of curved sections to which typical thermal
mathematical relationships are extraordinarily difficult to apply. By establishing the relationships
between thermal surface response, defect depth and location in a curved section we are able to
demonstrate the fundamental advantages that pulse transient IR thermography offers.
584 M. Krishnapillai et al.

FIGURE 1. Flat Plate Numerical FE Model and Thermal Response

References
1. Maldague XPV, Moore PO, editors. Infrared and thermal testing: Non-destructive Testing
Handbook. Vol. 3, American Society for Nondestructive Testing, 2001.
2. Perez I, Kulowitch P and Davis W. Modeling of pulsed thermography in anisotropic media.
Naval Air Warfare Center Aircraft Division Technical Report NAWCADPAX--99-38-TR,
1999.
3. Plotnikov YA. Modeling of the multi-parameter inverse task of transient thermography. 25th
Annual Review of Progress in Quantitative Non-destructive Evaluation Conference,
Snowbird, Utah. 1998.
 2T23. Novel testing and evaluation techniques 585

A NEW TECHNIQUE FOR THE MACHINING OF NATURAL CRACKS

N. P. Andrianopoulos and A. Pikrakis

Department of Mechanics, Faculty of Applied Sciences, National Technical University of Athens,
GR – 157 73, Hellas Tel.: +30 210 7721222
nandrian@central.ntua.gr

Experimental Fracture Mechanics is based on the study of specimens containing “machined-

natural” cracks, a term which is self-contradicting. In practice, an artificial (machined) notch of
considerable thickness is opened in the specimen by using various techniques and, then, a “natural”
crack extension of the notch is created through fatigue. This procedure is, in details, described by
ASTM standards [1], where an artificial notch is considered as acceptable when, among other
restrictions, it must have a radius of curvature of the tip equal to ~0.25X10-3m and the through-
fatigue crack extension has a quiet uncontrolled length of (1-2)X10-3m. Geometrically, this
“natural” crack satisfies two of the requirements of a slit to be a natural crack, i.e. no loss of mass
and small (but not near- zero) tip-radius. In many cases, it is not straight or forms a small angle
with the notch axis. In addition, this method cannot be applied in case of notches inclined to the
loading axis and special loading apparatuses are required. To facilitate and improve ASTM
standard many attempts have been presented (e.g. [2, 3]). However, an important side-effect of
this method has not been discussed. It is that fatigue causes changes of unpredictable severity in
the mechanical properties of the material, exactly in the tip area, where the final crack is expected
to initiate. In this area material fails and around the tip of the “natural” crack plasticized zones and
hardening processes are observed. Consequently, fracture criteria are based on a slippery ground.
In the present work a new technique of crack machining is proposed in order to avoid some
of these difficulties. For that, a steel mould is constructed (Fig.1). It consists of

Fig.1. A sketch of the mould. The fixing guide is not shown.

four identical plane plates and a guide to keep them firmly in the proper position. Between the two
plates of each face of the mould a thin sheet (thickness <20X10-6 m) of bronze is placed under a
small tensile load, in order to be plane. The width of the sheet of bronze is equal to the final crack
length. Then, the two pairs of the mould plates are put close to each other and the mould is fixed.
The material, under study, is melted and is cast into the mould. Finally, a plane specimen is
obtained with a straight crack of the desired length and
586 N. P. Andrianopoulos and A. Pikrakis

Figure 2. The crack-tip area in: (a) pre-fatigued and (b) cast specimen.

with radius of curvature equal to ~10-4m, i.e. at least, one order of magnitude smaller than that of
the ASTM standard.
This method was applied in case of an aluminum alloy with satisfactory results. In order to
compare the mechanical properties of the material, specimens pre-cracked according to the ASTM
technique and to the cast method, identical to their other respects, were submitted to micro-Vickers
hardness tests with load equal to 490.3mN. Three series of such tests were performed: One at the
close vicinity of the crack tip of the pre-fatigued specimen, one at the same area of the cast
specimen and the last one in remote areas of both specimens for reference reasons. Since hardness
number is an index of the strength of the material, it can serve, also, as a qualitative index of the
local changes of its mechanical properties due to fatigue. The results, obtained, give the same
hardness number for the last two series of experiments, i.e. cast specimens seem to keep the same
mechanical properties near and far the crack-tip. On the contrary, measurable deviations were
obtained from the near-tip area of the pre-fatigued specimens (see Table 1).

TABLE 1. Hardness data

The results indicate that hardening mechanisms activated in the tip-area of the pre-fatigued
specimens, caused considerable change in the hardness number of the order of (40.4-36.8)/
36.810%, which might be crucial in Experimental Fracture Mechanics.

References
1. ASTM, Standard E647-88a, Standard test method for measurement of fatigue crack
growth levels, pp. 646-665.
2. Tong, J., Yates J.R. and Brown, M.W., Fatigue Fract. Engng. Mater. Struct., vol. 17, 1261-
1269, 1994.
3. Scibetta, M., Lucon, E., Van Walle, E. and Valo, M., Int. J. of Fracture, vol. 117, 287-296,
2002.
 2T23. Novel testing and evaluation techniques 587

DISPLACEMENTS MEASUREMENT IN IRREGULARLY BOUNDED PLATES

USING MESH FREE METHODS

N. P. Andrianopoulos and A. P. Iliopoulos

Department of Mechanics, Faculty of Applied Sciences, National Technical University of Athens,
GR – 157 73, Hellas Tel.: +30 210 7721222
nandrian@central.ntua.gr

In the present work a combined experimental-numerical technique for the determination of the
displacement and strain fields in the surface of a deformable body with irregular boundaries is
described. The experimental part of this technique is based on the recording of the pair of images
of the surface before and after deformation of the body, by means of a digital camera. It is a new
approach for the pure grid method [1] since the surface is randomly gridded with a simple
permanent marker, while in pure grid method the grid has to be regular (Fig. 1).

FIGURE 1. A specimen with a central hole and a random grid of spots.

The boundaries of irregularities are first determined, by a channel extraction method (in this
case the red channel). The original color images are then converted to black and white bitmaps and
the spots are identified by an appropriate algorithm. By correlating the spots between the images,
we can easily determine the displacement of each of the nodes. In mesh free [2] methods a
component of the displacement field (say the x-displacement- u ) at any point x ( x , y ) is
approximated using the displacements at a number of nodes around it. These nodes constitute the
support domain of point at x , i.e.
n
T
uh x  ¦ I  x u
i 1
i i ĭ  x  U S with US ^u1 , u2 ,..., un ` (1)

In these equations, n is the number of nodes in the support domain of the point at x ,
ui is the component of the field displacement of the ith node in the support domain, U S is the

vector containing all the displacement variables at these nodes, iji(x) is the shape function of the

ith node and ĭ (x) is the matrix that collects all the shape functions values.
The strain at every point in the region of interest can then be computed by [1]:
588 N. P. Andrianopoulos and A. P. Iliopoulos

ªIi , x 0 º
« »
n īi « 0 Ii , y »
İh ¦ī u i i where «I » (2)
i
¬ i , y Ii , x ¼
In the case of irregular boundaries (i.e. holes, cracks), the shape function has to be re-formed in
order to approximate the boundaries [3, 4]. The visibility method is mostly used when dealing with
convex irregularities. The diffraction method is more appropriate when the body has linear
irregularities (i.e. cracks).
In the present paper we investigate the applicability of the method in two types of experiments,
i.e. uni-axial tension of a specimen with a hole and a specimen with a crack. In Fig. 2 we present
the the y component of the strain in a uni-axial tension experiment with a central hole.

FIGURE 2. Distribution of ey strain component on a specimen with a central hole under uni-
axial tension

References
1. Sevenhuijsen, P.J., Optics and Lasers in Engineering, vol. 18, 173-194, 1993.
2. Belytschko, T., Lu, Y.Y. and Gu, L., Int. J. Numer. Methods Eng., vol. 37, 229-256, 1994.
3. Belytschko, T., Fleming, M., Terry, T., J. Comput. Appl. Math., vol. 74, 111-126,1996.
4. Belytschko, T., Krongauz, Y., Organ, D., Fleming, M. and Krysl, P., Comput. Methods Appl.
Mech. Eng., vol. 139, 3-47, 1996.
 2T23. Novel testing and evaluation techniques 589

BIAXIAL STRENGTH TESTING ON MINI SPECIMENS

Robert Danzer, Peter Supancic1, Walter Harrer, T. Lube and Andreas Borger2
Institut fur Struktur- und Funktionskeramik, Montanuniversitat Leoben
Peter Tunner Straße 5, A-8700 Leoben, Austria
1Materials Center Leoben, A-8700, Leoben, Austria
2now at Treibacher Industrie AG, A-9330, Altrofen, Austria

isfk@unileoben.ac.at

Biaxial strength testing of brittle materials using the ball on three balls (B3B) test is a new method
for strength testing of disc or plate specimens [1] (see Fig. 1). The analysis of the stress fields and
the calculation of the effective volumes and surfaces for several types of specimens and testing
geometries has been performed recently. The ball on three balls testing method has several
advantages compared to the common three- or four point bending tests: the results are very stable
against small geometrical inaccuracies of the specimen or the test assembly, edge defects are not
relevant and there exists only a very low influence of friction [2]. This makes this type of
experiment ideal for testing very small specimens [3].

FIGURE 1. B3B sketch and corresponding stress field in the specimen.

In this paper ball on three ball tests were performed on mini specimens having a volume of
several cubic millimetres and less made of several types of advanced ceramic materials. The test
results are compared with results gained on larger ball on three ball specimens and on conventional
bending specimens. The results are discussed in the framework of the Weibull theory [4,5].
590 R. Danzer et al.

References
1. Morell, R., McCormick, N. J., Bevan, J., Lodeiro, M., Margetson, J., Brit. Ceram. Trans., vol.
98, 234-240, 1999Jones, G. A., Applied Fracture Mechanics, Soringer, Berlin, Germany,
1978.
2. Borger, A., Danzer, R., Supancic, P., J. Eur. Ceram. Soc., vol. 22, 1425-1436, 2002.
3. Lube, T., Manner, M., Danzer, R., Fatigue Fract. Eng. Mater. Struct., vol. 20, 1605-1616,
1997.
4. Weibull, W., J. Appl. Mech., vol. 18, 293-297, 1951
5. Danzer, R., J. Eur. Ceram. Soc., vol. 10, 461-472, 1993
 2T23. Novel testing and evaluation techniques 591

NUMERICAL SIMULATION OF A FRACTURE TEST FOR BRITTLE

DISORDERED MATERIALS

T. Auer and H. Harmuth

Christian Doppler Laboratory for Building Materials with Optimized Properties,
University of Leoben, A-8700 Leoben, Austria
thomas.auer@unileoben.ac.at, ha@unileoben.ac.at

For the characterisation of brittle disordered materials like building materials and refractories
fracture mechanical parameters are of great importance. For this purpose in the past a wedge
splitting test according to Tschegg [1] (Fig. 1) has established as a feasible testing method. This
test yields a specific fracture energy Gf (eq.(1)) and a nominal notch tensile strength VNT (eq.(2)).

FIGURE 1. Schematic representation of the testing procedure and the specimen shape (measures in
mm). Here FM, FH and GH are the machine force, the horizontal force
and the horizontal displacement, respectively.

G
1 u
˜ F dG
A ³0
Gf
(1)

F H max 6 ˜ F H max ˜ y
V NT 
b˜h b ˜h2 (2)
Here b and h are the width and the height of the ligament, respectively. A represents the
(projected) fracture surface area. The distance from the load point to the center of gravity of the
ligament is denoted by y. As a rule the test is performed until a minimum testing load is achieved;
therefore the specific fracture energy will be slightly larger compared with equation (2). The goal
of the numerical simulation of the testing procedure was to calculate the true specific fracture
energy, the tensile strength and Young´s modulus from the test results. For this purpose a finite
element simulation applying a fictitious crack model according to Hillerborg [2] was applied. A
strain softening behaviour according to Hordijk [3] was chosen (Fig. 2). From various simulations
mathematical relations have been derived to calculate the specific fracture energy, the tensile
strength and Young´s modulus from characteristics of the load/displacement curve. To give one
example for the results, the ratio of the notch tensile strength to the tensile strength is shown in Fig.
3 for the specimen shape of Fig. 1. Here the brittleness number B is used, which is defined by
equation (3):
592 T. Auer and H. Harmuth

FIGURE 2. Fictitious crack model according to Hillerborg, ft is the tensile strength, H the elastic
strain and x is the crack opening displacement, x2 is the ultimate crack opening displacement for
cohesive forces.

ft 2 ˜ h
B
Gf ˜E
(3)
In equation (3) ft is the tensile strength and E the Young’s modulus.

FIGURE 3. Dependence of the ratio VNT/ft on the brittleness number B.

With this approach it is possible to improve the existing evaluation of a wedge splitting test.
Important material parameters like tensile strength, specific fracture energy and Young’s modulus
can be determined with this procedure. Furthermore the whole load/displacement curve can be
plotted for any combination of material parameters without any further simulation.

References
1. Tschegg, E.K., Testing device and appropriate specimen shapes for tests to measure fracture
values, Austrian Patent Specification, AT 390 328 1-3, 1990.
2. Hillerborg, A., Fracture Mechanics of Concrete, Elsevier Science Publisher, 223-249, 1983.
3. Hordijk, D. A., Local approach to fatigue of concrete, Meinema, Delft, Netherlands, 1991.
 2T26. Structural integrity 593

UNIFICATION OF THE OUT-OF-PLANE CONSTRAINT LOSS IN CENTRE-

CRACKED PANELS

B. Bezensek, A. Baron1 and J. W. Hancock1

University of Maribor, Faculty of Mechanical Engineering,
Smetanova 17, SI-2000 Maribor, SLO
1University of Glasgow, Mechanical Engineering Department, Glasgow G12 8QQ, UK

b.bezensek@eng.gla.ac.uk, annabaron@fsmail.net, j.hancock@eng.gla.ac.uk

Within the scope of constraint based fracture mechanics the near tip fields are characterised with
two parameters that capture the level of crack tip loading and constraint (Betegón and Hancock [1],
O’Dowd and Shih [2]). Constraint arguments have been developed under plane strain conditions,
which assume infinite thickness and restrain deformation in the lateral direction. Such conditions
promote the development of high stress triaxialities at the crack tip and are necessarily
conservative. However most engineering structures are of finite thickness, many of which operate
on the upper shelf. Fracture resistance under such conditions is expected to benefit from lateral
deformation, which may render the assumed plane strain conditions over-conservative.
The three-dimensional nature of the near crack-tip field in finite thickness plates under elastic-
plastic conditions has been examined on a centre-cracked panel geometry. Mean and direct stresses
have been determined on the mid-plane and through the thickness, using a three-dimensional finite
element model.
The proximity to plane strain conditions on the mid-plane of real plates has been determined
by evaluating the mean stress, Vm, and the out-of-plane deviator defined as V z /( V x  V y ) . By
comparison with the reference plane strain solution under equal remotely applied loads, it has been
shown that plane strain conditions are maintained in relatively thin plates up to loads approaching
half the limit load.
SSY
A constraint parameter, (V m  V m ) / V o , was defined for real plates in an analogous
manner to the Q parameter defined for plane strain conditions [2]. A good agreement with the two-
dimensional solutions was observed under low load levels, indicating that at low loads the
conditions in real plates are close to those described by plane strain analyses. With increasing load
levels the mean stress in real plates reduces more rapidly than in corresponding plane strain model
under equal loading. Two-dimensional plane strain analysis incorporating the in-plane constraint
loss therefore provides a conservative estimation of the material resistance to fracture. In real
geometries additional margins on load can be invoked due to loss of out-of-plane constraint.
The mean stress ahead of a crack was written as a series expansion in which the difference
field was separated into the in-plane and out-of-plane contributions to the total mean stress (Baron
[3]):

V m3 D V mSSY  Q V 0  ' V mop (1)

where V m3D is the mean stress obtained from the finite thickness model , V mSSY is the small scale
yielding reference solution, Q V 0 is the in-plane constraint loss and ' V
op
m is constraint loss due
op
to out-of-plane deformation. The values of ' V m depend on loading and thickness of the plate.
op
Significantly, it was shown that all values of ' V m can be described by a single consistent curve
594 B. Bezensek et al.

which is independent of loading and thickness. The implications for the fracture assessments using
failure assessment diagrams are discussed.

FIGURE 1: The reduction of mean stress due to out-of-plane constraint loss in real plates.

References
1. Betegón C., Hancock J.W., ‘Two-Parameter Characterisation of Elastic-Plastic Crack-Tip
Fields’, Transactions of the ASME, vol. 58, 104-110, 1991
2. O’Dowd N.P.,Shih C.F., Journal of the Mechanics and Physics of Solids, vol. 39, 989-1015,
1991
3. Baron A.K., Determination of Out-of-Plane Constraint in Centre-Cracked Panels, Final Year
Project, University of Glasgow, Glasgow, 2005
 2T26. Structural integrity 595

HIGH TEMPERATURE FAILURE ASSESSMENT OF WELDMENTS

B. Dogan, B. Petrovski and U. Ceyhan

GKSS Research Centre, D-21502 Geesthacht, Germany
bilal.dogan@gkss.de

The integrity and residual life assessment of high temperature structures rely on the determined
defects in components under service loading conditions. Attention has been paid to development of
tools to detect flows in service-components (e.g.NDE). Furthermore, significance and
determination of crack initiation and growth behavior using metallography and fracture
mechanical (FM) testing have been addressed. The operational and plant assessment experience
indicates that in the majority of cases where high temperature failure occurs, defects predominate
in the vicinity of weldments. The concepts used for time dependent fracture analysis of
homogeneous bodies are commonly applied to high temperature crack growth in weldments that
show multi-crack initiation and crack branching (1,2). This calls for study of deformation behavior
and applicability of fracture parameters for high temperature assessment of weldments.
Most of the data obtained to date on high temperature materials has been reported on creep
crack growth (CCG), ignoring the creep crack initiation (CCI). To this end, the only available
testing standard for Measurement of Creep Crack Growth Rates in Metals, ASTM E1457-00, does
not address the CCI and testing of weldments. Recent European collaborative effort (e.g. EC
projects CRETE, ECCC, FITNET, and ESIS TC11, WG on High Temperature Testing of Welds)
concentrate on CCI and CCG testing and analysis of industrial specimens and assessment that will
contribute to development of European Codes of Practices (CoP) (3).
The present paper reports on the high temperature deformation and determination of CCI and
CCG using fracture mechanics specimens and its significance in defect assessment of components.
The deformation and crack growth behavior of similar welds of P22 and high strength martensitic
steel Grade P91 steels, are studied at 550°C and 600°C, respectively. The weldments produced
with a spectrum of industrially relevant properties produced in butt welded pipe joint. Therefore,
the studied material cover a spectrum of microstructures and ductility to give representative data
that applies to a range of weld types. The crack growth tests were carried out on compact C(T) type
and C-shape CS(T) specimens, under constant loads and constant displacement rates. The
emphasis is placed on the applicability of available test procedures, loading modes and the crack
tip parameters in assessing crack initiation and growth. It is aimed to contribute establishing
guidelines for acceptable methodologies for testing, analysis and assessment of weldments using
European methods, Time Dependent FAD (TDFAD) and Two Criteria Diagram (2CD), at high
temperatures. The TDFAD method is applied to P91 base and similar weld data as shown in
Figures 1 and 2. TDFAD have been considered for various times of 100 to 100,000 h. Kr and Lr
are calculated for crack initiation times for 'a=0.2 mm and 0.5 mm for experimental specimens.
596 B. Dogan et al.

Figure 1 – TDFAD Diagram of P91 BM at 600 °C with experimental data for crack growth 'a=0.2
mm (solid symbols) and 0.5 mm (hollow symbols).

Figure 2. - TDFAD Diagram of P91 WM at 600 °C with experimental data for crack growth
'a=0.2 mm (solid symbols) and 0.5 mm (hollow symbols).

References
1. B.Dogan, "High Temperature Defect Assessment Procedures", Int.Journal of Pressure
Vessels and Piping, Vol.80, 2003, pp.149-156.
2. B.Dogan, "High Temperature Design Approaches for Welds", Proc. 2nd Int. Conf. Integrity
of High Temperature Welds, IOM communications ltd, Ed. T.H. Hyde, 2003, pp.461-473.
3. B.Dogan, K.Nikbin and B.Petrovski, "Development of European Creep Crack Growth
Testing Code of Practice for Industrial Specimens", Proc. EPRI Int. Conf. Materials and
corrosion experience for fossil power plants, Nov.18-21, 2003, Isle of Palm, SC, USA.
 2T26. Structural integrity 597

POST-TENSIONED GLASS BEAMS

C. Louter, J. van Heusden1, F. Veer, J.Vambersky1, H. de Boer1 and J. Versteegen2

Faculty of Architecture, Delft University of Technology
1Faculty of Civil Engineering, Delft University of Technology
2Pieters Bouwtechniek

Po Box 5043, 2600 GA Delft

p.c.louter@bk.tudelft.nl

At Delft University, glass researchers have developed a revolutionary safety concept for glass
beams. This safety concept shows some analogy with reinforced concrete; glass beams are
reinforced and/or post-tensioned by adding (stainless) steel to the layout of the beam.
Prototypes of 3 and 7.2 m length have been built and tested in 4-point bending tests. The
results show that cracked and failed beams show significant post initial crack strength, see figure 1.
This has been described by Bos et al. in [1], and Louter et al. in [2].

Figure 1: Left: cracked reinforced glass box girder

This paper will focus on post-tensioned glass beams. The crucial aspect of post-tensioning
glass beams is the transmission of the forces from the post-tensioning rod into the head of the glass
beam. Due to the inability of glass to deform plastically, glass is very susceptible to local peak
stresses. Therefore the transmission has to be gradual and peak stresses have to be avoided.
Research has focused on the transmission of the forces at the head of the glass beams. Four
different methods to apply the post-tensioning forces on the head of a glass beam have been
developed (see figure 2), analyzed with Finite Element Models (see figure 3) and tested in
compression tests.
The results show that micro-cracks at the glass surface were not as critical as expected and
prestress levels of 80 MPa can be reached. Assembly methods and tolerances turned out to be the
most crucial aspects for post-tensioned glass beams.
598 C. Louter et al.

Figure 2: Developed methods to apply the forces on the head of the beam

Figure 3: Finite Element Models of glass beam head.

References
1. Bos, FP, Veer, FA, Hobbelman, GJ & Louter, PC(2004). Stainless steel reinforced and post-
tensioned glass beams. In Pappalettere, C (Ed.), International conference on experimental
mechanics / icem12 / advances in experimental mechanics. (pp. #1-#9). Bari: Politecnico di
Bari.
2. Christian Louter, Jan Belis, Freek Bos, Fred Veer, Gerrie Hobbelman. Reinforced glass
cantilever beams,. In Proceedings Glass processing Days 2005, Tampere, Finland, Tamglass
ltd.Oy.
 2T26. Structural integrity 599

STRUCTURAL INTEGRITY OF A NPP USING THE MASTER CURVE

APPROACH

D. Ferreno, I. Gorrochategui, M. Scibetta1, R. Lacalle, E.van Walle1 and F. Gutierrez-Solana

University of Cantabria
ETS Ingenieros de Caminos, Av/Los Castros s/n, 39005, Santander, Spain
1SCK-CEN

Boeretang 200, 2400 Mol, Belgium

ferrenod@unican.es

In this paper it is described the work performed on the steel coming from a Spanish BWR Nuclear
Power Plant (NPP) vessel currently operating in order to analyse its structural integrity assessment
from different procedures. The classical procedures for the characterization of the cuasistatic
fracture behaviour in the ductile to brittle transition region [1] are based on conservative
correlations obtained from Charpy impact tests. In this work, on the contrary, direct measurements
of cuasistatic KJC fracture toughness were performed. The results were analysed according to the
recent Master Curve methodology [2] in order to obtain the Reference Temperature, T0, describing
fairly the response of the material in such a region.
The experimental scope of the project was focused as an attempt to satisfy the next goals: to
characterise the different materials which constitute the vessel, base and weld materials, both non-
irradiated and irradiated conditions, evaluating in this way the effect of the irradiation induced
damage on the Reference Temperature; to study the effect of the specimen geometry comparing
results coming from PCCv and 1TCT specimens; to validate the process of reconstitution of 1TCT
specimens with electron beam welding and PCCv, 10 mm implant, with arc stud welding. In all the
cases several orientations were analysed. Additionally a study of the behaviour of the base material
in dynamic regime, in the transition region, by means of instrumented Charpy impact test on
precracked specimens was performed.
The obtained results were applied in the Structural Integrity assessment of the vessel under
study. It was focused in obtaining the operation limit curves, P-T curves, for the different
normalised scenarios. For this task it has been used, first of all, the procedure described in the
American Standard Design Code [1], currently valid for the Spanish NPPs designed in the United
States.
The development of the Master Curve methodology enables the resolution of this problem in
alternative ways. Thus, the P-T curves were also calculated following the indications of the N629
ASME Code Case [3] which partially includes the principles of the Master Curve. Moreover, in an
attempt to evaluate the conservatism of the different classical methodologies, the mentioned P-T
curves were determined incorporating the fully Master Curve approach in the ASME method and,
finally, according to the SINTAP European Structural Integrity Procedure [4] used now in the new
FITNET [5] procedure. The four alternatives have been compared in order to determine the level
of conservatism as a function of the refinement of the data used.
600 D. Ferreno et al.

References
1. ASME Boiler and Pressure Vessel Code Section XI, Apéndice G.
2. ASTM E1921-03: “Test Method for the Determination of Reference Temperature T0 for
Ferritic Steels in the Transition Range”, Annual Book of ASTM Standards, Section 3, Vol
03.02.
3. ASME Boiler and Pressure Vessel Code – Code Case N-629, “Use of Fracture Toughness
Test Data to Establish Reference Temperature for Pressure Retaining Materials”, Section III
and XI, Division 1,1999.
4. SINTAP, Structural Integrity Assessment Procedures For European Industry, Final
Procedure, November 1999.
5. FITNET, Fitnet For Service Procedure, Draft Working Document, February 2005.
 2T26. Structural integrity 601

FRP CONSOLIDATION FOR MASONRY ARCHES BY USING BRIDGED

CRACK MODEL

G. Ferro, M. Ipperico, V. Pignata and A. Carpinteri

Department of Structural and Geotechnical Engineering, Politecnico di Torino,
Torino, Italy
ferro@polito.it

This paper applies the Fracture Mechanics approach to the study of masonry arches consolidated
by means of FRP. The softening and fracturing process in shallow masonry arches is considered at
first, assuming a purely elastic constitutive law for the material coupled with a fracture crisis
condition.
The discrete model for a cracked beam element with only one reinforcement proposed by the
first author for the analysis of the constitutive beaviour to bending of reinforced concrete beams is
extended to the case of masonry arches. The fibres are considered, as reinforcement bars in a
concrete cross-section, for the bridging action that they develop between the crack faces
controlling the opening and the propagation. Such bridging action is therefore modelled by means
of concentrated closure forces acting at the crack mouth, where FRP is placed.

FIGURE 1. Cracked beam element reinforced by FRP.

References
1. Modena, C., Valdemarca, M. and Valluzzi, M. R., L’Edilizia, vol. 5-6, 26-32, 1999.
2. Franciosi, V., Giornale del Genio Civile, Fasc. 7-8-9, 215-234, 1980.
3. Carpinteri, A., Scienza delle costruzioni, Pitagora Editrice Bologna, Vol. 1-2, 1995.
4. Carpinteri, A., Carpinteri, An., in Proceedings of the 6th International Brick Masonry
Conference, Roma, 1982, 502-510.
5. Ceriolo, L., Meccanica delle strutture in muratura rinforzate con FRP materials:
modellazione, sperimentazione, progetto, controllo, Edizioni Libreria Cortina Padova, 2000.
6. Carpinteri, A., “Application of Fracture Mechanics to Concrete Structures”, Journal of the
Structural Division, ASCE, Aprile 1982.
7. Carpinteri, A., “A Fracture Mechanics Model for Reinforced Concrete Collapse”, IABSE
Colloquium Final Report, Delft, 2/4 Giugno, 1981.
602 G. Ferro et al.

8. Carpinteri, A., Di Tommaso, A., Viola, E., Proceedings of the AIMETA Conference, Palermo,
1980.
9. Carpinteri, A., Stiffness Loss and Fracture Resistance of a Cracked Beam with Circular
Cross-Section, Internal Report n. 56 ISCB, 1981.
10. Carpinteri, A., “Stability of fracturing process in RC beams”, Journal of Structural
Engineering, ASCE 110, 1984, pp. 544-558.
11. Ferro, G., Theor. Appl. Fract. Mech. 38, 177-190, 2002.
12. Tada, H., Paris, P. C. and Irwin, G., The Stress Analysis of Cracks Handbook , second edition,
Paris Productions Incorporated (and Del Research Corporation), St. Luis, Missouri, 1985.
 2T26. Structural integrity 603

STRUCTURAL RELIABILITY ANALYSIS OF PIPE SUBJECTED TO REELING

Hugo A. Ernst, Richard E. Bravo1 and Federico Daguerre2

Tenaris Group - Center for Industrial Research
Dr. Simini 250 (2804), Campana, Argentina
hae@siderca.com
1Tenaris Group - Center for Industrial Research

Dr. Simini 250 (2804), Campana, Argentina

sidbrv@siderca.com
2Tenaris Group - Tamsa

Km 433.7 Carretera Via Xalapa, Veracruz (91697), México

federico.d@tamsa.com.mx

Reeling is a process used to install offshore lines that produce large cyclic plastic deformation on
the pipes. In this work the method to perform a structural reliability analysis (SRA) for a tube
subject to reeling is considered in detail.
When cyclic deformation is considered the fracture mechanics parameters, necessaries to
perform the SRA, are not clearly defined. For this reason, the present work will be focus on
addressing the problem of the evolution of the applied and material fracture mechanics parameters
through strain cycles and propose a methodology to perform a SRA for such cases.
An incremental method, based on Rice’s model [1], was proposed to calculate CTOD (crack
face displacement at a single point) evolution. No attempt was made to describe stress-strain fields
ahead of crack tip though a single parameter.
An experimental program was carried out using single edge notch tension (SENT) specimens
with fixed grips (no rotation at the ends) to study the effects of the strain history on the crack
driving force parameter.
Using the proposed model previously mentioned, CTOD values were calculated along the test,
based on the experimentally determined load vs. CMOD values.
Experimental results and the prediction from the model are compared in Fig. 1.

FIGURE 1. Comparison experimental / theoretical CTOD.

Regarding the material fracture mechanics parameters, the Material Memory (history
dependent) R-curve criterion, see Fig. 2, was validated by the experimental data, for up to two
604 H. A. Ernst et al.

cycles. It means that the CTOD value needed to cause further crack extension is the maximum
reached in the previous cycle.

FIGURE 2. Material Memory R Curve

As a result of the current work, a methodology to assess the structural integrity of pipes
subjected to the strain history of the reeling process is proposed.
It is necessary to obtain the relation between stress and strain through the deformation cycles in
order to determine the corresponding applied loads or stresses. The cyclic stress-strain relation can
be obtained experimentally for the specific material, or theoretically using a model that describes
the stress response to strain cycles.
Once the stress-strain relationship is obtained for a particular sequence of strain cycles,
together with the material properties and tube and defect geometry, the applied fracture mechanics
parameters can be obtained following the proposed methodology previously mentioned.
The material resistance evolution criterion presented has to be taken in account for cyclic
deformation. For the adopted criterion the R curve does not shift to a new origin.
A failure criterion has to be defined. This criterion, together with the previous items, allows the
determination of the critical defect sizes for a pipe subjected to a reeling process.

References
1. Rice, J.R., Mechanics of Crack Tip Deformation and Extension by Fatigue, ASTM STP 415,
American Society for Testing and Materials, Philadelphia, 247-309, 1967.
 2T26. Structural integrity 605

NETWORK SEISMIC CAPABILITY ASSESSMENT OF POWER HIGH

VOLTAGE ELECTRIC EQUIPMENT

Ion Manea, Ciprian Diaconu1, Constantin Radu2 and Mihai Negru3

Ion Manea, Senior researcher –SC ICEMENERG Craiova, Romania
manea53@yahoo.co.uk
1Ciprian Diaconu, Eng. – CN Transelectrica, Romania

cdiaconu@transelectrica.ro
2Constantin Radu, Eng. – SC SMART SA, Romania

cradu@transelectrica.ro
3Mihai Negru, Professor - University of Craiova – Mechanical Faculty negrum@mecanica.ucv.ro

The good operation of the power system should be ensured under normal and limit operating
conditions as well as in case of seism or short-circuit events. From this point of view, special
problems appear at the switching equipment with column type construction such as high voltage
circuit breakers. At this type of equipment, due to their characteristic construction and specific
tasks to carry out, depending on the location in the network, network topology and type of
switching events the mechanical stress can vary over a very wide range. These events have
cumulative effects and are leading, over time, to weariness of the structure and a seism or serious
short-circuit, due to their violent actions, can have destructive effects. Consequently, with a view
to ensure a high reliability, it’s a good idea that the principal switching equipments should be
subject of some experimental tests in order to assess their capability to stand out to future events.
The combined analysis by experimental modal analysis and finite element analysis offers a
good solution to solve this problem. The main problem is to determine the mathematical model of
equipment having to base experimental data obtained by measurements carried out on the
parameters well characterizing the equipment evolution in certain conditions. The equipment is
excited in a well-defined condition and determining the excitation and response evolution lows
will be identified a minimum number of parameters that are intrinsic characteristics of the system.
From experimental data it will be determined the natural frequencies and modal shapes that
constitute the main criterion for validation of the FEM mathematical model realized with a
program like as ANSYS. A correct FEM model allows us to assess the dynamic response of the
equipment to various excitations, in the particular case to a seismic motion applied to base.
A primary image of the equipment state will be realized from results of measurements of
vibratory response to functional tasks and experiments of modal analysis in view to determination
of the modal parameters. All weariness or cracks of the resistance structure can be identified from
modifications of the natural frequencies and modal shapes. A good image gives the distribution of
the vibratory response on equipment structure to natural tasks like closing or opening operations.
A serious analysis with FEM programs will be realized only when distribution of the vibratory
response and experimental modal analysis indicate that the equipment presents serious deviations
from characteristic natural frequencies and modal shapes.
The paper presents the working method for seismic capability assessment of the high voltage
switching equipment according to international norms IEC61166/1993 and IEC-TS 61463/2000.
There is presented one application carried out on a SF6 circuit breaker pole of 420kV/2500A.
In annex it is presented the circuit breaker pole and their first modal shapes in the seismic
frequency domain realized with ANSYS.
606 I. Manea et al.

Keywords: experimental modal analysis, finite element analysis, seismic qualification

REFERENCES
1. IEC61166/1993, High-Voltage alternating current circuit breakers; guide for seismic
qualification of high-voltage alternating current circuit breakers.
2. IEC-TS 61463/2000, Bushings – Seismic qualification.
 2T26. Structural integrity 607

FKM GUIDELINE “FRACTURE MECHANICS PROOF OF STRENGTH FOR

ENGINEERING COMPONENTS” – OVERVIEW AND EXTENSION TOPICS

B. Pyttel, I. Varfolomeyev1 and M. Luke1

Institut für Werkstoffkunde TU Darmstadt, Staatliche Materialprüfungsanstalt,
Grafenstr. 2, 64283 Darmstadt, Germany
1Fraunhofer-Institut für Werkstoffmechanik, Wöhlerstr. 11, 79108 Freiburg, Germany

pyttel@mpa-ifw.tu-darmstadt.de

The German Guideline “Fracture Mechanics Proof of Strength for Engineering Components” [1]
has been released 2001 as a result of activities sponsored by the Research Committee Mechanical
Engineering (FKM), task group “Component Strength”. The Guideline describes basics for the
integrity assessment of cracked components subjected to static or cyclic loading and provides a
step-by-step computational procedure for the use in engineering practice. The Guideline was
formulated based on a number of national and international reference documents, in particular
SINTAP [2] and BS 7910 [3]. Since 2004 it is also available in English. Subsequently, the
computational methods and solutions used in [1] were implemented in a PC program FracSafe
(www.fracsafe.de).
The latest 3rd edition of the Guideline (2005) includes several new topics. These allow for the
consideration of
• mixed mode loading,
• special effects at cyclic loading (variable load amplitude, short cracks),
• dynamic fracture effects,
• stress corrosion cracking,
• probabilistic aspects in fracture mechanics calculations.
In addition, the compendium of the stress intensity factor and limit load solutions is extended
and corrected according to the state-of-the-art. Some new examples and case studies are included
to demonstrate the application of the procedure to engineering problems.
This paper gives an overview of the Guideline [1] and describes new features available since
its 1st edition presented by Hodulak et al. [4]. Figure 1 shows the structure of the document and
gives emphasis to the new chapters and extension topics..
608 B. Pyttel et al.

FIGURE 1.Structure of the FKM Guideline.

References
1. FKM Guideline “Fracture Mechanics Proof of Strength for Engineering Components”,
VDMA Publ. House, 1st Edition, 2001 (German), 2nd Revised Edition, 2004 (German and
English), 3rd Extended Edition, 2005 (German).
2. SINTAP: Structural Integrity Assessment Procedure for European Industry, Report BE95-
1426, 1999.
3. BS 7910: Guide on Methods for Assessing the Acceptability of Flaws in Metallic Structures,
British Standards Institution, 1999.
4. Hodulak, L., Blauel, J.G., Berger, C., Pyttel, B., Gerdes, C.P., In Proceedings of the 14th
European Conference on Fracture, Cracow, edited by A. Neimitz et al., EMAS Publications,
2002, Vol. 1, 691-698.
 2T26. Structural integrity 609

STATIC AND DYNAMIC BEHAVIOR OF A 3D-PERIODIC STRUCTURE

J. Rishmany, L. Renault, C. Mabru, R. Chieragatti and F. Rezaï Aria1

Ecole Nationale Supérieure d’Ingénieurs de Constructions Aéronautiques
1, Place Emile Blouin, 31056 Toulouse, France
1Ecole des Mines d’Albi-Carmaux

Route de Teillet, 81013 Albi, France

rjihad@ensica.fr

This paper lies within the context of assessment of damage of an air-to-air compact exchanger
subjected to cyclic pressure, temperature and vibrations. The objective is to provide a predictive
lifetime model both for accelerated testing and in-service applications. The model should describe
the macroscopic behavior and the ageing of the interfaces between the core of the exchanger and
all elements attached to the core. These interfaces constitute the critical sites where damage might
occur. The modeling reported here is limited to the description of the core behavior in order to
evaluate later the efforts that the core generates on the interfaces. The stiffness of the core is
calculated in all three directions via an analytical method. In addition, a FE calculation is
performed using the FE code Samcef. Also, static tensile tests are carried out on specimens in all
three directions.
The core of the exchanger is assembled from a series of flat sheets and corrugated fins in a
sandwich construction. Thus it forms a periodic structure whose unit cell is presented on Fig. 1.

FIGURE 1. Unit Cell of the exchanger core

The analytical approach consists of replacing a unit cell by an equivalent system of

discontinuous masses and springs with their specific stiffness (k). Thus, calculating the stiffness of
the equivalent system provides the stiffness of the unit cell. In a similar manner, the stiffness of the
structure is obtained based on the stiffness of the unit cells. Tables 1 presents a comparison of
analytical calculation, FE calculation, and experimental results of the stiffness along two directions
(y) and (z). Stiffness along x-direction can be obtained by analogy with respect to z-direction.
610 J. Rishmany et al.

TABLE 1. Comparison of analytical, numerical, and experimenta results

Results show a disagreement between analytical and experimental stiffness along y-direction.
This is due to the presence of different types of boundary conditions on the unit cells related to
their position in the structure. Therefore, a new analytical/numerical approach is proposed taking
into consideration the variation of stiffness among the unit cells. The model remains simple in
terms of modeling and efficient in computational effort. In this method, different substructures
with different number of unit cells are considered. The stiffness of each substructure is
independently found by a FE calculation and then, is expressed analytically in terms of the
stiffness of the unit cells. The number of substructures considered is equal to the number of unit
cells having different stiffness. The stiffness of different unit cells is found by the resolution of the
linear system of equations.
The present model can give very accurate results (Table 2) compared to a complex and much
more time-consuming finite-element model of the real structure. These results are later used to
study the dynamic behavior of the structure.

TABLE 2. Stiffness comparison along y-direction

 2T26. Structural integrity 611

ENVIRONMENTAL EFFECT ON PIPELINE STEELS: A FITNESS FOR

SERVICE PERSPECTIVE

J. A. Alvarez, F. Gutierrez-Solana and S. Cicero

University of Cantabria
ETS Ingenieros de Caminos, Av/Los Castros s/n, 39005, Santander, Spain
alvareja@unican.es

Structural integrity assessment of components have had a big development in the last years. New
tools, as the Failure Assessment Diagram used at FITNET procedure [1, 2] have provided a
friendly efficient methodology for assessing all kind of components. This work applies the new
approaches to assess the structural integrity of pipelines subjected to high pressure hydrogen [3].
This analysis involves several specific factors. The first one is the material embrittlement due
to the hydrogen whose main consequence is a decrease on fracture toughness. The last critical
factor is the influence of crack geometry in the failure process because different geometries can
lead to different failure mechanisms. This paper analyses fractures under different conditions, from
sudden failures to leak before break conditions.
This paper also analyses the influence of component geometry in the safety conditions,
including subcritical crack growth previous to final failure [4, 5].
Fig. 1 presents, as an example, the safety evolution of a pipe containing a stable crack (3 mm)
supporting increasing hydrogen pressure leading to final fracture and the evolution of a crack
supporting constant pressure which provokes leak before break as a final state.

FIGURE 1. Failure Assessment Diagram on EAC assesment

References
1. Turbull, A.., Koers, R., Gutiérrez-Solana, F., and Álvarez, J.A. Proceedings of the
OMAE2005-67567, edited by ASME, 2005.
2. FITNET Consortium, FITNET: European Fitness for Service Network, EU´s Framework 5,
Proposal No. GTC-2001-43049, Contract No. G1RT-CT-2001-05071.
3. Gutiérrez-Solana, F., Elices, M. “Current solutions to Hydrogen damage in steels”. ASM
International Conference. Washington D.C. 1982.
612 J. A. Álvarez et al.

4. ISO 7539 Part 6, 1998, “Preparation and use of pre-cracked specimens for tests under
constant load or constant displacement”, International Standards Organisation, Geneva.
5. Álvarez, J.A., Gutiérrez-Solana, F. “An elastic-plastic fracture mechanics based methodology
to characterize cracking behaviour and its application to environmental assisted proceses”,
Nuclear Engineering and Design, 1999, pp. 185-202.
 2T26. Structural integrity 613

FINDING THE AUSTRALIAN RAILWAY LOAD SPECTRUM DESIGN AND

ASSESSMENT OF LIGHT WEIGHT & DURABLE RAILWAY STRUCTURAL
COMPONENTS

R. Jones and J. Baker

AM (Retd), Monash, Clayton Campus
Wellington Road, 3070, VIC, Melbourne, Australia
rhys.jones@eng.monash.edu.au

Containerised general freight is carried on flat top wagons predominantly designed to American
Associated Railways fatigue spectrum. Australian geographic conditions are different to American
conditions which could realize a more benign fatigue spectrum. Lighter wagons designed for
Australian conditions could carry more freight for the same gross mass. This investigation contains
scientific investigation in the data extraction and innovation in the instrumentation system and its
processing.
One major challenge facing the Australian rail industry is to reduce weight without
compromising safety or durability. With the potential use of new materials and designs Australian
engineers need a means to accurately assess the durability and safety of new designs.
The science in this investigation is the extraction of a uniquely Australian Rail Fatigue
Spectrum for use in the design of rolling stock to suit Australian conditions.
Designing for durability depends on knowing the load spectrum and the first step is the
identification of typical Australian loading scenarios. Bolster and sideframe design and testing and
certification has been adopted from the US AAR Standards which does not allow the significant
differences between the Australian and US operations to be addressed. Having identified the
loading spectrum, the program will concentrate on the design, manufacture and testing of
indigenous products.
Using documented experience, freight wagon bogie sideframes and a bolster have been strain
gauged for vertical and lateral loads whilst the spring nests between the two structures have been
monitored for deflection. A National Instruments SCXI system umbilicalled to a data acquisition
card and a laptop computer form the data processing and recording system. National Instruments
LABView has been used as to configure the acquisition and processing systems.
The complete system recorded data on a Freight Australia train enroute Melbourne to Sydney
and return. The system immaturity requires that Monash and Freight Australia staff be present
during the acquisition so data collectors were required to be trained in safe operations in and
around freight trains. Data was recorded onto the laptop hard disk memory and then periodically
burnt onto compact disk.
The science in this investigation is the fitment of strain gauges which are immune to noise and
the data analysis to produce an Australian Spectrum.
Strain gauges were fitted to a freight wagon bogie to record vertical and lateral strains. The
bridges were classified full bridges with some gauges recording strain and others forming dummy
bridge arms. All arms of the bridge were exposed to the same temperature eliminating temperature
compensation.
The data was preprocessed and stored using LabView software. The software is GUI based and
built.
614 R. Jones and J. Baker

The data has been validated as sound, processed through FFT and filtered for best fit before
rain pair counting. The Fast Fourier Transformation showed concentrations of energy levels below
1 cycle per second.
Currently the data is being arranged into a Road Environment Percentage Operational
Spectrum (REPOS) to compare with American Associated Railways information. This is expected
to be completed by October 2005.

Michael Jones, John Baker and Professor Rhys Jones fitting the spring nest potentiometers
 2T26. Structural integrity 615

STRUCTURAL INTEGRITY ASSESSMENT OF COMPONETS WITH LOW

CONSTRAINT

S. Cicero, F. Gutierrez-Solana and J. A. Alvarez

University of Cantabria
ETS Ingenieros de Caminos, Av/Los Castros s/n, 39005, Santander, Spain
ciceros@unican.es

When performing Structural Integrity assessments using Failure Assessment Diagrams, a point
with coordinates Kr and Lr is representative of the component situation. The first one represents its
situation against fracture and the second one against plastic collapse:

KI P
Kr Lr
K IC , PL (1)

The fracture toughness value, KIC, is obtained in tests under high constraint conditions in such a
way that the obtained value is a lower bound of the material resistance to fracture. This working
scheme often leads to overconservative results when the component being assessed has lower
constraint conditions, as shown in Fig.1, where point A represents the component at failure. The
point should lead over the Failure Assessment Line (point B) but actually it is far away from it,
leading to a safety factor bigger than two (defined as OA/OB).

FIGURE 1. Assessment of a component with low constraint

Shallow cracks in components subjected to tension have lower hydrostatic and maximum
principal stresses near the crack tip than deep cracks in bending stress fields. These lower stresses
near the crack tip lead to increased load carrying capacity compared to that estimated using the
fracture toughness obtained from deeply cracked bend specimens [1]. This is referred to as 'loss of
constraint' and some fracture assessment procedures [2,3,4] contain methods for estimating the
increased load carrying capacity under these conditions. More generally, low constraint occurs
near blunt notches and at small cracks emanating from such notches. Some theories have been
developed to take into account the notch effect [5,6,7].
However, no procedures have been developed to assess components with loss of constraint due
to both phenomena despite this problem is quite common in metals fabrication processes. An
616 S. Cicero et al.

overall procedure to assess the 'in-plane' loss of constraint is presented [8]. It demonstrates the
independence of the phenomena described above and proposes a new formulation and its
validation through laboratory tests.

References
1. Cicero, S., Ainsworth, R.A, In Proceedings of the OMAE2005-67567, edited by ASME,
2005.
2. British Energy Generation Limited, R6: Assessment of the integrity of structures containing
defects, Revision 4, 2001.
3. FITNET Consortium, FITNET: European Fitness for Service Network, EU´s Framework 5,
Proposal No. GTC-2001-43049, Contract No. G1RT-CT-2001-05071.
4. SINTAP: Structural Integrity Assessment Procedures for European Industry, Brite-Euram
Project No. BE95-1426, Contract No.BRPR-CT95-0024, Final Report, 1999
5. Kim, J.H., Kim, D.H. and Moon, S.I., Materials Science and Engineering A, vol. 387-389,
381-384, 2004.
6. Taylor, D., Cornetti, P. and Pugno, N., Engineering Fracture Mechanics, vol. 72, 1021-1038,
2005.
7. McClung, R.C., Chell, G.G., Lee, Y-D., Russel, D.A. and Orient, G.E., Development of a
practical methodology for elastic-plastic and fully plastic fatigue crack growth, NASA,
Report NASA/CR-1999-209428.
8. Cicero, S., Evaluación de la Integridad Estructural de Componentes Sometidos a
Condiciones de Bajo Confinamiento, PhD Thesis, Universidad de Cantabria, Santander,
Spain, To be published.
 2T26. Structural integrity 617

LIFE ASSESSMENT OF SUPERHEATER TUBES FABRICATED FROM

2.25CR-1MO STEEL

S. Fujibayashi
Idemitsu Engineering Co., Ltd
Makuhari Techno Gardeb B23, 1-3 Nakase Mihama-ku, Chiba 261-8501 Japan
shimpei.fujibayashi@si.idemitsu.co.jp

In recent years, most of plant owners have tried to extend the continuous run length and shorten a
turnaround period in order to enhance the competitiveness. In the case of high temperature
components, in which creep is a major damage mechanism, creep resistance decays depending
upon the localized condition to which a component is exposed. Thus, condition assessment should
be performed with a suitable interval to evaluate the remaining operability. However,
conventional techniques to predict the remnant life requires long hours especially for low stressed
components like superheater tubes in a boiler. Choosing the most damaged tube among numerous
ones is not necessarily easy. Since the distribution of tube skin temperatures is rather erratic, a
spot inspection cannot be recommended. Therefore, a non-destructive screening technique would
be useful for a timely judgement. Due to carbides coarsening associated with high temperature
exposure, hardness decreases continuously with operating hours. In the present work, the
effectiveness of hardness measurement to find the weakest tube fabricated from 2.25Cr-1Mo steel
was confirmed. The mean rupture life against hardness compensated stress, defined as /Hv, for
virgin and service-exposed materials was correlated with Manson-Haferd Parameter by the
following equation.

MHP = (log tr-16.053)/(T-380) = -0.02975-0.01828 x log (/Hv) –0.01146 x log 2 (/Hv),

where tr is the mean rupture life, T is the temperature in K and Hv is the hardness prior to a test.
Most of rupture data for virgin and service-exposed tubes in actual boilers were predicted
with the accuracy of a factor of two.
However, the above relationship cannot be applied to the materials hardened by heat
treatment or cold work prior to creep tests, showing noticeably lower creep resistance. In order to
elucidate the reason for lowered creep resistance of a hardened material, precipitates evolution
associated with creep exposure was observed using as-received and cold-worked materials.
Rupture life decreased with the extent of cold-work and became halved when 15% of prestrain was
given and resultant hardness increased to 214 Hv from the original value of 158 Hv. Rapid
coarsening of carbides observed in a prestrained material should be ascribed to higher dislocation
density. Namely, additional paths for solute atoms should contribute to faster coarsening of
carbides. Poor creep strength of a quenched and tempered plate or weld metal of 2.25Cr-1Mo steel
with high hardness values might be also attributable to the effect of pipe diffusion.
In addition, it was found that the influence of a testing environment was quite remarkable,
especially when creep properties at low stresses are examined. Though it is generally accepted that
too much acceleration in stress would generate a pessimistic result, extraordinary long tests in air
would also lead to too much conservatism. To achieve the further life extension, the oxidation
effect upon the data obtained in air should be taken into account. The correction of existing data
by NIMS (No.3B) was made using the Kachanov’s damage mechanics and the equation to predict
metal loss during a creep test derived by Bueno and Marino. The mean creep strength expected in
vacuum, trv, is expressed by following equation using the Larson-Miller Parameter.
618 S. Fujibayashi

LMP = (log trv+20) T = 23,661+1,832 log –1,835 log2

By converting the raw testing data at 6000C and 22MPa, in which rupture life was 113,445
hours and stress at rupture was supposed to be doubled simply due to oxidation, the expected life in
vacuum became an order of magnitude longer. It should be addressed that lowering testing
temperatures in air dose not necessarily generate reliable results. In the iso-stress rupture tests,
metal loss caused by oxidation increased with decrease in testing temperature. Short-term tests at
higher temperatures reduced the metal loss and rupture datum obtained at 7000C was consistent
with those at lower temperatures.
The following equation was generated to correlate the current hardness and remaining creep
strength in vacuum.

LMP =(log trv+20) T = 18,858-6,183 log (/Hv) –1,807 log2 (/Hv)

It was confirmed by the experiments using the service-exposed superheater tube in argon that
the above equation still contained a safety margin.
 2T26. Structural integrity 619

PREDICTING CLEAVAGE FRACTURE IN PRESENCE OF RESIDUAL

STRESSES; A NUMERICAL CASE STUDY

S. Hadidi-Moud1, C. E. Truman and D. J. Smith

University of Bristol, UK, 2-Ferdowsi University, Iran
Department of Mechanical Engineering, University Walk, Bristol, BS8 1TR,
1Department of Mechanical Engineering, P.O. Box 91775-111, Mashhad, Iran

Saeid.Hadidi-Moud@Bristol.ac.uk, C.E.Truman@Bristol.ac.uk, David.Smith@Bristol.ac.uk

Background
Recently, a local stress based approach to fracture, developed by Hadidi-Moud et al [1] was
applied to predict cleavage fracture probability in ferritic steels. Predictions were validated by
experimental data from Smith et al [2] and Garwood and Smith [3]. The model captured the
influence of warm pre-stressing on low temperature fracture by using distribution parameters
calibrated to as received data, as shown by Hadidi-Moud et al [4]. This was justified by noting that
the plastic zone stress field automatically contained the prior load history effect and thus accounted
for its contribution to fracture.
The model was also used to explore the ability to predict fracture due to several load history
events. For example local compression followed by warm pre-stressing and subsequent loading to
fracture at low temperature. Where load history only included warm pre-stressing, thus introducing
compressive RS field at the tip region, enhancement in toughness was predicted, consistent with
the stress field in the plastic zone. For the combination of local compression (introducing tensile
RS in this case) and WPS, it was expected to obtain predictions shifted to the left of the WPS
predictions (less relative enhancement if any). However no difference was observed in the
predictions compared to the WPS alone, suggesting that the influence of local compression was
completely wiped out by the WPS loading. These findings were also supported by experimental
evidence (Hadidi-Moud et al [5]).
Procedure
The above study initiated a new research plan to investigate the interaction effect when the
influence of initial stress is rather significant. This work presents the results of a numerical study
conducted by introducing an RS field introduced as the initial stress field with significant extent
and or level of stresses defined by the user. The FE model was adopted from a previous work by
Bouchard et al [6] that was used for validation of RS fields in welds. It is a cylindrical vessel with
a part through circumferential crack on the outer surface, 50 mm long and 5 mm through the 19.6
mm thick wall at its deepest point. A533B ferritic steel data at room and low (-170qC) temperature
was used [3]. A combination of pressure on the inner surface and its corresponding axial tension on
the cross section of the pipe was used to simulate the loading.
A series of analyses were performed to provide the stress within the plastic zone for model
predictions. Toughness was calculated from the load and crack geometry using the approximate
analytical formulae available from Anderson [7]. The consistency of the predicted toughness
distributions with the corresponding stress distributions in the plastic crack tip region was also
investigated. The load cycles included IRS-CF, fracture at low temperature (CF) following
introduction and redistribution of an initial RS field (IRS), LUCF, introducing residual stresses by
preloading and unloading at room temperature (LU) followed by CF, and IRS-LUCF. For the
analyses with IRS two levels of initial stress were introduced, a “short range” field of local nature
620 S. Hadidi-Moud et al.

and a “long range” field of global nature with significant level and extent compared with the local
field. The latter was used to address the interaction effect.
Results
For all the performed analyses the stress distributions along the crack extension path through
the deepest point of the crack front were plotted and compared for various analyses. The relative
levels of stress were consistent with the predicted failure probability. The predicted distributions
are summarized in Figure 1. Various failure predictions1. As shown, the interaction of initial
stresses with the stresses arising from loading history and final loading is dependant of the
significance and the extent of these stresses. Figure suggests that at short range initial stresses,
warm pre-stressing has eliminated the influence of initial RS field whereas in presence of
significant IRS the resulting predictions for IRS-LUCF and LUCF are clearly different suggesting
the noticeable influence of interaction of the stress fields with mechanical loading on final fracture.

Figure 1. Various failure predictions

References
1. Hadidi-Moud, S., Mirzaee-Sisan, A., Truman, C.E., and Smith, D.J., Fatigue Fract. Engng.
Mater. Struct., vol. 27, 931-942, 2004.
2. Smith DJ, Hadidi-Moud S and Fowler H, Engineering Fracture Mechanics, Vol. 71, Issues
13-14, 2015-2032, 2004.
3. Smith D.J., Garwood S.J., Int. J. Pres. Ves. & Piping. Vol. 41, 297-331, 1990
4. S Hadidi-Moud, A Mirzaee-Sisan, C E Truman and D J Smith, Key Engineering Materials,
Vol. 261, 69-74. Advances in Fracture and Failure Prevention, 2004
5. Hadidi-Moud S., Mahmoudi A. H., Truman C. E. and Smith, D.J., In proceedings of the 9th
Int. Conf. on Mechanical behaviour of Materials (ICM9), Geneva, Switzerland, 2003
6. Bouchard, J. and Bradford, C., In proceedings of ASME Pressure Vessel and Piping
conference (PVP2001), Vol. 422: Fracture and Fitness, Atlanta, USA, 2001
7. Anderson, T. L., Fracture Mechanics, CRC Press, USA, 1994
 2T26. Structural integrity 621

A NECESSARY CONDITION FOR CLEAVAGE ON LABORATORY

SPECIMENS AND STRUCTURES

Vincent Le Corre, Stéphane Chapuliot, Suzanne Degallaix1 and Antoine Fissolo

CEA Saclay - Bat. 607 DEN/DM2S/SEMT/LISN
F-91191 Gif-Sur-Yvette Cedex
1Lille Mechanics Laboratory (LML, UMR CNRS 8107) Ecole Centrale de Lille BP 48

59651 Villeneuve d'Ascq Cedex

Le_Corre.Vincent@ec-lille.fr

This study concerns the nuclear pipe integrity. It is focused on the fracture in the ductile to brittle
transition region of ferritic steels. An essential problem is the transferability of failure criterions
from specimen to in-service structures. In order to ensure the integrity of a structure, we have to
estimate the existing margin towards the risk of brittle fracture. The question is : for a given
specimen or structure with a given crack geometry and with given loading conditions, at which
temperature cleavage may occur? In order to answer we have to discern the necessary conditions of
cleavage appearance.
This study is based on the results of the Euro-Fracture Toughness Dataset [1]. First, tests on
notch tensile specimens have been numerically analyzed. A critical cleavage stress has been
defined has the threshold stress in the fitting formula :

m
ª §V V · º
Pr 1  exp «  ¨¨ 11 th
¸¸ »
«¬ © Vu ¹ »¼
(1)

FIGURE 1. Fracture behaviour of NT specimens

The second part of the study concerns the fracture toughness tests on CT specimens. Three-
dimensional elastic-plastic finite element analyses were performed using the CAST3M software to
study the crack front stress fields on CT specimens. The crack propagation was simulated by nodes
slackening method controlled by experimental law J-Integral vs. crack length. In order to evaluate
the volumes involved in the cleavage process, we used as main criterion the threshold stress Vth
defined on NT specimens. As recommended by [2] , through this volume we try to put in evidence
the role of the yield stress and of plasticity constraint in cleavage fracture. As main result we put
622 V. L. Corre et al.

forward the linear relationship between the logarithm of the failure probability and the volume Vth,
as predicted by the weakest link theory.

FIGURE 2. CT mesh for numerical propagation simulation and experimental evolution of Vth for 4
geometries of CT specimens gathered at 0°C

This result shows the pertinence of a threshold stress criterion for the appearance of cleavage.
What is more, we can draw the evolution of the slope F with yield stress to show that the
dependence to temperature can be written :

m
1 § V Y (T )  V Y 0 ·
F (T ) ¨¨ ¸¸
V0 © Vu ¹ (2)
where VY is the yield stress, VY0 is the yield stress at the limit temperature for cleavage appearance
and m is found the same as in (1). Vu is slightly different.

References
1. Heerens J., Hellmann D., Development of the Euro fracture toughness dataset.. Engineering
Fracture Mechanics, Vol. 69, 421-449, 2002.
2. Chapuliot S., Le Corre V., Investigations on a stress based criteria for brittle fracture,
submitted in Fatigue and Fracture of Engineering Materials and Structures (2005)
 2T26. Structural integrity 623

SAFETY ASSESSMENT OF COMPONENTS WITH CRACK-LIKE DEFECTS

Yu. G. Matvienko and O. A. Priymak

Mechanical Engineering Research Institute of the Russian Academy of Sciences
4 M. Kharitonievsky Per., 101990 Moscow, Russia
matvienko7@yahoo.com and
ETC CKBN
31 B. Serpuhovskaya Street, 142100 Podolsk, Moscow Region, Russia
info@etc.ckbn.ru

Safety assessment methods play an important role in the industrial realization of fracture
mechanics applications. If the engineering component contains a crack or crack-like defect,
fracture mechanics approach must be employed. Therefore, the comparison between the applied
crack driving force and the material’s fracture toughness has to be carried out on the basis of crack
tip parameters such as the stress intensity factor, the J-integral, the crack tip opening displacement
or etc. In this case, the critical and acceptable state of the damaged component can be predicted in
terms of an applied stress or a crack-like defect size. The aim of this paper is to give a brief survey
of the basic principles of a failure assessment diagram and its application in the safety analysis of
components with defects.
The cohesive zone model and criterion of average stress in the cohesive zone ahead of the
crack/notch tip have been used by Matvienko [1, 2] to describe failure assessment diagrams for
cracked and notched bodies. The normal stress distribution at the notch tip, used in the criterion of
the average stress, is similar to that crack under uniform remote tensile stress but shifted from the
notch tip to a point of abscissa to U / 2 , i.e. r t U / 2 . It should be pointed out that such stress
distribution was suggested for blunt cracks when the distance ahead of the crack tip is much
smaller than the crack length and greater than the crack tip radius U . The fracture criterion leads
to the failure assessment diagram of notched bodies in terms of the stress intensity factor and
applied stresses V
2 1 / 2
§ V · ª § V coh · 2 1 º
K 1notch K 1C 1  ¨¨ ¸¸ «1  ¨ ¸ 2»
© V coh ¹ «¬ © V ¹ K t »¼ (1)

Here, K 1C is the fracture toughness, K t is the elastic stress concentration factor, V coh is the
cohesive stress ahead of the notch tip. The stress intensity factor at the notch tip is denoted
as K1notch . The cohesive stress V coh is determined by the yielding criterion at the boundary
of the cohesive zone according to flow law of von Mises for a body with a finite notch [3].
The failure assessment diagram (1) can be adopted for a component with a crack-like defect (or
blunt crack) as well as a notch. To determine an acceptable (safe) region, it should be reasonable to
introduce safety factors (e.g. [4, 5]) in the fracture criterion. The following condition should be
fulfilled if detected or assumed crack-like defect of a certain size should be assessed as acceptable
624 Yu. G. Matvienko and O. A. Priymak

2 2 1 / 2
K § V · ª §V · 1 º
K1notch d 1C 1  ¨¨ a ¸¸ «1  ¨¨ coh ¸¸ 2
»
SFK © V coh ¹ «¬ © V a ¹ K t »¼
(2)

where V Y is the yield stress, SF K is safety factor against fracture. The acceptable applied stress
V a is suggested to be less than V Y / SF Y , where SFY is safety factor against plastic collapse.

For the evaluation of critical conditions for the acceptable stress, the safety factor SF K in Eq.
(2) is set to unity. Any reliable method to calculate the stress intensity factor K 1 notch can be
employed if these methods are validated.

The safety factor SF K can be calculated by making an assumption that the applied acceptable
stress should be not less than the yield stress of material for an engineering component with a
crack-like defect of the acceptable size [4]. In this case, the safety factor SF K can be presented
taking into account Eq. (2) as

2 2
§ V / SF Y · §V · 1
1  ¨¨ Y ¸¸ 1  ¨¨ coh ¸¸ 2
© V coh ¹ © VY ¹ Kt
SF K SF Y
2 2
§ V · § V coh · 1
1  ¨¨ Y ¸¸ 1  ¨¨ ¸¸ 2
© V coh ¹ © V Y / SF Y ¹ Kt (3)
Thus, the right-hand side of Eq. (2) defines the acceptable region in the failure assessment
diagram. If the assessment point falls within this region, the component with a crack-like defect is
acceptable, i.e. it fulfils the required safety demands. For the special case of a crack ( K t o f )
the failure assessment diagrams (1) and (2) are transferred to the failure assessment diagram for
component with a sharp crack, and the safety factor (3) becomes the safety factor against fracture
of cracked component.
Safety analysis of pipelines and pressure vessels with defects is considered from the point view
of the proposed approach.

References
1. Matvienko, Yu. G., Int. J. Fracture, vol. 124, 107-112, 2003
2. Matvienko, Yu. G., Int. J. Fracture, vol. 131, 309, 2005
3. Matvienko, Yu.G., Int. J. Fracture, vol. 98, L53-L58, 1999
4. Matvienko, Yu. G. and Makhutov, N.A., Int. J. Pres. Ves. Piping, vol. 76, 441-444, 1999
5. Brickstad, B., Bergman, M., Andersson, P., Dahlberg, L., Sattari-Far, I. and Nilsson, F., Int.
J. Pres. Ves. Piping, vol. 77, 877-881, 2000
 2T26. Structural integrity 625

NUMERICAL ANALYSIS OF SURFACE CRACKS IN STEAM GENERATOR

TUBES

Zdenko Tonkovi, Ivica Skozrit and Jurica Sori

Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb
I. Luia 5, 10000 Zagreb, Croatia
ztonkov@fsb.hr, ivica.skozrit@fsb.hr, jurica.soric@fsb.hr

Operating experience with steam generators has shown that axial surface cracks present one of the
most common causes of loss of steam generator tube integrity. An accurate computation of the
plastic limit pressure represents a key for the prediction of structural integrity and reliability of
pressurized tubes. A greater number of existing solutions for limit pressure of a cracked tube have
been developed either analytically, based on a simple equilibrium stress field, or empirically, based
on test data [1]. These solutions are generally shown to be too conservative for components with
part-through-thickness defects. Recently, a finite element based plastic limit pressure expression
for cylinders with external axial semi-elliptical surface cracks has been developed in Ref. [2].
However, the proposed expression is applicable to a very limited range of crack dimensions, and
therefore the new extended solutions are desirable.
In this paper, three-dimensional non-linear finite element analyses have been carried out to
determine the plastic limit loads for external axial semi-elliptical surface cracks in VVER steam
generator tubes under internal pressure p. The tube geometry and loading are shown in Fig. 1. The
outer radius of the VVER tube R0 is 8 mm and the wall thickness t is 1,5 mm. The crack is assumed
to have a semi-elliptical shape described by a length 2c, depth a and normalized crack length,

defined as U c / R m t . Six different crack lengths were considered, 2c = 5, 10, 20, 30, 40
and 50 mm, and four values of the ratio of the crack depth to the tube thickness were selected, a/t =
0.2, 0.4, 0.6 and 0.8.

FIGURE 1. Geometry and dimensions of a tube subjected to internal pressure with an external
axial semi-elliptical surface crack

These thick tubes are made of austenitic steel 08X18H10T that corresponds to AISI 321 grade.
The applied material model employs the J2 deformation theory of plasticity and the small strain
formulation. Mechanical properties of the austenitic steel 08X18H10T, obtained experimentally
[3], are summarised in Table 1. The finite element analysis is performed by using the commercial
FE package ABAQUS/Standard [4]. A typical finite element mesh applied in the analysis is shown
in Fig. 2.
626 Z. Tonkovi et al.

TABLE 1: Mechanical properties of the austenitic steel 08X18H10T (Timofeev et al. [3])

a) b)
FIGURE 2. Typical FE mesh for a tube with external axial surface crack
(2c = 10 mm, a/t = 0.6): (a) whole mesh; (b) crack tip mesh

The plastic limit pressure solutions have been developed on the basis of the finite element limit
load analyses employing elastic-perfectly plastic material behaviour. Using these solutions, a new
analytical approximation of the plastic limit pressure has been developed for a wide range of
cracks

pL ( 2 / 3 ) V Y ln
Ro
Ri
>
A0  A1 a / t   A2 a / t 
2
@ (1)

The values of the plastic limit pressure obtained by this study are compared with the empirical
solutions presented in Ref. [1] and with finite element based plastic limit pressure expression
developed in Ref. [2]. The proposed analytical approximation of limit pressure provides very
useful tools for assessing the integrity of pressurized tubes. The effects of crack depth and length
on the failure pressure are evaluated for the considered tube geometry.
References
1. Miller, A.G., Int J Pres Ves Pip, vol. 32, 191–327, 1988.
2. Kim, YJ., Shim, DJ., Nikbin, K., Kim, YJ., Hwang, SS. and Kim, JS., Int. J. Pres. Ves. Pip.,
vol. 80, 527-540, 2003.
3. Timofeev, B.T., Karzov, G.P., Blumin, A.A. and Anikovsky, V.V., Int. J. Pres. Ves. Pip., vol.
76, 393-400, 1999.
4. ABAQUS User's guide and theoretical manual, Version 6.5, Hibbitt, Karlsson & Serensen,
Inc., 2005.
 2T28. Mesofracture mechanics 627

TENSILE SIMULATION OF POLYMERIC MATERIAL CONSIDERING THE

MESO-SCALE STRUCTURE

Akira Shinozaki, Kikuo Kishimoto and Inoue Hirotugu

Department of Mechanical and Control Engineering, Tokyo Institute of Technology
2-12-1, O-okayama, Meguro-ku, Tokyo, 152-8552, Japan
sinozaki@mep.titech.ac.jp, kkishimo@mep.titech.ac.jp, inoueh@mep.titech.ac.jp

Several studies related to meso-scale structure and mechanical properties have been made. De
Gennes[1] discussed the motions for one chain performing wormlike displacements inside a
strongly corss-linked polymer gel. Theodorou and Suter[2] suggested the method for the detailed
atomistic modeling of well-relaxed amorphous glassy polymers. Using molecular dynamics
simulations, Stevens[3] studied the effect of interfacial bond density and network size on
interfacial fracture. Hiroo[4] studied the coarse-graining is applied to flexible polymer chains.
Yashiro[5] discussed the mechanical properties from atomistic scale. Marc [6] studied the
orientation and crystallization of polyethylene during uniaxial extension. These works have been
clarified some aspects of mechanical properties.
In fact, the mechanical properties of polymers are strongly influenced by meso-scale (10-9~10-
3 m) structure, such as entanglement, molecular weight distribution, orientation, etc. It is important

to understand the relationship between mechanical properties of macro-scale and meso-scale

structure. But the study of relationship between mechanical properties and meso-scale structure is
few. This study therefore aims to clear this relationship by the simulation considered the meso-
scale structure by chain network model. Large strain deformation of this model is evolved via
molecular dynamics analysis.
Actually, molecular chains of Polymeric materials have many kinds of length, andentangled
each other. The complex meso-scale structures are composed by molecular chain as described in
Fig. 1. In this study, the structures are viewed molecular chain network model (Fig. 2). The
molecular chain which is the most basically element of the network model, is represented by
simple model such as the mass-spring model (Fig. 3). N successive atoms are combined into a
single mass-spring segment.

FIGURE 1. Entangled Polymer chains. FIGURE 2. Network model.

628 A. Shinozaki et al.

FIGURE 3. Segment. FIGURE 4. After concentrated model.

FIGURE 5. Deformation process of model A FIGURE 6. Stress-Stretch ratio curve

In this work, two kinds of network model are generated. One is composed by linear chains
(model A (Fig.2)) and another model has branches (model B). The network model is not
equilibrium. Therefore the network model is concentrated (Fig.4). Then, the concentrated network
model is simulated (Fig. 5). The simulation is assumed tensile test. Stress-Stretch ratio curve (Fig.
6) is obtained.
From the simulation, it is clear that the branch strongly affects the young’s modulus and yield
stress.

References
1. P.G. de Gennes: J. Chem. Phys, Vol. 55, 572, 1971
2. D.N. Theodorou, U.W,suter: Macromolecules, Vol. 18, P 1467, 1985
3. M.J.Stevens: Macromolecules, Vol 34, P2710, 2001
4. F.Hiroo, T. Jun-ichi, D. Masao: J. Chem. Phys, Vol. 116, P 8183, 2002
5. Y.Kisaragi, I.Tomohiro, T. Yoshida: Internatinal Journal of Mechanical Sciences, Vol. 45, P
1863, 2003
6. Marc S.Lavine, Numan Waheed, Gregory C. Rutledge: Polymer, Vol. 44, P1771, 2003
 2T28. Mesofracture mechanics 629

MICROFRACTURE AND STRAIN LOCALIZATION: A COMPUTATIONAL

HOMOGENIZATION APPROACH

C. Dascalu, G. Bilbie and R. Chambon

Laboratoire Sols-Solides-Structures, CNRS – INPG – UJF
BP 53, 38 041 Grenoble Cedex 9
Cristian.Dascalu@hmg.inpg.fr, bilbie@geo.hmg.inpg.fr, Rene.Chambon@hmg.inpg.fr

The aim of this work is to study the appearance instable behaviours (like strain localisation bands)
in elastic solids, as a consequence of micro-fracture. A two-scale approach of computational
homogenization (e.g. Kouznetsova et al. [2], Miehe [3]) is considered. The macroscopic behaviour
is obtained by localization/homogenization exchanges with representative elementary volumes.
Periodic boundary conditions are assumed for each elementary cell.
In order to properly understand the relation between micro-fracture and strain localization, in
this work we restrict our analysis to non-dissipative behaviours. On the level of the microstructure,
we consider a large strain hyper-elastic material containing traction-free microcracks with
unilateral contact.
The appearance of instabilities is indicated by the loss of ellipticity of the equilibrium
equations. To clearly identify the microcracks influence on the macroscopic stability, we assume
stable microscopic constitutive laws. In this case, the macroscopic instabilities are completely due
to the presence of the microcracks in the elementary cells.

FIGURE 1. Computational localization/homogenization scheme. A deformation gradient F is

imposed on the boundary of a cell and the resulting mean Piola-Kirchhoff stress P is returned.

The homogenization procedure is represented in Figure 1. If M ( X ) is the deformation map of

an elementary volume, F  X its gradient and P  X  the first Piola-Kirchhoff stress, the boundary
conditions on the boundary are:

M 
X   M  X  F ˜ X 
 X  ; t ( X  )  t(X  )
(1)

where t P ˜N with N the external normal. We denote by F  X  and P  X  the

corresponding mean values. This procedure provides an implicit (numerical) law P P  F  .
The computations are done with finite elements, for different hyperelastic laws. A finite-difference
formula together with the obtained implicit law allow for the computation of the coarse-scale
630 C. Dascalu et al.

acoustic tensor Q (e.g. Rice [4]). The appearance of macroscopic instabilities is indicated by the
loss of ellipticity :

d e t Q  0
(2)
For a hyperelastic law of the type of that considered by Abeyaratne and Triantafyllidis [1], we
represented in Fig. 2 the stability/instability regions in the plane 1  F1 1 ,1  F 2 2  for the
elementary cell with two microcracks.

FIGURE 2. a) Stability/instability regions b) REV with two cracks

We show that, even for a stable and non-dissipative microscopic behaviour, the presence of
microcracks can give rise to an unstable macroscopic behaviour, typical for bands of localization
of the deformation. We present macroscopic localization behaviours obtained with two-scale finite
element computations.

References
1. Abeyaratne, R., and Triantafyllidis, N., J. Appl. Mech., vol. 51, 481-486, 1984.
2. Kouznetsova, V., Geers, M. G., and Brekelmans, W. A. M., Int. J. Numer. Meth. Engng.,
vol.54, 1235-1260, 2002
3. Miehe, C., Comput. Methods Appl. Mech. Engrg., vol. 192, 559-591, 2003.
4. Rice, J.R., Theoretical and Applied Mechanics, edited by W.T. Koiter, North-HollandPubl,
Amsterdam, 1976, 207-220
 2T28. Mesofracture mechanics 631

STRAIN AND FRACTURE AT MESOSCALE OF COATED MATERIALS

S. Panin
Institute of Strength Physics and Materials Science SB RAS
2/1 Academicheskii pr., 634021, Tomsk, Russia
svp@ms.tsc.ru

Recently surface-strain mapping technique have found a wide application in experimental

mechanics due to intensive development of image acquisition equipment (high–resolution TV- and
photo–cameras) as well as possibility of fast and precise processing of experimental data with the
use of personal computers. Most interesting and essential results on developing both experimental
techniques and interpretation of results in terms of mechanics were achived in S.Corolina
University in prof. M. Sutton’s group [1]. In so doing most attention was paid to fatigue fracture of
heterogenious materils. Interesting results on mathematical processing of experimental data
obtained by surface–strain mapping were gained in Great Britain in Prof. P.Whithers’ group [2].
In researches of Prof. V.Panin to carry out within the framework of physical mesomechanics,
were shown that surface layers of a loaded solid are an individual structural level of strain origin
and development. Prof. V.E. Panin [3,4] has shown that special state of surface layers causes fast
shear stability loss under loading and gives rise to initial acts of plastic deformation there. In so
doing the propagation of defect flows of a non-dislocation nature in surface layers plays the
determining role. The development of the given processes is much pronounced in materials where
dislocation plasticity is restrained by elements of internal structure.
If surface layers are subjected to various treatment procedures or a protective (strengthening)
coating is deposited onto material surface, the pattern of strain origin and development is
substantially changed. The restriction of dislocation plasticity results in increase of local stress
concentration level, while the character of strain development is governed by pattern of their
occurrence and relaxation.
The target of experimental study of coated materials being loaded under different schemes was
associating shape and numerical parameters of loading diagram and characteristic patterns of strain
development (localization). The estimation of strain localization can be obtained by calculating
strain components (main plastic shear, strain rate intensity, shear and rotation component of
distortion tensor) by numerical differentiation of data obtained by television–optical observation of
surface of a loaded heterogeneous material.
Another important reason for carrying out the investigations is optimizing modes for coating
deposition as well as structural and geometrical factors. Frequently cracking of coatings occurs
under loading which does not give rise to catastrophic failures of surface–hardened parts of
machine. It is of importance to find out cracking pattern that diminishes the risk of strain
localization while the estimation of the latter can be performed by processing optical images
gained at the “coating–substrate” interface.
It is shown that most optimal from the point of view of behaviour under mechanical loading is
formation of coatings with nonflat “coating-substrate” interface and presence of a gradient
transient layer. The latter hinders crack propagation into lower substrate layers and increasing
composition strength by the additivity law.
Experimental prove has given to the fact that fine cracking of the coating allow to avoid
substantial strain localizations that can keep exploitation properties of surface–hardened parts of
machine that does not completed with catastrophic consequences (fast failure).
632 S. Panin

It is clearly demonstrtated that in materials with high-strength low-ductile coatings the

interface is the site of fatigue crack origin. The decrease in distinction of elastic moduli of a
coating and substrate results in nucleation of the crack on surface of the coating.

References
1. Sutton, M.A., Helm, J.D., Boone, M.L., International journal of fracture, No. 109, 285, 2001.
2. Clocksin, W.F., Chivers, K.F., Torr, P.H.S. et. al., In Proceedings International Conference on
New Challenges in Mesomechanics, edited by Pyrz and Sih. Aalborg University Press, Vol. 2,
2002, pp. 467.
3. Panin, V.E., Theor. and Appl. Frac. Mech., Vol. 37, 261, 2001.
4. Panin, V., Physical Mesomechanics, Vol. 4, No. 3, pp. 5, 2001.
 2T32. Micromechanisms in fracture and fatigue 633

RELATING CLEAVAGE CRACK NUCLEATION TO CRACKED CARBIDES IN

A533B STEEL

A. Kumar and S. G. Roberts

University of Oxford
Department of Materials, Parks Road, Oxford OX1 3PH, U.K.
akshay.kumar@materials.ox.ac.uk, steve.roberts@materials.ox.ac.uk

At low temperatures, ferritic steels exhibit a change in fracture behaviour, from ductile fracture
with high fracture energies accompanied by a high degree of plasticity, to brittle cleavage fracture
whereby fracture occurs at transonic speeds at low fracture energies accompanied by very low
degree of plasticity [1, 2]. This change from ductile to brittle cleavage fracture behaviour occurs
over a range of temperatures, over which the behaviour is characterised by a competition of ductile
microvoid coalescence and brittle cleavage fracture. The ductile to brittle transition temperature is
dependant on strain rate, stress triaxiality due to notch acuity, microstructural and compositional
factors [1, 2, 3]. The pervasiveness of ferritic steels in structural engineering and the nuclear power
industry, means that a thorough understanding of the micromechanism of cleavage fracture is
fundamental for the development of accurate models for predicting fracture stresses and lifetimes
of components.
The ductile to brittle transition behaviour of ferritic steels has been investigated by many
researchers for over half a century. Stroh [2], Cottrell [3] and Lindley et al. [4] have developed
theories describing the micromechanism of cleavage, in other words, the crack nucleation and
propagation that precedes brittle cleavage fracture in these steels. Nowadays it is believed that
cleavage in ferritic steels most probably starts in brittle second phase particles like carbides,
fractured in the stress field of dislocations piled up at the particle ferrite interface. While there is
some experimental evidence confirming fracture of grain boundary cementite during cleavage of
ferritic steels, the evidence in the literature for fractured spheroidal brittle phase carbides dispersed
throughout the ferrite matrix, is rather limited, ambiguous and sometimes contradictory [4, 5].
Moreover, studies on notched ferritic steel specimens by Stroh [2] and Zener [6] have shown that
crack nucleation involving the attainment of a critical value of local plastic strain independent of
the macroscopic applied tensile load, is the most difficult step in the cleavage fracture process. Yet
studies by Hendrickson et al. [7] and Knott [8] have shown the influence of applied tensile stress in
fracture, which suggests that cleavage fracture is propagation controlled whereby the catastrophic
propagation of a microcrack is dependant on a normal applied tensile stress. Hence as Bošansky
and Šmida [5] as well as Wang and Chen [9] point out that the origin of nuclei of cleavage cracks,
the mechanism controlling cleavage fracture and the fracture criterion are still subjects of
discussion.
The purpose of this study was to investigate if, and under what loading conditions, spheroidal
brittle carbides crack and act as the sites for cleavage crack nucleation. Also this study attempts to
identify the controlling event during cleavage fracture and over the brittle to ductile transition
temperatures.
To investigate the role of spheroidal carbide particles in cleavage crack nucleation, uniaxial
tensile tests have been carried out at temperatures from the “lower shelf” up to the transition to
ductile behaviour on notched samples of spheroidised A533B steel, a commonly used reactor
pressure vessel grade ferritic steel. Different spheroidisation heat treatments have been selected in
order to obtain a microstructures with a variety of carbide particle distributions, ferritic grain sizes
and Vickers hardness values. Fracture surfaces and cross-sections of completely fractured as well
634 A. Kumar and S. G. Roberts

as part-fractured specimens have been examined by a Scanning Electron Microscope (SEM) and
Energy Dispersive X-ray analysis (EDX) to identify cleavage nucleation sites and understand
carbide fracture and its propagation into the ferrite. The results of the tests are reported and their
implications for cleavage microcrack nucleation and controlling steps of cleavage fracture process
in steels containing spheroidised brittle particles are discussed.

References
1. Orowan, E., Rep. Prog. Phys., vol. 12, 185-232, 1948
2. Stroh, A.N., Adv. Phys., vol. 6, 418-465, 1957
3. Cottrell, A.H., Trans. Am. Inst. Min. Metall. Petrol. Engrs., vol. 212, 192-203, 1958
4. Lindley, T.C., Oates, G. and Richards, C.E., Acta Metall., vol. 18, 1127-1136, 1970
5. Bošanský, J. and Šmida, T., Mat. Sci. Eng., A vol. 323, 198-205, 2002
6. Zener, C., The micromechanism of fracture, In Fracturing of Metals, American Society for
Metals, Cleveland, Ohio, 1948
7. Hendrickson, J.A., Wood, D.S. and Clarke, D.S., T. Am. Soc. Metal, vol. 50, 656-676, 1958
8. Knott, J.F., J. Iron Steel Inst., vol. 204, 104-111, 1966
9. Wang, G.Z. and Chen, J.H., Int. J. Fracture, vol. 89, 269-284, 1998
 2T32. Micromechanisms in fracture and fatigue 635

MICRO-ENERGY RATES FOR DAMAGE TOLERANCE AND DURABILITY OF

COMPOSITE STRUCTURES

C. C. Chamis and L. Minnetyan

NASA Glenn Research Center, Clarkson University
Cleveland, OH 44135, Potsdam, NY 13699
Christos.C.Chamis@nasa.gov, levon@clarkson.edu

For effective structural health monitoring for durability, it is important to quantify damage
tolerance of a candidate structure. Since continuous fiber composites are able to arrest cracks and
prevent self-similar crack propagation, composite structures have received a great deal of
consideration for design with emphasis on damage tolerance. However, a number of design
parameters such as fiber orientation patterns, choices of constituent material combinations, ply
drops and hybridization, result in complex design options for composite structures. Thus, it is
necessary to evaluate damage initiation in a composite structure and its fracture propagation
characteristics for achieving a rational damage tolerant design.
Compared with homogeneous materials, damage initiation and progression characteristics of
fiber composites are much more complicated. Composite structures often contain some pre-
existing or induced flaws in matrix and fibers after fabrication of composites. At lower stresses,
matrix is likely to be cracked because of flaw-induced stress concentrations and cause the matrix
flaws to propagate across the composite. With the use of established material modeling and finite
element modes, and considering the influence of local defects, through-the-thickness cracks and
residual stresses, computational simulations have made it possible to evaluate the details of
progressive damage and fracture in composite structures. In a computational simulation, damage
evolution quantifier such as the damage volume, exhausted damage energy, and the Damage
Energy Release rate (DERR) are used to quantify the structural damage tolerance at different
stages of degradation. Low DERR levels usually indicate that degradation takes place with minor
resistance by the structure. Structural resistance to damage propagation is often dependent on
structural geometry and boundary conditions as well as the applied loading and the state of stress.
In certain cases, such as the room temperature behavior of composites designed for high
temperature applications, internal damage initiated as microcracks in the matrix become enlarged
to be externally visible. Thus, matrix cracking and its effect on damage propagation/damage
tolerance need be evaluated. Some simulations have been successful in predicting damage
tolerance and failure load of composite structure by considering ply stresses and the corresponding
stress limits for matrix crack growth. Damage initiation, growth, accumulation, and propagation to
fracture are studied. Since the complete evaluation of ply and subply level damage/fracture
processes is the fundamental premise of computational simulation, a microstress level damage
index is added for the identification and tracking of subply level damage processes. Computed
damage regions are similarly correlated with ultrasonically scanned damage regions. Simulation is
validated by comparison with test data from Acoustic Ultrasonic (AU) testing. Results show that
computational simulation can be used with suitable Non-Destructive Evaluation (NDE) methods
for credible in-service monitoring of composites. A complete set of micromechanics equations
used is published elsewhere for evaluating microstresses.
Computational simulation is implemented via integrating three modules: (1) composite
mechanics, (2) finite element analysis, and (3) damage progression tracking. The composite
mechanics module is designed to analyze fiber composite structures with an updated composite
mechanics theory. Its main function is to calculate ply and composite properties of laminates from
636 C. C. Chamis and L. Minnetyan

the fiber and matrix constituent characteristics and the composite lay-up. Additionally, it
determines the composite structural response and ply stresses from the FEM analysis results. In
simulation, the composite mechanics module is called before and after each finite element analysis.
The finite element analysis module is able to process linear and nonlinear static and dynamic
analysis. Four-node anisotropic thick shell elements are usually used to model laminated
composites. The finite element analysis module accepts laminate properties from the composite
mechanics module and performs the structural analysis at each load increment. After structural
analysis, the computed generalized node stress resultants and deformations are provided to the
composite mechanics module. The composite mechanics module computes the developed ply
stresses for each ply and checks for ply failure modes at each node. Failure criteria applied to
detect ply failures are based on the maximum stress and modified distortion energy (MDE) criteria
for combined stress effects.
The overall evaluation of composite structural durability is carried out in the damage
progression module that keeps track of composite degradation for the entire structure. The damage
progression module relies on the composite mechanics module for composite micromechanics,
macromechanics and laminate analysis, and calls the finite element analysis module for global
structural analysis. If excessive damage is detected, the incremental loads are reduced and the
analysis is restarted from the previous equilibrium stage. Otherwise, if the increment of loads is
acceptable, another finite element analysis is performed but the constitutive properties and the
finite element mesh are updated to account for the damage and deformations from the last
simulation. Simulation is stopped when global structural fracture is predicted.
If the ply is subject to combined stresses, its microstresses are obtained by simply
superimposing results of all corresponding stress components. Ply transverse fractures usually
begin in the matrix between fibers due to the elevated stress levels from stress concentration.
Microstress level damage tracking is able to quantify the type of damage in the matrix by
comparison of microstresses with constituent stress limits. A microstress damage index is defined
as a binary number with 14 bits in the damage progression module.
In this paper, the adhesive bond strength of lap-jointed graphite/aluminum composites is
examined by computational simulation. Computed microstress level energy release rates are used
to identify the damage mechanisms associated with the corresponding acoustic emission (AE)
signals. Computed damage regions are similarly correlated with ultrasonically scanned damage
regions. Results show that computational simulation can be used with suitable NDE methods for
credible in-service monitoring of composites.
 2T32. Micromechanisms in fracture and fatigue 637

MICROMECHANICAL OBSERVATION OF FRACTURE PROCESS IN

MORTARS

E. Schlangen and O. Copuroglu

Delft University of Technology, CiTG, Microlab,
P.O. Box 5048, 2600 GA Delft, The Netherlands
e.schlangen@citg.tudelft.nl

Cement-based materials like concrete and mortars are quasi-brittle materials. These materials show
a softening behaviour when tested in uni-axial tension. An example of such a behaviour is shown
in Fig. 1, which is a result obtained from a tensile test on mortar in the special micro-tensile testing
machine explained in this paper. The shape of the curve and the area under the curve which is a
measure for the fracture energy, is strongly related to the heterogeneity of the material, see for
instance Schlangen and van Mier [1]. At the weak spots in the material (generally the interface at
grain boundaries) first micro-cracks develop in the material when a load is applied. Subsequent
loading will result in a localized crack that will propagate through the material. This mechanism is
believed to take place in all cement based materials, but is hard to observe. Stable fracture in a
tensile test can only be realised with a closed loop hydraulic testing machine. Observations of
micro-cracking and localization of these cracks is a difficult task. However to be able to improve
materials a complete understanding of the fracture process is necessary. Furthermore durability is a
hot issue in cement based materials. Ingress of water and ions result in internal reactions of
concrete and embedded steel reinforcement which can result in fast degradation of concrete
structures [2]. Understanding the fracture process, the way micro-cracks and localized cracks
propagate through the material, is thus important in order to control transport through the material
and design the material in such a way that a long service life of structures is guaranteed.

FIGURE 1. Typical observed load-deformation curve of mortar.

The present study deals with a new developed technique to be able to observe the fracture
process (what happens at the crack tip) during a stable tensile test. A micro-tensile testing machine
as shown in Fig. 2 is used to test various mortar materials. The mortars are chosen in such a way
that components in the mortar have different local properties. This will then result in different
fracture mechanisms and also a different load-deformations response. The tensile tests are
performed on small specimens (15x15 mm cross section) with one notch. The crack mouth
opening displacement measured on the specimen is used as feed back signal to control the test. The
micro-tensile testing machine is constructed in such a way that it fits insight an ESEM. With the
638 E. Schlangen and O. Copuroglu

ESEM online images of the development of micro-cracks and propagation of the crack tip are
obtained.

FIGURE 2. Micro tensile testing machine(left) and SEN-mortar specimen (right).

The results of the various tests will be presented in the paper. A relation will be made between
the local properties in the material and the fracture mechanism that takes place. The results of this
investigation will provide very useful information to verify numerical models that simulate
fracture processes on the micro-scale of cement-based materials, see for instance Schlangen and
Copuroglu [3].

References
1. Schlangen, E. and van Mier, J.G.M., Cem. Conc. Composites, 14: 105-118, 1992.
2. Francois, R. and Arliguie, G., Magazine of Concrete Research, 51, 2, pp. 143-150, 1999.
3. Schlangen, E. and Copuroglu, O., In Proceedings of the 5th International Conference on
Computation of Shell and Spatial Structures, June 1-4, 2005, Salzburg, Austria
 2T32. Micromechanisms in fracture and fatigue 639

MICRO-FRACTURE MAPS IN PROGRESSIVELY DRAWN PEARLITIC

STEELS

J. Toribio and F. J. Ayaso

Department of Materials Engineering, University of Salamanca
E.P.S., Campus Viriato, Avda. Requejo 33, 49022 Zamora, Spain
Tel: (34-980) 54 50 00; Fax: (34-980) 54 50 02
toribio@usal.es

One of the major concerns in civil engineering construction and maintenance of prestressed
concrete structures is the fracture performance of the high-strength prestressing steel wires which
are the fundamental components in the afore-said composite material (prestressed concrete), since
these wires suffer the highest levels of stress and may be damaged by the combined action of both
mechanical and environmental agents.
High-strength prestressing steels are manufactured from a previously hot rolled bar with
pearlitic microstructure which is heavily cold drawn in several passes to obtain the commercial
prestressing steel wire with increased yield strength obtained by a strain-hardening mechanism.
Thus the final commercial product has undergone strong plastic deformations able to modify its
microstructure. Thus, although cold drawing improves the (traditional) mechanical properties of
the steel (i.e., those properties useful for regular service), the microstructural changes during
manufacture [1-3] may affect the fracture performance of the material, specially in the presence of
stress raisers like cracks or notches.
In this paper the fracture performance of axisymmetric notched samples taken from
prestressing steels with different levels of cold drawing is studied. To this end, a real manufacture
chain was stopped in the course of the process, and samples of all intermediate stages were
extracted. Thus the drawing intensity or straining level (represented by the yield strength) is treated
as the fundamental variable to elucidate the consequences of the manufacturing route on the
posterior fracture performance of the material. In addition, since samples with very different notch
geometry are considered, the effect of stress triaxiality on fracture performance is also analysed.
Fractographic analysis of the samples by means of scanning electron microscope (SEM)
showed the microscopic topographies after failure and allowed the assembly of micro-fracture
maps (MFM) covering the whole fracture surface and containing relevant information on the
micromechanisms of fracture in the material, as shown in Fig. 1 for one of the notched geometries
(that with the maximum stress triaxiality). In the matter of the influence of the stress triaxiality
(constraint) on fracture processes in the vicinity of notches, the behaviour is more ductile in the
case of blunt notches than in the case of sharp notches. With regard to the influence of the cold
drawing degree, the fracture behaviour becomes more ductile as the strain hardening level of the
steel increases, so that cleavage is predominant in the fracture area of slightly drawn steels,
whereas the presence of micro-void coalescence (MVC) increases with the degree of cold drawing
and becomes predominant in the case of the most heavily drawn steels in which cleavage can
hardly be detected as an extended region.
640 J. Toribio and F. J. Ayaso

FIGURE 1. Micro-fracture maps (MFM) in notched specimens

References
1. Dewey, M.A.P. and Briers, G.W., J. Iron and Steel Inst., vol. 2, 102-103, 1966.
2. Embury, J.D. and Fisher, R.M., Acta Metall., vol. 14, 147-159, 1966.
3. Langford, G., Metall. Trans., vol. 1, 465-477, 1970.
 2T32. Micromechanisms in fracture and fatigue 641

A BRIEF HISTORY OF FRACTOGRAPHY

S. P. Lynch and S. Moutsos

Dept. of Materials Engineering, Monash University
Clayton, Vic. 3168, Australia
stan.lynch@dsto.defence.gov.au

Fracture surfaces have, no doubt, been studied throughout the history of mankind, probably
starting with observations on stone-age tools. In the 16-18th centuries, the macroscopic
appearance of fracture surfaces was used to assess the quality of metallic materials, with studies by
Réaumur in 1722 [1] being the most notable (Fig. 1(a)). However, it was not until 1943 that
fracture surfaces were first examined at high-magnifications (using optical microscopy up to
1,000x) (Fig. 1(b)) (Zapffe and Moore [2]), and that the first attempts were made to examine
replicas of fracture surfaces using transmission-electron microscopy (TEM) (Barrett and Derge
[2]). Early replicas had poor fidelity and resolution, and it was not until 1956 that Crussard et al.
[3] pioneered high-resolution TEM fractography using shadowed, direct-carbon replicas (Fig. 2).
This technique (and its subsequent variations) revolutionised fractography and led to a plethora of
studies in the 1960’s and 70’s. It therefore seems appropriate to commemorate the 50th
anniversary of high-resolution electron fractography with a review of how it, and subsequent
scanning electron microscopy (SEM) and other techniques, have led to a better understanding of
fracture processes. Such understanding has been invaluable in failure analysis and in developing
improved materials. Milestone observations for a number of important modes of fracture in inert
environments including cleavage, brittle intergranular fracture, dimpled overload fractures, and
fatigue fractures, are described first, followed by examples of key observations for fractures
produced in embrittling environments (Fig. 3). Fractographic features and aspects of fracture
modes that are not well understood are also discussed.

FIGURE 1. (a) Sketches by Réaumur [1] showing a cleavage fracture, described as “little mirrors
of irregular shape and arrangement” - one of seven classes of iron identified by him according to
their fracture surface-appearance in 1722, and (b) Optical micrograph of cleavage fracture of iron
showing river lines and twins (Zapffe and Moore, 1943 [2]).
642 S. P. Lynch and S. Moutsos

FIGURE 2. Two of the first high-resolution transmission electron micrographs of fracture surfaces
of steels showing (a) fine river lines on a cleavage fracture, and (b) a dimpled overload fracture
surface (Crussard et al., 1956 [3]). Such observations led to improved understanding of ductile
and brittle behaviour, and clearly showed that dimples resulted from nucleation, growth, and
coalescence of nano/micro-voids.

FIGURE 3. SEM of fracture surfaces produced by cracking of a high-strength martensitic steel in

(a) liquid mercury, and (b) air, showing that crack growth can occur by a more localised
microvoid-coalescence process in embrittling environments than in air. Such observations (in both
hydrogen and liquid-metal environments) led to the development of localised-plasticity theories
for environmentally assisted cracking (e.g. Beachem, 1972 [4], Lynch, 1988 [5]) involving, for
example, adsorption-induced dislocation-emission (AIDE).

References
1. de Réaumur, R. A. F., see Smith C.S., A History of Metallography, MIT Press, 1988.
2. Zapffe, C. A. and Moore, G.A., Trans AIMME, vol. 154, 335-359, 1943, and discussion
therein by C.S. Barrett and G. Derge, 353-355.
3. Crussard, C., Borione R., Plateau J., Morillon Y. and Maratray F., J. Iron and Steel Inst., vol.
183, 146-177, 1956.
4. Beachem, C. D., Metall. Trans., vol. 3, 437-451, 1972.
5. Lynch, S.P., Acta Metall., Overview no. 74, vol. 36, 2639-2661, 1988.
C. SPECIAL SYMPOSIA/SESSIONS

C1. Nanomaterials and Nanostructures

 1. Fracture and Fatigue at the Micro and Nano scales 645

SIZE EFFECTS IN LEAD FREE SOLDER-JOINTS

A. Betzwar-Kotas, G. Khatibi, A. Ziering, P. Zimprich, V. Groeger, B. Weiss

and H. Ipser1
Institute of Materials Physics, University of Vienna
Strudlhofgasse 4, A-1090 Vienna, Austria
1Institute of Inorganic Chemistry, University of Vienna

Waehringerstrasse 42, A-1090 Vienna, Austria

brigitte.weiss@univie.ac.at

Due to the environmental and health concerns the usage of lead based solders in electrical and
electronic equipment is restricted. Recent regulations require the replacement of these solders by
non toxic lead free solders, thus numerous scientific and technical investigations has been
performed on the properties of the solder joints produced using these new materials [1].
Investigations on the effect of specimen size on properties of the bulk solders materials have
shown that the bulk data can be used only if the solder joint size is larger than a representative
volume [2]. Thus available bulk material data are not reliable if one considers the extremely small
volumes (<10-12 m³) being used in new microelectronic devices. While several investigations have
been conducted on the thermomechanical integrity and reliability of solder joints, very few
systematic experimental studies on the effect of the reduced solder size and geometry on the
mechanical performance are available [3].
In this study the effect of gap size (or volume) on tensile and fatigue behaviour of solder joints
was investigated. Model solder joints of Sn3.5Ag /Cu and Sn10InAg / Cu with gap sizes varying
between 50 - 800 µm were prepared by the reflow process. Selected specimens were aged at
suitable time and temperatures to study the influence of microstructure and the IMC (intermetallic
compound) growth on the mechanical response of the joints. SEM (scanning electron microscopy)
investigations were performed to determine the thickness of the IMC layer and the evolution of the
microstructure of the solder. Tensile testing was performed to assess the bonding strength of the
joints. The strain was measured by a non contacting laser speckle extensometer with a gauge
length of 15mm. The actual elongation of the solder joints was calculated considering the initial
thickness of the solder gap. The obtained stress-strain curves showed a strong effect of gap size on
tensile properties of the solder joints (Fig. 1). Fracture surface investigations of the tensile samples
gave an insight into the failure mechanisms of the solder joints and their dependence on the
thickness and morphology of the IMC s, solder gap size and their ratio.
The fracture sensitivity and crack growth behaviour of the solder joints as a function of the gap
size was studied using a special fatigue testing technique. The applied experimental set up allows
in- situ observation of the crack nucleation sites and the crack propagation path in the solder or
solder joint. The results of the fatigue experiments showed a dependence of the crack growth
behaviour on the gap size. The fracture path followed mainly the IMC / solder interface which is
known as the location of highest stress concentration in the joints. Increasing the thickness of the
solder gap lead to an enhanced degree of plastic deformation in the solder volume which in some
cases resulted a deflection of fracture path from the solder / IMC interface into the volume of the
solder.
646 A. Betzwar-Kotas et al.

FIGURE 1. Dependence of tensile strength to solder gap size.

Our experimental results of the mechanical size effects in the micro solder joints is supported
by a theoretical model describing the relationship between the mechanical properties and
characteristics size and structure of the solder joints.

References
1. Plumbridge W. J., Matela R. J., Westwater A. Structural Integrity and reliability in
Electronics, Kluwer Academic Publishers, London, 2003
2. Bonda N.R. and I.C. Noyan, IEEE Transactions on Components, Packaging, and
Manufacturing Technology – part A, vol. 19, issue 2, 208-212, 1996
3. Shen Y.-L. et al., Acta Materialia, vol. 53, issue 9, 2633-2642, 2005
 1. Fracture and Fatigue at the Micro and Nano scales 647

MICRO-SCALE SIMULATION OF IMPACT RUPTURE IN POLYSILICON

MEMS

Alberto Corigliano, Fabrizio Cacchione, Attilio Frangi and Biagio De Masi1

Department of Structural Engineering. Politecnico di Milano
Piazza Leonardo da Vinci, 32. 20133 Milano, Italy
1MEMS Business Unit. STMicroelectronics.

Via Tolomeo 1, 20010 Cornaredo, (Milano), Italy

alberto.corigliano@polimi.it, cacchione@stru.polimi.it, attilio.frangi@polimi.it,
biagio.demasi@st.com

The issue of Micro Electro Mechanical Systems (MEMS) mechanical reliability is becoming
increasingly important. This implies that sufficiently accurate methodologies to determine the
mechanical response of MEMS subject to various loading conditions must be developed.
This paper focuses on the problem of the mechanical response of polysilicon MEMS to impact
loading provoked typically by accidental drop of devices in which MEMS are introduced.
An explicit dynamic FE code was developed to simulate the mechanical responses of LPCVD
and PEVCD polysilicon specimens produced with THELMA™ process, Corigliano et al. [1],
subject to impact loading.
A Voronoi tessellation (Espinosa et al. [2], [3], Ballarini [4]) is first performed in order to
divide the polysilicon specimen into virtual crystalline grains. The 2D FE mesh is subsequently
obtained by means of an automatic mesh generator using six-noded triangular elements (Fig. 1).

FIGURE 1. Voronoi tessellation of a polysilicon specimen.

To take into account the anisotropy of the material, every grain is linear elastic up to rupture
with spatial orientation and elastic constants of single crystal silicon.
By means of a relaxation algorithm which suitably calibrates the damping coefficient the code
can also perform static analyses (Fig. 2).
The ongoing research activity concerns the development and implementation in the code of a
suitably defined interface law to simulate intergranular fracture and rupture criteria for single
crystals to simulate transgranular rupture.

Fig. 2. Deformed and undeformed configuration of a tensile specimen.

648 A. Corigliano et al.

References
1. Corigliano, A., De Masi, B., Frangi, A., Comi, C., J. of Microelectromechanical Systems, vol.
13, 2, 200-219, 2004.
2. Espinosa, H.D., Zavattieri P.D., Mechanics of Materials, vol. 35, 333–364, 2003.
3. Espinosa, H.D., Zavattieri P.D., Mechanics of Materials, vol. 35, 365–394, 2003.
4. Mullen, R.L., Ballarini R., Yin, L., Heuer, A.H., Acta Materialia, vol. 45, 2247-2255 1997.
 1. Fracture and Fatigue at the Micro and Nano scales 649

NANOINDENTATION OF CNT REINFORCED EPOXY NANOCOMPOSITES

Dimitris C. Lagoudas, Piyush R. Thakre and A. Amine Benzerga

Department of Aerospace Engineering, Texas A & M University, College Station, TX
Address: 3141 TAMU, College Station, TX 77843-3141
d-lagoudas@tamu.edu

The nanoindentation technique has been proven to be a useful tool for the evaluation of
deformation behavior at nano and micro scale [1, 2, 3]. Load and displacement sensing
nanoindentation is used for determining mechanical properties such as hardness and elastic
modulus of materials in small volumes. The most widely used methodology by Oliver and Pharr
[4, 5] enables the determination of elastic modulus and hardness by analyzing the unloading
segment of the indentation curve as a purely elastic recovery process. However time dependent
deformation, including indentation creep and viscoplastic behavior accompanies elastic
deformation [2, 3]. Another significant problem with the method is that it does not account for
pile-up of material [1] around the contact impression which is observed in many elastic-plastic
materials. Consequently, the modulus and the hardness values calculated by the Oliver-Pharr
method can be overestimated.
In the present work, we investigate the influence of the above factors on the elastic modulus
and the hardness of Carbon Nanotubes (CNT) reinforced epoxy specimens of small size. The test
has been performed on neat epoxy (EPON 862) and single walled carbon nanotubes (SWCNT)
reinforced epoxy composites. Two types of SWCNTs were investigated: pristine and silane-
functionalized with varying weight percentages. Micro-structural characterization was performed
using a high resolution SEM and TEM. Silane functionalization helped in better dispersion of
single walled carbon nanotubes as evident from SEM micrograph in Fig.1.

FIGURE 1: FE-SEM micrographs of SWCNTs before reinforcing in epoxy matrix (a) Pristine
SWCNTs (b) Silane functionalized SWCNTs.

A 90° Cube Corner tip was used for nanoindentation and atomic force microscopy mode on
nanoindenter. Specific experiments were performed to find the appropriate holding time, loading
and unloading rate to minimize the effects of time dependent deformation. Optical microscopy and
atomic force microscopy were performed to observe the indent impression and the material pile-
up.
650 D. C. Lagoudas et al.

FIGURE 2: Nanoindentation results showing variation of elastic modulus and hardness with
contact depth

One of the main results of this study is that there is a very small influence on the elastic
modulus and hardness of the nanocomposite as a result of functionalization and the degree of
dispersion of SWCNTs (Fig. 2). Another important result, also shown in Fig. 2 is that both
modulus and hardness increase with decreasing indentation depth below a critical value of about
500 nm. This size effect suggests a scale transition in the mechanisms that govern both elastic and
inelastic deformation in these materials.

References
1. Oliver, W.C., Pharr, G.M., J. Mater. Res. , vol.19, No. 1, 3-20, 2004.
2. Li, X., Gao, H., Scrivens, W.A., Fei, D., Xu, X., Sutton, M.A., Reynolds, A.P., Myrick, M.L.,
Nanotech. , vol. 15, 1416-1423, 2004.
3. Dutta, A.K., Penumadu, D., Files, B., J. Mater. Res., vol. 19, 158-164, 2004.
4. Oliver, W.C., Pharr, G.M., J. Mater. Res. , vol. 7, No. 6, 1564-1583, 1992.
5. Fischer-Cripps, A.C., Nanoindentation, Springer, NewYork, USA, 2002.
 1. Fracture and Fatigue at the Micro and Nano scales 651

DIFFUSION KINETICS AND MULTIVARIANT PHASE TRANSFORMATION IN

SHAPE MEMORY ALLOYS

D. R. Mahapatra and R. V. N. Melnik

Mathematical Modelling and Computational Sciences,
Wilfrid Laurier University, Waterloo, ON, N2L3C5, Canada
droymahapatra@wlu.ca, rmelnik@wlu.ca

We develop a general modeling framework to analyze the dynamics of multivariant phase

transformation in three-dimensional samples of shape memory materials. The model is based on a
systematic representation of the Gibbs free energy in stress, temperature and n-dimensional order
parameter space. Here n indicates the number of martensitic variants (M). Landau theory has
earlier been successfully applied to solve the problem of phase-field evolution in phase
transforming solids (see e.g. the works by Artemev et al.[1] and Ichitsubo et al. [2] and the
references therein). In the context of diffusion-less transformation in solid described by the Landau
theory of first-order kinetics, the order parameters are the scaled ensemble average of the atomic
order variables [3]. The diffusion kinetics based on the time-dependent Ginzburg-Landau equation
is thus the homogenized form of the spatio-temporal fluctuation due to atomic reordering. Our
main objective in this paper is to study how this diffusion kinetics influences the stress-temperature
induced dynamics of phase transformation in microscopic and larger length-scale without
attempting to solve a molecular dynamic problem in a coupled manner. Toward this direction we
finally arrive at the thermodynamic conservation law whose characteristics is controlled by the
phase kinetics.
As a starting point we first address the problem of 3D free energy representation in context of
n-variant martensitic transformation. Our free energy model [4] follows the steps similar to those
obtained in the works of Levitas and Preston [5,6] to arrive at the invariance and uniqueness
properties of the free energy. Also the number of order parameters can be reduced with the help of
the symmetry properties across the transformation surfaces. In addition to this we relax some of the
conditions involving many unknown material constants, which finally leads to a precise
description of austenite-martensite transformation energy barrier and variant-variant interaction
energy. In our approach the previously observed non-physical minima of the free energy wells are
eliminated. We perform detailed analysis of the proposed free energy model in context of
austenite-martensite (A-Mk) transformation and variant-variant (Mi-Mj) transformation with sharp
interfaces. A link between the evolution of texture using this approach and the classical condition
for microstructural compatibility within the framework of continuum deformation theory [7] is
discussed. Analytical characterization is then carried out on the existence of diffused interfaces
under quasi-static stress-temperature conditions by considering time-independent Ginzburg-
Launday equation. Results are compared with those reported by Levitas et al. [8] in context of
critical nuclei formation in NiAl. Next we turn our attention to the interplay between the diffused
states of the variants and the nonlinear thermoelasticity. Relationship between the physical time
scale of the atomic reordering process and the time scale at which the parbolicity of the
thermodynamic conservation is influenced and completely switched to elliptic problem is
analyzed. Numerical simulations on the evolution of microstructure are reported. Method of
introducing dislocation mechanism within the present framework will be discussed.
652 D. R. Mahapatra and R. V. N. Melnik

References
1. Artemev, A., Wang, Y. and Khachaturyan, A.G., Three-dimensional phase field model and
simulation of martensitic transformation in multiplayer systems under applied stresses, Acta
Mater. Vol. 48, 2503-2518, 2000.
2. Ichitsubo, T., Tanaka, K., Koiwa, M. and Yamazaki, Y., Kinetics of cubic to tetragonal
transformation under external field by the time-dependent Ginzburg-Landau approach,
Physical Rev. B, vol. 62, 5435, 2000.
3. Borg, R.J. and Dienes, G.J., An Introduction to Solid State Diffusion, Academic Press,
London, 1988.
4. Mahapatra, D.R. and Melnik, R.V.N., A dynamic model for phase transformations in 3D
samples of shape memory alloys, LNCS. Springer-Verlag, vol. 3516, 25-32, 2005.
5. Levitas, V.I. and Preston, D.L., Three-dimensional Landau theory for multivariant stress-
induced martensitic phase transformations. I. Austenite o martensite, Physical Rev. B, vol.
66, 134206, 2002.
6. Levitas, V.I. and Preston, D.L., Three-dimensional Landau theory for multivariant stress-
induced martensitic phase transformations. II. Multivariant phase transformations and stress
space analysis, Physical Rev. B, vol. 66, 134207, 2002.
7. Bhattacharya, K., Microstructure of Martensite, Oxford University Press, 2003.
8. Levitas, V.I., Preston, D.L. and Lee, D.-W., Three-dimensional Landau theory for
multivariant stress-induced martensitic phase transformations. III. Alternative potentials,
critical nuclei, kink solutions, and dislocation theory, Physical Rev. B, vol. 68, 134201, 2003.
 3. Nanoscale Deformation and Failure 653

EBSD ANALYSIS ON DEFORMATION OF NANOCRYSTALS

IN ECAP-PROCESSED COPPER

Hidehiko Kimura, Yoshiaki Akiniwa1, Keisuke Tanaka1 and Takaaki Ishida2

EcoTopia Science Institute, Nagoya University, Nagoya, Japan
h_kimura@mech.nagoya-u.ac.jp
1Department of Mechanical Science and Engineering, Nagoya University, Nagoya, Japan
2Graduate School of Mechanical Science and Engineering, Nagoya University, Nagoya, Japan

Nanocrystalline materials produced by severe plastic deformation techniques have captured great
attention because of the excellent mechanical properties. Equal channel angular pressing (ECAP)
is especially attractive because the simple technique can produce large bulk samples with little
residual porosity, elevating the strength without adding expensive alloying elements. Materials
processed by ECAP consist of nanoscopic and submicron grains. Though ECAP holds many
advantages in terms of practical application over other severe plastic deformation process that can
produce only nanocrystalline surface layers, the deformation and damage mechanism in the
characteristic microstructure of ECAP-processed materials is not clarified. For the detailed
microscopic observation, electron backscattering diffraction (EBSD) technique is a useful tool that
enables investigation based on crystallographic orientation. In this paper, EBSD analysis was made
on pure copper processed by ECAP in order to investigate the mechanism of deformation and
damage in the microstructure with nanoscopic and submicron structures, with special attention to
the grain boundary sliding of nanocrystals.
An orientation image microscope was utilized for the analysis of crystallographic orientation
with the scanning resolution of 30 to 50 nm. The inverse pole figure map of as-received pure Cu is
presented in Fig. 1. Many twin boundaries are observed. Figure 2 is the IPF map of ECAP-
processed Cu after 12 passes by route Bc. No twin boundaries are observed after ECAP. The
transgranular strain is expected to be small especially in small grains with little orientation change
within the grains. The average grain diameter, d, decreased from 350 Pm of as-received Cu to less
than 200 nm. The tensile strength was doubled from 205 MPa to 400 MPa.

FIGURE 1. IPF map (as-received). FIGURE 2. IPF map (ECAP). FIGURE 3.Change of
GOS.

The average of the misorientation spread within one grain calculated from EBSD analysis is
the grain orientaion spread, GOS. If GOS is large, the grain is expected to hold large transgranular
strain. Figure 3 shows the change of GOS with the number of ECAP pass, M. With increase in M,
GOS decreases indicating that the recrystallization is more readily completed after 12 passes than 4
654 H. Kimura et al.

passes. The ratio of high angle boundaries is also larger after 12 passes than 4 passes. Therefore,
the following investigation was conducted on Cu with 12 ECAP passes.
EBSD observation was made on the surface of tensile specimen at plastic strain, Hp, of 0, 4, 8
and 16 %. The change of orientation, T, was measured for all the grains at each Hp. Figure 4 shows
the results of the representatives with 3 nanoscopic grains (solid marks) and 4 relatively coarse
grains with d > 1 Pm (open marks). With the increase in Hp, the orientation change tends to increase
regardless of the grain size. It is clear that T is larger in nanoscopic grains than in larger grains
especially at higher Hp. The result of all the measured grains is shown in Fig. 5 for T between Hp = 0
and 16 % in relation to d. The orientation change is obviously larger for smaller grains. Below the
range of d < 1 Pm, the orientation change tends to decrease linearly with d. Large T in small d
region is considered to result either from the increase in misorientation within the grains, observed
as plastic deformation by slip, or from grain rotation by grain boundary sliding without increase in
the transgranular misorientation. Figure 6 shows the change of GOS between Hp = 0 and 16 % in
relation with d for all the measured grains. 'GOS slightly increases with d, indicating that plastic
deformation by slip is more likely to be introduced in coarser grains. Judging from the results in
Fig. 5 and 6, it is concluded that grain rotation occurs among nanoscopic grains in ECAP-
processed Cu. The rotation angle of nanoscopic grains is measured to be about 4 degrees under Hp
= 16 %.
EBSD observation was found to be useful for the quantitative analysis of deformation
mechanism in ECAP-processed materials with nanoscopic and submicron grains. By the
combination of EBSD and atomic force microscopy, grain boundary sliding mechanism among
nanoscopic grains will be clarified. The fatigue damage accumulation process and crack
propagation behavior will also be precisely investigated taking into consideration the effect of
microstructure such as crystallographic orientation, grain boundary and subsequent local strain
distribution.

FIGURE 4. Orientation change T vs H. FIGURE 5. T vs d. FIGURE 6. 'GOS vs d.

 3. Nanoscale Deformation and Failure 655

THE EFFECT OF EXTENSIONAL STRAINS ON MOLECULAR

ORIENTATION, POLYMER FREE VOLUME DISTRIBUTION AND
CRYSTALLIZATION

Hai Dong1,2, Ruilan Guo1 and Karl I. Jacob1,3

1School of Polymer, Textile and Fiber Engineering
2 School of Material Science and Engineering
3G.W.W School of Mechanical Engineering

Georgia Institute of Technology

Atlanta, GA 30332-0295
hai.dong@mse.gatech.edu, gt5546a@mail.gatech.edu, karl.jacob@ptfe.gatech.edu

Mechanical properties and morphological attributes of polymeric fibers and films are highly
depended on amount of draw induced in them [1]. Morphological characteristics, such as free
volume distribution, crystallite size and orientation are nanoscale attributes that determine the
mechanical and physical properties of the resulting polymeric material. In particular, stretching
enhances the crystallization rate by orders of magnitude. Thus, understanding the changes in
nanoscale morphological parameters when a polymer sample is subjected to axial extensional
strains is of significant importance in predicting mechanical properties as well as to design
polymeric materials for specific applications.
Two nanoscale aspects will be explored: free volume evolution and nucleation and growth of
crystallites. Free volume concept has long been used to understand mechanical and rheological
properties as well as to explain and interpret many aspects related to polymers such as the glass
transition, chain dynamics, physical aging and transport behavior [2-4]. The amount of free
volume will affect many commercial polymer shaping operations, such as fiber spinning and
injection molding [5]. Although the total free volume offers a quantitative basis for relating some
properties of polymers [5,6], the distribution of free volume may also play a significant role in
governing the underlying physics, such as in diffusion. Consequently, there have recently been
some efforts in characterizing the free volume distribution by experimental techniques, utilizing,
for example, positron annihilation lifetime (PAL) spectroscopy [7]. These experimental techniques
are sometimes laborious and involve some ambiguities in interpreting the results quantitatively in
terms of free volume distribution. In this work, the nature of free volume evolution and its
distribution in a linear model polymer resembling polyethylene under extensional strain, using a
combination of Molecular Dynamics (MD) and Voronoi tessellation has been investigated. It was
observed that free volume is highly dependent on the probe size. When molecular orientation due
to stretch increases, the total number of voids in the sample decreases along with the number
average void size, while the number of larger unoccupied regions in the polymer increases which
become more elongated due to stretch. The Voronoi tessellation shows that although the overall
free volume is decreasing during stretching, the reduction is not evenly distributed within the
region. Free volume associated with atoms located away from the ends of molecular chains is
decreasing while the free volume associated with atoms located at the molecular ends increases
with stretch. Results from this computational work are in good agreement with several
experimental observations reported in the literature.
Nucleation and crystal growth in polymers are significantly enhanced by extensional strains,
resulting in better mechanical properties. Nucleation of a single molecular chain was compared
with stress induced crystallization of polymer in the bulk, showing differing kinetics of nucleation
and crystal growth as well as nanoscale structural differences in the resulting morphology. The
656 H. Dong et al.

evolution of crystallinity as well as the resulting crystalline orientation were quantified with MD
simulations for various extensional strain levels. Thus, patterns of relationship between stain,
molecular orientation, degree of crystallinity and amount of free volume were developed.
Although MD simulations were carried out at very high strain rates, the results show good
qualitative agreement with experimental results.
Amount of free volume in polymers is very small, of the order of 1%, using the experimental
probe size in PAL. During the nucleation and growth phase, crystalline regions are also very small.
However, such nanoscale morphological structures have shown to have significant influence on
the behavior of polymers. The effect of axial extensional strains on the evolution of these
nanoscale structures and the mechanism for enhancement of polymer mechanical properties will be
the focus of this talk.

References
1. Dong, H., Jacob, K.I., Macromolecules, vol. 36, pp. 8881-8885, 2003.
2. Wang, B., Wang Z. F., Zhang M., Liu W. H., Wang S. J., Macromolecules, vol. 35, 3993-
3996, 2002.
3. Haraya, K., Hwang S. T., J. Membr. Sci., vol. 71, 13-27, 1992.
4. Kobayashi, Y., Haraya, K., Hattori, S., Sasuga, T., Polymer, vol. 35, 925-928, 1994.
5. Engelsing, K., Mennig, G., Mechanics of Time-Dependent Materials, vol. 5, 27, 2001.
6. Doolittle, A. K., J. Appl. Phys., vol. 22, 1471, 1951.
7. Doolittle, A. K., The Technology of Solvent and Plasticizers, Wiley, New York, 1954.
8. Kobayashi Y., Zheng W., Mwyer E. F., McGervey J. D., Jamieson A. M., Simha R.
Macromolecules, vol. 22, 2302-2306, 1989.
 3. Nanoscale Deformation and Failure 657

MICROROTATION-AUGMENTED ENERGY-MINIMIZATION FOR 3D

NANOCRYSTALLINE CU STRUCTURES

M. A. Tschopp and D. L. McDowell

School of Materials Science and Engineering
Georgia Institute of Technology
Atlanta, GA 30332-0245
gtg874p@mail.gatech.edu

Molecular dynamics simulations are frequently used to study deformation mechanisms at the
nanoscale. However, the lattice orientations of grains in the starting nanocrystalline (nc)
configurations are typically based on either random orientations (Van Swygenhoven and Derlet
[1]; Schiøtz et al. [2]) or orientations specifically chosen to be high angle grain boundaries
(Yamakov et al. [3]), which may not be either global or local minimum energy configurations.
Our hypothesis is that global or local minimum energy configurations are more representative of
actual nc grain structures.
Consequently, the objective of this research is to use atomistic simulations to study the grain
boundary character and deformation response for more energetically realistic initial configurations
in 3D nc-Cu with an algorithm that augments traditional methods for forming polycrystals during
the energy minimization process with microrotational perturbances of the lattices.
In our simulations, we examine the effect of lattice microrotations on the generation of 3D
Voronoi-constructed nc grain and grain boundary structures with an embedded-atom method
potential for Cu (Mishin et al. [4]). The microrotation algorithm couples a modified Monte Carlo
method with molecular statics to show that an initial grain structure with high angle grain
boundaries will evolve towards a lower energy grain structure via the introduction of small
rotations of grains prior to energy minimization for each Monte Carlo step. In this way, symmetric
and asymmetric coincident site lattice (CSL) boundaries naturally become more prominent within
the nc sample due to their lower energy; this is particularly relevant to nc grain ensembles where
the surface energy is dominant.
Fig. 1 shows a schematic of the microrotation algorithm used to generate the starting nc grain
structure for deformation simulations. In this schematic, the initial configuration is generated with
random grain lattice orientations. The microrotation algorithm randomly selects a grain and
rotates the lattice orientation to create a trial configuration. Through an iterative process, the
minimum energy grain boundary structure for this rotation is found using a nonlinear conjugate
gradient energy minimization algorithm. Last, an energy criteria decides whether to accept or
reject the trial microrotation based on a comparison of the potential energy of the trial
configuration to that of the previous configuration. This procedure repeats until the configuration
converges to either a global or local energy minimum.
The simulation results compare an initial configuration generated using random grain lattice
orientations with a configuration that was evolved using the microrotation algorithm to show how
the microrotation algorithm affects two specific parts: (i) the nanocrystal grain boundary
character statistics and (ii) the inelastic deformation response of Cu. The first part examines the
effect of the microrotation algorithm on CSL grain boundary statistics in nc grain structures and
compares the CSL statistics computed with the microrotation algorithm to experimentally
determined CSL grain boundary statistics. The second part compares the effect of the
microrotation algorithm on the atomistic deformation characteristics of nc grain structures; we
incorporate different mean grain sizes to compare the effect of the microrotation algorithm in
658 M. A. Tschopp and D. L. McDowell

nanoscale regimes dominated by either: (i) grain boundary-mediated deformation processes or (ii)
dislocation nucleation and emission processes.

FIGURE 1. Flow chart for the microrotation algorithm.

The significance of these results is that generating and deforming a more energetically realistic
nc configuration provides a more accurate understanding of grain boundary character and
deformation mechanisms with decrease of grain size into the nc regime. In addition, the
microrotation algorithm addresses CSL content in nc materials for atomistics in a more realistic
fashion.

References
1. Van Swygenhoven, H., Derlet, P.M. and Hasnaoui, A., Physical Review B, vol. 66, 024101-1-
4, 2001.
2. Schiøtz, J., Vegge, F., Di Tolla, F.D., Jacobsen, K.W., Physical Review B, vol. 60, 11971-
11983, 1999.
3. Yamakov, V., Wolf, D., Phillpot, S.R. and Gleiter, H., Acta Materialia, vol. 49, 2713-2722,
2001.
4. Mishin, Y., Mehl, M.J., Papaconstantopoulos, D.A., Voter, A.F. and Kress, J.D., Physical
Review B, vol. 63, 224106-1-16, 2001.
 3. Nanoscale Deformation and Failure 659

MECHANICS AND ELECTROMECHANICS OF SINGLE CRYSTALLINE

PIEZOELECTRIC NANOWIRES

Min-Feng Yu, Zhaoyu Wang, Jie Hu and Abhijit Suryavanshi

Department of Mechanical and Industrial Engineering, University of Illinois at Urbana-
Champaign, Urbana, IL 61801, USA.
mfyu@uiuc.edu

Significant interest exists for the scientific study and the commercial application of piezoelectric
materials with nanoscale dimensional properties [1-4]. With the recent advance in the synthesis of
piezoelectric nanomaterials of high crystalline quality and low dimensionality [5-6], it is now
possible to study the size and dimensionality effect in mechanics and electromechanics related to
piezoelectricity. We report here such a study with single crystalline barium titanate (BaTiO3)
nanowires. Specifically, the mechanical property and the piezoelectric behaviour of individual
BaTiO3 nanowires having diameters down to several tens nanometer are studied with
piezoresponse force microscopy (PFM) and with tensile electromechanical testing method.
The BaTiO3 nanowires utilized in the study are synthesized through a solid state reaction route
at high temperature in sodium chloride medium [5]. Fig.1 shows the representative electron
microscopy images of the nanowires. The excellent size uniformity and high crystallinity are
shown. Additionally, in situ TEM Energy Dispersive X-ray spectroscopy has confirmed that that
the nanowire is composed of Ba, Ti and O with Ba/Ti ratio close to 1.

FIGURE 1. (a) Scanning electron microscopy image of an individual BaTiO3 nanowire; (b)
Transmission electron microscopy image of a nanowire showing the clearly resolved perfect lattice
structure in nanowire; (c) Diffraction image taken from an individual nanowire showing its high
crystallinity.

In order to reveal the nanoscale piezoelectric behavior in nanowire, PFM, which is a technique
developed based on atomic force microscopy (AFM), is used. In PFM, a small ac testing signal is
applied across the sample with a conductive AFM tip. Due to the converse piezoelectric effect, the
applied electrical field induces a vertical or shear local displacement of the sample, which induces
a vertical or torsional deflection of the AFM probe cantilever. The induced displacement (down to
as small as 0.1 pm) can be measured with an ultrahigh sensitivity lock-in amplifier. PFM thus
allows the detection of magnitude and phase of the piezoelectric response, which then allows the
determination of the piezoelectric constants, as well as the local polarization state and domain
structure in the sample. Figure 2 shows the measured piezoresponse results from an individual
660 M.-F. Yu et al.

BaTiO3 nanowire having a diameter of 95nm. The measured d11 and d15 piezoelectric responses
indicate that the BaTiO3 nanowire is primarily polarized along the length direction. (We define the
direction perpendicular to the length direction as direction 1 and the length direction as direction 3
following the traditional nomenclature.) The d15 response is close to 60 times high as the d11
response, indicating the strong shear response behaviour of the nanowire, a property similarly
existed in bulk BaTiO3 crystal [7]. Additional PFM studies, such as domain switching dynamic
study, hysteresis loop study and PFM imaging, have confirmed that BaTiO3 nanowire is polarized
along the length direction and has minimal polarization at zero bias along the direction
perpendicular to the length direction.
For the quantitative study of the mechanical and piezoelectric properties of BaTiO3 nanowire,
a uniaxial tensile electromechanical testing flexure stage is developed. Integrated with the stage
include a miniaturized capacitive displacement sensor and a MEMS based force sensor, which
provides the sub-nanometer displacement resolution and nano-Newton force resolution required
for the electromechanical characterization of BaTiO3 nanowire. The description of this
measurement tool and the result obtained from the tensile testing will be further detailed in the talk.

FIGURE2. The acquired d11 (a) and d15 (b) piezoelectric response curve.

References
1. Naumov, I. I., Bellaiche L. and Fu, H., Nature, 432, 737, 2004
2. Dawber, M., Jung D. J., and Scott, J. F., Appl. Phys. Lett. 82, 436, 2003.
3. Ghosez P.and Rabe K. M., Appl. Phys. Lett. 76, 2767, 2000.
4. Alexe M., Harnagea C., Hasse D., and Gösele U., Appl. Phys. Lett. 79, 242, 2001.
5. Urban, J. J., Yun W. S., Gu Q., and Park, H. J. Am. Chem. Soc. 124, 1186 (2002).
6. Mao, Y., Banerjee S., and Wong S. S., J. Am. Chem. Soc. 125, 15718, 2003.
7. Jaffe B., Piezoelectric Ceramics, Academic Press, London, 1971.
 3. Nanoscale Deformation and Failure 661

MULTISCALE SIMULATION FOR HIGH SPPED PROPAGATION OF

DISORDERED REGIONS

W. Yang, X. Li and Z. Guo

Department of Engineering Mechanics, Tsinghua University
Beijing, 100084, China
yw-dem@mail.tsinghua.edu.cn

We propose a new multi-scale simulation scheme which seamlessly combines the conventional
molecular dynamics (MD) with the continuum mechanics formulated under the material point
method (MPM). In MPM, modified interpolation shape functions are adopted to reduce artificial
forces on the background hierarchical grids. The method is applicable to several kinds of potentials
including the Lennard-Jones, EAM and a bonding-angle related potential for silicon. Examples of
high energy Cu-Cu and Si-Si cluster impact are presented. The kinetic energy of the cluster is the
critical process parameters. The evolution of displaced atoms is found to depend on the underlying
lattice structures. For the case of Cu-Cu cluster impacts, stacking faults play an important role. The
displaced atoms, visualized in the method of “local crystalline order”, propagate in an anisotropic
manner. This implies the anisotropy in energy transformation process through impacts with multi
interactions among cluster and surface atoms. With the help of the present multi-scale scheme, the
computation capacity implemented in a personal computer could reach a system composed of 900
millions atoms. The case of Si-Si cluster impacts is also examined where the damage spreads in a
more isotropic manner. For the case of cluster impact on a beam with pre-existing cracks, the high
energy impact may seal the crack with amorphous matters and induces an efflux of atoms ahead of
the crack mouth, as shown in Figure 1.

FIGURE1. Local crystalline orders at 2.5ps (left), 5.0ps (center) and 7.5ps (right) after impact

A multiple time step algorithm, called reversible reference system propagator algorithm, is
introduced for the long time molecular dynamics simulation. In contrast to the conventional
algorithms, the multiple time method has better convergence, stability and efficiency. The method
is validated by simulating free relaxation and the hypervelocity impact of nano-clusters. The time
efficiency of the multiple time step method enables us to investigate the long time interaction
between lattice dislocations and low-angle grain boundaries. By use of the MTS method, we
examine the convergence and stability of rRESPA method, and show its performance in some
different systems. The MTS method obviously has better convergence and stability than the
662 W. Yang et al.

conventional algorithms. Moreover, this method updates the slow degrees of freedom less
frequently than the fast ones, and thus speed up the simulation. By simulating the interaction
between lattice dislocations and GBs, one observes not only the complete interaction process of
dislocations and GBs, but also phenomena such as dislocations transmission through the GB,
dislocations absorption into the GB, and dislocations coalescence into a micro-crack. These
phenomena will provide help in the study of the interaction of dislocations and GBs.
The multiple time step method is also used to investigate the generation of dislocation loops
from a Frank-Read source. Setting in a background of a bicrystal lattice, the formed dislocation
loops may pile-up at the grain boundary and then reinitiate at the neighbouring grain. Various
kinds of partial dislocations and branched stacking faults may also manifest, such as shown in
Figure 2, giving a rich variety for the propagation of disordered regions.

FIGURE 2. Snapshot of interaction between lattice dislocation and GB (14.5 ps)

 3. Nanoscale Deformation and Failure 663

SURFACE-STRESS-DRIVEN PSEUDOELASTICITY AND SHAPE MEMORY

EFFECT AT THE NANOSCALE

W. Liang and M. Zhou

Woodruff School of Mechanical Engineering
Georgia Institute of Technology, Atlanta, GA 30332
min.zhou@me.gatech.edu

The pseudoelastic deformation of some shape memory alloys (SMAs) such as Au-Cd, Au-Cu-Zn,
Cu-Zn-Al, and Cu-Al-Ni proceeds through the reversible movement of twin boundaries (Ren and
Otsuka [1]). The behavior of these materials is commonly referred to as rubber-like due to its
resemblance to the behavior of soft and pseudoelastic rubber (Otsuka and Wayman [2]). A similar
behavior and a shape memory effect (SME) are discovered in single-crystalline Cu nanowires with
lateral dimensions between 1.76 and 3.39 nm through molecular dynamics simulations. This
behavior at the nanoscale is due to reversible crystallographic lattice reorientations through the
movement of twin boundaries within the FCC crystalline structure (Fig. 2), allowing Cu nanowires
to exhibit recoverable strains of up to more than 50% which are well beyond the recoverable
strains of 5-8% of most SMAs. The reorientation is driven by high internal stresses resulting from
the surface stress and high surface-to-volume ratios of the nanowires. This phenomenon only
occurs in nanowires within the size range of 1.76-3.39 nm and is not observed in bulk Cu.
Furthermore, it is temperature-dependent and hence gives rise to an SME. Specifically, the critical
temperature for spontaneous reorientation upon unloading increases from 100 to 900 K as the wire
size increases from 1.76 to 3.39 nm, making it possible to design nanoscale components of varying
sizes for operation over a wide range of temperature. Such an objective is more difficult to achieve
with conventional bulk SMAs since their transition temperatures (martensite start and finish
temperatures, austenite start and finish temperatures) only vary with material structures and
composition, not size. Moreover, the nanowires have very short response times which are on the
order of nanoseconds due to their extremely small dimensions compared with bulk SMAs. These
unique properties can lead to important applications at the nanoscale, including sensors,
transducers, and actuators in nano-electromechanical systems (NEMS).
664 W. Liang and M. Zhou

FIGURE 1. Reversible lattice reorientations upon loading and unloading in single-crystalline

Cu nanowires; (a) original <110>-axis/{111}-surface wire with rhombic cross-sections, = 70.5°
and = 109.5°, (b) stretched <001>-axis/{001}-surface wire with square cross-sections, (c) (110)
atomic plane containing the > @ wires axis and the long diagonal ([001]) of the rhombic cross-
section in the original wire, (d) the same (110) atomic plane after lattice reorientation, containing
the new wire axis ([001]) and a diagonal ( [110] ) of the new square cross-section.
Atomistic simulations have also yielded the same rubber-like behavior and SME in Au
nanowires with similar levels of recoverable strains and different critical temperatures and critical
sizes. Since other FCC metallic (e.g., Pt and Ag) nanowires have similar structures and properties
as those of Cu and Au nanowires (Rodrigues and Ugarte [3]), it is tempting to speculate that a
similar behavior may also be found in them, with comparable levels of recoverable strains as those
for Cu and Au wires. Of course, their critical temperatures and critical sizes would depend on their
surface stress levels and the differences between their {001} and {111} surface energies. If such an
effect is proven true, these materials could provide a whole class of nano building-blocks with
unique rubber-like behaviors and SMEs operative over a wide range of temperatures and sizes.

References
1. Ren, X. and K. Otsuka, Nature, vol. 389, 579-582, 1997.
2. Otsuka, K. and Wayman, C.M., Shape Memory Materials, Cambridge University Press, New
York, 1998.
3. Rodrigues, V. and Ugarte, D., In Simulation and Modeling of Mechanical Deformation of
Nanowire in Nanowire Materials, edited by Z.L. Wang, Kluwer Academic/Plenum
Publishers, New York, 2003, 177-209.
 3. Nanoscale Deformation and Failure 665

THERMOMECHANICAL BEHAVIOR OF ZINC OXIDE NANOBELTS

A. Kulkarni and M. Zhou

G. W. Woodruff School of Mechanical Engineering
Georgia Institute of Technology
801 Ferst Drive, Atlanta, GA 30332-0405, U.S.A.
min.zhou@me.gatech.edu

Belt-like ZnO nanostructures with rectangular cross-sections have recently been synthesized
through vapor deposition (Pan et al.) [1]. These nanobelts can serve as building blocks for
functional nano-systems. The integration of these nanostructures in systems requires
understanding of their inherent properties, functionalities and behavior. An atomistic framework is
developed to evaluate the thermomechanical behavior of these nanobelts. Molecular dynamics
simulations are performed to characterize the response to uniaxial tensile loading. The ultimate
tensile strength, strain at failure and Young’s modulus are obtained as functions of temperature,
size and growth orientation. The results are compared with the behavior of bulk ZnO.

FIGURE 1. (a) unit cell of ZnO, (b) orientations of nanobelts, and (c) loading scheme

The nanobelts in three growth directions are generated by assembling the unit wurtzite cell

along the [0001] , [0110], and [2110] crystalline axes, see Fig. 1(a)-(b). The interactions
between the atoms are described by a Buckingham type potential which accounts for both
electrostatic and short range pair-wise effects. Following the geometric construction, dynamic
relaxation is carried out to yield free-standing nanobelts at 300 K. Two distinct configurations are
observed. When the lateral dimensions are above 10 Å, nanobelts with rectangular cross sections
are observed. Below this critical size, tubular structures involving two concentric shells similar to
double-walled carbon nanotubes are observed in the [0001] and [0110] orientations. The tubular
structures have low energy [0001] lateral surfaces, indicating a transformation driven by surface
energy minimization.
Approximation to quasi-static tensile loading is achieved through the specification of boundary
velocities at the ends of the nanobelt during loading steps and subsequent relaxation at fixed
lengths during equilibrating steps (Fig.1 (c)). Deformations of belts with [2110] and [0110]
orientations exhibit three stages, including initial elastic stretching, wurtzite-ZnO to graphitic-ZnO
666 A. Kulkarni and M. Zhou

structural transformation, and cleavage fracture. On the other hand, [0001] belts do not undergo
any structural transformation and fail through cleavage along [0001] planes. Calculations show
that the ultimate tensile strength (UTS) and Young’s modulus of the belts are size-dependent and
are higher than the corresponding bulk ZnO values. Specifically, as the lateral dimensions decrease
from 30 to 10 Å, significant increases between 38-76% and 31-53% are observed for the UTS and
Young’s modulus, respectively, as shown in Table 1. This effect is due to the size-dependent
compressive stress induced by tensile surface stress in the nanobelts (Diao et al.) [2].

TABLE 1. Young’s modulus and UTS as functions of nanobelt lateral dimensions (T = 300 K)

[0110] and [2110] nanobelts with multi-walled tubular structures have higher values of
elastic moduli (~430 GPa) and UTS (~36 GPa) compared to their wurtzite counterpart, echoing a
similar trend in multi-walled carbon nanotubes. The influence of temperature on properties is also
determined. Between 100 to 1000 K (note that Tm =2250 K ), approximately 18%, 9% and 51%

decreases in the Young’s modulus are observed for the [0001] , [0110], and [2110]
orientations. This observation is consistent with what is reported for bulk ZnO (Gadzhiev) [3].

References
1. Pan, Z.W., Z.R. Dai, and Z.L. Wang, Science, vol 291(9), 1947-1949, 2001.
2. Diao, J., K. Gall, and M.L. Dunn, Nano Letters, vol 4(10), 1863-1867, 2004
3. Gadzhiev, G.G., High Temperature, vol 41(6), 778-782, 2003.
 3. Nanoscale Deformation and Failure 667

NATURAL MODES OF C60 CAGE VIA CARBON-CARBON BONDING

ELEMENT

Pan Zeng, Xue-Gui Yang and Jing Du

Department of Mechanical Engineering, Tsinghua University
Beijing 100084, PR China
zengp@mail.tsinghua.edu.cn

Science and technology related to C60 cage have attracted the attention and research interest of
thousands of scientists and engineers over the past decade (Kroto et al. [1]). This attraction stems
from C60’s beautiful symmetry, dazzling chemical and physical properties, and potential
applications in nanotechnology (Zhang et al. [2]). The simulations are widely used to study or
predict the vibrational behaviours of C60 cage (Wu et al. [3], Jiang et al. [4]). The available
approaches can be mainly catalogued into the group theory model and the ab inito calculation
(Adams et al. [5], Koruga et al. [6]). However, there has not yet been a more completed
investigation for vibration of C60 cage in Raman domain from both the computational model and
the experiments. So a direct computational model of covalent atom structure is our objective.
To this end, a new bond-based direct computational approach is developed to investigate the
mechanical behaviours of C60 cage. This approach works through element technology and
molecular mechanics (Chang et al. [3]), called Carbon-Carbon Bonding Element (CCBE), which is
defined on a bond with two atoms (Fig.1). There are 6 DOFs defined to describe the force-energy
relationship of co-valent bond via molecular mechanics. Further the cluster model with a bond-
based hexagon ring structure is built (Fig.2). Three important characteristics of covalent atom
structure, i.e., length, angle, energy, are totally considered in computational model. The force
constants can be identified by Raman spectra of C60 cage. The CCBE has some important
characteristics, for example, it is a discrete model, and the constants of element matrix can be
directly related to the force constants of co-valent bond. As for the scale of computation, the CCBE
also has an obvious dominance. For a system with N atoms the DOFs will be 3N, and the equation
to be solved will be linear when only considering a small deformation near original equilibrium
position.

FIGURE 1. Representation of CCBE FIGURE 2. The cluster model with CCBE

A C60 cage has a novel structure with 60 carbon atoms. Its 174 DOFs should be corresponding
to total 174 natural modes. But only about 20 orders of modes can be directly detected by means of
Raman spectrum. The modelling technology of Carbon-Carbon Bonding Element is used to
analyze the total natural modes of C60 cage. So all information of natural modes of C60 cage has
been presented and investigated (Fig.3 and 4). The comparison with experiment of Raman
spectrum (Koruga et al. [6]) shows that the modelling method based on Carbon-Carbon Bonding
668 P. Zeng et al.

Element is convenient and efficient in predicting the vibration behaviour of C60 cage (Table 1).
This study represents an effort to develop a precise numerical method for applications in
nanostructure modelling.

FIGURE 3. Ag(1) natural mode of C60 FIGURE 4. Hg(1) natural mode of C60

TABLE 1. The natural frequencies of C60 molecule in the Raman domain

References
1. Kroto, H. W., Heath, J. R., et al., Nature, vol.318, 162, 1985
2. Zhang, Q. L., O’Brien, S. C., et al., Chem. Phy. Letters, vol.90, 525, 1986
3. Wu, Z C, Jelski, D A, and George, T. F., Chem. Phy. Letters, vol.137, 291-294, 1987
4. Jiang, Q., Xia, H., Zhang, Z and Tian, D., Chem. Phy. Letters, vol.192, 93-96, 1992
5. Adams, G. B., Page, J. B., et al., Physical Review B, vol.44, 4052, 1991
6. Koruga, D., Hameroff, S., et al., Fullerene C60, Amsterdam: North-Holland, 1993
 11. Deformation and Fracture at the Nano Scale 669

FRACTURE OF NANOCRYSTALLINE ALUMINUM

C. San Marchi, S. L. Robinson, N. Y. C. Yang and E. J. Lavernia1

Sandia National Laboratories, 1University of California, Davis
7011 East Ave., MS-9402, Livermore CA 94551
cwsanma@sandia.gov

Bulk nanocrystalline (NC) alloys are an exciting new class of materials that have only recently
established their potential for real-world applications. The tensile yield strength of nanocrystalline
5083 aluminum produced by consolidation of cyromilled powders is about twice that of
conventional 5083 as shown in Fig. 1. The material is extruded which accounts for the anisotropy
in tensile properties and has a grain size on the order of 50 nm. The elongation to failure is lower
than for conventional coarse-grained 5083, however not much different from similar high strength
alloys such as 5082-H19: 370 MPa yield strength, 4% elongation [1]. In addition, these materials
are thermally stable; heating tensile bars to 500C for 2 hours results in only modest loss in room
temperature strength.

FIGURE 1. Tensile stress-strain curves of nanocrystalline 5083 aluminum.

The fracture surfaces reveal that deformation is primarily ductile and that second phases
(oxides, silicides and intermetallics) contribute to damage. These second phases are also highly
refined due to the cyromilling process and are typically submicron in size. In addition, these
second phase particles are believed to provide the excellent thermal stability observed for these
nanocrystalline alloys produced by crymomilling. Fracture toughness tests in 3-point bending are
underway and preliminary fracture toughness values are in the range of 15 to 20 MPa m1/2 for
nanocrystalline 5083.
670 C. S. Marchi et al.

FIGURE 2. Fracture surface of nanocrystalline 5083 tested in tension, longitudinal orientation.

References
1. Hatch, J.E., Aluminum: Properties and Physical Metallurgy, American Society for Metals,
Metals Park OH, 1984.
 11. Deformation and Fracture at the Nano Scale 671

WEAR AND FATIGUE IN SILICON STRUCTURAL FILMS FOR MEMS

APPLICATIONS

Daan Hein Alsem, Robert Timmerman1, Eric A. Stach2, Christopher L. Muhlstein3,

Michael T. Dugger4 and Robert O. Ritchie5
Department of Materials Science and Engineering, University of California at Berkeley
Materials Sciences Division, Lawrence Berkeley National Laboratory
One Cyclotron Road MS 72/150, Berkeley, CA 94720
dhalsem@lbl.gov
1Department of Applied Physics, University of Groningen

Nijenborgh 4, 9747 AG Groningen, The Netherlands

r.timmerman@student.rug.nl
2School of Materials Engineering, Purdue University

501 Northwestern Avenue, West Lafayette, IN 47907

eastach@purdue.edu
3Department of Materials Science and Engineering, Pennsylvania State University

202B Steidle Building University Park, PA 16802

clm28@psu.edu
4Materials and Process Sciences Center, Sandia National Laboratory

PO Box 5800, Albuquerque, NM 87185-0889

mtdugge@sandia.gov
5Department of Materials Science and Engineering, University of California at Berkeley

Materials Sciences Division, Lawrence Berkeley National Laboratory

One Cyclotron Road MS 62/203, Berkeley, CA 94720
roritchie@lbl.gov

Recent advances in the application of micromachined structures have increased the demand for
more reliable structures. Stiction, fatigue and microwear [1] are large issues in the reliability of
microelectromechanical systems (MEMS), especially now that designs become more demanding.
Silicon is an excellent construction material at the micro-scale, because of highly developed
processing methods directly related to semiconductor electronics processing and its high strength.
However, it is an inherently brittle material and reliability is the limiting factor as far as
commercial and defence applications are concerned. Since the surface to volume ratio of these
structures is very large, surface forces become dominant and classical models for failure modes
cannot always be applied.
Whereas bulk silicon is not susceptible to fatigue, micron scale silicon is. This poses
considerable limitations to MEMS design and reliability. Muhlstein et al. [2] have posed a
“reaction-layer” fatigue mechanism where a stress induced thickened oxide layer accommodating
moisture assisted cracking is thought to cause fatigue in silicon at the microscale. Even though
more experimental confirmations for this mechanism has been found [3], the question of why the
oxide layer thickens under influence of high local cyclical stresses is still uncertain, and an
alternative fatigue mechanism has been suggested by Kahn et al. [4].
This presentation will present a new series of data on fatigue resonator devices, as used in [2]
and [3], fabricated at the Sandia SUMMITTM polysilicon process. These devices differ from the
previously used devices in that their post-release silicon oxide layers are much thinner (~2 nm vs.
672 D. H. Alsem et al.

~30 nm in previous studies). These results will provide additional insight into the mechanism
causing fatigue of micro-scale polysilicon.
Additionally, detailed wear debris investigation have been conducted on side-wall friction
devices run in ambient air, also fabricated in the Sandia SUMMITTM polysilicon process. Electron
microscopy techniques were used to image and analysis wear debris in these silicon MEMS. This
study showed that the debris particle size is ~50-100 nm and the debris forms agglomerates a large
as ~500 nm. Comparing abrasive wear tracks, wear debris, grain size and surface roughness, it
follows that after an initial adhesive wear mechanism, where ~50-100 nm particles are removed
from the surface by fracture through the grains (size ~400 nm) an abrasive wear mechanism regime
occurs and creates wear plough tracks with the width similar to the size of the debris particle
agglomerates. Surface roughness can aid in adhesion processes, but since it is considerably smaller
(~10 nm) than the wear particles and wear tracks, surface asperity fracture does not appear to
contribute to the wear debris in later stages of wear. It is interesting that the debris particles are
indeed amorphous and have a high oxygen content, but do not completely consist of SiO2. This
fact in combination with the post-release oxide layers in these films being only ~2 nm, tells us that
debris particles are formed by silicon from the film that amorphizes and subsequently partly
oxidizes [5].
Some numerical support for the suggested mechanisms, as well as additional experimental
evidence on sub-surface cracking of the silicon during wear will be presented in order to provide
insight into the potential role of delamination as a micro-scale wear mechanism.

References
1. A. D. Romig Jr. , M. T. Dugger, P. J. McWhorter, Acta. Mater., vol 51, 5837, 2003
2. C. L. Muhlstein, E. A. Stach, and R. O. Ritchie, Acta. Mater., vol 50, 3579, 2002
3. D.H. Alsem, E. A. Stach, C. L. Muhlstein and R. O. Ritchie, Appl. Phys. Lett., vol 86, 41914
- 1-3, 2005
4. H. Kahn, R. Ballerini, J. J. Bellante, and A. H. Heuer, Science, vol 298, 1215, 2002
5. D.H.Alsem, E.A. Stach, M.T. Dugger, M. Enachescu, R.O. Ritchie, Journal of Thin Solid
Films, submitted, 2005.
 11. Deformation and Fracture at the Nano Scale 673

INDENTATION INDUCED THROUGH THICKNESS FILM FRACTURE ON

ENGINEERING ALLOYS

D. F. Bahr, K. R. Morasch and A. Alamr

Mechanical and Materials Engineering, Washington State University
PO Box 642920, Pullman WA 99164-2920 USA
bahr@mail.wsu.edu

In thin film systems, failure often occurs via fracture mechanisms, with either through thickness
cracking or interfacial delamination leading to failure of the device or layer. In spite of this
common failure mechanism, many of the testing methods for analyzing fracture in thin film
systems are focused on the applied problems of thin film adhesion measurements. While this is an
important goal in many industries, particularly in the microelectronics community, for engineering
applications there many situations where film fracture thorugh the thickness of the film, and not
film adhesion, is the controlling factor in the life of a component. Measuring the stress at which
fracture occurs in these thin film systems requires testing methods amicable to both the small scale
of the films as well as the complex relationship between the mechanical properties of the film and
the substrate.
One method for testing thin film fracture is indentation driving fracture. A number of tests,
particularly those of Chechenin et. al. [1] and Weppleman and Swain [2] have shown it is possible
to produce indentation induced cracks in thin hard films on ductile substrates. Other studies have
extended these tests to a broad range of systems; Pang and Bahr studied TiO2 on Ti [3], Rodriguez-
Marek et. al. tested passive films on stainless steels [4], and Morasch and Bahr [5] and Bahr and
co-workers [6] examined Al2O3 on Al.
The current paper will focus on more recent examinations of engineering alloys, particularly
stainless steel and anodized aluminum. These films generate clear indentation induced fracture, as
shown in Fig. 1 for a conical indentation using a 1 m radius of curvature tip to indent a 200 nm
thick anodized aluminum film on samples of 1100 series aluminum. This fracture pattern clearly
occurs outside the contact radius, a, in this case the effective radius of fracture, c, is approximately
2.3 times the contact radius.

FIGURE 1. Indentation Induced Film Fracture in Anodized Aluminum

The load depth curves for these indentations can be analyzed in terms of energy, rather than
stress intensities, as has been done in previous work [3-6]. The load depth curve for an aluminium
substrate, an anodized film which exhibits a small radial crack in the contact region, and a large
674 D. F. Bahr et al.

circumerential crack. Using this energy analysis and the measured elastic properties of the porous
anodic film, the toughness of anodized aluminium oxide is 0.3 J/m2.
This paper will present an energy based analysis of the aluminium system as well as similar
studies of a stainless steel. The stainless steel system was tested during electrochemical film
growth, and shows clear examples of the effects of both alloy and surrounding environment on the
strength of passive films. Both in situ and ex situ testing was carried out to examine the effects of
environment on the testing procedures. Increasing the salt concentration in the electrolyte from
0.01M NaCl to 0.1M NaCl and changing the substrate on which the film was grown from 904L to
304 stainless steels reduced the applied tensile stress needed to fracture the film formed at a
metastable pitting potential from 1.74GPa to 1.63GPa and from 1.76GPa to 1.63GPa, respectively.
The measured strength of films measured in ambient conditions after removal from the electrolyte
was greater than when the films were measured in situ. However, the trends in film strength as a
function of environment varied by are the same between in situ and ex situ testing, suggesting the
two tests are both feasible methods of analyzing environment effects of film strength.

References
1. Chechenin, N.G., Bottiger, J., Krog, J.P. Thin Solid Films vol. 261, 228-235, 1995
2. Wepplemann, E., Swain, M.V., Thin Solid Films vol. 286, 111-121, 1996
3. Pang, M., Bahr, D.F. J. Mater. Res., vol. 16, pp. 2634-2643, 2001
4. Rodriguez-Marek, D., Pang, M., Bahr, D.F., Metall. Mater. Trans, A. vol. 34A, 1291-1296,
2003
5. Morasch, K.R. and Bahr, D.F., J. Mater. Res., vol. 20, 1490-1497, 2006
6. Bahr, D.F., Woodcock, C.L., Pang, M., Weaver, K.D., Moody, N.R., Inter. J. Frac. vol. 119/
120, 339-349, 2003
 11. Deformation and Fracture at the Nano Scale 675

SURFACE NANOSTRUCTURED ALUMINUM BY SEVERE PLASTICS

DEFORMATION

E. I. Meletis, K. Y. Wang1 and J. C. Jiang2

Materials Science and Engineering
The University of Texas at Arlington, Arlington, Texas 76019 USA
1Center for Advanced Microstructures and Devices
2Mechanical Engineering Department

Louisiana State University, Baton Rouge, Louisiana 70803 USA

meletis@uta.edu

Severe plastic deformation (SPD) has been known to be an effective method for producing
nanostructured materials [1]. Examples of SPD methods are mechanical alloying and mechanical
attrition to prepare nanostructured powders. The same concept can be used to develop a
nanocrystalline surface region in bulk materials thus enhancing surface sensitive properties [2].
Recently, we used ultrasonic shot peening (USSP) to surface harden commercially pure
aluminum (Al 1100). In this process, steel balls placed in the vial, vibrate at high frequency and
impact the surface inducing plastic deformation at high strain rate. The repeated multidirectional
impacts onto the material surface result in severe plastic deformation in the near surface layer
leading to nanocrystallization. Microhardness measurements as a function of depth revealed a two-
fold increase in surface hardness followed by a hardness gradient consistent with a gradient in
grain size. The hardness profile was stabilized after 1 hr of processing time.

FIGURE 1. Variation of hardness as a function of depth and USSP processing time at cryogenic
temperature.
676 E. I. Meletis et al.

FIGURE 2. Variation of friction with sliding distance and processing time.

The final hardness (or grain size) reflects the equilibrium between deformation hardening and
recovery from heating generated during processing. Due to low melting point of Al, recovery is
expected to have a significant effect on the minimum grain size attainable by this process impeding
further hardening. In the present study, an attempt was made to suppress the recovery process by
conducting USSP experiments at cryogenic temperatures. The objective was to manipulate the
kinetics and achieve a smaller grain size and thus higher hardness.
Fig. 1 presents the microhardness depth profile as a function of processing time at cryogenic
temperature. The microhardness achieved was 1.0 GPa that is about 25% higher than that obtained
with USSP conducted at room temperature (or 2.5 times higher than that of unprocessed 1100 Al).
Similarly, the microhardness shows a sharp initial increase and reaches a plateau after about 1 hr of
processing. Based on a Hall-Petch relationship, such hardness is expected to correspond to a grain
size of about 25 nm [3]. In our previous room temperature USSP experiments, the grain size at the
surface was found to be around 100 nm.
The tribological response, Fig. 2, shows that a processing time of more than 30 min results in a
rather stable initial friction. Also, a significant reduction in the coefficient of friction for a sliding
distance of about 600 m is observed. Thus, processing at lower temperatures offers the opportunity
to achieve smaller grain sizes and significant enhancement in mechanical properties.

References
1. Koch, C.C., Nanostruct. Mater., vol. 2, 109-129, 1993.
2. Gleiter, H., Prog. Mater. Sci., vol. 33, 223-315, 1988.
3. Zhou, F., Lee, J., Dallek, S. And Lavernia, E.J., J. Mater. Res., vol.16, 3451-3458, 2001.
 11. Deformation and Fracture at the Nano Scale 677

CONTRIBUTION OF LOCALIZED DEFORMATION TO IGSCC AND IASCC IN

AUSTENITIC STAINLESS STEELS

Gary S. Was, Zhijie Jiao and Jeremy T. Busby

University of Michigan
2355 Bonisteel Blvd., Ann Arbor, MI 48109 USA
University of Michigan
2355 Bonisteel Blvd., Ann Arbor, MI 48109 USA
P.O. Box 2008, Oak Ridge National Laboratory
Oak Ridge, TN 37831
gsw@umich.edu, zjiao@umich.edu, busbjt@ornl.gov

It is known that the irradiation-induced microstructrual and microchemical changes in austenitic

alloys may affect the irradiation-assisted stress corrosion (IASCC) susceptibility. However, a
recent study [1] shows that IASCC susceptibility is more likely to be linked to the mechanical
properties such as yield stress and hardness that arise from the microstructrual change. A higher
yield stress generally corresponds to a lower level of uniform elongation but a higher degree of
localized deformation. Localized deformation, induced by low stacking fault energy (SFE) and/or
irradiation damage, may play a key role in IGSCC and IASCC susceptibility of austenitic stainless
steels and nickel-base alloys. Proton irradiation and constant extension rate tensile testing (CERT)
were used to examine the potential impact of SFE and irradiation on deformation mode and
IASCC. Three model alloys (E : UHP-304, H : 304+Si and L : 304+Cr+Ni) having a spread in SFE
were selected for this study. Two batches of samples were irradiated with 3.2 MeV protons at
360°C to 1.0 and 5.5 dpa respectively, and then incrementally strained in 288°C Ar atmosphere to
3%, 7% and 12%. After each strain level, the degree of strain localization, as determined by
channel height, width and spacing were quantified using SEM and AFM on replicas of the surfaces
of the deformed samples. Results were compared to those from SCC tests in simulated LWR
environments to determine if localized deformation is a controlling factor in IASCC.
At 3% strain, irradiation dose has a significant positive effect on channel height. The higher the
dose, the larger is the average channel height, Fig. 1. However, as the strain increases, the
significance of this effect decreases. At 7% strain, the average channel height for alloy E at 1.0
dpa is nearly equal to that at 5.5 dpa. The average channel height follows the order, alloy H > alloy
E > alloy L, at 3% strain for both doses. Alloy L has the lowest height of the three alloys for all
conditions. The average channel width does not change much regardless of strain and dose. As
expected, the spacing between channels decreases with increasing strain levels.
At low dose, the average channel strain increases with increasing strain for all three alloys, Fig.
2. However, channel strain for alloys H and E saturates at 7% strain. Alloy L has the lowest
average channel strain of all three strain levels. Alloy H has the highest average channel strain at
low strain (3%). The magnitude of average channel strain follows the order of alloy H > alloy E >
alloy L. The order is inverse of that for the magnitude of stacking fault energy. The alloy with
lowest SFE has the highest average channel strain. At high dose, average channel strain for alloys
H and E is saturated even at 3% strain. Alloy L is still the lowest of all three alloys.
678 G. S. Was et al.

FIGURE 1. Average channel FIGURE 2. Average channel

height in alloys E, H and L at strain in alloys E, H and L at
doses of 1.0 and 5.5 dpa doses of 1.0 and 5.5 dpa

At low dose, the degree of strain localization in channels increases with increasing strain level
and it saturates at higher strain level. At high dose, strain localization is saturated at low strain
level (3%) for alloys H and E. At low strain levels (<3%), higher dose generally means a higher
degree of strain localization for the same strain level. At low dose, the degree of strain localization
in channels follows the order: alloy H > alloy E > alloy L, which is the inverse order of the
stacking fault energy. Alloy L has the smallest degree of strain localization at all strain levels and
both doses. The alloy with the highest SFE has the least degree of deformation. Alloy H with the
lowest SFE is most susceptible to IASCC, and alloy E has a moderate IASCC susceptibility. Alloy
L with the highest SFE is resistant to cracking. The degree of localized deformation at the point of
crack initiation is most likely to be the controlling factor for IASCC susceptibility.

References
1. K. Fukuya, M. Nakano, K. Fujii, T. Torimaru and Yuji Kitsunai, Journal of NUCLEAR
SCIENCE and TECHNOLOGY, Vol. 41, No. 12, p. 1218-1227Jones, G. A., Applied Fracture
Mechanics, Soringer, Berlin, Germany, 1978.
 11. Deformation and Fracture at the Nano Scale 679

A STUDY OF CRACK-DISLOCATIONS INTERACTION WITH 3D DISCRETE

DISLOCATION DYNAMICS

I. N. Mastorakos and H. M. Zbib

School of Mechanical and Material Engineering
Washington State University
Pullman, WA 99163
mastorakos@wsu.edu, zbib@mme.wsu.edu

Although in recent years there has been an attempt to model crack – dislocation interactions [1, 2],
the work has been limited to two dimensions and for certain special cases. In this work we address
the problem within a three-dimensional discrete dislocation dynamics frame-work coupled with a
dislocation density model for cracks.
Representing 2D cracks by dislocation density distributions is a well-known and widely
applied technique in planar cracks. Dislocation density represents the slope of the crack opening
displacement at a given position on the crack. Thus, the stress field due to the relative displacement
of an infinitesimal segment of the crack surfaces can be used as Green's function which can be
integrated over the crack surface to represent the stress field due to the crack at any given point in
the medium surrounding the crack. Furthermore, with the application of the superposition
principle, the stress-free crack surface condition can be satisfied. This technique can be extended to
3D cracks, where, instead of straight dislocations, infinitesimal dislocation loops are considered
(for a review of this technique see [3, 4]). In this case, the crack plane is considered as a
distribution of infinitesimal dislocation loops with Burgers vector b ( b , b , b ) . To describe
x y z

the displacement field produced by such a dislocation loop of area dS, a Cartesian coordinate
system Oxyz is defined so that the dislocation loop lies in the plane z z c (Fig. 1).
The problem that has to be solved is a planar crack of penny or elliptical shape, which is
embedded in an infinite body and is subjected to an arbitrary loading. Because of the nature of the
loading, the crack will experience both opening and shearing mode displacements. A continuous
distribution of infinitesimal dislocation loops, each of area dS and Burgers vector
b ( b , b , b ) is applied to the crack faces to model the three relative displacements of them.
x y z

The tractions on the crack plane induced by the distribution are:

V 3i ( x ) ³K ij ( x , x c) b j ( x c) d S
S
(1)

where i 13 , j is the number of loops, and Kij ( x, xc) is a known function. The boundary
condition to be satisfied is that the crack faces are stress – free, which leads to the following
integral equation:

³K ij ( x , x c ) b j ( x c ) dS  t i0 ( x )
S
(2)
In order to solve the above equation, we have to deal with the third degree singularity that
appears in the kernel Kij ( x, xc) . This type of singularity is universal, meaning that in all problems
680 I. N. Mastorakos and H. M. Zbib

involving crack representation by dislocations the singularity remains the same regardless of the
geometry concerned. Having solved the integral equation, the interaction of different shape cracks
with different dislocation types (edge, screw and mixed) and various loading types (fatigue, creep,
creep-fatigue, etc.) can be examined with the use of Discrete Dislocation Dynamics [5]. Figure 2
shows the stress distribution around a crack as a result of interaction with a nearby dislocation.

References
1. Van der Giessen, E, Deshpande, V.S., Cleveringa, H.H.M. and Needleman, A., J. Mech. Phys.
Solids, vol. 49, 2133-2153, 2001.
2. Deshpande, V.S., Needleman, A. and Van der Giessen, E., Acta Mater, vol. 51, 1-15, 2003.
3. Hills, D.A., Kelly, P.A., Dai, D.N. and Korsunsky, A.M., Solution of Crack Problems,
Kluwer Academic Publ., Dordrecht, 1996.
4. Weertman, J., Dislocation Fracture Mechanics, World Scientific, Singapore, 1996.
5. Zbib, H.M. and Diaz de la Rubia, T., Int. J. Plasticity, vol. 18, 1133-1163, 2002.
 11. Deformation and Fracture at the Nano Scale 681

NUMERICAL SIMULATIONS AND MEASUREMENTS OF CRACKS

PARALLEL AND NEAR INTERFACES IN GRADED STRUCTURES

Ivar Reimanis, Keith Rozenburg, John Berger, Matthew Tilbrook1 and Mark Hoffmann1
Metallurgical and Materials Enginering and Engineering Division
Colorado School of Mines Golden, CO 80401, USA
1School of Materials Science and Engineering, University of New South Wales, Sydney 2052,

New South Wales, Australia

reimanis@mines.edu

The driving force and mode mixity for cracks near interfaces was examined via experimental
measurements of the crack tip displacement field with moire interferometry, verified using
boundary element modeling, and further explored with finite element modeling. The stress
intensity factor was obtained by a collocation method in which the Westergaard crack tip
expansions are used for displacements [1]. The method of fundamental solutions was employed so
that displacements across the material interface could be used [2]. Boundary element simulations
provide verification of the method and allow a comparison with experiment. Systematic finite
element simulations enable a study of how residual stresses may affect crack trajectories and
driving forces.
Cracks near interfaces experience a stress field that is highly sensitive to the crack location.
The present work examines the major factors that influence crack growth when a crack is parallel
and very near an interface in a bimaterial configuration, such as that shown in Fig. 1. The elastic
mismatch, plastic mismatch and the residual stress gradient typically present in such structures
influence the crack driving force and trajectory. These are studied here with a combination of
experimental and numerical methods.
Conventional methods for applying moire interferometry to analyze cracks near interfaces are
limited because the displacement field on both sides of the interface cannot be utilized. For
example, in the Cu-W composites specimens examined here, the fracture process zone is on the
order of 100 Pm, and so for cracks within that distance from the interface ( e.g., t < 100 Pm in Fig.
1), the size of the displacement field necessary to extract the stress intensity factor through the
Westergaard expansions extends into the material across the interface. Expansions from the
method of fundamental solutions [2] are used to incorporate displacements from material across
the interface, and these expansions are coupled to those from the Westergaard expansions by
imposing continuity conditions across the interface. The process is verified with synthetic data
generated from a simple boundary element model. It is then applied to a test specimen (Fig. 1)
comprising a Cu-W bimaterial composite. Finite elements results on the same geometry are then
used to show that residual stresses play a large role in determining the crack driving force and
trajectory, but that this effect is mitigated by the presence of plasticity.
682 I. Reimanis et al.

FIGURE 1. Four point bend specimen used in fracture experiments and numerical simulations.

References
1. Sanford, R. J. ., Mechanics Research Communications, 6, 289-294, 1979
2. Fairweather, G, and Karageorghis, A. Advances in Computational Mathematics, 9, 69-95,
1998.
 11. Deformation and Fracture at the Nano Scale 683

DEFORMATION AND FAILURE PROCESSES OPERATING IN ULTRA-FINE

GRAIN METALS

K. Hattar, I. M. Robertson, J. Han1, T. Saif1, S. J. Hearne2 and D. Follstaedt2

Dept. of Materials Science and Engineering, University of Illinois, Urbana IL 61801.
1Dept of Mechanical and Industrial Engineering, University of Illinois, Urbana IL 61801.
2Sandia National Laboratory, PO Box 5800Albuquerque, New Mexico, NM 87185

hattar@uiuc.edu, jhhan@uiuc.edu, saif@uiuc.edu, ianr@uiuc.edu, DMFOLLS@sandia.gov,

SJHEARN@sandia.gov

Nanograined metallic materials typically have higher yield and tensile strength, lower stiffness and
better resistance to environmental degradation and wear than their large-grained analogs. We
currently lack an understanding of how the fundamental deformation processes change as a
function of grain size and therefore we interpret the macroscopic response by scaling-down the
processes that operate in large-grained systems. The disparity between model predictions and
experiment suggest we have yet to include the necessary physical processes. Molecular dynamics
computer simulations of deformation and failure processes in nanograined materials have been
used extensively to determine the deformation processes. It has been suggested that for high
stacking-fault energy materials such as Al, the deformation mode changes from perfect to partial
dislocation processes at a grain size of about 18 nm and to grain boundary mediated processes at
smaller grain sizes. A similar transition occurs in low stacking-fault energy materials except no
perfect dislocations are involved in the deformation mechanism. These results have been
summarized in the form of a deformation map[1]. Farkas et al.[2], using MD simulations, have
shown that both intergranular and transgranular fracture can occur and that the fracture process is
accompanied by dislocations and grain boundary mechanisms. Intergranular fracture occurs
through the linkage of nanocracks that form at triple junctions. They also show that grain
boundaries perpendicular to the crack propagation direction can arrest the crack. These
deformation and failure processes remain to be validated experimentally.
Using an in-situ TEM straining device[3], fabricated from silicon using microlithographic
techniques, it is possible to observe the deformation processes and simultaneously measure the
load and displacement for free-standing thin metallic films with a columnar grain structure. The
deformation and failure processes for Al and Au films of different thicknesses have been
investigated[4] and the results from these studies are the subject of this presentation.
For 125 nm thick Al films the average grain size was 130 nm but grains as small as 35 nm and
as large as 420 nm were also present. During tensile testing, no general plasticity was observed in
the gauge section and the fracture mode was intergranular with no accompanying dislocation
activity. Sharp cracks advanced rapidly through regions containing small grains but they were
arrested when a properly oriented large grain was encountered. The crack blunted by extending
along the boundary of the arresting grain. During this blunting process dislocation emission and
deformation twinning occurred in vicinal grains. Figure 2a shows a dislocation structure existing in
only a few grains along the fracture surface. Emission of perfect and partial dislocations occurred
from regions of the grain boundary and not from the crack tip. Voids or microcracks nucleated and
grew on grain boundaries ahead of the arrested crack, and crack advance occurred through linkage
of the microcracks with the primary crack. It was found that the progress of the arrested crack
could be halted in favor of nucleating and propagating a new crack.
For thicker Al films, dislocation activity occurred in the gauge section prior to the initiation of
the failure crack. The failure mode was again intergranular. It was accompanied by dislocation
684 K. Hattar et al.

activity that extended several grains into the gauge section, see Figure 2b. In contrast to the
thinner Al film in which the failure crack was perpendicular to the tensile axis, the failure crack in
the thicker film was at 45º. The thick Al film showed similar behavior to a 200 nm thick Au film.
The possible reasons for the difference in behavior (length scale effect, grain boundary distribution
and type, grain boundary and film chemistry, etc) will be discussed in this presentation.

FIGURE 2. Dislocation structure along fracture surface of a) 125 nm and b) 500 nm thick Al film
deformed in tension at room temperature.

References
1. Yamakov V., Wolf D., Phillpot S.R., Mukherjee A.K., and Gleiter H., Nature Mater. vol. 3,
43 (2004).
2. Farkas D., Van Swygenhoven H., and Derlet P.M., Phys. Rev. B vol. 66, 060101 (2002).
3. Haque M.A. and Saif M.T.A., Sensors and Actuators, A vol. 97-98, 239 (2002).
4. Hattar K., Han J., Saif M.T.A., and Robertson I.M., J. Mater. Res. vol. in press (2005).
 11. Deformation and Fracture at the Nano Scale 685

SIMULATION OF CROSS-SECTIONAL NANOINDENTATION IN

INTERCONNECT STRUCTURES WITH COHESIVE ELEMENTS

D. Gonzalez, J. Molina, I. Ocana, M. R. Elizalde, J. M. Sanchez. J. M. Martinez-Esnaola,

J. Gil-Sevillano, G. Xu1, D. Pantuso1, T. Scherban2, B. Sun1, B. Miner1, J. He1 and J. Maiz1
CEIT and TECNUN (Universidad de Navarra), P. Manuel Lardizabal 15,
20018 San Sebastián, Spain
1Intel Corporation, Hillsboro 97124 (OR), USA
2Tracey.Scherban@hotmail.com

The thermo-mechanical robustness of interconnect structures is a key reliability concern for

integrated circuits. The miniaturization process and the package/silicon interaction result in an
increase of the thermal stresses whilst new low-k materials with degraded mechanical properties
are used, increasing the likelihood of adhesion failure in the interconnect structures.
In this context, the cross-sectional nanoindentation technique (CSN and MCSN) [1-5] has been
recently developed by the authors as a technique to characterize the interfacial adhesion of real
interconnect structures. A Berkovich indenter is used to initiate fracture in the silicon substrate
near the interconnect structure. As a result, the cracks propagate through this structure,
preferentially along the weakest interfaces. An example is shown in Fig. 1, illustrating the outcome
of the experiment for two identical interconnect structures, where the material at the interface
indicated by the arrows was varied. The cracks generated in the corners of the indents propagate
through the interconnect structure and are finally arrested by the ductile copper capping. The
amount of delamination of the interface of interest could be directly used as a qualitative quick
monitor of good/bad adhesion in this case.

FIGURE 1. Examples of the MCSN technique applied to different structures: (a) good adhesion
and (b) bad adhesion of the interface indicated by the arrows.

In this work, these experiments have been modeled using the commercial finite element
program ABAQUS. The final goal is to obtain a numerical tool capable of predicting crack
propagation and crack path in real complex interconnect structures. This paper shows the
calibration of the model by comparison of the simulations with the experimental results obtained
on test samples.
In order to incorporate the fracture behaviour of the various materials and interfaces present in
the structure into the numerical analysis, user-defined elements have been created based on the
cohesive zone theory, which specifies the traction-separation law at the crack front. The cohesive
law has been directly derived from a potential function. Based on this theory, a non-standard finite
686 D. Gonzalez et al.

element (referred here as “cohesive element”) has been implemented as a user defined subroutine
in the general purpose finite element program ABAQUS.
The CSN experiments give insight into the preferred crack paths, as shown in Fig. 1. The
experiments are also used to calibrate the model as follows. Various possible crack paths are
introduced in the model through cohesive elements characterised by their fracture energy and
strength. As a function of these fracture properties, different crack paths are predicted in the
numerical simulations. By comparison of the numerical predictions of applied load and crack path
with the experimental results, the unknown cohesive fracture properties of the relevant interfaces
can be extracted. Fig. 2 shows the numerical simulation of a particular MCSN experiment. The
model is able to predict crack path and sudden changes in direction (kinking) depending on the
combination of mechanical and fracture properties of the materials and interfaces, as observed in
the experiments.

FIGURE 2. FEM modelling result of a particular MCSN experiment.

References
1. Sánchez, J.M., El-Mansy, S., Sun, B., Scherban, T., Fang, N., Pantuso, D., Ford, W., Elizalde,
M.R., Martínez Esnaola, J.M., Martín Meizoso, A., Gil Sevillano, J., Fuentes, M., Maiz, J.,
Acta Materialia, vol. 47, 4405-4413, 1999
2. Elizalde, M.R., Sánchez, J.M., Martínez-Esnaola, J.M., Pantuso, D., Scherban, T., Sun, B.,
Xu, G., Acta Materialia, vol. 51, 4295-4305, 2003
3. Scherban, T., Pantuso, D., Sun, B., El-Mansy, S., Xu, G., Elizalde, M.R., Sánchez, J.M.,
Martínez-Esnaola, J.M., International Journal of Fracture, vol. 119/120, 421-429, 2003
4. Molina, J., Ocaña, I., González, D., Elizalde, M.R., Sánchez, J.M., Martínez-Esnaola, J.M.,
Gil Sevillano, J., Scherban, T., Pantuso, D., Sun, B., Xu, G., Miner, B., He, J., Maiz, J. In
Proceedings of the 11th International Conference on Fracture, Topic 14: Electronic
Materials, Paper 4002, 2005
5. Ocaña, I., Molina, J., González, D., Elizalde, M.R., Sánchez, J.M., Martínez-Esnaola, J.M.,
Gil Sevillano, J., Scherban, T., Pantuso, D., Sun, B., Xu, G., Miner, B., He, J., Maiz, J., In
Mater. Res. Soc. Symp. Proc., Vol. 863 © 2005 Materials Research Society, B1.2.1-B1.2.6,
2005
 11. Deformation and Fracture at the Nano Scale 687

FRACTURE BETWEEN TWO SELF-ASSEMBLED MONOLAYERS

K. M. Liechti and D. Xu
University of Texas, Aerospace Engineering and Engineering Mechanics 210 E. 24th Street,
WRW 110C 1 University Station, C0600 Austin, TX 78712-0235
kml@mail.utexas.edu

The objective of this work was to determine the interfacial toughness of .two silicon surfaces that
were each coated with self assembled monolayers (SAMs) and pressed together. The SAMs were
mixtures of two species. Each SAM made the same covalent bond with the Si, but differed in the
terminal group at the other end on the molecule. One of the SAMs had a very active end group
whereas the other did not. As a result, the degree of adhesion between the surfaces was controlled
by the areal density of the active end group.
Previous studies [1-4] of fracture control using SAMs used specimens in which the SAMs
were deposited on a suitable substrate and interacted at the other end with an epoxy. Thus the
current study removes the epoxy with a view towards simplifying interactions.
Zhuk et al. [1] used methyl (CH3) and carboxy (COOH) terminal groups on 15-carbon
alkanethiols to control adhesion between gold and epoxy. The thermodynamic work of adhesion
of epoxy on the coated surface was linearly proportional to the COOH/CH3 fraction in solution up
to about 80%, and was constant thereafter. A series of superlayer fracture experiments revealed
that the interfacial fracture toughness increased strongly with the thermodynamic work of
adhesion. The rate of toughening increased with the work of adhesion, suggesting that more and
more plastic dissipation was excited in the epoxy layer. Kent et al. [2] and Reedy et al. [3] used
mixed monolayers of dodecyltrichlorosilane (DTS) and bromo-undecyltrichlorosilane (BrUTS) to
control adhesion between aluminum and epoxy. Both make strong covalent bonds with the
aluminum. The methyl terminal group on the DTS again makes weak, non-polar interactions with
the epoxy. The authors indicate that the BrUTS forms an alkyl ammonium bromide compound
with the amine crosslinker that was used to cure the EPON 828 epoxy. As a result, an ionic bond
was achieved with the epoxy through R-NH2+Br-R’ bonds. Between 10 and 20% bromine
termination, there was a strong increase in the tensile and shear strengths of the aluminum/epoxy
interface as determined by cruciform and napkin ring shear experiments. Asymmetric double
cantilever experiments were used to determine the toughness of the interface, which increased
linearly with the bromine fraction. The linear relationship was ascribed to the linear increase in the
thermodynamic work of adhesion with bromine fraction [2]. In contrast to the gold/epoxy
experiments, any plastic dissipation effects were apparently the same for all bromine fractions,
even though the toughness of the aluminum/BrUTS/DTS/epoxy was much higher than that of the
gold/COOH/CH3/epoxy interface in moist environments. The fracture mode-mix in the
asymmetric double cantilever beam experiments was -8° at a reference length of 10 Pm, whereas it
was about 50° in the superlayer experiments. In working with the same mixture of SAMs and an
epoxy, Mello and Liechti [4] found that increasing the bromine content altered the fracture
mechanism and increased the interfacial toughness of a sapphire/epoxy interface. The traction-
separation law for the sapphire epoxy interface became dependent on mode-mix with the addition
of a SAM to the sapphire surface. This was in contrast to the uncoated sapphire specimens whose
traction-separation law was independent of mode-mix, had a 50% higher maximum traction and a
ten-fold lower critical separation. It seemed that the bromine interaction with the epoxy was high
enough to pull epoxy chains from the epoxy matrix.
688 K. M. Liechti and D. Xu

In the current study, laminated beam specimens were fabricated by functionalizing Si(111)
surfaces of small Si beams with mixture self-assembled amine and carboxylic acid monolayers of
low defect density. Each areal density of hydrogen bonds distinguished a separate monolayer
adhesive. The beams were prepared and then pressed together in a controlled environment The
laminated beams were then placed in the miniaturized loading device and loaded in displacement
control while the reactive loads were measured. Crack opening displacements were be measured
by crack opening interferometry [5] using an infrared microscope. As currently configured, the
interferometer measures crack opening displacements (COD) to with a resolution of 30nm, 300nm
from the crack front. The fracture surfaces were probed via XPS, ISS and AFM. The
spectroscopic data was compared with that generated following formation of the monolayer
adhesives and used to determine the locus of failure or the weak link in the various monolayer
adhesives that are considered. AFM was used to determine fracture surface topography.
At the time of abstract submission, preliminary and encouraging results have been obtained.
These are expected to mature by the time the conference proceeding paper is due.

References
1. Zhuk, A. V., Evans, A. G., Hutchinson, J. W. and Whitesides, G. M., J. Mater. Res., 13, 3555-
3564, 1998.
2. Kent, M. S., Yim, H., Sorenson, J., Matheson, A., Reedy, E. D. and Majumdar, B.,
Proceedings of the 23rd Annual Adhesion Society Meeting, 323-325, 2003.
3. Reedy, E.D., Kent, M.S. and Moody, N. R. Proceedings of the 23rd Annual Adhesion Society
Meeting, 502-504, 2003.
4. Mello, A. W. and Liechti, K. M. J. Appl. Mech., to appear 2005.
5. Liechti, K. M. Chapter 4 in Experimental Techniques in Fracture, (ed. J. S. Epstein), VCH
Publishers, New York, 1993.
 11. Deformation and Fracture at the Nano Scale 689

NANOTUBE NANOACTUATOR

Min-Feng Yu, Jie Hu, Zhaoyu Wang and Abhijit Suryavanshi

Department of Mechanical and Industrial Engineering, University of Illinois at Urbana-
Champaign, Urbana, IL 61801, USA.
mfyu@uiuc.edu

Having provided the foundation for sensing and actuation applications, piezoelectricity has been
intriguing subjects for many studies. The application of piezoelectric materials for high-resolution
actuation and positioning in scanning probe microscopy has enabled the manipulation and
characterization of matters at nano- and atomic scale. Its application in high-resolution sensing has
allowed unprecedented ease in control and measurement of force and displacement. However, as
an indispensable part for any technology development, the study of nanoscale sensor and actuator
that make use of piezoelectricity and that take advantage of the significant progress in the
development of nanomaterials is still lacking.
The inherent unique properties of nanomaterials possess potentially better engineering figure
of merits for developing new materials and creating new devices with better performance [1].
Here, we report the study of piezoelectricity in BN nanotube, which reveals that BN nanotube is
piezoelectric. The study corroborates directly with a recent theoretical study by Mele and Král
predicting the existence of piezoelectricity in BN nanotube [2].
BN nanotube, structurally similar to carbon nanotube, was made through high temperature
chemical vapor deposition growth [3]. Its mechanical rigidity was found to be comparable to
carbon nanotube (Young’s modulus around 1 TPa), and its lattice structure was found to be mostly
zigzag. The diameter of the BN nanotubes was between 30 and 150 nm, and its length up to tens
micrometers. Most of the synthesized BN nanotubes were found to be high quality, as shown in
Fig. 1. We prepare a BN nanotube sample for atomic force microscopy (AFM) study by
depositing the nanotubes on an Au coated conductive substrate.

FIGURE 1. Transmission electron microscopy images showing (a) the multi-walled structure and
(b) the highly graphitized lattice structure of an BN nanotube.

Piezoresponse force microscope (PFM), one of the new techniques developed from AFM, was
applied to study the local piezoelectric response in BN nanotube based on the converse
piezoelectric effect. For the measurement, an ac testing signal is applied between a conductive
AFM probe (which is in contact with the top surface of an individual BN nanotube) and the
conductive substrate (which is in contact with the back surface of the BN nanotube). The induced
structural deformation is then directly measured from the deflection of the AFM cantilever with the
aid of high sensitivity electronics. Shown in Fig. 2 is the measured piezoelectric response from a
selected spot on a BN nanotube. A linear response of the piezoresponse amplitude on the applied
690 M.-F. Yu et al.

ac amplitude is clearly seen. This response corresponds to a piezoelectric coefficient of 0.09 pm/V
for this BN nanotube. The piezoelectric response is also found to increase with the increased DC
bias additionally applied on the BN nanotube, as shown in Fig. 3. The piezoelectric coefficient
increases from 1.4 pm/V at DC bias of 2 V, to 1.9 pm/V at 5 V and to 3.4 pm/V at 10 V. Such
dependence is indicative of the improved polarization alignment in BN nanotube according to the
traditional point of view on piezoelectricity. However, for BN nanotube, to understand the exact
origin of the polarization and the observed piezoresponse behavior calls for further investigation.

FIGURE 2. (a) An AFM image showing a 63 nm diameter BN nanotube on an Au coated substrate,

(b) the measured local piezoelectric response from the BN nanotube.

FIGURE 3. The dependence of the piezoelectric response on the applied DC bias.

In this talk, our extensive study on the piezoelectric behavior of BN nanotubes will be further
detailed, which includes the study of the piezoresponse phase change behavior related to the
polarization state, the converse piezoresponse behavior under mechanical load and the direct
piezoelectric effect of individual BN nanotubes.

References
1. R. H. Baughman, et al., Science, 284, 1340 (1999).
2. E. J. Mele and P. Kral, Phys. Rev. Lett. 88, 056803 (2002)
3. C. C. Tang, Y. Bando, and T. Sato, Chem. Phys. Lett. 362, 185 (2002).
 11. Deformation and Fracture at the Nano Scale 691

NANOCRACK DETECTION IN VIBRATING NANOWIRES

R. Ruoff, L. Calabri, N. Pugno, X. Chen, W. Ding and K. Kohlhaas

Department of Mechanical Engineering, Northwestern University, Evanston, IL 60208, USA
Department of Structural Engineering and Geotechnics, Politecnico di Torino,
Corso Duca degli Abruzzi 24, 10129, Italy
Department of Mechanics and Industrial Technology, Universita degli Studi di Firenze,
Via di Santa Marta 3, 50139 Firenze, Italy
r-ruoff@northwestern.edu, nicola.pugno@polito.it, lorenzo.calabri@unifi.it

Crystalline boron (B) nanowires (NWs) have been synthesized by the CVD method with
preformed metal catalyst particles. We have experimentally investigated their dynamical
resonance (i) and mechanical strength (ii). Both the two independent methods suggest the possible
presence of nanocracks in the tested B NWs. Nanocrack detection can in principle be achieved by
analytical calculations quantifying crack position and depth.
(i) The nanodynamical experimental apparatus is summarized in Fig. 1, [1]. The mechanical
resonances of cantilevered B NWs were excited and the resonance peak frequencies measured, Fig.
2. Shifts in the natural frequencies were observed, suggesting the possibility of the presence of
nanocracks; other possible shift causes, such as intrinsic NW curvature, non-ideal clamps,
presence of spurious masses, large displacements, etc., are also discussed. Assuming the existence
of a nanocrack, analytical calculations were performed to quantify its depth and position based on
the measured frequency shifts.
(ii) A newly developed and rapid electron beam induced deposition method was used to clamp
the B NWs and test them in tension inside an SEM with a home-built nanomanipulator [2]. High-
resolution SEM images were acquired at each loading step, and two independent methods of
analysis of each image were used to obtain the corresponding tensile load. The B NW geometries
were measured by TEM after the tests. The stress vs strain, Young’s modulus, and tensile strength
of the B NWs were obtained through data analysis. The strength measurements strongly suggest
the presence of nanocracks in the B NWs. Assuming the existence of a nanocrack, placed at the
fracture section, its depth is quantified by applying Quantized Fracture Mechanics [3] starting from
the measured fracture strengths.
The possibility of detecting nanocracks through mechanical resonance of nanostructures is
thus exemplified. The challenge on the experimental side in the future will be ensuring that other
contributors to shifts in mechanical resonance, are not present or have a strong enough different
functional dependence, that the nanocrack contribution can be “uncovered” if present.
We gratefully acknowledge the grant support from the NSF (#020079; #030450); ONR
#N000140210870 (support for W. Ding), and NASA BIMat URETI # NCC-1-02037 (support for
X. Chen).
692 R. Ruoff et al.

FIGURE 1. Experimental apparatus for driving mechanical resonance of NWs.

FIGURE 2. Mechanical resonance of a curved BNW naturally held at an AFM cantilever tip
without any deliberate fabrication of a clamp.

References
1. Chen, X., Zhang, S., Wagner, G., Ding, W., Ruoff, R.S, Journal of Applied Physics, vol.
95(9), pp. 4823-4828, 2003.
2. Yu, M.F., Lourie, O., Dyer, M.J., Moloni, K., Kelly, T.F., Ruoff, R.S., Science, vol. 287, 637-
640, 2000.
3. Pugno, N.M., Ruoff, R.S., Philosophical Magazine, vol. 84/27, 2829-2845, 2004.
 11. Deformation and Fracture at the Nano Scale 693

FRACTURE OF ATOMIC LAYER DEPOSITED NANOLAMINATE FILMS

N. R. Moody, J. M. Jungk1, T. M. Mayer1, R. A. Wind2, S. M. George2 and W. W. Gerberich3

Sandia National Laboratories, Livermore, CA 94550 USA
1Sandia National Laboratories, Albuquerque, NM 87158 USA
2University of Minnesota, Minneapolis, MN 55455 USA
3University of Colorado at Boulder, Boulder, CO 80309 USA

nrmoody@sandia.gov

Strength, friction, and wear are dominant factors in the performance and reliability of materials and
devices fabricated using microsystem technologies. While adequate for some applications, as-
fabricated strength and wear properties are insufficient for use of these devices in many dynamic
applications. Applying films is one method to enhance device performance and reliability. Atomic
Layer Deposition (ALD) is ideally suited for this application as it creates conformal and potentially
stress free films. [1] Of particular interest are ALD tungsten and alumina films where bulk
crystalline counterparts exhibit high hardness and good frictional resistance. However, recent
studies on single layer ALD tungsten films show they exhibit a pronounced susceptibility to
channel cracking and delamination while single layer ALD alumina films exhibit significantly
lower hardness than desired. [2,3]
We therefore began a study of atomic layer deposited nanolaminate-based composites using
alternating layers of nanocrystalline tungsten and amorphous aluminum oxide nanolaminates to
determine if properties could be tailored to meet microdevice needs. [4] The films were deposited
in four, eight, and sixteen bi-layer systems to form 100 nm thick films on silicon substrates. In all
samples, the alumina layer thickness was held constant at 4 nm and the metal layer was adjusted to
fill the remaining volume. (Table 1) Nanoscratch techniques were then employed to evaluate the
resistance to fracture of each film system at loads characteristic of microsystem operation. These
tests using a 50 nm radius Berkovich diamond tip revealed a strong reverse length-scale effect
where an increasing number of bilayers led to a decrease in fracture resistance. (Figure 1) Unlike
the results from single ALD tungsten films where scratch tests initiated extensive channel cracking
and film debonding, there was no widespread delamination in the nanolaminate films.
A micromechanical method, based on the crack lengths induced by the nanoscratch tests and
applied loads, was employed to provide a first order estimate of fracture toughness for the
nanolaminate films. [5] The stress intensity solution for a mode I crack under these conditions is
given by,

2 pb b 1 b
K Ic sin
1/ 2 a a
Sa  (1)
where p is the pressure opening the crack, b is the radius of the indenter in the groove, and a is the
half-crack length. Using this relationship, the fracture toughness for each nanolaminate film was
determined from the scratch track widths, crack lengths, and nanoscratch test loads. The values are
given in Table 1. These results show that a maximum in the fracture resistance exists between the
single layer blanket film of ALD tungsten and scale of bilayer nanolaminate structure of ALD
tungsten and ALD alumina thereby providing a means to tailor wear resistant film performance.
This work was supported by Sandia National Laboratories, a multiprogram laboratory operated by
Sandia Corporation, a Lockheed Martin Company for the United States Department of Energy's
694 N. R. Moody et al.

National Nuclear Security Administration under contract DE-AC04-94AL85000. WWG also

thanks NSF for its support.

FIGURE 1. Scratch fractures in single layer and bi-layer nanolaminate films. [2,4]

Table 1. Single layer and bi-layer film properties

Film Alumina Thick- Tungsten Thickness Volume Fraction - Fracture Tough-
ness (nm) (nm) Tungsten ness MPa-m1/2
SL W 200 100 0.33
4 BL 4 21.1 84.1 0.46
8 BL 4 8.4 67.7 0.35
16 BL 4 2.1 34.4 0.30

References
1. Mayer, T.M., Elam, J.W., George, S.M., Kotula, P.G. Goeke, R.S., Applied Physics Letters
82, pp. 2883-2885 (2003).
2. Moody, N. R., Jungk, J. M., Mayer, T. M., Wind, R. A., George, S. M., Gerberich, W. W.,
Proceedings ICF11, to be published.
3. N. R. Moody, Sandia National Laboratories, Livermore, CA, unpublished research.
4. Jungk, J. M., Moody, N. R., Mayer, T. M., Wind, R. A., George, S. M., Gerberich, W. W.,
Proceedings ICF11, to be published.
5. Hoehn,J. W., Venktaraman, S. K., Huang, H., Gerberich, W. W., Mater. Sci. Engng, A192/
193, pp. 301-308 (1995).
 11. Deformation and Fracture at the Nano Scale 695

INFLUENCE OF MICROSTRUCTURE, STRENGTH AND ADHESION ON AU

ELECTRODEPOSITS

Nancy Yang, J. Kelly, T. Headley and C. San Marchi

Sandia National Laboratories
nyyang@sandia.gov

Electrodeposition is extremely effective for applying uniform thin metal coatings on a complex
microdevices, which otherwise are inaccessible by other coating techniques. We are actively
developing electrical contact coatings for Ni-based microsystem using the LIGA* microfabrication
process 1-2, Fig. 1. The requirements for the coating are high strength, wear resistance and good
adhesion to the Ni substrate. The candidate coating that meets these criteria is electrodeposited Au
coating (referred to here as Hard Au), plated from a commercial sulfite electrolyte with organic
additive (Technigold 25E). In this study, we focus on the metallurgical factors that determine the
strength of the Au coating material and influence adhesion of the Au coating to the Ni microsystem
substrate.
Experimental results show that this Hard Au derives it high strength from nano-sized features
of its microstructure (Fig. 2). The hardness of the Hard Au is about 2 GPa, which is about 4 times
harder than coatings plated without the organic hardening additive from gold sulfite electrolytes
(Fig.3). Hardness and fracture resistance of the coatings are highly dependent on the combination
of the size of the microstructure and the stress in the coating. The nanocrystallite size (determined
from XRD peak broadening) is 50 nm. In addition, the microstructure exhibits small spherical
pores 10nm in diameter (Fig. 3) in the bulk and clusters of these pores along the Au-Ni interface.
These nanopores are characteristics of the Hard Au coatings, presumably related to the hydrogen
generation and entrainment during electroplating.
The resultant microstructure of the Hard Au coating is dictated by the concentration of the
organic hardening additive in the electrolyte and can be controlled to some extent by varying the
concentration. Hard Au coatings produced from electrolyte with the highest concentration of
hardening additive feature a significantly more complicated nanocrystalline microstructure, with
more twinning and more porosity, in addition to higher hardness. As a result these coatings have
more residual stress and are more susceptible to delamination from the substrate.
In this presentation, we will discuss correlation between microstructure and plating variables.
The effect of microstructure on coating strength and adhesion to the Ni-substrate will also be
discussed.

FIGURE 1. Contact spring made by LIGA* microfabrication process

696 N. Yang et al.

FIGURE 2. Typical microstructure of Hard Au

FIGURE 3. Hardness as a function of hardening additive in Hard Au electrodeposits.

References
1. J. J. Kelly, N. Yang, T. Headley and J. Hachman, J. electrochemical Society, 150 (6) C445-
C450, 2003.
2. C. San Marchi, N. R. Moody, M. J. Cordill, G. Lucadamo, J. J. Kelly, T. Headley, N. Yang:
"Structure-property relationships of Au films electrodeposited on Ni", Materials Research
Society Symposium Proceedings; 2004; v.821, p.5-10.
 11. Deformation and Fracture at the Nano Scale 697

FRACTURE OF SUBMICRON THIN METAL FILMS DURING CYCLIC

LOADING

S. Eve, D. Wang, C. Volkert, N. Huber and O. Kraft

Institut für Materialforschung II, Forschungszentrum Karlsruhe and Institut fur Zuverlässigkeit
von Bauteilen und Systemen Universitat Karlsruhe (TH)
Forschungszentrum Karlsruhe, Postfach 3640, 76021 Karlsruhe, Germany
oliver.kraft@imf.fzk.de

One current trend in microelectronics and MEMS is to develop micro devices based on plastic
substrates. This allows for new applications, for instance, by integrating such flexible devices into
the personal environment like textiles. Furthermore, the use of polymers offers a high potential for
cost effective production for mass products. Depending on the application, devices on plastic
substrates are fabricated by forming the polymer, and depositing and patterning one or several thin
films onto the substrate. One reliability issue in such devices is the stretchability of the thin films
during mechanical and/or thermal loading conditions. Under monotonic loading, however, it has
been shown that cracking of metal films on compliant substrates only occurs, when large strains
(>10%) are reached. In contrast, cyclic mechanical or thermal loading conditions with typical
strain ranges between 0.1% and 1% lead to damage formation such as cracking, surface
roughening and delamination [1,2].
For instance, a micro-spectrometer with a 400 nm thick gold film on 1.28 mm thick PMMA
substrate shows after 5 thermal cycles between -40 to +55°C severe damage. The thin metal film
and the substrate have very different thermal expansion coefficients. When applying a temperature
change, this results in high residual stresses and in cyclic deformation leading to cracking,
roughening, and also delamination of the film, as observed by scanning electron microscopy,
limiting the device lifetime.
The stability of the couple metal-film/polymer-substrate during thermal cycling and a good
comprehension of the relevant damage mechanisms are essential for the prediction and
improvement of the reliability of such systems. This includes an acceleration of the fatigue testing,
since standard thermal cycling tests often used for reliability assessments are restricted to a small
number of cycles and may take several days. Therefore, we employ mechanical methods to
characterize the fatigue behavior of thin metal films on an elastic substrate. These methods include
uniaxial tensile loading of metal films on polyimide substrates as described in [2]. Beside that, we
have developed a fatigue experiment, which allows for testing in an equi-biaxial loading condition.
This is to simulate the strain state that occurs during a thermal cycle in a given temperature range.
It is based on the ring-on-ring test; the specimen is supported by a ring and a concentric ring of
smaller diameter loads the opposite face. It produces a biaxial tensile strain and stress state inside
the smaller ring. For the film on substrate specimen, support and loading rings are applied on both
sides of the specimen in order to allow the mechanical simulation of thermal fatigue test with
alternating tensile and compressive strain in the film. This new experimental method has been used
to characterize the fatigue behaviour of films for different material combinations; i.e. aluminum
and gold as the film materials sputtered onto PolyMethylMethAcrylate (PMMA) and
PolyCarbonate (PC) for the substrate.
In both type of tests, uni-axial as well as equi-biaxial loading of the film/substrate composites,
the substrate material is expected to deform elastically while the film may undergo plastic
deformation in both tension and compression. Owing to this sample structure, the film stress
cannot be determined from the externally applied load. However, the stress-strain behavior of the
698 S. Eve et al.

films has been analyzed for several samples using in-situ X-ray diffraction during uni-axial
straining of the samples. The development of the fatigue damage has been studied during
interrupted tests, the samples were investigated using various techniques, including optical
microscopy, scanning, and transmission electron microscopy (SEM, TEM) as well as focused ion
beam microscopy (FIB). Quite recently, a study on the effect of length scale (both film thickness
and grain size) on fatigue-induced damage morphology in Cu films revealed that the dimensions of
the fatigue extrusions decreased with decreasing film thickness from 3.0 µm to 200 nm and that the
characteristic dislocation structures, such as those found in bulk fatigued metals, were only found
in the largest grains of the 3.0 µm thick film [3,4,5]. In contrast, only individual dislocations were
observed in films thinner than 1 µm. For a deeper understanding of these mechanisms, a more
detailed investigation of the influence of film thickness in the regime of 50 to 500 nm on the
fatigue mechanisms is currently being performed.

References
1. Kraft, O., Schwaiger, R., and Wellner, P., Material Science. Engineering A, vol. 319-321,
919, 2001
2. Mönig, R., Keller, R.R., and Volkert, C.A., Review of Scientific Instruments, vol. 75, 4997-
5004, 2004
3. Schwaiger, R., Dehm, G., and Kraft, O., Philosophical Magazine A, vol. 83, 693, 2003
4. Zhang, G.P., Schwaiger, R., Volkert, C.A., and Kraft, O., Philosophical Magazine Letters,
vol. 83, 477, 2003
5. Zhang, G.P., Schwaiger, R., Volkert, C.A., Arzt, E., and Kraft, O., Journal of Materials
Research, vol. 20, 201-207, 2005
 11. Deformation and Fracture at the Nano Scale 699

MICROMECHANICS OF DAMAGE EVOLUTION IN SOLID PROPELLANTS

N. Aravas1, F. Xu and P. Sofronis

University of Illinois at Urbana-Champaign,
Department of Mechanical and Industrial Engineering
158 Mechanical Engineering Building
1206 West Green Steet
Urbana, IL 61801
1Department of Mechanical and Industrial Engineering

University of Thessaly, Volos, Greece.

sofronis@uiuc.edu

Solid propellants are composite materials with complex microstructure. In a generic form, the
material consists of polymeric binder, ceramic oxidizer, and fuel particles (e.g. aluminum).
Damage induced by severe stress and extreme temperatures is manifested in particle cracking,
decohesion along particle/polymer interfaces, void opening or even polymer crazing at low
temperatures and inert propellants. In this work, the effect of damage due to void formation on the
material macroscopic response is investigated from a solid mechanics perspective. First, issues
related with the constitutive behavior of the individual phases in the absence of damage are
reviewed. Next, with the use of rigorous composite homogenization theory, a macroscopic
constitutive law is proposed that accounts for continuous void nucleation and growth upon
straining. Numerical calculations for the uniaxial tension test capture most of the experimentally
observed features, namely an initial elastic regime, a viscoplastic regime in which void formation
competes with hardening in the matrix, a softening regime, and a macroscopic volume expansion
which continuously increases with straining.
 11. Deformation and Fracture at the Nano Scale 701

DEFORMATION AND FAILURE MECHANISMS IN METALLIC

NANOLAYERED COMPOSITES

R. G. Hoagland, J. P. Hirth, and A. Misra

Los Alamos National Laboratory, Los Alamos, NM 87545
hoagland@lanl.gov

Layered composites composed of two or more dissimilar metals can be easily created with
precisely controlled layer thickness (to within one atomic layer) by various vapor deposition
techniques. For equithickness composites with layer thicknesses greater than 100 nm, typically,
flow strengths display a Hall-Petch type dependence on layer thickness, i.e., VY = kh-1/2 + V0,
where h is the layer thickness, k the Hall-Petch constant and V0, a background strength or friction
stress. For composites with thinner layers this relation is no longer obeyed, but instead the strength
becomes less dependent on h and reaches a maximum at an h of about 1 – 5 nm, as shown in Figure
1, Misra, et al [1]. Although, at maximum, the strength is lower than predicted by Hall-Petch, these
materials are still very strong, with flow strengths exceeding 1 GPa and, in some cases,
approaching the theoretical shear strengths of the constituents. Even at these high strength levels
the materials are deformable, plastically. Cu/Nb, for example is easily rolled to large reduction in
thickness except as noted below, Misra [2].

Figure 1. Flow strength, represented here in terms of hardness, of several multilayered composites
increases as the layer thickness decreases. In this graph, the abscissa is h-1/2 and so in the linear
portions on the left side of the graph, the data obey a Hall-Petch relation. Peak strength occurs in
the range of 1 - 5 nm.

We discuss the origins of the strengthening mechanisms in these materials. Important

differences exist in the types of operative strengthening mechanisms between systems where the
two components have the same crystal structure and are epitaxially related and systems where the
components have different crystal structures. In the former case, lattice mismatch, that creates
large coherency stresses with alternating signs from one layer to the next, is primarily responsible
for strength, Hoagland et al [3]. This observation suggests a simple description of maximum
strength in terms of lattice mismatch. Other factors include modulus mismatch and a form of work
hardening caused by accumulation of residual dislocations at interfaces due to repeated penetration
of the interfaces by the passage of slip from one material into another, Henager et al [4].
702 R. G. Hoagland et al.

In the latter case where the component materials have different crystal structure, the interfaces,
if weak in shear, attract glide dislocations. Once in the interface, these glide dislocations become
trapped because of dislocation reactions or core spreading. These effects reduce the stresses near
the dislocation core making it less able to nucleate, by whatever means, a glide dislocation in the
neighboring crystal. This reduces the ease of transport of slip across the boundary thereby making
the interface an effective obstacle to slip. We show evidence, derived from atomistic modeling, of
dislocation – interface interactions, Hoagland et al [4]. We discuss an elasticity model that enables
prediction of the magnitudes of the interactions. This model shows that some types of dislocations
interact much more strongly with shearable interfaces than others. The model also suggests that, as
the layer thickness is reduced, a greater fraction of existing glide dislocations will become trapped
in interfaces leading to dislocation starvation. This is consistent with observations of greatly
reduced ductility during rolling of Cu/Nb composite films with layer thicknesses of less than 4 nm.

Acknowledgements: The authors are pleased to acknowledge the support of this work by the U. S.
Dept. of Energy, Office of Basic Energy Sciences.

References
1. Misra, A, Hirth, JP, and Kung, H, Phil. Mag. A, vol. 82, 2935, 2002.
2. Misra, A. Zhang, X., Hammon, D., and Hoagland, R. G. , Acta Mat., vol. 53, 221, 2005.
3. Hoagland, R. G. Mitchell, T. E., Hirth, J. P., and Kung, H., Phil. Mag., vol. 82, 643, 2002.
4. Henager, C. H., Kurtz, R. J., and Hoagland, R. G., Phil. Mag., vol. 84, 2277, 2004.
5. Hoagland, R. G., Kurtz, R. J., and Henager, C. H., Scripta Mat., vol. 50, 775, 2004.
 11. Deformation and Fracture at the Nano Scale 703

DISLOCATION SOURCE SENSITIVITY OF PLASTICITY AND FRACTURE IN

TUNGSTEN

J. E. Talia and R. Gibala

Department of Mechanical Engineering
Wichita State University
Wichita, Kansas USA
George.talia@wichita.edu
Department of Materials Science and Engineering
University of Michigan
Ann Arbor, Michigan USA
rgibala@umich.edu

We have examined the effect of dislocation sources introduced on the surface of single-crystal and
polycrystalline tungsten on the plastic flow and fracture behavior at temperature in the range 77 K
to 590 K. The experiments utilize the phenomenon of surface film softening observed in many
body-centered cubic metals, such as observed by Sethi and Gibala [1-3]. In investigations like
these, it is observed that application of surface films of approximately 50-200 nm in thickness to
bcc metal substrates can decrease the yield and flow stress, increase the ductility, and
correspondingly reduce the large temperature dependence of the yield and flow stress at
homologous temperatures T below approximately 0.15Tm, where Tm is the absolute melting
temperature. The large temperature dependence of the yield and flow stress at T/Tm < 0.15 is
associated with the high Peierls-Nabarro stress of screw dislocations in the bcc structure. By
contrast, edge dislocations in bcc metals have high mobility at low temperatures. Mechanistically,
it has been shown that for the coated materials under applied stress, large densities of mobile edge
dislocations can be generated in the substrate metal at the film-substrate interface. These edge
dislocations can move into the substrate and effect plasticity at the reduced flow stresses observed.
In this investigation, additional use is made of surface modification (roughening) of the substrate
surface to afford control over the density of potential dislocation sources at the film-substrate
interface in coated materials. The results are described below.
The experiments have been done on high purity (> 99.999%) single crystals with an
approximate [213] tensile axis orientation that were grown by triple-pass electron-beam zone
melting of undoped polycrystalline tungsten in a vacuum of less than 5 x 10-5 Pa. These materials
were machined into tensile specimens of 3.2 mm in diameter and 25 mm in length, with a gage
diameter of 1.5 mm and gage length of 12.5 mm. The final specimens were electropolished and
then outgassed at temperatures in the range 1300 K to 1500 K at pressures below 10-6 Pa.
Experiments on polycrystalline materials were performed mainly on the as-received, warm-
worked tungsten of about 99.95% purity. The material had an elongated grain structure along the
tensile axis, with grains about 1-5 microns wide and about 10-50 microns long and an approximate
[110] crystallographic texture. For ease in reproducing deposition conditions from experiment to
experiment, the surface films chosen for these experiments were tungsten oxide films deposited
anodically at room temperature. Most of the results presented are for an 80 nm film deposited at 15
volts. Surface dislocation sources on the substrate surfaces were introduced by two different
methods: manual surface abrasion with metallographic papers of different grit sizes in the range
220-600 or, with better control, chemical etch pitting of the surface by metallographic techniques,
as used historically to disclose dislocations that emerge on specimen surfaces. The abraded or etch-
pitted surfaces were then oxide coated, as described above. Most mechanical testing was done in
704 J. E. Talia and R. Gibala

tension at a shear strain rate of 3 x 10-4/s, and various types of experiments were conducted over
the temperature range from 77 K to 590 K.
The experimental results fall into four broad categories. (1) Surface roughening by etch pitting
or surface abrading of the two tungsten materials reduced the plasticity and promoted a more brittle
fracture relative to corresponding baseline materials. These processes introduced surface flaws of
larger size than existed on the reference surfaces. (2) Deposition of surface oxide films on the
unetched or unabraded materials resulted in enhanced plasticity and reduced yield and flow
stresses relative to baseline properties. Thus, a somewhat conventional version of surface film
softening was observed in these experiments [1-3]. (3) When etch pitting or surface abrasion of the
substrates was combined with subsequent oxide coating of the roughened surface, greatly
enhanced plasticity over that given in (2) was observed. (4) To a first order, the extent of softening
in surface-roughened and oxide-coated material, whether measured by enhanced plasticity,
reduced flow stress, or extent of brittle fracture, correlated with the degree of surface roughening.
In effect, the presence of the coating converted roughening acting as flaws in (1) into potential
dislocation sources. Quantitative correlations in addition to the full complement of results are
given in the paper.
In summary, we have demonstrated that introduction of dislocation sources by etch pitting or
abrasion of the surface of tungsten in single-crystal and polycrystalline forms can promote
enhanced plasticity of the subsequently oxidized material. The experiments provide fundamental
data of the type needed for more complete constitutive analysis of dislocation dynamics and its
interaction with fracture processes. As the relative dimensions of the substrate decrease toward
nanometer length scales, the effects we see at millimeter dimensions should become more and
more significant.

References
1. Sethi, V.K. and Gibala, R., Acta Metall., vol. 25, 321-332, 1977.
2. Sethi, V.K. and Gibala, R., Phil. Mag., vol. 37, 419-429, 1978.
3. Sethi, V.K. and Gibala, R., in Surface Effects in Crystal Plasticity, edited by R. Latanision
and J. Fourie, Noordhoff, Leyden, The Netherlands, 1977, 599-610.
 11. Deformation and Fracture at the Nano Scale 705

DELAMINATION OF THIN METAL FILMS ON POLYMERS

A. Pundt, E. Nikitin, and R. Kirchheim

Universität Göttingen
Institut für Materialphysik, Tammannstr. 1, 37077 Göttingen, Germany
rkirch@ump.gwdg.de

Hydrogen absorption in niobium films on polycarbonate (PC) gives rise to large compressive in
plane stresses which finally lead to a delamination of the metal from the substrate. Delamination
occurs via straight and wrinkled buckles and can be observed with an optical microscope (cf. Fig.
1). By doping the Nb-film electrochemically with hydrogen the increasing stresses are measured
by determining the curvature of the PC-substrate. For a 200 nm thick Nb-film the corresponding
compressive stress is shown in Fig. 2 as a function of hydrogen concentration. The initial linear
increase of the stress is in accordance with the known volume expansion caused by H-atoms in Nb.
Deviations from the linear response start at the same time when buckles are formed.

FIGURE 1. Buckles formed in a 100 nm thick niobium film sputter deposited on a polycarbonate
substrate of 1 mm thickness. The Nb-film was doped electrochemically with hydrogen which
initiated compressive stresses in the adhering Nb-layer leading to partial delamination as buckles.
With increasing hydrogen content the network of buckles is changing.

The corresponding critical stress for buckling is measured for 50, 100 and 200 nm thick Nb-
films which were covered on both sides with a 10 nm thick Palladium film. It was shown by Pundt
et al. [1] that the critical stresses for buckling are much larger than the ones needed to exceed
Euler’s instability criterion. They are required to overcome the adhesion between Pd and PC. From
a simple energy balance between the adhesion energy and the released elastic energy the following
equation can be derived:

2JE
V cr r V in 
D (1  Q 2 )
(1)
where Vcr is the critical stress for buckle formation, Vin is the initial stress in the Nb-film (where
the positive sign is referring to tensile stress), D is the film thickness, J is the adhesion energy
between film and substrate and E and Q are Young’s modulus and Poisson’s ratio of the metal film.
In accordance with Eq. 1 measured critical stresses are plotted in Fig. 3 versus the inverse
square root of film thickness. The slope of the corresponding straight line yields a value of about 1
J/m2 for the adhesion energy and the intercept gives an initial compressive stress of about 600
706 A. Pundt et al.

MPa. The precision of the evaluated adhesion energy can be improved by measuring initial stresses
and/or the shape of the buckles.
The concept of producing compressive stresses in a film on a substrate by dissolution of atoms
or molecules in the film and measuring critical stresses for buckling is applicable to other systems
besides Nb on PC. For instance dry and hard lacquer films on metals buckle when they are exposed
to a vapour of Dichlormethan. In addition, a thin metal film can be placed between the hydrogen
dissolving metal, i.e. Niobium and the polymer. Then the adhesion between the thin film and the
polymer has to be used in Eq. 1.

FIGURE 2. Compressive stress within the

200 nm thick Nb-film as a function of FIGURE 3. Critical stresses
hydrogen concentration. Buckeling of the for buckling obtained from
film is observed at the same time when the stress measurements as
measured stress (open circles) deviates presented in Fig. 2 and
from the linear relationship The critical plotted versus inverse square
stress for buckling is defined as the stress, root of film thickness D cf.
where these deviation reach 10%. Eq. 1).

References
1. Pundt, A. Nikitin, E. Pekarski. P. and Kirchheim, R., Acta mater. 52, 1579-1587, 2004
 11. Deformation and Fracture at the Nano Scale 707

FRACTURE MECHANICS OF ONE-DIMENSIONAL NANOSTRUCTURES

Weiqiang Ding, Lorenzo Calabria, Kevin M. Kohlhaas, Xinqi Chen and Rodney S. Ruoff
Department of Mechanical Engineering, Northwestern University
2145 Sheridan Road, Evanston, IL 60208-3111
r-ruoff@northwestern.edu
aDipartimento di meccanica e tecnologie industriali, Universita di Frienze

Via di Santa Marta, 50139, Florence, Italy

One-dimensional (1D) nanostructures such as nanotubes and nanowires have attracted

considerable attention in recent years due to their promise of applications in sensing and materials
reinforcement. Over the past decade various 1D nanostructures has been synthesized. To develop
applications with these nanostructures, it is important to first understand their fundamental
properties. Our work focused on characterizing the mechanical properties of these novel 1D
nanostructures.
One of the methods we employed to characterize 1D nanostructures is tensile testing, which is
a commonly used testing method to characterize the tensile properties of materials. During quasi-
static tensile loading, a piece of material is loaded in tension until it fractures. The specimen
elongation and applied load is typically recorded during the loading process. The tensile strength,
Young's modulus and other tensile properties such as fracture strength and ultimate strain of the
material are then determined from the stress-strain relationship. Conventional tensile test machines
are not suitable for nanostructure characterization. Here we studied the tensile properties of 1D
nanostructures with our home-built nanomanipulator [1] inside a scanning electron microscope.
The two types of nanostructures we investigated are crystalline boron nanowires and multiwall
carbon nnaotubes (MWCNTs). The crystalline boron nanowires have been synthesized with the
chemical vapor deposition method. [2] Their average diameter was ~ 60 nm and they are tens of
microns in length (Figure 1a). Arc-grown MWCNTs were purchased from Materials and
Electrochemical Research Corp. They are 6 – 20 nm in outer diameter and 1 – 5 um in length
(Figure 1b).[3]

FIGURE 1. SEM images of (a) crystalline boron nanowires and (b) multiwall carbon nanotubes.

Nanoscale tensile tests on these nanostructures were performed with the nanomanipulator
inside a LEO 1525 field emission gun SEM. AFM cantilevers were used as the manipulation tool
and force sensing element. Individual nanostructures were picked up from the source and clamped
between two opposing AFM cantilever tips (Figure 2a) with the electron beam induced deposition
method.[4] Tensile load was then applied to each nanostructure tested through nanomanipulation.
With the increase of the tensile load the nanostructure eventually fractured.
708 W. Ding et al.

A MWCNT consists of many concentric cylindrical shells, and the interaction between the
individual shells is the relatively weak van der Waals force. When a MWCNT is tensile loaded, the
outer layer is primarily loaded and consequently fractured with inner layers being pulled out from
the outer shell fragment (Figure 2b), in a ”sword-in-sheath” manner.[5]

FIGURE 2. SEM images of (a) a boron nanowire being clamped between two AFM cantilever
tips under a tensile load; (b) sword-in-sheath fracture of a multiwall carbon nanotube.

A series of SEM images were taken during the tensile loading process. From image analysis
the cantilever deflection and nanostructure elongation were obtained. The diameter of the boron
nanowires were measured inside SEM. The diameter of MWCNTs were measured with high-
resolution TEM. The tensile stress and tensile strain of the nanostructure at each loading step were
obtained from data analysis, and a stress-strain diagram were plotted. From the stress-strain
relationship the tensile properties of the MWCNTs and boron nanowires were evaluated.
Acknowledgements
This work was funded by NSF EEC-0210120, and in part by ONR #N000140210870 (partial
support, W. Ding) and by the NASA BIMat URETI # NCC-1-02037 (support for X. Chen). The
SEM and TEM work was performed in the EPIC facility of NUANCE Center at Northwestern
University. We appreciate receiving the boron nanowires from C. Otten (Buhro group, Washington
University-St. Louis.)

References
1. Yu, M.F., Dyer, M.J., Skidmore, G.D., Rohrs, H.W., Lu, X.K., Ausman, K.D., Von Ehr, J.R.,
and Ruoff, R.S., Three-dimensional manipulation of carbon nanotubes under a scanning
electron microscope. Nanotechnology, 1999. 10(3): 244-252.
2. Otten, C.J., Lourie, O.R., Yu, M.F., Cowley, J.M., Dyer, M.J., Ruoff, R.S., and Buhro, W.E.,
Crystalline boron nanowires. Journal of the American Chemical Society, 2002. 124(17):
4564-4565.
3. http://www.mercorp.com/mercorp/nano.pdf.
4. Ding, W., Dikin, D.A., Chen, X., Wang, X., Li, X., Piner, R., Ruoff, R.S., and Zussman, E.,
Mechanics of hydrogenated amorphous carbon deposits from electron beam induced
deposition of a paraffin precursor. Journal of Applied Physics, 2005. 98: 014905.
5. Yu, M.F., Lourie, O., Dyer, M.J., Moloni, K., Kelly, T.F., and Ruoff, R.S., Strength and
breaking mechanism of multiwalled carbon nanotubes under tensile load. Science, 2000.
287(5453): 637-640.
 11. Deformation and Fracture at the Nano Scale 709

EFFECTS OF STRUCTURE AND BONDING AT SURFACES AND

INTERFACES ON FRACTURE

S. P. Lynch, S. Moutsos, B. Gable, S. Knight, D. P. Edwards and B. C. Muddle

Monash University Clayton, Vic. 3168 Australia, DSTO, P.O. Box 4331 Melbourne 3001
stan.lynch@dsto.defence.gov.au

It is well known that segregation of impurity elements at grain boundaries can facilitate
intergranular fracture. However, it is not so widely recognised that segregation of major alloying
elements in some materials, e.g. lithium and magnesium in aluminium alloys, may also be
embrittling. Moreover, it appears that certain structural arrangements may be required in addition
to segregation, and that ‘two-dimensional’ grain-boundary phase transitions are probably
responsible for ductile-to-brittle fracture transitions with decreasing temperature (e.g. Hart [1],
Lynch et al.[2]). For a given testing temperature, ductile-to-brittle fracture transitions may also
occur with increasing ageing time due to increasing levels of segregation. Intergranular chemistry
and structure can influence fracture resistance via effects on decohesion or dislocation emission
from crack tips or via effects on slip transmission across boundaries, but the relative importance of
these effects is not well established.
Evidence for the above hypotheses are reviewed, with a focus on recent work on a variety of
Al-Li alloys in different ageing conditions, and with grain sizes ranging from several hundred
nanometers to several hundred micrometers. High-resolution TEM and SEM fractography, thin-
foil TEM, atom-probe microscopy, and other techniques have been used. It is concluded that
particular grain-boundary structures and chemistries facilitate decohesion and dislocation-
emission from crack tips, and thereby promote low-energy intergranular fracture by a very
localised microvoid-coalescence process in some systems (Fig. 1). Ultra-fine grained alloys
exhibited similar behaviour to large grained alloys (Fig. 2).
Structure and bonding at crack tips could also be affected by adsorption of atoms from the
external environment (in addition to the intrinsic grain-boundary structure), with potential effects
on cohesion and dislocation emission. Exactly how structure and bonding at crack tips are affected
by adsorption is not known, but the dramatic phenomenon of adsorption-induced liquid-metal
embrittlement (e.g. Lynch [3]) demonstrates that significant effects must occur. Surface
relaxations and reconstructions are known to occur at planar surfaces, are known to be influenced
by adsorption, and can extend for several atomic distances beneath surfaces (e.g. [Van Hove [4],
Woodruff [5]). There may well be greater effects at stressed (non-atomically sharp) crack tips with
high-index, stepped surfaces, and at crack-tip/grain-boundary intersections, than at low-index,
unstressed surfaces. Thus, effects on dislocation emission (involving the creation of a surface step
and a dislocation core by shearing of atoms within a few atomic distances of crack tips) should be
expected. Fractographic evidence that adsorption-induced surface-lattice perturbations can
facilitate dislocation emission and decohesion from crack tips are outlined. Finally, the
implications of these results for fracture of nanocrystalline materials are considered.
710 S. P. Lynch et al.

FIGURE 1. (a) SEM, and (b) TEM of replica of low-energy intergranular fracture of a very
underaged 8090 Al-Li-Cu-Mg alloy tested at -196ºC. Lithium segregation at grain boundaries is
probably responsible for ‘brittle’ fracture. Note that the TEM of the replica (shadowed at a low
angle ~10º) reveals what appears to be very shallow dimples not resolved by SEM (with a field-
emission gun, low kV, and small working distance).

FIGURE 2 (a) TEM of thin-foil, and (b) SEM of low-energy intergranular fracture produced at
-196ºC, of a fine-grained mechanically alloyed 905-XL Al-Li-Mg-C-O alloy.

References
1. Hart, E. W., Scripta Metall., vol. 2, 179 - 182, 1968.
2. Lynch S.P., Muddle B.C., and Pasang, T., Philos. Mag. A, vol. 82, 3361-3373, 2002.
3. Lynch, S.P., Acta Metall., Overview No. 74, vol. 36, 2639-2661, 1988.
4. Van Hove, M.A., In Materials Science and Technology: A Comprehensive Treatment, edited
by R. W. Cahn et al., vol. 1, 485-453, 1991.
5. Woodruff, D. P. (Ed.), The Chemical Physics of Solid Surfaces, vol. 10, Elsevier, 2002.
 29. Reliability and Failure Analysis of Electronics and Mechanical Systems 711

APPLICATION OF THE NEW STATIC PHOTOELASTIC EXPERIMENTAL

HYBRID METHOD WITH NEW NUMERICAL METHOD TO THE PLANE
FRACTURE MECHANICS

Jai-Sug Hawong1 , Jeong-Hwan Nam2, O-Sung Kwon3 and Konstantin Tche3

1School of Mechanical Engr, Yeungnam Univ. Gyung San City, Gyung-Buk, Korea, 712-749
2Visiting Researcher, School of Mechanical Engr, Yeungnam Univ., Korea, 712-749
3Graduate Student, School of Mechanical Engr, Yeungnam Univ.,Korea, 712-749

jshawong@yumail.ac.kr

Dynamic photoelastic experimental method [1,2] and photoelastic experimental analysis for
othotropic material [3,4] have been widely studied [1,2] since 1950. Photoelastic experimental
method has been widely applied to the static and dynamic fracture mechaics [5,6] since 1970.
During this time, R.J. Sanford [7] suggested the non-linear least square method using Newton-
Raphson numerical method with Gaussian elimination method. We have used this method and this
conception for determination of stress intencity factor using photoelastic experimental method [8].
The non-linear least square method have been applied to the photoelastic experimental hybrid
method which can be used to obtain the static and dynamic stress intensity factor and stress
concentration factor [9,10,11]. The Newton-Raphson numerical method with Gaussian elimination
method have been used in the non-lineal least square method for the photoelastic experimental
method. In this research, the photoelastic experimental hybrid method with Newtion-Raphson
numerical method is called old technique. However, old technique is often diverged and unstable
in high stress distribution as vicinity of crack tip.
Therefore, the main aim of this research is to develop the photoelastic experimental hybrid
method with Hook-Jeeves’ numerical method(is called new technique in this research) which is
more precise and stabler than old technique and is diverged any case.
The detail aims of this research are as follows.
1 To develop the new photoelastic experimental hybrid method with Hook-Jeeves’
numerical method and to certify the validity of the new technique.
2 To certify the validity of application of the new technique to high stress distribution such
as plane fracture problem.
3 To certify the validity of the application of Hook-Jeeves’ numerical method to the static
photoelastic experimental method.
In this research, the new static photoelastic experimental hybrid method with Hook-Jeeves
numerical method has been developed: This method is more precise and stable than the
photoelastic experimental hybrid method with Newton-Rapson numerical method with Gaussian
elimination method. Using the new static photoelastic experimental hybrid method with Hook-
Jeeves numerical method, we can separate stress components from isochromatics only, stress
intensity factors can be determined and plane fracture mechanics can be analyzed. The new statics
photoelastic experimental hybrid method with Hook-Jeeves should be used in the full field
experiment than the photoelastic experimental hybrid method with Newton-Rapson with Gaussian
elimination method
712 J.-S. Hawong et al.

References
1. D. Post, Photoelastic Stress Analysis for an Edge Crack in a Tensile Field, Experimental
Stress Analysis, pp. 99~116, 1953.
2. Irwin, G.R., Proc. SESA, pp. 93~96, 1958.
3. M. Ramulu, Dynamic Crack Curving and Branching, Dissertation, University of
Washington, 1982.
4. J.S. Hawong, A.S. Kobayashi, M.S. Dadkhah, B. S.-J. Kang, and M. Ramulu, Experimental
Mechanics, pp. 146~153, 1985.
5. D.G. Smith and C.W. Smith, Engr. Fract. Mech, Vol. 4, pp. 357~366, 1972
6. Tada, H., Experimental Mechanics, Vol. 1, pp. 390~396, 1974.
7. Robert J. Sanford, Engr. Fracture Mech. pp. 621~633, 1979.
8. Jai Sug Hawong, Dong Chul Shin, Hyo Jae Lee, Experimental Mechanics, Vol. 16 no. 2, pp.
165~174, 2002.
9. J.S. Hawong, C.H. Lin, S.T. Lin, J. Rhee and R.E. Rowlands, Journal Stresses in Orthotropic
Composite Material, Vol. 29, No. 18, pp. 8~13, 1991.
10. Dong-Chul Shin, Jai-Sug Hawong and O-sung Kwon, Transactions of KSME(A). Vol. 25,
No. 3, pp.434~442, 2001.
11. Jai-Sug Hawong, Dong-Chul Shin and Un-cheol Back, Engineering Fracture Mechanics,
Vol. 71, pp. 233~243, 2004.
 29. Reliability and Failure Analysis of Electronics and Mechanical Systems 713

RISK ANALYSIS OF BURIED PIPELINE USING PROBABILISTIC METHOD

Ouk Sub Lee1, Dong Hyeok Kim2 and No Hoon Myoung3

School of Mechanical Engineering, InHa University
#253, Yonghyun-Dong, Nam-Ku, Incheon, 402-751, Korea
leeos@inha.ac.kr
Department of Mechanical Engineering, InHa University
#253, Yonghyun-Dong, Nam-Ku, Incheon, 402-751, Korea
kdonghyeok77@yahoo.co.kr
Department of Mechanical Engineering, InHa University
#253, Yonghyun-Dong, Nam-Ku, Incheon, 402-751, Korea
profmnh@yahoo.co.kr

Natural gas is currently one of the most widely used sources of energy and its use is increasing.
The buried pipelines transporting gas and fluid are mostly installed underground and near the
highly populated zones and have various types of defects such as corrosion and environment-
assist-cracking. The prediction of the remaining strength of pressurized pipelines containing
corrosion defects has been frequently carried out using deterministic methods. These methods use
the nominal values for both load and the resistance parameters. However, it is well known that the
load and resistance parameters have uncertainties resulted from the measurement of the
dimensions of defects, the manufacture of the pipe, the operating conditions of pipelines, and etc.
The major factors for the failure of pipelines transporting the high-pressure gas are known to
be mechanical damage and corrosion. The corrosion is a process through which metal is degraded
by the interaction with its environment. Several models are used to predict wall loss due to
corrosion. The technique to predict pipeline failure due to corrosion damage is necessary to
determine the corrosion tolerance when we design and want to keep the pipelines in reliable state,
Lee and Kim [1], Caleyo et al. [2].
For the case that the internal pressure is acting at the corroded pipeline, the failure pressure
model is used to assess for remaining strength of pipeline. The failure pressure model that is most
commonly used is ANSI/ASME B31G model. The failure pressure equation of ANSI/ASME
B31G model for the pipeline corrosion is classified by the corrosion shape of parabola and
rectangular. And the MB31G model is used for the purpose of reducing the conservatism and
assumption of ANSI/ASME B31G model [2].
In this paper, using this two of failure pressure models and three of wall loss model which is
corrosion model, the effects of varying boundary conditions on the failure probability of buried
pipeline are investigated and the results from FORM (first order reliability method) and SORM
(second order reliability method) are compared and systematically studied for the buried pipelines
with corrosion defects. And the FORM and SORM are used in order to estimate the failure
probability of the buried pipelines with corrosion defects.
When the FORM and SORM are used to estimate the failure probability, it is assumed that
every variable is normal distribution and the probability distribution is determined by its mean and
standard deviation. The failure probability is calculated using the FORM and SORM that is one of
the methods utilizing reliability index. The FORM and SORM are denoted from the fact that it is
based on a first-order Taylor series approximation and second order Taylor series approximation of
the LSF, respectively, which is defined as below, Halder and Mahadevan [3], Melchers [4].

Z RL (1)
714 O. S. Lee et al.

where R is the resistance normal variable, and L is the load normal variable. Assuming that R
and L are statistically independent normally distributed random variables, the variable Z is
also normally distributed.
The reliability index for the non-linear model is estimated by the process shown in Fig. 1. So
we can calculate the failure probability using the reliability index, and the reliability of buried
pipelines with corrosion defects can be estimated using the failure probability.

FIGURE 1. Process to determine of the reliability index.

References
1. Lee, O. S. and Kim, D. H., Key Engineering Materials, Vol. 270~270, 1688~1693, 2004.
2. Caleyo, F., Gonzalez, J. L. and Hallen, J. M., Int. J. of Pressure Vessels and piping, Vol. 79,
7786, 2002.
3. Halder, A. and Mahadevan, S., Probability, Reliability and Statistical Method in Engineering
Design, John Wiley & Sons, 2000.
4. Melchers, R. E., Structural Reliability Analysis and Prediction, John Wiley & Sons, 1987.
 29. Reliability and Failure Analysis of Electronics and Mechanical Systems 715

RELIABILITY ESTIMATION OF SOLDER JOINT BY ACCELERATED LIFE

TESTS

Ouk Sub Lee1, No Hoon Myoung2 and Dong Hyeok Kim2

School of Mechanical Engineering, InHa University
#253, Yonghyun-Dong, Nam-Ku, Incheon, 402-751, Korea
leeos@inha.ac.kr
Department of Mechanical Engineering, InHa University
#253, Yonghyun-Dong, Nam-Ku, Incheon, 402-751, Korea
kdonghyeok77@yahoo.co.kr
Department of Mechanical Engineering, InHa University
#253, Yonghyun-Dong, Nam-Ku, Incheon, 402-751, Korea
profmnh@yahoo.co.kr

The thermal stresses induced by difference in Coefficient of Thermal Expansion between FR-4
board and 63Sn-37Pb solder joint directly affect the reliability of 63Sn-37Pb solder joint. This
research, thus, focuses to investigate the crack initiation and propagation behaviour around solder
joint by imposing a designed Accelerated Life Tests Procedure on solder joint by using a newly
manufactured Thermal Impact Experimental Apparatus. The fracture mechanism of the solder
joint was found to be highly influenced by thermal stresses. Fig. 1 illustrates a solder joint
assembly deformed under thermal cycling. Arrows in Fig. 1 show thermal stresses induced around
a solder joint under thermal cycling. And solder joint is deformed by thermal stresses.. ' L is the
deformation of solder joint by thermal stresses.

Figure. 1 An illustration of thermo-mechanical deformation in solder joints under thermal

cycling.( h = height of solder joint, L is distance from neutral point to solder joint

In this paper, the reliability of solder joint was evaluated by using the modified Coffin-Manson
law and a failure probability model in terms of thermal frequency and temperature.

The total number of cycles to failure, N f , is assumed as being dependent on the plastic strain

amplitude, ' H p , the fatigue ductility coefficient, H cf , and fatigue ductility exponent, c . The
relationship among these variables in the Coffin-Manson law is described as Eq(1).

'H p
H cf ( 2 N f ) c
2 (1)
716 O. S. Lee et al.

The First Order Reliability Method(FORM) is denoted from the fact that it is based on a first-
order Taylor series approximation of the limit state function(LSF) which is defined as below.

Z RL (2)

where R is the resistance normal variable, and L is the load normal variable. Assuming that R
and L are statistically independent normally distributed random variables, the variable Z is
also normally distributed. The event of failure occurs when R  L , that is Z  0 . The
probability of failure(PF) is given as below.

(3)

where P Z and V Z are the mean and standard deviation of variable Z , respectively, new
variable U is U ( Z  P Z ) / V Z , and ) is the cumulative distribution function for a
standard normal variable and E is the safety index or reliability index.
The detail properties of silicon chip, PCB(FR4) and solder joint are shown in Table 2.

TABLE 1. Material properties of flip chip package.

References
1. Rao, R.T., Fundamentals of Microsystems Packaging, McGrawHill, New York, USA, 2001.
2. Achintya, H., Sankaran, M., Probability, Reliability and Statistical Method in Engineering
Design, John Wiley & Sons, New York, USA, 2000.
3. Deborah, D.L.C., Materials for electronic packaging, Butterworth-Heinemann, Boston, USA,
1995.
4. Michael, P., Handbook of electronic package design, Dekker, New York, USA, 1991.
 29. Reliability and Failure Analysis of Electronics and Mechanical Systems 717

ANALYSIS OF ENGINEERING PLASTIC BEHAVIORS IN THERMAL STRESS

CONDITION

Seon Il Ham, Dong Jun Choi and Sang Duck Park

Samsung Electronics Co., Ltd Customer Satisfaction Management Center
416 Maetan 3-Dong Paldal-Gu Suwon-Si Gyonggi-Do 442-742 KOREA
Phone: 82-31-200-1064 FAX: 82-31-200-2165, si.ham@samsung.com

Recently, various plastics are used for improvement of reliability and thermal stability of
electronic goods. But, it is difficult to choose suitable balance of grade resin to each other product
property. Selection of high efficiency resin is becoming inescapable circumstance according to the
development of product quality. Using an inexpensive and reasonable resin, that sustains a uniform
performance of product quality. It makes a robust product and increases a company's competitive
power. Hereupon, I introduce example that use ESPI methods of thermal deformation analysis of
product using in our company product. I try to refer to a structural weak point detection of thermal
condition of electronic goods instead of mechanical and chemical measurement of specimen type.
Keywords: ESPI(Electronic speckle pattern interferometry), NDT (Non contact destructive
testing)

Fig. 1 Tandem Laser Scanning Unit LD (Laser Diode) Base Specimen 1

Fig. 2 Tandem Laser Scanning Unit LD Base Specimen 2

Fig. 1 and Fig. 2 are test sample of colour laser printer unit. It makes engineering plastics
material NORYL. Top and bottom base units compare deformation before thermal stress and after
thermal stress conditions. LSU main frame and LD base unit are thermal deformation each
behaviours. It makes Optical Path Difference of laser printing unit. In this experiment, we make
adjust material property and bead structure of thermal stress condition.
718 S. I. Ham et al.

References
1. Jones. R., and Wykes. C., Holographic and speckle interferometry, Univ. Press, Cambridge,
1989
2. Vest CM. Holographic interferometry. New York: Wiley, 1979
 29. Reliability and Failure Analysis of Electronics and Mechanical Systems 719

A MECHANISTIC MODEL FOR THE THERMAL FATIGUE BEHAVIOR OF

THE LEAD-FREE SOLDER JOINTS

Ilho Kim, Tae-Sang Park and Soon-Bok Lee

Department of Mechanical Engineering
KAIST (Korea Advanced Institute of Science and Technology)
373-1, Guseong-dong, Yuseong-gu, Daejeon, 305-701, Korea(ROK)
sblee@kaist.ac.kr

There have been many publications on solder joint life prediction models in the last few years.
Many researchers have carried out mechanical fatigue and creep tests on bulk specimens in order
to develop a more accurate life prediction model. Some others measured the deformation of the
specimen and made the relationship between the deformation and the fatigue life. Other
researchers have carried out thermal cycling tests and made life prediction model using the FEA
results.
The mechanical tests with bulk specimen show different trends of material property with that
of solder ball joints. Because the representative volume affects material properties as an
experimental results as shown by Bonda and Noyan [1]. If we test the solder ball joints
mechanically, temperature condition of the specimen is isothermal. But actually solder joints are
subjected to the cyclic temperature condition with the thermal gradients. Thermal cycling tests
well describe the real operating condition. However it is very difficult to measure the stress and
strain of the solder joints. Generally, the stress and strain of the solder joints was predicted using
the finite element method. In this paper, we propose a new mechanistic life-prediction model that
can compensate the weaknesses in the mechanical fatigue test and the thermal cycling test.
The mechanical tests were performed with a micro-tension tester which has very high
sensitivity. Fig. 1 shows the micro-tension tester. Fig. 2 shows the specimen which consists of 9
solder balls (diameter: 760), upper PCB and lower PCB. The isothermal fatigue test was performed
under various temperature and loading conditions and the creep test was also performed.

Figure 1. Schematic diagram and a photograph of the micro-tension tester

Figure 2. Test specimen Figure 3. A schematic diagram of pseudo

power cycling tester
720 I. Kim et al.

Figure 4. Temperature profile of pseudo power cycling tests

Thermal cycling tests were performed with pseudo-power cycling test method, which applied
heats to the specimen by conducting as shown in Fig. 3. It has merits of both chamber cycling
method and the power cycling method. The pseudo-power cycling method can apply various
temperature ranges easily, and the temperature condition is more realistic. The temperature profile
described in this study is shown in Fig. 4. [3]
A new mechanistic life prediction model for the thermal fatigue was developed and its validity
has been approved using the finite element method and the experimental results.

References
1. Bonda, N.R., Noyan, I.C., IEEE transactions on components, packaging, and manufacturing
technology, vol 19, no 2, 208-212, 1996
2. Park, Tae-Sang, A study on mechanical fatigue behaviours of ball grid array solder joints for
electronic packaging, Doctoral Thesis, KAIST, DME985155, 2004
3. Kim, Ilho, A study on thermal fatigue behaviour of BGA package, Master’s Thesis, KAIST,
MME20023132, 2003
 29. Reliability and Failure Analysis of Electronics and Mechanical Systems 721

MECHANICAL BEHAVIOR OF METALLIC THIN FILM ON POLYIMIDE

SUBSTRATE

Dong-Cheon Baek, Sung-Yeol Kim and Soon-Bok Lee

Department of Mechanical Engineering
KAIST (Korea Advanced Institute of Science and Technology)
373-1, Guseong-dong, Yuseong-gu,
Daejeon, 305-701, Korea(ROK)
sblee@kaist.ac.kr

Stress state of metallic thin film on polyimide substrate under tensile loading was analyzed using
analytic solution and finite elements methods. To obtain the material properties of the film, tensile
test of copper film on Kapton substrate was performed as an illustrating example of metallic thin
film on polyimide substrate and the measurement technique of the mechanical behaviour of
metallic submicron film was verified. In addition, in situ tensile atomic force microscopy (AFM)
observations of the crack evolution are studied and successive damaging phenomena are described
and finally analyzed with the help of finite element calculations.

FIGURE 1. Tensile test apparatus

Thin films are used in many technical applications, such as semiconductor devices, micro-
electro-mechanical systems, hard or decorative coatings. The mechanical behavior of very thin
layers is difficult to study and sometimes strongly influenced by the substrate properties, as in
nanoindentation measurements reported by Bhattacharya and Nix [1]. While the stress state is very
complicated in nanoindentation, mechanical properties of thin film can be obtained directly from
uni-axial tensile testing. However, tensile testing of free standing film has many technical
difficulties, such as fabrication procedures depending on film materials and handling problem.
An alternative for tensile testing of thin films is to deform films still attached to their substrates
proposed by O. Kraft et al. [2]. The strain applied to both the film and substrate is the same, and
the measured force is the sum of the forces on both layers:

FT F f  Fs
(1)

where FT , F f and Fs are, respectively, total measured force of specimen, force on film and
force on substrate. The force on substrate can be obtained within the elastic range of substrate. To
722 D.-C. Baek et al.

minimize the contribution of the substrate to load, thin substrates of low elastic modulus with large
elastic range need to be used. This technique has the advantage of studying the thin film under the
real conditions of technical application and handling of specimens.
In the present paper, stress state of metallic thin film on polyimide substrate under tensile
loading was analyzed using analytic solution and finite elements methods considering warpage in
transverse section of loading due to the mismatch strain induced by different Poisson’s ratio. To
show the feasibility of this method, copper film on Kapton substrate was performed as shown in
Figs 1-2. Strain was measured directly on surface of specimen using dual microscope which is
optical pattern-matching apparatus made by Park et al. [3].

FIGURE 2. Stress-strain curve of Cu film

In addition, in situ tensile atomic force microscopy (AFM) observations of the crack evolution
are studied when an external tensile load is applied step-by-step to the specimen using small tensile
test device. The successive damaging phenomena are described and finally analyzed with the help
of finite element calculations.

References
1. Bhattacharya, A.K., Nix, W.D., International Journal of Solids and Structures, vol. 24, 1287-
1298, 1988
2. Kraft, O., Schwaiger, R., Wellner, P., Materials Science & Engineering. Properties,
Microstructure and Processing. A, Structural Materials, vol. 319, 919-923, 2001
3. Park, T.S., Baek, D.C., Lee, S.B., Sensors and Actuators. A Physical, vol. 115, 15-22, 2004
 31. Multiscaling in Molecular and Continuum Mechanics - Scaling in Time and Size from Macro to Nano 723

MACRO-, MESO- AND MICRO-DAMAGE MODEL BASED ON SINGULARITY

REPRESENTATION FOR ANTI-PLANE DEFORMATION

G. C. Sih and X. S. Tang

School of Mechanical Engineering, East China University of Science and Technology, Shanghai
200237, China
Department of Mechanical Engineering and Mechanics, Lehigh University, Bethlehem, PA 18015,
USA
School of Bridge and Structure Engineering, Changsha University of Science and Technology,
Changsha, Hunan 410076, China
gcs@ecust.edu.cn

Recent development of sub-micron devices in microelectronics and cell structure evolution in

microbiology has indicated the need to understand how defects or imperfections initiate or migrate
at the microscopic and atomic scale, where the chemical compositions are very unstable. Slight
changes can built up and lead to significant alterations of the macroscopic behavior. Such effects
have not received sufficient attention in the past at such lower scale. They are potential sources that
can effect the initial stresses and strains at the microscopic and atomic scale, the magnitude of
which depend on the evolution of the system. These initial disturbances will interact with energy
transmitted through the system by external means. Macroscopically speaking, a system may be
regarded to be in equilibrium but this does not necessarily apply to the lower scales. Hence,
modeling of the macro/meso/micro behavior of physical systems requires the capability to cover
scale range several orders of magnitude in lineal dimension.
In contrast to the current trend of using molecular dynamics for treating multiscale physical
systems where massive calculations are performed that involves millions and more atoms for
simulation, this work focuses attention near the singular tip of a defect or imperfection at the nano-
scale. Such an approach is justified from the findings of painstaking electron microscopy works
that have traced the crack-tip signatures to the nano-scale [1]. Instead of crunching out numbers of
unmanageable proportions, it suffices to model the dominant defect tip region. The same idea that
has been applied so successfully to the tip of a macro-crack in fracture mechanics although the
problem of multiscaling is considerably more challenging where effects at the micro-scale or
smaller must be related to those at the macro-scale by the use of a meso-scale zone because the gap
between macro- and micro-scale can be too coarse.
Demonstrated in this work is the development of a macro/meso/micro model that covers the
lineal scale of 10-11 to 100 by application of the volume energy density function. Boundary
constraints and defect geometries are shown to play a role at the smaller scale in the same way as
those at the macroscopic scale. Since the objective here is to illustrate the concept of the singularity
representation approach, anti-plane deformation is assumed. The corresponding case of in-plane
deformation [2,3] follows the same line of thought but more complicated. Since scale divisions are
chosen arbitrarily, discontinuities between any two scale ranges cannot be avoided in general in
view of the multiple choice of geometric and loading parameters that cannot be possibly be
anticipated in advance. Moreover, in order to justify the use of equilibrium mechanics theories, the
scale ranges have to be kept sufficiently small. The connection between scales becomes a major
issue in this approach. Scale invariant criteria based on force or energy must then be used. This is
reminiscent of the early works [4] dealing with size effects. What the model does is to simulate the
non-equilibrium process by a series of equilibrium states connecting two scale ranges at one time.
In this way, macroscopic behavior can be connected to imperfections at the dislocation level [5].
724 G. C. Sih and X. S. Tang

The connection necessitates the knowledge of how the homogeneity of a system changes as the
scale is switched. The explicit determination of this cross scale change in homogeneity is made
possible because of the simplicity of the anti-plane deformation model.

References.
1. Private communication with M. S. Bruemmer from Pacific Northwest National Laboratory,
Richland, WA, USA, February 23, 2005.
2. G. C. Sih and X. S. Tang, Simultaneity of multiscaling for macro-meso-micro damage model
represented by strong singularities, J. of Theoretical and Applied Fracture Mechanics, 42(3)
(2004) 199-225.
3. G. C. Sih and X. S. Tang, Scaling of volume energy density function reflecting damage by
singularities at macro-, meso- and micro-scopic level, J. of Theoretical and Applied Fracture
Mechanics, 43(2) (2005) 1-21.
4. G.C. Sih, Implication of scaling hierarchy associated with nonequilibrium:field and
particulate. Prospects of me-somechanics in the 21st century, G.C. Sih and V.E. Panin, eds.
Special issue of J. of Theoretical and Applied Fracture Mechanics, 37, (2002) 335-369.
5. G. C. Sih and X. S. Tang, Screw dislocations generated from crack tip of self-consistent and
self-equilibrated systems of residual stresses: atomic, meso and micro, J. of Theoretical and
Applied Fracture Mechanics, 43(3) (2005) (in press).
 31. Multiscaling in Molecular and Continuum Mechanics - Scaling in Time and Size from Macro to Nano 725

MULTISCALING EFFECTS IN TRIP STEELS

G. N. Haidemenopoulos and N. Aravas

Department of Mechanical & Industrial Engineering
University of Thessaly, Volos, Greece
hgreg@mie.uth.gr

Low-alloy TRIP steels are a relatively new class of steels with excellent combinations of strength
and formability, making them particularly suitable for sheet forming applications in the automotive
industry. The steels possess a multiphase microstructure containing ferrite, bainite and retained
austenite. During cold forming operations, such as stretch forming, the retained austenite
transforms to martensite under the action of the applied stresses and strains. This deformation-
induced martensitic transformation of the retained austenite is responsible for the transformation-
induced plasticity (TRIP) effects found in these materials. These effects include significant
improvements in ductility and formability.
TRIP steels offer a unique example for the study of multiscale effects in materials in the sense
that experimental observations and models, derived at different scale levels, can be combined for
the understanding and the design of these materials.
At the microstructural level, TRIP effects depend on the volume fraction and the distribution of
phases and more specifically the volume fraction and stability of retained austenite.
Themodynamic stability and strain-induced transformation of the austenitic dispersion are in turn
influenced by microstructural parameters, such as size and composition, and by macroscopic
parameters, such as stress state triaxiality. Computational thermodynamics and kinetics methods
are applied, at the microstucture scale, for the modeling of austenite stability [1]. Carbon
stabilization is taken into account during the austenite to bainite transformation, and austenite
stability is calculated as a function of annealing time. Typical results are shown in Fig.1 where the
calculated retained austenite volume fraction is plotted vs annealing time and the results are
compared with experimental measurements of retained austenite with a saturation magnetization
method. The results aid the design of heat treatments for the processing of TRIP steels.

FIGURE 1 Retained austenite as a function of annealing time

Multiphase TRIP steels are essentially composite materials with evolving volume fractions of
the individual phases. The effective properties and overall behavior of TRIP steels are determined
by using homogenization techniques for non-linear composites. A methodology for the numerical
integration of the resulting elastoplastic constitutive equations in the context of the finite element
726 G. N. Haidemenopoulos and N. Aravas

method is developed and the constitutive model is implemented in a general-purpose finite element
program. The model is calibrated by using experimental data of uniaxial tension tests in TRIP
steels[2]. The problem of necking of a bar in uniaxial tension is studied in detail. The constitutive
model is used also for the calculation of ``forming limit diagrams" for sheets made of TRIP steels;
it is found that the TRIP phenomenon increases the strain at which local necking results from a
gradual localization of the strains at an initial thickness imperfection in the sheet.

FIGURE 2. Forming limit curves for two different values of initial thickness inhomogeneities.The
solid lines correspond to the TRIP steel, whereas the dashed lines are for a non-transforming steel.
The dark triangles are experimental data.

Typical results of the calculations are shown in Fig.2 for a transforming (TRIP) and a non-
transforming steel for two different values of initial thickness imperfection. Included in Fig.2 are
experimental data from formability measurements. Work is underway to further refine the model
in order to describe more accurately the forming limits of the material.

References
1. A.I. Katsamas, G.N. Haidemenopoulos and N. Aravas, Steel Research Int., vol.75, No.11
,737-743, 2004.
2. I. Papatriantafillou, N. Aravas and G.N. Haidemenopoulos, Steel Research Int., vol.75,
No.11,730-736, 2004.
 31. Multiscaling in Molecular and Continuum Mechanics - Scaling in Time and Size from Macro to Nano 727

A HYPER-SURFACE FOR THE COMBINED RATE AND SIZE EFFECTS ON

THE MATERIAL PROPERTIES

Zhen Chen1, Luming Shen1, Yong Gan1 and H. Eliot Fang2

1Department of Civil and Environmental Engineering

University of Missouri-Columbia, Columbia, MO 65211-2200, USA

2Computational Materials & Molecular Sciences Department

Sandia National Laboratories, Albuquerque, New Mexico 87185-1411, USA

chenzh@missouri.edu

The recent interests in developing multiscale model-based simulation procedures have brought
about the challenging tasks of bridging different spatial and temporal scales within a unified
framework. However, the research focus has been on the scale effect in the spatial domain with the
loading rate being assumed to be quasi-static. Although material properties are rate-dependent in
nature, little has been done in understanding combined loading rate and specimen size effects on
the material properties at different scales. In addition, the length and time scales that can be probed
by the molecular level simulations are still fairly limited due to the limitation of computational
capability. Based on the experimental and computational capabilities available, therefore, an
attempt is made here to formulate a hyper-surface in both spatial and temporal domains to predict
combined size and rate effects on the mechanical properties of materials.
As shown in an asymptotic scaling analysis without considering the rate effect [1], the
relationship between the nominal strength V N and different sizes D of geometrically similar
structures exhibits a two-sided asymptotic support in the log D  log V N space. Hence, a
simple set of equations could be chosen to represent the size effect on the quasi-static strength V 0
in the spatial domain, as follows:

(1)

where Du is the specimen size at which the ultimate strength V u is reached, and Dm the
minimum macro-scale size beyond which the strength V m becomes size-independent. As can be
seen from Eq. (1), the slope of the size-dependent portion with D u  D  D m is given by

d log V 0  S log V u  log V m 

§ S log D  log D u ·
 cos ¨¨ ¸¸
d log D  2 log D m  log D u  © 2 log D m  log D u ¹ (2)

which can be normalized to be

d log V 0  log V u  log V m  S § S log D  log D u ·

/  cos ¨¨ ¸¸
d log D  log D m  log D u  2 © 2 log D m  log D u ¹ (3)
As can be found from Eq. (3), the normalized slope could be fully determined with the small-
S
size asymptotic limit and large-size asymptotic limit, and is in the range between  and zero.
2
728 Z. Chen et al.

Although the question on a reasonable small-size asymptotic limit remains open [1], the proposed
formulation might provide a simple means to characterize the size-dependence of certain materials
under quasi-static loading conditions.

To describe the dependence of the material strength on the strain rate H , Cowper and
Symonds’s model is adopted as follows:
1/ q
V H  § H p ·
1  ¨¨ ¸¸
Vo © H r ¹ (4)

in which H p denotes the plastic strain rate, V 0 is the quasi-static strength, and Hr and q are two
model parameters that can be determined with two experimental data points of V H  . The size
effect on the model parameters were not considered in the original model. As compared with the
plastic strain, the elastic strain can be neglected so that H p = H could be assumed. As can be found
from the open literature, however, there exists a critical strain rate for single-crystal metals, below
which the material strength becomes rate-independent. The critical strain rate is increased with the
decrease of the specimen size. In other words, the model parameters Hr and q in Eq. (4) could be
assumed to be size-dependent if this model is used to predict the rate-dependent strength with the
strain rate above the critical strain rate. Thus, a three-dimensional hyper-surface with respect to
both spatial and temporal domains could be formulated to describe combined size and rate effects
on the material strength, as below.
Based on Eq. (4), the hyper-surface is assumed to have a two-sided asymptotic support in the
V Vo H space. The small-size asymptotic limit and large-size asymptotic limit are
log  log
Vo H r
represented by Hrs , V os , q s  and Hrl , V ol , ql  , respectively. The size dependence of model
parameters Hr and q can therefore be described as follows:

log H r  log H rl § S log D  log D rs ·

1  sin ¨¨ ¸¸
log H rs  log H rl © 2 log D rl  log D rs ¹ (5.1)

q  ql § S log D  log D rs ·
1  sin ¨¨ ¸¸
q s  ql © 2 log D rl  log D rs ¹ (5.2)

for D rs  D  D rl with Drs and Drl being the specimen sizes at the small-size asymptotic limit
and large-size asymptotic limit, respectively. It follows from Eqs. (4) and (5) that the three-
dimensional hyper-surface takes the form of

ª § 1 º
log V H , D  log V D   log «1  10 * * ¨¨ log H  log H r D ·¸¸ »
© q D 
o
¬ ¹¼ (6)

in the log H  log D  log V space. To demonstrate the features of the proposed hyper-
surface, the mechanical properties of both tungsten and diamond specimens of various sizes under
various loading rates are investigated.
Reference
1. Bazant, Z.P. Scaling of Dislocation-Based Stain-Gradient Plasticity. J. Mech. Phys. Solids.
50:435-448, 2002.
 34. Cracks in Micro- and Nanoelectronics 729

A NEW METHOD FOR LOCAL STRAIN FIELD ANALYSIS NEAR CRACKS IN

MICRO- AND NANOTECHNOLOGY APPLICATIONS

B. Michel, D. Vogel1, N. Sabate2 and D. Lieske3

1Fraunhofer Institute for Reliability and Microintegration IZM Berlin
2University of Barcelona
3Infineon Technologies AG Dresden

Gustav-Meyer-Allee 25, D-13355 Berlin, Germany

bernd.michel@izm.fraunhofer.de, www.bernd-michel.com

The paper presents a new reliability approach based on local stress and deformation analysis by
means of digital image correlation methods. The so-called nanoDAC- (deformation analysis by
correlation technique) method is applied in connection with creep and fatigue crack evaluation
concepts. This will lead to improved lifetime estimations of microsolder joints in electronic
packaging and in MEMS and NEMS interconnection technologies as well.
Various kinds of flip chip, CSP and other packaging concepts for sensors, chips and MEMS
will be discussed based on the nanoDAC analyses of critical interconnection regions. Special
attention will be given to interface cracking taking also into account humidity and vibration effects
of the components.
Stress and Strain Measurements in Very Small Regions
A new approach to the local measurement of residual stresses in micro- and nanostructures is
described in the presentation. The presented technique takes advantage of the combined milling-
maging features of a focused ion beam (FIB) equipment to scale down the well-known classical
hole drilling method. This method consists of drilling a small hole into a solid containing inherent
residual stresses and measuring the displacements/strains caused by the local stress release, that
takes place around this hole. In the presented case, the displacements caused by the milling are
determined by applying digital image correlation techniques (DIC) to high resolution micrographs
taken before and after the milling process. The residual stress value is then obtained by comparing
the measured displacements with the analytical solution of the ideal displacement fields. The
feasibility of this approach and some generalization has been demonstrated on a micromachined
silicon nitride membrane showing that this method has very high potential for practical
applications in the field of MEMS, silicon chip interconnection technologies and related topics.
The authors call this new method FIBDAC. It received the 2005 Fraunhofer award for science
and technology. Besides the FIBDAC method the authors also present another example using the
DIC approach. It deals with fracture toughness evaluation by means of so-called nanoDAC
technique, i.e. nanodeformation analysis around tiny cracks using atomic force microscopy.
NanoDAC is a special microDAC variant of digital image correlation approach where local
deformation fields at the crack tip before and after loading steps are compared applying digital
image correlation methods providing these images by means of AFM, AFAM or other SPM
techniques. The methods enable to derive local stress intensity factors and more general fracture
quantities as well for micro- and nanostructures (e.g. MEMS, NEMS etc.).
730 B. Michael et al.

References
1. Vogel, D., Sabaté, N., Wunderle, B., Keller, J., Michel, B., Reichl, H., Nanoreliability for
Mechanically Loaded Devices, Int. Congress of Nanotechnology 2005 (ICN 2005), San
Francisco, USA, 1-4 Nov. 2005.
2. Michel, B., Keller, J, NanoDAC – a New Technique for Micro- and Nanomechanical
Reliability Analysis of Lead-free Solder Interconnects, Int. Conf. on Lead-Free Soldering,
Toronto, Canada, 24-26 May 2005.
3. Michel, B., Testing at Micro and Nanoscale, EuroSIME, European Conf. on Thermal,
Mechanical and Multiphysics Simulation and Experiments in Microelectronics, Berlin, 18-20
April 2005.
4. Auersperg, J., Vogel, D., Michel, B., Crack and delamination risk evaluation of thin silicon
based microelectronic devices, 11th Int. Conf. on Fracture (ICF 11) Torino, 20-25 March
2005.
 34. Cracks in Micro- and Nanoelectronics 731

EXPERIMENTAL INVESTIGATIONS FOR FRACTURE ANALYSIS OF

SOLDER JOINTS IN MICROELECTRONIC AND MEMS APPLICATIONS

H. Walter1,4, C. Bombach2, R. Dudek, W. Faust3 and B. Michel4

1AMIC GmbH Berlin, Germany, Volmerstraße 9B, D-12489 Berlin
2Nanotest und Design GmbH Berlin, Germany
3 Fraunhofer IZM Chemnitz, Germany
4 Fraunhofer IZM Berlin, Gustav-Meyer-Allee 12, D-13355 Berlin, Germany

hans.waltter@izm.fraunhofer.de

In Microelectronic and MEMS (Micro-Electro-Mechanical Systems) applications the volume of

solder joints decreases rapidly due to higher packaging density. In recent years, many solder alloys
have been developed and used. Many of lead-free-solders show a complex material behaviour due
to different mechanical and thermal properties throughout the sandwich which can influence the
mechanical and thermal reliability as well as the life time. For this reason, ensuring the solder joint
reliability is one of the most critical design aspects of electronic assemblies. To predict the failure
of solder joints with the help of the FE-Simulations tools, the description of solder behaviour is
needed. The mechanical behaviour of solder is non-linear and temperature dependent. Solder alloy
in consideration are above 0.5 of their melting point at – 40°C, so creep processes are expected
[1,2]. The failure behaviour of solder is a complex sequence and depends on microstructures, like
grain coarsening, micro-voiding, recrystallization, micro-cracking and macro-cracking on alloying
content, soldering temperature profile and dissolution of metallizations. Changes of the
microstructure can significantly effect the mechanical properties of the solders. Inhomogeneity of
solders, espacially of Sn-based lead free solders, can cause local fatigue driven multiple cracking
(3,4). Furthermore, plate-like intermetallic compounds (IMC) may cause crack initiation and
brittle fracture at interface to metallization, especially if the joint thickness becomes comperable to
the IMC-thickness respectively (Figure 1).

The failure criterion used for recording lifetime might vary based on either electrical failure or
mechanical cracks. For the analysis the temperature and stress dependent inelastic behaviour
(Creep and stress relaxation) no standard testing methods could be used for these applications.
combination of numerical testing methods and experimental are necessary. A comprehensive
mechanical characterization of solder alloy properties for tensile, shear, creep and fatigue
732 H. Walter et al.

properties were investigated. Thereby, this paper presents experimental test methods and results by
means of modified grooved lap specimens [5]. The analysis of shear stress and strain in solder
joints is of interest for development of analytical models that describe the shear deformations of
solder joints due to global thermal expansion mismatch between components and PCB. It could be
seen, that a couple of Sn-based lead free solder shows both primary and secondary creep behaviour
under cycling loading conditions. Primary creep and plastic strain may not be negligible in
applications with high temperature ramp rate or under thermal cycling conditions with short dwell
time.
For the description of the low cycle fatigue behaviour of solder alloys different failure
hypotheses are necessary. In addition to the defect free analysis concepts, a defect tolerant analysis
concept can be used. The last concepts based on fracture and damage mechanics hypotheses.
Fracture mechanics approaches using a critical crack tip parameters (J-Integral) with the assumes
that fatigue is limited to propagation of micro cracks [6]. Continuum damage approaches using a
viscoplastic constitutive framework with damage evolution capabilities. The application of
fracture mechanical concept enables an accurate interpretation of result of the FE-analysis to
estimate the reliability and lifetime duration.

References
1. Dudek,R.,Walter,H.,Döring,R.,Michel,B. : The World of Electronic Packaging and System
Integration, edited by Michel,B., Aschenbrenner,R., IZM special edition, Berlin 2005, 56 -65
2. Schubert,A.,Dudek,R.,Auerswald,E.,Gollhardt,A.,Michel,B.,Reichl,H.: Micromaterials and
Nanomaterials, Issue 03/2004, 30 – 41
3. Wiese,S. : The World of Electronic Packaging and System Integration, edited by Michel,B.,
Aschenbrenner,R., IZM special edition, Berlin 2005, 497-593
4. Schubert,A.,Dudek,R.,Auerswald,E.,Gollhardt,A.,Michel,B.,Reichl,H.: Proceedings of 53rd
Electronic Components Technology Conference , 2003, 603-610
5. Deplanque,S.,Nüchter,W.,Spraul,M.,Wunderle,B.,Dudek,R.,Michel,B., In Proceedings of
EuroSIME 2005, Berlin
6. Ghavifekr.H.B.,Michel .B., Sensors and Actuators, A 99, 2002, 183-187
 34. Cracks in Micro- and Nanoelectronics 733

SIMULATION OF INTERFACE CRACKS IN MICROELECTRONIC

PACKAGING

J. Auersperg1,2, B. Seiler3, E. Cadalen4, R. Dudek3 and B. Michel2

1AMIC Angewandte Micromesstechnik GmbH,

Volmerstraße 9B, D-12489 Berlin, Germany

2Fraunhofer Institute for Reliability and Microintegration Berlin (IZM), Dept. Mechanical

Reliability and Micro Materials,

Gustav-Meyer-Allee 25, D-13355 Berlin, Germany
3CWM Chemnitzer Werkstoffmechanik GmbH,

Otto-Schmerbach-Str. 19, D-09117 Chemnitz, Germany

3THALES Microelectronics S.A., Z.I. de bellevue

35221 Chateaubourg, France

juergen.auersperg@izm.fraunhofer.de

Increasing use under harsh environmental conditions - extreme temperatures, in particular often
lead to fatigue and failure of advanced electronic packages and related systems. As a result, its
thermo-mechanical reliability becomes more and more one of the most important preconditions for
adopting it in industrial applications.
Residual stresses from several steps of the manufacturing process, thermal and static and
dynamic mechanical loading conditions along with the fact that microelectronic packages are
basically compounds of materials with quite different Young's modules and thermal expansion
coefficients contribute to interface delamination, chip cracking and fatigue of interconnects.
Consequently, numerical investigations by means of nonlinear parameterized FEA, fracture
mechanics concepts are frequently used for design optimizations using sensitivity analyses [1]. So,
numerical design studies help to optimize designs of electronics applications at the earlier phase of
the product development processes. Unfortunately, this methodology typically accounts for
classical stress/strain evaluation or life time estimations of solder interconnects using modified
Coffin-Manson approaches. Delamination or bulk fracture mechanisms usually remain
unconsidered.
This contribution intends to figure out and discuss ways of using fracture mechanics numerical
approaches in connection with parameterized FEA attached to response surface method (RSM)-
based DOE-methodology. For improving such methods, the evaluation of mixed mode interface
delamination phenomena of several polymeric substrate/underfiller-, substrate/encapsulant- and
ceramics/encapsulant-specimens under bending have been combined with experimental
deformation measurements.
The simulation models were prepared to introduce residual stresses, satisfy the boundary/
loading conditions of the experiments, simulate the measured crack tip location vs. loading time
history and determine interface fracture parameters.
734 J. Auersperg et al.

FIGURE 1. Image analysis with uniDAC, sample molding compound on ceramics

Measured force vs. deflection curves, deformation fields as results of optical inspection and
deformation analysis as well as determined crack tip vs. deflection curves using the gray scale
correlation technique as shown in Fig 1 represent the input for the delamination modelling by
means of nonlinear FEM. The interface delamination investigations outlined here base on fracture
mechanics approaches for bimaterial crack problems. They take into account the special mixed
mode situations in case of interface fracture as discussed by Sun and Qian [1]. Because of the goals
of these investigations, the energy release rate G (ERR), the interface stress intensity factors K
(SIF) and their phase angle <(G) and <(K) are the parameters to be determined.
These values together with the methodology and software developed by Auersperg et al. [2-3]
enable costumers of FEA-based RSM/DOE approaches to evaluate the fracture toughness of
equivalent interfaces in various applications, electronics assemblies, for instance.

References
1. Sun, C. T., Qian, W, Int. J. Solids Structures, vol. 34. No. 20, 2595-2609
2. Auersperg, J., Kieselstein, E., Schubert, A., Michel, B., Trans. of the ASME, JEP, vol. 124,
318-322, 2002
3. Auersperg, J., Seiler, B., Cadalen, E., Dudek, R., Michel, B., In Proceedings of the EuroSime
2005, edited by L.J. Ernst, G.Q. Zhang, P. Rodgers, M. Meuwissen, S. Marco, O.de Saint
Leger, IEE Cat No. 05EX1050, Berlin, 2005, 197-200
 34. Cracks in Micro- and Nanoelectronics 735

AFM BASED FRACTURE ANALYSIS IN MICRO- AND NANOMATERIALS

Jürgen Keller1, Astrid Gollhardt, Dietmar Vogel and Bernd Michel2

Berliner Nanotest und Design GmbH,
Volmerstrasse 9, 12489 Berlin, Germany
Fraunhofer Institute for Reliability and Microintegration, Dept. Mechanical Reliability and Micro
Materials, Gustav-Meyer-Allee 25, 13355 Berlin, Germany
info@nanotest.org

Recent advances in microtechnology and the development of new electronics and microsystem
devices in automotive industry and communication sector have led to a strong need in material
characterization on the micro and nano scale. Defects occurring due to thermal or mechanical
material mismatches have to be analyzed by suitable methods. If they cannot be avoided
completely, e.g. microcracks in materials or at interfaces, evaluation of their effect on the life time
of the component is inevitable.
In addition the increasing interface-to-volume ratio in highly integrated systems and
nanoparticle filled materials and unsolved questions of size effect of nanomaterials are challenges
for experimental reliability evaluation. To fulfill this needs new strategies for reliability
assessment on the submicron scale are essential.
Under this prerequisite Scanning Probe Microscopy (SPM) serves as the basis for the
development of the nanoDAC method (nano Deformation Analysis by Correlation), which allows
the determination and evaluation of 2D displacement fields based on SPM data. In-situ SPM scans
of the analyzed object are carried out at different thermo-mechanical load states as shown in Fig. 1.
In the illustrated case a Scanning Force Microscopy topography signal serves as the imaging
technique. It is also possible to use other SPM image sources such as Phase Detection Microscopy
or Ultrasonic Force Microscopy.

FIGURE 1. SFM topography scans [4.6 µm u 4.6 µm] at a crack tip of a polymer CT (compact
tension) specimen; the scans are carried out at different load states F.

The obtained images are compared utilizing digital image correlation (DIC) based on grayscale
cross correlation algorithms [1,2]. This allows the tracking of local image patterns (compare to Fig.
1) of the analyzed surface structure. The measurement results of the nanoDAC technique are full-
field displacement fields. For the images of Fig. 1 the determined vertical (crack opening)
displacement field is illustrated in Fig. 2.
736 J. Keller et al.

FIGURE 2. Crack opening displacement field in vertical (y)-direction [µm] determined by means
of nanoDAC; in the background of the contour lines the AFM topography scan of Fig. 1 (left) is
illustrated.

Due to the application of SPM equipment deformations in the micro-, nanometer range can be
easily detected. Currently the accuracy of the nanoDAC method for displacement field
measurement is 1 nm for scan sizes of 2 µm, where the accuracy is determined by the
thermomechanical stability of the SPM system. The measurement technique can be performed on
bulk materials, thin films and on devices i.e. microelectronic components, sensors or MEMS/
NEMS. Furthermore, the characterization and evaluation of micro- and nano-cracks or defects in
bulk materials, thin layers and at material interfaces can be carried out.

References
1. Keller, J.; Vogel, D.; Schubert, A. and Michel, B. Displacement and strain eld measurements
from SPM images, in Applied Scanning Probe Methods, B. Bhushan, H. Fuchs, and S.
Hosaka, eds., pp. 253–276, Springer, 2004.
2. Vogel, D. and Michel, B. Microcrack evaluation for electronics components by AFM
nanoDAC deformation measurement, in Proceedings of the 2001 1st IEEE Conference on
Nanotechnology. IEEE-NANO 2001, pp. 309–312, 2001.
 34. Cracks in Micro- and Nanoelectronics 737

SIMULATION OF DEFORMATION AND FRACTURE BEHAVIOUR IN

MICROELECTRONIC PACKAGING

O. Wittler1, H. Walter, J. Keller2, R. Dudek, D. Vogel and B. Michel3

1Technische Universität Berlin, FSP Technologien der Mikroperipherik, Gustav-Meyer-Allee 25,

D-13355 Berlin, olaf.wittler@izm.fraunhofer.de

2AMIC GmbH, Volmerstraße 9B, D-12489 Berlin
3Fraunhofer IZM Berlin, Gustav-Meyer-Allee 25, D-13355 Berlin

The simulative and experimental analysis of fracture in microelectronic packages is a key issue of
current research and applications, because the mechanical integrity of a package and its electrical
interconnects is vital to the reliability of the complete microsystem. Dedicated simulations and
experiments can bring insight in the complex and critical field of mechanical failures in
microelectronic packages. The foundation of this is approach and focus of this paper is the
understanding and simulative description of deformation and fracture behaviour of selected
materials used for microelectronic packaging.
A key example is the AuSn interconnect, which is especially interesting for soldering of
optoelectronic and RF devices because of the fluxless soldering process [Sun, C. T., Qian, W, Int.
J. Solids Structures, vol. 34. No. 20, 2595-2609]. The key challenge for modeling of such
interconnects is the large variety of metals that can make up such interconnects. An example set-up
for a flip-chip interconnect and an eutectic AuSn interconnect is given in Fig. 1. Additionally these
set-ups can be varied. As can be seen three different AuSn phases are characteristic in this field:
eutectic AuSn, ] - Phase AuSn and pure Gold. Additionally it needs to be considered, that the
properties are process dependent. As there is only limited data available on deformation and
fracture behaviour of AuSn interconnects (i.e., [Lau, J. H. and Pao, Y. H., Solder Joint Reliability
of BGA, CSP, Flip Chip and Fine Pitch SMT Assemblies, McGraw-Hill, New York, 1997]),
additional research is done here.

FIGURE 1. Example of an eutectic (a) and flip-chip (b) interconnect

As the deformation behaviour varies locally, an approach is proposed to apply nanoindentation

on cross-sections of packages to characterise the local deformation behaviour. First results are
shown in Fig. 2, where the modulus is characterised as a function of the location inside the AuSn
solder interconnect. The data will be correlated to results obtained by tensile tests.
738 O. Wittler et al.

FIGURE 2.: Topographical scan of the indented AuSn bump (a) and the corresponding measured
indentation modulus values (b)

Additionally the elastic-plastic behaviour is analysed by a combined experimental and

simulative approach. The data obtained in this way will improve the analysis of fracture and
fatigue in AuSn solder interconnects.

References
1. Hutter, M., Hohnke, F., Oppermann, H., Klein, M., Engelmann, G., In Proceedings of the
54th Electronic Components & Technology Conference, IEEE, Piscataway, 2004, 49-57
2. Lau, J. H. and Pao, Y. H., Solder Joint Reliability of BGA, CSP, Flip Chip and Fine Pitch
SMT Assemblies, McGraw-Hill, New York, 1997
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 739

MIXED-MODE FRACTURE MODELED THROUGH A DISCRETE COHESIVE

ZONE MODEL-DCZM

D. Xie and A. M. Waas

Department of Aerospace Engineering
The University of Michigan
Ann Arbor, MI 48109-2140, USA
dcw@umich.edu

The Discrete Cohesive Zone Model (DCZM) is proposed to simulate fracture initiation and
subsequent growth when material nonlinear effects are significant using the finite element method.
Different from the widely-used Continuum Cohesive Zone Model (CCZM) where the cohesive
zone model is implemented within continuum type elements and the cohesive law is applied at
each integral point, DCZM uses 1D rod type elements and applies the cohesive law as the rod
internal force vs. nodal separation. A series of 1D interface elements are placed between node
pairs along the intended fracture path to simulate fracture initiation and growth. Dummy nodes are
introduced within the DCZM extract information regarding the mesh size and the crack path
orientation. To illustrate the DCZM, three popular fracture test configurations (double cantilever
beam -DCB, end notched flexure - ENF and mixed mode bending - MMB) are examined, and
results are presented that show mesh independence. Good agreement between the present approach
and previously published results is shown.
Results and Discussion
The proposed DZCM element was examined through application to analyze three popular
fracture test configurations; the DCB, ENF and MMB configurations. The (DCB) configuration
geometry is described in Figure 3 and the dimensions are: length (2L)=10 mm; width (B)=20 mm;
thickness (h)=1.5mm and the distance between the load point to the crack tip (a0)=30 mm. The
laminate properties in the analysis are: E11=135.3GPa; E22=E33=9.0GPa; G12=5.2GPa; Q12
=Q13=0.24; Q23=0.46 and fracture toughness for mode I is GIC=0.28 N/mm.
Figure 1 shows the meshes used to examine mesh sensitivity. Two uniform meshes have 100
elements and 400 elements, respectively, see Figure 1(a) and Figure 1(b). Two bias meshes with
400 elements have bias factors of 1.05/0.98 and 1.15/0.95, respectively, see Figure 1(c) and Figure
1(d). The insets show the details around the initial crack tip. The “crosses” indicate the position of
the DCZM elements placed along the intended crack path.
Figure 2 shows the results for the mesh sensitivity study. The thick dashed line is obtained
from a VCCT simulation which represents the linear elastic solution. The VCCT result is shown in
the figure as a reference. The DCZM results are shown in thin sold lines with different color. For
all these cases, the normal cohesive strength (VC) and the cohesive stiffness (both Kx and Ky) are
chosen to be 5.7MPa and 150×1012N/mm, respectively. To ensure mode I fracture, the GIIC and
shear cohesive strength, Wc, are set to be large compared to GIC and normal cohesive strength, Vc,.
The displacement increment is 0.01mm. It can be seen clearly that, (1) the shape of the four curves
agrees with each other; (2) the maximum loads are predicted very close to one another. The
conclusion from this is that the current DCZM approach is insensitive to the mesh size. Similar
results for the ENF and MMB configurations have been computed.
Conclusions
A new computational approach, the Discrete Cohesive Zone Model (DCZM), to model crack
initiation and growth has been presented and validated. For the examples studied, the present
740 D. Xie and A. M. Waas

approach is not sensitive to the mesh size, nor the load increment. The computations were carried
out on a SunBlade-100 machine with one processor (UNIX environment). For the model with 400
elements, the CPU time ranges from 93 to 101 seconds for the DCB (120 steps), 385 to 399
seconds for the ENF (240 steps and contact), and 188 to 205 seconds for the MMB (300 steps).
This is a drastic reduction in computation time compared to the CCZM approach which would take
on the order of hours for these corresponding simulations. Therefore, DCZM is a robust, efficient
and powerful tool that makes it possible to apply cohesive zone ideas in engineering analysis for
large scale crack growth problems when nonlinear effects are significant.
Acknowledgement
DX and AMW wish to thank the Aerospace Engineering Department at the University of .
Michigan for support.

Figure 1 FEA meshes for DCB of Mode I

Figure 2 Results for mesh sensitivity study

 43. Interfacial Fracture in Composites and Electronic Packaging Materials 741

SIGNIFINCE OF K-DOMINANCE IN DELAMINATION CRACKING IN

COMPOSITE LAMINATES

C. T. Sun and Z. Yang

School of Aeronautics and Astronautics, Purdue University
West Lafayette, Indiana 47907 USA
sun@purdue.edu

Delamination in composite laminates is basically an interfacial crack between two orthotropic

materials possibly oriented in different directions. One of the characteristics of interfacial cracks
between two dissimilar materials is the oscillatory behavior of the stress field near the crack tip [1-
2]. Classical fracture mechanics has been adopted for modeling delamination in composite
materials; stress intensity factors are commonly used to characterize delamination fracture
toughness. In using fracture mechanics with the stress intensity factors, one must make sure that
the near tip singular stress field be dominant in the fracture process zone
Different specimen geometries are known to have varying extents of K-dominance [3]. The K-
dominance region for interfacial cracks is also a function of the degree of material mismatch across
the interface [4]. O'Dowd et al. studied the K-dominance for two types of specimens made of
isotropic bimaterials [5]. Using a mixed-mode delaminating beam test specimen, Becker et al. [6]
demonstrated that the region of K-dominance near the crack tip was very limited. To the authors'
knowledge, however, no research work has been reported on the K-dominance of test specimens
made of advanced composite materials.
In the present paper, the significance of the K-dominance issue was addressed for delamination
in multidirectional composites. Two popular specimens for fracture testing of composites, the end-
notched flexure (ENF) specimen and the edge-delaminated (ED) specimen, were chosen for study.
The concept of degree of K-dominance was introduced to quantify the size of K-dominance region,
which was determined by comparing the K-based singular stress field and the full field stress
obtained with the finite element analysis. Effects of the specimen configuration, material
properties, loading condition, stacking sequence, and off-axis fiber orientation on the size of K-
dominance region were studied.
In order to investigate the K-dominance in a specimen, we compare the K-based singular stress
field with the full field stresses that can be either analytical solutions or numerical results generated
by other approaches such as the finite element method as is used in the present study. To
quantitatively characterize the K-dominance, we introduce the concept of degree of K-dominance
for mode I as

V zzk
DK I
V zzk  V zzf  V zzk
(1)
The degrees of modes II and III are defined in a similar manner.

We select a value, say 0.9, to determine the K-dominance region, r d Rk , within which we
have

DK I t 0 . 9 , DK II t 0 .9 , or DK III t 0 .9
742 C. T. Sun and Z. Yang

This is equivalent to saying that the K-based field is deemed to be able to characterize the full
field within the distance Rk along the interface ahead of the crack tip at which the degree of K-
dominance reaches 0.9. It is important to note that the K-dominance regions are not necessarily of
the same size for the three stress components.
Finite element analysis was performed for the ENF specimen with the layup
>0 10 / 90 / 0 9 / 0 //ș/ 0 9 / 90 / 0 10 @ where // indicates the location of the crack. As the stresses

V zz and V yz are very small, only the results based on the degree of K-dominance DK II are
plotted. Figure 1 shows the size of the K-dominance region R k for the stress component V xz
normalized by specimen thickness, t. It is seen that the size of K-dominance region varies with the
off-axis.

Figure 1 Size of K-dominance region vs. off-axis angle and specimen thickness

angle T . In the case of t / a 0 .2 for 0 / 90 interface the size of K-dominance region is

about a quarter of the AS4 composite ply thickness or only 4 times the diameter of a typical carbon
fiber. Consequently, it is questionable to use the stress intensity approach in LEFM for a composite
specimen when there is a big material mismatch and the specimen is very thin compared to the
crack length.

References
1. Williams, M. L., Bulletin of the Seismological Society of America, vol. 49, 199-204, 1959.
2. Qian, W. and Sun, C. T., Composites Science and Technology, vol. 57, 637-650, 1997.
3. Knott, J. F. 1973. Fundamentals of Fracture Mechanics, Butterworth, London.
4. Gu, P. 1993. "Multilayer material with an interface crack," Journal of Applied Mechanics, 60:
1052-1054.
5. O'Dowd, N. P., C. F. Shih, and M. G. Stout. 1992. "Test Geometries for Measuring Interfacial
Fracture Toughness," International Journal of Solids and Structures, 29: 571-589.
6. Becker Jr., T. L., J. M. McNaney, R. M. Cannon, and R. O. Ritchie. 1997. "Limitations on the
Use of the Mixed-Mode Delamination Beam Test Specimen: Effects of the Size of the Region
of K-dominance," Mechanics of Materials, 25: 291-308.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 743

EVALUATION OF INTERFACE TOUGHNESS BETWEEN SUBMICRON

ISLAND AND SUBSTRATE

Hiroyuki Hirakata, Takayuki Kitamura, Shohei Matsumoto and Yoshimasa Takahashi

Department of Engineering Physics and Mechanics, Kyoto University
Kyoto 606-8501, Japan
hirakata@kues.kyoto-u.ac.jp

Micro-electronics and micro-mechanical devices consist of low-dimensional components such as

islands, lines and films, and they have many intrinsic bi-materials interfaces. Stress concentration
near an interface edge due to the deformation mismatch sometimes causes delamination, which
brings about fatal malfunction of the device. On the other hand, the size required for the
components becomes submicron-scale in order to shrink the device. It is necessary, therefore, to
evaluate the interface strength among the submicron-components and the substrate. The size of the
stress-concentrated region near an interface edge, which affects the delamination crack initiation,
is dependent on the length scale; the field shrinks as the size of the component decreases [1]. For
submicron-components, the region size also becomes submicron-meter order. It is not clear
whether such a small region actually dominates the delamination or not. The purpose of the study
is to elucidate the mechanism dominating the crack initiation at the interface edge between a
submicron island and a substrate.
In this study, an experimental method [2, 3] for evaluating interface strength of a submicron
island on a substrate is developed using a modified atomic force microscope (AFM), which can
control and measure precisely the load and the displacement of the tip. This technique is applied to
chromium (Cr) islands of submicron scale on a silicon dioxide (SiO2) substrate. A loading tip is
positioned at the surface of the SiO2 substrate near the dot at first, and it is dragged horizontally
along the SiO2 surface under a constant displacement rate. The lateral as well as the normal load
and displacement of the tip are continuously monitored during the test by means of the modified
AFM. Figure 1 shows the lateral load Fl – displacement Gl curve. After the tip hits the island Fl
increases almost proportionally with Gl. The Cr island then is abruptly separated from the substrate
along the interface without nonlinearity in the load-displacement relation. No damages are
observed near the tip-contacted region of the island.
Figure 2 shows distributions of Vy and Wxy along the interaface at the crack initiation. A mixed-
mode singular stress field appears in the region of about 100 nm near the free-edge for both stress
components, and the intensity of Vy is higher than that of Wxy. The scatter in the intensities among
the specimens is relatively small. This suggests that interface strength for crack initiation between
the island and the substrate is dominated by the singular stress field, and the critical stress
intensities, KVC and KWC, can be evaluated as about 60 MPaPm0.36 and 16 MPaPm0.57,
respectively.
744 H. Hirakata et al.

FIGURE 1. Relationship between lateral load and lateral displacement.

FIGURE 2. Stress distributions along interface between Cr and SiO2 at the crack initiation.

References
1. Kitamura, T., Hirakata H. and Itsuji, T., Engng. Fract. Mech., vol. 70, 2089-2101, 2003
2. Hirakata H., Kitamura, T. and Yamamoto, Y., Int. J. Solids Struct., vol. 41, 3243-3253, 2004
3. Hirakata H., Kitamura, T. and Yamamoto, Y., JSME Int. J. A, vol. 47, 324-330, 2004
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 745

THREE-DIMENSIONAL THERMAL STRESS ANALYSIS CONSIDERING THE

STRESS SINGULARITY FOR BONDED STRUCTURES

H. Koguchi
Nagaoka University of Technology
1603-1 Kamitomioka, Nagaoka, Niigata, 940-2188 Japan
koguchi@mech.nagaokaut.ac.jp

It is well known that stress singularity occurs at a vertex in joints under mechanical loadings and
thermal loadings. In the present paper, a singular stress field at a vertex in three-dimensional
bonded structures as shown in Fig.1 is analyzed. Stress analysis and eigen analysis are performed,
and the intensity of singularity is determined varying mechanical property of thin layer
sandwitched between two substrates. Furthermore, thermal stresses at the vertex on an interface in
the bonded structures are expressed appropriately using the results in eigen analysis, and the
intensity of singularity is determined and arranged for the order of stress singularity using a useful
parameter.
In joints using adhesives, the thermal residual stresses will influence on the strength of joints.
In the present study, the residual stress distribution in three-dimensional bonded stuructures is
calculated by BEM, and a boundary integral equation is described as follows.

cij  P  u j  P  ³ U  P , Q  t dS  Q   ³ T  P , Q  u dS  Q 
*
ij j
*
ij j
* * (1)

where Uij* and Tij* are fundamental solutions for displacements and tractions. Observation point,
P, and source point, Q, are located on the boundary. tj and uj are traction and displacement
vectors, respectively. Using the fundamental solution of two-phase materials, mesh division on the
interface is unnecessary, and displacements and stresses at any points in the domain are determined
accurately. The thickness of the adhesive is fixed at 0.5mm. Materials used in the analysis and
elastic moduli are shown in Table 1. In this analysis, Material 1 and Material 3 have the same
property, and the stress distribution near the vertex of Material 1 and Material 2 is investigated. In
this analysis, the order of stress singularity is deduced from an eigen analysis of FEM. Then, we
can expressed stress distribution near the vertex in joints as follows (Koguchi [1],[2]).

Fig.1 A model for bonded structures

746 H. Koguchi

n
O j 2
V TT ¦ KTT T , I  r
j 1
j  KTT 2 T , I   KTT 3 T , I  ln r  KTT 4 T , I  ln r 
(2)
'
Munz and Yang [3] presented that the thermal stresses normalized by V 0 ' D ˜ ' E ˜ ' T
could be used for arranging the intensity of singularity in two-dimensional joints. We determine
the coefficients, KTTk (k=1,2,3,4) using a least square method. After that, we arrange the
coefficients against the order of stress singularity, O, using the following parameter
1
V 0c ' T ^1 E 1D 1  1 E 2D 2 ` . Furthermore, we make an approximation of coefficients
against 'D=D1-D2. Finally, we will deduce explicitly the expressions for thermal stress near stress
singularity vertex. We will show how to evaluate the distribution of thermal stress near the vertex
at an arbitrary temperature.

Table 1 Material properties used in analysis

Fig. 2 Intensity for power-law singularity

References
1. Prukvilailert, M., Koguchi, H., Int. J. Solids Structures, vol.42, 3059-3074, 2005.
2. Koguchi, H., Int. J. Solids Structures, vol.34, 461-480, 1997.
3. Munz, D. and Yang, Y. Y., J. Appl. Mech, vol.59, 857-861, 1992.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 747

CENTER OF DILATATION AND PENNY-SHAPED CRACK IN VISCOELASTIC

BIMATERIAL

K. T. Chau, R. C. K. Wong1 and Y. Z. Sun

Department of Civil and Structural Engineering, The Hong Kong Polytechnic University,
Kowloon, Hong Kong, CHINA
1 Department of Civil Engineering, The University of Calgary, Calgary, Alberta, CANADA

cektchau@polyu.edu.hk

This paper derives analytically the stress intensity factors (SIFs) of a frictional interfacial penny-
shaped crack in a viscoleastic bimaterial subject to the action of a center of dilatation (see Fig. 1).
This work is an extension of the 2-D problem considered by Chau and Wong [1].The standard
linear viscoelastic solid or “three-parameter-viscoelastic-model” is adopted. Using the
correspondence principle, the problem is formulated in Laplace transform space. The problem is
decomposed into two Auxiliary Problems: (I) a bimaterial containing a center of dilatation; and (II)
an interfacial crack in the bimaterial subject to tractions that cancel those induced by the Auxiliary
Problem I. The elastic solution of Auxiliary Problem I has been given by Yu and Sanday through
the use of Galerkin vector in 3-D space, which can easily be extended to that of a viscoelastic
bimaterial subject to a sudden applied center of dilatation. Auxiliary Problem II can be solved by a
Fourier transform technique proposed by Shifrin et al. [2] through the solution of a
pseudodifferential equation subject to an arbitrary boundary traction condition. The applied
traction from Auxiliary Problem I is first approximated by polynomials, and balancing coefficients
yield a system simultaneous equations. The stress intensity factors can then be obtained. The
approximate inverse of Laplace transform of Schapery is used to obtain the solutions in time.
Creeping tests for the Swiss Central Alps shales have been adopted for viscoelastic parameter
calibration. Frictional and overburden effects are also incorporated. For the limiting case of
homogenous material, our solution agrees with the isotropic case [3]. When the number of terms
needed for polynomials interpolation equal 15, the solutions converge to steady solution. In
contrast to the 2-D cases, all mode I, II and III may appear. The maximum modes I and II stress
intensity factors normally appear at the point closest to the center of dilatation, which mode III
becomes maximum. Short term crack stability does not necessarily rule out long term crack
propagation. If the overburden stress and the friction on crack surface are neglected, both modes I
and II SIFs are induced at the crack tips. When both friction and overburden pressure are included
in the analysis, tensile cracking is unlikely whilst shear cracking remains possible. The maximum
mode II SIF strongly depends on the orientation as well as the distance of the center of dilatation
from the crack tip. Both overburden pressure and crack face friction decreases chance of crack
propagation.
Acknowledgements: The work reported in this paper was fully supported by PolyU research
project (No. G-T911).
748 K. T. Chau et al.

Figure 1. Sketch of the viscoelastic model containing a interfacial penny-shaped crack.

Figure 2. Convergence check on the stress intensity factors d.

References:
1. Chau K.T. and Wong R.C.K. (2004). Interactions of a center of dilatation and an interface
crack in two-dimensional visoelastic bimaterial. Submitted to the Keer’s special issue of
Mechanics of Materials.
2. Shifrin E.I., Brank B. and Surace G. (1998). Analytical-numerical solution of elliptical
interface crack problem. Int. J. Fract. 94 (3), 201-215.
3. Chau K.T., Yang X. and Wong R.C.K. (2000). Interactions of a penny-shaped crack with a
center of dilatation in an elastic half-space. Mech. Mater. 32, 645-662.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 749

FRACTURE ANALYSIS ON POPCORNING OF PLASTIC PACKAGES DURING

SOLDER REFLOW

S. W. R. Lee and D. C. Y. Lau

Electronic Packaging Laboratory
Center for Advanced Microsystems Packaging
Hong Kong University of Science & Technology
Clear Water Bay, Kowloon, Hong Kong
Tel: +852-23587203, Fax: +852-23581543
rickylee@ust.hk

Moisture-induced cracking during the solder reflow process is a critical reliability problem for
plastic packages. Figure 1 shows the schematic diagram of a plastic ball grid array (PBGA)
package with either a crack at the die attach or at the interface between solder mask and copper.
These types of internal cracking are mainly resulted from a combination of thermal expansion
mismatch of the package and the moisture-vaporized pressure acting on the crack surface inside
the package. Such a phenomenon is known as the “popcorning” problem (Kuo et al. [1]).
In this study, the finite element simulation is performed to determine the deformation field of the
plastic package during the solder reflow process. Fracture mechanics is employed to characterize the
crack tip properties after the crack initiation. Two different methods, namely, the crack tip opening
displacement (CTOD) and virtual crack closure technique (VCCT) are used to determine the strain
energy release rate, stress intensity factors, and phase angle at the crack tip for different crack lengths
and temperatures. Two different sets of loading are imposed to the cracked structure. The first one is
the thermal stress loading that is due to the thermal expansion mismatch of the package. The second
one is the steam pressure loading. During the solder reflow process in surface mounting, the moisture
absorbed in the plastic package is vaporized and exerts a pressure on the internal crack surfaces of
the package. The vapor pressure is estimated following the equation below:

(1)
where C is a constant (0.485 atm), Ti = (273+85)°K, and Tf (K) is the reflow temperature.

In the present analysis, eight-node plane strain elements are used to model the plastic package.
The cracks in the die attach adhesive and along the interface between solder mask and copper are
simulated by an array of double nodes. In addition to uniform temperature loading, normal traction
is applied to the crack surfaces to account for the steam pressure originated from the absorbed
moisture in the package. It should be noted that, due to the thermal mismatch between materials
750 S. W. R. Lee and D. C. Y. Lau

and the applied pressure, both types of cracks have mixed-mode fracture characteristics.
Furthermore, since the crack surfaces are no more traction-free, the conventional J-integral
formula may not be valid (Rice [2]). In order to evaluate the fracture parameters at the crack tip,
two other approaches are employed. According to Hutchinson and Suo [3], the relationship
between stress intensity factors (SIFs, i.e., KI and KII) and CTODs (Gx and Gy) can be expressed in
a complex variable form as

(2)
where i is the imaginary unity, r is the distance from the crack tip, and x and y are the directions
parallel and perpendicular to the crack, respectively (Fig. 2). Besides, E* and H denote the effective
properties of materials adjacent to the crack. Subsequently, by the formula of Malyshev and
Salganik [4], the strain energy release rate can be obtained as

(3)
Another computational fracture mechanics approach, namely, VCCT, is used as well in this
study for comparison. With this method, a single element at the crack tip is artificially closed.
Consequently, the strain energy release rate may be evaluated by calculating the work per unit
length required to close the crack at the tip. In order to reduce the cost of computation, an
assumption of self-similarity at the crack tip was introduced by Rybicki and Kanninen [5].
Therefore, the formula to evaluate the strain energy release rate by the modified VCCT can be
written as

(4)
where Fx and Fy denote the nodal forces in the x and y direction, respectively (Fig. 3). In this study
the strain energy release rates derived from both Eqs. (3) and (4) are evaluated and compared (Fig.
4). The good agreement verifies the accuracy of results obtained from the present finite element
analysis with fracture mechanics approaches.

References
1. Kuo, A. Y., Chen, W. T., Nguyen, L. T., Chen, K. L. and Slenski, G., “Popcorning-A Fracture
Mechanics Approach,” Proceedings of 46th Electronic Components and Technology
Conference, IEEE Publications, New Jersy, 1996, 869-874.
2. Rice, J. R., “Elastic Fracture Mechanics Concepts for Interfacial Cracks,” ASME J. Appl.
Mech., vol. 55, 98-103, 1988.
3. Hutchinson, J. W. and Suo, Z., “Mixed Mode Cracking in Layered Materials,” Adv. Appl.
Mech., vol. 29, 64-187, 1992.
4. Malyshev, B. M. and Salganik, R. L., “The Strength of Adhesive Joints Using the Theory of
Cracks,” Int. J. Fract. Mech., vol. 1, 114-128, 1965.
5. Rybicki, E. and Kanninen, M., “A Finite Element Calculation of Stress Intensity Factors by a
Modify Crack Closure Integral,” Eng. Fract. Mech., vol. 9, 931-938. 1997.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 751

DELAMINATION OF PB-FREE FLIP CHIP UNDERFILL DURING 2ND LEVEL

INTERCONNECT REFLOW

Soonwan Chung, Zhenming Tang and Seungbae Park

Department of Mechanical Engineering, SUNY at Binghamton
Binghamton, NY 13902-6000
sbpark@binghamton.edu

As the need for the use of Pb-free solder in electronic packaging fields increase, many researches
have been conducted to ensure the reliability of the new solder systems. Currently, Sn-based
solders are generally accepted as the most promising Pb-free solders. Most of Sn-based solders
(e.g. Sn/Ag, Sn/Cu, Sn/Ag/Cu) have similar melting temperatures between 220 and 240. This
transition forces electronics industry to lose soldering hierarchy between the first and second level
packaging. Specifically, the first level interconnect (flip chip) will be melt again during second
level interconnect (BGA) reflow which allows up to 260. In the mean time, the first level
interconnect, which is encapsulated by underfill material, experiences phase change from solid to
liquid. Similar to most of metals, Sn-based Pb-free solders have volume expansion as much as
4.5% during the phase change. It puts hydrostatic pressure on the surrounding materials such as
underfill encapsulation, chip, and substrate and ultimately delaminate the interfaces or crack the
underfill material. The pressurized solder is interjected through the failure and make short circuit.
In our earlier work, the impact of phase change of Pb-free solder was numerically investigated
and the interjection of melted solder and the interfacial failure of underfill were observed in flip
chip plastic ball grid array packages as shown in Fig.1 [1]. In this study, adhesion characteristics
such as interfacial fracture toughness between underfill and chip passivation/solder mask are
measured in various temperatures (above melting temperature of Pb-free solder). Although the
adhesion strengths of underfill with Pb-free solder flip chip package components are available at
room temperature [2,3], there are little reports regarding adhesion strengths of underfill at high
temperature. It is general understanding that adhesion strength decreases as temperature increases
and the quantitative analysis is extremely required as a delamination criterion of the interfaces.
For the measurement of interfacial fracture toughness, CMM (Compact Mixed Mode) fracture
test [4] is performed. Interfacial edge cracked specimens of sandwich type are prepared by
cleaning the surface to be bonded and a piece of Teflon film is applied onto the surface of chip
passivation or solder mask for an initial crack. Underfill encapsulant is dispensed into a gap
generated by plastic shim and cured according to the recommended procedure. The die or substrate
plates are diced with the depth of 4mm and they are baked out in an oven at 100 to remove any
remaining moisture. Then aluminium blocks are bonded on to the substrate by using a room-
temperature-cured epoxy. The specimen is mounted in the loading fixture made of stainless steel
and tested in Instron testing machine under displacement control. The CMM specimen and fixture
are surrounded by the environment chamber which can regulate the temperature. The load-
displacement curve is recorded and the critical strain energy release rate is calculated by using the
compliance method as follows.

Pc2 wC
Gc
2b w a (1)
752 S. Chung et al.

where Pc is the peak load associated with the onset of crack growth, C is the compliance, which

G G a
is equal to P, is the displacement, b is the depth and is the crack length.

(a) solder interjection through void (b) Interfacial failure between underfill & substrate
FIGURE 1. Cross-section pictures around solder balls of FC-PBGA

References
1. Soonwan Chung, Zhengming Tang and Seungbae Park, “Investigation of phase change of flip
chip solders during the second level interconnect reflow,” 55th Electronic Components and
Technology Conference, May 31-June 3, 2005, Lake Buena Vista, FL.
2. Xiang Dai, Mark V.Brillhart and Paul S.Ho, “Adhesion measurement for electronic
packaging applications using double cantilever beam method,” IEEE Transactions on
Components and Packaging Technology, 23(1), 2000, pp.101-116.
3. Man-Lung Sham and Jang-Kyo Kim, “Adhesion characteristics of underfill resins with flip
chip package components,” Journal of Adhesion Science and Technology, 17(14), 2003,
pp.1923-1944.
4. Hock Lye John Pang, Xueren Zhang, Xunqing Shi and Z.P. Wang, “Interfacial fracture
toughness test methodology for adhesive bonded joints,” IEEE Transactions on Components
and Packaging Technologies, 25(2), 2002, pp.187-191.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 753

RELIABILITY OF INTERFACES BETWEEN COMPONENTS IN ADVANCED

ELECTRONIC PACKAGES UNDER SOLDER REFLOW PROCESS

T. Ikeda and N. Miyazaki

Department of Mechanical Engineering and Science
Graduate School of Engineering, Kyoto University
ikeda@mech.kyoto-u.ac.jp

In order to the downsizing of electronic packages, complicated structures of adhesively jointed

dissimilar materials are used in many advanced packages. In these packages, jointed interfaces are
sometimes delaminated by the combined effects of thermal stress, vapor pressure during solder
reflow process and so on. Qualitative analysis of the delamination is desired to develop effective
materials and to perform reliable design to protect the delamination. In this study, we focused on
the delamination in advanced electronic packages under solder reflow process.
The delamination toughnesses of interfaces in electronic packages are evaluated using the
stress intensity factors of an interface crack between dissimilar materials. If a coordinate system
around an interface crack is defined as shown in Fig. 1, the asymptotic solution of the stress
components along the x-axis in the vicinity of an interface crack tip is expressed as

(1)

where KI and KII represent stress intensity factors of the interface crack for Mode I and Mode II
deformations respectively, lk is a characteristic length, which can be taken as an arbitrary constant,
and H is a bimaterial constant [1].
The diffusion analysis of the moisture absorption of an electronic package was performed, the
results of which were utilized to calculate the vapor pressure in the package during the solder
reflow process. The moisture absorption of materials in an electronic package is assumed to be
dominated by Fick's law.
w------
C
= D ’2C (2)
wt

Fig. 1 Coordinate system around an interface crack.

where C is moisture concentration, D is the diffusion coefficient, and t is time. In order to remove
the discontinuity of moisture concentration between dissimilar materials, moisture concentration is
normalized with the saturated concentration of respective materials.
754 T. Ikeda and N. Miyazaki

C
w=
Csat (3)
The vapor pressure in a small gap between materials during the solder reflow process is
predicted by

PV = PSV  C max t H PSV at TR °C  ½

°
C ¾
PV = max  C max < H PSV at TR °C  °
H ¿ (4)
where PV is predicted vapor pressure in the gap, PSV is saturated vapor pressure, Cmax is the
maximum moisture concentration, H is Henry’s law coefficient, TRC is solder reflow temperature
[2].
The delamination often occurs from a corner of a component in an electronic package as
shown in Fig. 2. It is considered that higher thermal residual stress causes a small delamination
around the corner and it is expanded by the vapor pressure under the solder reflow process. The
stress intensity factors along a crack front around a corner of a component in an electronic
package, those are induced by the combination of thermal stress and vapor pressure during solder
reflow process, are analyzed using the finite element method.
Generally, soft adhesive resin can reduce thermal residual stress but have relatively low
delamination toughness at high temperature. The balance between the stiffness of resin and the
delamination toughness is important to select a material in advanced electronic packages.
Presented qualitative evaluation of the delamination can provide usefull information for balancing
these conflicting factors.

Fig. 2 Delamination from corners of components in an electronic package.

References
1. F. Erdogan, “Stress distribution in a nonhomogeneous elastic plan with crack”, J. of Applied
Mechanics, vol. 30, 1963, pp232-236.
2. T. Ikeda, I. Arase, Y. Ueno, N. Miyazaki, N. Ito, M. Nakatake, and M. Sato, "Strength
evaluation of plastic packages under solder reflow process using stress intensity factors of V-
notch", J. Electronic Packaging, vol. 125, 2003, pp. 31-38.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 755

THREE-DIMENSIONAL STRESS INTENSITY FACTORS ANALYSES OF

INTERFACE CRACKS BETWEEN DISSIMILAR ANISOTROPIC MATERIALS

Masaki Nagai, Toru Ikeda, Noriyuki Miyazaki

Department of Mechanical Engineering and Science
Graduate School of Engineering, Kyoto University
ikeda@mech.kyoto-u.ac.jp

The down sizing of electronic devices, the incredible progress of micro-electro-mechanical system
(MEMS) and the development of composites materials demand the effective technique to evaluate
the reliability of interfaces between dissimilar materials. The main cause of interface fracture is
thermal stress due to the difference of thermal expansion of jointed materials. Many of these
materials are anisotropic materials like composite materials, single crystals and so on.
Hwu [1] proposed the asymptotic solution of stress around an interface crack between
dissimilar anisotropic materials using the Stroh formalism [2], and defined the stress intensity
factors that are compatible with those for a crack in homogeneous material.

­ K II ½  iH D ­ V 12 ½
° ° § r · ° °
K ® KI ¾ lim 2 Sr / ¨ ¸ /  1 ® V 22 ¾
°K °
ro0
© lk ¹ °V °
¯ III ¿ ¯ 32 ¿ (1)

/ >O 1 , O2, O3@

(2)

where r is the distance from a crack tip, and lk is an arbitrary characteristic length. denotes
the diagonal matrix. O1, O2, and O3 are eigenvectors of the interface. These stress intensity factors
of an interface between dissimilar anisotropic materials are very useful to evaluate the strength of
jointed interfaces.
We proposed a new numerical method based on the energy method to calculate the stress
intensity factors of an interface crack between dissimilar anisotropic materials [3]. In this paper,
we extended the previous method for three-dimensional problems. This numerical method is
usually used with the finite element method. In three-dimensional problems, the cost of preparing
huge input data is troublesome for actual users, especially in industries. In the presented method,
the stress intensity factors are automatically calculated using the contour integral method and the
moving least square method. Users of this method need to prepare a small amount of data for the
SIF calculation.
J-integral of an interface crack in three-dimensional structure is calculated by following
equation.

(3)
756 M. Nagai et al.

FIGURE 1. Contour integral of a three-dimensional crack.

where W is the density of strain energy, Ti is traction, ui is displacement, * is the contour of

integral and A is the area inside the contour as shown in Fig. 1. To reduce the cost of data
preparation, each term of this equation is automatically calculated from the nodal displacements
along a crack front using the moving least square method.
The J-integral must be separated into each mode of stress intensity factors. Because the stress
intensity factors of an interface crack between dissimilar anisotropic materials are not independent.
Each mode of stress intensity factors depends on others. We utilized the superposition method for
the separation. In this technique, asymptotic solutions shown as Eq. (1) for three sets of stress
intensity factors are superposed on the target problem which should be solved. If we superposed
three independent asymptotic solutions on the target problem, three modes of stress intensity
factors will be obtained.
We applied this technique to some typical three-dimensional problems of interface cracks
between anisotropic materials. Obtained stress intensity factors did not depend on the path of
contour integral except too close path around a crack tip. This method can calculate the stress
intensity factors of an interface crack between dissimilar materials in a 3D jointed structure
accurately. This method needs only nodal displacements along a crack front obtained by a
commercial FEM code and small amount of data those indicate the shape of a crack and the
locations of contours of J-integral.

References
1. Hwu. C, Explicit Solutions for Collinear Interface Crack Problems, International Journal of
Solids and Structures, 30, 1993, 301-312.
2. Stroh, A. N., 1962, Steady State Problems in Anisotropic Elasticity, Journal of Mathematical
Physics, 41, 77-103.
3. Toru Ikeda, Koh Yamanaga, Masaki Nagai and Noriyuki Miyazaki, Stress Intensity Factors
Analysis of an Interface Crack between Anisotropic Dissimilar Materials under Thermal
Stress, Proceedings of European Congress on Computational Methods in Applied Sciences
and Engineering (ECCOMAS 2004), Jyväskylä Finland, 2004, CD-ROM.
 43. Interfacial Fracture in Composites and Electronic Packaging Materials 757

MOLECULAR DYNAMICS OF INTERFACIAL FRACTURE

T. E. Tay, V. B. C. Tan and M. Deng

Department of Mechanical Engineering, National University of Singapore,
9 Engineering Drive 1, Singapore 117576
mpetayte@nus.edu.sg

There is growing realization that many fracture phenomena straddle various length scales. This is
particularly so in interfacial fracture between dissimilar materials and in composites, where
conventional continuum mechanics approaches do not explicitly model the interfacial failure
mechanisms at the molecular level. Consequently, several issues such as the effects of local mode
mixity on fracture toughness have not been satisfactorily resolved. With recent advances in
molecular dynamics (MD) and multi-scale modelling techniques, the chasm between molecular or
nano-scale characteristics and macro-scale properties relevant to fracture is beginning to be
narrowed, and may eventually be bridged.
This presentation begins with a brief review of MD and multi-scale simulation and their
relevance to interfacial fracture and damage in fiber-reinforced composite materials. Much of the
MD literature in fracture concerns metallic systems, with few publications in MD of polymeric
systems [1-2]. There are also very limited experimental investigations of crack initiation in
polymeric interfacial systems at the molecular level [3]. Although many challenges remain, some
pertinent parameters such as strength of adhesion may be related to the molecular and chemical
interactions. We present recent results of a classical atomistic MD simulation of a diffused
polymer interface (Fig. 1) [1], and a coarse-grain MD simulation of a highly cross-linked epoxy-
silane system (Fig. 2). For the latter case, ab initio MD is used to develop the intermolecular force
fields, which are in turn employed in the coarse-grain (CG) bead-spring models. The CG technique
enables modeling of highly cross-linked polymer systems and permits bond scission, a prelude to
the nucleation of micro-voids and cracks.

FIGURE 1. A polycarbonate-silane (PC-AMPTES) interface atomistic MD model.

758 T. E. Tay et al.

FIGURE 2. Failure at an epoxy-silane-glass interface in a coarse-grain MD model.

Reference
1. Deng M., Tan V.B.C. and Tay T.E., Polymer, vol. 45 (18), 6399-6407, 2004.
2. Stevens, M.J., Macromolecules, vol. 34, 1411-1415, 2001.
3. Kent, M.S., Reedy, E.D., Yim, H., Matheson, A., Sorenson, J., Hall, J., Schubert, K., Tallant,
D., Garcia, M., Ohlhausen, T. and Assink, R., J. Maters. Res., vol. 19 (6), 1682-1695, 2004.
C. SPECIAL SYMPOSIA/SESSIONS

C2. Engineering Materials and Structures

 4. Fracture and Fatigue of Elastomers 761

NUCLEATION, GROWTH AND INSTABILITY OF THE CAVITATION IN

RUBBER

E. Bayraktar1,2, K. Bessri1 and C. Bathias1

1ITMA/CNAM- Arts et Métiers, School of Mechanical Engineering, Paris, France
2Supmeca/LISMMA-Paris, School of Mechanical and Manufacturing Engineering, France

bayraktar@supmeca.fr, kamel.bessri@cnam.fr, bathias@cnam.fr

Elastomeric Matrix Composites (EMCs) subjected to static and fluctuating loads basically fail due
to the nucleation and growth of defects. In fact, higher hydrostatic pressure influences mechanical
behaviour of EMCs. In other words, the change in behaviour of EMCs due to the nucleation of
cavitations under the hydrostatic pressure is evaluated for understanding of the mechanics
underlying the damage mechanism. Two types of specimens are used in this study; Natural rubber,
NR vulcanised and reinforced by carbon black and Synthetic rubber (styrene-butadiene-rubber,
SBR).
It is generally accepted that under static loading conditions, elastomeric Matrix Composites
(rubbers) are considered as isotropic hyperelastic incompressible materials. Because a rubber
material element cannot be extended to infinite stretch ratio, a damage mechanism at large strain is
considered. Cavitation in rubber particles plays an important role in the toughening mechanism of
rubber-modified plastics [1-7]. Indeed, cavitation in elastomers is thought to be initiated from
flaws, which grow primarily due to a hydrostatic tensile stress, and ahead of the crack there will be
not only a high stress perpendicular to the plane of the crack but also significant stress components
in the other direction. [1-3, 5, 7].
The most popular idea about the cavitation phenomenon says that the cavitation is related to
the existence of the gas bubbles trapped in the material during the production stage and the
growing of the cavities would then be the result of the growing the gas bubble. Instable failure
mechanism at the end of the cavitation is not well known too. Here, an experimental study was
carried out on the cavitation phenomenon of the pancake shaped specimens from NR and SBR
(smooth and notch). Initiation and propagation stages of cavity and also the instability conditions
of the cavities have been determined by simple static tensile testing.
In order to complete the exploration of the cavitation in NR and SBR, we have conducted a
comparative study on the effect of the hydrostatic pressure and the effect of the stress triaxiality at
the tip of the notch. So, tensile tests conducted on the smooth and notched specimens, of whom the
depths varied from 2 to 8 mm, showed that a very strong hydrostatic pressure at the centre of the
specimen governs the damage. However, the effect of the stress triaxiality increases at the tip of the
notch. So, a real competition between the hydrostatic pressure and the stress triaxiality should be
attended just from the beginning of the deformation, which characterises the damage mechanism
of the NR and SBR. In-situ observations of uniaxial tensile testing are also presented by using X-
rays computed tomography, CT, (medical scanner) and compared with those of the scanning
electron microscopy (SEM).
762 E. Bayraktar et al.

References
1. Rivlin R. S. and Thomas A. G., Journal of polymer science, 10, Vol. 291-318, 1953.
2. Chang W. J. and Pan J., J. of Mater. Science, Vol. 36, 1901-1909, 2001.
3. Lake G. J., Thomas A. G. and Lawrence C. C., Polymer, Vol. 33, (19), 4069-4074, 1992.
4. Bayraktar E., Antolovich S. and C. Bathias, In Proceedings of ICFC3 on the 3rd International
Conference on Fatigue of composites, 13-15 September 2004, Kyoto -Japan
5. Gdoutos E. E., Schubel P. M. and Daniel I. M., Strain, Vol. 40, 119-125, 2004.
6. Jancar J., Dianselmo A. and Dibenedetto A. T., Polymer, Vol. 34, (8), 1684-1694, 1993.
7. Bayraktar E., Montembault F. and Bathias C., Journal of Materials Science and Technology,
Vol. 20 (1), 27-31, 2004.
 4. Fracture and Fatigue of Elastomers 763

ENGINEERING FRACTURE MECHANICS FOR CRACK TOUGHNESS

CHARACTERISATION OF ELASTOMERS

Katrin Reincke1, Wolfgang Grellmann1 and Gert Heinrich2

1Institute of Materials Science, Martin-Luther University Halle-Wittenberg,D-06099 Halle/Saale,

Germany2 Leibniz Institute of Polymer Research, Hohe Straße 6 D-01069 Dresden, Germany
katrin.reincke@iw.uni-halle.de

Elastomer materials are used in a wide application range and subjected to different loading from
which material failure can result. Because this failure is caused by initiation and propagation of
cracks, application of fracture mechanics methods for material assessment is obvious and the
assessment of materials fracture properties have fundamental importance. Many papers in the
literature deal with fracture mechanical investigations of elastomers under cyclic loading.
However, fracture mechanical experiments under quasi-static and especially impact-like loading
have been described to a lesser extent.
This paper shows performance, analysis and results of different methods of technical fracture
mechanics and their possibilities for the assessment of crack resistance behaviour of elastomeric
materials. Experimental methods that are described in detail are the instrumented notched tensile
impact test (NTI) and a quasi-static fracture mechanics test. By using these two tests, it is possible
to get information on the materials resistance against unstable and stable crack initiation and
propagation, respectively. The instrumented NTI test (see Fig. 1) can be favourably performed
with thin and/or flexible specimens. Due to recording of the load–extension (F–l) diagram, a
fracture mechanical analysis is possible for these materials/specimens. This means, one can
determine toughness parameters Jd describing the resistance against unstable crack propagation
under impact-like loading conditions (pendulum hammer speed up to 3.7 m/s) also for elastomers.

FIGURE 1. Experimental configuration of the instrumented notched tensile impact test (a) and a
schematic representation of a load–extension diagram of a razorblade-notched elastomeric
specimen

An example of application of instrumented NTI tests is shown in Fig. 2. Two elastomeric

materials were exposed different times to lye at a temperature of 95 °C. In Fig. 2 a, one can see Jd
as a function of exposure time; Fig. 2 b shows the dependence of the conventional tear resistance
W on the exposure time. It can be seen that the resistance against initiation and propagation of an
unstable crack decreases with increasing exposure time; a similar result was found for W.
However, while for Jd no significant differences between the two elastomers were found, this
means, both materials have the same crack resistance behaviour, for the value W, elastomer 2
shows higher tear resistance. In contrast to Jd, in the industrial praxis W is used for a description of
764 K. Reincke et al.

the resistance of elastomeric materials against tearing of an existing crack. In this case, if one uses
only W, an overestimation of the crack resistance behaviour of elastomer 2 is done.

FIGURE 2.Crack toughness related to unstable crack propagation Jd (a) and tear resistance W (b)
as function of exposure time for two different elastomers

Furthermore, a quasi-static fracture mechanics test will be described in the paper. This test is a
single-specimen test, where the deformation and fracture process is recorded with a video camera.
Aim of the experiment is to record crack resistance curves for a quantitative characterisation of the
stable crack initiation and propagation behaviour. Examples of such crack resistance curves are
shown in Fig. 3 for NR vulcanizates with different amounts of an organically modified nano-
dispersed layered silicate (OMLS). Increasing the amount of layered silicate leads to a change in
crack propagation behaviour. At low filler contents, the nature of crack propagation can be
described with stick-slip behavior, while at higher amounts a steady tearing occurs. This means,
the strain crystallization of the rubber matrix is disturbed.

FIGURE 3. Crack resistance curves of NR compounds filled with different contents of an

organically modified layered silicate
 4. Fracture and Fatigue of Elastomers 765

MULTIAXIAL FATIGUE CRACK INITIATION ON FILLED RUBBERS :

STATISTICAL ASPECTS

L. Laiarinandrasana, A. Bennani and R. Piques

Centre des Matériaux – Ecole Nationale Supérieure des Mines de Paris- UMR CNRS 7633
BP87 – F 91003 Evry Cedex - FRANCE
lucien.laiarinandrasana@ensmp.fr

Nowadays, white reinforcement fillers such as precipitated silica particles are more and more used
in many industrial products. In particular, they are reported to increase tear resistance and to reduce
internal heating. In this work, two precipitated silica reinforcements, which essentially differ in
their specific surface area, are considered. They have been incorporated into a natural rubber
matrix. Two reinforced rubber materials have thus been elaborated.
Experiments on both filled rubber materials consist of cyclic tension/compression and torsion
tests, carried out on dog-bone shaped specimens. All tests have been pursued until crack initiation
defined as a visible millimetric crack on the outer surface of the specimens. Microscopic
examinations are utilised in order to find out the location of the crack and to depict the microcrack
growth mechanisms. It seems that, first, void nucleation occurs at the interface between coarse
inclusions and the rubber matrix; then, the crack propagates up to 1mm in size through the
"composite" filled rubber materials.
The mechanical behaviour of both materials has been modelled by means of finite element
(FE) analyses. This allows assessment of the local stress state inside the specimen. Then, with the
help of local parameters computed by FE analysis, the fatigue lifetime is characterised. In
particular, it has been reported [1] that the fatigue lifetime plots of both compounds seemed to
merge into a unique curve, which will be considered as the deterministic one.
The statistical aspects are carried out by determining the mean amount of particles per unit
volume, D by using a quantitative mapping technique. Thus, 1/D corresponds to the mean volume
Vu containing one inclusion. By considering that the probability to find no inclusion within a
volume V reads:

§ V ·
P exp ¨¨ 
© Vu ¹ (1)
it comes out that the volume VX that has the probability X to contain one particle is :

VX V u ln(1  X )
(2)
where ln stands for naeperian logarithm.
Then, void nucleation occurs when the largest principal stress in the vicinity of a coarse
particle is greater than a critical value Vc. Let VV be the volume of material in the investigated
geometry where the stress is greater than Vc. The aforementioned statistical approach can be
applied by setting: VX = VV.
It allows to plot fatigue curves (Vmax versus number of cycles to crack initiation) with various
amounts of risks of crack initiation: e.g. 10%, 50% and 90%. Experimental scatter will be
compared with 10% and 90% risks fatigue curves bounds. The statistical treatment of the fatigue
database has been implemented as a post-processor into a FE code. The extension of this
766 L. Laiarinandrasana et al.

probabilistic approach to fatigue lifetime plots accounting for the mean/minimum stress effects is
also investigated.

Reference
1. Laiarinandrasana, L., Bennani, A., Cantournet, S. and Bomal, Y., to appear In Proceedings of
the Fourth European Conference on Constitutive Modelling of Rubber, Stockhölm 27-29
June 2005
 4. Fracture and Fatigue of Elastomers 767

FRACTURE CRITERIA OF RUBBER-LIKE MATERIALS UNDER PLANE

STRESS LOADINGS

A. Hamdi1*, M. Nait-Abdelaziz1, N. Ait Hocine2 and P. Heuillet3

1LML, UMR CNRS 8107, Polytech’Lille, Villeneuve d’Ascq, France
2Laboratory of Rheology, Brest, France
3LRCCP, 94400 Vitry Sur Seine, France
*adel.hamdi@polytech’lille.fr

The use of rubber like materials is now very widespread in industry. Thus, establishing a failure
criterion which allows to predict the ultimate properties of the rubbery parts in structures, is of a
great interest to make easier the dimensioning of such components. Our general purpose in this
work is to build such a criterion.
The first part of this paper deals with some fracture elongations criteria reported in the
literature. Indeed, since the first work of T.L. Smith [1] on the ultimate properties of rubbers,
several authors [2-5] have proposed some fracture criteria based upon strains, stresses or energy.
These are examined using experimental data determined from simple and biaxial tension tests on
unfilled and carbon black filled elastomers. Four materials have been tested under quasi-static and
monotonic loadings, at room temperature and under strain rate of 0.015 s-1.
By examining the obtained results, the limits of these criteria which have been initially
established to fit unfilled elastomers ultimate properties are highlighted.
So, in the second part, an original failure criterion based on an equivalent elongation concept is
introduced. This equivalent elongation IJseems to be linearly dependent on a given biaxiality ratio
ln( O2 b )
n . Thus, from such an evolution, analytical expressions of the principal elongations at
ln( O1b )
break can be written as function of the biaxiality ratio n and IJ evolution parameters. A quite good
agreement is highlighted when comparing the failure experimental data with the analytical
expressions.

FIGURE 1. Evolution of the normalized equivalent elongation as function of the biaxiality factor n
768 A. Hamdi et al.

References
1. Smith T.L., “Dependence of the Ultimate Properties of a GR-S Rubber on Strain Rate and
Temperature”, Journal of polymer science, vol. XXXII, 99-113, 1958.
2. Smith T.L., Rinde J.A., “Ultimate Tensile Properties of Elastomers. V. Rupture in
Constrained Biaxial Tensions”, Journal of polymer science, part. A2, vol.7, 675-685, 1969.
3. Dickie R.A., Smith T.L., “Ultimate Tensile Properties of Elastomers. VI. Strength and
Extensibility of a Styrene-Butadiene Rubber Vulcanizate in Equal Biaxial Tension”, Journal
of polymer science, part. A2, vol. 7, 687-707, 1969.
4. Kawabata S., “Fracture and Mechanical Behavior of Rubber-like Polymers Under Finite
Deformation in Biaxial Stress Field”, Journal Macromol. Sci. -Phys., B8 (3-4), 605-630,
1973.
5. Nevière R., Pfiffer A., Stankiewicz F., “A strain based design criterion for solid propellant
rocket motors” Congress OTAN : Symposium on Combat Survivability of Air, Sea and Land
Vehicules, Aalborg, Danmark, 23-26 September 2002.
 4. Fracture and Fatigue of Elastomers 769

PREDICTION OF RUBBER FATIGUE LIFE UNDER MULTIAXIAL LOADING

A. Zine1, N. Benseddiq1, M. Nait-Abdelaziz1 and N. Ait Hocine2

1Lille University of science and technology, France;
2University of Brest, France

Moussa.nait-abdelaziz@polytech-lille.fr

The process of fatigue failure of rubbers is generally described by two phases : crack initiation and
crack propagation. This study concerns the crack initiation in such materials submitted to a cyclical
loading. Concerning this aspect, either criteria based upon maximum stretch or strain energy
density have been developed in the literature [1, 2, 3]. More recently, a parameter predicting the
onset of primary crack and its probable orientation has been introduced by Mars [4, 5]. This
criterion is based on the so-called “cracking strain energy density (CSED)” Wc and postulate that
crack initition will occur in the plane in which the value of Wc is a maximum. The cracking strain
energy density parameter represents only the portion of strain energy density available to initiate a
& with
crack in a given plane. It is defined incrementally as the dot product of traction vector V
&
the strain increment vector dH on this material plane (figure 1):
& & & &
dW c V .d H ( r T .V ).( d H.r )
(1)
The CSED parameter can be written, assuming finite strains, as:
& & & &
U R T C S dE R U R T (2E  I) S dE R
dW c
& & & &
Uo RTC R Uo R T (2E  I) R (2)
&
U / U o is the ratio of the deformed mass density to the undeformed mass density, R is the unit
vector in the undeformed configuration, defining material plane of interest, S is the 2nd Piola-
Kirchhoff stress tensor, E is the Green-Lagrange strain tensor and C is the Green deformation
tensor.

Figure 1 : Wc/W as function of the plane orientation and of the maximum stretch ratio

In this work, assuming finite strains, this parameter has been analytically determined in the
cases of simple tension, biaxial tension and simple shear loading. These expressions have been
derived by assuming a neo-hookean strain energy density function W. As shown in fig. 1, for the
770 A. Zine et al.

particular case of biaxial tension, the results obtained from such relationships denotes the
capability of this criterion to predict the orientation plane in which the primary crack would be
expected to occur in a material.
The criterion has been then implemented in a finite elements code and results obtained for
classical strain states have been compared to analytical results. As shown in figure 2, a good
agreement has been highlighted. The results have also shown that the ratio Wc/W is not dependent
on the analytical form of the strain energy density function. Fatigue experiments have also been
achieved on 2 kinds of rubbers on PS and uniaxial tension specimen. The results show that the
cracking strain energy density is good predictor of rubber fatigue life.

Figure2: evolution of the CSED as function of the stretch ratio O1

References
1. Cadwell, S.M., Merrill, R.A., Sloman, C.M., Yost, F.L. (1940), Dynamic fatigue life of
rubber, Industrial and Engineering Chemistry, Analytical Edition; 12, pp.19-23.
2. Beatty, J.R. (1964) Fatigue of rubber, Rubber Chemistry and Technology; 37, pp.1341-1364.
3. Roberts, B.J, Benzies, J.B. (1977), The relationship between uniaxial and equibiaxial fatigue
in gum and carbon black filled vulcanizates In Proceedings of Rubbercon ’77, vol. 2.1.. pp.
2.1-2.13.
4. Mars, W. V.( 2002). Cracking Energy Density as a predictor of fatigue life under multiaxial
conditions, Rubber Chemistry and Technology, 75, pp. 1-18.
5. Mars, W. V., A. Fatemi, A. (2001) Criteria for Fatigue Crack Nucleation in Rubber Under
Multiaxial Loading. Constitutive Models for Rubber II, D. Besdo, R. Schuster, J. Ihlemann
(eds.), Swets and Zeitlinger, Netherlands, pp. 213-222.
 4. Fracture and Fatigue of Elastomers 771

MODELING OF BIAXIAL FATIGUE OF NATURAL RUBBER

Shanyan Dong1, Claude Bathias2, Karine Le Gorjo3, F. Hourlier4 and J. F. Vitorri5

1 ITMA, Ph.D. Student, Conservatoire National des Arts et Métiers, 2 rue Conté, 75003, Paris,

France. Tel: 00 33 1 40 27 26 58. dongshanyan@hotmail.com.

2 ITMA, Professor and director of ITMA. Conservatoire National des Arts et Métiers, 2 rue Conté,

75003, Paris, France. Tel: 00 33 1 40 27 23 22. bathias@cnam.fr.

3 Project manager. HUTCHINSON S.A. – Centre de Recherche, rue Gustave Nourry, BP 31,

45120, Châlette-sur-Loing, France.

4 PAULSTRA 26, bld de Péringondas, F-28207, Châteaudun Cedex, France.
5 Technocentre RENAULT, 1 rue du Golf, Guyancourt Cedex 78 288, France.

Numerous rubber components experience multiaxial cyclic stresses in service. From a practical
point of view, there is a great need for predicting the effects of complex loading histories on the
fatigue life of rubber parts. This paper studies the modeling of fatigue natural rubber under biaxial
loading by using a specimen called pancake, in which both uniaxial and multiaxial loading can be
introduced. The specimen is shown in Fig. 1 left. The filled nature rubber is fixed between two
steel plates to form a shape of pancake. A tomography method is used to examine the fatigue
damage inside the specimen, which is helpful to have a good understanding of the fatigue damage
initiation and propagation during the fatigue test. The medical tomographic scanner system of our
laboratory is shown in Fig.1 right. The work of this paper include the following aspects: stress-
strain response analysis, stress-strain behavior with load cycles, the fatigue life and the observation
of failure plane associated with the nucleation of fatigue cracks, and their subsequent growth, the
exploration of the fatigue damage initiation and propagation inside the specimen by tomography
method and the observation of the crack surface by Scanning Electron Microscope.

FIGURE 1. Pancake specimen and tomography scanner system.

First, a FEA simulation is carried out to study the stress-strain response of the specimen under
multiaxial loading. A plane strain condition prevails in the pancake specimen. The biaxial effect on
the stress-strain response is discussed. A comparison between the FEA results and experimental
results is made and a satisfied agreement is obtained.
On the pancake specimen, fatigue tests are conducted over a wide range of load histories,
including pure tension, pure torsion, proportional and non-proportional combined tension-torsion
772 S. Dong et al.

loading. A set of video cameras are set up around the specimen to monitor the nucleation of fatigue
cracks and their subsequent growth. Then the effect of the biaxility on fatigue failure plane and
crack growth rate are analyzed. The ability of existing multiaxial equivalences criteria (maximum
principal strain, strain energy density, octahedral shear strain) to predict fatigue behavior is
explored.
In order to study the fatigue damage inside the specimen in specific stage of fatigue process,
first, we accomplish certain fatigue test without stop to know the whole fatigue process. Then we
repeat the same condition on another specimen. After some specific cycles of test according to the
fatigue stage which we are interested, we take off the specimen from the fatigue test machine, and
then we scan the specimen in the tomographic scanner system to examine the damage inside at this
cycle. After the scanning, we continue the fatigue test until next number of cycle which we want,
and repeat the scanning. We repeat this operation several times until the specimen is cracked. Thus
we can get a history of the fatigue damage inside in different stage of fatigue process. From the
damage histories, we can find that the damage first inside the specimen near the center. With the
propagation of the first damage, other damages appear. Two or more damage nearby can coalesce
to form a bigger damage. There is often a major damage which propagates faster than the others.
The major damage may dominate the fatigue process.
The crack surface is observed using Scanning Electron Microscope. Then the mechanism of
the crack propagation under biaxial loading is discussed.
 4. Fracture and Fatigue of Elastomers 773

MODELING OF CRACK PROPAGATION IN ELASTOMERIC MATERIALS

USING CONFIGURATIONAL FORCES

T. Horst and G. Heinrich

Leibniz Institute of Polymer Research Dresden
Hohe Straße 6, D-01069 Dresden, Germany
horst@ipfdd.de

Service life prediction of elastomeric materials is clearly of high practical and scientific interest
and has attracted the attention of chemists, engineers and physicist. The failure of components is
often associated with the initiation and extension of cracks. Hence, the fracture mechanics
approach is very fruitful in understanding such destructive processes.
Fracture toughness of elastomeric materials is much higher than surface energy due to the
existence of dissipative processes with and without time scale in the bulk. Tailoring the materials
by blending the elastomers and adding fillers can influence this value significantly. These non-
catastrophic dissipative processes were induced mostly near the crack tip as a result of polymer-
polymer, polymer-filler and filler-filler interactions due to blend morphology and filler structure
preventing the breakage of polymer chains and thus the extension of the crack, see Hamed [1].
Characterizing fracture of elastomeric materials tearing energy was introduced as the amount
of energy G required to advance a fracture plane by one unit area. Separating the dissipated energy
Ediss outside of the process zone by thermodynamical consideration of fracture, see Maugin and
Trimarco [2], the total dissipation, that is the total bulk energy lost ) per unit time, can be written
as

) G ˜ v  E diss
(1)
provided that the local dissipative behaviour is described by means of internal variables of
state where G is the configurational force of non-Newtonian nature acting as a driving force on the
crack tip as generalization of the J-integral, and v is the crack tip velocity. Following the
argumentation of irreversible thermodynamics, see e.g. Maugin [3], an equation describing crack
propagation can be derived from a crack dissipation potential Dcrack via

w D crack (G )
v
wG (2)
The observed S-shape curve in the plot G vs. v, see Fig. 1, derived from a nonconvex crack
dissipation potential is capable of explaining steady as well as stick-slip tearing where the crack
advances in a continuous and discontinuous manner respectively, see Maugis [4]. This general
point of view can be modified for the application to elastomeric materials by accounting for
temperature and inertia effects, emphasized by Persson et al. [5] resulting in a scenario similar to
the ductile-brittle transition.
774 T. Horst and G. Heinrich

FIGURE 1. Qualitative relation between the configurational force and the crack tip velocity

Based on these considerations we emphasize the influence of viscoelastic properties on the

crack propagation taking into account structural aspects of elastomeric materials. The application
of these concepts in Finite Element Analysis leads to further understanding of the peculiarities
observed in fracture.

References
1. Hamed, G.R., Rubber Chem. Technol., vol. 67, no. 3, 529-536, 1994
2. Maugin, G.A. and Trimarco, C., In Configurational Mechanics of Materials, edited by
Kienzler, R. and Maugin, G.A., Springer, Wien, 2001
3. Maugin, G.A., The Thermodynamics of Nonlinear Irreversible Behaviors, World Scientific,
Singapore, 1999
4. Maugis, D., J. Mater. Sci., vol. 20, 3041-3073, 1985
5. Persson, B.N.J., Albohr, O., Heinrich, G. and Ueba, H., J. Phys.: Condens. Matter, in press
 4. Fracture and Fatigue of Elastomers 775

DETERMINATION OF INTER-FIBRE-FAILURE IN COMPLEX, REINFORCED

COMPOSITES

V. Trappe and H. Ivers

Federal Institute of Materials Research and Testing (BAM)
Unter den Eichen 87, 12205 Berlin, Germany
volker.trappe@bam.de

Fibre Reinforced Plastics (FRP) produced as complex textile composites are increasingly
employed for weight and cost reasons in transportation systems (aircraft, railway, automotive).
With the rapid development of modern manufacturing methods, there is a need for new measuring
methods and realistic theoretical approaches for design and calculation purposes.
Advanced FRP structures have to endure high mechanical and environmental loading.
Therefore the durability and reliability depends much more on the micro mechanical properties as
on the global strength. For example pure intralaminar micro cracking – without any delaminations
of layers and fibre fracture leads to a distinct strength reduction of carbon fibre reinforced plastics
(CFRP). Micro cracking is non-detectable for conventional X-ray-radiography and ultra sonic
measurements [1].
State of the art: layer-wise strength analysis
Determination of the inter fibre fracture of a single layer is typically employed to estimate the
layer-wise strength of a complex laminate. Thereby, the influence of the neighbouring layers is
ignored. This often leads to large deviations between theory and reality (s. Fig. 1). Additionally,
the problem occurs that for many semi-finished materials with complex textile reinforcement, i.e.
fabrics, nettings or knitting fabrics, no single layer strength can be stated. Thus, the indication of
material parameters (inter-fibre-break-strength), essential for the construction, is not possible and
an understanding of the failure processes in complex layer composites impractical.

Non destructive evaluation, X-ray refraction and effects

X-ray refraction topography [2] is caused by the effect of refraction at the interface of
materials of different refractive indices as well known from visible light passing glass lenses. In
the experimental set-up a collimated X-ray beam passes the sample. At a fixed angle the refracted
signal is measured and additionally a signal proportional to the absorption. A characteristic
refraction value C is determined, which is proportional to the surface per unit volume. The
intensity of the refracted beam will increase if the difference of the refractive index rises at the
observed interfaces. Hence the intensity will be higher for materials with debonded fibres or pores
776 V. Trappe and H. Ivers

than without (s. Fig. 2). By calibration the real as well as the relative inner surfaces are measured.
The relative increase 'C is sufficient in most cases and used in further investigations. Scanning the
whole area of the sample gives a topographic map of inner surfaces (s. Fig. 2).

Solution
A new method for a quantitative determination of transverse and shear strength in a complex
laminate solves the denoted problems by “online-refraction”. Therefore a combination of
mechanical loading and non destructive-testing with the X-ray refraction technique results in
fundamentally new understanding of the micro-mechanical properties of FRP. The objective is to
understand the nonlinear mechanical failure behaviour of this material class by measuring the
physical properties employed in computational models for failure predictions. The experimental
data show, how the nonlinear and linear mechanical characteristics of GFRP laminates with +/-45 -
and/or 0 /90 woven fabrics of linen style correlate with the increase of the "online" refraction
(increase of micro crack density 'C). The layer-wise strength analysis reflects neither
quantitatively nor qualitatively these failure processes.

References
1. TRAPPE, V., HARBICH, K.-W., ERNST, H.: „Damage State of CFRP characterized non-
destructively by X-Ray-Refraction-Topography and Ultrasound“; 48th International SAMPE
Symposium, Long Beach, USA 2003, Proceedings Vol. 48-1 S. 1228-1239
2. HENTSCHEL, M.P., HARBICH, K.-W., LANGE, A.: „Non-destructive evaluation of single
fibre debonding in composites by X-ray refraction“, NDT&E International, Vol. 27, 5 (1994)
p. 275
 4. Fracture and Fatigue of Elastomers 777

THE TEST FREQUENCY DEPENDENCE OF THE FATIGUE BEHAVIOR OF

ELASTOMERS

Z. Major1, Ch. Feichter2, R. Steinberger2 and R. W. Lang1,2

1Institute of Materials Science and Testing of Plastics, University of Leoben
2Polymer Competence Center Leoben GmbH

Leoben, Austria
major@unileoben.ac.at

In engineering applications elastomers are frequently exposed to complex combinations of

mechanical loads (monotonic, static, intermittent and cyclic loads). A better understanding of the
material resistance against crack initiation and propagation becomes of increasing practical
importance. Real elastomer parts and components are loaded over a very wide range of loading
frequency (1 to 105 Hz) [1]. It is also a well known phenomenon that viscoelastic materials may
reveal a hysteretic type heating during cyclic loading [2]. Furthermore, this temperature increase is
influenced by various factors, i.e., loading frequency, amplitude, the viscoelastic loss and the heat
capacity and conductivity of the elastomer type [3 and 4].
Hence, the main objectives of this paper are: (1) to develop and implement adequate test
methods and data reduction schemes to determine the test frequency dependence of the crack
initiation and growth resistance of elastomers under cyclic (fatigue) loading conditions and (2) to
investigate the effect of hysteretic heating in elastomers.
The tests were performed on several elastomer types typically used in engineering
applications. The fatigue tests were run on a high rate servohydraulic test system (MTS 831.59
Polymer Test System). A pure shear specimen configuration with a faint waist in the mid-section
was used in this study. The test frequency was varied between 0.2 and 50 Hz keeping all other test
conditions constant.

FIGURE 1. Frequency dependence of the hysteretic heating during cyclic loading.

The specimen temperature and the spatial and temporal temperature distribution was measured
during the cyclic loading using a full-field thermal analysis system (Altair, Cedip, F). Based on
these results, to simplify the experimental set-up single non-contact temperature sensors were
applied to the near crack tip region (near field) and in the specimen ligament (far field) for
recording the temperature.
A significant test frequency dependence of both the crack growth behavior and the hysteretic
heating was observed. Moreover, the temperature increase and the crack growth was found to
778 Z. Major et al.

strongly depend on the elastomer type. To gain more insight into the dependence of the heating
effect on the viscoelastic parameters, the hysteretic stress-strain curves were also recorded and
analyzed. In addition, dynamic characterization tests were also performed to determine the
amplitude and temperature dependence of the loss modulus values for various elastomer types.

FIGURE 2. Frequency dependence of the tearing energy, T, at crack propagation rate, da/dN, of
10-4 mm/cycles.

References
1. Heinrich, G., et al., 2005, Crack Propagation in Rubber-like Materials, ICF 11, Torino
2. Strobl, G, 1997, Physics of Polymers, Springer, Berlin.
3. Mars, W.V., Fatemi, A. and Cooper, W. V., 2004, Rubber Chemistry and Technology, Vol.
77, 392-408.
4. Kerchman, V and Shaw, Ch., 2003, Rubber Chemistry and Technology, Vol. 76, 386-405.
 5. Integrity of Dynamical Systems 779

Invited Contribution
NONLINEAR MODEL FOR REINFORCED CONCRETE FRAMES LOADED
BY SEISMIC FORCES

D. Kovacevic
Civil Engineering Department, Faculty of Technical Sciences
University of Novi Sad, Serbia & Montenegro
dusan@uns.ns.ac.yu

This paper is a review of one possibility for numerical modeling of in-plane reinforced concrete
frames loaded by seismic actions. Objective of this research is the formulation of one enough
sophisticated and, for engineering practice, reasonably convenient numerical model. Particularly,
the main goal is to adopt the "optimal" model - compromise between complexity and quality of
approximation.
Physical discretization and mathematical approximation is based on finite element method
(FEM) concept. Finite element (FE) stiffness and geometric matrix are defined by LaGrange
updated formulation. Mass matrix is adopted as a lumped mass matrix neglecting the rotational
inertia. Damping matrix is assumed as a proportional to initial stiffness matrix of FE system.
Beam FE cross-section is divided in certain number of finite thickness concrete and steel
layers, which behavior under the cyclic loading is modeled by corresponding uniaxial constitutive
rules, Fig. 1.

FIGURE 1. Reinforced concrete cross-section approximated by finite layers and uniaxial

constitutive rule for concrete

The uniform cracks distribution, i.e. "smeared cracks approach" is accepted in zones where
concrete tension strength is reached. Steel-concrete bond relation is modeled indirectly - by so-
called "tension stiffening effect".
Linear strain distribution is adopted along the height of the cross-section, because of negligible
shear influence in the total deformation field for the typical beam. In this way, the well-known
"shear locking" effect (overestimated the participation of shear deformation in the total
deformation energy) is avoid. In contrast to shear in strict sense, axial and shear stresses interaction
influence, as a cause of inclined cracks appearance must be included in model.
The proposal for inclusion the frame joint deterioration, as well as, interaction of shear and
flexural forces (inclined cracks effects), in this model, is given additionally.
Straight, plane, two-joint beam FE approximates a RC element (beam, column). Two
displacements and one rotation per joint are the degrees of freedom (DOF). Third degree
780 D. Kovacevic

L'Hermite polynomials and a linear functions are adopted as the interpolation functions for
transversal and longitudinal displacements field.
For numerical integration of dynamic equilibrium, the Newmark integration procedure (with
5ms increment) is applied as well as a modified Newton-Raphson iterative procedure for balancing
the residual loads.
As an illustration of previous propositions, presented are the results of the four numerical tests
- one linear and three nonlinear analysis of one simple reinforced concrete frame loaded by three
seismic actions: San Fernando, Parkfield and Imperial Valley earthquakes.
Attained results indicate that proposed numerical concept is one "good compromise" solution.
Compromise is made between the accuracy, as an essential parameter, and, on the other hand,
simplicity, as everyday design practice task. The objective of presented research was to find the
middle way in nonlinear analysis of mentioned structures.
Opposite to standard beam FE models [1], that include only the concrete and reinforcement
behavior modeling, suggested beam FE model considers frame joint deterioration as well as
interaction of shear and flexural forces too, Kovaevi [2].
Final verification and implementation of this modeling concept proves in parametric test
analysis i.e. in comparison with the experimental data and results obtained by application of some
complex model. It is direction of future research on this modeling concept.

References
1. Behavior and Analysis of Reinforced Concrete Structures under Alternate Actions Inducing
Inelastic Response, Vol. 1: General Models, CEB Bulletin d'information No 210, Lausanne,
1991.
2. Kovaevi, D., In Numerical Modeling of Reinforced Concrete Frames Loaded by Seismic
Actions, Ph.D. Thesis, Civil Engineering Faculty, University of Belgrade, 2001.
 5. Integrity of Dynamical Systems 781

Invited Lecture
MONITORING THE DURABILITY PERFORMANCES OF CONCRETE AND
MASONRY STRUCTURES BY ACOUSTIC EMISSION TECHNIQUE

A. Carpinteri and G. Lacidogna

Department of Structural Engineering and Geotechnics
Politecnico di Torino
Corso Duca degli Abruzzi 24, 10129 Torino, Italy
alberto.carpinteri@polito.it
giuseppe.lacidogna@polito.it

The contribution of non-destructive and instrumental investigation methods is currently exploited

to measure and check the evolution of some negative structural phenomena, such as damage and
cracking, and to predict their subsequent developments. The choice of the technique for controlling
and monitoring reinforced concrete and masonry structures is strictly correlated to the kind of the
structure to be analysed and on the data to be extracted, Carpinteri and Bocca [1].
Among these methods, the non-destructive methodology based on Acoustic Emission (AE)
proves to be very effective, Carpinteri et al. [2]. This technique makes it possible to estimate the
amount of energy released during the fracture process and to obtain information on the durability
performances of the monitored structures, Carpinteri and Lacidogna [3,4].
The AE monitoring equipment adopted by the authors consists of PZT transducers fitted with
a preamplifier and calibrated on inclusive frequencies of between 100 and 400kHz. The
transducers are connected to switchboards equipped with an amplifier and a pass-band filter, one
threshold level measuring system, a recorder and an oscillation counter. The threshold level of the
signal recorded by the system, fixed at 100PV, is amplified to up to 100mV. The amplification
gain, determined as the ratio between the voltage at output Eu and that at input Ei according to the
formula dB = 20 log10 Eu/Ei, turns out to be 60 dB. This signal amplification value is the one
commonly adopted in the assessment of AE events in concrete. Oscillation counting capacity is
assigned a limit of 255 every 120 seconds of signal recording. In this manner, a single event is the
result of 2 recorded minutes [4].
From the literature (Ohtsu [5], Shah and Zongjin [6]) we find that the amplitude of the direct
non-amplified signal is of the order of 100PV, neglecting the attenuation phenomena due to the
distance of signal generation; hence, it can be assumed that the measurement system has the ability
to detect the most significant AE events of material cracking. By means of this system, the
intensity of a single event is by definition proportional to the number of counts N recorded in the
time interval (Event Counting). Clearly, this hypothesis is fully justified only in the case of slow
crack growth, Holroyd [7]. With this system, in addition, a localization procedure based on time
delays measured by few (e.g., 2-6) spatially distributed AE sensors is applicable to localize the
main sources of the damage process.
By means of the AE technique, an extensive experimental analysis on masonry and reinforced
concrete structures under service loads has been carried out [2,3]. In some cases the main cracks
sources are localized in the volume or on the surface of the structure. For surface sources, direct
observations confirm the validity of the localization procedure.
Strictly connected to the energy detected by AE is that dissipated by the monitored structure.
The energy dissipated during crack formation in structures made of quasi-brittle materials plays a
fundamental role in the behaviour of a structure throughout its life. Strong size effects are clearly
782 A. Carpinteri and G. Lacidogna

observed on energy density dissipated during fragmentation. Recently, a multiscale energy

dissipation process has been shown to take place in fragmentation, from a theoretical and fractal
viewpoint, Carpinteri et al. [8]. Based on Griffith’s assumption, of local energy dissipation being
proportional to the newly created crack surface area, the fractal theory shows that the energy will
be globally dissipated in a fractal domain comprised between an Euclidean surface and volume.
According to the fractal concepts, an ad hoc theory is herein employed to detect and monitor
concrete and masonry structures performances by means of the AE technique. The fractal theory
takes into account the multiscale character of energy dissipation and its strong size effects. This
makes it possible to introduce a useful energetic damage parameter for structural assessment based
on a correlation between AE activity in a structure and the corresponding activity recorded on a
small specimen obtained from the structure and tested to failure [2]. Moreover, by applying Fractal
and AE criteria, the safety of structures undergoing damage and degradation processes can be
efficiently evaluated.

References
1. Carpinteri, A., Bocca, P., Damage and Diagnosis of Materials and Structures, Pitagora Ed.,
Bologna, Italy, 1991.
2. Carpinteri, A., Lacidogna, G. and Pugno, N., In Proceedings of 5th International Conference
on Fracture Mechanics of Concrete and Concrete Structures (FraMCos-5), edited by V. C.
Li, C. K. Y. Leung, K. J. Willam, S. L. Billington, 2004, 31-40.
3. Carpinteri, A., Lacidogna, G., Journal Facta Universitatis, Series: Mechanics, Automatic
Control and Robotics, vol. 19, 755- 764, 2003.
4. Carpinteri, A., Lacidogna, G., System for the assessment of safety conditions in reinforced
concrete and masonry structures, Italian Patent N. To 2002 A000924, deposited on 23
October 2002.
5. Ohtsu, M., Magazine of Concrete Research, vol. 48, 321-330, 1996.
6. Shah, P., Zongjin, L., ACI Materials Journal, vol. 91, 372-381, 1994.
7. Holroyd, T., The Acoustic Emission & Ultrasonic Monitoring Handbook, Coxmoor
Publishing Company’s, Oxford, 2000.
8. Carpinteri A, Lacidogna G, Pugno N., International Journal of Fracture, vol. 129, 131-139,
2004.
 5. Integrity of Dynamical Systems 783

Invited Lecture
BIFURCATION CONTROL OF PARAMETRIC RESONANCE IN AXIALLY
EXCITED CANTILEVER BEAM

H. Yabuno and M. Hasegawa

Graduate School of Systems and Information Engineering
University of Tsukuba
Ten-no dai, Tsukuba 305-8573 JAPAN
yabuno@esys.tsukuba.ac.jp

Bifurcation control of parametric resonance in an axially excited cantilever beam is theoretically

and experimentally investigated. In the cases when the excitation frequency is in the neighborhood
of twice the natural frequencies, the parametric resonance is produced through the trans-critical
bifurcation and the cantilever overcomes also saddle-node bifurcation at the finite amplitude
which is a discontinuous bifurcation [1] [2]. In state of the passage through resonance, jumping
phenomena occurs due to the discontinuity of the bifurcation. Our control objectives is to avoid
the occurance of the jaump phenomena in the state of the passage throught resonance. We have
proposed a bifurcaion control method for parametric resonance in a single degree of freedom
system. In this research, the method is expanded to the paremetric resonance produced in the
acontinuous system, i.e., cantilever beam. Then, a control method for the avoidance of the
paremetric resonance is proposed. Also, the validity of the mehod is experimentally confirmed.
First, we derive the equation of motion of the cantilever beam by taking into account the effect
of the curvature nonlinearrity [3]. The governing equation is theoretically analyzed by using the
method of multiple scales [4]. The obtined averaged equation is autonomus for which the
bifurcaion analyisis is easily performed. The above mentioned discontinuous bifurcations (saddle-
node bifurcation and subcritical pitchfork bifurcation) are investigated with respect to the
excitation frequency and their bifurcation points are detected.

FIGURE 1 Analytical model

Next, we consider a control method for the avoidance by using a piezo actuator. We derive the
equation of motion under the effect of the bending moment by the actuator. It is assumed that the
bending moment is proportional to the input voltage. The averaged equation under the control
input is derived again by the method of multiple scales. We set the control input to be
proportional to the velocity of the beam and the cubic nonlinear term with respect to deflection.
The design of the feedback gains are designed based on the averaging equation.
Furthermore, we use an experimental set-up as in Fig. 2. We confirm the validity of the
theoretically proposed control system. The displacement of the beam at a point is measured. By
784 H. Yabuno and M. Hasegawa

this data is feedback to actuate the piezo actuator. Under the control, we observe the passage
through resonance, and compare with the case of no control. As a result, it is confirmed that the
proposed control method is valuable for the avoidance of the jumping phenomena in the case of
passage through resonance.

FIGURE 2 Experimental setup

References
1. Yabuno, H., Nonlinear Dynamics, Vol. 12, 263-274,1997.
2. Yabuno, H., Murakami, T., Kawazoe, J., and Aoshima, N., ASME J. Vibration Acoustics,
Vol 126, 149-162, 2004.
3. Anderson, T. J., Nayfeh, A. H., and Balachandran, B., ASME J. Vibration Acoustics, Vol
118, 21-27, 1996.
4. Nayfeh, A. H. and Mook, D. T. , Nonlinear Oscillations Wiley New York 1979.
 5. Integrity of Dynamical Systems 785

Invited Lecture
ADAPTIVE PROPERTIES OF DYNAMIC OBJECTS

I. I. Blekhman and L. A. Vaisberg

“Mekhanobr–tekhnika” Corp., Institute for Problems in Mechanical Engineering of RAS
Bolshoy pr. V.O., 61, St. Petersburg, 199178 Russia
blekhman@vibro.ipme.ru

By adaptability we understand an ability of objects to retain their individuality and main

functional properties, accommodating to the change in outer conditions. Then in case of technical
objects we mean that their main qualitative and quantitative parameters of functioning remain
within the limits, acceptable for the consumers.
In most cases adaptability contains such fundamental notions as stability and steadiness.
Speaking roughly, we will say that the stationary condition (particularly, the equilibrium
position) or motion is stable if in time it does not change much under the action of certain
excitations.
We will say that the stationary states or regimes of motion of an object are steady if certain
(significant for all practical purposes) parameters of those states or motions remain within a certain
domain A when the parameters of the system and the parameters of outer excitation remain within
a certain domain B.
The notion of adaptability differs essentially from that of self-organization by which is usually
meant the emergence of new properties in a system, the latter becoming the integration of separate
objects interacting with each other. Self-organization does not mean that the object is intended to
fulfill a certain practical aim. Though, there are cases when the adaptability of the system is
provided by its ability to be self-organized. Self-synchronization of dynamic objects (briefly
discussed in the report) is a bright example of it. It should be noted that for some strange reason the
phenomenon of self-organization eludes the attention of the adherents of synergetics, while this
phenomenon is one of the expressive examples of self-organization.
The more complicated and highly organized the object, the more adaptive, as a rule, it is. The
adaptability of living organisms is well known. It can be explained by a natural selection: only
those organisms which were able to adapt to the changes in the outer conditions have survived in
the process of evolution.
It is surprising, however, that some technical systems also possess a certain adaptability. This
presentation is devoted to the description of some examples of the adaptability of mechanical
objects that the authors came across in their scientific and engineering practice. Perhaps specialists
in other branches can give other, no less dramatic examples.
We consider here the following examples of adaptation in technical systems:
1 Self-synchronization of dynamic objects, in particular, vibro-exciters, i.e. unbalanced
rotors, actuated by the asynchronous engines. This phenomenon consists in the fact that
two or more objects, generating oscillation or rotation with different frequencies, when
united into a single system, generate oscillations with equal or multiple frequencies.
2 Adaptive properties of the devices with self-synchronized vibro-exciters – an auto-balance
beam - and vibrational machines. In the former devices the inertial elements of a beam
automatically compensate the changing unbalance of the rotor, while in the latter the
786 I. I. Blekhman and L. A. Vaisberg

rotors of the exciters automatically record the changing position of the center of gravity of
the solid body they are installed on.
3 A pendulum with a vibrating axis of suspension (the Stephenson-Kapitsa pendulum) is
automatically located in the direction of the most intensive oscillations of the axis of
suspension.
4 The particle which is being deformed in the field of a standing wave tends to approach
either the nodes or the loops of the wave, depending on its elastic and inertial properties,
and also on the oscillation frequency.
5 The profiles of a working chamber of crushers are worn out in the process of work in such
a way that the wasted energy, the rate of wear and the strain of the critical parts are
minimized. It is remarkable that after a certain period of work, the so called working
period, the profiles acquire a configuration, obtained as optimal, as a result of a rather
intricate theoretical investigation.
6 A chain of generators of a certain type with a sequentially reduced oscillation frequency
possesses the following adaptive property. If the links between first generator, which
possesses the highest frequency of the generated oscillations and the next one, function
adequately, then the first generator will dampen the vibrations of the other generators. In
case the first generator stops working or its link with the next generator is broken, the latter
“wakes up” and dampens by its oscillations all the other generators, located further and so
on. This effect underlies the adaptation properties of the system of exciting the rhythm of
heart contractions. It should be noted that the adaptation properties of some machines are
not accidental, but are the result of the purposeful work of investigators, designers and
inventors, though based only on intuition. In fact, any well designed machine possesses, in
one way or another, adaptive properties.
It is natural that the adaptability of a technical device can be provided by the use of the means
of control and of mechatronics. These approaches, very expensive as a rule, can be characterized as
a struggle against nature, while in this presentation we consider it to be the use of its gifts.
One of the effective means of providing the adaptive properties of the machines is the transfer
from the cinematic bonds in their mechanism to the dynamic ones, i.e. to the enrichment of the
additional degrees of freedom. There are two glowing examples of it.
In the closing part of the presentation some general principles are considered which underlie
the effects of adaptation in the mechanical systems. In particular, there are suppositions that in a
number of cases adaptability is a consequence of the extreme properties of the appropriate
processes.
Along with that the authors realize that these interpretations just give the answers to the
question “how” but not to the questions “why”, which is characteristic of modern physics. The
brilliant Euler and his contemporaries, as well as many outstanding scientists nowadays believe
that the presence of the extreme principle is the design of the Creator. The researchers who are of
materialistic views argue that these principles are the produce of the human brains, because
practically all the physical regularities can (if one tries) be interpreted mathematically, as a
realization of certain extreme statements. The authors of this presentation would rather share the
first point of view.
 5. Integrity of Dynamical Systems 787

Invited Contribution
INFLUENCE OF ADDENDUM MODIFICATION COEFFICIENT ON THE
GEAR'S LOAD CAPACITY

I. Atanasovska and V. Nikoli-Stanojevi

Faculty of anagement in Industry, JNA 63, 37000 Krusevac, Serbia,
Faculty of Mechanical Engng, Univ. of Kragujevac, Sestre Janji 6, 34000 Kragujevac, Serbia
ivanacvejic@beotel.yu

There are many factors that influence at the mechanical behavior of gears. The gear’s mechanical
phenomenons are greatly conditioned by teeth’s profile. One of the values that is most important
and influential is the addendum modification coefficient value (x1, x2).
This paper gives the description of the procedure developed for researching the influence of
addendum modification coefficient value on the load capacity of cylindrical involute gears with
straight teeth. The numerical method – Finite Element Method (FEM) – is used for the
determination of stress state for meshed teeth’s flanks and roots. The model of meshed gears’ teeth
contact in FEM is made to enable the simultaneously monitoring teeth’s flanks stress state and
teeth’s roots stress state.
It’s important to notice that the accuracy of FEM calculations is greatly conditioned by
accuracy of model’s geometry and also by the choice of the boundary conditions that simulate the
real working conditions in the best way. According to this, the FEM models each consist of three
gear’s teeth are developed and described in this paper. The special program for tooth’s profile
drawing is developed and built in present FEM software. This program makes possible the
developing of the FEM models with ideal gears’ geometry both for period with two tooth pairs in
contact as well as for single tooth pair meshing period.
During the developing of the FEM model for meshed gear pair the special attention is paid to
the simulation of contact conditions on meshed teeth’s flanks. The “point-to-surface” contact finite
elements are chosen for tooth contact simulation. Very high level of coincidence between the
behavior of contact zone obtained by the developed FEM procedure and the behavior of contact
zone at real gear pairs. The particular characteristics of the “point-to-surface” contact finite
elements are used during the FEM calculations.
During the developing of gears’ meshing simulation this paper pays special attention to the
developing of the procedure for load distribution calculation and for obtaining its influence on the
load capacity of meshed gear pair. This is in accordance with the fact that the teeth stiffness is the
most influential value for load distribution among simultaneously meshing tooth pairs. Also, this is
in accordance with the fact that the load distribution is the most influential mechanical
phenomenon for the deformation and stress state of gear pairs as well as for theirs load capacity.
The stiffness of tooth pair directly depend on the values of the loads trensferred by the tooth pairs
that is simultaneously in contact. As a consequence, for period with two tooth pairs in contact the
developed procedure used iterations for obtaining the values of stiffness and load for each tooth
pair in contact. The FEM models made for a single real gear pair are used for analysis and
verification of developed procedure. Obtained results are compared with corresponds many
different aspects. The very high coincidence is obtained.
In order to compare the stress states of meshed teeth's flanks and roots during the meshing
period for gear pairs with different values of addendum modification coefficients the comparative
diagrams are made and shown in this paper. All values useful for the researching of mechanical
phenomenons at gear pairs (tooth stiffness c', tooth deformation u', load distribution q, teeth’s
788 I. Atanasovska and V. Nikoli-Stanojevi

flanks stress state VH, teeth’s roots stress state VF1, VF2) are monitored through the comparative
diagrams. Many conclusions about the influence of addendum modification coefficient at tooth
defformation and gear load capacity are obtained from that diagrams. These diagrams are used for
making the special procedure and diagram sets which enable one to make the right choice for the
values of the addendum modification coefficients for a particular gear pair in accordance with the
general goal – increasing the gear's load capacity. The exponential functions which describe how
the maximum teeth’s flanks stress values and the maximum teeth’s root stress values depend on
the values of addendum modification coefficients for a particular gear pair are defined using
numerical methods.
Many conclusions about the influence of addendum modification coefficient at tooth
defformation and gear load capacity are obtained from that diagrams. These diagrams are used for
making the special procedure and diagram sets which enable one to make the right choice for the
values of the addendum modification coefficients for a particular gear pair in accordance with the
general goal – increasing the gear's load capacity. The exponential functions which describe how
the maximum teeth’s flanks stress values and the maximum teeth’s root stress values depend on
the values of addendum modification coefficients for a particular gear pair are defined using
numerical methods.
The developed procedure for choosing the optimal values of addendum modification
coefficients described in this paper gives possibility for easy recognizing the influence of these
values on the all gear’s mechanical phenomenons. It is so especially for the influence of described
values on the load distribution which is extremely important factor for gear’s load resistance, as
well as for reducing vibrations, shocks and noise during the gear pair working
 5. Integrity of Dynamical Systems 789

Invited Contribution
MICROMECHANICAL MODELLING OF FRACTURE-INDUCED ANISOTROPY
AND DAMAGE IN ORTHOTROPIC MATERIALS

Vincent Monchiet, Ion-Cosmin Gruescu, Djimedo Kondo and Oana Cazacu1

Laboratoire de Mecanique de Lille, UMR CNRS 8107, Universite Lille 1, France
Vincent.Monchiet@ed.univ-lille1.fr, ic.gruescu@ed.univ-lille1.fr, kondo@univ-lille1.fr
1GERC, University of Florida, Shalimar, USA

cazacu@gerc.eng.ufl.edu

Matrix cracking is commonly recognised as one of the main inelastic deformation mechanisms of
Brittle Matrix Composites. The modelling of such phenomenon still presents some difficulties
which are mainly related to the description of the interaction between the initial anisotropy and the
cracks-induced anisotropy. The present study concerns a new micro-macro approach of the non
linear behavior and damage propagation in this class of materials.
First, we present an analytical method for the determination of Eshelby tensor S (or
equivalently the Hill tensor P) associated to an arbitrarily oriented crack embedded in an
orthotropic elastic medium (see the representative elementary volume on Figure 1). The crack is
modeled as an infinite elliptic cylinder along a symmetry axis of the solid matrix, and with a low
aspect ratio. The new analytical results provided by the present study show the strong interaction
between the primary anisotropy and the crack-induced anisotropy. Several validations of the new
results are presented through their agreement with some existing results in literature for particular
configurations of the cracks system. The incorporation of these analytical results in a Mori-Tanaka
homogenization scheme allows to determine the macroscopic elastic properties of the orthotropic
matrix containing a general distribution of cracks.
A last part of the study is devoted to a modelling of the damage process in initially orthotropic
materials. For this purpose, we propose a physically-motivated damage criterion based on an
energy release concept and accounting for the interaction between the primary anisotropy and the
induced damage. Identification of the model requires 5 parameters with clear physical meaning.
For illustration purpose, these parameters are calibrated on experimental data provided by Gasser
et al. on a Ceramics Matrix Composites (See Figure 2). The predictive capabilities of the
anisotropic damage model and comparisons for off-axes loading experimental data on the same
material provide a large validation of the proposed approach. Evolution of the overall stiffness
tensor components will be also presented at the conference.

FIGURE 1 : Representative Elementary Volume : anisotropic solid matrix weakened by cracks

790 V. Monchiet et al.

FIGURE 2 : Tensile loading and calibration curve on uniaxial axis loading ( M 0q) )

FIGURE 3 : Model predictions and validation on off- axis loading test ( M 20q) ) and M 45q) )

Reference
1. Gasser, A. "Sur la modélisation du comportement mécanique des composites céramique-
céramique à temperature ambiante", Ph.D. Thesis, ENS Cachan, 1994. In French
 5. Integrity of Dynamical Systems 791

Invited Lecture
VIBRATION CONTROL DEVICES AND THEIR APPLICATION

K. Nagaya
Department of Mechanical Engineering, Gunma University
Kiryu, Gunma 376-8515, JAPAN
nagaya@me.gunma-u.ac.jp

The vibration control is of importance for preventing fractures of machines and structures.
Recently various control methods have been developed. However, vibration control devices have
not been discussed thoroughly. The present report introduces various vibration control devices
developed in our laboratory[1]-[13].
1 Vibration isolation systems using electromagnetic actuator
It is difficult, in general, to suppress the transmissibility from the base to the vibrating body
to be less than one by using usual vibration control methods such as PID, optimal regulator, etc. In
order to isolate the base vibration, we presented a permanent-electromaget actuator [1],[2] and the
method of control, in which vibration disturbances are cancelled directly, and velocity feedback is
performed to suppress resonance peaks. When the actuator and control method are used, the
vibration disturbance is perfectly isolated, and the transmissibility becomes almost zero.
2 Vibration suppression for structures using magnetic dampers
A vibration absorber is available to control vibrations in wide frequency range. In order to
suppress vibrations we present a magnetic damper. The sound noise due to plate vibrations is
suppressed, and when an appropriate arrangement is made for the absorber, high frequency
vibrations can also be suppressed [3].
In order to suppress vibrations of large flexible structures such as aircraft wings, a dam per will
be applicable. Since eigenfrequencies of an aircraft wing are small, it requires the following for the
damper(1)The damping force has to be large for low frequency because the vibration frequency of
the wing is small, (2)it works for small displacement, (3) its size has to be small which can be
stacked in the wing, and (4) passive damper is desirable because the maintenance has to be free in
the damper. This article presents a new type linear damper, which satisfies the conditions as just
mentioned. The damper consists of a number of thin plates with slits. MR fluid is filled in the
damper, so that magnetic fields freeze MR fluid. The resisting torque is generated when the plates
slides due to the shear of slits of the plates. The damper works against the bending moment of the
wing. Theoretical and experimental results are obtained[5].
3 Tunable absorber for suppressing vibrations
Since, vibration amplitude for machines and structures is zero at anti-resonance frequency. In
order to create the anti-resonance, we provided a tunable absorber. Auto tuning control method is
developed[4]. The absorber is utilized to a milling machine.
4 Silencer consisting two-stage Helmholtz resonator
A Helmholtz resonator has advantages over other noise control methods, in that it does not
require energy to function and it can be applied to high-frequency noise. In this article, a noise
reduction method based on Helmholtz resonator is discussed in reference to application to a
blower. However, as the frequency of noise generated by a blower, although high, varies over time,
the Helmholtz resonator is ineffectual if applied directly. In order to reduce varying high-
frequency noise, a new type of silencer, comprised of a two-stage rotary auto-tuning resonator, is
792 K. Nagaya

proposed[8]. A fast Fourier transform (FFT) analysis of noise obtained by a noise meter is used for
control.
5 Counteracting moment devices for controlling vibrations of buildings
The present article provides a method of vibration control for a beam carrying a mass at its tip
subjected to earthquakes. A vibration isolation mechanism consisting of a gear train for the beam is
presented[6]. Theoretical analysis for the beam is developed, and to validate the method and the
analysis, experimental tests are carried out for a model of the present mechanism. The device is
also applied to multi-level buildings[7]. It is clarified that the vibration displacements and the
moments in the beam are suppressed significantly.
6 High-TC superconducting levitation systems and its vibration control
The modeling and analysis of a high TC suoerconducting levitation system are discussed[9]-
[11]. First, the analytical expressions for obtaining the non-linear levitation force given by the
present author are clarified; and then a vibration control method is presented. In which feedback
currents involving frequency weights are used. Optimal coefficients of the transfer function of the
controller are obtained by minimizing the cost function. Hence, the spillover instability due to the
non-linear vibrations will be reduced. Numerical calculations have been carried out for some
typical problems. To validate the present method, experimental tests have been carried out.
The superconducting levitation system is also applied to pulse motors[12]-[13]. The levitation
system is somewhat new, because in our system, the superconductor supports at its one end, and
the permanent magnet supports the other end of the rotor shaft. This means that, the system
requires only one superconductor, and stable levitation is possible without levitation control. The
vibration control method is also presented. To validate the system, experimental tests have been
carried out.
References
1. K.Nagaya and H.Kanai, Journal of Sound and Vibration, 180-4 (1995), pp.645-655.
2. K.Nagaya and M.Ishikawa, IEEE Trans. on Magnetics, 31-1(1995), pp.885-896.
3. K.Nagaya and L.LI, J. Acoust. So. America, 104-3((1998), pp.1466-1473.
4. K.Nagaya A.Kurusu, S.Ikai and Y.Shitani, Journal of Sound and Vibration 228-4 (1999),
pp.773-792.
5. T. Pranoto, K.Nagaya and A.Hosoda, Journal of Sound and Vibration, 276 (2004), pp.919-
932.
6. K.Nagaya and T.Fukushima, Intelligent beam transforming earthquake force into vibration
control force and its optimal design”, Journal of Sound and Vibration, 218-3(1998), pp.445-
461.
7. K.Nagaya, T.Fukushima and Y.Kosugi, J. Acoust.Soc.America, 105-5(1999), pp.2695-2703.
8. K.Nagaya, Y.Hano and A.Suda, J. Acoust. Soc. America, 110-1(2001), pp.289-295.
9. K.Nagaya, IEEE Trans. on Magnetics, 32-2(1996), pp.445-451.
10. K.Nagaya and S. Shuto, IEEE Trans. on Magnetics, 32-3(1996), pp.1888-1996.
11. K.Nagaya, M.Tsukagoshi and Y.Kosugi, J.Sound Vib., 208-2(1997), pp.299-311.
12. K.Nagaya, Y.Kosugi, T.Suzuki and I.Murakami, IEEE Trans. on Appl. Superconductivity, 9-
4(1999), pp.4688-4694.
13. K.Nagaya, T.Suzuki, N.Takahashi, and H.Kobayashi, IEEE Trans Appl. Superconductivity,
11-4 (2001), pp.4109-4115.
 5. Integrity of Dynamical Systems 793

Invited Lecture
MEASUREMENTS OF DYNAMICAL SYSTEM INTEGRITY AND FRACTURE
MECHANICS

K. S. Hedrih
Faculty of Mechanical Engineering University of Nis, Institute of Mathematics SANU Belgarede
Yu-18000 Niš, ul. Vojvode Tankosia 2/22, Serbia and Montenegro
katica@masfak.ni.ac.yu, khedrih@eunet.yu

I. What is the Dynamical System ? Problems in dynamics have been fascinating physical scientists (and mankind
in general) for thousands years. Modern dynamical system theory has relatively short history.
It began with Poincaré (1880, 1890, 1899), who revolutionized the study of nonlinear differential equations by
introducing the qualitative techniques of geometry and topology rather than strict analytic methods to discus the global
properties of solutions of these systems.
Robert Devaney in his book [1] wrote: “The simple examples of dynamical systems show how dynamical systems
occur in the “real world” and how some very simple phenomena from nature yield rather complicated dynamical
systems”.
The answer is quite simple: take a scientific calculator and input any number whatsoever. Than, start striking one
of the function keys over and over again. This iterative procedure is an example of a discrete dynamical system:
x 0 , f x 0 , f  f x 0 , f  f  f x 0 ,......, f  f .... f  f x 0  .

The basic goal of the theory of Dynamical Systems is to understand the eventual or asymptotic behavior of an
iterative process. If the process is a discrete process such as iteration of a function, then the theory hopes to understand
the eventual behavior of the points x , f x , f 2 x , f 3 x ,..., f k x ,..., f n x  . Functions which determine
dynamical systems are also called mappings, or maps, for short. Mathematical phenomenology and analogy [2] is very
useful in the investigation of kinetic properties and processes in the dynamical systems with disparate nature.
A very important notation in the study of dynamical systems is the behavior or persistence of the system under
small changes or perturbations. This is the concept of structural stability.
The notion of structural stability is extremely important for practical application.
II. Characterization of a dynamical system:
Material model with material structures (construction with structural elements);
Mathematical description of material structure dynamics (geometry and material of construction, constitutive
stress-strain relation depending of material properties [3], time rate changes in the system structures, material
coefficients, coefficients of inertia tensor, coefficients of stiffness……and other material properties; coupled tensor
fields properties of material…);
Stress and strain states of dynamical system structure on the boundary surfaces (defined by boundary conditions
for free and for loaded system, external surface contour excitations, contour displacements);
Stress and strain states [4] of dynamical system structure at initial time moment or with all history before initial
moment on the observation in all material structures and on the boundary surfaces (by initial condition satisfying the
boundary conditions for free and for loaded system);
States of dynamical processes in dynamical system;
State of stress and strain in the material, and also state of displacements of structure points and also states of other
coupled fields depending on the type of material structure.
III. Types of dynamical system integrity [5,6]
Ideal integrity of dynamical system is when properties of the material structure system and its own kinetic
parameters do not change under external excitations;
Partial integrity of dynamical system is when properties of the material structure system or its own kinetic
parameters do not change under external excitations.
Modified integrity of dynamical system is when properties of the material structure system or its own kinetic
parameters change under external excitations in the proposed intervals.
794 K. S. Hedrih

Rheonomic integrity of dynamical system is when properties of the material structure system and its own kinetic
parameters are functions of time and do not change under the external excitations.
Integrity of subprocesses in dynamical systems (In the linear oscillatory system own modes are mutual
independent processes and single frequency processes with one own frequency each).
Integrity of the real structure material in dynamical system (Measurements of integrity of the real structure
material are defined by development of material sciences and the fracture mechanics theory [7] and [8]. Papers written
by the British scientist A. A. Griffith (1920, 1924) are of permanent importance for the early formative period of
fracture mechanics. He was the first to consider the energy balance approach to the crack problem, the important
aspects of which have been reviewed by Panasyuk (1993) [9].
IV. Chaotic Clock Models: A paradigm for vibrations and noise in machines and integrity of machines. For
examining natural clocks [10] of reductors (power transmission), as well as sources of nonlinear vibrations and noise
in its dynamics, it is necessary to investigate properties of nonlinear dynamics, and phase portraits, as well as
structures of homoclinic orbits, layering and sensitivity of this layering of homoclinic orbits and bifurcation of
homoclinic points. The natural clocks of nonlinear dynamics of coupled rotors are studied, as well as integrity of
machines with respect to the fatigue of the material structure of power transitions [11].
Key words: Homoclinic orbits and points, separatrice layering, trigger, coupled singularities, bifurcation, vector
method, mass moment vectors, phase plane and portrait, couple triggers, the form of number eight or it multiplicand,
Chaotic Clock Models, integrity, machines.
V. Single and multifrequency vibration regimes in the sandwich system with discontinuity. By using two
examples of free and forced vibrations of the elastically connected multi body systems and corresponding system with
a discontinuity in an elastic connection, we show some basic properties and measurements of the integrity of basic
dynamical system. The integrity of dynamical subprocesses in the behavior of the whole system and its subsystems or
in component processes and as response of the whole system to corresponding system with discontinuity have been
studied by using methods of Bernoulli’s particular integral and Lagrange’s method of constants variation. It is shown
that one- and two frequency subprocess regimes change into multi-frequency regimes induced by discontinuity in the
system, which represents the loss of integrity of the system structure and marks the appearance of the loss of integrity
of basic subprocesses [6]. The presence of multifrequency regimes in one of the modes of vibrations is an indicator of
discontinuity in dynamical system and of loss of previous integrity.
VI. Concluding remarks. For the measures of dynamical system integrity it is possible to take one of the sets of
kinetic parameters, the changes of which in the critical values range are sources of bifurcation processes or of the
appearance of nonprogrammed processes in dynamical system with possible appearance of structure discontinuity.
Acknowledgment
Parts of this research were supported by the Ministry of Sciences, Technologies and Development of Republic
Serbia trough Mathematical Institute SANU Belgrade Grants No. 1616 Real Problems on Mechanics and Faculty of
Mechanical Engineering University of Niš Grant No. 1828 Dynamics and Control of Active Structures.
References
1. Devaney R., (1989), An Introduction to Chaotic Dynamical Systems, Addison-Wesley Publishing Company, Inc.
p. 336.
2. Petrovi Mihailo, Fenomenološko preslikavanje, Srpska kraljevska akademija, Štamparija Planeta, Beograd, 1933.
3. Goroško O.A. and Hedrih (Stevanovi), K., Analitika dinamika (mehanika) diskretnih naslednih sistema, (Analytical
Dynamics (Mechanics) of Discrete Hereditary Systems), University of Niš, 2001, Monograph, p. 426. (in Serbian).

4. Raškovi, D., (1985), Teorija elastinosti (Theory of Elasticity), Nauna knjiga, 1985, 414.
5. Rega G. and. Lenci S., (2004), Identifying, Evaluating, and Controlling Dynamical Integrity Measures in Nonlinear
Mechanical Oscillators, Booklet of Abstracts, WCNA Orlando 2004, pp.13-14.
6. Hedrih (Stevanovi), K., Integrity of dynamical systems, Booklet of Abstracts, WCNA Orlando 2004, pp. 25-32.
7. Gdoutos E. E., Fracture Mechanics, An Introduction, Kluwer Academic Publishers, Dordrecht, 1993.
8. Hedrih (Stevanovi) K., Jovanovi D. B., (2003), Mehanika loma i ošteenja (Damage and fracture mechanics) –
matematika teorija – renik pojmova i prilozi, Mašinski fakultet u Nišu, str. 208.
9. Ukrainian Siciety on Fracture Mechanics 1992-2002, Spolom, Lviv 2002, p. 273.
10. Moon F,C.: Chaotic Clock Models: A Paradigm for Noise in Machines, Booklet of Abstaracts, IUTAM
Symposium on Chaotic Dynamics and Control of Systems and Processes in Mechanics, Università di Roma “La
Sapienza”, Roma, Italy, 2003, pp. 41-44.
11. Hedrih (Stevanovi) K., (2004), Homoclinic orbits layering in the coupled rotor nonlinear dynamics and chaotic
clock models: A paradigm for vibrations and noise in machines, Proceedings ICTAM04 Abstracts Book and CD-
ROM Proceedings, pp. 320, SM17S-10624.
 5. Integrity of Dynamical Systems 795

Invited Contribution
MODELING OF THE SURFACE CRACKS AND FATIGUE LIFE ESTIMATION

Katarina Maksimovic, Stevan Maksimovic and Vera Nikolic-Stanojevic1

VTI- Aeronautical Institute, Katanieva 15, 11000 Belgrade, Serbia,
1Faculty of Mechanical Engng, Univ. of Kragujevac, Sestre Janji 6, 34000 Kragujevac, Serbia.

tince@net.yu

Part-through cracks such as corner or surface cracks are one of the most common cracks in
structural components. The paper focuses to develop analytic expresses for the stress intensity
factor (SIF) for the surface crack in 3-D solid type structural components and crack growth. For
this purpose three-dimensional finite-element analyses were used to develop an analytic equation
for the stress-intensity factors. Traditionally, damages in structural components are assumed to
have an elliptic shape that are loaded with cyclic loads and load spectra. Semi-elliptic surface
cracks frequently initiate and grow in the vicinity of high stresses, stress concentrations, thermal
stresses and other non-linear stress fields. Accurate stress intensity factors for such cracks are
necessary for reliable prediction of fatigue crack growth rates or fracture. The slice synthesis
approach used herein to computation of surface flaw stress intensities. To validate the analytic
derived tress intensity facors for semi-elliptic surface cracks, finite element method is used. Three-
dimensional finite elements were used to model a plate containing a semi-elliptic surface crack.
The propagation of semi-elliptical surface initiated fatigue cracks has been considered. Analytic
model for the stress intensity factors, derived in this work, are used for crack growth analyses and
fatigue life predictions. Fatigue life under a load spectrum was predicted using these analytic stress
intensity factors.. The finite element analyses were made using MSC/NASTRAN, with 20-noded
isoparametric three-dimensional solid elements. In order to model the square root singularity at the
crack tip, three-dimensional prism elements with four mid-side nodes at the quarter points (a
degenerate cube with one face collapsed) were used and the separate crack tip nodal points were
constrained to have the same displacements [1]. The quarter-point displacement FEM was
employed in present work to evaluate stress intensity factors along semi-elliptic crack front. This
method uses the out-of -plane displacement value at the quarter-point behind the crack tip. With
the square-root singularity simulated this method has been verified to be of good accuracy for
calculating of the stress intensity factors of a variety of practical cracked geometries [2]. In order to
readily use these results in crack growth prediction computer routines, equations relating stress
intensity to crack shape, and depth to plate thicknes ratio, have been empirically established based
on the slice synthesis model

3 3 i i
§c· 2 §a·
KA V Sa ¦¦ A ij ¨ ¸
©a¹
¨ ¸
©t ¹
i 0 j 0 (1)

3 3 i i
§c· 2 §a·
KB V Sa ¦¦ B ij ¨ ¸
©a¹
¨ ¸
© t ¹
i 0 j 0 (2)
where: KA is the stress intensity at depth; KB is the stress intensity at surface, s -applied stress, a-
crack depth, c - half surface length, t -plate thicknes; Aij and Bij are the coefficients (represent
the displacements over the ntire crack face, the continuity expression is evaluated at 13 points).
Figure 1 shows typical finite element model of structural component under tensile loading. To
796 K. Maksimovic et al.

check he validity of the above method for SIF evaluation by the semi-analytic the slice synthesis
approach, comparison between the calculated SIF results of surface cracks in plate and the solution
obtained by finite elements is shown in figure 2. For this purpose plate with semi-elliptic surface
crack under tension stress s=83.3 [N/mm2] is analyzed. Geometry properties of this plate are:
w=60 mm, t=10 mm, c=10 mm, a=10 mm.

FIGURE 1. Finite element mesh of semi-elliptic crack

FIGURE 2. Comparisons between analytic and FE results for SIF calculations

Good agreement between analytic determined SIF with finite elements is evident. These
analytic determined SIF can be reliable used in fatigue life estimation of surface cracked
structures.

References
1. K. Maksimovic, V. Nikolic-Stanojevic and S. Maksimovic, Efficient Computation Method in
Fatigue Life Estimation of Damaged Structural Components, FACTA UNIVERSITATIS,
Vol. 4, No. 16, 2004.
2. R. S. Barsoum, Triangular quarter-point elements as elastic and perfectly-plastic crack tip
elements, Int. J. Numer. Meth. Engng., 11, pp. 85-98, (1977).
 5. Integrity of Dynamical Systems 797

Invited Contribution
STRUCTURAL DAMAGE DETECTION VIA THE SUBSPACE
IDENTIFICATION METHOD

Marina Trajkovic1,2, Dragoslav Sumarac3 and Marina Mijalkovic2

1Institute
of Mechanics, Ruhr-University Bochum, D-44780 Bochum, Germany
2Department of Civil Engineering and Architecture, Aleksandra Medvedeva 14, 18 000 Nis, Serbia

and Montenegro
3Department of Civil Engineering, Bulevar Kralja Aleksandra 73, 11 000 Belgrade, Serbia and

Montenegro
trajmarina@yahoo.com, sumi@eunet.yu, maca@gaf.ni.ac.yu

Based on theoretical preposition from reference Xiao et al. [1], concerning The subspace
identification method as one of the possible variants of inverse dynamic analyses, behaviour of real
structural systems with real load and really noise contaminated input/output data were investigated
in this work.
The change of dynamic parameters (stiffness, damping), as a consequence of structural history
(erosion, friction, fatigue, internal damages and cracks), has an impact on decrease of reliability
and serviceability of a structure or, drastically, causes its collapse. Having in mind that structural
damage is small or inside the system, hence the detection cannot be done visually, the authors
propose one useful and non-destructive dynamic parameters evaluation tool - vibration monitoring
of the structure.
The report of the original investigation conducted on real models with real impulse load in
laboratory is also given, Fig. 1. The hypothesis that the damage occurrence or the decrease of
integrity of structural system leads to the change of the dynamic characteristics of the structure is
tested during the investigation.

The authors propose the Scilab, a free MatLab clone, as a powerful tool for numerical
computations and data analysis. A special software for experiment monitoring and for
determination of relevant mechanical characteristics as well as location of possible damage of
construction has been developed (for complete code see Trajkovic [2]). The results of the
experimental part of the work, Fig. 2., were used as the entrance data for the software both in the
case of complete set of data (number of sensors which measure acceleration of the structure as a
response of the load is equal to the number of degrees of freedom) and in the case of incomplete
798 M. Trajkovic et al.

data (for the systems with a large number of degrees of freedom) where the missing values are
calculated through the iteration algorithm.

With the progression in programming languages and software packages, computer-aided

engineering methodology has significantly improved the practice of structural engineering. Hence,
the experimental results as well as the values obtained using numerical methods have been
compared with the results of structural analysis and simulation of the same model performed with
the finite-element software package ANSYS.
Considering that within the mentioned subspace identification method there still are some
unsolved issues and that it is a topical research issue in the world, the authors emphasise the
observed problems, and offer some ideas for potential overcoming of the method inadequacies.

References
1. Xiao, H., Bruhns, O.T., Waller, H., Meyers, A., Journal of Sound and Vibration 264/4, 601-
623., 2001.
2. Trajkovic, M., Investigation on dynamic properties of linear systems, Edition Academia,
Andrejevic Endowment, Beograd, Serbia and Montenegro, 2004.
3. Trajkovic, M., Monitoring dynamic properties of linear systems, Master thesis, Ruhr
University Bochum, Germany, 2002.
 5. Integrity of Dynamical Systems 799

Invited Lecture
CLOCK MECHANISM AS BASE OF ARTILLERY SAFETY AND ARMING
DEVICES

M. Ugrcic
Military Technical Institute, Ratka Resanovica 1, 11132 Belgrade, Serbia & Montenegro
marinkou@EUnet.yu, vti@vti.vj.yu

The coupled cogged pear consisted of the pallet and escape wheel is the most important part of the
clock mechanism (Fig. 1) that is the base of the function of all types of the mechanical watches. In
this case, the elastic force of compressed spring or gravitation’s force of bonded weights support a
driving (propulsion) and regular function of the clock mechanism.
Besides the mentioned, the clock mechanism has been used largely to design the safety and
arming devices for all modern types of fuzes for artillery projectiles because of the precision,
physical stability, and primordially its-self reliability. In this case, the axial forces of inertia or
centrifugal forces of bonded mobile masses drive the mechanism.
The safety and arming device (S&A device), as a subsystem for safety and arming is a special
part of artillery fuze consisted of the mechanism that takes broken mechanically the explosive train
in the fuze and absolutely makes it safe and secure during the storage, transportation, and handling
as well as in the beginning phase of projectile launching. The mechanism makes possibility to
reinstate the initial explosive train (arming) only then the projectile crosses to muzzle safety
distance (assigned to minimum 400 of calibers, i.e. projectile diameters).

FIGURE 1. 3-D view and components of the train of coupled gears in the clock mechanism

The determination of optimum design parameters of this very responsible and relatively
complex mechanism is not possible without an appropriate calculating method. The mathematical
modeling of the safety and arming device is based on main dynamics equations of motion for the
rigid material system and equations of the impact mechanics. Just a procedure is a rather
complicated and long, that was reason to show in the paper, as well as to explain, only the final
equations for motion of the mechanism elements. The mathematical model of the compound
movement of mechanism carried out for each phase of three main motions: coupled motion, free
motion, and impact.
On the basis of derived mathematical model of S&A device motion, the program code
OAMFOR for resolving of differential equations of the clock mechanism motion and numerical
800 M. Ugrcic

simulation of its function is developed. The program makes possibilities to optimize mechanism
and perform arming zone as well. Some calculation results with variation of the significant
parameters of the clock mechanism are shown (Fig. 2).
Finally, the S&A device have been tested experimentally by firing from real artillery systems
76 and 105 mm of calibers. Some of the experimental results are shown and compared with results
of numerical simulation. Both functions of S&A device safety and arming where successfully
approved.

FIGURE 2. Safety zone s depending on design parameters b and Ȝ of real clock mechanism
(numerical simulation)

References

1. Lowen, G.G., Tepper, F.R., Dynamics of the Pin Pallet Escapement, Tech. Report,
ARRADCOM, Dover, 1982.
2. Lowen, G.G., Tepper, F.R., Computer Simulation of Artillery S&A Mechanism (Involute Gear
Train and Straight-side Verge Runaway Escapement), Tech. report, ARRADCOM, Dover,
1982.
 5. Integrity of Dynamical Systems 801

Invited Contribution
TWISTING DEFORMATION EVOLUTION OF DRILLING ROPES

N. P. Puchko
Kiev State Shevchenko University
Ap.18, 16, St.Garmatna, Kiev, 03067, Ukraine
p_natalia@univ.kiev.ua

The drilling ropes are used for lifting or lowering of measurement instruments into boreholes. The
spatial orientation of instrument container affects the measurement accuracy of layer occurrence
depth and analysis of the layer properties. In this paper an evolution of twisting deformation of the
drilling rope in processes of lifting, lowering and stopping in the borehole is studied.
The specific character of rope deformation is conditioned by its rheological properties and
ability to come untwisted under a longitudinal loading.

The longitudinal - twisting deformations of ropes are described by equations

f
w 2U w 2U w 2T § w 2U w 2T ·¸
Ru (t  W )¨
U
wt 2
EF
wx 2
 nEF
wx 2
 EF
³
f
¨ wx
©
2
k
w x 2 ¸¹
dW  f ( x , t ) ,

f
w 2T w 2U 2
w 2U
U r2
w t2
kEF
w x2

 GJ P  k 2 EF  ww xT  kEF ³ R
2 U (t  W )
w x2

f
f 2

³ GJ
f
P RT (t  W )  k 2 EFRU (t  W )  ww xT d W  m( x, t ) ,
2
(1)

where U ( x, t ) and T ( x, t ) are the longitudinal and twisting displacements of the rope cross

sections, EF and GJp – instantaneous longitudinal and twisting rope rigidities, Ru (t  W ) and
RT (t  W ) are the relaxation kernels of the longitudinal and twisting deformations, is a linear
mass, k is an untwisting coefficient.
Owing to the significant viscosity of liquid solution in the borehole the longitudinal and
twisting vibrations of the rope with load are quickly faded. The rope untwisting (or twisting) is
conditioned by the action of the force of weight, the distributing force of external viscous friction
which is depended upon the velocity of rope lifting or lowering. It is described by the equation

§ 2 t ·
¨w T w 2T ¸ wT
GJ p ¨
¨ w x2

³ RT ( t  W )
wx
dW ¸  E
2
¸ wt
 k U ( g  a ( v )) 0
© f ¹ (2)

wT
E
where is a coefficient of external viscous friction, wt – a moment of a force of external viscous
friction U a(v) , – the longitudinal distributing force of external viscous friction.

The boundary conditions for the function T ( x, t ) under lowering have the next outlook:
802 N. P. Puchko

§ wT wT ·
T l (t ), t )  T 0 ( x) under x=l , ¨¨ E  GJ p ¸  km0 ( g  a (v ) under x l (3)
wt wx¸ © ¹
0

Under lifting

t
§ wT ·
T l (t ), t )  ˜ l(t ) dt
³ ¨¨© w x ¸¸¹
0 x l
,

§ wT wT ·
¨¨ E 0  GJ p ¸ km0 ( g  a (v)) under x l0 (4)
© w t w x ¸¹
where m0 – a mass of instrument container.
The solution of equation (2) is presented in the form

T ( x, t ) T0 (l )  ( x  l ) ) (t ) (5)
It is found by the direct methods like Galerkin’s method and has a view

§ t · § § t ··
¨ ¸ kg § U ( l0  l ) · ¨ ¨ ¸¸
T l0 , t  T 0 (l ) exp ¨  O (l ) dt ¸ 
¨ ³ ¸
¨ m0 
GJ p © 2 ¹¨ ¨ ³
¸ ¨ 1  exp ¨  O (l ) d t ¸ ¸
¸¸
© 0 ¹ © © 0 ¹¹ (6)
where l(t) and l0 are Lagrange’s coordinates of the upper and lower points of the drilling rope,

§ l0  l ·
O (l ) GJ p l(t ) ¨ E 0  ¸
© 3 ¹ .

At first glance it would seem that the observed in practice regularities of untwisting-twisting of
the drilling rope with instrument container during the processes of lifting, lowering and stopping in
the borehole are the paradoxical ones. But the obtained solution (6) explains these peculiarities.
Under a first lowering and stopping a transition process determined by the initial untwisting of
rope occurs. Under next liftings, lowerings and stoppings the process of untwisting-twisting has
become periodic. In the case of the ideal elastic rope a curve described the twisting displacement
evolution of the measurement instruments container was obtained by the calculation.
The relaxation of strain, a magnitude of lifting or lowering velocities and duration of stopping
distinctly affects the magnitudes of untwisting-twisting. However in the case of sufficient duration
of these stages the rope behaviour approximates to the calculated data.
 5. Integrity of Dynamical Systems 803

Invited Lecture
HEREDITARY STRAIN THEORY OF SYNTETIC AND STEEL ROPES

O. O. Goroshko
Kiev State Shevchenko University
Ap.32, 22, av.Lesnoj, Kiev, Ukraine, 02166
dtm@univ.kiev.ua

The steel and syntetic ropes are widely used as the elements of transport, sea, drilling and other
structures. Millions tones of hightquality steels and organic materials are used for the production of
power ropes every year. The distinguishing properties of the ropes are: 1)untwisting, when the
ropes are under tension; 2) hereditary (rheological) character of deformation which is due to
contact interaction of wires and the changes of the lay angles.
A model of natural twisting rod is selected as a model, which describes the rope deformation.
Then longitudinal and twisting deformations U ( x, t ) and 4( x, t ) are coupled by the relationship

U ( x, t ) u ( x , t )  k 4 ( x, t ) (1)
where u(x,t) – elastic rope lengthening.
The untwisting rope moment of rope longitudinal tension is defined by the dependence

M ( x, t )  k P ( x, t ) (2)
where P(x,t) is the acting force.
The coefficient k in formulas (1) and (2) has a dimension of length and defines the connection
between the lengthening and twisting. At the same time k is an arm of an twisting moment.
The relationship between a force, a moment and deformations in the rope cross sections is
defined by the equation in the relaxation form

§ t ·
¨ wu w u ( x,W ) ¸
P ( x, t ) EF ¨
¨wx
 Ru (t  W )
³wx
d W ¸,
¸
© f ¹
§ t ·
¨ wT w T ( x,W ) ¸
M ( x, t ) GJ p ¨
¨ wx
 RT (t  W )
³wx
d W ¸.
¸
© f ¹ (3)
According to d’Alambert principle and the relationships (3) the dynamical equations of
longitudinal and twisting deformations have an outlook

t
w 2M w 2U w 2T § w 2U w 2T ·
Ru (t  W )¨ 2  k 2 ¸ dW  f ( x, t )
U
wt 2
EF
wx 2
 kEF
wx 2
 EF
³
f
¨ wx
© w x ¸¹

t
w 2T 2
w 2U w 2U
U r2
wt 2
GJ p  k 2 EF  ww xT2
 kEF
wx 2
 kEF
³R u (t W )
w x2
dW 
f
804 O. O. Goroshko

t
§ w 2T ·¸
( t  W )  k 2 EFR u (t  W )

³ ¨¨© GJ
f
p RT
w x 2 ¸¹
dW  m ( x, t )
(4)

Here Ru (t  W ) and RT (t  W ) are the relaxation kernels of the longitudinal efforts and
twisting moments, which are defined experimentally, EF and GJp – instantaneous longitudinal and
twisting rope rigidities, f ( x, t ) and m( x, t ) are the distributing force and moment.

In a general case the problem of nonlinear dependence of the kernals Ru (t  W ) and RT (t  W )

upon the longitudinal rope strains is discussed. It was established that the time of rope relaxation
may vary by an order depending upon the longitudinal strains and it may decrease upon increasing
strain. The kernals of relaxation of standard hereditary bodies are used for the description of the
relaxation of longitudinal strain and twisting moments under fixed average value of strain.
The dependency of the rhelogical properties of syntetic ropes upon a humidity is discussed.
The experimental data for the definition of the rheological parameters of the dry and humid
syntetic ropes applied in the trawl technique are given. The dynamic deformation of the ropes
under a variation of their length with regard to nonholonomic conditions at the point x l (t ) of
rope running on a cylinder is investigated. The nonholonomic conditions take the form:
t
w U (l , t ) 
U (l (t ), t )
³
0
wx
l (t ) dt
,
t
w T (l , t ) 
T (l (t ), t )
³
0
wx
l (t ) dt
(5)
These conditions determine a carrying mechanism of the rope deformation through the
boudary point of rope running on or running out a cylinder. A nature of elastic crawling over a
rope under its passage through a pulley is explained. The strain relaxation in a rope reeling part at
lifting and its manifestation in the processes of the dynamical deformation at the next rope
lomering is taken into account. The critical velocities of the load lifting by rheological ropes under
excess of which the dynamical process has become nonsteady are defined. Under these velocities
the amplitudes of vibrations of dynamical efforts increase.
The conditions under which the rheological dynamical equations (4) can be put into the
canonical form were established. There were determined the velocities of propagation of two
groupes of longitudinal-twisting waves in the rheological rope. The applied problems of static and
dynamical ropes deformation are solved. There were considered the questions of fatigue
destruction of ropes and ways of the fatigue strenght estimation of ropes.
 5. Integrity of Dynamical Systems 805

Invited Lecture
BRITTLE AND DUCTILE FAILURE IN THERMOVISCOPLASTIC SOLIDS
UNDER DYNAMIC LOADING

R. C. Batra and B. M. Love

Department of Engineering Science and Mechanics, M/C 0219
Virginia Polytechnic Institute and State University
Blacksburg, VA 24061, USA
Tel. 540-231-6051;
rbatra@vt.edu

Plane strain transient finite thermomechanical deformations of heat-conducting particulate

composites comprised of circular tungsten particulates in nickel-iron matrix are analyzed using the
finite element method to delineate the initiation and propagation of brittle/ductile failures by the
nodal release technique. Each constituent and composites are modeled as strain-hardening, strain-
rate-hardening and thermally-softening microporous materials. Values of material parameters of
composites are derived by analyzing deformations of a representative volume element whose
minimum dimensions are determined through numerical experiments. These values are found to
be independent of sizes and random distributions of particulates, and are close to those obtained
from either the rule of mixtures or micromechanics models.
Brittle and ductile failures of composites are first studied by homogenizing their material
properties; subsequently their ductile failure is analyzed by considering the microstructure. It is
found that the continuously varying volume fraction of tungsten particulates strongly influences
when and where adiabatic shear bands (ASB) initiate and their paths. Furthermore, an ASB
initiates sooner in the composite than in either one of its constituents.
We have studied the initiation and propagation of a brittle crack in a precracked plate deformed
in plane strain tension, and a ductile crack in an infinitely long thin plate with a rather strong defect
at its center and deformed in shear. The crack may propagate from the tungsten-rich region to
nickel-iron-rich region or vice-a-versa. It is found that at the nominal strain-rate of 2000/s the
brittle crack speed approaches Rayleigh’s wave speed in the tungsten-plate, the nickel-iron-plate
shatters after a small extension of the crack, and the composite plate does not shatter; the minimum
nominal strain-rate for the nickel-iron-plate to shatter is 1130/s. The ductile crack speed from
tungsten-rich to tungsten-poor regions is nearly one-tenth of that in the two homogeneous plates.
The maximum speed of a ductile crack in tungsten and nickel-iron is found to be about 1.5 km/s.
We have also performed mesoscale analysis of ductile failure in tungsten heavy alloys
comprised of tungsten particulates immersed in nickel-iron matrix. Plane strain tension/
compression, plane strain simple shear and axisymmetric deformations have been analyzed. It is
found that the mesoscale analysis predicts the initiation of an ASB much sooner than that predicted
by the analysis of the equivalent homogenized body. Furthermore, the ASB initiation criterion for
a particulate composite is quite different from that for a homogenized body. Even though the ASB
initiation criterion for a monolithic material has been satisfied at numerous points, a coherent shear
band develops much later. The location and orientation of the ASB can be determined by
analyzing results of the complete simulation rather than when the analysis is in progress.
For the multiscale analysis we first analyze transient coupled thermo-mechanical deformations
of a homogenized body with values of thermophysical material parameters equivalent to those of
the particulate composite. Time histories of deformation variables on the bounding surfaces of the
centrally located 1 mm x 1 mm subregion of the 5 mm x 5 mm region are recorded. These are then
806 R. C. Batra and B. M. Love

used to study plane strain coupled thermomechanical deformations of the 0.5 mm x 0.5 mm
subregion of the 1 mm x 1mm region with the remaining part comprised of the equivalent
homogeneous material of the 5 mm x 5 mm body. It is found that the multiscale analysis of the
problem gives an ASB initiation time of ~ 22 µs as compared to ~ 58 µs in the equivalent
homogenized body and ~ 50 µs in the macroanalysis of deformations of the 0.5 mm x 0.5 mm
region containing randomly distributed 50% volume fraction of 50 µm diameter tungsten
particulates.

References:
1. R. C. Batra and B. M. Love, Adiabatic Shear Bands in Functionally Graded Materials, J.
Thermal Stresses, Vol. 27, 1101-1123, 2004.
2. R. C. Batra and B. M. Love, Crack Propagation due to Brittle and Ductile Failures in
Microporous Thermoelastoviscoplastic Functionally Graded Materials, Engineering Fracture
Mechanics (in press).
 5. Integrity of Dynamical Systems 807

Invited Contribution
SOME ASPECTS OF DYNAMIC INTERFACIAL CRACK GROWTH

R. R. Nikolic and J. M. Veljkovic

Faculty of Mechanical Engineering, Sestre Janjic 6, 34000 Kragujevac
DP ZASTAVA MASINE, Trg Topolivca 4, 34000 Kragujevac
Serbia and Montenegro
inikolic@ptt.yu, vkatarina@ptt.yu

The scientific describing of the crack initiation and crack growth mechanics on bimaterial
interfaces is of vital importance for understanding the failure processes in materials like
composites and ceramics. An important failure mechanism in fiber or whisker reinforced ceramic
composites is the debonding between the matrix and the reinforcing phases. This process of
debonding can take place either quasi-statically or dynamically, depending on the loads nature that
the composite structure is subjected to. If the interface is weakened by existence of flaws, they can
serve as initiators of a crack, which, under adequate circumstances, can propagate in an unstable
manner. Such situations lead to necessity of analyzing the dynamic interfacial crack growth. To
formulate the mechanism of initiation and dynamic crack growth in bimaterial systems, it is
necessary to know the whole 3D structure of the field around the tip of the moving interfacial
crack.
In this paper is presented the asymptotic analysis of the strain field around the tip of a crack
that propagates along an interface. The asymptotic methodology is applied in order to reduce the
problem to solving the Riemann-Hilbert problem. By that the displacements potentials are
obtained that are used for explicit determination of the stress field in the vicinity of the tip of the
nonuniformly propagating crack. The fields at the crack tip are determined for velocities within
interval zero and less than two transversal wave velocities. The obtained analytical results were
compared to experimental results of dynamic growth in different bimaterial systems, obtained by
the optical method Coherent Gradient Sensor (CGS) and by high-speed photography. This
comparison serves for demonstrating the necessity of using the complete expression in the analysis
of the optical data and in exact prediction of the dynamic fracture behavior.
The stress and displacement components, for each material, are expressed in terms of
displacement potentials. By applying the asymptotic methodology the equations of motion are
reduced to a system of coupled partial differential equations in terms of displacement potentials.
The term coupled equations is used in the sense that the higher order solution will depend on the
lower order solutions. Combination of the homogeneous and particular solution satisfies the
boundary conditions on the crack faces and along the interface.
By applying the boundary conditions on the interfacial crack faces and the bonding conditions
along the interface the problem of nonuniform crack growth is reduced to solution of the Riemann-
Hilbert problem. As a solution of this problem the displacement potentials are obtained. Those
equations are then used to present the stress distribution as a function of the angle at the crack tip.
These stress distributions as a function of angle for different velocities at the crack tip on the
interface were not met in the literature, yet. These diagrams are shown here for the first time. From
them (Fig.1a) one can notice the stress Vrr jump across the interface under the pure Mode I
conditions. In pure Mode II conditions this jump does not exist. The same result is obtained for the
case of the stationary crack. The maximum of stress VTT, in conditions of pure Mode I is moved
towards the material 1 with the velocity increase, what corresponds to conditions of the mixed
mode for the stationary crack.
808 R. R. Nikolic and J. M. Veljkovic

By solving the Riemann-Hilbert equation and Stieltjes transforms the asymptotic elasto-
dynamic field is obtained. The most important characteristics of this field is that in appearance of
two terms in these equations that are completely different from terms found in solution for the
crack which propagates nonuniformly in the homogeneous medium. The first term is completely
related to the interfacial nature of the crack growth, since it depends on the oscillatory index H. The
second term depends on the complex dynamic stress intensity factor that is characteristic for
interfacial crack, and on the crack tip velocity and its time derivative.
In this paper is given the new approach to asymptotic analysis of the strain field around the
interfacial crack tip. The programming routine Mathematica was applied and the solutions are
compared to those obtained by Yang et al. [1] and Liu, Lambros and Rosakis [2]. Based on
obtained results and that comparison we concluded that the complete expression from [2] has to be
applied for the analysis of the dynamic interfacial crack growth. This also proves superiority of the
analytical results obtained by Mathematica routine relative to numerical results obtained from
references.

FIGURE 1. a) Angular stress distribution for bimaterial combination PMMA/steel for Mode I
conditions. b) Asymptotic analysis results, for the crack velocity v=720 m/s; c) Intereferogram
obtained for impact load, obtained by the CGS method. Bimaterial combination Homalite-100/Al
Red lines are contours of the asymptotic analysis results

References:
1. Yang, W., Suo Z. and Shih C. F., Proc. R. Soc. Lond. A 433, pp. 679-697, 1991.
2. Liu C., Lambros J. and Rosakis A. J., J. Mech. Phys. Solids Vol. 41. pp. 1887-1954, 1992.
 5. Integrity of Dynamical Systems 809

Invited Contribution
ON STABILITY PROBLEMS OF PERIODIC IMPACT MOTIONS

S. Mitic
Faculty of Occupational Safety University of Nis
Yu-18000-Nis, Zetska 4/9, Serbia and Montenegro, Yugoslavia
fax: +381 18 49 962
smitic@znrfak.znrfak.ni.ac.yu, nimitic@bankerinter.net

Behavior of mechanical systems exhibiting impacts is of great relevance in many practical

engineering problems. Mechanical systems with impact interactions have wide applications in
engineering as the most intensive source of mechanical influence on materials, structures and
processes. Periodic oscillations with impacts take place very often in machine dynamics, structure
machine, vibration engineering, vibration ecology problems as well as in ambient buildings and
city structures etc. Oscillatory systems exhibiting impacts are of great relevance in many practical
engineering problems.
Stability problems of one group of periodic impact motions can be solved in the analytical
form. “Accurate” method of “adjustment” by Kobrinski [1] was first developed and later different
models of impacts introduced by Babicki [2], Masri and Caughay [3] and other were developed. If
dynamic model of vibro-impact system is described with differential equations whose solutions
can be found, the coordinative increments of positions and speeds can be expressed with help of
small increments of integration constants and phases. After determining small increments of
coordinative positions and speeds for the two following intervals, there should be used adjustment
conditions, i.e. limit conditions and periodical conditions expressed in shape of small increments
of coordinative positions and speeds as well as time, in the moment of impact which appears
between that two following intervals. Then the recurrence links are obtained between small
increments of arbitrary constants for two following intervals in shape of linear systems of
homogeneous equations, of final differences in relation to small increments. The question of
periodic vibro-impact process stability is solved considering recurrent relations, which link small
disturbances of integration constants. These constants can be expressed in the shape of the product:
GC ji h j E i , where h j does not depend on i interval, and E is a constant number. So the
question of stability is reduced to the problem of determining E . System of linear equations with
small disturbances, has obtained, is homogeneous and has nontrivial solution if determinant of
system is equal zero. The roots of characteristic determinant characterize stability of examined
vibro-impact regime. For vibro-impact system with n / 2 degrees of freedom, characteristic
determinant has n roots. If root modules of characteristic equations are less then one, then for the
great number of impacts the movement is stable. The case when module of at least one root of
characteristic equation is equal to one or it is higher and then bifurcation appears. On the basis of
this method, the analytical procedure for determining stability of the periodic motion pendulum
with rigid barriers, which limit the amplitude variation from its central position, was treated. It is
considered in both the normal (downward) position and in the upright (inverted) position. The
overall dynamics of the harmonically excited pendulum include impacts with the rigid arresters.
This method is used for determining the characteristic equation for assessing stability of vibro-
impact pendulum system’s periodic motion. Also, this method is used for determining the
characteristic equation for assessing stability of the vibro-impact system’s periodic motion that
consists of mass, spring, damper and a rigid arrester set at a particular distance from the mass’s
810 S. Mitic

equilibrium position. The mass, acted upon by one or two asynchronous excitation forces, thus
inducing the mass’s oscillatory motion interrupted by impacts upon the rigid arrester.
New direction in developing theory about vibro-impact systems appears with development of
bifurcation and chaos theory, which is enabled with improving computer technique and with using
numerical and experimental methods (Moon and Rand [4], Shaw and Holmes [5], Peterka and
Kotera [6], etc.). After 80‘s, a great number of authors are occupied with different models of vibro-
impact systems where basic oscillatory system is described with linear differential equations, but
complexity of problem as nonlinear exists because of the influence of impacts which provoke free
oscillations of system additionally in relation to forced oscillations. In later years, it can be noticed
from accessible literature that Peterka and Kotera gave significant contribution in their works, and
they are giving generalized solutions about transitions between different types of impact motion
and ways of how motion of system develops from periodic to chaotic.
It is characteristic of vibro-impact systems that at simple deterministic system with periodic
initiative, reactions are various, from possibilities of periodic to chaotic motion. In the case of
vibro-impact system, Peterka [7], after analytical procedure on the basis of adjustment” method
(small disturbances of amplitude and phase), with use of numerical procedure, were determined
limits of stability with determining number E .
The transition between periodic and chaotic impact motion on the limit of saddle-node
bifurcation is very sensitive to system parameter change, which come from the character of saddle-
node instability. Laws of transition from periodic to chaotic impact motion are explained with help
of their areas of existence and stability, bifurcation characteristics and phase trajectories, of
symmetrical and asymmetrical periodic process, and phase trajectories for vibro-impact system
where oscillator is described with Duffing’s differential equation with one periodic initiative force,
where the initiative of scratching impacts can be seen.
In relation to stereotype oscillatory theory by which reaction is periodical if initiative is
periodic, in theory of vibro-impact systems that becomes unbearable, reaction can be of chaotic
character, because of bifurcation occurrence, and because of unpredictable additional impacts
which reduce kinetic energy and amplitudes of motion system. Although the system is returning to
the condition before bifurcation, initiative force is providing energy, amplitudes are growing and
additional impacts are arising again with possible effect of chaotic motion.
Keywords: stability, vibro-impact system, periodic motion, chaotic motion, bifurcation, phase
trajectories.

References
1. Kobrinskii, A.E., Kobrinskii A. A., Vibroudarni sistemi, Moskva, Nauka, 1973.
2. Babicki.I. , Teorija vibroudarnih sistema, Moskva, 1978.
3. Masri S.F., Caughay T. K.,O. S. of the I.D., Trans. of ASME, september, 586-592,1966.
4. Rand, R.H. and Moon, F.C., B. C.. Int. J. Non Linear Mechanics, Vol.25, No.4: 417-432,1992
5. Shaw S.W., Holms P.J., A P. F.I. O. , Journal of Applied Mechanics, Vol. 50/849-657,1983.
6. Peterka. F. and Kotera, T., R.of E., Acta Technica CSAV. No.1 : 92-117, 1982.
7. Peterka, F., D. of M.S., EUROMECH-2nd ENOC. Prague 1996
 5. Integrity of Dynamical Systems 811

Opening Invited Lecture

DYNAMICAL INTEGRITY OF NONLINEAR MECHANICAL OSCILLATORS

S. Lenci and G. Rega

Dip. Architettura, Costruzioni e Strutture, Università Politecnica delle Marche, via Brecce
Bianche, 60131, Ancona, Italy
Dip. Ingegneria Strutturale e Geotecnica Università di Roma “La Sapienza” via A. Gramsci 53,
00197, Roma, Italy
Lenci@univpm.it, Giuseppe.Rega@uniroma1.it

This work overviews and continues recent investigations of the authors [1, 2] on the dynamical
integrity of nonlinear mechanical systems and other oscillators. In fact, it has been realized that
attractors must be paralleled by uncorrupted basins for safe practical applications [2, 3]. Eroded
basins constitute a critical state for the structure corresponding to its impending failure, which can
be instantaneous in the presence of perturbations or uncertainties. On the other hand, using the
structure only when erosion is prevented may be too much conservative, because it can actually
survive safely well above the relevant threshold if the erosion is not sharp.
These considerations triggered detailed investigations of the safe basin erosion, which is
basically constituted by four steps [2].
1 Choice of the right definitions of “safe basin” and integrity measure. They may vary, e.g.,
(i) if one is interested in transient or steady dynamics, and (ii) if fractality of the basin can
be accepted or must be disregarded, and have to be chosen according to the specific
problem. Also the availability of fast algorithms for computing the integrity measure plays
a role, because all of the following steps are very time consuming.
2 Investigation of the basins evolution. This item can be done only by extensive and
systematic numerical simulations. However, key information, either exact or approximate,
have to be obtained by studying the global bifurcations responsible for the main
metamorphoses of basin boundaries, which sometimes enlighten on and explain in detail
the hidden complex dynamical mechanisms.
3 Construction of the erosion profile for a varying system parameter, e.g., the excitation
amplitude. This step is crucial because it permits to estimate how the dynamical integrity
depends on the considered parameter. Moreover, it is very important in practical
applications because it possibly allows for optimal choices of the parameter, if the erosion
profile has maxima, or suggests to stay far apart from dangerous sharp falls of the profile.
Indeed, it has been noted [2] that the erosion profiles are usually sharp in the neighborhood
of resonance frequencies, whereas they are dull elsewhere.
4 Study of the collapse of the safe basin. In fact, the erosion ends with the onset of out-of-
well phenomena, namely, with the occurrence of scattered attractors, for multi-well
hardening oscillators, or escape (which can have different physical meaning, e.g.,
overturning for rigid blocks, capsizing for ships, collapse to the substrate for
electrodynamically actuated microbeams, etc.), for softening oscillators. According to the
various cases, the structure itself may collapse or survive (in a desirable or unwanted way)
after this event.
The second point that will be addressed is how it is possibly to maintain the dynamical
integrity under severe conditions, or, equivalently, how it is possible to reduce or shift the safe
basin erosion. In this respect, a technique previously developed by the authors [1] for controlling
812 S. Lenci and G. Rega

nonlinear dynamics and chaos will be applied. The method consists in optimally eliminating
appropriate homo/heteroclinic bifurcations by varying the shape of the excitation, and it is based
on the observation that the erosion phenomenon is usually triggered by the global bifurcation of a
given (usually hilltop) saddle, although it can also involve other topological events [2, 3]. The
elimination, or shift in parameter space, of this bifurcation – which is the key point of the method –
is thus expected to have positive effects in terms of dynamical integrity of the system.
Several uncontrolled and controlled erosion profiles of various mechanical oscillators will be
built, investigated in detail and compared with each other to the aim of verifying the performances
of control. Indeed, control method and dynamical integrity are intimately connected. In fact, the
control permits to shift erosion, i.e., to maintain integrity. On the other hand, the ability of control
in shifting the erosion profile is an important tool to check its practical effectiveness.
Apart from overall developments, in this paper previous works are extended in at least two
specific directions. First, further mechanical systems are considered and compared with each other,
possibly in different dynamical regimes. This permits to enlighten on various aspects which were
not adequately studied.
Second, special attention is devoted to a new class of solutions, namely the rotating solutions
of the mathematical pendulum, which are different from fixed points and oscillating solutions (of
various mechanical systems) considered in previous works, and are likely related to different
topological phenomena. Moreover, they are important in nowadays practical applications.
Another novelty of this class of solutions is that they are born by a saddle-node bifurcation at a
certain level of excitation amplitude. This is quite different from the solutions studied in previous
works, which have no triggering bifurcation and exist also in the unforced case. In terms of erosion
profiles this is expected to entail strong differences. In fact, while in the latter case the integrity
measure is a decreasing function of the excitation amplitude and provides the well known “Dover-
cliff” erosion profile [3], in the former case the profile initially increases due to the onset of
solution basin up to reaching one or more local maxima, then it is eroded by the incursion of unsafe
fractal tongues from outside and eventually collapses. This question, which has also some effects
in terms of normalization of the erosion profiles needed to compare different systems, will be
discussed in detail.

References
1. Lenci, S. and Rega, G., A unified control framework of the nonregular dynamics of
mechanical oscillators,” J. Sound Vibr., vol. 278, 1051-1080, 2004.
2. Rega, G. and Lenci, S., Identifying, Evaluating, and Controlling Dynamical Integrity
Measures in Nonlinear Mechanical Oscillators, in press, Nonlinear Analysis, T.M.&A, 2005.
3. Thompson, J.M.T., Chaotic Behavior Triggering the Escape from a Potential Well, Proc.
Royal Soc. London A, vol. 421, 195-225, 1989.
 8. Modelling of Material Property Data and Fracture Mechanisms 813

FATIGUE CRACK INITIATION AND PROPAGATION AT HIGH

TEMPERATURE IN A SOFTENING MARTENSITIC STEEL

B. Fournier1,2, M. Sauzay1, M. Mottot1, V. Rabeau1, A. Bougault1 and A. Pineau2

1CEA, DEN-DMN-SRMA, Batiment 455, 91191 Gif-sur-Yvette cedex, France.
2ENSMP, Centre des Matériaux P.-M. Fourt, UMR CNRS 7633, BP 87, 91003 Evry, France.

benjamin.fournier@cea.fr

Martensitic steels of the 9Cr1Mo family have been selected for advanced power generation
systems, in particular because of their high thermal conductivity and low thermal expansion
coefficient [1]. Typical in-service conditions with repeated start- and stop-operations lead to
loadings of creep-fatigue type at high temperature (between 673 and 873 K) , with very long hold
times (typically one month). Submitted to cyclic loading these martensitic steels are known to
soften both under pure fatigue and under creep-fatigue loading [2,3]. Pure fatigue and creep-
fatigue mechanical tests were performed for different cyclic and creep applied strains. Fig. 1 shows
that the number of cycles to failure decreases with increasing applied creep strain but the time to
failure strongly increases. The interaction between fatigue and creep damage is illustrated by the
map drawn in Fig. 2.
SEM and optical observations of both the body and the fracture surfaces and of longitudinal
sections were performed. In pure fatigue, the usual striations are observed but the observation of
the tortuous propagation paths of secondary cracks (Fig. 3) reveals the strong influence of
environment and recalls the paths usually observed under fatigue-corrosion interactions. In creep-
fatigue tests both transgranular and intergranular propagation are observed on the same sample.
The environment is found to have a strong influence on the crack propagation. The more damaging
effect of hold times in compression is attributed to the environment effect [4], the mean stress
effect and a higher crack nucleation rate. The number of secondary cracks is the smallest for pure
fatigue tests, and is higher for compression than for tension hold times. Short secondary cracks are
found to be completely filled with oxide, whereas for long secondary cracks the amount of oxide
depends on the nature of the hold period. In addition, the growth of the oxide layer is found to be
strongly different depending on the type of loading. WDS measurements reveal that both inner and
outer oxidation take place and differences in the oxide layer compositions are observed. An
attempt is made to use the results of these mechanical tests and metallurgical observations to
model the creep-fatigue-oxidation interactions in this steel.

FIGURE 1. Evolution of a) the number of cycles to failure and b) the time to failure for increasing
creep strain in creep-fatigue tests with 'Ht=0.7% at 823K.
814 B. Fournier et al.

FIGURE 2. Interaction between fatigue and creep damage.

FIGURE 3. Observation of secondary cracks in longitudinal sections for a) a pure fatigue test at
823K and b) a creep fatigue test at 823K.

FIGURE 4. Observation of secondary cracks on the specimen surface, with the oxide layer etched
away, in a creep-fatigue test at 823K with a) compression hold time and b) tension hold time.

References
1. Swindeman, R.W. et al., Pressure Vessels and Piping, 507-512, 2004.
2. Kim, S., Weertmann, J.R., Metall. Trans. A19, 999-1007, 1988.
3. Lukas, P., Kunz, L., Sklenicka, V., Mat. Sci. Eng. A129, 249-255, 1990.
4. Hecht, R.L., Weertman, J.R.. Metall. Trans. A24, 327-333, 1993.
 8. Modelling of Material Property Data and Fracture Mechanisms 815

TRANSFERABILITY OF CLEAVAGE FRACTURE PARAMETERS BETWEEN

NOTCHED AND CRACKED GEOMETRIES

C. Bouchet1, B. Tanguy1, J. Besson1, A. Pineau1 and S. Bugat2

1Ecole des Mines de Paris, Centre des Materiaux, Evry Cedex, France
2EDF Les Renardieres, Ecuelles, Moret-sur-Loing, France

Benoit.Tanguy@ensmp.fr

In the recent past, the local approach to fracture has shown its capacity to be a predictive tool in
many complex cases of structural integrity assessment [1]. Brittle fracture of steels is one of the
fields where local approach has brought a lot of understanding (e.g. effect of prestraining, effect of
dynamic loading [2]). In the nuclear industry, local approach to fracture is often used as a
complementary tool to the code (e.g. ASME) in order to reduce the empirical margins and to
contribute to a better understanding of the physical mechanisms which can lead to fracture.
Most of the successful models devoted to the description of cleavage fracture are based on the
weakest link assumption. One of the pioneering models devoted to low temperature cleavage
fracture, the Beremin model [3], has linked the fracture probability with the defects population
through the Weibull stress concept. Besides its physical bases, the success of this model relies on
the small number of parameters that has to be calibrated. In the original study, notched tensile
specimens with different radii were used to calibrate the parameters over a given range of
temperatures, the model was then applied successfully to predict lower shelf fracture toughness
values of an A508 cl.3 steel. However the capacity of the model to predict cleavage fracture
toughness values with the same set of parameters when temperature increases, i.e. on the lower
part of the transition region, has been questioned in the litterature[4]. One underlying problem is
that for the same temperature, extensive plasticity may develop in notched tensile specimens
whereas it remains confined in deep-cracked geometries because of constraint conditions. On the
other side, the use of different temperatures for parameters calibration and toughness predictions
raises concerns about the temperature effect on the parameters, which were assumed to be
temperature independent in the original model.
In order to investigate the transferability of cleavage fracture parameters in the lower part of
the transition region, a large experimental program was carried out on an A508 cl.3 pressure vessel
steel. The scope of this study is firstly to give some insights on the evolution of cleavage fracture
mechanisms with increasing temperature for notched tensile and deep-cracked geometries.
Secondly, based on a large number of specimens to deal with statistical aspects, several strategies
have been carried out to calibrate the cleavage fracture parameters using the Beremin model.
Experimental investigations include a large set of quasi-static mechanical tests : notched
tensile (NT) specimens with 3 different radii tested at five temperatures (-150°C, -130°C, -100°C,
-70°C,-40°C) covering pure cleavage fracture up to full ductile behavior, deep-cracked CT
specimens with 2 different thicknesses (12.5 and 25 mm) tested over a large temperature range,
with elastoplastic fracture toughness values ranging from 45 up to 600 MPa.m1/2 . Range of
macroscopic failure strains for notched tensile specimens are given in Tab. 1. Besides mechanical
tests, a fractographic investigation was done for each tested specimen using scanning electron
microscopy. Both the nature of the site at the origin of the cleavage initation and its position on the
fracture surface was studied. Evolution of the nature of the cleavage triggering sites with
temperature is given for the different geometries. It is shown that for a given temperature and for a
fracture mode which remains macroscopically brittle, manganese sulfide inclusions play an
816 C. Bouchet et al.

increasing role in the notched tensile geometries whereas such inclusions were not found in deep-
cracked geometry.

TABLE 1. Macroscopic failure strain range for the tensile notched specimens (*:fully ductile)

Based on Axisymetric and 3D F.E. simulations, the local parameters (principal maximum
stress, plastic strain and triaxiality ratio) were calculated at the site of cleavage initiation for all the
tests performed.
Several strategies were then carried out to determine the Beremin model parameters. In a first
step, the parameters were determined at a fixed temperature and using the different notched tensile
geometries (with the 3 different radii). In a second step, the results obtained in the whole range of
temperatures for one given geometry were used. Finally, the whole set of experimental data was
used. The values of the different sets of parameters are then discussed. Capacity of the model to
predict the fracture toughness data is then discussed based on the observed fracture
micromechanisms.

References
1. Burstow, M.C., Beardsmore, D.W., Howard, I.C., Lidbury, D.P.G., Int. J. Pressure Vessels
and Piping, Vol. 80, 775-785, 2003.
2. Minami, F., Kazushige, A., J. of Pressure Vessel Tech., Vol. 123, 362-372, 2001.
3. Beremin, F.M., A., Met. Trans., Vol. 14A, 2277-2287, 1983
4. Wiesner, C.S., Goldthorpe M.R., Journal de Physique IV, Vol. 6, C6-295—C6-304, 1996
 8. Modelling of Material Property Data and Fracture Mechanisms 817

RELATION BETWEEN CRACK VELOCITY AND CRACK ARREST

M. Hajjaj1,2, C. Berdin1, P. Bompard1 and S. Bugat2

1Laboratoire de Mecanique des Sols, Structures et Materiaux, Grande Voie des Vignes, 92295

Châtenay-Malabry, France
2Electricite de France, R&D Division, Département MMC, Les Renardieres, 77818 Moret-sur-

Loing cedex, France

hajjaj@mssmat.ecp.fr, berdin@mssmat.ecp.fr, bompard@mssmat.ecp.fr, stephane.bugat@edf.fr

Initiation fracture toughness is a material parameter well established to predict crack initiation. An
improved fracture design for security component could be obtained in predicting crack arrest.
Arrest fracture toughness can be used for that purpose. However, this parameter although studied
for a long time [1] is still questionable, even though a standard test procedure was developed in [2].
The influence of dynamic effects on crack arrest are often invoked and the assessment of arrest
toughness under static analysis could lead to a dependence with specimen geometry [ 2].
The purpose of this paper was to determine the mechanical state ahead of a crack tip during
crack propagation at high velocity and arrest under quasi-static loading, in order to later study a
local fracture criterion.
Crack arrest experiments were performed using a notched disk taken from a A533 Cl.B type
steel, submitted to a thermal shock. This test was initially developed at Ecole des Mines de Paris
[3]. The specimen consists in a ring with a radial fatigue pre-crack issued from on its outer surface
(Fig. 1). It is first cooled at -196°C and its inner surface is then warmed up at about 600°C in 10s
using an electromagnetic inductor. After brittle crack initiation, crack propagates towards areas of
increasing toughness due to temperature rise. Crack arrest occurs after a few centimetres of high
speed propagation. Temperature field is measured by thermocouples introduced inside the
specimen. Crack velocity, which is about 600m.s-1,is measured by crack gages. A deceleration
during the last few millimetres of propagation was noted. Detailed experimental results are
reported elsewhere [Hajjaj 4].
Finite element modeling was used in order to obtain mechanical states at initiation, during the
propagation and at arrest of the crack. The element size at crack tip was down to 50 Pm.
Temperature loading was prescribed according to the experimental measurements. Temperature
dependence of the elastic-plastic behaviour of the material was accounting for. Strain rate
dependence was modelled through a Cowper-Symonds law. The release node technique was used
to represent the crack propagation. Different assumptions were made about the crack velocity: an
imposed constant crack velocity during crack propagation up to crack arrest, or crack deceleration
prescribed as measured in experiments. Considering high crack velocity, dynamic computations
were performed.
818 M. Hajjaj et al.

FIGURE 1. Precracked ring specimen : a crack length, W, width of the ring (left); Stress intensity
factor assessment during crack propagation (right).

Stress and displacement analyses at the crack tip showed that elastic singularity is dominant as
expected in viscous material [Kalthoff 1990 5]. The results are represented in terms of stress
intensity factor (Fig. 1). Crack velocity history has a major influence on the mechanical state ahead
of the crack tip: the stress intensity factor, i.e. the opening stress ahead the crack tip, highly
increases and then decreases since the crack decelerates. Post-arrest analyses showed that it
oscillates at the frequency of the first opening mode of the cracked specimen. In the case of the
constant crack velocity, this vibration mode occurs only after crack arrest, whereas, in the case of
the crack deceleration, it takes place during the last millimetres of propagation at low velocity.
Different computations showed that this is not explained by different times to arrest. Comparison
between stress intensity factors at arrest crack (the last points of the two curves in Fig. 1) indicates
that there can be a large uncertainty of the mechanical state at arrest assessment, due to crack speed
assumption. It seems therefore essential to account for the actual crack velocity in order to predict
crack arrest, either in generation or propagation modelling.

References
1. Bluhm, J.I., In Fracture V, edited by H. Liebowitz, Academic Press, New York, 1969, 2-60.
2. Kalthoff , J.F., In Crack Dynamics in metallic materials, edited by J.R. Klepaczko, Springer-
Verlag, New York, 1990, 69-254.
3. Bouyne, E., Joly, P., Houssin, B., Wiesner, C.S., Pineau, A, Fat. Fract. Eng. Mat., vol. 24,
105-116, 2001.
4. Hajjaj, M., Bugat, S., Berdin, C., Bompard, P., in Proceedings of ASME Pressure Vessels and
Piping Conference, 2004.
5. Freund, L.B., Dynamic fracture mechanics, Cambridge university Press, 1990.
 8. Modelling of Material Property Data and Fracture Mechanisms 819

MECHANISMS OF DAMAGE AND FRACTURE IN TRIP ASSISTED

MULTIPHASE STEELS

G. Lacroix, Q. Furnemont, P.J. Jacques and T. Pardoen

Departement des Sciences des Materiaux et des Procedes, Universite catholique de Louvain,
IMAP, Place Sainte Barbe 2, B-1348, Louvain-la-Neuve, Belgium
lacroix@imap.ucl.ac.be, pardoen@imap.ucl.ac.be, jacques@imap.ucl.ac.be

A wide range of applications, especially in the automotive industry, require both large ductility and
high strength, two properties that are unfortunately antagonist. TRIP-assisted multiphase steels, i.e.
steels consisting of an intercritical ferrite matrix with retained austenite, bainite and martensite
dispersed phases [1] provide extremely good combination of these two properties. In TRIP steels,
the mechanical contribution adds to the chemical free enthalpy as a driving force for martensitic
transformation of the metastable austenite. This transformation acts like an additional work-
hardening source that allows to increase both ductility and tensile strength [2]. The martensite
transformation of metastable austenite has been reported to have either a beneficial or a detrimental
effect on fracture toughness [3,4], depending probably on the complex relationship between
austenite stability, austenite volume fraction and martensite brittleness.
In this work, three different steel compositions have been studied in order to vary the initial
austenite content and two heat treatments have been defined in order to vary the austenite stability.
Fracture test were carried out on precracked Double Edge Notched Tension (DENT)
specimens in order to characterize the damage mechanisms and the toughness of these steels. At a
microscopic scale, the size of the transformed zone at the crack tip was measured by Orientation
Imaging Microscopy (OIM). The damage sites and crack path have been characterized in the near
crack tip region by SEM. The macroscopic toughness was measured by loading several DENT
specimens to reach different amount of crack advance. The Crack Tip Opening Displacement (c)
and the J integral at cracking initiation were determined. These two parameters quantify the
cracking initiation toughness. Some DENT specimens with different ligament lengths were also
loaded up to fracture. These tests allowed to estimate the essential work of fracture (we) which is a
more physical parameter for quantifying the resistance to crack propagation in thin plate than the
tearing resistance extracted from a JR curve.
From the combination of the microscopic and macroscopic results, the damage mechanisms
and the initiation fracture process were determined precisely (see Fig. 1): the transformation of all
the austenite in martensite takes place very early in the fracture process zone during the crack
opening process[5], then damage develops by decohesion between the ferrite and the martensite
[5], followed by the coalescence along the interface between the ferrite and the second phase.

FIGURE 1. Damage mechanisms and fracture process (F=ferrite, B=bainite, A=austenite and M=
martensite lathes).
820 G. Lacroix et al.

Figure 2 compares the we values and the Jc values. Whereas Jc seems to be strongly linked to
the austenite content, we presents another behavior that can be explained by the different austenite
stability. The necking and the damage contribution have been partitioned for the J integral and we
[6]. Both quantities contain a contribution cracking from necking and from material separation.
The separation of these two contributions has allowed to understand the variations shown on Fig. 2
as well as the difference between we and Jc.

FIGURE 2: Macroscopic toughness values (We and J) as a function of the austenite content.

References
1. Jacques, P.J., Cornet, X. , Harlet, Ph., Ladrière, J. and Delannay, F., Metall. Trans. A, 1998,
29A, 2382.
2. Jacques, P. J., Furnémont, Q., Mertens, A. and Delannay, F., Philosophical Magazine A, vol.
81 (7), 1789-1812, 2001.
3. Chen, J.H., Kikuta, Y., Araki, T., Yoneda, M. and Matsuda, Y., Acta Metall., 1984, 32, 1779.
4. Antolovich, S.D. and Singh, B., Metall. Trans., 2, 1971, 2135-2141.
5. Jacques, P. J., Furnémont, Q., Pardoen, T. and Delannay, F., Acta Materialia, vol. 49, 139-
152, 2001.
6. Pardoen, T., Hachez, F., Marchioni, B., Blyth, P.H. and Atkins, A.G., J. Mech Phys solids, 52,
423-452, 2004.
 8. Modelling of Material Property Data and Fracture Mechanisms 821

THE ROLE OF SUB-BOUNDARIES IN THE BRITTLE FRACTURE OF

POLYCRYSTALLINE MATERIALS

Gareth Hughes1, Peter Flewitt1, Fabio Sorbello1, Gillian Smith2 and Alan Crocker2
1Interface Analysis Centre, University of Bristol, Bristol, BS28BS, UK
2Department of Physics, University of Surrey, Guildford, GU27XH, UK
1gareth.hughes@bris.ac.uk , 2gillian.smith@surrey.ac.uk

Grain boundaries are resistant to the propagation of cleavage cracks in polycrystalline materials.
Indeed, the importance of grain boundary orientation on the propagation of cracks, and in
particular brittle cracks has been recognised by various workers [1,2]. In a polycrystalline material,
the effective cleavage planes and therefore cracks in adjacent grains do not meet each other in a
line in the common boundary, except in special circumstances. Therefore if the polycrystal is to
separate into two parts some additional failure at the grain boundary must occur; this can take the
form of multiple cleavage, brittle intergranular failure or ductile fracture [3]. However, in different
metals and alloys, a range of other boundaries and interfaces are encountered, which can modify
the propagation of brittle cracks, and in particular cleavage. Examples are twins observed in bcc
and hcp metals, martensite and bainite in ferritic steels and more generally, interphase boundaries.
Although the importance of these boundaries on the initiation and propagation of cracks has been
recognised for many years, there has been little attempt to address the interactions involved.
Various techniques are available to examine the detail of these local crack boundary
interactions ranging from simple optical microscopy to high resolution scanning electron
microscopy and in some cases, scanning probe techniques. More recently, Hughes et al. [4] have
demonstrated that focussed ion beam (FIB) techniques, which combine high resolution imaging
and milling, can be used to study the interaction between cleavage and grain boundaries in hcp
materials. In this paper, we consider the influence of twin boundaries in bcc and hcp metals and
alloys. The experimental results are then compared with the predictions obtained from microscale
geometrical models, as described by Crocker et al. [3].
Several materials have been investigated to consider different twin sub-boundaries. These
include a bcc nominal iron-3%-silicon steel, hcp zinc and Zircaloy. Each material has been heat
treated to optimise the grain size and the brittle fracture mode so that the role of the twin
boundaries on crack propagation can be investigated over a range of temperatures. The main
instrument for investigation of the fracture surfaces has been the FEI Strata FIB201 focussed ion
beam system, capable of producing high resolution images with microstructural information from
both surface and sub-surface regions.
Specimens of both hcp zinc and the bcc Fe-3%Si transformer steel were fractured at low
temperature, spanning the lower shelf end of the brittle to ductile transition curve. The
experimental observations for hcp zinc have shown that cleavage cracks can propagate either
across twins or along twin boundaries. Hence the computer models to describe fracture in hcp zinc
have been revised to incorporate both of these interactions between {10-12} twins and either basal
(0001) or prismatic {10-10} cracks. Initially, the models have one twin per grain, which, when
combined with either basal or prismatic cleavage contributions, predicts that a small amount of
grain boundary accommodation is required. This prediction is consistent with the lack of observed
grain boundary failure at low temperatures, where twinning is observed. Further investigations into
the effect of twins and sub-boundary failure will be discussed through the examination of the
brittle fracture of commercial Zircaloy alloy.
822 G. Hughes et al.

Again, for the lower temperature fractures of bcc Fe-3%Si steel, {112} twins in the material
are visible, emergent on {001} cleavage facets, where there is limited cracking along the twin
interface. This is also accompanied by a localised form of accommodation, which has little effect
on overall propagation of the cleavage crack. Again the computer models have been revised to
accommodate the role of {112} twins.
The implication of these results from experiments conducted over a range of temperatures
spanning the brittle to ductile transition for both hcp and bcc materials will be discussed, with
respect to the fracture mechanisms. The links between the geometric modelling and experimental
work will be examined, with a view to explaining how the models have been developed, and
comparisons between modelling predictions and experimental findings will be discussed.
Finally, progress on experimental work and related modelling on the influence of bainitic
substructure, developed in the prior austenite grains of an A508 steel, on the propagation of cracks
will be discussed.
Acknowledgements
The project from which this work is taken involves collaborations with Professors JF Knott
(University Birmingham, UK) and V Randle (University of Wales-Swansea, UK) and is supported
by the EPSRC. Professor Flewitt would like to thank BNFL British Nuclear Group, for allocating
him time to work at the University of Bristol.

References
1. Crocker, A.G., Smith G.E., Flewitt P.E.J. and Moskovic R., Materials Science Forum, 294-
296, 674, 1999
2. Argon A.S. and Qiao Y., Philos. Mag. A, 82A, 3333-3347, 2002
3. Crocker A.G., Flewitt P.E.J.and Smith G.E., Int. Mat. Rev. 50, 99, 2005
4. Hughes G.M., Smith G.E., Crocker A.G. and Flewitt P.E.J., Mats. Sci. and Tech., to be
published, 2005
 8. Modelling of Material Property Data and Fracture Mechanisms 823

THREE-DIMENSIONAL MODELLING OF FRACTURE IN POLYCRYSTALS

Gillian Smith1, Alan Crocker1, Gareth Hughes2 and Peter Flewitt2

1Department of Physics, University of Surrey, Guildford, Surrey GU2 7XH, UK
2Interface Analysis Centre, University of Bristol, Bristol BS2 8BS, UK
1gillian.smith@surrey.ac.uk, 2gareth.hughes@bristol.ac.uk

A great deal of research has been carried out on modelling the fracture of polycrystalline materials.
However, this has been largely restricted to 2-D models and special cases of 3-D models, e.g.
Smith et al. [1], Crocker et al. [2]. Using 2-D models, brittle cleavage cracks in adjacent grains
meet at a point in their common grain boundary. A crack can therefore propagate from grain to
grain without the necessity of any grain boundary failure. In real materials this is not the case.
Cleavage cracks in adjacent grains do not in general meet in a line in their common grain boundary
so that some grain boundary failure or an equivalent accommodation mechanism, such as multiple
cleavage or ductile tearing, must occur. For this and many other reasons it is important to develop
satisfactory 3-D models, Smith et al. [3]. The simplest that have been used are in the form of a
body-centred cubic array of identical, regular tetrakaidecahedra (14-hedra). These have suggested
that of the order of 30% of brittle fracture is in the form of accommodation grain-boundary failure.
This does not include complete grain boundaries that fail because of some inherent weakness. The
figure of 30% is much greater than most reported experimental values and therefore there is a need
for more realistic 3-D models. One approach has been to construct prismatic grains on a random
array of 2-D polygonal cells. This is a good approximation of the structure of columnar grains in
weld metal and again suggests that in brittle fracture about 30% of accommodating grain boundary
failure is needed.[1] However, as the prisms are effectively of infinite height, this figure gradually
increases as fracture propagates.
A general 3-D model has therefore been developed. This has the form of a cube within which
randomly positioned grain nuclei are generated. Grains grow radially from these nuclei until they
meet their neighbours and thus generate randomly-shaped grains with flat polygonal faces.
Information about these grains and their connectivity is held in the computer using parent-child
links, relating grains to faces (F), edges (E) and vertices (V).
This paper will describe applications of an example of this model with 45 grain nuclei. Only 5
of these grains are completely enclosed within the model, which at first was surprising but in fact is
to be expected. The remaining 40 grains meet the surface and of these 29 have one or more faces in
common with the interior grains. Striking features of the grains are the occurrence of many
relatively small faces, several elongated faces, and vertex angles very different from the ideal
equilibrium angle of 109.5o. The interior grains have 13 to 15 faces, 33 to 39 edges and 22 to 26
vertices, the averages being 14.2, 36.6 and 24.4, satisfying Euler's formula, V – E + F = 2. These
values are close to those for the regular 14-hedron: F = 14, E = 36, V = 24. The average number of
edges per face is 5.03 for the model compared with 5.14 for the 14-hedron.
To fracture the model a stress axis is defined and each grain is allocated a crystallographic
orientation and therefore cleavage planes appropriate to the material being investigated. Relative
energies are then assigned for the available fracture mechanisms depending on the temperature. A
crack is nucleated in the most favourably oriented grain and it is assumed that the projection of the
fracture front on the plane perpendicular to the stress axis increases radially with time. When this
crack reaches a grain boundary the weakest path is chosen for propagation. As fracture progresses
824 G. Smith et al.

the amounts of transgranular and intergranular brittle and ductile fracture and their associated
energies are recorded.
When applied to a model representing ferritic steels three orthogonal {001} cleavage planes
are used. The amount of grain boundary failure around the grain in which the initial cleavage crack
nucleates is then about 15% but as fracture propagates this soon rises to 27%, similar to the 30%
found in earlier models. In this case it is assumed that multiple cleavage and ductile failure do not
occur. The model is also being used to investigate the fracture of polycrystalline zinc. It was
anticipated that this would be a relatively simple exercise as it was expected that cleavage only
occurred on the basal (0001) plane. However experimental work on zinc has demonstrated that
cleavage also occurs on prismatic planes and along twin boundaries so that the situation is
complex, Hughes et al. [4].
Currently the surface grains in the above model are being studied and a crack nucleated in one
of the interior grains is being allowed to propagate to the surface of the model. Cracks are also
being nucleated at the surface of the model rather than in the interior. Different weighting factors
are being examined for cleavage, grain boundary and ductile failure and distributions of fracture
energies are being introduced. The alternative accommodation mechanisms of multiple or stepped
cleavage and ductile failure are also being considered. Grain substructures, such as twin
boundaries and bainitic boundaries, are being introduced following experiments that demonstrate
the importance of these features. It is also proposed to use larger models with more complete
interior grains. Selected results from this work will be reported and compared with the associated
experimental research.
Acknowledgements
The project upon which this paper is based, which involves collaboration with Professor John
Knott at the University of Birmingham and Professor Valerie Randle at the University of Wales at
Swansea, is supported by EPSRC. Professor Flewitt would like to thank BNFL British Nuclear
Group for allocating him time to work at Bristol University.

References
1. Smith, G. E., Crocker, A. G., Moskovic, R. and Flewitt, Philos. Mag. A, vol. 82, 3443-53,
2002.
2. Crocker, A. G., Flewitt, P.E.J. and Smith G. E., Int. Mater. Rev., vol. 50(2), 99-124, 2005.
3. Smith, G., Crocker, A. and Flewitt, P., In Proceedings of the 15th European Conference on
Fracture, CD-ROM, KTH. Stockholm, 2004.
4. Smith, G., Hughes, G., Flewitt, P., and Crocker, A., In Proceedings of the 2nd International
Conference on Multiscale Materials Modeling, Naser M. Ghoniem ed., UCLA, 2004, 282-4.
 8. Modelling of Material Property Data and Fracture Mechanisms 825

ANTI-WING CRACK GROWTH FROM SURFACE FLAW IN REAL ROCK

UNDER UNIAXIAL COMPRESSION

Robina H. C. Wong1, Y. S. H. Guo2 , L. Y. Li3, K. T. Chau1 , W. S. Zhu2 and S. C. Li2

1Civil and Structural Engineering Depart., The Hong Kong Polytechnic University, Hong Kong
2Geotechnical & Structural Engineering Research Center, Shandong University, 250061,China
3The China University of Mining and Technology, Beijing, China

cerwong@polyu.edu.hk

According to unofficial estimation, within the 20th century 3.5millions people die in natural
disaster, of which over 40% was by earthquakes. Earthquakes remain one of the worst natural
disasters posing threats to human civilization. For example on 21 September 1999, an earthquake
of magnitude of 7.6 occurred near Chi-Chi Taiwan, causing about 2317 death, 8722 injured, and
51761 buildings collapses [1]. A surface rupture, which is more than 100 km long is associated
with the earthquake. The first 90 km surface rupture running roughly followed the known
Chelungpu fault in north-south direction, then the surface rupture turned east creating a new
fracture zone going through the Shihkang Dam, at where waterfall was formed from a flat riverbed,
and the maximum surface displacement of the surface rupture was about 7 m in both vertical and
horizontal directions. This newly formed fault arouse much international attention regarding its
formation mechanism and many studies are still ongoing. The fault rupture mechanism is still not
fully understood. The rupture mechanisms from pre-existing 3-D fault is worth to investigate for it
has both academic merit and of societal need.
A series of experiments have been conducted on 3-D opened surface fault (or flaw) in both
artificial (PMMA) and real rocks (marble, gabbro and sandstone) under uniaxial compression [2-
3]. It was found that the mechanisms of fracture propagation in homogeneous (PMMA) material
may not be the same as in inhomogeneous materials (real rock). Under the uniaxial compression,
wing crack (tensile crack) and petal crack are very common observed in the PMMA specimen (Fig.
1a) and in some of marble specimens (Fig. 1b). However it was observed in other real rock
specimens, tensile cracks induced along the opposite direction of the wing crack first (Fig. 1c to
1e). This is the so called anti-wing crack while the wing crack or compressive crack induced at the
later stage (Fig. 1c & 1d). This phenomenon has not been reported before. According to the stress
field around the flaw tip on the homogeneous material [4-5], the maximum tangential stress VTmax
is near the position of T S/2(90q). The definitions of I and T are given in Fig. 2. It is further
supported by the observation form the PMMA specimen where the wing crack propagates upward
while the lower one propagates downward. However, it is not clear why anti-wing tensile cracks
induced in most of the real rocks first. According to the acoustic emission (AE) measurements (see
Fig. 1e), the AE signals supported the observation from the real rock specimen where anti-wing
crack would induce first. This interest phenomenon is worth to further investigate.
826 R. H. C. Wong et al.

FIGURE 1 Crack growth form flaw tip (1a to 1e). Acoustic essmision (AE) sensors were mounted
on the surface of specimen to detact the AE signal (1e). The AE signle induced around the flaw
area (1f). The surface of real rock specimens is painted to produce a big contrast for crack
observation.

FIGURE 2 The stress field around a partially circular crack as a function of orientation I and
location angle T. The curves shows the maximum normalized stress VT/Vcr various angle T at
different orientation I

Acknowledgements: The study was supported by the Hong Kong Polytechnic University (A-
PG46) and the National Natural Science Foundation of China to Zhu-WS (40272120, 50229901).

Reference
5. Lin, C.W., Lee, Y.L., Huang, M.L. Lai, W.C., Yuan , B.D. and Huang, C.Y., Eng. Geology,
71, 13-30, 2003
6. Wong, R.H.C., Huang, M.L., Jiao, M.R., Tang, C.A. and Zhu, W., Key Eng. Materials, 261-
263, 219-224, 2004
7. Wong, R.H.C., Law, C.M., Chau, K.T. and Zhu, W.S., Int. J. of Rock Mech. & Min. Sci., 41(3),
360, 2004.
8. Murakami, Y. and Natsume, H., JSME Int. J. A-Solid Mech., 45(2), 161-169, 2002
9. Chau, K.T., Wong, R.H.C., Wong, Y.L., Lai, K.W., Third Int. Conf. Continental Earthquake,
Beijing, China, July 12-14, 60, 2004
 8. Modelling of Material Property Data and Fracture Mechanisms 827

MECHANICAL BEHAVIOR MODELING IN THE PRESENCE OF STRAIN

AGING

J. Belotteau1,3, C. Berdin1, S. Forest2, A. Parrot3 and C. Prioul1

1MSSMat, Ecole Centrale Paris, Grande voie des Vignes, 92295 Châtenay Malabry Cedex, France.
2Centre des Materiaux P.M. Fourt, Ecole des Mines de Paris, BP87, 91003 Evry Cedex, France.
3EDF R&D/MMC, Site des Renardieres, 77818 Moret s/ Loing Cedex, France.

belotteau@mssmat.ecp.fr, berdin@mssmat.ecp.fr, samuel.forest@ensmp.fr, aurore.parrot@edf.fr,

prioul@mssmat.ecp.fr

In many fields some structural materials are subjected to strain aging, which gives rise to
inhomogeneous yielding such as Piobert–Lüders’ bands and Portevin – Le Châtelier instabilities.
These phenomena occur in steels containing interstitial elements in solid solution such as carbon or
nitrogen which segregate to dislocations during aging and pin them. Strain aging in ferritic steels
induces a loss in ductility and fracture toughness at mild temperatures [1]. To study the influence
of strain aging on ductile tearing, it is possible to simulate plastic strain instabilities using a local
constitutive equation [2] taking into account the interaction between solute atoms and dislocations
responsible for strain aging [3]. This study first requires the identification of the model parameters,
which needs to correlate the experimental values obtained from tensile tests to the intrinsic
behavior of the material and then requires a good interpretation of the upper and lower yield
stresses in the presence of static strain aging. So we have performed a numerical study of the
Lüders’ plateau featuring the influence of several parameters such as meshing, boundary
conditions or behavior law on the nucleation and growth of Lüders’ bands. The influence of the
local band development on the plateau has been investigated. Correlations between the plateau
stress level and the local behavior will be discussed.
As proposed by Tsukahara and Iung and for the sake of simplicity the aging process is
introduced in the local constitutive equation by an overstress necessary to the unlocking
mechanism [4]: at yield point (V0) the stress drops down to V1 and then follows a linear isotropic
strain hardening behavior, as shown in Fig. 1. During the simulation of a tensile test, the specimen
exhibits a strain localization band oriented around 50° from the tensile direction, growing from an
“artificial” defect until filling the whole specimen. The band spread is related to a plateau at nearly
constant stress on the V–H macroscopic curve.
In a first part we have investigated the shape of the Lüders’ peak and plateau. Correlations can
be made between the stress variations and the band spreading along the specimen (Fig. 2). Several
structure effects are examined: boundary conditions, type and location of the initial defect,
meshing.
In a second part the notions of upper and lower yield stresses, usually used in industry are
questioned. A correlation has been checked between the lower yield stress, i.e. the plateau stress
level, and the intrinsic parameters of the local behavior law V0 and V1. Different forms of
constitutive equations have been tested. The respective effect of structure and local material
behavior on the stress-strain curve can then be clearly separated. This offers the possibility of
further modeling of ductile tearing in the presence of strain aging.
828 J. Belotteau et al.

FIGURE 1. Local behavior to simulate Lüders’ bands. Introducing local softening allows strain to
localize.

FIGURE 2. Simulation of Lüders’ bands: relation between plateau singularities and the
progression of the strain band. Dependence of lower yield stress on prescribed load conditions.

References
1. D. Wagner, J.C. Moreno, C. Prioul, J. M. Frund and B. Houssin, J. Nucl. Mater., vol. 300
Issue 2-3, 178-191, 2002.
2. S. Graff, S. Forest, J. -L. Strudel, C. Prioul, P. Pilvin and J. -L. Béchade, Mater. Sci. Eng A.,
vol. 387-389, 181-185, 2004.
3. Y. Estrin and L.P. Kubin, J. Phys. III, vol. 1, 929-943, 1991.
4. H. Tsukahara and T. Iung , Mater. Sci. Eng. A, vol. 248, 304-308, 1998.
 8. Modelling of Material Property Data and Fracture Mechanisms 829

ON THE LOCAL CONDITIONS FOR CLEAVAGE INITIATION IN FERRITIC

STEELS

Jorg Hohe, Valerie Friedmann and Dieter Siegele

Fraunhofer Institut fur Werkstoffmechanik
Wohlerstr. 11, 79108 Freiburg, Germany
Joerg.hohe@iwm.fraunhofer.de
Valerie.friedmann@iwm.fraunhofer.de
Dieter.siegele@iwm.fraunhofer.de

The present study is concerned with an investigation of the local conditions for initiation of
cleavage fracture in ferritic steels which have been analyzed in the context of a research program
directed to different methods of cleavage fracture assessment (Hohe et al. [Swindeman, R.W. et al.,
Pressure Vessels and Piping, 507-512, 2004.]). The investigations are performed experimentally
and numerically using a German nuclear grade 22NiMoCr3-7 (A 508 class 2) pressure vessel steel
as a material example under forged conditions. The heat treatment of the material consists of an
austenization at 900°C, quenching in water, tempering at 650°C for 7.5 hours and subsequent air
cooling.
Following the basic characterization of the material, a variety of fracture mechanics specimens
is tested. In the experimental investigation, specimens with different geometries including SE(B)-,
C(T)- and CC(T)-specimens, different thicknesses ranging from 10 mm to 50 mm and different
crack depths ranging from a/w = 0.13 to a/w = 0.52 are tested. The test temperatures are ranging
from -120°C to 0°C. The tests are performed and evaluated according to ASTM standard E 1921
[ASTM E 1921, Standard Test Method for Determination of the Reference Temperature T0 in the
Transition Range, ASTM, West Conshohocken, PA, 2002.] or similar, where the standard is not
applicable. In this context, significant constraint effects are observed, resulting in a strong
specimen geometry dependence of the measured fracture toughness KJc. In order to assess this
effect, the local mechanical conditions at the cleavage initiation sites are investigated.
To determine the fracture mechanism, all specimens tested are analyzed by fractographic
methods. It is observed that all specimens fail by pure transgranular cleavage. Nevertheless, at
higher test temperatures (T -30°C), some ductile tearing precedes the cleavage failure. The
cleavage triggering sites in all cases are situated in the interior of the ligament at a distance of
approximately 20 to 170 m (depending on the load level at failure) ahead of the crack front. Larger
distances are found at higher test temperatures due to the higher failure loads.
Subsequently, the fracture mechanics tests are simulated numerically using the finite element
method. Full three-dimensional, geometrically nonlinear analyses with extremely fine meshes are
performed for an accurate determination of the local stress and strain fields in the cleavage fracture
process zone directly ahead of the crack front. Special interest is directed to the maximum
principal stress I which is assumed to control the propagation of a freshly initiated micro-crack
(Ritchie et al. [])Ritchie, R.D., Knott, J.F. and Rice, J.R., J. Mech. Phys. Solids, vol. 21 (1973)
395.. At the cleavage origin, approximately constant values of this quantity are found
irrespectively of the specimen geometry and the test temperature and thus the failure load level
(see Fig. 1). Nevertheless, a deterministic criterion based on the level of the maximum principal
stress I alone does not form a sufficient criterion for cleavage initiation since the critical value f is
already reached for rather small load levels at some positions rather close to the crack front.
As an extended concept, Chen et al. [Chen, J.H., Wang, Q., Wang, G.Z. and Li, Z., Acta. Mat.,
vol. 51, 1841-1855, 2003.] have proposed an alternative three-parameter criterion. This criterion
830 J. Hohe et al.

takes the equivalent plastic strain e as well as the stress triaxiality coefficient e/kk into account. The
equivalent plastic strain is assumed to control the nucleation of micro-cracks which might serve as
cleavage initiation sites whereas the stress triaxiality coefficient controls the possibility of blunting
and thus the criticality of freshly nucleated micro-cracks for triggering of cleavage fracture. For the
specimens in the present experimental database, a large scatter of both quantities at the cleavage
origin is observed when the failure load is reached. On the other hand, it can be shown that the
levels of both quantities at cleavage initiation are coupled rather than being independent. Based on
these results, a framework for the assessment of cleavage fracture in ferritic steels based on the
local mechanical fields is proposed.

FIGURE 1. Local cleavage stress for different specimen geometries and failure loads.

References
1. ASTM E 1921, Standard Test Method for Determination of the Reference Temperature T0 in
the Transition Range, ASTM, West Conshohocken, PA, 2002.
2. Chen, J.H., Wang, Q., Wang, G.Z. and Li, Z., Acta. Mat., vol. 51, 1841-1855, 2003.
3. Hohe, J., Tanguy, B., Friedmann, V., Stöckl, H., Böhme, W., Vafolomeyeva, V., Hebel, J.,
Burdack, M., Fehrenbach, C., Schüler, J., Sguaizer, Y., Siegele, D., Kritische Überprüfung
des Mastercurve-Ansatzes im Hinblick auf die Anwendung bei deutschen Kernkraftwerken,
Fraunhofer Institut für Werkstoffmechanik, Report No. S8/2004, 2005.
4. Ritchie, R.D., Knott, J.F. and Rice, J.R., J. Mech. Phys. Solids, vol. 21 (1973) 395.
 8. Modelling of Material Property Data and Fracture Mechanisms 831

UNIFIED CONSTITUTIVE EQUATIONS TO DESCRIBE ELASTOPLASTIC

AND DAMAGE BEHAVIOR OF X100 PIPELINE STEEL

T. T. Luu1,2, B. Tanguy1, J. Besson1, A. Pineau1 and G. Perrin2

1Ecole des Mines de Paris, Centre des Materiaux, Evry Cedex, France
2Applied Mechanics Division, Institut Francais du Petrole, Rueil-Malmaison, France

luu@mat.ensmp.fr, btanguy@mat.ensmp.fr

Economic gas transportation over long distances requires the use of high pressure and
consequently the use of high strength steels such as grade API X100 which is investigated in this
study. This steel is produced using Thermo-Mechanical Controlled Process (TMCP) as thick
plates which can be rolled into large diameter tubes. The resistance to ductile crack propagation of
these new steels must be determined to access the integrity of pipelines. This is done using Charpy
tests, for which the upper shelf energy (USE) is determined, or drop weight tear tests. The purpose
of this study is to establish unified constitutive equations able to represent the plastic and rupture
behaviour of plates and tubes ; the model must account for plastic anisotropy (induced by the
production process) and for the pre-strain resulting from rolling plates into tubes (about 2% [1]).
To illustrate the pre-strain effect [2], load-displacement curves issued from instrumented Charpy
tests for the plate and the tube for two configurations (L-T and T-L) are reported on Fig. 1 where
the value of corresponding USE are given. The maximum load is obtained for the tube (Fig. 1a T-
L direction) due to pre-strain. However, for the T-L solicitation a sharp drop of the load in the case
of the tube is observed once the ductile crack has initiated. This results in a strong decrease of the
USE for the tube.

FIGURE 1. Load-displacement curves for the dynamic Charpy test at 20°C.

(a) L-T direction, (b) T-L direction

Material behavior was investigated using several specimen geometries: smooth tensile,
notched tensile (NT) with different radii, compact tension. Oligocyclic tests were used to access
isotropic and kinematic hardening contributions to the overall plastic behavior. Tensile tests were
conducted along long (L), transverse (T) and short (S) directions to characterize the anisotropy of
the material. Tensile properties (yield strength (YS), ultimate yield strength (UTS) and uniformed
elongation (UE) for plate and tube along three directions are reported on table 1.
832 T. T. Luu et al.

TABLE 1. X100 steel plate and tube mechanical properties

The strategy to describe the elastoplastic and damage behavior was the same than the one used
in [3]. The Gurson ductile damage model was used to describe the initiation and the propagation of
ductile damage; the anisotropic elastoplastic behavior was described using the model proposed in
[4]. For plastic strain less than UE, the parameters of the elastoplastic model were identified on
smooth tensile tests for the three directions (L, T, S). For larger plastic strains, inverse
identification using notched tensile tests was used. Comparisons between experiments, the
anisotropic model and von Mises criterion are reported on Fig. 2 for the plate. Moreover,
considering a pre-strain of 2%, the constitutive equations well describe the behavior of notched
specimens cut off from the tube. The parameters of the Gurson model were identified using the
notched tensile tests and used to simulate Charpy and compact tension tests.

FIGURE 2. Normalized load (F/S0) vs radial displacement (')s/) along the short direction in
the notch. Experiments (symbols), von Mises simulations (dashed lines) and simulations with the
anisotropic model (solid lines)

References
1. Qui, H., Enoki, M., Hiraoka, K., Kishi, T., Eng. Frac. Mech., Vol. 72, 1624-1633, 2005.
2. Hagiwara, N., Masuda, T., Oguchi, N., J. of Pressure Vessel Tech., Vol. 123, 355-361, 2001.
3. Rivalin, F., Besson, J., Di Fant, M., Pineau, A., Eng. Frac. Mech., Vol. 68, 347-364, 2001
4. Bron, F., Besson, J., International Journal of Plasticity, Vol. 20, 937-963, 2004
 8. Modelling of Material Property Data and Fracture Mechanisms 833

ESTIMATION OF LOWER BOUND ENGINEERING FRACTURE TOUGHNESS

IN THE DUCTILE TO BRITTLE TRANSITION REGIME

R. Moskovic and R. A. Ainsworth

British Nuclear Group, Berkeley, Gloucestershire GL 13 9PB, United Kingdom.
British Energy, Barnett Way, Barnwood, Gloucester GL4 3RS, United Kingdom.
Robert.Moskovic@magnox.co.uk, bob.ainsworth@british-energy.com

Ferritic steels are used for the manufacture of components that form pressure boundaries of nuclear
power plant. The mechanism of fracture of ferritic steels changes with increasing temperature and
reducing specimen thickness from brittle, cleavage fracture to ductile. In order to assess the
structural integrity of a component, fracture toughness properties are measured in a laboratory on
specimens that are smaller than the component.
In order to use fracture toughness results measured on small specimens in fracture mechanics
assessments of plant components, it is necessary to apply thickness or size correction to cleavage
fracture toughness values. These thickness corrections can be derived either from statistical
analyses of cleavage fracture toughness databases or by adopting a framework based on the Master
Curve. In this approach, the cleavage fracture toughness, K25, of a reference thickness B = 25mm
is related to fracture toughness, KB, for a thickness B by an equation:

K25 = Kmin + (KB – Kmin)(B/25)1/4 (1)

where Kmin is a minimum toughness, usually taken as 20MPam. This equation implies that if
section thickness of a component to be assessed is significantly greater than the thickness of the
specimens used to measure the cleavage fracture toughness then the resulting estimates of the
lower bound values will be substantially lower.
In the Master Curve approach, the thickness correction is the same for all ferritic steels
independently of the microstructure and tensile properties of the steel. However, where cleavage
fracture toughness data are available for several different sizes of specimen over a range of
temperatures, it is possible to use statistical techniques to determine the specimen size dependence
of cleavage fracture toughness. This paper illustrates application of this approach to the Euro
fracture toughness database obtained from tests performed on quenched and tempered pressure
vessel steel DIN 22NiMoCr37. There are data sets consisting entirely of cleavage initiation
fracture toughness values for several specimen sizes at three different temperatures: -154ºC, -91ºC
and -60ºC. In addition, there is a data set of cleavage initiation fracture toughness values for a
single specimen size at –110ºC. These data were analysed in two different ways. First, the data for
each test temperature were analysed to estimate the thickness dependence of cleavage fracture
toughness for specific temperatures. The random scatter in the data was modelled by a Weibull
probability distribution and the Weibull shape parameter at each temperature was assumed to be
the same for all specimen sizes. The analyses showed that the Weibull shape parameter varied
with temperature in the range from 3.08 to 5.85. The thickness dependence of cleavage fracture
toughness also varied with temperature. At two temperatures, -154ºC and -60ºC, the coefficient
for thickness dependence was of the order of 1/15. At –60ºC, this value was approximately 1/6. A
further analysis at –60ºC in which the Weibull shape parameter was allowed to vary with specimen
size yielded a different shape parameter for each specimen thickness and the coefficient for
thickness dependence was approximately 1/33.
834 R. Moskovic and R. A. Ainsworth

The statistical analyses have shown that the specimen thickness correction may be smaller than
assumed in the Master Curve. In this approach, the effect of thickness on cleavage fracture
toughness is rationalised by the weakest link concept. With an increase in the thickness and the
associated crack length there is an increasing likelihood of sampling the weakest microstructure.
Hence, it should be possible to sample the weakest link either by testing a thick specimen or a set
of small specimens with the same aggregate crack length as the thick specimen. In the EURO
fracture toughness database, there are at least 30 values of fracture toughness for each specimen
thickness at each temperature. These data can also be used to assess the effect of specimen
thickness on the lower bound values of cleavage fracture toughness. For the purpose of these
analyses, it was assumed that the lowest eleven values, comprising approximately 10% of the data,
are in the lower tail of the distribution. In these analyses, the data were re-arranged as follows:
• The data obtained at the same temperature for different specimen thicknesses were
arranged in ascending order and the lowest eleven values were extracted.
• Data obtained at the same temperature for different specimen thicknesses were statistically
analysed as a function of thickness. Standardised residuals were computed and arranged in
ascending order and the lowest eleven values were extracted.
• Values of cleavage fracture toughness for different specimen thicknesses were adjusted
using the size correction K25 = KB(B/25)¼. The calculated values of K25 were arranged in
ascending order and the lowest eleven values were extracted.
Percentages of values in the lower tails of the distribution for different specimen thicknesses
were compared. At –154ºC and –91ºC there was a weak trend in percentage to increase with
specimen thickness consistent with a small effect of specimen thickness on the lower bound
cleavage fracture toughness. At -60°C, there was no trend.
In addition to the above statistical analyses, the cleavage fracture toughness values were
examined graphically. The values for each specimen thickness were ordered, ranked and
cumulative values of failure probability for each data point calculated. The cumulative failure
probabilities were plotted against measured values of J. Examination of the plots showed that the
curves for different specimen thicknesses converged below approximately 0.1 cumulative
probability consistent with no thickness effect on cleavage fracture toughness in the tails of the
distribution. At higher cumulative probability values, the curves diverged with increasing J
consistent with specimen thickness effect on cleavage fracture toughness in the main body of the
data.
 8. Modelling of Material Property Data and Fracture Mechanisms 835

CLEAVAGE FRACTURE MICROMECHANISMS RELATED TO WPS EFFECT

IN RPV STEEL

S. R. Bordet1, B. Tanguy1, S. Bugat2, D. Moinereau2 and A. Pineau1

1Centre des Materiaux, Ecole des Mines de Paris, UMR CNRS 7633,

BP 87, 91003 Evry Cedex, France

2EDF R&D, Dept. MMC, Les Renardieres, 77818 Moret-sur-Loing Cedex

S.R.Bordet.98@cantab.net, benoit.tanguy@ensmp.fr, stephane.bugat@edf.fr,

dominique.moinereau@edf.fr, andre.pineau@ensmp.fr

Since the first investigations four decades ago, a large number of experiments on ferritic steels has
confirmed the existence of a warm pre-stress (WPS) effect, which describes the effective
enhancement of the cleavage fracture toughness at low temperature following the application, at a
higher temperature, of a stress intensity factor (SIF) which exceeds the fracture toughness of the
virgin material at low temperature. These experiments allowed for the establishment of the so-
called ‘conservative principle’, which states that no fracture will occur if the applied SIF decreases
(or is held constant) while the temperature at the crack-tip decreases, even if the fracture toughness
of the virgin material is exceeded. In structural integrity assessments involving a prior overload or
a thermal transient, such as that of a nuclear pressure vessel subjected to a pressurized thermal
shock (PTS) consecutive to a loss of coolant accident (LOCA), such a principle is of great
importance in the evaluation of the safety margins. Three main reasons have been advanced to
explain the WPS effect: the blunting of the crack tip at high temperature, the formation of high
compressive stresses on elastic unloading, and a change in the cleavage fracture micromechanisms
induced by plastic deformation. While all these factors certainly contribute to the effective
toughness enhancement following WPS, their relative incidence on the fracture risk is not easily
established. In this paper, we choose to mainly focus on the cleavage fracture micromechanisms
following WPS, as a first step towards better quantifying the individual contributions of crack tip
blunting and residual stresses.
This paper presents different WPS fracture test results obtained on 18MND5 (A533B) RPV
steel, supplemented by isothermal tests on the virgin material for comparison:
• toughness tests in compact tensile (CT) geometry following different thermo-mechanical
cycles between 20 and –150°C: LCF (Load-Cool-Fracture), LUCF (Load-Unload-Cool-
Fracture) and LCIKF (Load-Cool-Increasing applied K- Fracture).
• notched tensile (NT) tests (notch radius = 2.4 mm) plastically deformed in traction,
compression, and traction followed by compression, at room temperature, and then broken
in liquid nitrogen.
Extensive fractographic investigations using electron microscopy revealed two competing
particles for triggering macroscopic cleavage:
• titanium nitride (TiN) particles, often bound or clustered with other particles such as
manganese sulfides (MnS) and oxides (CaO, MgO, Al2O3 etc.).
• grain-boundary carbides.
In the fracture toughness tests, the particles responsible for global cleavage fracture are almost
exclusively TiN particles, whether the specimen was pre-stressed or not. This is also the case for
the notched tensile tests on the virgin material at liquid nitrogen temperature. Significant changes,
however, were observed in the pre-deformed NT specimens. While the TiN particles are still the
836 S. R. Bordet et al.

main critical particles in the NT specimens pre-deformed in compression alone, these particles are
replaced by carbides in specimens tested in traction-compression or traction alone. More
specifically, for specimens pre-strained above 20% deformation in traction, global cleavage
initiates at grain boundaries (the presence of carbides is not easily observed) in vicinity of holes
formed around cracked (TiN, oxides) or detached (MnS) particles, which serve as stress
concentrators. A material aspect of the WPS effect in 18MND5 steel is therefore the deactivation
of brittle TiN particles at high temperature: matrix straining under low stress enables breaking the
particle without causing unstable propagation. Because the yield stress increases as the
temperature is decreased, there exists, for a given geometry, a transition temperature below which
the breaking of a sufficiently large TiN particle will systematically trigger global instability. This
assertion is confirmed by tensile tests performed on the virgin material at –150°C: arrested ferrite
microcracks, originating from broken TiN particles and a few grains long, were observed on the
specimens. All arrested cracks were surrounded by a region of extensive ductile crack growth,
which is indicative of crack formation at an early stage, characterised by low stress level.
The experimental observation that the TiN particles are still the main initiators for global
cleavage in the CT geometry, despite relatively large pre-loads, is explained by the fact that the
large strain region remains very localized at the crack tip. At a blunted, yet highly constrained,
crack tip, the peak stress occurs at some distance from the crack tip (about twice the CTOD), and
keeps approximately constant while moving away from the crack tip under increasing load.
Reloading at low temperature not only globally raises the stress level, which favours cleavage
propagation, but also, for a sufficiently large applied SIF, causes the peak stress to sample material
little affected by the WPS at high temperature, hence the continued participation of the TiN
particles in cleavage initiation. A material effect of WPS is therefore also present in the CT
geometry, through a more or less complete exhaustion of the potentiality to nucleate cleavage
microcracks (i.e. through TiN particle cracking) in the WPS affected region.
These findings are useful for understanding some of the difficulties with cleavage toughness
modelling, notably in the ductile-to-brittle (DBT) transition, of steels of similar composition.
Recent attempts to model the transition using local approach (Beremin type) cleavage models had
to allow for an apparent temperature dependence of the parameter Vu, to take into account a
possible change of micromechanisms. For 18MND5 steel, this apparent temperature dependence
can be attributed to a modification of the cleavage initiation micromechanisms, from cleavage
initiated by TiN particle breaking at low temperature to cleavage started at grain boundary carbides
in the transition, possibly helped by ductile damage (holes around inclusions).
 8. Modelling of Material Property Data and Fracture Mechanisms 837

MODELLING OF FATIGUE DAMAGE IN ALUMINUM CYLINDER HEADS

R. Salapete1,2, B. Barlas3, E. Nicouleau2, D. Massinon3, G. Cailletaud1 and A. Pineau1

1Centre des Materiaux, Ecole des Mines de Paris, UMR CNRS 7633, B.P.87, F-91003 Evry cedex
2RENAULT, 67 Rue des Bons Raisins, 92508, Rueil Malmaison, France
3Fonderie Montupet, 67 rue J. de la Fontaine, 60181 Nogent-sur-Oise, France

raphael.salapete@ensmp.fr

Because of the regular increase in specific power, automotive cylinder heads, in particular those
used in Diesel direct injection engines, are exposed to increasing thermomechanical loadings.
Those cylinder heads are usually made of cast aluminum alloys. Cyclic loads corresponding to the
alternative heating and cooling of the operative engine generate fatigue cracks, initiating from the
fire deck (the inter valve bridge region) which can lead ultimately to engine failure (Fig. 1).
Beyond a limit level of crack propagation, some essential functions of the cylinder heads, in
particular the separation between the different fluids circuits (intake, exhaust, water cooling, and
oil) are no longer fulfilled.
The failure assessment in the design stage of an engine is therefore of crucial importance. In
the present study, a failure assessment method for cylinder heads made of A319 type aluminum
alloy is presented. The method includes three stages: structural analysis using an
elastoviscoplasticity model, crack initiation model and, finally, crack propagation assessment.
Elastoviscoplasticity model
A cyclic viscoplastic constitutive equation including kinematic and isotropic hardening laws
and taking also into account the progressive ageing of the A319 type aluminum alloy, is used [1-3].
Structural computations are made thanks to finite element codes, Zebulon or Abaqus / Zebulon
interface (Zmat) in order to represent the cyclic behavior of the material during thermomechanical
loadings.
Fatigue crack initiation model
A fatigue crack initiation model based on continuum damage mechanics was developed. This
model was established from a large number of isothermal and anisothermal tests carried out on
fully machined specimens. The formulation uses a Wohler type curve and accounts for ageing and
temperature effects. It is implemented as a post-processor of the Zebulon code and can be used on
an Abaqus data base. It provides a fairly reasonable estimation for the appearance of 0.1 –1mm
class cracks.
Fatigue crack propagation model
The prediction of the life spent in crack propagation from 0.1 –1mm defect up to failure is
derived from fracture mechanics concepts. A modified stress intensity factor is introduced to
account for plasticity on fatigue crack propagation [4]. Isothermal fatigue crack growth tests were
performed on specimens containing an artificial short crack (0.150 mm < a < 1 mm). A “master”
curve giving the variation of crack growth rate, da dN , with the amplitude of the effective stress

intensity factor 'K eff was thus derived. This curve includes the temperature and crack closure
effects.
A weight function method was used to calculate the stress intensity factor corresponding to
corner cracks initiated either from the ports counter borings or to cracks crossing the region
838 R. Salapete et al.

between them, as observed in Fig. 1. The number of cycles corresponding to crack propagation
was determined by integrating the “master” curve.
Having in hand such an approach, the life prediction can now be obtained as the result of
initiation and propagation of a crack. This last point is crucial, since one can discriminate between
dangerous propagating cracks, and cracks which will stop before gaz from combustion reaches the
water jacket. Several significant examples are shown in the paper.

FIGURE 1 : inter valve bridge failure

References
1. Nicouleau-Bourles, E., In Etude expérimentale et numérique du vieillissement d’un alliage
d’aluminium. Application aux culasses automobiles, PhD thesis, Ecole des Mines de Paris –
(1999).
2. Barlas, B., In Etude du comportement et de l’endommagement en fatigue d’alliages
d’aluminium de fonderie, PhD thesis, Ecole des Mines de Paris – (2004).
3. Salapete, R., Barlas, B., Nicouleau, E., Massinon, D., Cailletaud, G., Pineau, A., In Fatigue
life predictions of aluminum cylinder heads. Fatigue Design, Senlis, November 2005, to
appear
4. Haigh, J.R., Skelton R.P., A strain intensity study of type 304 and 316 stainless steels.
Fatigue crack growth and failure. Materials and Science Engineering, Vol 36, pp 133-137
(1978).
 8. Modelling of Material Property Data and Fracture Mechanisms 839

LOCAL APPROACH TO HIGH TEMPERATURE DUCTILITY MODELING IN

6XXX ALUMINIUM ALLOYS

D. Lassance, D. Fabregue, F. Delannay and T. Pardoen

Universite catholique de louvain, Materials Science and Processes Department
Unité IMAP, Reaumur, Place Sainte Barbe 2, B1348 Louvain-La-Neuve, Belgium
Tel : +32 10 472417, Fax : +32 10 474028, pardoen@imap.ucl.ac.be

Low and medium strength AlMgSi alloys are commonly used for hot extrusion. For high
productivity it is necessary to obtain the alloy in a soft condition so that the extrusion rate can be
increased without overloading the press. Such AlMgSi alloys can be considered as two phase
alloys in which one phase is the Al matrix and the second one is constituted by different particles
among which the Fe rich intermetallics. The second phase particles may have important
consequences on the deformation process, because not only they affect the flow stress of the
material, but also they are likely to influence the ductility and change the final microstructure. In
the case of commercial aluminium extrusion alloys, a homogenisation heat treatment of the as-cast
material is required to improve ductility and enable efficient extrusion. The optimisation of such
extrusion processes depends on a thorough understanding of the homogenisation kinetics and, in
particular, of the morphological evolution of the intermetallic inclusions during the
homogenisation process. Moreover damage initiation occurs in such alloys by decohesion or
fracture of inclusions as referred by Toda and Kobayashi [1]. By performing a long
homogenisation heat treatment at a sufficiently high temperature, during which the -Al5FeSi phase
transforms to the more rounded meta-stable cubic -Al12(FeMn)3Si phase or equilibrium hexagonal
phase, this unfavourable effect can be improved.
The aim of this work is to improve the understanding and the control of the damage resistance
during hot extrusion. To achieve this goal, the link between hot deformation behaviour and
microstructural evolution during the homogenisation treatment of the 6xxx aluminum alloys serie
is studied and compared with the behavior at room temperature. Experiments show that during
deformation, different void populations are nucleated by different particle groups. The first type of
void come from the AlFeSi particle/matrix decohesion, the second one corresponds to the
AlFeMnSi particle/matrix decohesion whereas the third one corresponds to pennyshape voids
initiated by the multiple fractures of elongated AlFeSi particles, which accelerates the damage
progression. In the case where two (or more) populations of voids are present, the strain at the
onset of coalescence (and thus the ductility) tends to decrease. Such an influence of multiple void
populations on the deformation process has already been observed experimentally [2].
In order to take this into account and to relate quantitatively microstructure and ductility, FE
simulations based on an enhanced micromechanics-based model are used. The model used is a
Gurson type constitutive model considering several populations of cavities. The model consists in
using a Gologanu type model for the matrix including the first population. Then the local strains
close to the cavity are extracted from the results of the solved equations. These strains are used to
solve the Gurson equation locally giving the evolution of the second population of voids close to
the first cavity . Using a modified Thomason model, one can predict the onset of coalescence [3].
This model is well suited for the present case as it incorporates void shape effects which is key to
capture the impact of the / transformation on ductility.
The results of ductility calculations show that the model correlate correctly the experimental
values at room temperature and high temperature. The evolution of ductility with the AlFeMnSi
particles content evolution is properly predicted.
840 D. Lassance et al.

References
1. Toda, H., and Kobayashi, H., Materials Science Forum, vol. 426-432, 393-398, 2003.
2. Marini, B., Mudry, F, Pineau, A., Engng. Fracture Mech., vol.22, 989-996, 1985.
3. Fabregue, D., and Pardoen, T., to be published
 9. Micromechanisms in Fracture and Fatigue 841

SMALL FATIGUE CRACK GROWTH IN STEEL -COMPRESSOR DISKS OF

AIRCRAFT ENGINES.

A. A. Shanyavskiy and A. Yu. Potapenko

State Centre for Safety Flight of Civil Aviation, Moscow, Russia
124340, Moscow, Airport Sheremetievo, P.O. Box 54, SCSFCA
shana@flysafety.msk.ru

It was the problem for in-service fatigue crack growth from the steel-disk base of the high-pressure
engine compressor. The fractographic analysis have been performed to examine the crack growth
mechanism and to reproduce crack growth period in term of number of engine flights.
It was shown that the crack growth performs in low-cycle-fatigue area because of cascade of
the interacting small cracks, Fig.1.

FIGURE 1. Fatigue fracture surface on the in-service fatigued steel- disk.

The statistic analysis of in-service fatigue cracks detection in disks in time, simulation of the
crack growth in disks on the basis of the well known kinetic curve for the disk-material have
shown that the crack growth period on the distance 4.0mm can be performed in service without
disk in-flight failure, Fig.2.
842 A. A. Shanyavskiy and A. Yu. Potapenko

FIGURE 2. Fatigue striation spacing, G, against crack growth depth, a, for (1) tested specimens, (2)
reproduced from the striation spacing measurements in-service fatigued disks and modeled as
striation spacing dependence on the 'K2. Results are shown for two critical crack length 4.0 mm
and 4.5 mm.

The inspection interval for disks non-destructive testing was recommended. The new disk
design for stress level diminishing has been discussed.
 9. Micromechanisms in Fracture and Fatigue 843

MICROMECHANISMS OF DAMAGE IN MULTIAXIAL FATIGUE OF AN

AUSTENITIC-FERRITIC STAINLESS STEEL

Ahmed El Bartali, Véronique Aubin, Suzanne Degallaix and Laurent Sabatier

Laboratoire de Mécanique de Lille (UMR CNRS 8107)
Ecole Centrale de Lille
59651 Villeneuve d’Ascq Cedex, France
Ahmed.Elbartali@ec-lille.fr

Austenitic-ferritic stainless steels, also called duplex stainless steels, are designed to combine the
properties of both ferritic and austenitic phases in order to optimize mechanical strength, ductility
and corrosion resistance. They are therefore used for components submitted to corrosive
environments as for instance in the chemical, petrochemical or nuclear industry or for the transport
of chemical products.
Increasing safety requirements together with tighter economic restrictions lead to harder
industrial requirements towards significant weight reduction by simultaneous improvement of the
mechanical properties. In order to achieve this goal, a thorough understanding of the mechanical
behaviour of these materials under complex mechanical loading is necessary. This knowledge
bases upon understanding of the strain physical mechanisms at a microstructural level.
Specifically, the identification of constitutive laws under complex cyclic loadings requires a
multiaxial fatigue experimental data base. Low cycle fatigue tests with the continuous observation
of the surface phenomena and the quantitative analysis of the strain fields allow identifying
physical mechanisms of strain distribution, crack initiation and damage evolution in fatigue.
One purpose of the present work is to complete the data base of fatigue results available in our
laboratory (Aubin [1]). Tension-compression fatigue tests were carried out using cylindrical
specimens and tension-compression/torsion fatigue tests were performed with tubular specimens.
All the specimens were obtained from the same rod of a X2 Cr Ni Mo 25 7 duplex stainless steel
containing 40 vol.% austenite and 60 vol.% ferrite.
An optical method was developed in order to register online the strain field and the damage on
the surface of the specimens during the fatigue tests. By using a digital image correlation technique
the method identifies and extracts correspondent small zones in the deformed and in the reference
images (Hild et al [2]). The displacement vector field is determined, its gradient delivers the strain
field. Figure 1c shows the calculated longitudinal strain obtained from the image of a specimen
after 0.27% total strain in simple tension (fig. 1b) and from the reference image (fig. 1a) of the
undeformed specimen.

FIGURE 1. Strain fields measurement by digital image correlation

844 A. El Bartali et al.

For the optical strain analysis a very smooth cylindrical notch was machined on the specimen
surface. In order to follow the strain evolution and the surface damage, each specimen was
polished and etched. The observations were carried out at the scale of the grain size (about 10 Pm).
An in-situ observation device composed from CCD camera equipped with macro objective was
conceived and mounted on the frame of the testing system. It allows the observation of the
specimen during testing and the acquisition of images for the strain field analysis (Fig. 2). The
observation zone (about 100 Pm x 100 Pm) is large enough to be considered as a Representative
Elementary Volume (REV) of the macroscopic behaviour of the two-phase material.

FIGURE 2. In-situ observation device

Strain fields were calculated with an image correlation software. Analysis of the strain fields
and their evolution during the cyclic loading shows a heterogeneous strain distribution and the
localization of the plastic strain at the grain scale. Moreover, it allows to identify the initiation sites
of micro-cracks and to understand the micro-crack propagation process.

References
1. Aubin, V., Quaegebeur, P. and Degallaix, S., Materials Science and Enginerring A, Vol
346, 208-215, 2003
2. Hild, F., Raka, B., Baudequin, M., Roux, S. and Cantelaube ,F., Applied Optics, Vol. 41, No
32, 2002
 9. Micromechanisms in Fracture and Fatigue 845

MULTISCALE MODELING OF FRACTURE AND PLASTICITY IN LAYERED

STRUCTURES

Alexander Hartmaier, Nils Brodling and Huajian Gao

Max Planck Institute for Metals Research
Heisenbergstr. 3, 70569 Stuttgart, Germany
Hartmaier@mf.mpg.de

Plasticity in the geometrical confinement of a multilayered composite material is studied with a

dislocation-based model, where the mutual interaction of dislocations in the elastically
inhomogeneous layered structure is defined within the theory of elasticity on the continuum level.
The interface between two different layers is described as a cohesive zone, which allows crack
initiation and delamination to be studied on a rather fundamental level, taking into account
information on atomic binding energies across the interface. With this combined dislocation
dynamics-cohesive zone model we investigate whether a pre-existing crack in a ductile layer of the
composite structure will blunt or rather propagate in a brittle fashion. It is seen that the length scale
defined by the geometrical confinement of the plastic zone in the layered structure plays a critical
role in determining the behaviour of the entire structure.
In the two-dimensional model underlying this work, the stress and strain fields of straight edge
dislocations are calculated according to Mura's solution of the dislocation eigenstress in an infinite
bi-material [1]. The correct elastic boundary conditions for a plastic layer embedded in an elastic
material are established by dividing the problem into calculating stresses in an infinite elastic
medium and then correcting the tractions across the crack plane and the cohesive zone by
appropriate counter forces. In this work, these counter forces are produced by a surface layer of
virtual dislocations with continuous (non-lattice) Burgers vectors. The distribution of these virtual
dislocations that establishes the correct tractions across surface plane and cohesive zone is
calculated with a boundary integral method. The Green's function for the boundary integral is
constructed following the approach of Gutkin and Romanov [2]. Once the surface distribution of
the virtual dislocations is known, the corresponding stress and strain fields within the medium can
again be calculated using Mura's equations. The boundary conditions along the cohesive zone
comprise a region where the crack is fully open and where consequently the tractions across the
crack plane vanish. Ahead of this crack the cohesive zone is located, where the atomic bonds are
strained, but not yet completely separated. The surface tractions in this region are defined by the
elastic displacements across the cohesive zone, which are given by the displacement fields of the
dislocations and the boundary conditions, and the force-separation characteristics of the atomic
bonds. With this model the complex interplay between dislocation mediated plasticity and crack
opening, crack blunting and, finally, crack advancement is studied. Fig. 1 shows a sketch of the
combined dislocation dynamics-cohesive zone model.
The first applications of this model have been aimed at determining whether there exists a
critical thickness of the ductile layer below which crack advancement within this layer occurs and
above which the layer shows bulk-like behaviour. Our simulations have shown that the fracture
toughness of such layered structures increases continuously with the thickness of the ductile layer
until the bulk limit is met. The results are in good qualitative agreement with the simple analytical
model by Hsia, Suo and Yang [3]. The dependence of this transition thickness on material
parameters like dislocation density and distribution, dislocation mobility, and the relative
stiffnesses of elastic and plastic layers are investigated as well as the influence of the external
loading conditions and the temperature. Further studies will include initiation of secondary cracks
as a consequence of slip localization and subsequent dislocation pile-ups at interfaces (see Fig. 1).
846 A. Hartmaier et al.

FIGURE 1. Sketch of the combined dislocation dynamics-cohesive zone model. A primary crack
(cohesive zone 1) in the ductile layer (material 1) is loaded which causes yielding of the
surrounding material. Depending on the geometry and material properties it is possible that either
crack advancement occurs or ductile blunting of the crack tip. In the latter case slip localization
and stress concentrations from dislocation pile-ups against the elastic layer (material 2) may cause
opening of a secondary crack (cohesive zone 2).

References
1. Mura, T., In Advances in Materials Research, vol. 3, chap 1, edited by H. Herman,
Interscience Publishers New York, 1968, 164-201.
2. Gutkin, M.Yu. and Romanov, A.E. , physica status solidi, vol. 125, 107-125, 1991.
3. Hsia, K.J., Suo, Z. and Yang, W., J. Mech. Phys. Solids, vol. 42, 877-896, 1994.
 9. Micromechanisms in Fracture and Fatigue 847

CRITICAL AND FRACTURE PLANES OF 18G2A STEEL UNDER NON-

PROPORTIONAL COMBINED BENDING AND TORSION

A. Karolczuk and E. Macha

Technical University of Opole, Department of Mechanics and Machine Design
ul. Mikolajczyka 5, 45-271 Opole, Poland
karol@po.opole.pl, emac@po.opole.pl

Many multiaxial fatigue failure criteria of materials are based on the critical plane approach [1-3].
In these criteria, it is assumed that the fatigue failure of material is caused by stresses and/or strains
acting in one plane (called critical) within the material. The proper orientation of that plane in
fatigue failure criteria based on the critical plane concept must be established for fatigue life
calculation. In the paper the damage accumulation method was used to determine the critical plane
orientations. In this method, the critical plane is the plane where the maximum damage degree,
computed according to the selected fatigue failure criterion, is the highest. The critical plane
orientations are often compared to the experimental fatigue fracture plane orientations.
Fatigue tests were performed in the high cycle fatigue regime (HCF). Round cross-section
specimens made of 18G2A (Fig.1, a)) steel were subjected to different combinations of constant
and variable amplitude bending Mg,a and torsion Ms,a. The fatigue tests were performed under
bending and torsion moments control system. Two approaches were used to calculate stress
courses from histories of moments. In the first approach, stresses and strains were computed using
simple elastic beam theory (nominal stresses). In the second approach, time courses of moments
were used to calculate stresses and strains histories taking into account plastic strains and non-
linear stresses distribution along cross-section of specimen on the basis of the algorithm described
in the paper. The computed, by two approaches, histories of loading were used to calculate critical
plane orientations. It was assumed that the orientation of the critical plane is controlled only by
shear or tensile fatigue mechanism. Additionally, the theoretical critical plane positions were
compared to the experimental fatigue fracture plane orientations.
It was assumed that fatigue fracture plane orientation can be determined from the crack line
position detected on the free surface of the specimen. Detection of the fatigue crack and
measurement of its orientation were performed on the basis of photos of the specimen surface. The
photos of specimens were taken using a microscope with magnification of 60 times, connected
directly to a computer. The obtained picture of a surface segment was 1.8 x 1.8 mm (macroscopic
scale) with a resolution 167x167 pixels per mm2. Points representing cracks have been
approximated by a regression line using the least squares method. The slope coefficient of the
regression line has been used to calculate the experimental value of the angle Dexp (Fig. 1).
848 A. Karolczuk and E. Macha

&
FIGURE 1. Specimen geometry and scheme measurement of angle Dexp. (a) Shear s and normal
&
n directions in the plane defined by angle D (b)

From the obtained results and analysis the following conclusions can be drawn:
1 For proportional constant and variable amplitude loading the influence of plastic strains on
the critical plane orientation is negligible.
2 For non-proportional constant amplitude loading the plastic strains should be taken into
account to determine the critical plane orientation.
3 Accumulated damage degrees according to analyzed fatigue failure criteria are much more
higher in elastic model than they are in elasto-plastic model.

References
1. Carpinteri A., Spagnoli A., International Journal of Fatigue, vol. 23, 135-145, 2001.
2. Morel F., International Journal of Fatigue; vol. 22, 101–119, 2000.
3. Das J., Sivakumar M., Engineering Failure Analysis, vol 7, 347-358, 2000.
 9. Micromechanisms in Fracture and Fatigue 849

SLIP PROCESSES AND FRACTURE IN IRON CRYSTALS

V. Pelikan, P. Hora, A. Machova1 and M. Landa1

CDM-Institute of Thermomechanics AS CR, Pilsen, 301 14 Veleslavínova 11
1Institute of Thermomechanics AS CR, Prague 8, 182 00 Dolejškova 5

pelikan@cdm.it.cas.cz, hora@cdm.it.cas.cz, machova@it.cas.cz, ml@it.cas.cz

The brittle-ductile behavior of cracks has long been an area of intensive study. In particular, body
centered cubic (bcc) iron has been studied often in the recent past in the framework of continuum
(e.g. Rice et al. [1-3]) and atomistic models (e.g. Mullins et al. [4-8]), and as well in experiments
(e.g. Marsh et al. [9], Šmida and Bošanský [10]) due to the structural steels applications. The
models consider usually plane strain conditions (2D) along the crack front and temperature of 0 K.
However, free sample surface (where plane stress conditions are expected in the normal direction)
may influence the ductile-brittle behavior, as well as the thermal atomic motion. Such studies
require 3D atomistic simulations.
This contribution is devoted to crack simulations by molecular dynamic (MD) technique in 3D
bcc iron crystals. We use our new MD code for parallel processing in MPI. Interatomic interactions
in bcc iron are described using N-body potentials of Finnis-Sinclair type (Ackland et al. [11, 12]).
The 3D codes have been tested in perfect samples under simple uni-axial tension and in thermal
simulations. The simulated thermal expansion in bcc iron agrees well with experimental data (see
Machová [13]), as well as the phonon frequency spectra [11, 12].
In crack simulations we consider a central pre-existing Griffith (through) crack loaded in
tension mode I. The relatively long crack is embedded in a thin bcc iron crystal having the basic
cubic {100} orientation. The crack is introduced by removing part of atoms from the central plane,
i.e. its initial blunting corresponds to the lattice parameter. Crack surface lies on a (001) plane,
crack front is oriented along the [010] direction, and the direction of the potential crack extension
is [100]. The interatomic interactions across the free crack faces are not allowed. Surface
relaxation has been performed before a loading to avoid its influence on crack tip processes.
We performed the 3D crack simulations at temperatures of 0 K and 300 K. At these
temperatures, the samples were loaded symmetrically in the <001> directions by prescribing
external forces Fext distributed homogeneously at individual atoms lying in several surface layers,
similar to [11]. The samples were loaded slowly, gradually in time. While at the temperature of 0 K
brittle crack initiation has been observed, at 300 K dislocation emission and slip processes on
{110} and {112} planes have been detected. The slip processes start at the free sample surfaces,
which is in agreement with our stress analysis and MD simulations by Zhou et al. [14]. The
process begins on {110} planes via emission of a curved dislocation (~ quarter circle loop) from
the corners, where the crack penetrates the free sample surface. We further show that the slip
processes on the inclined {110} planes cause crack tip blunting and hinders crack growth. The slip
processes on the oblique {112} planes make jogs in the crack front and enable a slow plastic crack
growth. Our results are in agreement with continuum predictions that the microscopic processes at
the crack front generally depend also on the mutual orientation of the crack and accessible slip
systems (see e.g. Pokluda and Šandera [15]).
850 V. Pelikan et al.

Acknowledgements
The work was supported by the Grant Agency at Academy of Sciences of the Czech Republic
under a grant IAA2076201. The authors thank Dr. F. Kroupa, Dr. V. Paidar and Alena
Spielmannová for very helpful discussions.

References
1. Rice, J.R., J. Mech. Phys. Solids, vol. 40, 239-271, 1992
2. Beltz, G.E. and Fisher, L.L., In Multiscale deformation and fracture in materials and
structures, edited by T.J. Chuang and J.W. Rudnicki, Kluwer, Boston, 2001, 237-242.
3. Argon, A.S., Xu, G. and Ortiz, M., In Fracture-Instability, Dynamics, Scaling, and Ductile/
Brittle Behavior, edited by R.L.B. Selinger et al., Materials Research Society, vol. 409,
Pittsburgh, 1996, 29-44
4. Mullins, M. and Dokainish, M.A., Phil. Mag. A, vol.46, 771-787, 1982
5. Kohlhoff, S., Gumbsh, P. and Fishmeister, H., Phil. Mag. A, vol. 64, 851-878, 1991
6. Shastry, V. and Farkas, D., Modelling Simul. Mater. Sci., vol. 4, 473-492, 1996
7. Machová, A., Beltz, G.E. and Chang, M., Modelling Simul. Mater. Sci. Eng., vol. 7, 949-974,
1999
8. Beltz, G.E. and Machová, A., Scripta Materialia, vol. 50, 483-487, 2004
9. Marsch, P.G., Zielinski, W., Huang, H. and Gerberich, W., Acta metall. mater., vol. 40, 2883-
2894, 1992
10. Šmida, T. and Bošanský, J., Materials Sci. Eng. A, vol. 287, 107-115, 2000
11. Machová, A. and Ackland, G.J., Modelling Simul. Mater. Sci. Eng., vol. 6, 521-542, 1998
12. Ackland, G.J., Bacon, D.J., Calder, A.F. and Harry, T., Phil. Mag. A, vol. 75, 713-732, 1997
13. Machová, A., Computational Materials Science, vol. 24, 535-543, 2002
14. Zhou, S.J., Beazley, D.M., Lomdahl, P.S., Voter, A.F. and Hollian, B.L., In Advances in
fracture research, edited by B.L. Karihaloo et al., Pergamon, New York, 1997, 3085-3094
15. Pokluda, J. and Šandera, P., Metall. Mater., vol. 33, 375-383, 1995
 9. Micromechanisms in Fracture and Fatigue 851

A DISCUSSION OF THE APPLICABILITY OF 'K-VALUES TO EXPLAIN THE

FATIGUE CRACK GROWTH BEHAVIOUR OF SHORT CRACKS

A. Tesch, H. Doker, K. H. Trautmann, R. Pippan1 and C. Escobedo2

German Aerospace Center, Institute of Materials Research; 51170 Köln; Germany
1Erich-Schmid-Institute for Materials Science of the Austrian Academy of Sciences;

8700 Leoben; Austria

2Airbus Deutschland GmbH; 21129 Hamburg; Germany

andreas.tesch@dlr.de

In a joint research project of AIRBUS Deutschland in Hamburg and the Institute of Materials
Research of the German Aerospace Centre (DLR) in Köln short corner cracks emanating from
holes were investigated.
Fatigue crack growth thresholds and fatigue crack growth rates (da/dN) of long cracks were
determined at different R ratios for Al 2524-T351 and Al 6013-T6.

FIGURE 1. Short crack growth results for different stress ratios R, compared with long cracks
growth data for 6013-T6.

For the short crack tests a very small edge notch was cut into one edge of the hole of M(T)
specimens (Fig.1) with the Focused Ion Beam (FIB) Technology. From these FIB-notches short
corner cracks were introduced by compression loads (R=20). The lengths of these initial cracks
varied between 30µm and 100µm. Fatigue crack growth tests were performed with R = 0,1 and R
= 0,7.
In Fig. 1, the experimentally determined da/dN-'K curves of short corner cracks are compared
with the long crack test data for the alloy 6013-T6. In this graph the stress intensity factor solutions
were taken from Murakami’s stress intensity handbook [1].
852 A. Tesch et al.

It seems that:
At the same R-values short cracks grow below the threshold of the stress intensity factor range
of long cracks.
At R = 0,7 short cracks propagate approximately 10 times faster than long cracks at the same
'K-values.
The difference in the crack propagation rate between the long crack data and the short crack
data becomes bigger at lower R-ratios (e.g. R = 0,1).
The short crack growth rate is not influenced by the stress ratio R.
It is well known, that short cracks don’t fulfil the rules of linear elastic fracture mechanics
(LEFM) [e.g. 2-5]. If one takes an effective stress intensity factor derived from the Dugdale model
pertaining to perfect plastic materials [6], most of the short crack characteristics in Fig.1 vanish:
At R = 0,7 short cracks propagate with the same propagation rate as long cracks at the same
'K-values and the cracks don’t propagate below the threshold of long cracks.
At R = 0,1 short cracks grow below the threshold of long cracks at the same stress ratio, but
not below the threshold of long cracks at higher stress ratios (e.g. R = 0,7).
The crack growth rates of short cracks are independent of R.
This paper discusses the influence of the different solutions to calculate the stress intensity on
the observed short crack behaviour. The formula of LEFM [1] as one limit, the aera concept of
Murakami and Endo [7], considerations of Riemelmoser and Pippan [2], and an effective stress
intensity factor derived from the Dugdale model pertaining to perfect plastic material [6] as the
other limit are compared.

References
1. Newman, J.C. and Raju, I.S., in Stress Intensity Factors Handbook, Editor-in-Chief Y.
Murakami, Pergamon Press, Oxford, England, 1987, 712-722.
2. Riemelmoser, F.O. and Pippan R., International Journal of Fracture, vol. 118, 251-270,
2002.
3. Suresh, S., Fatigue of Materials, second edition, Cambridge University Press, Cambridge,
United Kingdom, 2001
4. Newman, J.C., Jr., in Small-Crack Test Methods, ASTM STP 1149, edited by J. Larsen and
J.E. Allison, American Society for Testing and Materials, Philadelphia, 1992, pp. 6-33.
5. Taylor, D., in The Behaviour of Short Fatigue Cracks, EGF Pub.1, edited by K.J. Miller and
E.R. de los Rios, Mechanical Engineering Publications, London, pp. 479-490.
6. Cioclov, D.D., in Fatigue Life Simulation-Equivalent Initial Flaw Approach-EIFSIM Version
3.2d/2004 User’s Manual, Saarbrücken, Germany, 2005.
7. Murakami, Y. and Endo, M., Engineering Fracture Mechanics, vol. 17, 1-15, 1983.
 9. Micromechanisms in Fracture and Fatigue 853

SIMULATION OF CRACK GROWTH UNDER LOW CYCLE FATIGUE AT

HIGH TEMPERATURE IN A SINGLE CRYSTAL SUPERALLOY

B. Fedelich, Y. Kiyak, T. May and A. Pfennig

Bundesanstalt für Materialforschung und -prüfung (BAM)
Unter den Eichen 87, 12205 Germany
bernard.fedelich@bam.de

Blades in gas turbines or aero-engines undergo a combination of creep and cyclic

thermomechanical loading corresponding to start up, steady state operation and shut-down. The
blades of the first stages are nowadays usually made of single crystal Ni-base superalloys due to
their higher creep resistance. Film cooling allows for even higher service temperatures but the
stress concentrations at the cooling holes and the local damage induced by drilling the holes are
known to promote cracking. The prediction of the crack behaviour under high temperature cyclic
loading is thus an essential element of any blade lifing procedure.
Crack growth tests have been performed at 950°C with Single Edge Notch specimens of a Ni-
base single crystal superalloy. In particular, several cycle shapes, frequencies and orientations have
been used, thus allowing the assessment of the influence of these parameters on the crack growth
rate. The crack growth is monitored by the electric potential method. The formula of Johnson [1]
has been used to relate the electric potential to the crack length. The distance between the leads and
the crack plane has been treated as an adjustable parameter and identified from several marker
bands on the fracture surface. In addition, oxidation experiments have been carried out to
characterise the kinetics of the oxide scale growth at the same temperature.
On the other side, crack growth has been simulated with the FE program Abaqus in real test
conditions by the node release technique. The FE nodes are released on the basis of the measured
crack growth rate. To limit the computational time, the simulations have been performed in either
plane strain or stress with a 2D sub-model. The appropriate boundary conditions for the sub-model
have been identified from computations with a 3D model of the whole specimen. The mesh of the
3D specimen is showed in Fig. 1a and an example of the predicted crack opening profile after 20
cycles can be seen in Fig1b.
The simulation results are compared with the test results on the basis of the computed crack tip
opening displacement (CTOD). The CTOD is evaluated by the 45° rule, which yield values that
are remarkably insensitive to mesh refinement in contrast to the opening of individual nodes, as
can be observed on Fig. 2. The crack is propagated until a stabilized value of the CTOD is
obtained. This is usually the case when the crack has crossed the initial plastic zone, as already
reported by several authors, e.g., González-Herrera and Zapatero [2]. The oxidation of the crack
flanks has been also taken into account, using the oxide scale measurements. The procedure
provides a direct evaluation of the effects of cycle form, crystal orientation, plasticity and oxide
induced crack closure.
854 B. Fedelich et al.

FIGURE 1. a) 3D mesh used to identify the boundary conditions on the 2D sub-model. b)

Equivalent plastic strain distribution at maximum load after 20 cycles.

FIGURE 2. First node opening displacement and CTOD evaluated by the 45° rule for two mesh
refinements.

A constitutive law for Ni-base single crystal superalloys has been developed and implemented
in the UMAT user-subroutine of Abaqus. It corresponds to a simplified version of a
micromechanical model previously developed by Fedelich [3]. The model is based on the slip
system theory and accounts for octahedral and cubic slip, for dislocation multiplication as well as
kinematical hardening. Furthermore, the effects of rafting during long duration tests can be
predicted.

References
1. Johnson, H. H., Materials Research & Standards, vol. 5, 442-445, 1965.
2. González-Herrera, A. and Zapatero, J., Engng. Fract. Mech., vol. 72, 337-355, 2005.
3. Fedelich, B., Int. J. Plasticity, vol. 18, 1-49, 2002.
 9. Micromechanisms in Fracture and Fatigue 855

FATIGUE CRACK GROWTH FOR DIFFERENT RATIOS OF BENDING TO

TORSION IN ALCU4MG1

D. Rozumek and E. Macha

Technical University of Opole
ul. Mikoajczyka 5, 45-271 Opole, Poland
drozumek@po.opole.pl and emac@po.opole.pl

The paper contains the fatigue crack growth test results obtained under proportional bending with
torsion in AlCu4Mg1 aluminium alloy [1]. Specimens with rectangular cross sections and
dimensions: length l = 90 mm, height w = 10 mm and thickness g = 8 mm were tested [2]. Each
specimen had an external unilateral notch with depth 2 mm and radius U = 0.2 mm. The tests were
performed under the stress ratio R = -1. The notches in the specimens were cut with a milling cutter
and their surfaces were polished after grinding. The tests were realized on a fatigue test stand
MZGS-100 where the ratio of torsion moment to bending moment was MT(t)/MB(t) = tgD
= 3 / 3, 1 and 3 (Fig. 1) and loading frequency was 29 Hz. The total moment

M t  M T2 t   M 2
B t  was generated by force on the arm 0.2 m in length.

FIGURE 1. Loading of the specimen.

Unilaterally restrained specimens were subjected to cyclic bending with torsion with the
constant amplitude of moment Ma = 7.92 N˜m. Crack development was observed on the specimen
surface with the optical method. The fatigue crack increments were measured with a digital
micrometer located in the portable microscope with magnification of 25 times and accuracy 0.01
mm. At the same time, a number of loading cycles N was written down. J-integrals were calculated
with the finite element method (FEM) using the computer program FRANC3D and FRANC2D.
The test results were shown as graphs of the crack length a versus the number of cycles N, and
crack growth rate da/dN versus the 'J integral range. 'JI was compared with the 'JIII for different
ratios of bending to torsion (Fig. 2). The experimental results for elasto-plastic range at the
constant rate value da/dN shown in Fig. 2 were described with the following formula
2 2
§ 'J I · § ' J III ·
¨ ¸ cos D  ¨ ¸ sin D 1
¨ J ¸ ¨ J ¸
© Ic ¹ © IIIc ¹ (1)
where JIc = 0.68JIIIc.
856 D. Rozumek and E. Macha

The graphs in Fig. 2 present the experimental results for stress ratio R = - 1. In Fig. 2 it can be
noticed that together with an increase in plasticity there occurs a deflection of experimental points
in direction of mode I. The bigger plasticity, the greater deflection of points is.

FIGURE 2. Experimental data for different ratios of bending to torsion for R = - 1.

The test results (Fig. 2) fatigue crack growth rate da/dN = 7.6˜10-8 m/cycle and da/dN =
1.4˜10-7 m/cycle were described with graphs 1 and 2, respectively.

References
1. Rozumek D., Proc. of the 7th Int. Conf. on Biaxial/Multiaxial Fatigue and Fracture, DVM
Berlin, Germany, 2004, 489-494.
2. Macha E., Rozumek D., Proc. of the 15th European Conference of Fracture, Advanced
Fracture Mechanics for Life and Safety Assessments, Stockholm, KTH, 2004, CD-ROM, 8ps

Acknowledgements: This work was supported by the Commission of the European Communities
under the FP5, GROWTH Programme, contract No. G1MA-CT-2002-04058 (CESTI)
 9. Micromechanisms in Fracture and Fatigue 857

DUCTILE DAMAGE MODELS APPLIED TO ANISOTROPIC FRACTURE OF

AL2024 T351

D. Steglich, W. Brocks and T. Pardoen1

Helmholtz Association of National Research Centres (GKSS), D-21502 Geesthacht, Germany,
dirk.steglich@gkss.de
1Université catholique de Louvain, B-1348 Louvain-la-Neuve,

Belgium, pardoen@pcim.ucl.ac.be

Al2024-T351 is an aluminium alloy renowned for its good mechanical properties, damage
tolerance and resistance against corrosion. Therefore it has been used for decades in aircraft
applications, especially in fuselage skins. Airplane designers estimate the fracture resistance of
fuselage components by ductile crack growth resistance tests, which have to be performed on
different specimen geometries and sizes. These test campaigns are time consuming and expensive.
The development of predictive models for ductile crack extension is thus essential in order to
improve structural design and maintenance and accelerate the insertion of new materials or
assembling methods.
Micro-mechanically based damage models accounting for both void nucleation and void
growth such as the Tvergaard or Rousselier models allow for the description of size and geometry
effects. These are often encountered when trying to transfer rupture parameters from small to large
test specimen. The models are commonly restricted to isotropic material behaviour with respect to
both deformation and damage. Metal alloys which have undergone extensive plastic deformation
by rolling, extrusion, etc. exhibit a significant anisotropy of mechanical properties, however.
Predictions of fracture resistance for rolled panels loaded in longitudinal and transverse direction
have been performed by accounting for the plastic anisotropy and assuming a scalar damage
variable. Due to the cubic symmetry of fcc and bcc metals, the yielding behaviour may still be
considered as almost isotropic, while ductility shows a significant reduction, if loading is applied
in the thickness direction. This effect is considered by orientation dependent damage parameters.
Generally, the identification of model parameters is a crucial point. It can be done
phenomenologically by fitting numerical simulations to respective results of mechanical tests. The
specific feature of micro-mechanical models is, however, that the damage parameters can be
directly identified from microstructural investigations.
Optical microscopy (OM), scanning electron microscopy (SEM) and synchrotron-radiation
based X-ray microtomography (XTM) have been used to obtain quantitative information. Figure 1
shows a reconstructed view of the investigated volume. The particle clusters appear white, while
the surrounding matrix is grey. It is obvious that the particles are aligned in a network-like
structure. These particle clusters separate the matrix into “matrix domains”. The domains have a
disc-like shape, with the S-direction as the shortest axis and almost identical dimensions in L- and
T-direction, see figure 2.
In the present contribution, the model developed by Pardoen and Hutchinson based on the
growth and coalescence of axisymmetric voids embedded in a work hardening matrix material is
chosen to simulate the failure of round notched bars. The effect of varying model parameters is
studied by cell model calculations. Special emphasis is laid on a careful characterisation of the
precipitate and particle morphology. From this microstructural information, a simplified model is
derived, which allows to directly evaluate model parameters for loading in S-direction. In a second
step, these parameters are modified based on simple geometrical assumptions and used to predict
failure in L-direction.
858 D. Steglich et al.

FIGURE 1. Particle clusters (white) in the investigated material in an XDM analysis.

FIGURE 2. Model of the microstructure accounting for direction-dependent failure mechanisms

 9. Micromechanisms in Fracture and Fatigue 859

FATIGUE AND FRACTURE PROCESSES IN SEVERE PLASTIC DEFORMED

RAIL STEELS

F. Wetscher, R. Pippan and R. Stock1

Erich Schmid Institute of Material Sciences, Austrian Academy of Sciences, Jahnstraße 12, A-
8700 Leoben, Austria
CD-Laboratory for Local Analysis of Deformation and Fracture, Jahnstraße 12, A-8700 Leoben
1Voestalpine Schienen GmbH, Kerpelystraße 199, A-8700 Leoben, Austria

wetscher@unileoben.ac.at

On the surface of rails a severely deformed layer is formed during the application due to the
permanent loading of a rail-wheel contact [1]. In this deformed layer cracks are formed and may
grow to a critical length. Therefore, it is of great importance to determine the fracture properties of
these plastically deformed regions to enhance finite element calculations for live time predictions
and define service intervals.
In this study, the rail steel UIC 900A has been severely deformed by Equal Channel Angular
Pressing (ECAP) [2] with a tool angle of 120° using Route A. The samples for ECAP had a size of
10x10x70mm and were deformed at room temperature. From these deformed samples, CT-
specimens for fracture toughness tests and fatigue tests were machined. The investigation of the
microstructure of this fully pearlitic steel subjected to severe plastic deformation has shown that a
nanostructured lamellar structure parallel to the shear plane evolves and the mechanical strengths
increases markedly [3, 4]. Hence, two critical directions for the crack were defined as shown in
Fig. 1.

Figure 1 Sketch of the ECAP deformed sample and the machined CT specimens

It can be expected that the crack cannot easily grow through the lamellae in the case of
specimen “A” whereas in specimen “B” almost no obstacles will hinder the progress of the crack.
Figure 2 shows that already after three ECAP passes the crack does no longer follow the original
direction in a fracture toughness experiment in the case of the sample orientation “A”. If the
sample orientation is “B”, the crack goes almost perfectly straight through the sample.
860 F. Wetscher et al.

Figure 2 Tested CT-Specimens after 3 ECAP passes with different orientation

Fatigue crack propagation curves were measured for different values of R using a direct
current potential drop technique.
All experiments were performed for different numbers of ECAP passes and therefore for
different shear strains. The results as a function of the direction and the strain are explained in
terms of the developing microstructure.

References
1. Wang, L., Pyzalla, A., Stadlbauer, W. and Werner, E.A., Mat. Sci. Eng. A., vol. 359, 31-43,
2003
2. Segal, V.M., Mat. Sci. Eng. A., vol. 197, 157-164, 1995
3. Wetscher, F., Vorhauer, A., Stock, R. and Pippan, R., Mat. Sci. Eng. A., vol 387-389, 809-
816, 2004
4. Wetscher, F., Tian, B., Stock, R. and Pippan, R., to be published in Mat. Sci. Forum
 9. Micromechanisms in Fracture and Fatigue 861

DAMAGE EVOLUTION IN TORSION SPECIMENS DEFORMED AT FORGING

TEMPERATURES

Gernot Trattnig, Reinhard Pippan and Siegfried Kleber1

Erich Schmid Institut of Material Science of the Academy of Sciences,
CD Laboratory for the local Analysis of Deformation and Fracture, Austria, Leoben
1Böhler Edelstahl, Kapfenberg, Austria

trattnig@unileoben.ac.at, Siegfried.Kleber@bohler-edelstahl.at, 1pippan@unileoben.ac.at

Torsion experiments at forging temperatures were carried out in order to analyse the dependency
of the damage evolution on forming parameters like testing temperature, strain, strain rate and
prior heat treatment.
The testing material was the nickel-based superalloy NiCr20TiAl (Böhler L306 VMR).
Experiments were carried out with a high temperature torsion testing machine at forging
temperatures between 900 and 1100 °C. In this way shear strains up to 3000 percent were
achieved. Flow curves for certain testing temperatures and strain rates are gained, furthermore the
influence of prior heat treatment was examined.

FIGURE 1. (a) Used torsion specimen geometry; (b) ruptured specimen after 25 revolutions at
1000 °C testing temperature with prior heat treatment; (c) SEM image of a cut through the
symmetry axis of specimen (b) with apparent voids.

The deformation of the torsion specimen (Fig. 1.a) leads to void nucleation, void growth and
void coalescence and subsequently to rupture of the torsion specimen (Fig. 1.b). The shear strain in
the specimen increases from cero in the centre to the maximum value at the surface. By analysing a
cut through the symmetry axis of a deformed specimen different deformation stages can be
examined with a single specimen.
Fig. 1.c shows a scanning electron microscope (SEM) image of a cut through the symmetry
axis of the specimen in Fig. 1.b. This specimen was deformed at a testing temperature of
862 G. Trattnig et al.

1000 °C to a shear strain of approx. 1200 % with prior heat treatment. By digital image analysis the
void area fraction and the number of voids were gained.
The void area fraction was taken as an indicator for the accumulated damage. By varying
experimental parameters like testing temperature, strain rate, prior heat treatment and shear strain,
their influence on the damage could be evaluated. In this way it is possible to evaluate different
damage respectively fracture criteria, e.g. presented by Clift et al. [1], Wifi et al. [2], Venugopal-
Rao et al. [3].
In addition circumferential notches were induced in torsion specimens with a lathe and
sharpened by razor blade polishing. The void density in un-notched torsion specimens was
compared to the void density in front of the growing crack in these notched torsion specimens.

References
1. Clift, S.E., Hartley, P., Sturgess, C.E.N., and Rowe, G.W., Int. J. of Mechanical Sciences,
vol. 32, 1-17, 1990.
2. Wifi, A.S., Abdel-Hamid, A., and El-Abbasi, N., J. of Materials Processing Technology, vol.
77, 285-293, 1998.
3. Venugopal-Rao, A., Ramakrishnan, N., and Krishna-kumar, R., J. of Materials Processing
Technology, vol. 142, 29-42, 2003.
 9. Micromechanisms in Fracture and Fatigue 863

MICROSTRUCTURAL EFFECTS ON SHORT FATIGUE CRACK

PROPAGATION AND THEIR MODELLING

H. J. Christ, O. Duber, W. Floer, U. Krupp, C. P. Fritzen1, B. Kunkler1 and A. Schick1

Universität Siegen, Institut für Werkstofftechnik, 1Institut für Mechanik und Regelungstechnik
57068 Siegen, Germany
hans-juergen.christ@uni-siegen.de

Cyclically loaded components in structural applications often undergo a stress amplitude which is
close to the fatigue limit of the material used. Under such conditions, crack initiation and short
crack propagation is considered to play an important role. It is well established that in high cycle
fatigue (HCF) cracks propagate as stage I cracks during a rather large fraction of fatigue life (up to
90% of the number of cycles until failure, Nf). The propagation behaviour of these so-called
microstructurally short fatigue cracks is strongly affected by microstructural properties, such as the
grain size and the presence of phase boundaries, and cannot be described using linear-elastic
fracture mechanics. Because of the significance of crack initiation and early crack propagation for
cyclic life, new concepts and experimental methods have to be applied in order to provide a robust
and reliable fatigue life assessment. Results from two interdisciplinary research projects are
reported in this presentation, which were carried out by the authors during the past seven years, in
order to shed light on the relevant mechanisms and to illustrate the implementation of these
findings in a life prediction model.
In order to study the effect of grain boundaries as obstacles to short crack propagation, a
metastable beta titanium alloy (TIMETAL®LCB) in the solution annealed single-phase condition
was used. Push-pull fatigue tests were conducted at room temperature in laboratory air at various
stress amplitudes and stress ratios on electrolytically polished, cylindrical specimens. A shallow
notch was used to predefine the crack initiation sites. Crack growth was monitored and evaluated
using optical and scanning electron microscopy (in combination with EBSD/OIM). Furthermore,
crack closure was determined by means of laser interferometry (ISDG).
It was found that crack initiation occurs mainly at grain boundaries on the surface and can be
attributed to the elastic anisotropy of the material [1]. Crack growth is strongly affected by the
microstructure of the material, in the sense that grain boundaries act as microstructural barriers
which decelerate or even stop propagating cracks. The barrier strength of a grain boundary
depends on the crystallographic misorientation of the adjacent grains, as shown by orientation
measurements using EBSD. Since stage I crack growth takes place on slip planes, the
misorientation between the slip planes in the grains involved is the decisive factor and not the
orientation difference between the lattices of the grains. Furthermore, crack closure measurements
showed that for low stress amplitudes roughness-induced crack closure is predominant, whereas at
higher stress amplitudes plasticity-induced crack closure prevails [2].
A duplex stainless steel (X2 CrNiMoN 22-5-3) with a two phase austenitic-ferritic
microstructure was applied to characterize the role of phase boundaries. Major differences in the
propagation behaviour of short cracks were noticed as compared to LCB, which can be related
mainly to the presence of two phases in the duplex steel. Crack initiation in the duplex steel occurs
primarily because of the interaction of slip bands, which develop in the softer austenitic phase,
with JD-phase boundaries. The crack propagation is determined not only by the crystallographic
misorientation between adjacent grains but also by the spacial arrangement of the constituent
phases. Also the phase of the grain that contains the crack tip is an important factor, since this
phase determines the plastic zone at the crack tip. At low stress amplitudes, crack growth takes
864 H. J. Christ et al.

place in the harder D-phase in single slip at 45° to the stress axis, whereas in the softer J-phase,
crack advance occurs in double slip and perpendicular to the stress axis. Therefore, the crack can
alternate between stage I and stage II crack growth depending on the yield strength of the current
grain [3].
On the basis of the results obtained for the titanium alloy a numerical model was developed,
which is capable to describe the strong interaction of the propagating short fatigue crack with grain
boundaries. Crack closure is implemented as well. The model calculates the crack growth in stage
I and hence predicts fatigue life on a physical basis. It also allows to simulate the short crack
growth phenomena found in the duplex steel. The influence of the spacial arrangement of the two
phases as well as the impact of the different yield strength is correctly reproduced [3,4]. A
comparison between experimentally observed data and the results of the simulation calculation is
given in Fig. 1.

FIGURE 1: Experimentally observed and simulated crack growth in a duplex steel [4]

References
1. Hu, Y.M., Floer, W., Krupp, U., Christ, H.-J., Materials Science and Engineering, vol. A278,
170-180, 2000.
2. Krupp, U., Floer, W., Lei, J., Hu, Y.M., Christ, H.-J., Phil. Mag A, vol. 82, 3321-3332, 2002.
3. Düber, O., Künkler, B., Krupp, U., Christ, H.-J., Fritzen, C.-P., Journal of Fatigue, in print.
4. Künkler, B., Düber, O., Krupp, U., Christ, H.-J., Fritzen, C.-P., in 11th International
Conference on Fracture 2005, CD-ROM.
 9. Micromechanisms in Fracture and Fatigue 865

MICROMECHANICAL ASPECTS OF TRANSGRANULAR AND

INTERGRANULAR FAILURE COMPETITION

I. Dlouhy and M. Holzmann

INSTITUTE of Physics of Materials, Academy of Sciences of the Czech Republic,
Zizkova 22, 61662 Brno, Czech Republic
idlouhy@ipm.cz

Occurrence of intergranular initiation of brittle fracture could be taken as significant simple

measure of the negative influence of impurities content. This micromechanism of failure has been
usually responsible for strong decrease of mechanical properties, anomalous fracture behaviour or,
at least, comparably larger and in structural applications unacceptable data scatter.
The intergranular fracture initiation itself has to be taken as result of competition between two
stress-controlled fracture micromechanisms: cleavage and intergranular. For action of these
micromechanisms a criterion of some local fracture stress acting over some microstructurally
susceptible distance (volume) has to be reached. What is the cause that forces the metal grain
boundary to fail in intergranular manner rather than transgranular? As known, it is a cohesion
strength that decreases at the same time as the cleavage fracture stress is kept on the same level.
But a number of questions are connected with relation of microscopic cohesion strength and local
fracture stress, with effect of pre-strain due to cold deformation or constraint phenomena etc. An
interconnection of fractographic methods with analysis of data from simple suitable specimen (e.g.
Charpy type) seems to be the effective tool for solution of these problems.
The aim of paper can be seen in analysis of micromechanical aspects of brittle fracture
initiation connected with intergranular decohesion as micromechanism that is well susceptible to
metallurgic cleanliness and cumulative degradation of material. Causes and characteristics
governing the intergranular fracture initiation and occurrence of this fracture micromechanism in
competition with cleavage one should be also addressed.

FIGURE 1.Comparison of FTTD for both steel melts after and step cooling treatment

A CrNi steel of commercial quality and the same steel with increased content of impurity
elements have been used for investigation. Step cooling treatment has been applied in order to
induce intergranular embrittlement and brittle fracture initiation in both steel melts. Except for
standard specimen geometry for three-point bending the pre-cracked Charpy type specimens were
applied for determination of fracture mechanical properties. Fractal analysis was applied in order
866 I. Dlouhy and M. Holzmann

to characterise the quantitative morphological differences in fracture surfaces. Relation of cleavage

fracture stress and critical stress for intergranular failure has been followed.
Fracture behaviour of the steel. The typical temperature dependences of fracture toughness are
shown in Fig. 1 for the steel A and R after step-cooling treatment. The as received state is here
represented by thin dashed lines only. It follows from comparison of main trends that except for
the scatter in fracture toughness values a very limited embrittlement occurred and only slight shift
of transition region was observed for melt A.
Quantitave characterisation of fracture surfaces. Change of fractal dimension, DF, is shown in
Fig. 2 for the investigated states. The separate points represent data for steel A in initial state and in
embrittled state. The scatter bands are shown for initial state represented by full lines only, for the
state after embrittling ageing by shaded field. It follows from comparison that practically the same
DF values appears to be the main characteristics of brittle failure for both microstructures. Based
on this finding and based on knowledge shown in previous part, the “embrittled” state of steel A
posses “still acceptable properties” and embrittlement process is just in stage of conversion to
intergranular initiation. Evident differences have been found when comparing both embrittled
microstructures of the steel melt A and R, the higher value of fractal dimension corresponds to
higher fracture roughness due to intergranular failure mechanism.

Figure 2.Change of fractal dimension, DF, with distance from crack tip, x, – steel A with initial
microstructure and after ageing, steel R after step cooling

Critical (fracture) stress quantification. To determine the local fracture stress experimentally
two procedures were used in the paper some quantitative results will be described in more detail.
It has been found that the (local) fracture stress is lower for steel with intergranular initiation
when compared to the steel with pure cleavage failure mechanism. Fractal dimension
corresponded well to brittle fracture morphology reflecting, on quite susceptible level, the changes
from pure cleave to intergranular fracture. The fracture stress may be used as local criterion of
fracture and as one of parameters for identification of critical impurity levels causing intergranular
embrittlement.
Authors gratefully acknowledge to Grant Agency of the Academy of Sciences (GAAV
200410502) for the financial support.
 9. Micromechanisms in Fracture and Fatigue 867

DEFECT IN ULTRA-FINE GRAINED MG-BASED ALLOYS DEFORMED BY

HIGH-PRESSURE TORSION

J. Cizek, I. Prochazka, B. Smola, I. Stulikova, R. Kuzel, Z. Matej and V. Cherkaska

Faculty of Mathematics and Physics, Charles University in Prague, Czech Republic
V Holešovikách 2, CZ-180 00 Praha 8 Czech Republics
R.K. Islamgaliev, O. Kulyasova
Ufa State Aviation Technical University, Russia
Ufa 450 000, Russia
jcizek@mbox.troja.mff.cuni.cz

Applications of Mg-based alloys at elevated temperatures are limited by the low melting point of
Mg. This difficulty can be overcome by an addition of rare earth elements. A number of novel
promising Mg-based hardenable alloys with high creep resistance at elevated temperatures have
been developed, e.g. Mg-Gd, Mg-Mn-Sc etc. Despite the favorable strength and thermal stability, a
disadvantage of these alloys consists in a low ductility, which is not sufficient for industrial
applications. Grain refinement is known as a way how to improve ductility. It has been
demonstrated that an extreme grain size reduction can be achieved by methods based on severe
plastic deformation (SPD). In the present work we used high pressure torsion (HPT), which is the
most efficient in grain size reduction among the SPD-based techniques, for preparation of selected
Mg-based alloys with ultra fine grained (UFG) structure. Microstructure investigations and defect
studies of HPT deformed UFG Mg-based alloys are presented in this paper. The extraordinary
properties of UFG materials are closely related with defects (grain boundaries, dislocations)
introduced by HPT. Positron lifetime (PL) spectroscopy [1] is a well-established non-destructive
technique with high sensitivity to open volume defects. It enables identification of the defect types
present in the material studied and determination of defect densities. Thus, PL spectroscopy
represents an ideal tool for defect studies of UFG materials. In the present work PL spectroscopy
was combined with X-ray diffraction (XRD), microhardness measurements, and direct
observations of microstructure by TEM.
Typical TEM image of HPT deformed Mg-10wt.%Gd alloy is shown in Fig. 1 as an example.
It exhibits homogeneous UFG structure with grain size around 100 nm, mainly high-angle type
grain boundaries, and high dislocation density. High number of dislocations leads to a significant
broadening of XRD profiles. Two components were resolved in PL spectrum of HPT deformed
Mg-10wt.%Gd alloy. Namely the free-positron component and a contribution of positrons trapped
at dislocations, which represent dominant positron trapping sites. Spatial distribution of
dislocations was investigated.
868 J. Cizek et al.

FIGURE 1. A TEM image of Mg-10wt.%Gd alloy deformed by HPT.

References
1. Hautojärvi, P., Corbel, C., In Proceedings of the International School of Physics “Enrico
Fermi”, Course CXXV, edited by A. Dupasquier, A.P. Mills, IOS Press, Varena 1995, 491-
532.
 9. Micromechanisms in Fracture and Fatigue 869

MODELLING CRACK-TIP SHIELDING EFFECTS IN PARTICLE

REINFORCED COMPOSITES

Jana Hornikova, Pavel Sandera and Jaroslav Pokluda

Brno University of Technology, Technická 2, Brno, Czech Republic
pokluda@fme.vutbr.cz

When a crack propagates in particle reinforced composites, the particles induce various shielding
effects at its tip. These effects can either increase or decrease the crack driving force which is
reflected in the change of fracture toughness of a composite. The increase in this value is
particularly important in case of brittle-matrix composites [1]. The paper is focused on the
calculation of selected local shielding effects caused by rigid particles and holes at the tip of a
crack under remote mode I loading conditions.
A numerical two-dimensional analysis based on the finite-element code ANSYS was
employed to investigate a superposition of influences of both the modulus difference and the crack
tip tilting, induced by a local mode II component. The influence of size and spacing of particles
distributed in a regular network around the crack tip was also considered in the numerical analysis.
The effective stress intensity factor Keff was calculated for many crack tip positions in an
appropriately chosen area between two particles of the network. Then, by assuming the translation
symmetry, the results were generalized for all possible positions within the network. The presence
of inclusions generally induces the mixed mode I + II at the tip of the straight crack and, in this
case, the crack tilting appears to minimize the mode II loading. The ratio between the averaged
stress intensity factors (SIFs) at deflected and straight crack fronts expresses a relative amount of
toughening [2]. Spherical particles of the same interparticle spacing l 60 µm , of different
diameters d  6, 180 µm and of 20-times higher Young moduli than the matrix (rigid
inclusions) are considered in the model. Moreover, particles of negligibly small moduli (holes) are
also studied for comparative reasons. A geometrical configuration of a standard CT specimen was
modeled using a mesh of six-node triangular elements in the investigated region. The central nodes
of elements at the pre-crack tip were shifted in order to simulate a singularity of the stress field.
Several thousands possible positions of the crack tip were analyzed within an investigated area in
between a pair of spherical particles, according to the scheme in Fig. 1. This area was chosen to be
long enough to reproduce the translation symmetry and to generalize the results to a periodical
square network of particles. This can be done by multiplication of normalized effective SIFs in the
points lying within both the left-hand and the right-hand parts of the investigated region associated
by the translational periodicity. To assess the crack driving force, the effective SIF [3] normalized
to the remote K If was used as follows:

12
K eff K 2
I  K II2  K If
(1)
The tendency to the crack tip tilting by an angle at the onset of an unstable fracture was
assesses by using the well-proved criterion [4] based on the minimization of the mode II affected
crack growth:

2
§D · KI § KI ·
4 tan ¨ ¸  ¨ ¸ 8
© 2¹ K II © K II ¹ (2)
870 J. Hornikova et al.

The Keff - factor can be considered to be invariant, i.e. independent on the kink angle , when
assuming only a formation of an elementary kink within the process zone during the KIc test. A
further propagation of the kink causing an additional shielding is beyond the frame of this analysis.

FIGURE 1. CT specimen and the investigated region (hatched area) of crack tip positions between
two particles.

In spite of a significant crack tip tilting induced by spherical holes, the resulting effect is a
slight increase in the crack driving force or the Keff factor (anti-shielding). However, a possible
crack-tip trapping in the holes was not taken into account. On the other hand, the rigid particles
cause a significant decrease in the Keff-value (shielding), while leaving the crack geometry
practically unaffected. These results are in a qualitative agreement with previously published
simple models [5] concerning the interaction of a crack with a single particle.

References
1. M. Kotoul and R. Urbiš: Engng. Fract. Mech. 68 (2001), p. 89.
2. K.T. Faber and A.G. Evans: Acta Met. 31 (1983), p. 565.
3. J. Pokluda, P. Šandera and J. Horníková: Fat. Fract. Engng. Mat . Struct. 27 (2004), p. 141.
4. D. Broek: Elementary Engineering Fracture mechanics (Martinus Nijhoff Publishers, The
Hague 1984).
5. F. Erdogan, G. D. Gupta and M. Ratwai: J. Appl. Mech. 12 (1974), p. 1007.
 9. Micromechanisms in Fracture and Fatigue 871

EARLY STAGES OF FATIGUE DAMAGE IN 316L STEEL

J. Man, K. Obrtlik, J. Polak and P. Klapetek1

Institute of Physics of Materials, Academy of Sciences of the Czech Republic
Žižkova 22, 616 62 Brno, Czech republic
1Czech Metrology Institute

Okružní 31, 638 00 Brno, Czech Republic

man@ipm.cz, pklapetek@cmi.cz, obrtlik@ipm.cz, polak@ipm.cz

The study of the nature and the physical mechanisms of fatigue damage evolution have started
more than 100 years ago by Ewing and Humfrey [1]. These authors firstly performed the optical
microscopic observations on flat specimens of polycrystalline Swedish iron fatigued in rotating
bending. They observed the localized cyclic slip in surface grains and the formations of
pronounced surface markings. Their most important finding – the identification of fatigue cracks
within the rough surface relief – has activated hitherto persisting interest in mechanisms of fatigue
damage.
The localization of the cyclic plastic strain in so-called persistent slip bands (PSBs) and the
formation of pronounced surface relief at the locations where PSBs intersect originally flat surface
is a general and very important feature of early fatigue damage of crystalline materials. The sharp
surface slip markings (called persistent slip markings – PSMs) consist of extrusions and intrusions.
Fatigue cracks nucleate in these locations later in the fatigue life (for recent review see Polák [2]).
Thus, for the understanding of fatigue crack initiation, knowledge of the PSMs topography and its
evolution is essential.
The purpose of the present work was to study growth and crystallographic pattern of intrusions
and extrusions within individual grains of polycrystalline austenitic 316L stainless steel fatigued
with constant plastic strain amplitude at room temperature. For this purpose two modern
experimental techniques were employed: atomic force microscopy (AFM) and electron
backscattering diffraction (EBSD). In agreement with our recent study by Polák et al. [3] and Man
et al. [4], unambiguously documenting limitations of AFM technique in observation and
quantitative evaluation of the surface relief topography, extrusion growth was systematically
monitored by direct observation of the metallic surface and plastic replica technique has been used
to reveal intrusions. Experimental details were similar to those described in our previous work [3,
4]. Experimental material was 316L steel supplied by Uddeholm with average grain size of 39 µm.
Experimental results and their discussion
Inverse pole figure as obtained by the analysis of EBSD measurement for all 32 surface grains
in 316L steel showed that no specific orientation of grains was preferred in this surface relief
study. The comparison of the observed and calculated trace angles of PSMs revealed that in the
majority of grains only 1 slip system (ss) was activated (primary ss in 20 grains, secondary ss in 7
grains). Only within 5 grains both the primary and the secondary ss were detected and in 1 grain no
PSMs were developed at the moment of arrest of the fatigue test (after 6750 cycles ~ 15% Nf).
872 J. Man et al.

FIGURE 1. Typical configurations of extrusion/ intrusions pairs observed in 316L steel.

FIGURE 2. Growth of extrusions and intrusions in 316L steel.

Statistical evaluation of results obtained by AFM systematic observations of identical areas

within individual grains in 316L steel using both direct specimen observation and plastic replicas
can be summarized as follows:
(i) Intrusions developing at PSB/matrix interface are always preceded by extrusions regardless
of the orientation of individual grains of a polycrystal.
(ii) Four typical configurations of extrusion/intrusion pairs were detected (see Fig. 1).
Intrusions denoted A and B are produced at the side of the extrusion where the emerging active slip
plane is inclined to the surface at an obtuse and an acute angle respectively (see also Fig. 2).
(iii) The first intrusions appear after 200–1000 cycles (~ 0.4–2.2% Nf) at the moment when the
“static” extrusions develop (Fig. 2), predominantly but not exclusively at the side of an extrusion
where the emerging active slip plane is inclined to the surface at an acute angle (position B denoted
in Fig. 1). They grow faster in comparison with stage II of extrusion growth.
(iv) Typical morphology of mature PSMs developed after 6 750 cycles (~ 15% Nf) consists of
ribbon-like extrusions accompanying by two thin parallel intrusions going simultaneously along
PSM/matrix interface.
Experimental data on the morphology and growth of extrusions and intrusions are discussed
and compared with the predictions of the recent theoretical models and computer simulations of
surface relief evolution and fatigue crack initiation which are based on irreversibility judged from
intra-bulk behaviour of entire PSBs.

References
1. Ewing, J.A. and Humfrey, J.C.W., Phil. Trans. Roy. Soc. (London), A200, 241–250, 1903.
2. Polák, J., In Comprehensive Structural Integrity, Vol. 4, edited by I. Milne, R. O. Ritchie and
B. Karihallo, Elsevier, Amsterdam, 2003, 1–39.
3. Polák, J., Man, J. and Obrtlík, K., Int. J. Fatigue, 25, 1027–1036, 2003.
4. Man, J., Obrtlík, K. and Polák, J., Mater. Sci. Eng., A351, 123–132, 2003.
 9. Micromechanisms in Fracture and Fatigue 873

AB INITIO STUDY OF ELASTICITY AND STRENGTH OF NANO-FIBRE

REINFORCED COMPOSITES

M. Cerny and J. Pokluda

Faculty of Mechanical Engineering, Brno University of Technology
Technická 2, Brno 616 69, Czech Republic
cerny.m@fme.vutbr.cz

Ab initio calculations of elastic moduli and theoretical uniaxial strength of composite lamina
having continuous nano-fibre reinforcements are performed using pseudo-potential plane-wave
code of Kresse et al. [1]. Obtained results are used to verify validity of macro-scale empirical
relations for composites (rules of mixtures) [2] on the nano-scale. All quantities are computed
from the dependence of crystal energy on a suitable deformation parameter. Results for tungsten
nano-fibres in niobium matrix will be presented as a particular example of the ab-initio analysis.
A model of the nano-composite is built by periodic repeating of 4x4x1 bcc-based supercell
displayed in Fig. 1.

FIGURE 1. The supercell

The crystal basis of our supercell contains 32 atom in both A (solid circles) nad B (open
circles) ¢001² planes. The gray solid circles belong to other (adjacent) supercells. The dashed
contours define considered interfaces between tungsten (W) wire and niobium (Nb) matrix in
investigated lamina models of different atomic concentration of W.
Calculation procedure consists of following steps: first, computation of the equilibrium volume
of each lamina model and determination of the lamina bulk modulus. Atomic positions within the
cell were relaxed in order to minimize the interfacial stresses. Second, the elongation of the crystal
system in ¢001² direction to simulate uniaxial loading applied in a direction parallel to the lamina
fibres. The system energy is minimized at any elongation by varying lengths of lateral cell edges
(the p1 and p2 translation vectors lengths). The dependence of total energy on the relative
elongation yields the uniaxial stress applied on the system as well as the Young modulus value.
The dependence of the computed composite bulk modulus B on the atomic concentration of W
(nearly corresponding to the volume fraction Vf of tungsten fibres) is depicted in Fig. 2 by solid
symbols. It seems to nicely follow the simple rule of mixture for an ideal composite
874 M. Cerny and J. Pokluda

B V f B f  (1  V f ) Bm (1)
where Bf and Bm are bulk moduli of the fibre (W) and matrix (Nb), respectively [2]. The line in
Fig. 2 follows eq. (1) for experimental values Bf = 310 GPa and Bm = 170 GPa.

FIGURE 2. The dependence of the bulk modulus on the atomic concentration of tungsten in the
Nb-W composite.

This result confirms on the atomistic level that deviations from eq. (1) observed for real
composites are caused by their imperfections, particularly by reduced interface cohesion.
Similarly, computed values of Young modulus as well as maximal values of uniaxial stress exhibit
simple linear dependence on atomic concentration of W. On the other hand, atomic volume
linearly decreases with W concentration.

References
1. Kresse, G. and Hafner, J., Phys. Rev. B vol. 48 13115, 1993; Kresse, G. and Furthmller, J.,
Phys. Rev. B vol. 54 11169, 1996; Kresse, G. and Furthmller, J., Comput. Mat. Sci. vol. 6 15,
1996.
2. Holiday, L., Composite materials, Elsevier, Amsterdam-London-NewYork, 1966.
 9. Micromechanisms in Fracture and Fatigue 875

STRENGTH AND FRACTURE OF ULTRA-FINE GRAINED ALUMINUM 2024

ECAP METAL

Kee Bong Yoon, Young Wha Ma1, Jeong Woo Choi1 and Seon Hwa Kim2
Professor, Department of Mechanical Engineering, Chung Ang University
221 Huksuk Dongjak, Seoul 156-756, Korea
1PhD Students, Graduate School, Chung Ang University

221 Huksuk Dongjak, Seoul 156-756, Korea

2Professor, Department of Materials Engineering, Soon Chun Hyang University

646 Eupnae Shinchang Asan, Chungnam 336-745, Korea

kbyoon@cau.ac.kr, roadpilot@wm.cau.ac.kr, go-omtang@wm.cau.ac.kr and
seonhwa@sch.ac.kr

When subjected to severe shear deformation by ECAP (Equal Channel Angular Pressing),
microstructure of Al2024 becomes extremely refined. To measure the strength of this ultra fine
grained metal, the small punch (SP) testing method was employed as a substitute for the
conventional uniaxial tensile testing since the size of metal bar processed by ECAP were limited to
12 mm in transverse direction as shown in Fig. 1. The small punch tests were performed with
specimens in longitudinal and transverse directions of Al 2024 ECAP metal. For comparing the
strength values with those assessed by SP tests, the uniaxial tensile tests were also conducted with
specimens in longitudinal direction, in which specimen sampling is possible. Failure
characteristics were investigated using scanning electron microscopy (Fig. 2). The surfaces of the
tested SP specimens showed that failure mode was shear deformation (Fig. 3). Based on this
observation it was argued that the conventional equations proposed for assessing the strength of the
material by the SP test were improper to assess those of Al2024 ECAP metal.
Using the geometry shown in Fig. 4, an equation for assessing the strength of Al 2024 ECAP
metal was proposed and was proven to be accurate by comparing the strength measured by the SP
test with that from the standard tensile test. Difference of the strength values between the SP test
and the standard test was less than 4.5%. Hence using the equation proposed in this study the
strength in T-direction was obtained with the same accuracy. Difference of microscopic fracture
characteristics in L and T directions and its relationship to the strength values were also discussed.

FIGURE 1. Specimen machining and directions of small punch testing for L- and T-specimens.
876 K. B. Yoon et al.

FIGURE 2. SEM observation of the tested SP specimen of Al 2024 ECAP metal.

FIGURE 3. Cross-section of
the interrupted small punch FIGURE 4. Measuring
test specimen of L-specimen parameters for assessing the
of Al 2024 ECAP metal. strength of the metal by SP
tests.

References
1. Choi, J.W., MS Thesis, School of Mechanical Engineering, Chung Ang University, 2005
2. Kim, Y.S., Theory of Plasticity, Sigmapress, Korea, 2003
 9. Micromechanisms in Fracture and Fatigue 877

FATIGUE LIFETIME OF BEARING STEEL IN ULTRA-HIGH-CYCLE REGION

Ludvik Kunz, Petr Lukas, Marian Cincala1 and Gianni Nicoletto2

Institute of Physics of Materials, ASCR, Žižkova 22, 616 62 Brno, Czech Republic, e-mail:
kunz@ipm.cz, lukas@ipm.cz
1University of Zilina, Velky diel, 01026, Žilina, Slovak Republic, e-mail:

marian_cincala@kmi.utc.sk
2Dept. of Industrial Engineering, University of Parma, Parco Area delle Scienze 181/A, 43100

Parma, Italy,
nick@me.unipr.it

Fatigue initiation mechanism in high-cycle and ultra-high-cycle region is a topic of increasing

importance due to rising demands on lifetime of engineering components. Car engines, high-speed
trains and turbines are typical examples of engineering systems loaded up to gigacycle region
[Bathias, C., Paris, P.C., Gigacycle fatigue in mechanical practice, M. Dekker, New York, USA,
2005.]. According to the valid standards, e.g. [ISO Standard 12107:2003(E) Metallic materials –
Fatigue testing – Statistical planning and analysis of data.], S-N curves are generally limited to 107
cycles. It is assumed, that the fatigue limit can be determined as a horizontal line below which no
failure takes place. Stair case method is frequently applied to the experimental determination of
fatigue limit on the basis of 107 cycles. In contrast to low-cycle and conventional high-cycle
fatigue, where the mechanisms of fatigue damage, crack initiation and propagation are more
explored, progress in ultra-high-cycle fatigue is highly desirable.
The experimental fatigue lifetime data in the ultra-high-cycle region have been shown to
exhibit sometimes a “two-stage” or “stepwise” S-N curve [Murakami, Y., Nomoto, T., Ueda, T.,
Murakami, Y., Fat. Fract. Engng Mater. Struct., vol.22, 581-590, 1999.,Wang, Q.Y., Baudry, G.,
Bathias, C., Berard, J.Y., In Proc. of the EUROMAT 2000, edited by Miannay, D. et. al, Elsevier
2000, 1083-1088.]. Majority of these observations has been performed on bearing steels by
Japanese researchers using rotating bending loading, e.g. [Tanaka, K., Akiniwa, Y., Fat. Fract.
Engng Mater. Struct., vol.25, 775-784, 2002.,]. It has been pointed out that the two-step S-N
curves are related to two different types of crack initiation: surface or internal at inclusions. Two-
step S-N curve was theoretically predicted even for one phase materials [Mughrabi, H., Fat. Fract.
Engng Mater. Struct., vol.22, 633-641, 1999.]. Recently some authors expressed the opinion that
the two-step S-N curve is related rather to the type of loading than to the material [Marines, I.,
Dominguez, G., Baudry, G., Vittori, J.-F., Rathery, S., Doucet, J.-P., Bathias, C., Int.J. of Fatigue,
vol.25, 1037-1046, 2003.].
The aim of this contribution is to report on experimentally determined S-N curves of bear-ing
steel in rotating bending and in tension-compression in region where a step is expected, to analyze
the crack initiation and to throw more light on the two step S-N curve phenomenon.
Hypereutectic bearing steel according to the Czech standard SN 14 109 and corresponding to the
European 100Cr steel, E52100 or SUJ 2 steel was used for the study. Two types of mechanical cyclic
loading were used. Tension-compression fatigue tests under controlled load were conducted in
resonant machine Amsler. The cycling was characterized by sine load wave with a frequency of 190
Hz. The stress ratio R was chosen equal to -1. The second loading system was rotating bending one,
operating at a frequency 50 Hz. Experiments were performed at room temperature in the ambient air
atmosphere.
878 L. Kunz et al.

Experimentally determined S-N curves in tension-compression and rotating bending are shown
in Figs 1a and 1b. It can bee seen that there is a clear difference in the curve shape. Despite of
considerable scatter of data, the experimental points corresponding to tension-compression test can
be approximated by continuously decreasing dependence, full line in Fig.1a. All specimens failed
by internal initiation. Data generated by rotating bending exhibit typical two-step curve and clear
separation of groups of points corresponding to the surface and internal initiation.

FIGURE 1a. S-N curve, tension-com-pression loading. FIGURE 1b. S-N curve, rotating ben-ding.

FIGURE 2. S-N curve based on tension-com-pression and corrected rotating bending data.
Fracto-grafic analysis including location of crack initiation made it possible to correct the
rotating bending results in terms of the local stress amplitude at the initiation site. Fig.2 shows the
corrected results for rotating bending together with tension-compression data. It can be seen that
the number of cycles to fracture increases continuously with decreeing stress amplitude.
In the discussion, details of crack initiation on inclusions of different type and the serious
scatter of lifetime data, which is inherent to the very high strength bearing steels, will be addressed.

References
1. Bathias, C., Paris, P.C., Gigacycle fatigue in mechanical practice, M. Dekker, New York,
USA, 2005.
2. ISO Standard 12107:2003(E) Metallic materials – Fatigue testing – Statistical planning and
analysis of data.
3. Murakami, Y., Nomoto, T., Ueda, T., Murakami, Y., Fat. Fract. Engng Mater. Struct., vol.22,
581-590, 1999.
4. Wang, Q.Y., Baudry, G., Bathias, C., Berard, J.Y., In Proc. of the EUROMAT 2000, edited by
Miannay, D. et. al, Elsevier 2000, 1083-1088.
5. Tanaka, K., Akiniwa, Y., Fat. Fract. Engng Mater. Struct., vol.25, 775-784, 2002.
6. Shiozawa, K., Lu, L., Ishihara, S., Fat. Fract. Engng Mater. Struct., vol.24, 781-790, 2001.
7. Mughrabi, H., Fat. Fract. Engng Mater. Struct., vol.22, 633-641, 1999.
8. Marines, I., Dominguez, G., Baudry, G., Vittori, J.-F., Rathery, S., Doucet, J.-P., Bathias, C.,
Int.J. of Fatigue, vol.25, 1037-1046, 2003.
 9. Micromechanisms in Fracture and Fatigue 879

CALCULATION OF K-FACTOR AND T-STRESS FOR CRACK AT

ANISOTROPIC BIMATERIALS

Michal Kotoul, Tomas Profant and Oldrich Sevecek

Brno University of Technology, Faculty of Mechanical Engineering
Technická 2, CZ 616 69, Brno, Czech Republic
kotoul@fme.vutbr.cz

The increasing use of fibre-reinforced composites in high performance structures has brought a
renewed interest in the analysis of cracks in anisotropic materials. Most matrices of the advanced
composite material are brittle. They prone to cracking under very low applied stresses and failure
frequently occurs in the form of multiple matrix cracking. The orientations of these cracks may
vary depending on the relative position of the reinforcement in relation to the load. The stress field
in the neighbourhood of crack is governed by the overall anisotropic material response. Therefore,
first the overall anisotropic parameters must by obtained by some homogenisation technique to
calculate the stress field, which drives the crack in the matrix material. The existence of material
interfaces in composites, especially in laminates, brings another problems in the analysis of cracks
– the problem of crack terminating at the interface of two anisotropic solids and the problem of
interfacial crack in anisotropic solids. These problems are also encountered in the technology of
protective coatings.
For the assessment of crack behaviour in the aforementioned situations it is essential to
investigate the stress field near the crack tip. Although the finite element (FE) analysis is capable
of capturing the singular stress behaviour near a corner or a crack tip in homogeneous regions with
a refined mesh of conventional elements, this traditional FE approach fails to accurately capture
the appropriate singular behaviour near a corner or a crack tip at the junction of dissimilar
materials. A very promising approach to an accurate calculation of the near crack tip fields consists
in the application of so-called two-state (or mutual) conservation integrals. Among these two-state
conservation laws, the two-state J-integral proposed by Chen and Shield [Chen, F.H.K.and Shield,
R.T., Z. Angw. Math. Phys. (ZAMP), vol. 28,1-22, 1977.] has been widely employed for obtaining
stress intensities and elastic T-stresses for cracks, as well for finding dislocation strength. The two-
state conservation integrals, e.g. [Chen, F.H.K.and Shield, R.T., Z. Angw. Math. Phys. (ZAMP),
vol. 28,1-22, 1977.]-[Choi, N.Y.and Earmme, Y.Y, Mech. Materials, vol. 14, 141-153, 1992.], in
conjunction with a displacement-based FEM provide an efficient tool for calculating stress
intensities and elastic T-stresses for crack, or for calculating dislocation strength. This method is
capable of extracting the near-tip information directly from the far-field deformation where the
numerical fields are more accurate. This is a major advantage over the singular finite elements, and
other various special techniques such as the boundary collocation, the singular hybrid FEM or the
enriched FEM. Basically, the two state conservation integral can be based upon the well-known J-
integral or upon the M-integral proposed by Budiansky and Rice [Budiansky, B. and Rice. J.R.,
ASME J. Appl. Mech., vol. 40, 201-203, 1973.] and Knowles and Sternberg [Knowles. J.K.and
Sternberg, E., Arch. Rat. Mech. Anal., vol. 44, 187-211, 1978.].
Let ‘A’ and ‘B’ denote two independent elastic states for the plane problems. Another elastic
state ‘C’ can be obtained by superposing ‘A’ and ‘B’. Then one can write the path-independent
integrals

JC J A  J B  J  AB  , M C M A  M B  M  AB  (1)
880 M. Kotoul et al.

The integrals J(A,B) and M(A,B) result from the mutual interaction between two elastic states ‘A’
and ‘B’. These are conservation integrals for two equilibrium states, since the area version of these
contour integrals vanishes identically for the domains with no singularities. In applications, one of
the elastic states ‘A’ and ‘B’ comes for example from FE computation and the second is called the
auxiliary solution. The success of the mutual integral method is crucially linked to the existence of
`the auxiliary solutions in the form of the complementary eigenfunction. The existence of these
complementary solutions was proved with the aid of the eigenfunction series by Im and Kim [Im,
S.and Kim, K.-S., J. Mech. Phys. Solids, vol. 48, 129-151, 2000.]. Once the mutual integral is
computed from the far-field deformation together with appropriate auxiliary solution, the
individual stress intensity factors and/or T-stresses may be determined. While the auxiliary
solution was found for semi-infinite and finite (interfacial) crack for the general case of anisotropic
medium, see e.g. Beom and Atluri [Beom, H.G.and Atluri,S.N., Int. J. Fract., vol. 75, 163–183,
1996.] and Kim et al. [Kim, J.H., Moon, H.J. and Earmme, Y.Y., Mech. Mater., vol. 33, 21–33,
2001.], it is still unknown for bi(multi)material wedge for the present.
The paper will address the application of the concept of mutual two-state integrals to the
calculation of stress intensities and T-stresses for crack arrested at an anisotropic/anisotropic
material interface. Numerical procedures for estimation of general stress intensity factors based on
finite element method will be suggested and tested. As a first step of the analysis a numerical
procedure based upon the implicit method suggested by Papadakis and Babuska [Papadakis, J.P.
and Babuska, I., Comput. Methods Appl. Mech. Eng., vol 122, 69-92, 1995.] will be used for the
determination of eigenvalues and eigenvector in Williams-like asymptotic expansion. Implicit
methods do not have a closed form for the eigenequation, they are slower but they can be used for
anisotropic materials as well as multi-material wedges. They use the variational formulation of the
problem which provides a functional of sesquilinear form. The classical FEM approximation is
then used leading to a system of homogeneous algebraic equation for eigenvalues and
eigenvectors. In second step, the construction of auxiliary solution for multi-material wedge will
be attempted. Finally, the stress distribution in the vicinity of the singular stress concentrators will
be obtained by combination of the developed asymptotic analytical and corresponding numerical
computations based on finite element method.

References
1. Chen, F.H.K.and Shield, R.T., Z. Angw. Math. Phys. (ZAMP), vol. 28,1-22, 1977.
2. Yau, J.F., Wang, S.S. and Corten, H.T, ASME J. Appl. Mech., vol. 47, 335-341, 1980.
3. Shih, C.F.and Asaro, R.J, ASME J. Appl. Mech., vol. 55, 299-316, 1980.
4. Kfouri, A.P., Int. J. Fract., vol.30, 301-315, 1986.
5. Choi, N.Y.and Earmme, Y.Y, Mech. Materials, vol. 14, 141-153, 1992.
6. Budiansky, B. and Rice. J.R., ASME J. Appl. Mech., vol. 40, 201-203, 1973.
7. Knowles. J.K.and Sternberg, E., Arch. Rat. Mech. Anal., vol. 44, 187-211, 1978.
8. Im, S.and Kim, K.-S., J. Mech. Phys. Solids, vol. 48, 129-151, 2000.
9. Beom, H.G.and Atluri,S.N., Int. J. Fract., vol. 75, 163–183, 1996.
10. Kim, J.H., Moon, H.J. and Earmme, Y.Y., Mech. Mater., vol. 33, 21–33, 2001.
11. Papadakis, J.P. and Babuska, I., Comput. Methods Appl. Mech. Eng., vol 122, 69-92, 1995.
Acknowledgement. The support through the grant GACR 101/05/0320 is gratefully
acknowledged.
 9. Micromechanisms in Fracture and Fatigue 881

INTERACTION OF MICROCRACKS WITH GRAIN BOUNDARIES:

SYSTEMATICAL INVESTIGATION OF THE MECHANISMS

M. Marx, W. Schaf and H. Vehoff

Institute of Materials Science and Methods, Saarland University
Building 43B, D-66041 Saarbrücken, Germany
m.marx@matsci.uni-sb.de

Fatigue is based on the initiation and propagation of microcracks. This period can occupy up to
90% of the lifetime of a cyclic loaded structure. It is well known that short fatigue cracks propagate
faster than long cracks at the same level of the stress intensity factor 'K [1,2,3]. Deshpande,
Needleman and Van der Giessen showed that the behavior of short cracks is determined by the
internal stresses under cyclic loading due to the dislocation structure in the vicinity of the crack tip
[4]. However, in this investigation the influence of grain boundaries and phase boundaries was not
included. Due to the emission and movement of dislocations during crack propagation grain
boundaries influence the crack propagation rate, it is observed that short cracks may stop in the
front of grain boundaries. Therefore the estimation of lifetime of a broad range of fatigued
structures and components on the basis of the stress intensity factor is difficult.

FIGURE 1. Crack initiated by Focused Ion Beam (FIB) in front of a grain boundary

The interaction between microcracks and micro structural elements like grain boundaries and
phase boundaries is described qualitatively quite well by the models of Tanaka and Navarro and
De Los Rios [5,6]. However, the influence of grain boundary parameters like the misorientation of
the grains at the grain boundary or grain boundary segregation is missing in these models. To
create a physically more fundamental model it is necessary to investigate the interaction of
microcracks with different grain boundaries and different crack parameters systematically. These
measurements are not possible by observing natural cracks. Therefore a new method was
developed to initiate microcracks with a predicted crack length and distance from the grain
boundary at grain boundaries with different misorientation angles. The grain boundaries were
characterized completely by orientation imaging microscopy. Afterwards cracks were initiated
near the grain boundaries of interest by focused ion beam (FIGURE 1). The crack growth during
cyclic loading was investigated by replica technique and by in situ loading in SEM with a
resolution of about 20 nm. The influence of the grain boundary on the crack tip opening
displacement and crack propagation rate (da/dN) was measured quantitatively.
882 M. Marx et al.

FIGURE 2: Crack propagation rate as function of the distance from the grain boundary,
sample 1: small angle boundary (13°), sample 2: large angle boundary (32°)

A significant difference was found in the crack propagation rate for small angle and large angle
boundaries (FIGURE 2). Additionally the Electron Channeling Contrast Imaging (ECCI)-
technique was used to get further information on the interaction of cracks and grain boundaries. By
in situ loading the crack propagation mechanisms were investigated step by step. The further aim
of these investigations is a better lifetime prediction and a prolongation of the lifetime by grain
boundary engineering.

References
1. Suresh, S., Ritchie, R.O., Int. Metals Rev., vol. 29, 445, 1984
2. Ritchie, R.O., Peters, J.O. Mat. Trans., vol. 42, 58, 2001
3. Suresh, S., Fatigue of materials, Cambridge University Press, Cambridge, UK, 1998
4. Deshpande, V.S., Needleman, A., Van der Giessen, E., Cata Materialia, vol. 51, 1, 2003,
5. Tanaka, K., et. al., Eng. Fracture Mech., vol 24, 803, 1986
6. Navarro, A., De Los Rios, E.R., Phil. Mag. A, vol. 57, 15, 1988
 9. Micromechanisms in Fracture and Fatigue 883

DISLOCATION ARRANGEMENTS IN CYCLICALLY STRAINED INCONEL

713LC

M. Petrenec, K. Obrtlik and J. Polak

Institute of Physics of Materials, Academy of Sciences of the Czech Republic
Zizkova 22, 616 62 Brno, Czech Republic
petrenec@ipm.cz, obrtlik@ipm.cz, polak@ipm.cz

INCONEL 713 LC is a cast polycrystalline superalloy used to produce turbine wheels of small
internal-combustion turbines [1]. The dislocation structure, which is formed in the material during
cyclic loading, is closely connected with fatigue damage of the material. The fatigue processes of
numerous structural materials in the initial stage are characterized by heterogeneous dislocation
configuration that becomes unstable during cycling. Localized bands with a specific substructure
are formed. Due to the localized deformation the characteristic surface relief is developed and
subsequently fatigue cracks are initiated [2]. In the studies of the dislocation structures of CSMX4
single crystal the specific dislocation arrangement corresponding to cyclic strain localization was
observed only recently [3]. The aim of this contribution is to study the dislocation arrangement of
INCONEL 713 LC cyclically strained in symmetrical cycling at room and high temperatures. The
work is a part of a complex project directed to the relationship between the internal structure, the
stress-strain response and the fatigue life of polycrystalline nickel base superalloys [1,4].
Experimental, results and discussion
Polycrystalline INCONEL 713 LC superalloy was supplied by PBS Velká Bíteš, a.s. in the
form of cast rods. Chemical composition is 11.90 Cr, 5.75 Al, 4.57 Mo, 1.96 Nb, 0.7 Ti, 0.19 Fe,
0.10 Zr, 0.08 Co, 0.05 C, 0.013 B, the rest is Ni (all in wt %). The structure includes rough
dendrites and shrinkage pores up to 0.4 mm in diameter. The average grain size was 4.2 mm [1,4].
The specimens were fatigued in electrohydraulic MTS 810 testing machine in symmetrical push-
pull loading with constant total strain amplitude Ha to fracture. The tests were conducted at room
temperature (300 K) and at 773 K, 973 K and 1073 K in the air [4]. Internal dislocation structures
were observed in Philips CM-12 transmission electron microscope, operating at 120 kV with a
double tilt holder using the technique of oriented foils [5]. Specimens were sectioned in the gauge
area parallel to the loading axis. Bright field imaging conditions were mostly adopted and the
diffraction patterns and Kikuchi lines were used to determine the grain orientation (stress axis S.A.
and foil plane F.P.).
Though the details of the dislocation structures slightly differed depending on the testing
temperature, the general feature, i.e., the thin dislocation rich bands passing through the J channels
and the J´ precipitates parallel to the primary slip plane (111), were common at all the
temperatures. Contrary to room temperature cycling, where ladder-like structure of the bands
developed (Fig. 1), only thin bands passing through the matrix and precipitates (Fig. 2) are present
at 1073 K. In the previous studies [6,7], the arrangement of dislocations in the bands parallel to the
primary slip plane was observed only exceptionally. Dislocation arrangements observed in the
present work are consistent with surface relief study by AFM in IN 713LC at different
temperatures [4]. Parallel slip markings are common feature of the surface relief produced by
cyclic straining at all the temperatures. In the room temperature cycling, the localization of the
cyclic slip into very thin bands results in building up the sharp surface relief in the form of
persistent slip markings consisting of extrusions and intrusions. The sharp persistent slip markings
are thus closely connected with the specific dislocation arrangements in the thin bands running
884 M. Petrenec et al.

parallel to the primary slip plane similarly as in case of single and polycrystals of simple f.c.c. and
b.c.c. metals [2,8].

FIGURE 1. Detail of a slip band in a single-slip-oriented grain (300 K).

FIGURE 2. Detail of slip bands in a single-slip-oriented grain (1073 K).

References
1. Obrtlík, K., Man, J., Polák, J., In Proceedings of 7th European Conference on Advanced
Materials and Processes (EUROMAT 2001), Associazione Italiana di Metallurgia, Milano,
2001, No. 894.
2. Polák, J., Cyclic Plasticity and Low Cycle Fatigue Life of Metals, Academia/Elsevier, Prague/
Amsterdam, 1991.
3. Obrtlík, K., Lukáš, P., Polák, J., In Low Cycle Fatigue and Elasto-Plastic Behaviour of
Materials, K.-T. Rie and P.D. Portella (Eds.), Elsevier, Amsterdam, 1998, 33–38.
4. Obrtlík, K., Man, J., Petrenec, M., Polák, J., In Proceedings of the Eight International Fatigue
Congress (FATIGUE 2002), edited by Blom, A. F., vol. 2/5, EMAS, West Midlands, UK,
963–970.
5. Obrtlík, K., Polák, J., Komrka, J., Scripta Metall. Mater. 28, 495–499, 1993.
6. Monier, C., Bertrand, C., Dallas, J. P., Trichet, M. F., Cornet, M., Mater. Sci. Eng. A188, 133-
139, 1994.
7. Decamps, B., Brien, V., Morton, A.J., Scripta Metall. Mater. 31, 793-798, 1994.
8. Polák, J., In Comprehensive Structural Integrity, edited by Milne, I., Ritchie, R. O.,
Karihaloo, B., vol. 4, Elsevier, Amsterdam, 2003, 1-39.
 9. Micromechanisms in Fracture and Fatigue 885

CRACK INITIATION AND FRACTURE OF METAL MATRIX COMPOSITES

K. Unterweger and O. Kolednik

Erich Schmid Institute of Materials Science, Austrian Academy of Sciences
Jahnstrasse 12, A-8700 Leoben, Austria
kolednik@unileoben.ac.at

An experimental study is performed to investigate the role of the composite architecture on the
mircomechanisms of crack initiation and fracture of particle reinforced metal matrix composites
(MMCs) under monotonic loading. In-situ tensile tests are performed in the scanning electron
microscope (SEM), and the SEM-micrographs of different deformation stages are analyzed by an
automatic local deformation measurement system. The procedure, which is described in Tatschl
and Kolednik [1], yields the local deformation and strain fields at the different stages with high
accuracy and lateral resolution. In this way, the influence of the local composite architecture on the
damage evolution in the material can be studied in great detail: the occurrence of first plasticity at
particle corners, the formation of shear bands, particle fracture and decohesion, final fracture.
The investigated material is a powder-metallurgically processed MMC with an Al6061 alloy as
matrix and 10 or 20 vol. % SiC particles as the reinforcing phase. The particles have a mean
diameter of 10 and 100µm, respectively. The specimens were solution annealed at 530°C for one
hour, quenched and subsequently aged at 175°C for 15min, 8h, and 200h in order to yield an
under-aged, a peak-aged, and an over-aged condition. For comparison, the un-reinforced matrix
materials are also tested.
As examples of the analyses, in Fig. 1 local strain maps are presented for the over-aged MMC
with a volume fraction of 10% small and coarse particles at two different stages of deformation.
Plotted are the local relative strains in loading direction Nxx, i.e., the local strains Hxx divided by the
global strain. By comparing two stages of deformation, it becomes evident that for the MMC with
small particles the pattern of the local relative strains remains constant during the deformation.
Neither the local values of Nxx change significantly, nor an emergence of new shear bands can be
found.
It has been found that even in the pure matrix material the deformation is rather inhomogeneously
distributed and shear bands develop which are orientated under approximately 45° to the tensile
axis. The deformation pattern of the MMCs with small particles appears similar, but the
deformation pattern is more diffuse than in the pure matrix material. Particle fracture happens very
rarely and does not induce far-reaching shear bands in the matrix. Along with a good particle
distribution and an obviously good interfacial bonding between matrix and reinforcements, the
MMCs have therefore a high strength and a good ductility. The MMCs with coarse particles
exhibit a much lower ductility than the MMCs with small particles. The reason is that they deform
due to the development of particle-damage controlled shear bands in the matrix which take the
main part of deformation. In the case of the peak-aged material only these damage controlled
bands are observed, whereas in the under- and over-aged MMCs, additionally matrix shear bands
develop independently from particle damage. This explains the poor mechanical properties of the
peak-aged MMC with coarse particles.
886 K. Unterweger and O. Kolednik

Figure 1. Local relative strains in loading direction (horizontal) for: MMC with 10µm particles at a
global strain of 1.94% (a) and 5.07% (b); pixel-size = 0.04µm. MMC with 100µm particles at a
global strain of 1.17% (c) and 2.72% (d); pixel-size = 0.232µm.

Noticeable further improvements of the ductility and fracture toughness of MMCs can be only
obtained if the local deformation and crack growth behavior is explored and related to the local
composite architecture, see also Unterweger and Kolednik [2,3].
Acknowledgements
The authors acknowledge gratefully the financial support of this work by the Fonds zur
Förderung der wissenschaftlichen Forschung under the project number P14333-N02.

References
1. Tatschl, A., Kolednik, O., Mat. Sci. Eng., vol. A339, 265-280, 2003.
2. Unterweger, K., Kolednik, O., Materials Science Forum, vol. 482, 215-219, 2005.
3. Unterweger, K., Kolednik, O., Z. Materialkde, in press.
 9. Micromechanisms in Fracture and Fatigue 887

MECHANICAL BEHAVIOUR OF ULTRA-FINE GRAINED AUSTENITIC

STAINLESS STEEL

Stephanie Brochet, Angeline Poulon-Quintin, Jean-Bernard Vogt, Jean-Christophe Glez1 and

Jean-Denis Mithieux1
Université de Lille I, LMPGM, UMR 8517, 59 655 Villeneuve d’Ascq, France
1UGINE&ALZ Research Center, BP 15, 62 330 Isbergues, France

stephanie.brochet@ed.univ-lille1.fr

There is an increasing interest in the development of fine grained materials because this could
appear as a solution for providing higher mechanical strength alloys, allowing thus reduction in
component mass and raw material saving. The origin in the improvement of yield stress is directly
connected with the the Hall-Petch relationship:

k
Re Re0 
d (1)
Several processes are now available for the elaboration of ultra fine grained materials among
which grain refinement reached by the recrystallisation of a highly cold rolled material.
One interesting point of ferrous alloys is that they can exhibit phase transformation, either
induced by temperature variations or by plastic deformation. Taking advantage of this last point,
the tendency is to use metastable austenitic stainless steels for which microstructure undergoes to a
partial martensitic transformation J Æ D’ during cold rolling resulting in a fine grained
microstructure after recrystallization.
With this aim in view, a new grade, type 316L modified, is developed by UGINE&ALZ.
Successful thermo-mechanical processes give rise to microsize grained microstructures. These
new grades have to satisfy a large panel of mechanical properties which must be investigated and
connected with the microstructure as well as process parameters.
The objective of the present work is to study the effect of the grain size and of the phase
stability on the monotonic and cyclic response of a metastable austenitic stainless steel.
Two austenitic microstructures, illustrated on the figure 1, are studied: the first one with a grain
size of about 2µm and the second with a grain size of 20µm.

FIGURE 1. Optical micrograph of two studied microstructures tested in monotonic and cyclic
experiments. Estimated grain size: a) 2µm b) 20µm.
888 S. Brochet et al.

For tensile tests, the decrease of one decade of the grain size leads to an increase of 250MPa of
the yield stress as well as the ultimate tensile strength values with keeping good value of
elongation.
Low cycle fatigue tests are carried out on thin sheet specimens (one millimetre in thickness)
under total axial strain control ranging from 0,4% to 1,2% and using a fully push pull mode (RH=-
1).
The study of the cyclic stress response shows an important influence of the grain size:
- The cyclic accommodation, at the beginning of the test (less than 5% of the lifespan), consists
of a cyclic softening for the fine grained microstructure whereas a cyclic hardening always
precedes this stage on the coarse grained microstructure. The stress levels are 250MPa higher with
the small grained microstructure.
- A secondary cyclic hardening is observed for the higher strain levels. The intensity of this
cyclic hardening increases with the strain amplitude, and is much significant with the coarse
grained microstructure.
The fatigue resistance is evaluated trying to take into account both the plastic strain and the
stress amplitude. The mechanical behaviour is discussed from X-ray diffraction and TEM analyses
of the microstructures before and after fatigue tests. Both small and large grained microstructures
undergo phase transformation. The martensitic transformation is direct J Æ D’, H martensite is not
observed.
Nevertheless, even if the fatigue behaviour is function of the grain size, the amount of
martensite formed due to the austenite instability during cyclic strain, seems to be nearly
equivalent for both microstructures after fatigue tests for a same strain amplitude. This suggests
that cyclic plasticity mechanisms differ between the fine and large grained steels.
 9. Micromechanisms in Fracture and Fatigue 889

TRIBOLOGICAL PROPERTIES AND WEAR MECHANISMS OF WEAR

RESISTANT THERMALLY SPRAYED COATINGS

Sarka Houdkova, Frantisek Zahalka and Radek Enzl

Škoda Research Ltd.
Tylova 57, Plzeò, 316 00, Czech Republic
sarka.houdkova@skoda.cz

The tribological properties of parts surface, namely their wear resistance and friction properties,
are in many cases determining for their proper function. To improve surface properties, it is
possible to create hard, wear resistant coatings by thermal spray technologies. Using these versatile
coatings it is possible to increase parts lifetime, reliability and safety. To predict their behavior,
lifetime and application area of thermally sprayed coatings it is necessary to completely understand
the relationships between technology, process parameters, microstructure and properties of the
coatings. In the case of thermally sprayed coatings evaluation it is necessary to take into
consideration their unique lamellar microstructure. Together with materials characteristics, such as
hardness, Young modulus of elasticity or fracture toughness, the coatings porosity, cohesive
strength, content of oxides and other microstructure defect also plays its role [1,2].
In the article, the set of thermally sprayed coatings were tested by tree different method to
prove their wear resistance to various load conditions. The sliding wear conditions were
represented by pin-on-disc wear test [ASTM G 99], the abrasion wear condition by dry sand rubber
wheel test [ASTM G 65] and slurry wear test [ASTM G 75].
During the pin-on-disc test, the coefficient of friction between Al2O3 pin and grinded coatings
surface wear recorded. After the test the wear tracks profiles were measured and wear rate for each
coating was calculated according to [2]:

V
K
s.L (1)
where K is wear rate
V is volume loss
L is load
s is sliding distance
In the case of dry sand rubber wheel test and slurry wear test, the coated samples were
weighted periodically in pre-set intervals. It enables to construct a wear curves, representing
coatings wear resistance.
The results pin on disc measurement showed, that the lowest coefficient of friction were 0,365
for WC-Co coating, followed by 0,576 for NiCrSiB coating and 0,647 for Cr3C2-NiCr coating.
The highest value of coefficient of friction, 0,857, was measured for steel coating AISI 316L.
The wear rates, results from pin on disc test, can be seen from Table 1. The highest wear
resistance were proved for both cermet coatings, WC-Co and Cr3C2-NiCr. The worst results
showed AISI 316 L coating. Combined with the results of coefficient of friction measurements, the
best, from tested coatings, for sliding wear is WC-Co coating, followed by Cr3C2-NiCr coating.
890 S. Houdkova et al.

TABLE 1. Pin on disc wear rates.

In the Fig. 1, the wear curves, based on the results of dry sand rubber wheel test can be seen.

FIGURE 1. Dry sand-rubber wheel abrasion wear test results.

The slurry wear test showed very similar results to dry sand abrasion test. Again, as the best
alternative for solution of abrasion wear are cermet coatings, based on hard particles in ductile
matrix. The advantage of WC-CoCr and Cr2C3-NiCr coating is in their resistance to corrosion
attacks.

References
1. EricksonL. C.- Hawthorne H. M.- Troczynski, T.: Wear 250, 2001, p. 569-575
2. Holmberg K.- Matthews A.: Coatings Tribology, Elsevier, Amsterdam, 1998

The article was prepared thanks to financial support of project no. MSM 4771868401.
 9. Micromechanisms in Fracture and Fatigue 891

CRACK PROPAGATION RESISTANCE AND DAMAGE MECHANISMS IN

NUCLEAR GRAPHITE

A. Hodgkins, J. Marrow, P. Mummery, A. Fok1 and B. J. Marsden1

School of Materials, Materials Performance Centre, University of Manchester
1School of Mechanical and Civil Engineering, Nuclear Graphite Research Group,

University of Manchester
Andrew.hodgkins@postgrad.manchester.ac.uk, James.Marrow@manchester.ac.uk
Paul.Mummery@manchester.ac.uk, Alex.Fok@manchester.ac.uk,
Barry.Marsden@manchester.ac.uk

Sharply rising resistance to crack propagation in polygranular nuclear graphite, immediately

following crack initiation from a stress concentrator has been suggested to be due to bridging by
filler particles in the wake of the crack. In other brittle materials in which such rising resistance
(R-curve) is observed, the length of the bridging zone is linked to the crack extension over which
rising resistance occurs. This is termed Stage I of the R-curve. Following this stage II of the R-
curve, a plateau, is commonly observed (Fig. 1).
Tomography and optical strain mapping techniques have been used to monitor damage, and the
permanent strains due to microstructure damage and crack propagation. The aim of this work is to
develop a model for the effects of microstructure on resistance to crack propagation.
Crack propagation tests have been carried out using optical microscopy and high resolution X-
ray tomography to determine the cause of the continued rise in resistance in stage II. The
observations show that there is a physical bridging zone, characterised by small ‘pinning’
ligaments in the ‘binder’ phase in the first 5 to 6 mm from the crack tip. These, and microstructure
damage ahead of the notch, are the likely cause of the stage I R-curve. The tomography
observations have shown the effect of microstructure damage and have also indicated the existence
of bridging regions up to 40mm behind the crack tip (Fig 2). Optical microscopy shows that the
crack is actually continuous in these regions (Fig.3), however. It is suggested that frictional
contact may occur between the two surfaces of the crack, and is the cause of the stage II R-curve.
The crack opening displacement in these regions is below the resolution limit of tomography, thus
apparent bridging is observed.

Fig 1: R-curve for compact tension specimens. Crack monitoring by compliance, surface strain
mapping (ESPI) and 3D tomography (XRìT).
892 A. Hodgkins et al.

Fig 2: Tomography images of a propagating crack. Crack bridging is observed.

Fig 3: Frictional contact, 40mm behind the crack tip.

 9. Micromechanisms in Fracture and Fatigue 893

ENVIRONMENT-ASSISTED CRACKING OF HIGH-STRENGTH MAGNESIUM

ALLOYS WE43-T6

A. Ahmad and T. J. Marrow

School of Materials,
University of Manchester, M1 7HS, UK
azmi.ahmad@postgrad.manchester.ac.uk, james.marrow@manchester.ac.uk

Some magnesium alloys experience environment assisted cracking (EAC) in ambient atmospheric
air, and hydrogen embrittlement is the dominant mechanism [1-3]. The highest susceptibility has
been found in aluminum-containing alloys (6% max) [4-5]. EAC has also been observed in pure
magnesium [3]. The EAC sensitivity of aluminum-free rare earth-containing alloys, such as
WE43 has not been previously investigated in detail. Previous work has shown that plastic strain
nucleates EAC in WE43-T6 by cracking of the intergranular intermetallic [6]. This paper reports a
quantitative study of the crack nucleation mechanism in WE43-T6. This is part of a project that
aims to predict the probability of nucleation and propagation of cracks at notches and micro-
shrinkage porosity in cast engineering materials.
Notched specimens of WE43-T6 alloy (4.2 wt% Y, 2.3 wt% Nd, 0.7 wt% Zr, 0.8 wt% HRE,
bal. Mg) were static-fatigue tested in ambient air. Static fatigue was only observed at stresses
close to the dynamic tensile fracture stress, but stable cracks were observed at the notch root in
run-out specimens, which did not fail within 100 hours (1). The statistical distributions of stable
cracks initiated at the notch root below the failure stress and the clusters of intergranular
intermetallic phase have been analysed, and compared. The strain at notch root, as a function of
applied stress, was measured using Electronic Speckle Pattern Interferomery (ESPI) (2), and has
been used to interpret the fracture behaviour of different notch geometries.
The longest crack length in each surviving specimen was independent of the notch tip strain
and was consistent with the longest intermetallic cluster size. At low notch strain, the cracks are
shown to come from the same population, which is the same as the intermetallic cluster population.
With increasing strain, the maximum crack length and crack density increases and the crack
population distribution changes (3). The critical event that controls specimen failure in static
fatigue is concluded to be coalescence of crack nuclei, rather than propagation of a single critical
defect.

FIGURE 1: Cracks nucleated by brittle intermetallic clusters, a) metallographic section at notch

tip, b) Scanning electron microscopy of notch tip.
894 A. Ahmad and T. J. Marrow

FIGURE 2: ESPI measurement of notch tip strain: a) strain map of the notch, b) the peak strain at
the notch root as a function of stress.

FIGURE 3: (a) Effect of notch tip strain on maximum observed crack length at notch root, (b)
maximum observed crack length compared to the intermetallic cluster population.

References
1. Oryall, G. and Tromas, D., Corros. Sci., 27, 335-341, 1971.
2. Chakrapani, D.G. and Pugh, E.N., Metal. Trans., 7A, 173-176, 1976.
3. Stampella, R.S., Procter, R.P.M. and Ashworth, V., Corros. Sci., 24, 325-341, 1984.
4. Makar, G.L., Kruger, J. and Sieradzki, K., Corros. Sci/, 34, 1311-1342, 1993.
5. Emley, E.F., Principles of Magnesium Technology, London, 1966.
6. Marrow, T.J., Bin Ahmad, A., Khan, I.N., Sim, S.M.A. and Torkamani, S., Mater. Sci. and
Engng A, vol. 387-89, 419-423, 2004.
 9. Micromechanisms in Fracture and Fatigue 895

EFFECTS OF SURFACE FINISH ON THE FATIGUE LIMIT IN AUSTENITIC

STAINLESS STEELS (MODELLING AND EXPERIMENTAL OBSERVATIONS)

M. Kuroda, T. J. Marrow and A. Sherry

School of Materials, Materials Performance Centre, University of Manchester
Masatoshi.Kuroda@manchester.ac.uk, James.Marrow@manchester.ac.uk,
Andrew.Sherry@manchester.ac.uk

The fatigue limit of austenitic stainless steels has been reported to be influenced by surface
roughness, work hardening, microstructure and residual stresses induced by working. However,
the available data in the literature is complicated by the interactions of these factors, which results
in difficulty in obtaining a fundamental understanding of the mechanism. Therefore, models and
experiments which can elucidate the contribution of each surface effect to the fatigue limit is
required. Microstructure models for the effect of these factors on short fatigue cracks have been
developed for aluminium alloys [E R de los Rios, M Trull, A Levers, “Modeling fatigue crack
growth in shot-peened components of Al 2024-T351”, Fatigue Fract Engng Mater Struct, 23
(2000) 709-716. , C Vallellano, A Navarro, J Dominguez, “Fatigue crack growth threshold
conditions at notches part I: theory”, Fatigue Fract Engng Mater Struct, 23 (2000) 113-121. ], but
have not been sustantially verified by experiment, not demonstrated to be applicable to other
alloys. This is the aim of this investigation, which has focussed on the parameters of surface
roughness (i.e. stress concentration factor in terms of notch depth and root radius), residual stress
(i.e. depth profiles) and microstructure (i.e. distribution of crack arresting/retarding boundaries).
The response surface methodology (RSM) coupled with central composite design (CCD) was
employed to prepare austenitic stainless steel fatigue specimens with a controlled range of surface
characteristics. A wide range of conditions were produced by changing the final lathe cutting
conditions of spindle speed, feed rate and cutting depth. The fitting of the response surface model
for the data was statistically studied by analysis of variance (ANOVA). The response surface
model (1) adequately represented the largest peak to valley height (roughness Ry) and the axial
residual stress (determined by X-ray Diffraction), and was interpolated to design fatigue specimens
of significantly different roughness and surface residual stress.
The residual stress profiles below the surface were measured by X-ray Diffraction (2), and
employed with surface profile measurements (3) and microstructure characterisation to predict the
threshold fatigue stress for crack propagation as a function of short crack length. The staircase
method was used to measure fatigue limit. This was used to test the model predictions, together
with measurement of the stable crack nucleus population below the fatigue limit. Electron
backscatter diffraction was used to characterise the microstructure and near surface plastic
deformation. The effects of annealing, to remove plastic strain and residual stress effects, have
also been investigated. The results show that the simple microstructure model does not necessarily
apply to austenitic stainless steels, and the effects of additional factors such as the deformation
structure must also be considered.
E R de los Rios, M Trull, A Levers, “Modeling fatigue crack growth in shot-peened
components of Al 2024-T351”, Fatigue Fract Engng Mater Struct, 23 (2000) 709-716.
C Vallellano, A Navarro, J Dominguez, “Fatigue crack growth threshold conditions at notches
part I: theory”, Fatigue Fract Engng Mater Struct, 23 (2000) 113-121.
896 M. Kuroda et al.

Fig. 1: Response surface of roughness and surface axial residual stress for the combined effects of
feed rate and cut depth. Conditions 1 to 6 for fatigue study are identified.

Fig. 2: Variation of axial residual stress below the machined surface for various conditions.

Fig. 3: Surface roughness characterised by scanning electron microscopy stereo imaging.

 9. Micromechanisms in Fracture and Fatigue 897

INTERGRANULAR STRESS CORROSION CRACK PROPAGATION IN

SENSITISED AUSTENITIC STAINLESS STEEL (MICROSTRUCTURE
MODELLING AND EXPERIMENTAL OBSERVATION).

T.J. Marrow, L. Babout, A.P. Jivkov, P. Wood, D. Engelberg, N. Stevens, P.J. Withers
and R.C. Newman
School of Materials, University of Manchester, UK and
Department of Chemical Engineering and Applied Chemistry, University of Toronto, Canada
James.Marrow@manchester.ac.uk

Stress corrosion cracking is a life-limiting factor in many components of nuclear power plant
in which failure of structural components presents a substantial hazard to both safety and economic
performance. Uncertainties in the kinetics of short crack behaviour can have a strong influence on
lifetime prediction, and arise due both to the complexity of the underlying mechanisms and to the
difficulties of making experimental observations.
This paper reports a research programme into the dynamics and morphology of intergranular
stress corrosion cracking in austenitic stainless steels, which makes use of recent advances in high
resolution X-ray microtomography. In particular in-situ, three dimensional X-ray tomographic
images of intergranular stress corrosion crack nucleation and growth in sensitised austenitic
stainless steel (FIGURE.1) provide evidence for the development of crack bridging ligaments,
caused by the resistance of non-sensitised special grain boundaries. (FIGURE.2)
A simple grain bridging model, introduced to quantify the effect of crack bridging on crack
development, has been assessed for thermo-mechanically processed microstructures via statically
loaded room temperature simulant solution tests and as well as high temperature/pressure
autoclave studies. Thermo-mechanical treatments have been used to modify the grain size, grain
boundary character and triple junction distributions, with a consequent effect on crack behaviour.
Two and three-dimensional finite element models of intergranular crack propagation have been
developed (FIGURE.3), with the aim of investigating the development of crack bridging and its
effects on crack propagation and crack coalescence.

FIGURE. 1: Successive intervals during in-situ tomography observations of stress corrosion crack
growth showing failure of crack bridging ligaments
898 T. J. Marrow et al.

FIGURE. 2: Fractography of ductile crack bridging ligament due to low energy grain boundary
segments, associated with annealing twins.

FIGURE. 3: Three dimensional intergranular stress corrosion crack model, predicting the
development of crack bridging ligaments.
 9. Micromechanisms in Fracture and Fatigue 899

IDEAL STRENGTH OF NANOSCALE THIN FILMS

Takayuki Kitamura, Yoshitaka Umeno and Akihiro Kushima

Department of Mechanical Engineering and Science, Kyoto University.
Yoshidahonmachi Sakyo-ku Kyoto, Kyoto, Japan.
kitamura@kues.kyoto-u.ac.jp

With the development of nano fabricating technology, we can now obtain materials with structures
in nano scales, and there is increasing interest in understanding the mechanical properties of such
materials from atomistic point of view. The ideal strength was originally defined as the stress or
strain at which perfect crystal became mechanically unstable with respect to arbitrary
homogeneous deformation[1]. There have been quite a few numbers of researches investigating
the ideal strength of materials[2-7] since this is a fundamental mechanical property of the material.
However, nano-structured materials do not have the three dimensional periodicity as perfect
crystals, and investigating the ideal strength of perfect crystal is not sufficient for understanding
the mechanical property of such materials. It is required to investigate the effect of structures on
the mechanical property. As a first step, we can consider evaluating the ideal strength of the
material with high symmetry which is a basic structure. Here we expanded the definition of ideal
strength to include the strength of highly symmetric structured materials namely ideal structure.
We conduct ab initio calculation for precise evaluation of ideal strength of the material with ideal
structure to obtain the fundamental mechanical property of nano-structured materials.
In this study, we evaluated ideal strength of Si nanoscale thin films with p(2×1) asymmetric
(100) surface under tensile deformation to investigate the effect of surface structure on mechanical
strength. We conducted tensile simulation for films with 6, 10, and 14 atomic layers. Fig. 1 shows
relationships between tensile stress and the strain of Si nanofilms under [011] tension. The result
of the bulk is also shown for comparison. At low strain region, where the strain is less than 0.1,
you cannot observe significant difference in the curve between the results of nanofilms and the
bulk.

FIGURE 1. Tensile stress-strain curve of Si nanofilms under [011] tension.

900 T. Kitamura et al.

FIGURE 2. Change in charge density of Si nanofilms under [011] tension.

The maximum stresses of Si nanofilms are lower than that of the bulk and decrease as the
thickness of the film decreases. Fig. 2 shows change in electronic density of Si nanofilms during
tension. The isosurfaces with the charge density of 0.425 X 103 nm-3 are drawn. In 14 layered film,
charge densities of all bonds decrease equally at unstable deformation as in the bulk. In the case of
10 and 6 layered films, breakings of bonds propagate along dashed lines and lead to fracture. This
difference in the fracture mechanism causes the thinner film to have lower ideal strength.
However, the ideal strength of Si nanofilm with 14 atomic layers which is 1.8 nm thick is about
96% of that of the bulk. These results indicates that the effect of surface structure on the
mechanical property can be observed when the thickness of the film is fairly small, less than 1.8
nm.

References
1. Born, M. and Huang, K., Dynamical Theory of Crystal Lattices, Oxford UP, 1954.
2. Ogata, S., Li, J. and Yip, S., Science, 298, 807-811, 2002.
3. Sob, M., Wang, L. and Vitek, V., Mater. Sci. Eng., A 234-236, 1075-1078, 1997.
4. Sandera, P., Pokluda, J., Wang, L and Sob, M., Mater. Sci. Eng.. A 234-236, 370-372, 1997
5. Clatterbuck, D., Chrzan, D. and Sob, M., Acta. Mater., 51, 2271-2283, 2003.
6. Roundy, D. and Cohen, M., Phys. Rev., B 64, 212103, 2001.
7. Umeno, Y. and Kitamura, T., Mater. Sci. Eng., B 88(1), 79-84, 2002.
 9. Micromechanisms in Fracture and Fatigue 901

TOUGHNESS VARIABILITY

R. Bouchard, G. Shen and W.R. Tyson

Materials Technology Laboratory
Natural Resources Canada
568 Booth St., Ottawa, Canada K1A 0G1
btyson@nrcan.gc.ca

The weakest-link model of brittle fracture initiation introduced by Beremin et al. [1] has had
substantial success in describing the inherent variability (scatter) in fracture toughness values for
steel samples failing by cleavage. A project was initiated at MTL/CANMET to explore the
applicability of the weakest-link model to Charpy absorbed energy. In the course of this work,
fracture toughness tests using standard three-point-bend bars were performed. The results of these
tests were unexpected, and are reported here.
The tests were performed using a carbon/manganese structural steel in the as-rolled condition.
The microstructure was polygonal ferrite with slightly banded pearlite (volume fraction 19%), with
average ferrite grain size 12:m (surface) and 14.5:m (centre) as shown in Fig. 1.

Fig. 1. Optical micrographs of plate microstructure.

Fracture toughness tests were performed following ASTM E 1921 [2], which allows
determination of a reference temperature T0 in the transition range at which the median toughness
of 1T size specimens is 100 MPa—m. The test temperature (-110qC) was chosen to be close to T0,
and a total of 26 tests were carried out. All results were valid according to E 1921. Data was ranked
and cumulative probability calculated as described in E 1921. The results are shown in Fig. 2.

Fig. 2. Weibull plot of fracture toughness results

902 R. Bouchard et al.

The weakest-link model predicts that results should fall on a straight line of slope four on a
Weibull plot such as Fig. 2. However, the slope of the best-fit line through the data of Fig. 2 is
1.86, differing significantly from 4. Possible reasons for the discrepancy are discussed. The most
likely reason is material inhomogeneity, although there is no microstructural evidence for this and
samples were taken from a nominally uniform region of the plate.

References
1. Beremin, F. M., “A Local Criterion for Cleavage Fracture of a Nuclear Pressure Vessel
Steel”, Metallurgical Transaction A, Volume 14A, 1983, pp. 2277-2287.
2. “E 1921-02: Standard Test Method for Determination of Reference Temperature, T0, for
Ferritic Steels in the Transition Range”, Annual Book of ASTM Standards, Volume 3.01,
America Society for Testing and Materials, PA, U.S.A., 2002.
 9. Micromechanisms in Fracture and Fatigue 903

THERMO-MECHANICAL BEHAVIOUR OF NANOSTRUCTURED COPPER

C. Duhamel, S. Guerin, M. J. Hytch and Y. Champion

Centre d'Etudes de Chimie M?tallurgique - CNRS
15 rue Georges Urbain 94407 Vitry-sur-Seine, cedex, France
cecilie.duhamel@glvt-cnrs.fr, sandrine.guerin@glvt-cnrs.fr, martin.hytch@glvt-cnrs.fr,
champion@glvt-cnrs.fr

A material is an “ultrafine grained material” when substantial deviation(s) of its behaviour is

observed compared to the “standard” coarse grained materials properties. In plasticity, ultrafine
grained metals and alloys exhibit at room temperature new behaviours such as near-perfect elasto-
plasticity, strain rate sensitivity and high mechanical performances. For copper, these features are
detected for grain size below around 300 nm ; this threshold is connected to the length scale
associated to the lattice dislocation of the metal, Champion et al [1]. Nanostructured metals and
alloys open interesting perspectives of applications as far as their special mechanical behavior can
be controlled by orientation of their nanostructure and chemistry.
One of the main key-point that impedes the use of nanostructures, is the problem with ductility,
which limits the shaping and the use when plastic deformation occurs. Ductility may happen in
nanostructures in relation with an increase of the strain-rate sensitivity, m, which occurs at room
temperature when grain size is below around 300 nm and at relatively low strain rate, H , below
10-4 s-1, Conrad and Jung [2]. However, as early shown by Woodford, [3] and in connection with
the Hart criterion, Hart [4] significant elongation is obtained when the strain-rate sensitivity
becomes sufficiently high: for example, 100% of elongation should be expected when m is around
0.2 and as well known as superplasticity behavior, few hundred per cents may be expected when m
is larger than 0.3. Literature data show that significant values of m are obtained at extremely low
stain rate H < 10-5 s-1, for nanostructured and this effect is enhanced with decreasing the grain size
in the ultrafine domain, when lattice dislocations are still present.
Some strategies to increase ductility have been already examined based on the fabrication of
“complex” nano-architectures where regions are capable of stress relaxation, as for instance,
bimodal nano-micro grained size metals, severe plastic deformed metals with dislocations
saturated grain-boundaries and dentritic nanostructures which have the properties of limiting the
propagation of shear bands and associated mechanical instabilities.
On the other hand, direct use of strain-rate sensitivity promoting ductility may be also
exploited at high temperature. Preliminary studies showed as expected that m increases with the
temperature, with the main drawback that nanostructures exhibit a low thermal stability. Use of
nanostructures at moderate temperatures would necessitate investigations of the mechanisms
involved in the plastic deformation at various temperatures. Understanding of the mechanisms
would lead to engineering strategies involving the control of the structure and the chemistry, to
increase the stability combined with appropriate mechanical properties.
904 C. Duhamel et al.

FIGURE 1. strain-rate sensitivity m versus temperature for different strain rates.

To investigate the mechanisms of plasticity, measurements of the strain rate sensitivity were
carried out on nanostructures copper prepared by powder metallurgy from nanocrystalline copper
powders. Metals have grain size of 90 nm. Experiments were carried out between room
temperature and 120°C and for strain rate ranging from H =1u10-5 s-1 and 2u10-3 s-1 (Fig. 1).
Analysis of the data leads to parameters such as activation volume and energy which give an
insight of the micromechanisms involved in the plastic deformation.

References
1. Champion, Y., Langlois, C., Guérin, S., Lartigue-Korinek, S., Langlois, P. and Hÿtch, M.J.,
Mater. Sci. Forum vol. 482, 71-76, 2005.
2. Conrad, H. and Jung, K., Scripta Mat. available online 2005.
3. Woodford, D.A., Trans ASM, vol 62, 291-294, 1969.
4. Hart, E.W., Acta Metal. vol. 15, 351-355, 1967.
 9. Micromechanisms in Fracture and Fatigue 905

SOME INSIGHTS INTO FATIGUE CRACK INITIATION STAGE

H. Alush and Y. Katz

Negba, Beer Sheva 84230, Israel
katzroy@012.net.il

Fatigue crack initiation stage is competitive and requires a full screening of potential initiation
sites. In a more comprehensive view, the role of the mechanical driving force is generally
broadened, thus, deformation/environment interactions become significant. Besides better
fundamental understanding as related to the fatigue process other considerations emerged like the
distinction between initiations or propagation controlled processes. This has direct implications on
the fatigue life assessment. The current phenomenological study selected polycrystalline, pure
copper of 99.98 %, a typical FCC model system in the intensive volume of fatigue investigation.
The material with various impurity traces included at the most, aluminum of 100ppm. After heat
treatment microstructure of 50-80 µm grain size was tested at ambient temperature. Standard
mechanical properties were determined and strain controlled fatigue tests were performed. The
latter in the range of about 10-4 to 10-3 plastic strain amplitudes. The specimens consisted of
uniform and cylindrical geometry, 3mm in diameter and tension/compression cyclic tests of load
ratio, R=-1 at 3Hz were conducted. Four fold interrelated information levels were searched
namely, mechanical response, low energy dislocation structure development, slip upset and crack
initiation tracking. These were achieved by closed loop strain controlled devices, Transmission
Electron Microscopy (TEM), Scanning Electron Microscopy (SEM) and Micro Probe Microscopy
(MPM). In addition crack on a micron scale resolution has been tracked by one stage replication
technique. For some isolated cases crack initiation tracking was assisted by Acoustic Emission
(AE) spectra. The slip upset was observed by centering on fine scale features that enabled a
quantitative approach to be analyzed by defining the local strain. As addressed previously by
Harvey et al [1], the local strain has been defined by the slip upset height and the slip spacing given
for one cycle in equation (1):

h
f 'H p
s (1)
where h is the slip height or the displacement, s the slip spacing, p the imposed plastic strain
amplitude and f the slip upset efficiency. The effective term f is introduced since only a fraction of
the dislocations can emerge at the free surface. This due to forest dislocation interactions,
dislocation annihilation in the bulk or orientation constraint. In this context, the study was
extended also to residual stresses either thermal or by mechanical surface modification. The
mechanical response clarified the saturation stress and the corresponding saturation cycles. Low
energy dislocation structures caused by cyclic loading were substantiated by loop patch, veins and
cells as well as Persistent Slip Bands (PSB). Higher strain amplitudes resulted in elongated cells,
labyrinth and symmetrical cells. In the current specific system, crack initiation followed a
sequence of events. The pre crack plastic deformation was confined to discreet zones highly
related to crystal plasticity habits enhancing as such the crack onset. Consequently, micro crack
coalescence occurred in the micro domain. The MPM findings indicated that the initiation life
could be formulated with analogy to the total fatigue life based on a cumulative damage model.
For the sake of consistency, the MPM results confirmed the role of compressive residuals on the
local strain. Under the current imposed residuals in the order of 0.7y after 104 cycles the local strain
reduced by 50%. Concerning the formulation form of the fatigue initiation life, damage
906 H. Alush and Y. Katz

accumulation efficiency factor was introduced beside the imposed cumulative plastic strain. Note
that the present approach is geometrical, assisted by novel techniques. Finally, it seems advisable
to compare the current approach to Muras School [2, 3] that utilized an energetic model. This
comparative study actually supports the current quantitative approach that remains consistent with
imposed residuals affecting the fatigue process. The notion that fatigue crack initiation stage is
frequently less emphasized is mainly related to its complexity. In fine scale segments applications
like thin films, structural integrity aspects can in fact be dominated by the crack initiation life.
Accordingly, the following was concluded. Fine feature observations assisted by novel techniques
promise further insights into the fatigue initiation stage. Even in highly smooth surfaces plastic
dissipation is manifested by low energy dislocation structures and slips upset. Internal stored
energy or geometrical changes at the free surface resulted in the initiation of discontinuities. Post
crack initiation follows other level of considerations; crack tip shielding or micro crack
coalescence might influence the propagation kinetics. Nevertheless, the current study facilitated
beside experimental confirmation attempt in modeling, simulation or life predictions. At least the
hierarchy of the critical parameters as related to the initiation stage can be evaluated by cause and
further discussed.

References
1. Harvey,S.E., Marsh,P.G and Gerberich,W.W.,Acta.Metall.Mater,vol.42,3493-3502,1994.
2. Mura,T. and Nakasone,Y.,J.Appl.Mech.vol.57,1-6,1996.
3. Vekataraman,G., Chung,Y.M.,Nakasone,Y. and Mura,T.,Acta.Metall et Mater.vol.39, 2631-
2638,1991.
 9. Micromechanisms in Fracture and Fatigue 907

FATIGUE BEHAVIOUR OF METALLIC MATERIALS EXPOSED TO HIGH

PRESSURE HYDROGEN ENVIRONMENTS

Yoji Mine, Saburo Matsuoka, Yukitaka Murakami Chihiro Narazaki1 and Toshihiko Kanezaki1
Department of Mechanical Engineering Science, Kyushu University,
Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan
1Graduate School of Engineering, Kyushu University,

Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan

mine@mech.kyushu-u.ac.jp

In order to solve the global warming problem, development and commercialisation of fuel cell
(FC) systems are being promoted. In the FC systems, many stainless steels are used for the
components such as the liners of ultra-high pressure vessels, piping, bearings, and springs. Such
components are directly exposed to hydrogen environments under cyclic loading. It has been
reported that hydrogen degrades the mechanical properties, especially ductility of metallic
materials [1-3]. Moreover, the degradation of fatigue strength of metallic materials due to
hydrogen intrusion is a matter of concern from a practical point of view. In our previous
investigations [4-7], hydrogen was artificially (electro-chemically) charged into the specimens of
several candidate materials, and it was revealed that hydrogen affects the slip band morphology
and fatigue crack growth behaviour. In this study, hydrogen intrusion and fatigue crack growth
behaviour in stainless steels exposed to high pressure hydrogen environments were investigated at
room temperature and in laboratory air to clarify the effect of the intruded hydrogen on the fatigue
crack growth.
The materials used in this study were round bars made of austenitic (types 304, 316, 316L and
310S), ferritic (type 405) and martensitic (type 440C) stainless steels. The austenitic stainless
steels were solution-treated, ferritic stainless steel was annealed, and the martensitic stainless steel
was quenched and tempered. The Vickers hardness of type 304 and type 316L stainless steel was
HV = 176 (SUS304) and HV = 157 (SUS316L). Hydrogen was charged into specimens by
exposing them to high pressure hydrogen gas. The hydrogen exposure experiments were
conducted at a pressure of 18, 25 and 40 MPa, at a temperature of 383 to 388 K and for a holding
time of 100 to 115 h. Disks with a diameter of 7 mm and a thickness of 0.8 mm were used for the
hydrogen exposure experiments. For the comparison, cathodically-charged specimen for fatigue
test was prepared with platinum electrode at a current density of 27 A/m2 in a sulfuric acid solution
(pH = 3.5) at 323 K for 672 h. The hydrogen content was measured by the thermal desorption
spectrometry (TDS) at a heating rate of 0.33 K/s. The specimens for fatigue test were
manufactured and were polished with #2000 emery paper and buff. A small artificial hole was
drilled onto the specimen surface to limit the crack initiation sites. In the hydrogen-exposed and
cathodically-charged specimens, a small artificial hole was introduced after hydrogen charge.
Tension-compression fatigue tests were carried out at room temperature in laboratory air. All the
fatigue tests were conducted at a stress ratio, R = -1 and at a frequency of 1.2 to 2 Hz to control a
rise in the temperature. Crack growth behaviour was investigated by the replica method.
The conclusions can be summerised as follows:
(1) From the results of the hydrogen exposure experiments, the hydrogen content increased
with an increase in the hydrogen exposure pressure up to 40 MPa. The increment in hydrogen
content was dependent on the microstructure and chemical composition. Austenitic stainless steels
contained more hydrogen compared with martensitic and ferritic steels by approximately an order
of magnitude at the same hydrogen exposure conditions. Hydrogen thermal desorption spectra
908 Y. Mine et al.

showed exposing the stainless steels to high pressure hydrogen environments raised the height of
the lower-temperature peak of H2 ion intensity, i.e. diffusive hydrogen.
(2) In SUS304, the fatigue crack growth rates of the hydrogen-exposed and cathodically-
charged specimens were approximately twice higher than those of the uncharged specimen; on the
other hand, in SUS316L, acceleration in the fatigue crack growth of the hydrogen-exposed
specimen was observed at a short crack length below ~400 Pm. The aspect ratio for the crack
observed on the fracture surface, b/a significantly reduced in the hydrogen-exposed and
cathodically-charged specimens in SUS304, while in SUS316L, the aspect ratio of the hydrogen-
exposed specimen was slightly smaller than that of the uncharged specimen. These results suggest
that the concentrated hydrogen in the surface layer accelerates the fatigue crack growth in the
austenitic stainless steels. The differences in the crack growth acceleration and the aspect ratio for
the crack between SUS304 and SUS316L presumably arose from those in the content and
distribution of the intruded hydrogen and/or the susceptibility to hydrogen.
(3) A distinct difference in the appearance of slip bands was observed between the hydrogen-
charged (due to hydrogen exposure and cathodic charge) and uncharged specimens in both
SUS304 and SUS316L. In particular, the difference was pronounced in SUS304. In the uncharged
specimens, slip bands were densely and widely distributed; on the other hands, slip bands were
discrete in the hydrogen-charged specimens. The crack growth morphology in the hydrogen-
charged specimens was also different from that in the uncharged specimens. The fatigue cracks in
the hydrogen-charged specimens grew more straightly and were sharper compared with the
uncharged specimens. These results imply that the intruded hydrogen into austenitic stainless
steels influences the dislocation mobility to cause the slip localisation.

References
1. Vennett, R. M. and Ansell, G. S., Trans. ASM, vol. 60, 242-251, 1967.
2. Benson Jr., R. B., Dan, R. K. and Robert Jr., L. W., Trans. Metall. Soc. AIME, vol. 242, 2199-
2205, 1968.
3. Herms, E., Olive, J. M. and Puiggali, M., Mater. Sci. Engng. A, vol. 272, 279-283, 1999.
4. Kanezaki, T., Mine, Y., Fukushima, Y. and Murakami, Y., Proceedings of the 15th European
Conference of Fracture (ECF15), CD-ROM, 2004.
5. Uyama, H., Mine, Y. and Murakami, Y., Proceedings of the 15th European Conference of
Fracture (ECF15), CD-ROM, 2004.
6. Murakami, Y. and Matsunaga, H., Proceedings of the Third International Conference on Very
High Cycle Fatigue (VHCF-3), 322-333, 2004.
7. Murakami, Y., Proceedings of the 11th International conference on fracture (ICF11), CD-
ROM, 2005.
 9. Micromechanisms in Fracture and Fatigue 909

IN-SITU INVESTIGATIONS OF THE FRACTURE MECHANISMS AT VARIOUS

LENGTH SCALES

Zbigniew Pakiela, Witold Zielinski and Krzysztof J. Kurzydlowski

Faculty of Materials Science and Engineering, Warsaw University of Technology
Woloska 141, 02-507 Warsaw, Poland
zpakiela@inmat.pw.edu.pl

One of the main features of small crack growth are “slow-downs” observed when they approach
microstructural barriers such as grain boundaries. In-situ observations of crack growth can deliver
valuable information about fracture behaviour of the investigated materials.
In this paper the authors present results of the in situ observation of the crack growth in various
materials, at various length scales and under various modes of loading. In order to approach better
understanding of the fracture behaviour light microscope (LM) with in-situ recording device,
scanning electron microscope (SEM) equipped with a tensile test machine and transmission
electron microscope (TEM) with a straining holder have been utilized.
In situ investigations by light microscopy
Observations were carried on pure Cu of bimodal grain size with microcrystalline grains
embedded in submicrocrystalline matrix. As mechanical properties of the materials with coarse
and small grains are different, crack propagation should be different in the fine grained matrix and
coarse grained islands, what is important from material design point of view. The crack growth in
such materials is still not well described by classical fracture theory, so authors focused on this
problem in this paper. For quantitative description of strain field the Digital Image Correlation
method proposed by Peters and Ranson [1] was used.

FIGURE 1: a) experimental setup, b) strain field near a crack tip in nanostructureed Cu with
bimodal microstructure.

In situ SEM fracture studies

In situ SEM fracture studies were performed on the coarse grain Al sample directly in the
vacuum chamber of SEM, equipped with tensile test machine. The sample with a thickness of 0.2
mm was pre-notched by means of dimple obtained by electro polishing. The processes of strain
localization, crack initiation and propagation have been investigated with details showing
distribution of slip bands.
910 Z. Pakiela et al.

FIGURE 2. SEM images of the slip bands network evolution and the crack initiation in Al
specimen during in situ SEM straining: a) arrangements of the slip bands at the very early stage of
deformation; b) a nucleus of fracture.

In situ TEM deformation study

316 type stainless steel for the in situ TEM dislocation pile up against grain boundary studies
as an appropite material having low stacking fault energy has been chosen. An increase of
dislocation quantity in the pile up as well as slip of the dislocation in the opposite direction was
registered and analyzed in detail.

FIGURE 3. Dislocation network evolution during stainless steel in situ straining: a) dislocation
pile up against grain boundary: b) slip traces indicating dislocation motion

References
1. Peters, W.H., Ranson, W.F., Opt. Eng. vol. 21, 427–432, 1982.
 12. Interface Fracture and Behavior of Joints 911

ENVIRONMENTAL ATTACK AT POLYMER/METAL INTERFACES

A. J. Kinloch1, D. Bland1, K. T. Tan1 and J. F. Watts2

1Department of Mechanical Engineering, Imperial College London, Exhibition Road,

London SW7 2BX, UK

2School of Mechanical and Materials Engineering, University of Surrey, Guildford, Surrey, GU2

5XH, UK.
a.kinloch@imperial.ac.uk

There are many advantages that structural adhesives can offer compared to the more traditional
joining methods such as welding, bolting, mechanical fastening, etc. However, there are some
issues, which have so far limited the wider application of adhesives. The most important of these is
the lack of knowledge concerning the durability of adhesive joints upon exposure to an adverse
environment which invariably tends to lead to interfacial failure. This has led to extensive studies
aimed at evaluating the effects of moisture, and developing a better understanding of the
mechanisms of failure. Unfortunately, most of the laboratory test methods involve lengthy time-
scales and the results are often difficult to interpret quantitatively. Therefore, a most important
challenge facing the adhesive community is to develop accelerated test methods that can assess
quantitatively durability of adhesive joints in a relatively short-time scale and are able to predict
accurate the service life of adhesive joints under such conditions.
Experimental
The full details of the preparation of testing of the joints used for the fracture-mechanics tests
are given elsewhere Korenberg et al. [1]. An adhesively-bonded joint using a tapered-double
cantilever-beam (TDCB) specimen was employed. The substrates used throughout the present
work were manufactured either from steel (British Standard 970 070M55) or aluminium alloy
(British Standard 7075 (unclad)). The adhesive used was a hot-curing rubber-toughened epoxy
adhesive, which was based upon a dicyandiamide-cured diglycidyl ether of bisphenol-A epoxy.
Three types of joints were made: steel-steel, aluminium alloy-aluminium alloy and steel-
aluminium alloy joints. The last type will be denoted as ‘dissimilar-substrate’ joints. All substrates
were first degreased in a liquid bath of boiling 1,1,1-trichloroehtylene before they were grit-blasted
with 60Pm-78Pm mesh alumina particles. The specimens were tested under monotonic-loading
conditions, with a displacement rate of the crosshead ranging from about 0.1Pm/min to 10mm/min.
The tests were mainly conducted (a) in conditions of 21r1qC and approximately 55% relative
humidity (RH), and (b) in ‘wet’ conditions of immersion in water at 21r1qC.
Results and Discussion
The fracture energy, GC, of the various joints was measured in the two different environments
as a function of the crack velocity, a . The results were plotted on a double-logarithmic scale.
Three different regions of crack behaviour have been identified and have been labelled as ‘Region
I’, ‘Region II’ and ‘Region III’. ‘Region I’ was observed at low crack velocities and the crack grew
in a stable manner visually along the adhesive/substrate interface; and the data yielded linear
relationships between GC and a . In contrast, ‘Region III’ occurred at high crack velocities and
stick-slip crack propagation mode was observed, and the failure was mainly cohesive in the
adhesive layer. ‘Region II’ was the transition region between ‘Region I’ and ‘Region III’. The
reasons for these different regions, and the underlying mechanisms of failure which occurred, will
be discussed in the presentation. However, clearly ‘Region I’, where interfacial failure is observed,
912 A. J. Kinloch et al.

is where attention must be focused in terms of measuring the effect that an aqueous environment
may have upon the fracture performance of the joint.
In ‘Region I’, the relationship between GC and a in the ‘55% RH’ tests was found to be
controlled by the viscoelastic nature [3] of the epoxy adhesive, and was independent of the type of
substrate used. Also, in such tests the application of a stress and the presence of water molecules at
the crack tip was found to control the mechanism of failure of the joints. In the ‘wet’ environment
of immersion in water, all the various types of joint showed an inferior behaviour compared to the
tests undertaken in the ‘55% RH’ environment.
It is also noteworthy that, in ‘Region I’, the dependence of GC upon a for tests in liquid water
and water-saturated environments (i.e. the ‘RH’ environments) was very different. This implies
that the relationship log GC  log a in the ‘wet’ immersion tests was not merely dependent on the
concentration of water molecules. It is proposed that a subtle corrosion mechanism plays a crucial
role in the tests conducted in liquid water, i.e. the ‘wet’ tests. Such corrosion processes lead to the
production of hydroxyl ions at the crack tip which cause cathodic debonding to occur in the very
alkaline environment created in this region. This results in the segregation of charge-carrying
cations to the cathodic sites under the influence of the corrosion potential present in the system.
Indeed, such a model has been supported by detailed surface-analysis studies, where relatively
high concentrations of cations have been detected on the ‘wet’ fracture surfaces. Also, the physical
appearance of joints after the immersion tests exhibited classical cathodic debonding features. The
occurrence of corrosion was further supported by the good fit of the experimental data to the model
of environmental crack growth proposed by Wiederhorn [3].
The durability of the steel/epoxy/aluminium-alloy dissimilar-substrate joints tested in the ‘wet’
environment was relatively even poorer, when compared to that of the similar-substrate joints. It is
shown that this observation does not arise from residual stresses but from the occurrence of
additional cathodic corrosion occurring in the dissimilar-substrate joints in water. Indeed, a
relatively high concentration of hydroxyl ions was detected from the dissimilar-substrate joints
compared to the similar-substrate joints.
Finally, it is shown that the monotonically-loaded fracture-mechanics test methodology
provides a good accelerated test method for evaluating the durability of adhesive joints. It allows
an assessment of the environmental resistance of adhesive joints within a matter of days, as
opposed to the more typical accelerated ageing tests which involve exposing the joint, unstressed
in water, for many months.

References
1. Korenberg, C.F., Kinloch, A.J. and Watts, J.F., J. Adhesion, vol. 80, 169-201, 2004
2. Kinloch, A.J. and Young, R.J., Fracture Behaviour of Polymers, Applied Science Elsevier,
London, 1983.
3. Wiederhorn, S.S, J. Am. Ceramic Soc., vol. 50, 407-414,1969
 12. Interface Fracture and Behavior of Joints 913

MODELLING OF ELASTIC-PLASTIC PEEL TESTS FOR STRUCTURAL

ADHESIVES

A. J. Kinloch, H. Hadavinia, L. Kawashita, D. R. Moore and J. G. Williams

Department of Mechanical Engineering, Imperial College London, Exhibition Road,
London SW7 2BX, UK
a.kinloch@imperial.ac.uk

The adhesive fracture energy, Gc, of adhesive joints may be readily ascertained from linear-elastic
fracture-mechanics (LEFM) methods, and indeed a British Standard (BS7991-2001) now exists for
the LEFM Mode I value, GIc, largely as a result of the efforts of the European Structural Integrity
Society (ESIS) TC4 Committee, as described by Blackman and Kinloch [1]. Notwithstanding, the
LEFM test specimens are relatively complex and expensive to make and test, and many industries
would far prefer to deduce the value of Gc from the very common and widely used ‘peel test’.
The peel test is an attractive test method to assess the performance of a wide range of flexible
laminates and adhesive joints. However, although it is a relatively simple test to undertake, it is
often a complex test to analyse and thus obtain a characteristic measure of the toughness of the
laminate, or adhesive joint. The most successful approach that has been adopted is one based upon
applying a fracture-mechanics method using an energy-balance approach. A value of the adhesive
fracture energy, Gc, is thereby ascertained, which is the energy needed to propagate a crack
through unit area of the joint, either cohesively through the adhesive layer or along the bimaterial
interface. The value of Gc may be obtained via an analytical or a numerical analysis of the peel
test. However, the peel test invariably involves gross plastic deformation of the peeling arm, which
may account for up to about 80% of the measured peeling energy.
The value of Gc should be characteristic of the joint and, ideally, independent of geometric
parameters such as the applied peel angle, the thickness of the flexible substrate arm(s) being
peeled and the thickness of the adhesive layer. However, it is recognised that, since the value of Gc
includes plastic and viscoelastic energy dissipation which occurs locally at the crack tip, it will be
a function of the rate and temperature at which the peel test is conducted. (Only if such energy
losses are reduced to virtually zero, and the locus of joint failure is exactly along the bimaterial
interface, will the value of Gc be equivalent to the thermodynamic work of adhesion.)
The present paper will describe the use of novel analytical and numerical methods to deduce
the value of Gc from the elastic-plastic peel test.
Analytical Methods
The first part of the presentation will concentrate upon analysing the mechanics of the peel test
by applying a continuum fracture-mechanics method using an energy-balance approach. The
theoretical analytical methods from Kinloch et al. [2] and Georgiou et al. [3], which describe
models to ascertain values of the adhesive fracture energy, Gc, will be outlined and employd. As
will be seen, the current challenge is to model accurately any extensive plastic deformation which
may occur in the flexible peeling arm, since if this is not accurately modelled then the value of Gc
deduced may suffer a high degree of error.
Numerical Methods
Secondly, the paper will describe the use of a novel numerical method, based upon a finite-
element analysis (FEA) model, to deduce the value of Gc from the elastic-plastic peel test. A major
914 A. J. Kinloch et al.

feature of such a method would be its ready applicability to analysing more complex adhesively-
bonded components where gross plastic deformation of the substrates accompanies fracture. For
example, crush tests on adhesively-bonded aluminium-alloy tubes and box beams, as employed in
the front-end section of the new Aston Martin ‘Vanquish’ sports car.
The technique eventually chosen for numerically modelling the peel test was a FEA approach
employing the node-release method to simulate crack growth along a pre-defined crack path. The
main novel feature of this approach is that the critical equivalent plastic strain was chosen as the
criterion for releasing the node. However, this parameter is a strong function of the degree of
triaxiality at the crack tip. This was accounted for by using the FEA results to also obtain the stress
field at the crack tip, and then employing the Rice-Tracy [4] model to allow for the different
degrees of triaxiality found in the various types of peel test that were considered. The FEA model
is then used to compute the energy dissipated by the plastic deformation of the peel arm during
steady-state crack growth, and hence so deduce the value of Gc. The values of Gc so determined
are shown to represent a ‘material characteristic’ parameter, independent of such factors as the peel
angle, thickness of peel arm, etc.
Conclusions
The advantages and disadvantages of the different methods will be reviewed and a major
conclusion is that the values of Gc from the different modelling methods are found to be in very
good agreement, and are also in excellent agreement with values from standard LEFM test
methods.

References
1. Blackman, B.R.K. and Kinloch, A.J., in ‘Fracture Mechanics Testing Methods for Polymers,
Adhesives and Composites’ edited by D.R. Moore, A. Pavan and J.G. Williams, (Elsevier
Science, Amsterdam, 2001) 225-270.
2. Kinloch, A.J., Lau, C.C. and Williams, J.G., Int. J. Fracture, vol. 66, 45-70, 1994.
3. Georgiou, I., Ivankovic, I., Kinloch, A.J. and Tropsa, V., J. Adhesion, vol. 79, 239-265, 2003.
4. Rice, J.R. and Tracy, D.M., J. Mech. Phys. Solids, vol. 17, 201-217, 1969.
 12. Interface Fracture and Behavior of Joints 915

AN ALTERNATING CRACK GROWTH IN ADHESIVELY BONDED JOINTS

A. R. Akisanya
Engineering Department,
School of Engineering and Physical Sciences
King’s College, University of Aberdeen,
Aberdeen AB24 3UE, UK.
a.r.akisanya@eng.abdn.ac.uk

Adhesive joints are increasingly used in the assembly of components in electronic, automobile and
aerospace industries. A standard joint design is of sandwich type where two parts made from the
same or different material are bonded by a thin layer of adhesive. The reliability and fracture
characteristics of the joints under mechanical and thermal stresses during processing and service
constitute a major technical problem.
Cracks and flaws are inevitable in the manufacture of adhesive joints. The joint geometry,
material combination and service loads must be chosen to prevent crack growth; this is sometimes
difficult if not impossible. Hence, these parameters must be selected such that when crack growth
does occur the chosen crack path is that which maximises the macroscopic toughness (or strength)
of the joint. A sandwiched joint consisting of a brittle adhesive can fail in a variety of ways. The
crack may grow along one of the interfaces, within the adhesive, alternate between the adhesive
and the adjacent material or alternate between the two interfaces of the joint. The effect of remote
loading on interfacial fracture toughness has been extensively studied, see for example Chai [1],
Cao and Evans [2], Akisanya and Fleck [2], Banks-Sills and Schwartz [4]; the interfacial toughness
increases with increasing magnitude of the remotely applied shear. Further, the directional stability
of crack paths in adhesive joints has been related to the magnitude of the non-singular stress
(usually called the T-stress) near the crack tip, e.g. Fleck et al. [5], Chen et al. [6]. In this paper we
examine the effect of material combination and the remote mixed mode loading on the likelihood
of occurrence of the alternating crack in an adhesive joint.
Alternating crack trajectory has been observed by several authors during experimental testing
of sandwiched joints with a brittle adhesive [1, 3, 6]. A detailed understanding of this failure
mechanism is important since it significantly enhances the macroscopic toughness of the joint [3].
In a previous study (Akisanya and Fleck [7]) the effects of remote tension and thermal residual
stress in the adhesive layer on the wavelength of the alternating crack were examined for a
sandwiched joint subjected to a remote mode I loading. For a given material combination, the
wavelength of the alternating crack increases with increasing magnitude of the thermal residual
stress [7].
In the current paper we consider the effects of remote shear load on the alternating crack
trajectory. The near-tip region of the alternating crack is replaced by a circular domain loaded

remotely by the crack tip stress intensity factor K

f
 K If  iK IIf  for the homogeneous base
specimen neglecting the presence of the layer, as shown in Fig. 1. K f is related to the remotely
applied loads and the geometry, see Tada et al. [8]. Loading is achieved by imposing the
asymptotic crack tip displacement field for homogeneous body on the circular boundary of the
idealised geometry shown in Fig. 1b. The numerical solutions of the interfacial crack-tip stress
intensity factor are obtained. These solutions are combined with the well-established interfacial
fracture mechanics to explore the necessary and sufficient conditions for the interfacial crack to
kink into the adhesive layer in the presence of a pre-existing kink-type flaw. We find that for a
916 A. R. Akisanya

given size of flaw and material combination there is a range of K IIf / K If for which the alternating
crack is likely to occur. The lengths "1 and " 2 of the interfacial crack prior to kinking into the

adhesive (see Fig. 1a) are different for the two interfaces. An increasing magnitude of K IIf / K If
results in an increase in one of the lengths and a decrease of the other.

Figure 1. (a) An alternating crack in a sandwiched joint subjected to remote mixed mode
loading. (b) An idealised geometry of the alternating crack.

References
1. Chai, H., Int. J. Fract., vol. 32, 211-213, 1987.
2. Cao, H.C. and Evans, A.G., Mechanics of Materials, vol. 7, 295-304, 1989.
3. Akisanya, A.R. and Fleck, N.A., Int. J. Fract., vol. 58, 93-114, 1992.
4. Banks-Sills, L. and Schwartz, J., Int. J. Fract., vol. 118, 191-209, 2002.
5. Fleck, N.A., Hutchinson, J.W. and Suo, Z., Int. J. Solids Struct., vol. 27, 1683-1703, 1991.
6. Chen, B., Dillard, D.A., Dillard, J.G. and Clark Jr., R.L., Int. J. Fract., vol. 114, 167-190,
2002.
7. Akisanya, A.R. and Fleck, N.A., Int. J. Fract., vol. 55, 29 – 45, 1992.
8. Tada, H., Paris, P.C. and Irwin, G.R., Stress Analysis of Cracks Handbook, Del Research, St.
Lious, MO, 1985.
 12. Interface Fracture and Behavior of Joints 917

MEASUREMENTS OF INTERFACE FRACTURE AND MECHANICAL

PROPERTIES OF LOW-K DIELECTRIC THIN FILMS

F. Atrash and D. Sherman

Department of Materials Engineering
Technion-Israel Institute of Technology
Haifa, 32000 Israel
dsherman@tx.technion.ac.il

The residual stresses and the biaxial elastic modulus of low-k dielectric thin films, and in particular
the interfacial fracture energy of these films on a silicon substrate were all measured using a single
method. The method consists of depositing a thin metallic 'super layer' on top of cantilever beam
structures [1] of the low-k dielectric thin films, manufactured using the conventional micro
machining processes. The 'super layer' generates tensile stresses in the investigated thin layers.
When these beams are either released from the substrate or separated by spontaneous crack
propagation initiated from a natural precrack, a curvature of the bilayer beam structures is formed
and used to evaluate the above properties of the thin films and of the interface.
The low-k dielectric thin films in this investigation is 0.5, 1, and 1.5 Pm thick Novelus
CORAL© having a composition of 30at%C-20at%O-50at%Si. The CORAL© thin films were
deposited on top of nearly 1 Pm thick Cu and 50 nm thick Ta thin films (the latter serves as a
barrier layer), all on top of 725 Pm thick 6" (100) Silicon wafers. The metallic 'super layer' is
highly stresses e-beam evaporated Ti layers, having thicknesses of 150, 200, and 250 nm. This
constituted nine different bilayers, enabled to develop a comprehensive methodology to measure
the mechanical properties and to provide valuable information regarding the evaluation of these
properties.
We use Timoshenko Equation [2] for the curvature of bilayer thermostats:

1 6( h1  h2 )
3
' H res
R 2 E2 h 2E1 h13 2 2
h1    h2  3( h1  h2 )
E1 h1 E 2 h2 (1)

res
where the index 1 designates the investigated layer, 2 the metallic super layer, 'H H 2res  H1res .
The thicknesses of the constituents, Eq. (1), were measured using alfa-step profilometer. The
average radius of curvature, R, for each of the nine cases of the released or of the delaminated
bilayers was calculated based on about 60 test results for each case. The elastic modulus of the Ti
thin films was taken from the literature as 100±10 GPa. The residual stresses in the Ti thin films
res
(required to calculate the residual strains in that layer, H 2 ) on top of the above multilayered
structure was evaluated by means of Sotney Equation [3]; they are 677±66MPa, independent of the
layers' thickness (in the range of 150-250nm). Eq. (1) has, therefore, two unknowns: 'Hres (since
H1res is unknown) and [ E2 / E1 (since E1 is unknown). It can be rewritten as:

1 A1
R A2[  A3 / [  A4 (2)
918 F. Atrash and D. Sherman

where the Ai's are functions of the thicknesses. Selecting a range of [ yields a curve of 'H vs. [for
each set of specimens. Plotting 'H vs. [. for a group of specimens with same CORAL thickness
(h1) generates three intersections which yield the range of values for 'Hres and for [. Fig. 1 shows
the curves and the intersections obtained for the group with h1=1.5 Pm.

FIGURE 1. 'Hres vs. [for bilayers having 1.5 Pm thick CORAL film.

This group yields ECORAL 17.5 r 0.5 GPa and V res 45 r 5 MPa.
Knowing the material properties, it is now possible to calculate the interfacial fracture energy
of the (weak) interface between the CORAL and the copper. Using the decohesion-no-decohesion
criterion [1] of the nine bilayers it was evaluated to be 0.6-0.7 Jm-2, Fig. 2. More details of the
calculations, the effect of layer thickness, and details of the evaluation of the interfacial fracture
energy will be presented.

FIGURE 2. The decohesion-no-decohesion plot to evaluate the interfacial fracture energy.

References
1. Bagchi A., Lucas G., Suo Z. and Evans A., J. Mater. Res., 9, 1734, 1994.
2. Timoshenko S., Journal of the optical society of America, 11, 233-255, 1925.
3. Stoney G., Proceedings of the royal society, A82, 172-175, 1909.
 12. Interface Fracture and Behavior of Joints 919

INITIATION OF FRACTURE MECHANISMS AT THE FIBRE/MATRIX

INTERFACE

E. Martin1, B. Poitou1 and D. Leguillon 2

1LCTS, CNRS UMR 5801 Université Bordeaux 1, Pessac, France.
2LMM, CNRS UMR7607, Université P. et M. Curie, Paris, France.

LCTS, 3 Allée de la Boetie, 33600 Pessac, France,

Phone : (33)556 844 700, Fax : (33) 556 841 22,
martin@lcts.u-bordeaux1.fr

Interface is the key region which determines the fracture properties of a composite material.
Modelling of fracture mechanisms in the vicinity of the fibre/matrix interface must provide a better
comprehension of the role of the interface properties in order to design the adequate fibre coating
(Kerans et al.) [1]. For example, promoting the fracture of the interface under the influence of an
approaching matrix crack allows to protect the fibre from stress concentration. The control of this
fracture mechanism at the fibre/matrix interface leads to composites made with brittle matrices and
reinforced with brittle fibres which exhibit high toughness.
The aim of this paper is to provide conditions for the initiation of interfacial debonding in the
vicinity of a matrix crack in brittle composites. Previous authors focus on the competition between
crack penetration and crack deflection with the crack tip located at the bimaterial interface (He and
Hutchinson) [2]. However, experimental observations have shown that interfacial debonding may
occur ahead of the crack tip before the matrix crack reaches the interface (Xu et al.) [3].

FIGURE 1. a) Nucleation of a fibre crack in the vicinity of the matrix crack, b) Nucleation of an
interfacial crack in the vicinity of the matrix crack.

The fracture mechanism analysed is depicted in Fig. 1. An annular matrix crack is introduced
in a representative cell (composite cylinder with a fibre radius Rf and a volume fraction Vf ). The
distance between the crack tip and the fibre/matrix interface is denoted l. Upon axial loading and as
a result of the stress concentration induced by the matrix crack, an interfacial crack of length 2d or
a fibre crack of length p may nucleate. To evaluate the initiation stress of these nucleation
mechanisms, standard approaches of fracture mechanics are ineffective but use is made of an
initiation criterion (Leguillon) [4] which reveals efficient in such situations (Martin and Leguillon)
[5] : i) an energy balance provides an incremental condition in which the infinitesimal energy rates
of the classical Griffith's condition are replaced by finite energy increments, ii) an additional
strength condition (which states that the opening normal stress along the anticipated path of crack
nucleation is greater than the relevant strength) is imposed. Applying such approach provides the
*
initiation stress V i* (respectively V f ) of the deflection (respectively penetration) mechanism :
920 E. Martin et al.

V i* 
g l , R f , V f , E f , E m ,Q f ,Q m , V ic , G cf V ic  ,

V *f 
g l , R f , V f , E f , E m ,Q f ,Q m , V cf , G cf V cf  (1)

where ( E ,Q ) are the Young’s modulus and the Poisson ratio, ( V c , G c ) are the strength and the
toughness. A finite element procedure determines the value of (f,g) in order to analyse the
* *
competition between the two nucleation mechanisms which is given by the condition V i  V f .
Results show that the deflection criterion is expressed as :
2
G i
c
 C G c
f  D V  c
f w ith
C C l , R f ,V f , E f , E m ,Q f ,Q m , D D l , R f ,V f , E f , E m ,Q f ,Q m  (2)
As illustrated by Fig. 2, it is interesting to note that a high value of the interfacial toughness
G i can promote debonding provided the value of the fibre strength V cf is high enough.
c

FIGURE 2. Debonding criterion (the decohesion is predicted if the interfacial toughness Gic is
lower than the plotted lines)

REFERENCES
1. Kerans R.J., Hay R.S., Parthasarathy T.A., Cinilbulk M.K, J. Am. Ceram. Soc., 85, 2599-
2632, 2002.
2. He M.Y., Hutchinson J.W., Int. J. Solids Structures, 25, 1053-1067, 1989.
3. Xu L.R., Huang Y.Y., Rosakis A.J., J. Mech. Phys. Solids, 51, 461-486, 2003.
4. Leguillon D., Eur. J. of Mechanics – A/Solids, 21, 61-72, 2002.
5. Martin E., Leguillon D., Int. J. Solids Structures, 41, 6937-6948, 2004.
 12. Interface Fracture and Behavior of Joints 921

EFFECTS OF PLASTICITY AND RESIDUAL STRESS FOR CRACKS NEAR

INTERFACES

Ivar Reimanis, Keith Rozenburg, Matthew Tilbrook1 and Mark Hoffmann1

Metallurgical and Materials Enginering, Colorado School of Mines, Golden, CO 80401, USA
1School of Materials Science and Engineering, University of New South Wales, Sydney 2052,

New South Wales, Australia

reimanis@mines.edu

Cracks situated parallel to, and very near, the interface in layered, ductile-brittle composite
specimens were investigated with finite element analysis. Elastic, plastic and thermal properties of
Cu-W composites previously obtained from experiments were utilized in the model. A finite-
element model for simulating mixed-mode crack propagation in linear elastic materials was
modified to incorporate yielding. A routine for automatic crack extension and remeshing enabled
simulation of incremental crack propagation. Particular issues, including calculation of fracture
parameters, crack propagation direction under mixed-mode loading and retention of plastic strain
history, are addressed. The geometry and an example mesh close-up are shown in Fig. 1. Crack
propagation was simulated in homogeneous and layered Cu-W composites, employing thermal and
mechanical properties previously obtained from experiments. Two effects of plasticity on crack-tip
stresses are predicted: (i) compliance mismatch leads to stress intensity factor amplification or
‘anti-shielding’, and (ii) accumulation of plastic strains leads to increases in effective toughness.
Competition between these determines the structural reliability of the interface region.
Influences of thermal, elastic and plastic mismatch on critical load and crack-tip mode-mixity
were examined and reasonable predictions were obtained for crack-tip stresses and critical loads
for cracks near interfaces in the presence of plastic yielding. The following conclusions were
reached:
1 While the Irwin relation between crack-opening displacements and stress intensity factors
cannot be used when plastic deformation occurs at the crack tip, a non-linear relationship
may be obtained and used to calculate stress intensity factors.
2 Deflection of cracks near interfaces can occur due to mismatch in any one or more of
elastic, plastic and thermal properties. Plasticity tends to mitigate the effects of residual
stress on crack path.
3 The occurrence of plastic flow in the more ductile region increases stress intensity factors,
due to the enhanced compliance, and rotates crack-tip stresses toward the ductile region.
4 For cracks close to the interface, a decrease in crack-tip stress intensity factors may be
observed with crack extension, due to accumulation of plastic strain. This corresponds to
an increase in effective toughness due to the work absorbed during plastic deformation.
Plasticity has an important influence on crack-tip stresses and propagation paths for cracks
near interfaces. Accordingly, it is important to incorporate plastic deformation into models of
failure near interfaces and to consider the effects of prior strain history on crack-tip stresses during
simulations.
922 I. Reimanis et al.

FIGURE 1. (a) Layered copper-tungsten composite specimen configuration, showing dimensions

and positive deflection direction. Material 1 is the stiffer, more brittle composition while Material
2 is the more ductile. (b) Close-up of crack-tip, showing the path used for J-integral calculation in
simulations of homogeneous specimens.
 12. Interface Fracture and Behavior of Joints 923

TOUGHNESS OF A ±450 INTERFACE

L. Banks-Sills, Y. Freed, R. Eliasi and V. Fourman

Dreszer Fracture Mechanics Laboratory
Tel Aviv University
Ramat Aviv, 69978 Israel
banks@eng.tau.ac.il

Experiments are carried out to determine the delamination toughness for a crack along the interface
between two transversely isotropic materials. The material chosen for study consists of carbon
fibers embedded within an epoxy matrix. A crack is introduced between two layers of this
material, with fibers in the upper layer along the +450-direction and those in the lower layer along
the –450-direction both with respect to the crack plane. The Brazilian disk specimen is employed
in the testing.

FIGURE 1. Brazilian disk specimen.

This specimen shown in Fig. 1 allows for a wide range of mode mixities. A crack (or thin
notch) is placed along an interface between two layers with fibers on one side in the +450-direction
and on the other side in the –450-direction. The two layers are within a composite laminate strip
which was made from graphite/epoxy (AS4/3502) prepregs cured in an autoclave at a high
temperature and pressure. A plate approximately 12.4 mm thick was fabricated by Israel Aircraft
Industries with 15.4 mm wide and 25.4 µm thick Teflon (FEP fluorocarbon resin) strips introduced
periodically between two of the ±450 layers. The plate consists of an inner part of {0,-45,+45,0}s
each of nominal thickness 0.54 mm. Outer stiffening layers of ±450, 4.05 mm thick, were added to
prevent plate bending. Strips were cut from the plate and glued to aluminium partial disks to form
the specimen illustrated in Fig. 1. For details on specimen construction, see Banks-Sills et al. [1].
To calibrate the specimens, stress intensity factors are obtained which result from the applied
load, as well as residual curing stresses. It may be noted that all three modes are coupled, leading
to a three-dimensional problem. The finite element method and a mechanical M-integral are
employed to determine the stress intensity factors arising from the applied load (Freed and Banks-
Sills [2]). For the residual stresses, a three-dimensional thermal M-integral, together with the finite
element method, is used for stress intensity factor determination (Banks-Sills et al. [3]). The stress
intensity factors found for the applied load and residual stresses are superposed to obtain a local
interface energy release rate Gi , together with two phase angles \ and I .
924 L. Banks-Sills et al.

From the load at fracture, the critical interface energy release rate or interface toughness Gic
is determined as a function of the phase angles. There were twenty-six specimens. For each
specimen, the interface toughness is obtained at twenty-one stations along the crack front. These
results are plotted as the points in Fig. 2. A fracture criterion has also been developed for this
three-dimensional problem. It is the surface shown in Fig. 2. Points along the crack front for
nineteen of the specimens intersect the surface. It may be postulated for these specimens that some
of the points along the crack front have become critical and drag the rest of the crack front with
them as the delamination expands. For two of the specimens, all of the points are above the
surface; whereas for five of them, all of the points are below. This behavior is considered as
experimental scatter.

FIGURE 2. Delamination toughness Gic .

References
1. Banks-Sills, L., Boniface, V. and Eliasi, R., Int. J. Solids and Struct., vol. 42, 663-680, 2005.
2. Freed, Y. and Banks-Sills, L., Int. J. Fract. Mech., vol. 133, 1-41, 2005.
3. Banks-Sills, L., Freed, Y., Eliasi, R. and Fourman, V., Tel Aviv University Report 2005/1,
2005.
 12. Interface Fracture and Behavior of Joints 925

RESIDUAL STRESS INFLUENCE ON DISSIMILAR MATERIAL WELD

JUNCTION FRACTURE

P. Gilles and M.-F. Cipiere

Framatome-ANP
Tour AREVA Paris la Défense 92084, France
Josette DEVAUX
ESI-France
Le Discover, 84 Bd. Vivier Merle 69485 Lyon France
Philippe.gilles@framatome-anp.com

In most of nuclear reactors such as Pressurized Water Reactors (PWR) or Boiling Water Reactors
(BWR), heavy section components made in low alloy steel are connected with stainless steel
piping systems. The dissimilar material weld (DMW) junctions are performed between ferritic
nozzle ends and austenitic stainless steel piping, following a special manufacturing procedure to
ensure a good resistance of the joint. Post weld heat treatment (PWHT) is applied to reduce
residual stresses in the heat affected zone (HAZ), but whatever the process is, the difference in
thermal expansion coefficients induces residual stresses during the cooling stage. Furthermore,
differences in tensile properties may cause strain concentration at the weld to ferritic steel interface
which enhances the risk of crack initiation and extension. Several experiences from the field
confirm sensitivity to fatigue, corrosion or low toughness areas in this type of junction.
In the framework of the European Community Research and Development Programme a
project (ADIMEW, C. Faidy [1]) has been sponsored on the fracture behaviour of cracked stainless
steel/ferritic steel bimetallic welds. The objective was to contribute to the development and
verification of analysis methods which describe the behaviour of an external circumferential defect
in a DMW. The type of fracture is ductile, but the question of the influence of the residual stress
fields on tearing initiation has been raised since the material resistance is lowered when the crack is
close to the interface between the ferritic steel pipe and the weld. It has been shown (Devaux et al.
[2], Gilles et al. [3]) that this lowering of the “local” toughness is due the high degree of stress
triaxiality at the interface. In defect assessments tensile residual stress fields have to be taken into
account, except if their influence may be proven as negligible.
The present paper aims to estimate the influence of the residual stress field by numerical
simulation of the ADIMEW mock-up fracture behaviour and to recommend a simplified procedure
for including residual stress field in a J estimation scheme.
For the ADIMEW project, two DMW specimens have been manufactured of which one has
been tested in 4-point bending and one being utilised for the experimental determination of
welding residual stresses and for the generation of material property data for input to the analysis
procedures. The cracked section of the ADIMEW mock-up is shown in Fig. 1. To guarantee the
quality of the DMW's, a considerable experience and high degree of quality control has been
required within the weld manufacturing process, in accordance with FRAMATOME-ANP basic
nuclear specification.
926 P. Gilles and M.-F. Cipiere

FIGURE 1. Crack shape and location in the ADIMEW Dissimilar Weld Metal Junction

The analysis is conducted in three steps as explained below:

A refined numerical simulation of the welding process, which simulates each elementary step
of the mock-up manufacturing procedure by modelling each of the passes, has been conducted and
validated by comparison with the residual stress measurements using the neutron diffraction
technique (C. Ohms [4]). This simulation assumes axisymmetry.
The residual stress field is then transferred to a three dimensional mesh of the cracked DMW,
the crack being closed. Then crack is opened progressively simulating crack manufacturing.
Two finite element simulations of the fracture behaviour of the cracked mock-up under
bending are conducted with and without considering residual stress fields. Two methods are used:
one based on the crack driving force J and the other one using Rice and Tracey local approach
criterion for ductile tearing initiation.
The paper compares the predictions with the experiment and analyses the influence of the
residual stress field on the crack opening behaviour and the stress redistribution due to yielding.

References
1. Faidy C., “Structural integrity of dissimilar welds – ADIMEW project overview” In
Proceedings of PVP 2004, ASME Pressure Vessel and Piping Conference, 2004, USA.
2. Devaux, J., Mottet, G., Bergheau J.-M., Bhandari S. and Faidy C.: ‘Evaluation of the Integrity
of PWR Bimetallic Welds’, Jour. of Pressure Vessel Technology, 2000, ASME, 122, 386-
373.
3. Gilles Ph., Devaux J., Faidy C. – “ADIMEW project: Prediction of the ductile tearing of a
cracked 16” dissimilar welded junction” Proc. of PVP2004, ASME Pressure Vessel and
Piping Conference, 2004, USA.
4. Ohms C., Katsareas D. E., Wimpory R. C., Hornak P., Youtsos, A. G., “Residual stress
analysis in a thick dissimilar metal weld based on neutron diffraction”, PVP Vol. 479, ASME
Pressure Vessel and Piping Conference, July 2004, USA.
 12. Interface Fracture and Behavior of Joints 927

FRACTURE MECHANISMS OF A THIN ELASTIC PLASTIC LAMINATE

C. Bjerken, S. Kao-Walter and P. Stahle

Malmö University, Lund Institute of Technology
Dept. Solid Mechanics
pers@ts.mah.se

Packaging laminate has received much interest recently because of its use in liquid foor containers,
cf. [1-3]. This study considers a crack in a laminate consisting of a soft and a hard layer. The layers
are assumed to be thin so that the only significant fracture process is cross sectional necking in a
region ahead of the crack tip. The necking proceeds until the cross sectional area is very small
making any potential fracture of a remaining ligament insignificant to the macroscopic behaviour
of the structure. An experimental and numerical investigation of a laminate consisting of a 25 Pm
thick polymer sheet and a 9 Pm thick aluminium sheet demonstrates a remarkable increase of the
fracture toughness and the fracture strength as compared with that of the individual layers.

FIGURE 1. a) Geometry of the CCP test specimen. b) Results for aluminium and LDPE tested
separately and as a laminate.

Tensile tests were performed on the laminate, an aluminium sheet and a low-density
polyethylene (LDPE) sheet. The aluminium sheet and the LDPE sheet were identical to the two
layers of the laminate. All specimens had the same length h = 230mm and width b = 95mm. A
through thickness crack with the length a = 45mm was cut in the centre of the sheet and
perpendicular to the direction of the load. The tests were performed according to the ASTM
standard [4]
Figure 1 that the maximum load, 13N, carried by the aluminium sheet is at somewhat less than
1mm. At that extension the LDPE sheet carries around 1N. Thus one may expect that laminate
would carry a load of around 14N. However, the loading of the laminate is roughly 20N when it is
extended 1mm.
It is here assumed that, as the load is increased, crack growth is initiated in the aluminium
fracture first. It is further assumed that this compel the strains to localize in the LDPE layer at the
crack tip parallel with the necking region in the aluminium. Therefore the maximum load carrying
capacity of the LDPE is reached at continued crack growth in the laminate. The maximum load a
separate LDPE layer can carry is around 15N.
928 C. Bjerken et al.

To explore the features of this phenomenon elastic plastic FE analyses have been performed.
The necking region that develops ahead of the crack tip neck is assumed to be in a state of
approximately plane strain in a plane perpendicular to the direction of the crack. Therefore the
fracture process is viewed as a two dimensional problem of plane strain for a cross section
perpendicular to the crack ahead of the crack tip. Some de-cohesion between the two layers of the
laminate is assumed. The role of the interface toughness is studied. The plastic deformation is
simulated to the point when the cross section almost has vanished, i.e. almost to complete fracture.
The following observations were made:
- Onset of crack growth was observed to occur approximately at peak load. In the laminate
fracture of LDPE occurred after completed failure of the Al-foil.
-Peak load for the laminate is almost the same as the sum of the peak load for the Al-foil and
the peak load for the LDPE layer. This suggests that both materials reach peak stress in a small
region in the vicinity of the crack tip.
-The energy required before onset of fracture is unexpectedly large and around three times
larger than for the separate Al-foil layer. From the result the fracture toughness at onset of crack
growth

References
1. Kao-Walter, S. and Ståhle, P., In the Proceedings of SPIE, Third International Conference on
Experimental Mechanics, Xiaoping Wu, Yuwen Qin, Jing Fang, Jingtang Ke, Editors,
Volume 4537, Beijing, 2002, 253-256.
2. Lau, C. C., A Fracture Mechanics Approach to the Adhesion of Packaging Laminates,
Doctoral Thesis, Imperial College of Science, UK, 1993.
3. Tryding, J., In Plane Fracture of Paper, Doctoral Thesis, Division of Structural Mechanics,
Lund Institute of Technology , Sweden, 1996.
4. ASTM International, Standard Test Methods for Tensile Properties of Thin Plastic Sheeting,
D882-91, 1991.
 12. Interface Fracture and Behavior of Joints 929

CRACK-TIP PARAMETERS IN POLYCRYSTALLINE PLATES WITH

COMPLIANT GRAIN BOUNDARIES

R. Ballarini and Y. Wang

Department of Civil Engineering
Case Western Reserve University
Cleveland, Ohio 44106-7201
roberto.ballarini@case.edu

This paper presents two micromechanical models to study the statistics of the local stress intensity
factor of a cracked polycrystalline plate. The first is a finite element model based Monte Carlo
procedure where the plate's microstructure, which includes a finite number of crystals (or grains)
separated by a finite thickness interphase, is approximated as a Poisson-Voronoi tessellation (Fig.
1). The statistics of the effective elastic moduli of the uncracked plate are calculated as well as the
local stress intensity factors of the corresponding cracked plate (Fig. 2). This is done for selected
values of the parameters that quantify the ratio of elastic mismatch between the crystals and the
grain boundaries, and the expected number of grains in the plate. The results indicate that the
average values and standard deviations of the local stress intensity factors are independent of the
number of grains in the plate. The results of the Monte Carlo model suggest that the crack tip
parameters of cracked polycrystalline plates could be calculated using an efficient alternative
analytical model involving a long crack penetrating a circular inhomogeneity (Fig. 3). This
problem is solved using the method of continuously distributed dislocations and by expressing the
traction-free condition along the crack surfaces as a system of singular integral equations, which
are solved numerically. The results demonstrate that as long as the elastic mismatch between the
inhomogeneity and the surrounding material is interpreted correctly, then the approximate
analytical model is associated with averaged stress intensity factors that are in excellent agreement
with those of the polycrystalline microstructure. An attempt is made to apply the developed models
to interpret experimental data obtained from warm lake ice. It is concluded that proper
interpretation of data obtained from polycrystalline plates with compliant grain boundaries
necessitates stress analyses that incorporate explicitly the stochastic microstructure. However,
additional experimental data is required to resolve the issue that motivated the present study.
930 R. Ballarini and Y. Wang

References
1. Y. Wang and R. Ballarini, Meccanica, vol. 38, 579-593, 2003.
2. Y. Wang, “Crack-tip parameters in polycrystalline plates with compliant grain boundaries,”
Ph.D. thesis, Department of Civil Engineering, Case Western Reserve University, 2003.
 12. Interface Fracture and Behavior of Joints 931

EXTENDED FE SIMULATIONS OF CRACK GROWTH IN LAYERED AND

FUNCTIONALLY GRADED MATERIALS

C. Comi and S. Mariani

Dipartimento di Ingegneria Strutturale, Politecnico di Milano
Piazza Leonardo da Vinci 32, 20133 – Milano, Italy
claudia.comi@polimi.it , stefano.mariani@polimi.it

In this work we address some computational issues concerning failure analysis of layered and
functionally graded materials (FGMs). The gradual variation of the mechanical properties of
FGMs, unlike the abrupt change encountered in layered ones, is known to improve the failure
performance.
We focus on quasi-brittle materials subject to quasi-static loadings. As in Zhang and Paulino
[1], a cohesive model is used to simulate crack propagation in a linear elastic bulk. The relevant
traction-displacement discontinuity law is implemented within the context of the extended finite
element method (see, e.g., Sukumar et al. [2] and Mariani and Perego [3]). The approach allows to
follow the crack path independently of the background finite element mesh; this feature is
especially important for FGMs, since the gradation of the mechanical properties may lead to
complex crack paths also in a simple symmetric test set-up, Rousseau and Tippur [4].

FIGURE 1. 4PB test on a FG SEN specimen. (a): specimen geometry and sketch of the material
gradation; (b): adopted (coarse) spatial discretization.
932 C. Comi and S. Mariani

FIGURE 2. 4PB test on a FG SEN specimen: final crack path and relevant level sets of the in-plane
maximum principal stress.

Results concerning crack growth in functionally graded and layered glass-filled epoxy beams,
Rousseau and Tippur [5], are presented in order to assess the effects of mechanical gradation on
the failure mechanism. As for a single edge notched (SEN) specimen featuring a material gradation
parallel to the expected propagation direction and subject to four-point bending (4PB) (see Fig. 1),
first outcomes are shown in Fig. 2.

References
1. Z. Zhang and G.H. Paulino. International Journal of Plasticity, vol. 21, 1195 1254, 2005.
2. N. Sukumar, N. Mo?s, B. Moran and T. Belytschko. International Journal for Numerical
Methods in Engineering, vol. 48, 1549-1570, 2000.
3. S. Mariani and U. Perego. International Journal for Numerical Methods in Engineering, vol.
58, 103-126, 2003.
4. C.-E Rousseau and H.V. Tippur. Acta Materialia, vol. 48, 4021-4033, 2000.
5. C.-E Rousseau and H.V. Tippur. Engineering Fracture Mechanics, vol. 69, 1679-1693, 2002.
 13. Computational Fracture Mechanics 933

SIMULATION OF PLASTIC FATIGUE CRACK GROWTH BY A TWO SCALE

EXTENDED FINITE ELEMENT METHOD

A. Gravouil, T. Elguedj and A. Combescure

LaMCoS, Laboratoire de Mecanique des Contacts et des Solides
UMR 5514, INSA Lyon
Bat. Jean d’Alembert, 18,20 rue des Sciences 69621 Villeurbanne France
anthony.gravouil@insa-lyon.fr

A new technique for the finite element modelling of elastic-plastic fatigue crack growth with
frictional contact on the crack faces is presented. The eXtended Finite Element Method (X-FEM)
is used to discretize the equations, allowing for the modelling of arbitrary cracks whose geometry
are independent of the finite element mesh. This paper presents an augmented Lagrangian
formulation in the X-FEM framework in order to simulate elastic-plastic crack growth with
treatment of contact and friction. The original formulation, which takes advantages of a coarse and
a fine mesh close to the crack tip, is presented. In a second time the numerical issues such as
contact treatment and numerical integration are adressed, and finally numerical examples are
shown to validate the method.
The X-FEM uses the partition of unity in two ways: first to take into account the displacement
jump across the discontinuity far from the crack tip, and second to enrich the approximation close
to the front by considering the appropriate asymptotic fields [3] [4]. Several LEFM issues were
treated with the X-FEM such as elastic fatigue crack growth, crack growth with friction, arbitrary
3D fatigue crack growth with level set methods and dynamic crack growth. In recent works the
method has been applied to non-linear issues such as phase transformation, large strain analysis of
rubber like materials, and finally it has been modified by the authors to deal with plasticity [1] [2]
[7].
The main purpose of this contribution is, in the framework of the eXtended Finite Element
Method, to treat the case of multiple non-linearities with a two scale approach : a coarse
description of the structure is assumed (the X-FEM approach allows us to obtain an accurate
solution without meshing the crack) and a fine discretization close to the crack tip in order to
accurately describe the plastic zone. The presented method will focus on the case of plasticity
combined with frictional contact and is applied to fatigue crack growth analysis [9]. In the case of
material non-linearities, several issues have to be addressed. In a previous paper of the authors [8]
an appropriate elastic-plastic enrichment basis was developed to allow the X-FEM to deal with
plasticity [5].

1 ª T T T T T T º
ª
¬«
BD º
¼»
r n 1 « sin  cos  sin sin T  cos sin T  sin sin 3T  cos sin 3T »
¬ 2 2 2 2 2 2 ¼ (1)

In the present paper this approach is coupled with the treatment of frictional contact, and the
problems associated with plastic crack propagation are explored. The formulation is presented by
coupling an augmented Lagrangian method with the X-FEM, and the integration issues are studied
[6]. Several numerical examples are given to show the possibilities and the efficiency of the
method, and finally one concludes and presents possible future work.
934 A. Gravouil et al.

Figure 1.Crack propagation accross subdivided elements and evolution of the subelements
between the two configurations.

References
1. Hutchinson JW. Singular behaviour at the end of a tensile crack in a hardening material.
Journal of the Mechanics and Physics of Solids 1968; 16:13–31.
2. Rice JR, Rosengren GF. Plane strain deformation near a crack tip in a power-law hardening
material, Journal of the Mechanics and Physics of Solids, 1968; 16:1–12.
3. Moës N, Dolbow J, Belytschko T. A finite element method for crack growth without
remeshing, International Journal for Numerical Methods in Engineering 1999; 46(1):133–
150.
4. Babuska I, Melenk JM. The Partition of unity method, International Journal for Numerical
Methods in Engineering 1997; 40:727–758.
5. Fleming M, Chu YU, Moran B, Belytschko T. Enriched Element-free Galerkin methods for
crack tip fields. International Journal for Numerical Methods in Engineering 1997; 40:1483–
1504.
6. Simo JC, Laursen TA. An augmented lagrangian treatment of contact problems involving
friction. Computers and Structures 1992; 42(1):97–116.
7. Pan J, Shih CF. Elastic-plastic analysis of combined mode I, II and III crack-tip fields under
small-scale yielding conditions, International Journal of Solids and Structures
1992;29(22):2795–2814.
8. Elguedj T, Gravouil A, Combescure A. Appropriate extended functions for X-FEM
simulation of plastic fracture mechanics. Computer Methods in Applied Mechanics and
Engineering, accepted, 2005.
9. Combescure A, Gravouil A, Baietto-Dubourg MC, Elguedj T, Ribeaucourt R, Ferrié E.
Extended finite element method for numerical simulation of 3D fatigue crack growth.
Proceedings of the 31st Leeds-Lyon Symposium on Tribology 2004
 13. Computational Fracture Mechanics 935

ACCURATE DETERMINATION OF COHESIVE CRACK TIP FIELDS USING

XFEM AND ADMISSIBLE STRESS RECOVERY

B. L. Karihaloo1, Q. Z. Xiao1 and X. Y. Liu2

1Cardiff School of Engineering, Queen’s Buildings, The Parade, Newport Road,

Cardiff CF24 3AA, U.K.

2Division of Engineering Sciences, Institute of Mechanics, Chinese Academy of Sciences,

No.15 Beisihuanxi Road, Beijing 100080, China

karihaloob@cardiff.ac.uk (B L Karihaloo), xiaoq@cardiff.ac.uk (Q Z Xiao),
liuxy@lnm.imech.ac.cn (X Y Liu)

Cohesive zone (or crack) models have been extensively used in the study of localisation and failure
in engineering structures. Elices et al. [1] have discussed the advantages and limitations of these
models. De Borst et al. [2] have given a concise overview of the various ways in which the
cohesive zone methodology can be numerically implemented. The recently developed extended/
generalized finite element method (XFEM) (see, e.g., Daux et al. [3], Strouboulis et al. [4], and
Karihaloo and Xiao [5]) provides a proper representation of the discrete character of cohesive zone
formulations avoiding any mesh bias. Moes and Belytschko [6] and Wells and Sluys [7] analysed a
continuous cohesive crack that runs through an existing finite element mesh without mesh bias.
Remmers et al. [8] further studied the possibility of defining cohesive segments that can arise at
arbitrary locations and in arbitrary directions and thus allow for the resolution of complex crack
patterns including crack nucleation at multiple locations, followed by growth and coalescence.
Rubinstein [9] has shown that relatively small errors in the determination of the crack path
deflection angle can lead to a significant cumulative deviation of the crack path over a finite crack
length. Therefore a reliable analysis of crack propagation requires not only a suitable criterion of
crack growth but also an accurate evaluation of the crack tip field. The latter will be addressed in
this contribution.
We will first make a detailed analysis of the asymptotic field at the tip of a cohesive crack in
quasi-brittle materials for typical cohesive laws. This analysis will help us understand the structure
of the crack tip field, and at the same time, provide us with suitable crack tip enrichment functions
for the corresponding cohesive cracks. For traction free cracks, it has been shown that the
implementation of the accurate crack tip field as enrichment functions gives most the accurate
crack tip field using the XFEM (Liu et al. [10]).
We will then consider general cohesive cracks for which asymptotic fields are difficult to
obtain, and enrichment functions at the crack tip can only be chosen to meet the local displacement
conditions adjacent to the tip. In order to obtain accurate stresses, the statically admissible stress
recovery (SAR) scheme of Xiao and Karihaloo [11, 12] will be extended to cohesive cracks. SAR
uses basis functions, which meet the equilibrium equations within the domain and the local traction
conditions on the boundary, and moving least squares (MLS) to fit the stresses at sampling points
(e.g., quadrature points) obtained by the XFEM. It has been shown to be very powerful for
traditional FEM as well as XFEM for linear elastic problems with traction-free boundary
segments.
Typical cohesive crack problems with linear and nonlinear cohesive laws will be analysed and
compared with available results in the literature to illustrate the accuracy and applicability of the
methodology developed in this paper.
936 B. L. Karihaloo et al.

References
1. Elices, M., Guinea, G.V., Gómez, J. and Planas, J., The cohesive zone model: advantages,
limitations and challenges, Engng. Fract. Mech., 69, 137-163, 2002
2. De Borst, R., Gutierrez, M.A., Wells, G.N., Remmers, J.J.C. and Askes, H., Cohesive-zone
models, higher-order continuum theories and reliability methods for computational failure
analysis, Int. J. Numer. Meth. Eng., 60, 289-315, 2004
3. Daux, C., Moes, N., Dolbow, J., Sukumar, N. and Belytschko, T., Arbitrary branched and
intersecting cracks with the extended finite element method, Int. J. Numer. Meth. Eng., 48,
1741-1760, 2000
4. Strouboulis, T., Copps, K. and Babuska, I., The generalized finite element method, Comput.
Meth. Appl. Mech. Eng., 190, 4081-4193, 2001
5. Karihaloo, B.L. and Xiao, Q.Z., Modelling of stationary and growing cracks in FE framework
without remeshing: a state-of-the-art review, Comput. Struct., 81, 119-129, 2003
6. Moes, N. and Belytschko, T., Extended finite element method for cohesive crack growth,
Engng. Fract. Mech., 69, 813–833, 2002
7. Wells, G.N. and Sluys, L.J., A new method for modeling cohesive cracks using finite
elements, Int. J. Numer. Meth. Eng., 50, 2667-2682, 2001
8. Remmers, J.J.C., de Borst, R. and Needleman, A., A cohesive segments method for the
simulation of crack growth, Comput. Mech., 31, 69-77, 2003
9. Rubinstein, A.A., Computational aspects of crack path development simulation in materials
with nonlinear process zone, Int. J. Fract., 119, L15-L20, 2003
10. Liu, X.Y., Xiao, Q.Z. and Karihaloo, B.L., XFEM for direct evaluation of mixed mode SIFs
in homogeneous and bi-materials, Int. J. Numer. Meth. Engng., 59, 1103-1118, 2004
11. Xiao, Q.Z. and Karihaloo, B.L., Statically admissible stress recovery using the moving least
squares technique, In Progress in Computational Structures Technology, edited by B.H.V.
Topping and C.A. Mota Soares, Saxe-Coburg Publications, Stirling, Scotland, 2004, 111-138
12. Xiao, Q.Z. and Karihaloo, B.L., Statically admissible stress recovery for crack problems, In
Proc. ICF11, Turin, Italy, March 20 - 25, 2005.
 13. Computational Fracture Mechanics 937

A NEW GENERATION OF BOUNDARY ELEMENT METHOD FOR DAMAGE

TOLERANCE ASSESSMENT OF AEROSTRUCTURES

M. H. Aliabadi
Department of Aeronautics, Imperial College London, Prince Consort Road, London, UK.
m.h.aliabadi@imperial.ac.uk

This paper presents a brief review of recent progress in development of dual boundary element
formulations for damage tolerance assessment of aerostrucutres.
A new generation of boundary element formulations known as the Dual Boundary Element
Method was developed by Aliabadi and his students in 1992. for modeling crack growth in two-
and three-dimenional linear elastic problems. The formulation was subsequently extended to
nonlinear and transient problems. More recently, a new formulations have been developed which
allow the application of the BEM to thin walled strucutres widely used in engineering, for example
aircraft wings and fuselage panels, pressure vessels etc. The extension of the dual boundary
element method to thin-walled problems has provided for the first time a comprehensive modeling
tool suing BEM.
In this paper application of the BEM to linear and nonlinear stiffened panels are reviewed.
A flat square panel reinforced with three Z-stringers from the wing box of the B-52
Stratofortress is subjected to the transverse load. The panel and the stringers are modelled with 13
thin plates in total. Each plate is divided into 32 quadratic elements

Figure 1: BEM results a) stiffened flat panel, b) stiffnened curved panel, c) buckling of thin craked
panel.

Dirgantara and Aliabadi carried out a nonlinear fracture mechanics analysis of stiffened
curved panel. In the analysis, BEM shell mesh has 64 boundary elements and the stiffener is
modeled with 5 nodal points each.
Recently Purbolaksono and Aliabadi presented a dual boundary element formulation for
buckling analysis of plates with cracks. They presented the problem of a rectangular plate with
longitudinal central crack subjected to compression. Fig. 1 presents the changes in the buckling
mode of rectangular plate with aspect ratio 2.
Recently Di Pisa and Aliabadi presented the application of the DBEM to modeling a wing
section reinforced with stiffenerers which are riveted to the skin.
In this paper a brief review of the recent advances in the application of the boundary element
method to Damage Tolerance Assessment of Aerostructures structures was presented. The method
is shown to be capable of analyzing linear and nonlinear problems with boundary only
discretization.
938 M. H. Aliabadi

References
1. Portela,A., Aliabadi,M.H. and Rooke,D.P. The dual boundary element method: efficient
implementation for cracked problems, Int. J. Numer. Methods in Engng, 32, 1269-1287, 1992.
2. Mi,Y and Aliabadi,M.H. Dual boundary element method for three-dimensional fracture
mechanics analysis, Engineering Analysis, 10, 161-171, 1992.
3. M.H.Aliabadi A new generation of boundary elements in fracture mechanics, Int. J. Fracture,
86, 91-125, 1997.
4. Cissilino,A.P. and Aliabadi,M.H. Three-dimensional BEM analysis for fatigue crack growth
in welded components, Int. J.Pressure Vessel and Pipping, 70, 135-144, 1997.
5. Aliabadi,M.H. The Boundary Element Method, applications in solids and structures, Wiley,
Chichester, 2002
6. Dirgantara,T and Aliabadi,M.H. Numerical simulation of fatigue crack growth in pressurized
shells, International Journal of Fatigue, 24, 725-738, 2002.
7. DiPisa,C, Aliabadi,M.H. and Alaimo,A. Nonlinear Analysis of a reinforced panel undegoing
large deformation, Proceeding of the 5th International Conference on Boundary Element
Techniques, edited by Leitao,V and Alaiabdi,M.H., EC Publications, 2004.
8. Purbolaksono,J and Aliabadi,M.H., Dual boundary element analysis of cracked plates under
buckling loads, Int.J.Numer. Meth. Engng., 62, 537-563 2005
 13. Computational Fracture Mechanics 939

ROBUST STRESS INTENSITY FACTORS EVALUATION FOR 3D FRACTURE

MECHANICS WITH X-FEM

Hans Minnebo, Eric Bechet and Nicolas Moes

GeM - Institut de Recherche en Genie Civil et Mecanique, UMR CNRS 6183
Ecole Centrale de Nantes / Universite de Nantes
1, rue de la Noe, BP 92101 44321 Nantes CEDEX3
hans.minnebo@ec-nantes.fr

Fatigue life prediction for cracked parts with the classical finite element method is not an obvious
task for industrial structures. Indeed, it is necessary to take the crack geometry into account (its
surfaces and front) with enough accuracy. The stress intensity factors (SIFs) are then computed
and the crack can be propagated. Consequently, a remeshing step is necessary for the next time
step's mechanical problem. These remeshing procedures are difficult in 3D and are very time-
consuming.
The X-FEM (eXtended Finite Element Method) (Moes et al. [1]) allows to embed physical
surfaces of discontinuity into a mesh without modifying it, thanks to the use of two combined
level-sets (Stolarska et al [2]), the partition of unity method (Babuska and Melenk [3]) and an
enrichment with compact support. This method fits particularly well to crack analysis. The use of
specific enrichment functions for cracks, singular at the crack tip and discontinuous on the surfaces
enables one to carry a complete analysis of the crack without having to change the mesh.
Nevertheless, the enrichment functions are singular and thus difficult to integrate. So, it is
necessary to use a non-standard integration scheme in order to improve the accuracy of the weak
form's integration. We propose a modified integration scheme based upon a classical Gauss-
Legendre scheme to solve this problem (Bechet et al [4]). This scheme combines a coordinates
change with two variables changes adapted to the singular functions that are used for the crack tip.
For the 3D case, the cylindrical property of the singular problem is used to extend the 2D
integration method. We show increased convergence for both 2D and 3D cases.
The conditionning of the stiffness matrix is degraded, due to the use of these singular
enrichment functions. This brings into trouble conventional iterative solvers. In order to improve
the condition number, we use a preconditionning step based upon a Cholesky decomposition, that
makes a local orthogonalisation on the whole set of regular and enriched shape functions for each
node. Although local, this pre-conditionner improves greatly the conditionning.
The size of the enrichment zone is also discussed. On the one hand, the enrichment of too
many nodes will lead to a huge matrix with a high condition number. On the other hand, with the
enrichment of too few nodes, the representation of the problem will not be sufficient. So, an
intermediate size has to be chosen.
We propose a robust computation of the stress intensity factors, based on the X-FEM method,
which takes into account all the real boundary conditions : thermal, centrigugal... The robustness of
the SIFs evaluation is of primary importance for the estimation of the propagation path.
The improvements over standard FEM will affect the computation in that way :
an initial mesh is generated, not necessarily respecting the crack configuration, but with a
sufficient refinement around the crack tip ;
• several steps of propagation are made, with an update of the crack representation (using
level-sets) ;
940 H. Minnebo et al.

• when the crack tip is in a region where the refinement is non longer sufficient or if the
configuration is not suited for post-processing, a new mesh should be generated in the
same way as the initial one.
So, we hope to save CPU time for computational fatigue analysis, as we reduce the number of
remeshing steps. Those are also simpler to implement due to the lack of geometrical constraints on
the mesh.
Acknowledgements
The authors greatfully acknowledge the support of Snecma Moteurs.

References
1. Moes, N., Dolbow, J., Belytschko, T., A finite element method for crack growth without
remeshing., International Journal for Numerical Methods in Engineering, 46:131-150, 1999.
2. Stolarska, M., Chopp, D. L., Moës, N., and Belytschko, T., Modelling crack growth by level
sets and the extended finite element method, International Journal for Numerical Methods in
Engineering, 51(8):943-960, 2001.
3. Babuska, I., Melenk, J.M., The partition of unity method, International Journal for Numerical
Methods in Engineering, 40:727-758, 1997.
4. Bechet, E., Minnebo, H., Moes, N., Burgardt, B., Improved implementation and robustness
study of the X-FEM method for stress analysis around cracks, International Journal for
Numerical Methods in Engineering, accepted.
 13. Computational Fracture Mechanics 941

A MICRO-MACRO PARTITION OF UNITY METHOD FOR CRACK

PROPAGATION

P. A. Guidault1, O. Allix1, L. Champaney1 and C. Cornuault2

1LMT-Cachan (ENS Cachan/CNRS/University Paris)

61 avenue du President Wilson, F-94235 Cachan Cedex, France

2Dassault Aviation

78 quai Marcel Dassault, cedex 300, F-92552 Saint-Cloud Cedex, France

Christian.Cornuault@dassault-aviation.com, {guidault,allix,champaney}@lmt.ens-cachan.fr

In the last decade, two types of approaches have been designed to deal with effects which occur at
different scales. The first class of approaches is the class of multiscale strategies. The second one is
based on the Partition of Unity Method (PUM) first introduced by Melenk and Babuska [1] and all
its applications. By enabling one to enrich the kinematics of continuous media, the PUM makes
possible the introduction of discontinuities into the displacement field using a relatively small
number of additional degrees of freedom. One of the main advantages in this case is that the mesh
does not have to conform to the geometry of the crack. This technique greatly simplifies the
meshing and remeshing processes which, despite the improvement of meshing tools, remain
tedious tasks for engineers confronted with crack propagation situations. However, it does not
completely incorporate the multiscale aspect induced by the localization of strains in the cracked
zone. Generally, it requires further remeshing around the crack: thus, the remeshing problem is
only partially resolved. Furthermore, conditioning difficulties remain because of the treatment of
multiscale phenomena without separation. To overcome these two difficulties, the goal of our work
in cooperation with Dassault Aviation is to combine both types of approaches, a multiscale one and
a PUM-based one, in order to deal with crack propagation situations in an efficient way.

FIGURE 1. Three-point-bend specimen: microdisplacements (modeling of the crack by the X-

FEM) and macrodisplacements (thick lines).

The process involved is a combination of two techniques. The first technique consists in
applying the recently developed micro-macro approach [2] based on a homogenization technique.
The microscale is associated with local phenomena which occur around the crack. This scale is
much smaller than the macroscale which corresponds to the whole structure. This multiscale
approach ensures a correct global-local dialog between the macroscale and the microscale. The
second technique, based on the PUM, is used to define a proper representation of the local solution,
in terms of discontinuity and solution at the crack tip, on the microscale. The integration of
enrichment functions is obtained by the X-FEM, a PUM-based enrichment method introduced by
Belytschko and Black [3]. With this scale separation, the macroproblem keeps the same structure
throughout the calculation while the whole numerical effort is directed towards the microlevel [5].
942 P. A. Guidault et al.

In a technical point of view, the meshing and remeshing difficulties are thus resolved not only at
the global level but also at the local level.
In the micro-macro approach, the fact that a crack affects both the local level and the global
level raises the question of the kinematics and the description of the forces on the two scales.
Consequently, the choice of the macroscale and its associated discretization in order to include the
macroeffect of a crack is discussed. Moreover, in order to limit the use of the refined scale only to
where it is required, a decomposition of the domain into substructures and interfaces is proposed to
link two finite element descriptions. Finally, the integration of the X-FEM [4] as a local
enrichment method in such a multiscale frame is presented. Simulations of fatigue crack growth
will be shown in the field of linear elastic rupture. Our work in progress consists in applying the
strategy to ductile fracture mechanics and, more widely to damage.

References
1. Melenk, J. M. and Babuska, I., Computer Methods in Applied Mechanics and Engineering,
vol. 139, 289-314, 1996.
2. Ladeveze, P., Loiseau, O. and Dureisseix, D., International Journal for Numerical Methods in
Engineering, vol. 52(1-2), 121-138, 2001.
3. Belytschko, T. and Black, T., International Journal for Numerical Methods in Engineering,
vol. 45(5), 601-620, 1999.
4. Stolarska, M., Chopp, D. L., Moës, N. and Belytschko, T., International Journal for
Numerical Methods in Engineering, vol. 51(8), 943-960, 2001.
5. Guidault, P.-A., Allix, O., Champaney, L. and Navarro, J.-P., In Proceedings of the Seventh
International Conference on Computational Structures Technology, edited by B.H.V.
Topping and C.A. Mota Soares, Civil-Comp Press, Stirling, 2004, 443-444.
 13. Computational Fracture Mechanics 943

A DYNAMIC CRACK PROPAGATION CRITERIA FOR XFEM, BASED ON

PATH-INDEPENDENT INTEGRAL EVALUATION

Ionel Nistor, Serge Caperaa and Olivier Pantale

Ecole Nationale d’Ingenieurs de Tarbes
47, Avenue d’Azereix, TARBES, 65016, FRANCE
inistor@enit.fr, caperaa@enit.fr, pantale@enit.fr

The eXtended Finite Element Method (XFEM) has been successfully used for several years for the
numerical analysis of cracked structures under statically and later under dynamically loadings.
Based on Partition of Unity Method, the XFEM was developed at Northwestern University (Moes
et al. [1]) firstly as a method for analysing crack growth without remeshing, using special
enrichment functions to model discontinuous displacement fields. Since, the method was
continuously improved and applied to various domains of fracture mechanics. The dynamic crack
propagation is one of them and an important contribution for its modelling by XFEM was given by
Belytschko et al. [2].
In this paper we propose a crack evolution model for XFEM, based on dynamic energy release
rate computation using a path-independent integral. The numerical implementation of this model
was achieved in the home-made code DynaCrack (Nistor et al. [3]). Developped using Oriented-
Object programming framework, this code allows an explicit analysis of bi-dimensional structures
under dynamic loads, using XFEM. A particular approach concerning the enrichment functions
was adopted in order to obtain easily time-dependant solution. Only the Heaviside step function
was used and the crack-tip was restricted to cut one element at a time, passing from edge to edge.
So, the discontinuous displacement field for a N-nodes mesh, including NÃ enriched nodes, is

approximated, using the classical shape functions ) I and the enrichment one H , by:

u h
X  ¦ I I  X u I  ¦ I I  X H  X a I
I N I N *
(1)

where by uI are denoted the classical degrees of freedom and by a I the enriched ones.
A cohesive model is also incorporated in DynaCrack in order to model the material behaviour
in the cohesive zone between the mathematical crack-tip (i.e. the point where the crack opening
displacement is zero) and the physical crack-tip (i.e. the point of the complete separation of crack
faces). The introduction of the numerical crack-tip by this enrichment approach, gives us the
possibility to handle with cohesive model parameters in function of the considered material.
Adapted to our implementation of XFEM, the dynamic crack evolution model must provide
the crack advancing criteria, the propation direction and the crack speed. To answer these
questions, the crack evolution model implemented in DynaCrack is based on physical quantities,
analytically developed by Freund [4] and used by many authors for various finite-elements based
methods for computational fracture. The crack will propagate if at the crack-tip the current energy
release rate exceeds a critical limit, given as a material property.

For computing the current dynamic energy release rate, the path-independent integral J' ,
derived by Nishioka and Atluri [5], will be numerically evaluated:
944 I. Nistor et al.

J k' ³ >W
*  *c
@
 U n k  V ij u i , k n j d *  ³ U u u
S
i i ,k  U u i u i , k dS
(3)

where W and U are the strain and kinetic energy densities, n is the unit normal vector to

surrounding contour * directed away from the crack-tip, *c *c  *c represents the crack
edges inside of considered contour and S is the area inside of * .
The most important feature of the path-independent integral is the invariance of its value with
the chosen contour and this allows us to avoid the effects of the non-accurate solution for the
closely crack-tip field in XFEM using the far fields for integrating quantities from (3). Another
important feature of J ' -integral is given by its property to be related to the dynamic stress

intensity factors. The dynamic stress intensity factors are directely extracted from J'
components, using the equations system:

A I a K I2  A II a K II2 A IV a 
J 1' ; J 2'  K I K II
2P P (4)

where A I , II , IV a  are coefficients depending on the propagation speed crack a and P is the

shear modulus. The ratio between K I and K II values, computed at each time step, used in the
critical hoop stress criterion relation, gives the crack propagation direction.
The crack speed is provided by the numerical propagation algorithm, since the crack-tip
advances one edge at a time.
Several numerical examples demonstrating the main features of DynaCrack and the
computational efficiency of the proposed crack evolution model are presented in the last section of
the paper.

References
1. Moes N., Dolbow J., Belytschko T., A finite element method for crack growth without
remeshing, Int. J. Numer. Meth. Engng, vol. 46 (1), 133-150, 1999
2. Belytschko T., Chen H., Xu J., Zi G., Dynamic crack propagation based on loss of
hyperbolicity and a new discontinuous enrichment, Int. J. Numer. Meth. Engng, vol. 58,
1873-1905, 2003
3. Nistor I., Pantale O., Caperaa S. On the modelling of the dynamic crack propagation by
Extended Finite Element Method: Numerical implementation in DynELA code, COMPLAS
VIII, Barcelona, 2005 (to appear)
4. Freund L.B., Dynamic Fracture Mechanics, Cambridge University Press, 1998
5. Nishioka T. and Atluri S.N., On the computation of mixed-mode K-factors for a dynamically
propagating crack, using path-independent integrals J’, Engineering Fracture Mechanics,
Vol. 20, 193-208, 1984
 13. Computational Fracture Mechanics 945

TRUSS MODEL AS SIMPLE COMPUTATIONAL TOOL IN FRACTURE

MECHANICS

P. G. Papadopoulos, D. Plasatis and P. Lambrou

Department of Civil Engineering
Aristotle University of Thessaloniki, Greece
panaggpapad@yahoo.gr

The usual finite elements have complicated stiffness matrices and present particular difficulties in
handling nonlinear problems (Argyris [1]). The bar of a truss is the finite element with the simplest
possible local stiffness matrix (Absi [2], Fraternali et al.[3], Papadopoulos and Xenidis [4], Slaich
and Schäfer [5]), which can be written, in 2D, as:

(1)

where elastic stiffness, geometric stiffness, E elasticity modulus, A cross-section area and
undeformed length of the bar. The c = {cx cy} are direction cosines of bar axis, N axial force,
present length of the bar and I2 is the unit matrix in 2D.
Whereas, the simple global stiffness matrix of a truss can be written:

(2)
where B = (Bik) i : 1 … n, k : 1 … b is the Boolean linkage matrix of a truss and n, b are the
numbers of nodes and bars of the truss, respectively. Bik = -1 if node i is left end of bar k, Bik = +1
if node i is right end of bar k and Bik = 0 if there is no connection between node i and bar k.
The cross-section areas A of the bars are determined by considering the correspondence
between the stress-strain relations of Elasticity theory and the force-deformation relations of a truss
element [2-5].
The bars of the truss model obey nonlinear uniaxial stress-strain – laws [4]. Thus, the whole
truss can, in a simply way, describe material nonlinearities. On the other hand, by writing
equilibrium equations with respect to the deformed truss and updating the simple stiffness matrix
of the truss, within each step of an incremental loading procedure [4], we can, in a simple way,
take into account geometric nonlinearities.
A short, thus transparent, computer program with only about 250 Fortran instructions has been
developed for the nonlinear static analysis of a truss model in 2D, as well as in 3D.
The truss model can be proved a simple and useful computational tool in Fracture Mechanics.
Applications, by a truss model in 2D, are presented (Fig. 1). A plate with a crack is subjected to
uniaxial tension. And the following are clearly observed, in accordance with theory: 1) Stress
concentration at the crack, 2) Stress relief above the crack tip, and 3) Far from the crack, the stress
field approximates the uniform field of a plate without crack.
The formation of plastic zone at the crack tip and the crack propagation are also studied by the
truss model, and comparisons are made with other published experimental and computational
results.
946 P. G. Papadopoulos et al.

Figure 1. a. A plate, with a crack, subject to uniaxial tension. b. A truss model of one quarter of the
plate. c. One quarter of a plate without crack. Deformations, reactions and stresses of the bars in
kN/cm2. Uniform stress field. d. One quarter of a plate with crack. Deformations, reactions and
stresses of the bars in kN/cm2 near the crack (+tension, -compression).

References
1. Argyris, J.H., Editor. Fe.No.Mech. (Finite Elements in Nonlinear Mechanics). International
Conferences. Institute for Statics and Dynamics. University of Stuttgart, I. 1978, II. 1981, III.
1984.
2. Absi, E., Calcul Numerique en Elasticite. Eyrolles, Paris, 1978.
3. Fraternali, F., Angelilo, M., Fortunato, A., A lumped stress method for plane elastic problems
and the discrete-continuum approximation. International Journal of Solids and Structures,
vol. 39, 6211-6240, 2002.
4. Papadopoulos, P.G., Xenidis, H.C., A truss model with structural instability for the
confinement of concrete columns. Journal of EEE (European Earthquake Engineering), No 2,
57-80, 1999.
5. Schlaich, J., Schäfer, K., Design and detailing of structural concrete using strut-and-tie
models. Structural Engineering, 113-125, 1991.
 13. Computational Fracture Mechanics 947

FINITE ELEMENT MODELING OF COHESIVE CRACKS BY NITSCHE’S

METHOD

P. Hansbo and P. Heintz

Department of Applied Mechanics
Chalmers University of Technology, SE-412 96 Göteborg, Sweden
hansbo@am.chalmers.se, perhei@chalmers.se

In [1,2] a new finite element method, based on Nitsche’s method [5], was introduced, allowing for
discontinuities, internal to the elements, in the approximation across the interface; a method that
can handle both perfectly and imperfectly bonded interfaces without modifications of the code. For
the problem of linear elasticity, the case of an internal boundary * over which there was
continuity in traction but not necessarily in the displacement field, was studied in [2]. The problem
was formulated as seeking the displacement field u and the stress tensor V such that

in : 1 ‰ : 2 , (1)

in : 1 ‰ : 2 , (1)

u 0 on w : , (3)

on * , (4)

>u @  K V ˜ n on *. (5)
Here, [u] denotes the jump in the displacement field, n is a unit normal to the interface (its
direction being tied to the definition of the jump), and K is a compliancy tensor. The last condition,
which is a linear spring-type condition, clearly requires allowing for discontinuities in the
displacement field. For this simple model, it can be shown that our method enjoys the usual
stability and convergence properties of standard finite element methods. Numerical examples also
demonstrated the ability to simulate crack propagation using an oversimplified maximum tensile
stress model for deciding the crack propagation direction. In [3], Nitsche’s method was instead
combined with the use of material forces for performing more realistic brittle crack propagation
simulations. A typical example of brittle crack propagation using the Nitsche approach is given in
Figure 1. Notice the refinement at the crack tip, which is necessary for resolving the stresses close
to the tip.
In this contribution, we extend the method of [2] to the case of cohesive cracks with a softening
behavior under loading. The discontinuities are introduced when a failure criterion is met, so that
the softening behavior is displayed only by the interface while the continuum is assumed to remain
elastic. This is achieved by a cohesive law letting the compliances be functions of the jumps,
K=K([u]).
948 P. Hansbo and P. Heintz

FIGURE 1. Brittle crack passing a stiff inclusion.

We remark that a similar approach was taken by Mergheim and Steinmann [4], i.e., using
elements with internal discontinuities in the spirit of [1,2]. However, in [4] the standard finite
element formulation for handling interface springs were used, and the Nitsche framework was
avoided. This standard formulation will be badly conditioned in case the springs are extremely stiff
(which typically can lead to oscillations in the tractions [5]), and it also enforces a type of on/off
behaviour: either an element has cracked or it has not. In contrast, we will use the combined
Nitsche/spring formulation of [2] which in one single formulation encompasses the complete scale
from full continuity to full discontinuity. This allows us, in a sense, to use elements which are only
partially cracked since we can apply the full Nitsche method for the continuous part and the
cohesive law for the cracked part of a given element.
We will give a background to the method, the formulation for small elastic deformations, and
some numerical examples showing the performance of the method.

References
1. Hansbo A. and Hansbo P., Comput. Methods Appl. Mech. Eng., vol. 191, 5537-5552, 2002.
2. Hansbo A. and Hansbo P., Comput. Methods Appl. Mech. Eng., vol. 193, 3523-3540, 2004.
3. Heintz P., Chalmers Finite Element Center Preprint 2005-2.
4. Mergheim J. and Steinmann P., Int. J. Numer. Methods Eng., vol. 63, 276-289, 2005.
5. Nitsche J., Abh. Math. Sem. Univ. Hamburg, vol. 36, 9-15, 1971.
6. Simone A., Comm. Numer. Methods Eng., vol. 20, 465-478, 2004.
 13. Computational Fracture Mechanics 949

COMPUTING CRACK GROWTH IN QUASIPERIODIC ALLOYS

P. M. Mariano and F. L. Stazi

Dipartimento di Ingegneria Strutturale e Geotecnica,
Universita di Roma "La Sapienza",
paolo.mariano@uniroma1.it, furio.stazi@uniroma1.it

A standard crystalline lattice is characterized by periodicity and symmetries described by

crystallographic groups. However, there exist strange alloys for which periodicity is violated by
their own structure. They are characterized by a prominent influence at a gross level of atomic
effects. We may classify three classes of aperiodic - or quasiperiodic - crystalline alloys: (i) The
first class is the one of incommensurate modulated crystals (IMC) that can be obtained by primary
periodic structures by ‘displacing’ ideal locations of atoms in a way in which the original period of
the translational symmetry is incommensurate with respect to the ‘period’ of the modulation. (ii)
Incommensurate intergrowth compounds (IIC called also composite crystals) are characterized by
the presence of two or more (sub)lattices with periods mutually incommensurate. These sublattices
shift reciprocally to try to match one another. (iii) Finally, quasicrystals are intrinsically
quasiperiodic and satisfy symmetries forbidden by the classification of crystallographic groups:
they may display five-fold, eight-fold, ten-fold symmetries. Let us think of a planar quasicrystal
with pentagonal symmetry. From elementary geometry we know that we cannot fill the plane by
using only pentagons. So that we are forced to insert here and there topological alterations - i.e.
structures different from pentagons - called ‘worms’. Of course, the presence of worms alters
periodicity. For all these classes, local rearrangements of atomic clusters occur within each
crystalline cell. Such substructural rearrangements have influence on the gross mechanical
behavior and the interactions power conjugated with them cannot be neglected. To describe
substructural interactions, we follow the general format of multifield theories for complex bodies,
so that we start by assigning to each material element a morphological descriptor of the
substructural changes within the element itself and call such changes phason activity. For the
substructural events occurring in quasiperiodic alloys, a natural descriptor is a vector w attached at
each point. Substructural interactions are associated with w and its gradient and satisfy appropriate
balance equations. The energetic landscape is different for the three classes of quasiperiodic alloys
listed above. IIC and IMC may admit phason kinetics unless structural defects generate a ‘pinning’
effect that bars phason inertia and allows just phason modes of diffusive nature. On the contrary,
for quasicrystals, only diffusion phason modes are allowed. The elastic energy depends for IIC and
IMC not only on the macroscopic measures of deformation but also on w and its spatial gradient ’
w. The dependence on w is justified because it represents a relative displacement in IIC and IMC.
On the contrary, in the case of quasicrystals, the elastic energy does not depend on w but only on ’
w.
A unified description of the mechanics of quasiperiodic alloys is available [1]. It is based on a
parametrized variational principle of Lagrange-d'Alambert type including dissipative effects due to
phason friction always present in quasicrystals and in IMC and IIC when pinning effects occur.
When we analyze the behavior of cracks in quasiperiodic alloys, we get modiefied expressions
of the J-integral that account for the contribution of substructural interactions due to phason
activity and altering the energetic landscape around the tip of the crack as in general occur for
microstructures in all cases of complex bodies [2]. Preliminary numerical experiments on
quasicrystals, developed in the theoretical limit of vanishing phason friction display unusual
phenomena of clustering and self-organization of phonon (the standard deformative ones) and
phason modes around the tip of a crack. They influence the crack propagation.
950 P. M. Mariano and F. L. Stazi

Here we develop the issue and analyze the influence of phason modes on the growth of a
macroscopic crack.
To investigate numerically the topic we cannot make use of standard numerical procedures for
crack growth because they are based in part on the knowledge of exact solution for special cases.
Here we develop a general procedure that allow us to evaluate directly the vector driving force at
the tip of the crack without making use of stress intensity factors. We also find the direction of
propagation by means of a maximum dissipation principle. The procedure has general interest for
the mechanics of complex bodies but it is also non-standard for simple bodies.

References
1. Mariano, P. M., Mechanics of quasiperiodic alloys, submitted, 2005.
2. Mariano, P. M., Proc. Royal Soc. London A, vol. 461, 371-395, 2005.
3. Mariano, P. M., Stazi, F. L., Augusti, G., Comp. Str., vol. 82, 971-983, 2004.
4. Mariano, P. M. and Stazi, F. L., Crack growth in periodic and quasiperiodic alloys: computing
directly the driving force, in preparation, 2005.
 13. Computational Fracture Mechanics 951

X-FEM FOR 3D CRACKS IN SHAFT WITH CONTACT

S. Geniaut1,2, P. Massin1 and N. Moes2

1LaMSID, EDF R&D, UMR CNRS-EDF 2832

1, avenue du General de Gaulle, 92141 Clamart, France.

2GeM, ECN/universite de Nantes, UMR CNRS 6183

1, rue de la Noe, 44321 Nantes Cedex 3, France.

nicolas.moes@ec-nantes.fr, {samuel.geniaut, patrick.massin}@edf.fr

Meshing issues in the context of the finite element method (FEM) are often difficult ones to deal
with, as in the case of modelling helix-shaped cracks in rotor shafts (Andrier et al. [1]). As a matter
of fact, 3D automatic meshing programs often generate a large number of badly-shaped elements
which are not reliable and imply ill-conditioned stiffness matrix. Creating an adequate mesh
requires a considerable amount of user-time compared to the total computing time and these
meshing procedures are relatively costly. Besides, remeshing is necessary at each propagation step.
To avoid these difficulties, a recent method named eXtended Finite Element Method (X-FEM)
allows one to consider a crack in a unique and simple mesh within the classical framework of the
FEM (Moës et al. [2]). The crack, represented explicitly, is independent of the mesh, so the mesh
does not need to follow the geometry of the crack faces and remeshing is therefore avoided. Based
on the Partition of Unity Method (Melenk and Babuška [3]), X-FEM uses an enrichment of the
classical shape functions. To represent displacement discontinuities through the interface, a
generalized Heaviside function is introduced. Moreover, adding singular asymptotic fields at the
crack tip gives accurate results in linear elastic fracture mechanics. In addition, the level set
method (Osher and Sethian [4]) is a convenient way to describe a crack in 3D and propagating a
level set is well documented in the literature (Sukumar et al. [5]). As the general framework of
finite elements is kept, X-FEM can be easily extended to treat many problems such as cracks with
plasticity or hyper-elasticity, which explains the recent wide interest for this new approach.
However, to take into account the possible re-closing of the crack, interpenetrability must be
prevented. Indeed, surveys of real industrial cases have shown that contact between the crack faces
should not be neglected. Mixed method such as the Augmented Lagrangian Method (Alart and
Curnier [6]) using both Lagrange multipliers and regularization are well suited to solve that
problems. We propose in this paper an X-FEM approach with frictional contact. Few works deal
with contact and X-FEM. Among them, Belytschko et al. [7] use Lagrange multipliers to treat a
circular inclusion problem in 2D with X-FEM and frictionless contact. Dolbow et al. [8] propose a
formulation of the problem of a crack with frictional contact in 2D sharing similar features with the
Augmented Lagrangian Method [6] and satisfying alternatively the equilibrium and contact
conditions using an iterative scheme. In our extension to small displacements in 3D, we have
chosen a framework close to the Augmented Lagrangian Method using an hybrid (displacement-
pressure) and continuous formulation between two deformable solids for large displacements (Ben
Dhia and Zarroug [9]), adapted to X-FEM.
To adapt the mixed formulation of the contact problem, the following methodology has been
used. From the Lagrangian form of the virtual work principle for two solids in contact, introducing
the action-reaction principle, we obtain a simplified expression of the virtual work of contact
forces. The hybrid formulation for the contact problem based on pressure unknowns is directly
derived, incorporating weak forms of the contact laws. This formulation is well suited for a
discretization by the FEM leading to an assembly of elementary terms for contact and friction.
Appropriate choices of approximation spaces for the Lagrange multipliers are discussed in terms of
952 S. Geniaut et al.

convergence of the error. The Ladyshenskaja-Babushka-Brezzi (LBB) condition is verified by a

numerical test (Chapelle and Bathe [10]). The integration aspects are also studied. Indeed, with X-
FEM, the crack faces are viewed as a single surface which may cut a finite element. The
integration of contact terms on that non-meshed surface calls upon quantities related to the nodes
belonging to the elements that are cut by the crack. Besides, one of the main features of contact
with X-FEM is that under small displacements assumptions, no contact-nodes searching algorithm
is needed, because a geometrical point of the surface can be seen as two physical points, one on
each side of the surface. Therefore the displacement jump is expressed in terms of enriched
degrees of freedom introduced by X-FEM. An extension of the formulation when large sliding
occurs is under investigation.
As the FEM framework is preserved, the implementation within the industrial finite elements
software Code_Aster developed by EDF did not raise major difficulties linked to the architecture.
To date, this method has managed to deal with problem of cracks involving structures with
relatively simple geometries, when re-closing of the crack occurs. Once the validation process is
satisfied, real industrial cases such as rotors under rotative bending will be investigated. Taking
into account large displacements and plasticity at the crack tip is one of the future aims.

References
5. Andrier B., Garbay E., Hasnaoui F. and Massin P., In Proceedings of the Seventh
International Conference on Biaxial/Multiaxial Fatigue and Fracture, Berlin, Germany,
2004, 675-680.
6. Moes N., Dolbow J. and Belytschko T., Int. J. Numer. Meth. Engng., vol. 46, 131-150, 1999.
7. Melenk J.M. and Babuska I., Comp. Meth. Appl. Meth. Engng., vol. 139, 286-314, 1996.
8. Osher S. and Sethian J.A., J. Comput. Phys., vol. 79, 12-49, 1988.
9. Sukumar N., Chopp D.L., and Moran B., Engng. Frac. Mech., vol. 70, 29-48, 2003.
10. Alart P. and Curnier A., Comp. Meth. Appl. Meth. Engng., vol. 92, 353-375, 1991.
11. Belytschko T., Moës N., Usui S. and Parimi C., Int. J. Numer. Meth. Engng., vol. 50, 993-
1013, 2001.
12. Dolbow J., Moes N. and Belytschko T., Comp. Meth. Appl. Meth. Engng., vol. 190, 6825-
6846, 2001.
13. Ben Dhia H. and Zarroug M., Revue Européenne des Elements Finis, vol. 9, 417-430, 2002
14. Chapelle D. and Bathe K.J., Comput. & Struct., vol. 47, 537-545, 1993.
 13. Computational Fracture Mechanics 953

SOME IMPROVEMENTS FOR EXTENDED FINITE ELEMENT METHODS IN

FRACTURE MECHANICS

Patrick Laborde1, Julien Pommier2, Yves Renard2 and Michel Salaun3

1MIP, UPS Toulouse 3, 118 route de Narbonne, 31062 TOULOUSE cedex 4, France
2MIP, CNRS UMR 5640, INSA de Toulouse, 31077 Toulouse, France
3ENSICA, 1 pl. Emile Blouin, 31056 Toulouse cedex 5

laborde@mip.ups-tlse.fr, jpommier@insa-toulouse.fr, renard@insa-toulouse.fr,

msalaun@ensica.fr

Computer simulation of fracture processes remains a challenge for many industrial modelling
problems. In a classical finite element method, the non-smooth displacement near the crack tip is
captured by refining the mesh locally. The number of degrees of freedom may drastically increase,
especially in three dimensional applications. Moreover, the incremental computation of a crack
growth needs frequent remeshings. Reprojecting the solution on the updated mesh is not only a
costly operation but also it may have a troublesome impact on the quality of results.
In the last few years, Moës, Dolbow and Belytschko [1] introduced a numerical methodology
which has been developed by the name of XFEM - eXtended Finite Element Method. Not only
finite elements are enriched with the asymptotic crack tip displacement solutions, but also with a
step function which takes into account the jump of the displacement across the crack. Then, the
finite element mesh can be defined independently of the crack geometry. The partition of unity
chosen to localize the enrichment functions is linked to the mesh and is generally defined using
linear shape functions.
An advantage of the XFEM method is to obtain more accurate numerical results than classical
finite element one. However, the rate of convergence is not optimal with respect to the mesh
parameter h and is independent of the degree of the finite element method (see Fig. 1, and [2]).

In the classical XFEM method, only the nodes the nearest to the crack tip are enriched;
consequently the support of the additional basis functions vanishes when h goes to zero. So we
propose to enrich a whole fixed area around the crack tip independently of h. The expected optimal
rate of convergence is nearly reached. But the condition number of the linear system and the
number of unknowns increase.
A bonding condition is introduced on the enrichment area around the crack tip: for each
singular shape function, the equality of the corresponding degrees of freedom is prescribed. Doing
954 P. Laborde et al.

so, the numerical tests show an important lake of optimality. A simple analysis shows that this is
due to the transition layer between the enriched area and the rest of domain (see [3]).
To overcome it, we propose a nodal matching condition at the interface. Numerical tests show
that the optimal convergence rate is then obtained (see Fig. 2) even for high-order polynomial
functions. These tests concern bidimensional plate using degree one, two and three polynomial
finite element methods on triangular meshes with the Getfem++ [4] library.

References
1. Moes, N, Dolbow, J, and Belytschko, T., A finite element method for crack growth without
remeshing, Int. J. Numer. Meth. Engng., 46:131--150, 1999.
2. Stazi, F.L., Budyn, E, Chessa, J, Belytschko, T, An extended finite element method with
higher-order elements for curved cracks, Computational Mechanics, 31:38--48, 2003.
3. Laborde, P, Pommier, J, Renard, Y, Salaün, M, High order extended finite element method
for cracked domains, to appear in Int. J. Numer. Meth. Engng.
4. Pommier, J, Renard, Y, Getfem++, An open source generic C++ library for finite element
methods. http://www-gmm.insa-toulouse.fr/getfem
 14. Cohesive Models of Fracture 955

FAILURE PREDICTION OF ADHESIVELY BONDED T-PEEL JOINTS

A. Pirondi
Department of Industrial Engineering, University of Parma
Parco Area delle Scienze, 181/A - 43100 Parma, Italy
pirondia@me.unipr.it

Adhesive joining can offer significant advantages over traditional joining methods such as welding
or mechanical fastening in structural applications. Automotive and aerospace industries are
important examples. Joint fabrication procedures and component service loads may introduce or
initiate defects, whose evolution will control the performance and the reliability of the bonded
joint. In those cases, Fracture Mechanics (FM) can be used to assess the structural integrity of a
bonded joint [1]. The FM approach consists in the comparison of a parameter, function of load and
geometry of the cracked body (for example the strain energy release rate, G), with the fracture
resistance (Gc). The simulation of fracture therefore requires to implement a criterion that triggers
propagation when G=Gc. An attractive way to simulate the effect of a defect on joint strength is to
incorporate a model of the rupture process (i.e. the criterion to trigger propagation). In particular,
the fracture of bonded joints has been successfully simulated using the Cohesive Zone Model
(CZM) in [2-7].
In this work, CZM was used to simulate failure of T-peel bonded joints (Fig. 1) with 1.5 and
3mm thick adherends, respectively, bonded toghether with Loctite Multibond 330 adhesive [8].
The fracture toughness and load-opening behaviour recorded in previous experiments on bonded
DCB specimens [9] were taken as reference to calibrate CZM parameters.

FIGURE 1. Outline of half of the 1,5mm-thick T-peel joint tested in [8].

Two-dimensional and three-dimensional models were analysed using the FE code ABAQUS.
The failing interface was modeled with the cohesive elements available in this software. The
influence of: i) different cohesive law shapes, ii) elasto-plastic adherend behaviour, iii) modeling
the presence of the adhesive layer explicitly, was studied. The results obtained with cohesive
elements were initially compared with the corresponding ones obtained in a previous work [10]
where the adhesive layer was modeled with a series of non-linear springs attached to the interface
nodes of the cantilever (Fig. 2).
956 A. Pirondi

FIGURE 2. Comparison between load-displacement plots obtained modelling the adhesive layer
with nonlinear springs [10] and with cohesive elements.

References
1. Kinloch, A.J., Adhesion and Adhesives, Chapman and Hall, London, UK, 1986.
2. Hutchinson, J.W., Evans, A.G., Acta Mater., 48, 125-135, 2000.
3. Mohammed, I., Liechti, K.M., J. Mech. Phys. Solids, Vol. 48, 735-764, 2000.
4. Yang, Q.D., Thouless, M.D., Ward, S.M., J. Mech. Phys. Solids, Vol. 47, 1337-1353, 1999.
5. Knauss, W.G., Losi, G.U., J. Appl. Mech., Vol. 60, 793-801, 1993.
6. Hadavinia, H., Kinloch, A.J., Williams, J.G., In Adv. in Fract. and Damage Mech. II, M.
Guagliano and M.H. Aliabadi eds., Hoggar, Geneva, 2001, 445-450.
7. Sorensen, B.F., Acta Mater., 50, 1053-1061, 2002.
8. Rossetto, M., Private communication, Polytechnic of Turin, Turin, Italy, 2003.
9. Pirondi, A., Nicoletto, G., In Proc. IGF 2000, Bari, Italy, 2000.
10. Pirondi, A., In Proc. ECF 15, Stockolm, Sweden, 2004.
 14. Cohesive Models of Fracture 957

AN APPROACH FOR THE DETERMINATION OF MIXED MODE COHESIVE

LAWS

B. F. Sorensen and T. K. Jacobsen

Materials Research Department, Riso National Laboratory
4000 Roskilde, Denmark, bent.soerensen@risoe.dk
R & D Department, LM Glasfiber A/S
Rolles Møllevej 1 ,6640 Lunderskov, Denmark, tkj@lmglasfiber.com

Prediction of failure of structures and materials remains an important topic in engineering design.
In that respect, cohesive laws, describing the mechanical behaviour of a failure process zone, have
shown to be useful tools in numerical simulation of crack growth [1], in particular in structures in
which large-scale bridging develops [2]. The modelling has been rather advanced and is now
reasonably well established. However, the determination of cohesive laws, which in most case can
be regarded as material properties [3], still remains an unresolved issue. This is particular true for
mixed mode cracking. Mixed mode cohesive laws are sometimes interpolated from measurements
of pure mode I and pure mode II tests [4]. Furthermore, the normal stress is often assumed to
depend on the normal crack opening, but not on the tangential crack opening displacements. It is
difficult to evaluate the validity of such assumptions.
The present study aims at the development of an approach for the determination of mixed
mode cohesive laws. Special attention is devoted to large-scale bridging, since for such problems,
linear-elastic fracture mechanics solutions are invalid. The suggested approach is therefore based
on the path-independent J integral. A special test specimen, a double cantilever beam specimen
loaded with pure bending moments (DCB-UBM) is used, since, the J integral can be determined in
closed analytical form also under large-scale bridging for this test configuration:

1  Q  21M 
2
2 1  M 22  6 M 1 M 2
J
4B 2 H 3 E , (1)
here M1 and M2 denote the applied bending moments (positive signs are shown in Fig. 1), E and Q
denotes the Young's modulus and the Poisson's ratio, B is the specimen width and H is the beam
height. For plane stress, the terms 1-Q2 should be replaced by unity. The DCB-UBM specimen is
shown schematically in Fig. 1.

FIGURE 1. Double cantilever beam specimen loaded with uneven bending moments.
958 B. F. Sorensen and T. K. Jacobsen

Assuming that the cohesive stresses are derivable from a displacement potential [5], mixed
mode cohesive laws can be determined:

V n G n* , G t* 

wJ R G n* , G t*  V t G n* , G t* 

wJ R G n* , G t* 
wG n* wG t* (2)

* *
where Vn and Vt are the normal and shear stress, respectively, while JR, G n and G t are the
fracture resistance (the value of the J integral), the normal and tangential opening at the end of the
cohesive zone, see Fig. 1. It follows from (2) that mixed mode cohesive laws can be derived by
*
simultaneous measurements of JR, G n* and G t . That is the key idea.
Here we present theoretical basic for the approach. Furthermore, the practical implementation
of a DCB-UBM test fixture will be described. The approach will be illustrated for the problem of
delamination of unidirectional fibre composites. This is a mixed mode cracking problem involving
large-scale bridging. It is found, that the normal stress and shear stress both depend on both the
normal and the tangential crack opening displacements.

References
1. Tvergaard, V., and Hutchinson, J. W., J. Mech. Phys. Solids., vol. 41, 1119-35, 1993.
2. Bao, G. and Suo, Z., 1992, Applied Mech. Rev., vol. 45, 355-61, 1992.
3. Cox, B. N., and Marshall, D. B., Int. J. Fracture, Vol. 49, 59-76, 1991.
4. Yang, Q. D. and Thouless, M. D., Int. J. Fracture, vol. 110, 175-87, 2001.
5. Suo, Z., Bao, G., and Fan, B., J. Mech. Phys. Solids., Vol. 40, pp. 1-16, 1992.
 14. Cohesive Models of Fracture 959

THE USE OF CZM FOR COUPLED FATIGUE/PLASTICITY CRACK

PROPAGATION SIMULATION

Jl. Bouvard, F. Feyel and Jl. Chaboche

ONERA, DMSE-LCME
29 av de la Division Leclerc, BP72,92322 CHATILLON cedex France
Jean-Luc.Bouvard@onera.fr

In blading components, a large proportion of service life is taken up in initiating and growing
surface or near surface short cracks through the Stage I fatigue regime, that is when the crack
length is small compared to the scale of the crack tip plasticity.
Such fatigue cracks driven by mechanical and thermal stresses can lead to the catastrophic
failure of a component. This is why short cracks growth in Ni-base superalloys has received
considerable attention.
The main objective of this work is to develop refined mechanistic modelling techniques for the
coupling between oxidation, local inelastic behaviour and fatigue crack propagation, using
cohesive law formulations.
A cohesive zone method is a numerical tool developed for the mechanics of interfaces, that
was initially proposed to model crack initiation and growth. This method treats fracture as a
gradual process in which separation between incipient material surfaces is resisted by cohesive
traction. Its constitutive behaviour is specified through a relation between the relative opening
displacement U and a corresponding traction T at the same location, with being either the local
normal (n) or tangential (t) direction in the cohesive zone plane for 2D example.
We focused here to present a cohesive zone model that incorporates the energy released at the
crack tip, as well as closure effects induced by oxidation and cyclic plasticity/creep.
We developed there an hybrid formulation that incorporates the traditional fracture mechanics
approach in the framework of advanced damage mechanics concept.
Many cohesive models have been proposed under monotonic loading but these cohesive laws
are not able to model crack growth under cyclic loading. Consequently, for a structure, subjected to
constant amplitude loading, shake down and crack arrest would be automatically predicted due to
the stress fields redistribution and accommodation.
A specific damage law with time dependency will be then developed to simulate the crack
growth and to account for the different phenomena induced by the plasticity at crack tip[1-3].
The response of the analytical model under cyclic loading can be observed in Fig. 1. We can
notice that the damage increases during loading and remains constant during unloading.
We give in Fig. 2. the results concerning the use of this damage law in the case of a structure (a
smooth specimen with a precracking) considering a single crystal material.
960 Jl. Bouvard et al.

FIGURE 1. description of the damage law.

FIGURE 2. : plastic strain and Von Mises fields at crack tip.

References
1. De-Andres and al., “Elastoplastic finite element analysis of three-dimensional fatigue crack
growth in aluminium shaft subjected to axial loading”, Int. J. Of Solids And Struct., pp. 2231-
2258,1999.
2. Nguyen and al., “A cohesive model of fatigue crack growth”, Int. J. Of Fract., pp. 351-369,
2001.
3. Yang and al., “A cohesive zone model for fatigue crack growth in quasibrittle materials”, Int.
J. Of Solids And Struct., pp. 3927-3944, 2001.
 14. Cohesive Models of Fracture 961

DYNAMIC CRACK GROWTH : ANALYTICAL AND NUMERICAL CZM

APPROACHES

G. Debruyne, J. Laverne and P.E. Dumouchel1

EDF R&D LaMSID, 1 av du Général de Gaulle 92141 CLAMART, France
1LMM-UPMC, Paris VI University , 4 place Jussieu case 162, 75252 PARIS Cedex 5, France

gilles.debruyne@edf.fr, pierre-emmanuel.dumouchel@edf.fr, jerome.laverne@edf.fr

I. Industrial motivations.
Service life extension of PWR vessels is an important issue for a number of nuclear operators.
This aim is partially subjected to improvement of safety margins. These margins may be enhanced
by acceptance of a limited amount of crack growth, starting from an initial flaw. It is therefore
necessary to predict not only the crack initiation but also the crack arrest. It is also of great interest
to compare dynamic and static analysis and to know if quasi-static prediction is a conservative way
or not. Analytical investigations for actual structures are rather complex to carry on. An academic
problem theoretical analysis is therefore useful, in a first step, to understand the different features
of dynamic crack propagation and arrest. Moreover, numerical investigations may usefully
complete the analysis.
II. Problem statement.
The peeling of a thin film bonded on a rigid surface (cf fig 1.a) is a good example to check
such features. This problem has been analyzed in a dynamic way by Freund (1989) with a Griffith
criterion. It is revisited here with some slightly different conditions, where the film exhibits some
weak toughness zones and activates, during the peeling process, a surface energy described by a
Griffith or a Dugdale-Barenblatt model (cf fig 1.b). The analytical results are compared with a
finite element model using cohesive zones.

The inextensible film is initially completely bonded and tighted by a tension force. A small
deflection is then performed (with prescribed force or displacement) to one boundary, which leads
to the film debonding. The toughness all along the bonded surface is uniform, except on a weak
zone where a toughness jump arises.
III. Analytical investigation.
The 1D peeling problem is analytically performed with several configurations : displacement
or force control, Griffith or Dugdale debonding criterion, and several toughness zones
distributions. Dynamic and static investigations are both carried on (cf Fig 2.a and 2.b). Theoretical
962 G. Debruyne et al.

backgrounds to settle the main problem equations, starting from Hamiltonian principle, are briefly
outlined. Fields solutions are completely described and waves fronts (including debonding zone)
are exhibited.

IV. Numerical approach with CZM.

Following the ideas presented in Francfort and Marigo [1], we assume that the fracture of a
brittle body is governed by a principle of least energy. The total energy of a loaded structure is the
sum of its (elastic) strain energy and its surface energy defined on its discontinuity surface, minus
the potential of the dead loads (when forces are prescribed). The CZM relation between the density
of the surface energy and the jumps displacement is taken as a exponential (Barenblatt-type) or
a linear (Dugdale-type) function of the jump displacement. Two specifics finite element are used,
lying on a know crack path (cf Laverne [3]). The first one is an interface element where the jumps
are defined as linear functions of the nodal displacements. The second one is a finite element with
an embedded discontinuity, taken into account by enhanced terms for displacements and strains.
The previous academic 1D problem is now extended to a most realistic plane strain
configuration simulating the Double Cantilever Beam test. The initial flaw is blunted so that the
crack onset velocity is controlled by the bluntness parameter. Wave absorbing conditions are
prescribed on longitudinal boundaries to roughly fit in with the film problem. In order to analyze
the effects of wave reflections to crack kinematics, these conditions are released and a numerical
investigation is carried out. The infuence of the following parameters is considered : cohesive law
shape, including or not loading rate dependence, crack bluntness, interface or embedded
discontinuity element, weak toughness zones.

References
1. Francfort, G.A. & Marigo, J. -J., J. Mech. Phys. Solids, vol. 46(2), 1319-1342, 1998.
2. Freund L. B., Dynamic Fracture mechanics, Cambridge Monographs on Mechanics and
Applied Mathematics.
3. Laverne J. Thèse de l’université Paris 13, 2004.
 14. Cohesive Models of Fracture 963

SIMULATION OF PRE-CRITICAL CRACKING IN CONCRETE USING 3D

LATTICE MODEL

H.-K. Man and J. G. M. van Mier

Swiss Federal Institute of Technology (ETH) Zurich, Institute for Building Materials
ETH Hönggerberg, CH-8093 Zürich, Switzerland
man@ifb.baug.ethz.ch, vanmier@ifb.baug.ethz.ch

In this paper a discrete three-dimensional computational model of plain concrete structures at

meso-level is used to simulate fracture experiments and to analyse cracking in the pre-critical
stage.
In the past and the present many different computational approaches at meso-level have been
developed with the goal to study the mechanical behaviour of concrete, i.e. to simulate and to
understand fracture, either using the discrete or the continuum modelling approach, mostly done in
two-dimensions.
In the used lattice model, the continuum is replaced by a network of beam elements. Derived
from the statistical physics, it is a simple and efficient tool for understanding the fracture behaviour
of brittle heterogeneous solids such as concrete. Fracture is simulated by a sequential removal of
elements in each loading step, and gives insight and understanding into the fracture behaviour. In
the past these kinds of simulations have been conducted in two dimensions, but recently the model
has been successfully generalized to three dimensions which, in terms of fracture processes, is
more realistic, see Lilliu and van Mier [1], Lingen [2].
As case study prisms subjected to displacement-controlled three-point bending fracture
experiments will be simulated and analysed. For this purpose three-dimensional (triangular regular
and random) lattice models are developed. For constructing 3D lattices, an alternative method will
be introduced. This approach is a simple, but an efficient way to generate random 3D lattices.
For simulating fracture experiments at the meso-level the material properties of concrete as a
three-phase material (matrix, aggregate and interfacial transition zone) have to be considered and
they are included into the lattice.
In the present study various numerical experiments are concucted and the following parameter
variations are studied;
• scaling of specimen size by maintaining the same particle structure and
• the variation of thickness (full 3D-scaling).
A new method is introduced to find out where cracks preferably emerge. Analyses are
conducted mainly during microcracking stage (on a typical load-displacement diagram of concrete
it is the stage between the linear response and maximum load). Based on these investigations and
the analysis of several parameters and distributions, it is possible to make predictions and explains
where cracks nucleates, and how (why) fracture propagates.
964 H.-K. Man and J. G. M. van Mier

FIGURE 1. Boundary conditions and crack patterns of a three-point bend specimen, generated
with a 3D random lattice (the green color indicates cracking).

References
1. Lilliu, G., van Mier, J.G.M., Engineering Fracture Mechanics, vol. 70, 927-941, 2003.
2. Lingen, F.J., Design of an Object Oriented Finite Element Package for Parallel Computers,
PhD Thesis, Delft University of Technology, The Netherlands, 2000.
 14. Cohesive Models of Fracture 965

EFFECT OF COHESIVE LAW AND TRIAXIALITY DEPENDENCE OF

COHESIVE PARAMETERS IN DUCTILE TEARING

I. Scheider1, F. Hachez2 and W. Brocks1

1GKSS Research Centre Geesthacht, Max-Planck-St. 1, D-21502 Geesthacht
2PCIM, Université Catholique de Louvain, Place Saint Barbe 2, B-1348 Louvain-la-neuve

ingo.scheider@gkss.de

The cohesive model has been introduced for numerical crack propagation analyses of ductile
materials almost thirty years ago [1]. The two parameters of the cohesive model, the cohesive
stress, 0, and the cohesive energy, 0, can be explained on a physical basis and determined by
experiments [2]. However, up to now almost no literature exist about a metallurgical or
micromechanical basis for the shape of the constitutive law c(), called traction-separation law
(TSL) with being the material separation, and therefore many different shape functions are given
in literature instead [3]. On the other hand, it has been proven that the shape of the TSL may have a
significant effect on the result of the crack propagation analysis [4].
The present paper addresses the effect of the shape of the TSL without considering
micromechanical phenomena. For an illustration of the problem, two different fracture specimens,
namely a C(T) and an M(T) specimen, are simulated using two different traction-separation laws.
One is a widely used cubic polynomial shape (e.g. [1]), the other obeys a function with a high
initial stiffness and a wide region with constant stress [5]. Both functions are shown in Fig. 1.

FIGURE 1. Functions of two different traction-separation laws. The cohesive energy, *0, is equal
for both functions.

The material under consideration has the following elastic-plastic parameters: E = 210000
MPa, Q = 0.3, Y = 300 MPa, n = 10 (power hardening material law). The cohesive parameters are 0
= 1000 MPa and 0 = 100 kJ/m². The decreasing part of the TSL with constant stress starts at a
separation of 75% of the critical separation 0.
The results of the crack propagation analyses are displayed in Fig. 2, showing that the crack
resistance is highest with the constant stress function for both the C(T) and the M(T) specimen.
Obviously, the cohesive parameters cannot be chosen independently of the TSL. If the given
parameter set is assigned to the polynomial law, then the corresponding parameters for any other
function must be fitted in order to give the same global response for the C(T) specimen.
Consequently, simulations of the behaviour of an M(T) specimen using both TSLs with the
respective parameters are expected to result in an identical load-displacement curve. It turns out
that the parameters for the constant stress function have to be modified significantly to reproduce
the load-displacement curve, namely 0 = 824 MPa and *0 = 120 kJ/m². The results are also shown
966 I. Scheider et al.

in Fig. 2a. In a second step, the M(T) specimen behaviour has been simulated with this parameter
set to check whether the load-displacement curve obtained by applying the polynomial law is
reproduced by the constant stress function with the modified parameters. As can be seen in Fig. 2b,
this is not possible. The load-deformation curve calculated with the constant stress function is
significantly below the results obtained by the polynomial function.

FIGURE 2. Reproduction of the results achieved by the polynomial law with modified parameters
of the constant stress function. a) Parameter fitting using the C(T) specimen, b) Results of the
M(T) specimen with the fitted parameters.

As this result puts the transferability of the cohesive parameters into question, experimental
validations are necessary, since no decision can be made about the “correct” traction-separation
law. Additional improvements are expected, if the local triaxiality is taken into account and the
cohesive parameters made dependent on this value, as studied in [6].

References
1. Needleman, A., J. Appl. Mech., vol. 54, 525-531, 1987
2. Cornec, A., Scheider, I. and Schwalbe, K.-H., Eng. Fract. Mech., 70, 1963-1987, 2003
3. Brocks, W., Cornec, A. and Scheider, I., In: Comprehensive Structural Integrity. Edited by I.
Milne, R.O. Ritchie, B. Karihaloo, 2003, Elsevier, Amsterdam, 127-209.
4. Scheider, I. and Brocks, W., Key Eng. Mat., vols. 251-252, 313-318, 2003
5. Scheider, I., In First M.I.T. Conf. on Comp. Fluid and Solid Mech., edited by K.J. Bathe,
Elsevier, Amsterdam, 2001, 460-462
6. Brocks, W. and Siegmund, T., Int. J. Fract., vol. 99, 97-116, 1999
 14. Cohesive Models of Fracture 967

MODELING QUASIBRITTLE MATERIAL CRACKING WITH COHESIVE

CRACKS: EXPERIMENTAL AND COMPUTATIONAL ADVANCES

J. Planas, J. M. Sancho, A. M. Fathy, D. A. Cendon and J. C. Galvez

Universidad Politécnica de Madrid
ETS de Ingenieros de Caminos, Canales y Puertos,
Profesor Aranguren sn, 28040 Madrid, Spain.
jplanas@mater.upm.es, jose.sancho@upm.es, adel@mater.upm.es,
dcendon@mater.upm.es, jcgalvez@caminos.upm.es.

Fracture processes in concrete and other quasibrittle materials can be realistically described by
means of the cohesive crack model which, as introduced by Hillerborg in his celebrated fictitious
crack model, can be viewed as a constitutive assumption for the fracturing behavior of the material
[1].
The present paper discusses two important aspects concerning the application of the cohesive
crack to practical problems. The first aspect is the experimental determination of the cohesive law,
which is necessary to make accurate predictions; the second aspect is the computational procedure
to use to carry out the predictions.
Recent improvements of the test procedure developed in [2] for concrete are presented, which
are currently being considered for standardization by ACI Committee 446, and an improved
experimental setup is described together with various series of tests. The method relies on the
combination of results from diagonal compression splitting tests and three-point bending tests on
notched beams. Closed-form formulas are given to carry out the inverse analysis of the tests to
compute a bilinear approximation of the cohesive law. The influence of the shape of the stress vs.
crack opening curve on the test reliability is discussed in terms of the brittleness of the specimen,
and limits on the specimen size for other quasibrittle materials are deduced.
The paper next addresses the computational aspects of the practical application of the cohesive
crack model. Recent improvements of the method developed in [3] are described that show that
classical finite elements with embedded cohesive cracks can be formulated that provide a robust
method for cohesive crack growth analysis while keeping the implementation simple (i.e. any
finite element code accepting user-defined finite elements can be used).
Contrary to other strong discontinuity approaches (SDA) the proposed method does not require
enforcing crack path continuity (crack tracing or tracking) or defining exclusion zones. Such
numerical resorts are required, respectively, to avoid crack locking and to prevent cracking of
elements close to a main cohesive crack.
The proposed method combines simple features such as
1 constant strain elements which incorporate
a) strong discontinuity kinematics and
b)local crack equilibrium (strong form of equilibrium condition),
which implies that the stiffness matrix is, in general, non symmetric (SKON formulation
according to Jirásek’s classification [4]---static and kinematically optimal nonsymmetrical).
2 cohesive models with simple formulation based on
a)central-force law (crack traction vector parallel to the relative displacement of crack
faces)
968 J. Planas et al.

b)damage-like behavior (unloading to the origin) and

c)Rankine criterion for crack initiation and crack orientation.
The approach further incorporates two basic ingredients that proved essential to avoid crack
locking and spurious cracking:
3 the crack location in the finite element is selected to minimize the asymmetry of the
tangent stiffness matrix of the element (the crack does not go through the centre of gravity
of the element), and
4 the crack is allowed limited adaptability, which means that the crack is allowed to reorient
itself according to rotation of the stress axes, but only as long as the cohesive crack
opening is small: after a threshold crack opening is reached, the crack orientation is kept
constant.
Evidence is provided that shows that the method leads to mesh insensitive results, free from
mesh bias effects, and to robust behavior, both in 2D and in 3D cases. The effect of the various
ingredients of the model are evaluated and it is shown that the combination of central forces and
crack adaptation is essential for good performance.

References
1. Planas, J., Elices, M., Guinea, G. V., Gómez, F. J., Cendón, D. A. and Arbilla, I. Engineering
Fracture Mechanics, vol. 70(14), 1759-1776, 2003.
2. Planas, J., Guinea, G.V., and Elices, M., International Journal of Fracture, vol. 95, 367-378,
1999.
3. Sancho, J.M., Planas, J. and Cendón, D.A., In Fracture Mechanics of Concrete Structures,
edited by V.C. Li et al., Ia-FraMCoS, 2004, 107-114.
4. Jirásek, M. International Journal of Solids and Structures, 35, 4133-4145, 1998.
 14. Cohesive Models of Fracture 969

PINWHEEL MESHES AND BRANCHING OF COHESIVE CRACKS

P. Ganguly and K. D. Papoulia

Department of Theoretical and Applied Mechanics
and School of Civil and Environmental Engineering
Cornell University Ithaca NY 14853, USA
kp58@cornell.edu

We consider the use of cohesive interface models in a dynamic finite element setting to simulate
crack branching under mode-I loading conditions as performed in an experiment by Sharon et al.
[1] (Figure 1.)

Figure 1. Experimental setup

We follow the approach of implementing the cohesive interface law in a zero thickness
cohesive interface finite element positioned along bulk element edges. Since in this problem the
crack pattern is not known in advance, the possibility of prepositioning cohesive surfaces is
precluded and every edge of the bulk elements is considered as a potential fracture surface. The
crack propagation path can then be resolved as part of the solution of the governing equations.
We use an adaptive approach in which cohesive surfaces are only inserted when they are
needed, i.e., the effective initial stiffness of the cohesive model is infinite. We denote this type of
model as initially rigid. Special considerations in its finite element implementation are essential to
its numerical behavior. It was pointed out by Papoulia et al. [4] and Sam et al. [5] that in explicit
dynamics using a rigid model, the correct time convergence rate can be obtained only if the
condition that the nodal forces be continuous at the moment of activation is satisfied.
We use a scaled down version of the above experiment [2, 3] to reduce the number of degrees
of freedom. A pinwheel mesh [6] is used to triangulate the domain with 8 mesh edges resolving the
cohesive zone. The simulations are performed for various initial stretches as shown in Figure 2. A
mesh-shape parameter was varied to measure the mesh dependence of the crack path. Dependence
of crack path on this parameter should decrease for finer meshes. Our numerical simulations
validate this result and further suggest that the crack path computed in the pinwheel mesh is more
stable as the mesh is refined compared to a structured mesh which is used in [2].
970 P. Ganguly and K. D. Papoulia

Figure 2. Crack branching simulations for different initial strains.

References
1. Sharon, E. and Fineberg, J., Microbranching instability and the dynamic fracture of brittle
materials, Physical Review B, 54(10):7128-7139, 1996.
2. Zhang, Z. and Paulino G. H., Extrinsic cohesive modeling of dynamic fracture and
microbranching instability in brittle materials. Unpublished.
3. Miller, O., Freund, L. B. and Needleman, A., Energy dissipation in dynamic fracture of brittle
materials. Modeling and Simulation in Material Science and Engineering, 7, 573-586, 1999.
4. Papoulia, K. D., C.-H. Sam and S. A. Vavasis, Time continuity in cohesive finite element
modeling, Int. J. Num. Methods. Eng., 58(5): 679-701, 2003.
5. Sam, C.-H., K. D. Papoulia and S. A. Vavasis, Obtaining initially rigid cohesive finite element
models that are temporally convergent, Eng Frac Mech, 72(14): 2247-2267, 2005.
6. Ganguly, P., S. A. Vavasis and K. D. Papoulia, An algorithm for two-dimensional mesh
generation based on the pinwheel tiling, SIAM J. Scientific Computing, in press.
 14. Cohesive Models of Fracture 971

A DYNAMIC CRACK GROWTH SIMULATION USING COHESIVE ELEMENTS

M. Anvari and C. Thaulow

Department of Engineering Design and Materials, NTNU
Richard Birkelandsvei 2B, 7491 Trondheim, Norway
Majid.Anvari@immtek.ntnu.no
Christian.Thaulow@immtek.ntnu.no

Rate sensitive and triaxiality dependent cohesive elements are used to simulate ductile crack
growth under quasi static and dynamic loading conditions. A number of finite element analyses are
performed for a center-cracked specimen made of aluminum alloy (6XXX-series). Load-
displacement curves for different cases and considerations are presented and the results are
discussed. The analyses show that, for a cracked specimen, ignoring material rate dependency can
consequence to quite high energy absorption predictions.
When the speed of loading on a structure is high, four additional influences on ductility
compared to quasi-static cases have to be taken into consideration as addresses by El-Magd and
Brodmann [1]: rate dependent plasticity, adiabatic heating, mass inertia, and decrease of local
failure strain.
In finite element representation of cohesive zone modeling (CZM), failure happens in cohesive
elements which are defined as interface elements between undamaged continuum elements. The
stress in the cohesive elements is controlled by a separation dependent law, named traction
separation law (TSL). The TSL used in simulating ductile fracture is meant to represent the process
of void initiation, growth and coalescence. The idea of the present contribution is to obtain it by
studying the stress-elongation response of a single element obeying Gurson type constitutive
equation [2]. TSL generally consists of cohesion and decohesion parts. The opening at which a
cohesive element fails totally is called critical separation, G0, and is one of the fracture parameters.
The other parameter is the maximum traction or cohesive strength, S. The area under the TSL is
the energy absorbed by the cohesive element, *0 and is known as the cohesive energy:

G0
T G  dG
*0
³ 0 (1)
Among the aforementioned phenomena that are important in dynamic fracture, the effect of
strain rate on damage has been considered in the present article. To consider the effect of strain rate
on cohesive element properties, a single plane strain Gurson type element (unit cell) has been
loaded under variety of strain rates. A rate dependent version of the model known as “complete
Gurson model” developed by Zhang et al. [3] has been used. It has been mentioned by variety of
authors, e.g. Hancock and Mackenzie [4] and Brocks et al. [5], that stress state, for example stress
triaxiality, influences the mechanical response of ductile materials. Therefore, the effect of stress
triaxiality on TSL has been examined by applying different values of constant stress biaxialities on
the unit cell, too. In all the cases, the variation of maximum traction and the energy absorbed are
investigated. These values are considered as cohesive strength and cohesive energy as functions of
stress triaxiality and strain rate. The results are then used for crack growth simulations of the
aluminum M(T) specimen. Since the measures of strain rate and stress triaxiality are not available
in the cohesive elements, they are calculated automatically from the adjacent continuum elements
and transferred to them. In the beginning of each calculation, the cohesive parameters are adjusted
to the values transferred. Both quasi-static and transient dynamic analyses are performed and the
load-displacement curves are obtained in different cases.
972 M. Anvari and C. Thaulow

It is shown that cohesive elements have the potential of performing ductile crack growth
simulation not only in static cases, but also in high speed dynamic loading. The cohesive model
presented can be used for both small and large scale yielding and takes effects of different strain
rates and triaxialities into account. Unloading at the crack tip, which happens because of stress
waves, has been implemented by irreversible separation behaviour of the cohesive elements.
The results of the analyses show that generally, ignoring constraint or local strain rate on TSL
makes the analysis underestimate the toughness. The toughness of the structure increases under
dynamic loading because of the inertia and strain rate sensitivity, but for a cracked specimen,
ignoring rate sensitivity of material in high speed of loadings can lead to quite high energy
absorption predictions. Depending on the load speed, material properties, the structure dimensions
and the crack length, the effect of phenomena like elastic waves, strain rate, adiabatic heating and
inertia forces might be different.

References
1. El-Magd, E. and Brodmann, M., Mat. Sci. Eng. A, vol. 307, 143-150, 2001
2. Gurson, J., J. Eng. Mat. Tech., vol. 99, 2-15, 1977
3. Zhang, Z.L., Thaulow, C. and Ødegård, J., Eng. Fract. Mech., vol. 67, 155-168, 2000
4. Hancock, J.W. and Mackenzie, A.C., J. Mech. Phys. Solids, vol. 24, 147-169, 1976
5. Brocks, W., Sun, D.Z. and Honig A., Int. J. Plast., vol. 11, 971-989, 1995
 14. Cohesive Models of Fracture 973

A NEW COHESIVE ZONE MODEL FOR MIXED-MODE DECOHESION

M. J. van den Bosch , P. J. G. Schreurs and M. G. D. Geers

Netherlands Institute for Metals Research (NIMR), Delft, The Netherlands
Eindhoven University of Technology, Eindhoven, The Netherlands
Dep. of Mech. Eng., Eindhoven University of Technology, The Netherlands
m.j.v.d.bosch@tue.nl

This paper focuses on the frequently adopted exponential cohesive zone law of Xu and Needleman
(Xu and Needleman [1]). In spite of its wide applicability, this cohesive zone law gives an
unrealistic physical response under most mixed-mode loading conditions. This conclusion is based
on investigations of the influence of the coupling parameters and the assessment of the dissipated
energy under mixed-mode loading conditions.
One of the assessment methods is to firstly load the cohesive zone in normal direction until
'n,max. At that time an energy Wn has been dissipated. The second step is to break it completely in
tangential direction by shearing it until 't >> Gt. In the second step an energy Wt has been
dissipated, resulting in a total dissipated energy of Wtot (= Wn + Wt). The loading sequence is
shown in Fig. 1.

FIGURE 1. Loading sequence of a cohesive zone. Step 1: load the cohesive zone in normal
direction until ' n  ' n, max and subsequently break it in shear: ' t !!G t .

Tn and Tt are the tractions in normal and tangential direction respectively. The tractions are a
function of both the normalized normal and tangential opening displacements 'n/Gn and 't/Gt.
Where Gn and Gt are characteristic lengths. The initial areas under the curves represent the normal
and tangential work-of-separation: In and It. The shaded areas in Fig. 1 represent the energies
dissipated after loading in normal and tangential direction: Wn and Wt.
The total dissipated energy as a function of 'n,max is depicted in Fig. 2a for the exponential
cohesive zone law of Xu and Needleman. The required energy to break the cohesive zone has a
minimum, which is not likely to occur unless specific experimentally verified mechanisms in the
interface have proven to lead to this kind of behavior. Remedies for this physical problem are not
yet available in literature. Thus, a modified exponential cohesive zone law, based on the law of Xu
and Needleman, is formulated. The total dissipated energy, in case of this modified cohesive zone
974 M. J. van den Bosch et al.

law, is shown in Fig. 2b. The Wtot for the modified cohesive zone law shows a more realistic
behavior for most practical cases: a smooth increase from It to In.

FIGURE 2. The total dissipated energy when the cohesive zone is first loaded in normal direction
until n,max and then broken in shear (see also Fig. 1) for the exponential cohesive zone law of Xu
and Needleman (a) and the modified exponential cohesive zone law (b). In both cases: In = 100
[Jm-2] and It = 80 [Jm-2].

It is shown that the dissipated energy, under mixed-mode loading conditions, can be predicted
by an analytical expression. The mode-I parameters of the cohesive zone law, In and Gn, can be
determined by double-cantilever beam experiments, as described by Sørensen and Jacobsen [2].
The mode-II parameters, It and Gt, might be determined by a similar approach. Results of
experimental mixed-mode bending (MMB) tests (Benzeggagh and Kenane [3]) are used to verify
the predictability of the modified cohesive zone law under mixed-mode loading conditions.

References
1. Xu, X. P., Needleman, A., Mod. Sim. Mat. Sci. Eng., vol. 1, 111-132, 1993
2. Sørensen, B. F., Jacobsen, T. K., Eng. Frac. Mech., vol 70, 1841-1858, 2003
3. Benzeggagh, M.L., Kenane, M. Comp. Sci. and. Tech., vol 56, 439-449, 1996
 14. Cohesive Models of Fracture 975

COHESIVE-ZONE MODELING OF CRACK GROWTH IN SPECIMENS WITH

DIFFERENT CONSTRAINT CONDITIONS

C. R. Chen1, O. Kolednik2 and F. D. Fischer3

1 Materials Center Leoben
2 Erich Schmid Institute of Materials Science, Austrian Academy of Sciences
3 Institute of Mechanics, University of Leoben and Erich Schmid Institute

A-8700 Leoben, Austria

kolednik@unileoben.ac.at

This paper deals with the tri-dimensional (3D) finite element modeling of ductile crack growth in
smooth-sided compact tension (CT) and double edge notched tension (DENT) specimens with
smooth side-surfaces using the cohesive zone model. The main problem is that the cohesive zone
parameters, the separation energy, *, and the cohesive strength, Tmax, vary along the crack front
and also with the crack extension. The purpose of the study is, therefore, to investigate how the
cohesive zone parameters and the in-plane and out-of-plane crack tip constraint are interrelated.
The material is a pressure vessel steel 20MnMoNi55, and the specimen thickness is 10 mm.
The slant shear-lip fracture near the side-surfaces is modeled as a normal fracture along the
symmetry plane of the specimen. The cohesive zone parameters are determined by fitting the
simulated crack extensions found by the 2D and 3D computations to the experimental data of the
multi-specimen tests. The effects of element size and load increment on the results of the
simulations are also studied.
The results show that * affects both the moment of fracture initiation and the crack growth
rate, whereas Tmax strongly affects the crack growth rate, but has little effect on the fracture
initiation. For the same cohesive zone parameters, the crack tip triaxiality near the midsection is
lower in DENT specimens than in CT specimens. When the separation energy is set constant for
CT and DENT specimens, the cohesive strength for the DENT specimens should be significantly
lower than that for the CT specimens in order to make the simulated crack extensions near the
midsection fit to the experimental data, see Fig. 1.
For constant cohesive zone parameters, the simulated crack extension shows a strong tunneling
effect, however for the CT-specimen the crack extension near the side surfaces is still too large
(Fig. 1). For a better fit between simulated and experimental crack growth, both the cohesive
strength and the separation energy near the side-surface should be considerably reduced compared
to the values near the midsection (Fig. 2). When the same cohesive zone parameters are applied to
the 3D model and a plane strain model, the stress triaxiality in the midsection of the 3D model is
much lower and the von-Mises equivalent stress is distinctly higher than in the plane strain model.
976 C. R. Chen et al.

FIGURE 1. The cohesive zone parameters that make the simulated crack extensions near the mid-
section fit to the experimental data: (a) * = 180 kJ/m2 and Tmax = 3.14 Vy for CT; (b) * = 180 kJ/
m2 and Tmax = 2.58 Vy for DENT.

FIGURE 2. (a) Assumed variations of the cohesive strength and the separation energy in thickness
direction of the CT-specimen; (b) Comparison of the simulated crack extension curves with the
experimental data.

Acknowledgements
The authors acknowledge gratefully the financial support of this work by the Materials Center
Leoben under the project numbers SP7 and SP14.

References
1. Chen, C.R., Kolednik, O., Scheider, I., Siegmund, T., Tatschl, A., Fischer, F.D., Int. J.
Fracture, vol. 120, 517-536, 2003.
2. Chen, C.R., Kolednik, O., Heerens, J., Fischer, F.D., Eng. Fract. Mech., in press.
3. Chen, C.R., Kolednik, O., Int. J. Fracture, in press.
 14. Cohesive Models of Fracture 977

EFFECT OF ANISOTROPIC PLASTICITY ON MIXED MODE INTERFACE

CRACK GROWTH

V. Tvergaard and B. N. Legarth

Department of Mechanical Engineering, Solid Mechanics
Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark
viggo@mek.dtu.dk

Crack growth in metals gives rise to plastic work in the material surrounding the crack-tip, which
contributes significantly to the fracture toughness, such that the macroscopic work of fracture is
much larger than that of the local fracture process. For mode 1 crack growth with isotropic
plasticity this has been studied numerically by Tvergaard and Hutchinson [1], modeling the local
fracture process by a traction-separation law along the crack plane, with a specified work of
separation per unit area. Subsequently, the same authors have used the method to account for the
effect of a T-stress on crack growth [2].
Crack growth along an interface between an elastic-plastic solid and a rigid solid has been
analysed by Tvergaard and Hutchinson [3], using a traction-separation law to represent the local
fracture process, and Tvergaard [4] has extended this analysis to consider crack growth between
dissimilar elastic-plastic solids. Due to the mismatch of elastic properties on either side of the
interface the elastic crack-tip fields are characterized by an oscillating stress singularity, and the
analyses under conditions of small scale yielding make use of these elastic fields as boundary
conditions on the outer edge of the region analysed.
Plastic anisotropy is often important in practice, due to texture development during large
plastic straining in a metal forming process, or due to microstructure effects, as e.g. inclusions
elongated in a particular direction (Legarth [5]). Several different models of anisotropic plasticity
have been discussed by Kuroda and Tvergaard [6] in relation to studies of the forming limits for
thin metal sheets with anisotropic properties. For crack-tip blunting one of these anisotropic yield
surfaces (Hill [7]) has been used to the study the evolution of the crack-tip fields (Legarth et al. [8],
[9]). Subsequently, two of the models (Hill [7], Barlat et al. [10]) have been used in numerical
studies of crack growth (Tvergaard and Legarth [11]), with the local fracture process modeled by a
traction-separation law, as in [1]-[4]. In these applications to fracture mechanics elastic-
viscoplastic versions of the anisotropic plasticity models have been applied, so that strain-rate
sensitivity is accounted for. Recently, the crack growth analyses for anisotropic plasticity have
been extended by Tvergaard and Legarth [12] to incorporate an anisotropic plasticity model with a
vertex-type plastic flow rule, proposed by Kuroda and Tvergaard [13].
In the studies to be presented here, the effect of anisotropic plasticity is analysed for crack
growth along an interface between a ductile material and an elastic material that does not yield
plastically. While the crack growth analyses in [9] have considered only mode I loading
conditions, the present analyses, with anisotropic plasticity only on one side of an interface, are
carried out for various degrees of mixed mode loading. As in previous analyses for isotropic
plasticity [3,4], it is found that plastic flow near the crack-tip results in much increased resistance
to crack growth when mode II conditions dominate, but the main focus here is on differences
resulting from anisotropy.

References
1. Tvergaard, V. and Hutchinson, J.W., J. Mech. Phys. Solids, vol. 40, 1377-1397, 1992.
978 V. Tvergaard and B. N. Legarth

2. 2. Tvergaard, V. and Hutchinson, J.W., Int. J. Solids Structures, vol. 31, 823-833, 1994.
3. Tvergaard, V. and Hutchinson, J.W., J. Mech. Phys. Solids, vol. 41, 1119-1135, 1993.
4. Tvergaard, V., J. Mech. Phys. Solids, vol. 49, 2689-2703, 2001.
5. Legarth, B.N., Int. J. Mech. Sci., vol. 45, 1119-1133, 2003.
6. Kuroda, M. and Tvergaard, V., Int. J. Solids Structures, vol. 37, 5037-5059, 2000.
7. Hill, R., Proceedings of the Royal Society of London A 193, 281-297, 1948.
8. Legarth, B.N., Tvergaard, V. and Kuroda, M., Int. J. of Fracture, vol. 117, 297-312,
9. 2002.
10. Legarth, B.N., Tvergaard, V. and Kuroda, M., On-line publication of WCCM V –
11. ISBN 3-9501554-0-6, 2002. http://wccm.tuwien.ac.at/
12. Barlat, F., Lege, D.J. and Brem, J.C., Int. Journal of Plasticity, vol. 7, 693-712, 1991.
13. Tvergaard, V. and Legarth, B.N., Int. J. of Fracture, vol. 130, 411-425, 2004.
14. Tvergaard, V. and Legarth, B.N., Int. J. Solids Structures (2005) (in print).
15. Kuroda, M. and Tvergaard, V., J. Mech. Phys. Solids, vol. 49, 1239-1263, 2001.
 16. Environment Assisted Fracture 979

CHARACTERISATION OF TG-SCC IN PURE MAGNESIUM AND AZ91

ALLOY

N. Winzer1, G. Song1, A. Atrens1, W. Dietzel2 and C. Blawert2

1Materials Engineering, The University of Queensland, Brisbane, Qld Australia 4072
2GKSS-Forschungszentrum Geesthacht GmbH, Germany

a.atrens@minmet.uq.edu.au
wolfgang.dietzel@gkss.de

Our recent critical review [ ] of the stress corrosion cracking (SCC) of magnesium aims to provide
a foundation for the safe and effective use of magnesium alloys, including practical guidelines for
the service use of Mg alloys in the atmosphere and/or in contact with aqueous solutions. This is to
provide support for the rapidly increasing use of Mg in industrial applications, particularly in the
automobile industry. These guidelines should be firmly based on a critical analysis of our
knowledge of SCC based on (1) service experience, (2) laboratory testing and (3) understanding of
the mechanism of SCC, as well as based on an understanding of the Mg corrosion mechanism.
Early reports have led to the perception of high SCC resistance for pure Mg and Mg-Mn alloys.
Subsequent research on pure Mg, Mg-Al alloys and Zr containing alloys have shown that SCC is a
significant issue, and that SCC can occur for loading equivalent to 50% of the yield stress for many
combinations of alloy + (common) environment. The review [1] indicated that the following must
be rationalised in any comprehensive mechanism of Mg SCC.
• "Pure Mg suffers transgranular stress corrosion cracking (TGSCC),
• "Mg alloys suffer TGSCC in distilled water.
• "TGSCC in Mg alloys is related to hydrogen,
• "Small additions of sulphate and bromide ions to distilled water decrease significantly the
threshold stress intensity and significantly increase the steady state crack velocity.
These observations may be rationalized in terms that the environment influence may be
characterized in terms of an effective hydrogen fugacity, and it is the aim of our experimental
program to quantify this approach. A number of recommendations are given [1] for the prevention
of SCC of Mg alloys exposed to the atmosphere or aqueous solutions. One of the most important
recommendations might be that the total stress in service (i.e. the stress from the service loading +
the fabrication stress + the residual stress) should be below (and during service should remain
below) a threshold level, which, in the absence of other data could be (conservatively) estimated to
be ~ 50% of the tensile yield strength.
On the basis on the critical review [1], a research program has been devised as illustrated in
Fig. 1. The work reported in this paper is part of that research program. Transgranular Stress
Corrosion Cracking (TG-SCC) of Mg and Mg-alloys has been investigated for ZE41 and AZ91
alloys in distilled water, NaCl solution and NaCl+K2CrO4 solution, using the linearly increasing
stress test (LIST) [ ], Fig. 2. ZE41 and AZ91 represent the two principal classes of Mg alloys;
zirconium-containing alloys and others respectively. Pure Mg was investigated to study the
influence of second phases. The study investigated the influence of stressing rate on crack velocity,
which was characterised using the DC potential drop (DCPD) technique. Fractography and surface
composition analysis using X-Ray Photon Spectroscopy (XPS) was carried out comparing samples
fractured in air, subject to TGSCC in solution and for specimens pre-exposed to SCC
environments.
980 N. Winzer et al.

FIGURE 1. Schematic of research program on environment influences for TGSCC of Mg Alloys.

FIGURE 2. Schematic drawing showing the principle of the LIST apparatus. The load on the
specimen is increased linearly by means of a lever principle and a linearly moving load on the right
hand side of the lever. The lever is maintained horizontal via a linear actuator and a servocontroller
by means of a displacement signal from one end of the lever arm.

References
1. N. Winzer, A. Atrens, G. Song, E. Ghali, W. Dietzel, K. U. Kainer, N. Hort and C. Blawert,
Advanced Engineering Materials, 2005 to appear in August issue.
2. A. Atrens, C. C. Brosnan, S. Ramamurthy, A. Oehlert and I. O. Smith, Meas Sci Technol
vol 4, 1281-1292, 1993.
 16. Environment Assisted Fracture 981

HYDROGEN EMBRITTLEMENT AND CRACKING OF 18MN-4CR STEELS

A. Balitskii
Karpenko Physico-Mechanical Institute of National Academy of Sciences of Ukraine
Naukova Street, Lviv, 79601, UKRAINE
balitski@ipm.lviv.ua

The number of generators retaining ring failures caused by the fracture of these details increases
with the power of electric machines. Since hydrogen is used as a cooling agent, these accidents
leads to fires, cause damage of generators, or even result in their complete failure. Statistical data
of this sort are mainly accumulated for 18Mn-4Cr-type steels [1-8].
The temperature dependence of the fracture toughness of 18Mn-4Cr-type steels is
characterized by presence of the minimum at about 550 oC (Fig.1). After thermal treatment
performed at 400 oC, fracture toughness abruptly drops to its minimum value and then somewhat
increases but does not reach the original value. At temperatures higher than 600 oC, fracture
toughness begins to decrease again. If tempering temperatures are not higher than 400 oC, we
observe plastic fracture. At higher tempering temperatures (e.g. 600 oC), we have a mixed type of
fracture, i.e., a combination of plastic (intragranular) and brittle (intergranular) types of fracture.
The time to failure of 18Mn-4Cr-type steels is the same in dry hydrogen and in dry air (Fig.2).

FIGURE 1. Temperature dependences of fracture toughness and sub critical crack growth rates in
specimens from18Mn-4Cr steel [9].
FIGURE 2. Long term strength of 18Mn-4Cr steel in dry hydrogen and in air (1), in distilled water
without oxygen (2), in humid hydrogen (3), in water saturated with oxygen (4), and in a nitrate
solution (5) [10].

Humid hydrogen is a much more aggressive medium than distilled water without oxygen.
Saturation of water with oxygen and nitrated leads to asharp decrease in the time of specimes
fracture. The diagrams of structural strength of the materials of a rotor retaining ring (traditional
and promising) presented in Fig.3 illustrated the advantages of new 18Mn-18Cr steel.
982 A. Balitskii

FIGURE 3. KIC – VVT diagrams ({ – KJC, z – KIC) (a) (in air), Kfc – VVT (b), Kth – VVT (c)
(electrolytical hydrogenation with current density 100 /m2) of steels for rotor-retaining ring unit:
1 – 3,5NiCrMoV; 2 – 8Mn8Ni4Cr; 3 –18Mn-4Cr; 4 – 18Mn-18Cr.

To increase the crack growth resistance of retaining ring materials in hydrogen environments
admixtures of La (up to 0.05 vol. %), Ce (up to 0.1 vol. %) in N-containing 18Mn-4Cr have been
introduced during experimental melting. They have a positive influence on quantuity, geometry and
distribution regularity of non-metallic inclusions, improve the metal quality, and change its dislocation
structure. Applying Ca (up to 0.05 vol. %) for deaccidation of steels significantly influences the
geometry and composition of nonmetallic inclusions, observed on fracture surfaces of specimens.
Experimental steels with high content of Cr, Cu and other alloying components demonstrated
increasing resistance to corrosion-mechanical fracture and long-term stength in hydrogen-containing
environments.

References
1. Balitskii, A.I., Modern Materials for Powerful Turbogenerators, National Academy of
Sciences of Ukraine, Lviv, 1999.
2. Balitski A., Krohmalny O., Ripey I., Intern. J. of Hydr. Energy, vol.25, 2, 167-171, 2000.
3. Balyts’kyi O.I., Mater. Sci. , vol.33, 4, 539-552, 1997.
4. Balitskii A.I., Mater. Sci. , vol.34, 4, 113-120, 1998.
5. Balitskii A.I., Mater. Sci. , vol.35, 4, 485-490, 1999.
6. Balitskii A.I., Mater. Sci. , vol.36, 4, 541-545, 2000.
7. Balitskii A.I., Makarenko V.G., Shved M.M., Shokov N.A. In Abstr. of IV Sem. on Hydrogen
in Metals, vol.2, Moscow, 1984, 137.
8. Balitskii A.I., Shokov N.A., In Abstr. of 2nd Symp. on Fracture Mechanics, vol.II, Zhytomir,
1985, 61.
9. Scarlin R. B., Albrecht J., Speidel M. O., In Proc. 8th Int. Brown Boweri Symp., Plenum
Press, New-York, London, 1984, 453–461.
10. Speidel M. O., VGB Kraftwerktechnik, vol.61, 5, 1981, 417–427.
11. Lukas P., Kunz L., Bartos J., Mat.Sci.Eng.,vol.56, 1982, 11-18.
12. Shuju H., Xiaofang L., Shufeng Z., Yuanchun H., Haicheng G., Corrosion, vol. 55, 12, 1999,
1182 – 1190.
 16. Environment Assisted Fracture 983

TRANSIENT STRESS AND EAC OF STEAM TURBINE DISC STEEL

A. Turnbull and S. Zhou

National Physical Laboratory
Queens Road, Teddington, Middlesex, TW11 0LW, UK
alan.turnbull@npl.co.uk

For economic reasons, the power industry is now operating its 500 MW coal-fired plants on a two-
shift cycle in which the turbines are on-load for 16 hours per day and off-load overnight and at
weekends. The concern with ‘two-shifting’ is the impact on environment assisted cracking of the
associated transients in stress, water chemistry and temperature. On-load, with well-controlled
water chemistry, the condensate on the low-pressure turbines will be free of oxygen with chloride
and sulphate levels both up to about 300 ppb. Off-load, the condensate would essentially be pure
water but aerated unless there is nitrogen blanketing. The stress off-load would be zero. Ideally, to
fully simulate two-shifting in laboratory testing, the combined influence of transient stress and
water chemistry would be evaluated but there are technical difficulties in synchronising the
changes in the stress, temperature, oxygen, and anion (chloride and sulphate) concentrations. For
the purpose of assessing the impact of transient stress on crack propagation, the environment was
held constant, viz. deaerated 300 ppb Cl- +300 ppb SO42- solution at 90 °C. Separate measurement
to examine the effect of transient water chemistry at constant stress are underway.
Fatigue precracked compact tensions specimens were contained in a environmental chamber
and the test solution ciculated from a 22 L reservoir that was refreshed weekly to maintain good
water chemistry. The oxygen concentation was monitored on-line and was typically 1 ppb. A
trapezoidal load cycle was used to reflect service behaviour. It is desirable to sensibly accelerate
the process to maximise the number of stress cycles per day. In this case, a 20 min period (a typical
hot start period) was fixed for both the load rise and fall time and for zero load. The maximum load
was set to give a K value of 40 MPa m1/2 for 100 mins. Later, this hold period was varied to assess
the impact on crack growth. In parallel with the transient load tests, two tests at constant stress
intensity factors of 30 MPa m1/2 and 40 MPa m1/2 were undertaken. A pulsed high resolution/high
stability DCPD method was used to monitor crack length.
Unusual behaviour was experienced in tests of the disc steel under constant load. No crack
growth was observed for well-controlled conditions extending to 9 months. However, in one test
an uncontrolled excursion in oxygen and chloride solution lasting just 1 hour, by which time the
environmental conditions had been restored to normal, induced a crack growth rate of 9.0x10-12 m/
s that was sustained for a further 4 months, after which it decreased to 2.3x10-12 m/s. In another
test, no cracking was observed after 9 month under constant load. To possibly activate the crack
growth process, a single trapezoidal load was applied. No crack extension was detected in the next
2.8 months after the transient load application but the crack then started growing at a rate of
2.2x10-12 m/s.
An example of the results from transient loading is shown in Fig. 1.
984 A. Turnbull and S. Zhou

FIGURE 1. Crack growth rates under trapezoidal loading for turbine disc steel in deaerated 300
ppb Cl- +300 ppb SO42- solution at 90 °C. Loading wave 1 has hold time of 100 mins and loading
wave 2 has hold time of 300 min.

There is no apparent static load contribution to the crack growth rate; the crack growth is not
sustained under static load condition (the slight decay in apparent crack length is a reflection of
corrosion product we believe) and there is no effect of hold time on the cylic crack growth rate.
The cyclic crack growth rate is high, about 10-6 m/cycle. With typically one cycle per day in
service except weekends, and accounting for any stress corrosion cracking component at
maximum load (to allow for chemistry transients) a crack growth rate of 0.4 mm per year would be
estimated. Crack inspection intervals should be adjusted accordingly under two-shifting operation.
With proposals for four-shifting under consideration, life expectancy will be very significantly
reduced.
 16. Environment Assisted Fracture 985

IRREVERSIBLE HYDROGEN TRAPPING IN WELDED BETA-21S TITANIUM

ALLOY

D. Eliezer1, 2, E. Tal-Gutelmacher1, C. E. Cross2 and Th. Boellinghaus2

1Dept.Materials Engineering, Ben-Gurion University of the Negev, Beer-Sheva 84501, Israel
2Federal Institute for Materials Testing and Processing (BAM), Berlin D-12200, Germany

deliezer@bgu.ac.il

ȕ - 21S titanium alloy is ranked among the most important advanced materials for a variety of
technological applications, due to its combination of a high strength/weight ratio, good corrosion
behavior and oxidation resistance. However, in many of these technological applications, this alloy
is exposed to environments which can act as sources of hydrogen, and consequently, hydrogen-
induced cracking and property degradation, hydrogen-induced ductile-to-brittle transition
associated with a change in the fracture mode from ductile, micro-void coalescence to brittle,
cleavage have to be considered[1-4]. In the aged E-21S alloy, the susceptibility to hydrogen induced
cracking and the decrease in the alloy's strength has been attributed to the E-phase precipitated
during the aging and the hydrogen-induced stabilization of the E-phase[1, 2, 5]. Hydrogen-induced
intergranular cracking in the cathodically pre-charged E-21S alloy was significantly influenced by
the preferential Į precipitation at E grain boundaries[5]. Even without hydriding, the Į -E
interfaces could provide trapping sites and the accumulation of hydrogen at these interfaces could
result in fracture. Pound[6] revealed that in aged E-Ti alloys the relationships among the trapping
constants, resistance to hydrogen embrittlement and grain boundary are critical in determining the
role of trapping in hydrogen embrittlement of these alloys.
Very few data are available in literature for intrinsic hydrogen effects in welded E-Ti alloys
and almost no reports on trapping characteristics. Thus, in this work hydrogen was introduced to
the alloy via the arc shielding gas (argon with pre-mixed amount of hydrogen (5% H2) at a flow
rate of 20 L/mi), using a GTA welding process for full penetration, bead-on-plate welds on E-21S
plate. GTAW allowed us to investigate the hydrogen evolution phenomenon from the welded E-
21S alloy and, in particular, to determine whether the trapping characteristics can change in
correlation with the unique morphology induced in the alloy’s microstructure by the welding
process. NDT analyses exposed no porosity or any other volumetric defects in the welds. The
amount of hydrogen inside the welded specimens, measured by Leco, was 455 ± 29 wt. ppm, in
comparison to 56 ± 6 wt. ppm in the as-received material. SEM microstructure investigations
showed quite clearly the boundary between the dendritic microstructure in the fusion zone and the
equiaxed grain microstructure in the HAZ with coarser quenched grains in comparison to the base
metal. The microstructure developed in the weld metal varied noticeably from the edge to the
centerline of the weld. EDS analysis revealed that enrichment of Mo and depletion of Ti takes
place in the interdendritic regions.
Hydrogen desorption and trapping characteristics were determined by means of thermal
desorption spectroscopy (TDS) and were supported by other experimental techniques, such as
LECO hydrogen determinator, XRD and SEM microstructure investigations. Detailed technical
descriptions of the Leco and TDS system, as well as the quantification of several trapping
parameters are given elsewhere[7]. TDS analyses are conducted on the welded E-21S titanium
specimens using different temperature ramps of 3, 5 and 7 ºC/min. The results were summarized in
Table 1.
986 D. Eliezer et al.

TABLE 1. Summary of TDS parameters for GTA welded -21S alloy.

The TDS plots were characterized by only one desorption peak, which occurs at very high
temperatures (623-670ºC). The calculated energy for hydrogen desorption was found to be
approximately 111 kJmole-1. This is a very high value, indicating that the residual hydrogen
induced to the material from the gas shield during the welding process, is probably trapped at an
irreversible, strong trapping site. Furthermore, even at a slow rate of 3 ºC/min, less than half of the
absorbed amount of hydrogen is desorbed. XRD patterns of the weld metal revealed the existence
of Į -phase in addition to the E-phase. Therefore, similar to aged E-Ti alloys, irreversible trapping
might take place at the Į -E interface[8], resulting from possible misfit strain, or some other cause.
Another assumption could be that the dendritic arms spacing might act as a possible irreversible
trap site, for hydrogen accumulation at the interface might occur due to the high residual stresses
induced by the welding process. Since hydrogen evolution and trapping can be affected by several
factors, i.e., the unique microstructure; the shape, morphology and size of the dendrites, elements
segregation and the change in the chemical composition, phase transitions that might occur,
residual stresses induced by the welding process, the reasons for such deep trapping associated
with welding have to be investigated further.

References
1. Nelson, H.G., In Hydrogen Effects in Metals, edited by A.W. Thompson, N. R. Moody, TMS,
Warrendale (PA), 1996, 699.
2. Hardwick, D.A., In Hydrogen Effects in Metals, edited by A.W. Thompson, N. R. Moody,
TMS, Warrendale (PA), 1996, 735.
3. Teter, D.F., Robertson I.M. and Birnbaum, H.K., Acta Mater., vol. 49, 4313, 2001.
4. Sofronis, P. et. al., In Hydrogen Effects on Material Behavior and Corrosion Deformation
Interactions, edited by N.R. Moody, A.W. Thompson, E.E. Ricker, G.S. Was, R.H. Jones,
TMS-AIME, Warrendale (PA), 2003, 537.
5. Young, G.A. and Scully, J.R., Scripta Metall. Mater., vol. 28, 507, 1993.
6. Pound, B.G., Acta Mater., vol. 45, 2059, 1997.
7. Tal-Gutelmacher E., et. al., Mater. Sci. Eng. A, vol. 52(8), 230, 2004.
8. Young, G.A. and Scully, J.R., Corrosion, vol. 50, 919, 1994.
 16. Environment Assisted Fracture 987

EAC IN HIGH STRENGTH STEELS FOR GAS TRANSPORTATION

G. Gabetta and R. Bruschi

1 Eni S.p.A. Div.E&P, 5° Palazzo uffici, Via Emilia 1, 20097 San Donato Milanese (Mi) Italy
giovanna.gabetta@agip.it
Snamprogetti, Via Toniolo 1, 61032 Fano (An) Italy
Roberto.bruschi@snamprogetti.eni.it

The use of high grade steels for pipelines (X80-X120 API 5L) gives the possibility of competitive
gas transportation over long distance (3000-7000 km), with high capacity (15-30 Gsm3/year) and
high pressure (10-15 MPa). Pipelines with those characteristics mast cross hostile environments,
such as: permafrost, swamps, seismic activities, slopes, hydro-geo hazards.
Long distance pipelines require large investment (of the order of 10 Billion US$), and 20-30
years of safe and efficient operation to have an adequate return of the investment. With the above
requirements, materials risk must be minimal, in other words defect tolerance of the material
becomes an important issue.
Underground pipeline currently in operations are mainly made of low alloy steels of the API
specifications (for instance, X65 and X52), showing a ferritic-pearlitic structure with elongated
inclusions, mainly MnS. These materials have been proved susceptible to Transgranular Stress
Corrosion Cracking (TGSCC) in diluted solutions of near neutral pH, with composition similar to
the underground water. This cracking phenomenon has been firstly studied by Parkins [ 1,2 ] as a
consequence of field observations [ 3 ]. Cracking was mainly observed under disbonded region of
the pipeline coating, where cathodic protection is not effective. Hydrogen embrittlement due to the
diffusion of atomic hydrogen produced in the cathodic reaction at the crack tip has been identified
as the most probable active cracking mechanism, but the influence of the different variables
(metallurgy, stress, environment) is not yet fully understood [ 4 ]. An important effect of time
dependent stress variations, such as pressure fluctuation and/or land slide movements, was
demonstrated [ 5,6 ].
To obtain steels with higher strength, suppliers act modifying microstructure and composition,
so that the so-called micro-alloyed steels are produced. The goal of the producing process is to
increase the strength, while decreasing (or not increasing) the toughness, taking into account the
larger wall thickness, required for long distance transportation. In addition, there is a need for a
transition temperature lower than -10°C (it can depend on external environmental: in artic
conditions, this temperature can be lower), and of an adequate weldability.
Due to differences in microstructure and composition, the comparison between old grades and
new grades steels (and between new grades from different mills) is not immediate. Moreover, there
is a lack of significant field experience, as Environmentally Assisted Cracking for pipelines often
presents a long incubation time, difficult to predict and to handle. In particular, the damage can
only be detected after many years of service and with only a short time interval remaining between
the first detection and the final failure. In the frame of the TAP (High Pressure Gas Transportation)
project, a first goal is to ensure that X100 steel will not "behave" worse than X65 steel in service.
Experimental data obtained up to now are however somewhat contradictory concerning the effect
of metallurgy: a coarse microstructure is definitely less resistant to EAC than a fine microstructure,
but there is an effect of the different phases, as for instance that of interstitial elements, that could
be quite important.
To obtain data on the new material, a pilot section was built in Sardinia, in a military firing
range. A 48" pipeline 750m long, designed for 15MPA, will be "operated" for about two years.
988 G. Gabetta and R. Bruschi

The objective of the test is to verify the constructability of X100 pipeline, and to test the
performance of X100 steels (joints will be provided by three different suppliers), under different
working factors in realistic environments. Typical pipe defects will be simulated in the pilot
section, to obtain field feedback on hydrogen embrittlement and EAC problems, at different levels
of cathodic protection and mechanical damage. Pressure fluctuations will be applied to simulate
about 25 years of operation of a trunk line for gas transportation.
In the mean time, pipe samples from each pipe mill (having different microstructure) are
tested in laboratory. Part of testing is dedicated to investigate the effect of crack nucleation and to
compare the behaviour of X65 "old" steel to the behaviour of the new X100 steel. There is a need
for additional efforts in order to validate specific test techniques and to develop procedures that can
yield quantitative results ("figures"), and - most important - that can provide users with reliable
figures.
To pursue the goal for X100 steel, a deeper analysis of results and information will be
necessary. Laboratory tests and literature results are important. The phenomenon of TGSCC was
observed not only in pipeline steels, but in other components, such as for instance in nuclear
pressure vessel steels [ 7 ]. Tests and discussion are underway since about 30 years on that subject.
People involved in the design and building of large structure are facing new challenges and are
looking for the application of the scientific findings.

References
1. Parkins, R.N., "Mechanisms of Stress Corrosion Cracking" Cap.8.2, in "Corrosion" L.L.Sheir,
Ed., Newness-Butterworths, London, 1963
2. Parkins, R.N., Metals Performance, vol. 24, N°8, 1985.
3. TransCanada Pipelines: "Report on 1987 Pipe Integrity Program, SCC Research Program and
Planned 1988 SCC Research Program", 1988
4. Zheng, W., Revie, R.V., Dinardo, O., MacLeod, F.A., Tyson, W.R., and Kiff, D.: "Pipeline
SCC in Near-Neutral pH Environment: Effect of Environmental and Metallurgical
Variables", EPRG 1996
5. Gabetta, G., DiLiberto, S., Bennardo, A., "Laboratory tests reproduce transgranular stress
corrosion cracking observed in field", paper N°00372, Corrosion 2000, Orlando, Fl, March
26-31, 2000
6. Gabetta, G., Di Liberto, S., Bennardo, A., Mancini, N., "Strain rate induced stress corrosion
cracking in buried pipelines", British Corrosion Journal, vol.36, n°1, 2001
7. M.Elboujdaini and W.Revie, "Stress Corrosion Cracking on Canadian Pipelines", ICG-EAC
25th Anniversary celebration, May 11-16 2003, Ottawa, Canada
 16. Environment Assisted Fracture 989

HIGH TEMPERATURE FATIGUE CRACK GROWTH IN TITANIUM

MICROSTRUCTURES

H. Ghonem
Department of Mechanical Engineering
University of Rhode Island
Kingston, Rhode Island 02881 (USA)
ghonem@egr.uri.edu

Titanium alloys are interesting to several industrial applications including transportation and
biotechnology’s, aerospace designers having been particularly interested because of the
combination of their fatigue and time-dependent mechanical properties, good formability, high
specific properties due to a relatively low density, and high resistance to aggressive environmental
and impact loads. Titanium based alloys represent 30% and 40% of the total mass in commercial
and military engines, respectively, a considerable amount of this mass being devoted to high
temperature rotating disc and blade components.
The improvement of the engine performance being directly related to the increase of the
operative temperature, Ti-rich Ti-Al alloys (Al < 12.4 wt %) which account for this goal are the
most critically studied. These alloys are used under cyclic conditions in aggressive environments
such as ambient air at elevated temperature, and can be submitted to the superposition of constant
loading and large number of small amplitude perturbations.
This paper examines the loading frequency effects on elevated temperature fatigue crack
growth mechanisms in titanium alloys in two distinctive microstructures, Widmanstatten and fully
lamellar.
In this examination, the individual influences of environment and creep are identified in terms
of both the macroscopic crack growth rates and detailed fractographic observations of the crack
growth path as a finction of loading frequency and the applied '. .
Results in both microstructures indicate the existence of two regions of microstructurally
sensitive and insensitive transgranular crack growth. The transition between these two regions has
been found to depend on the testing temperature.
In the Widmanstatten microstructure, lower frequencies promoted an increase in cleavage
fracture while an addition of hold time at maximum load resulted in prior-E grain boundary
fracture attributed to creep damage. In addition, the transition to intergranular fracture at high '.
has been found to take place at higher frequencies in vacuum, yielding a higher crack growth rate
than in air which is indicative of a faster, more widespread viscoplastic flow without the presence
of oxidation.
The controlling feature in the fracture mechanism in this microstructure was identified as the
intercolony boundaries. In fully lamellar microstructure, the crack growth rate is found to be
insensitive to variations in lamella size. Furthermore, the presence of hold time within the loading
cycle results in a fracture failure dominated by prior E intergranular fracture and DE interface
decohesion.
The role of the loading frequency is correlated in this microstructure to the associated crack tip
shear activity and transmission at the DE interfaces. These interfaces are identified as the
controlling fracture feature in fully lamellar microstructures. Experimental results as a function of
990 H. Ghonem

temperature and frequency are detailed and a discussion focusing on the role of environment in
observed fracture mechanisms is presented.
 16. Environment Assisted Fracture 991

CORROSION DAMAGING AND CORROSION FATIGUE ASSESSMENT IN

THREE-LAYERED METALLIC MATERIAL

I. M. Dmytrakh and V. V. Panasyuk

Karpenko Physico-Mechanical Institute of National Academy of Sciences of Ukraine
5 Naukova Street, Lviv, 79601, UKRAINE
dmtr@ipm.lviv.ua, panasyuk@ipm.lviv.ua

The layer-like structures are widely used for corrosion protection of hull constructions in a power
engineering and refinery industry. The work presents the corrosion and corrosion fatigue studies of
three-layered metallic material with the aim to assess the possible corrosion damaging and
corrosion fatigue crack growth behaviour under operating conditions.
The subject of study consists of a basic (hull) material and two fused layers by thickness of
3.5mm each. The nominal chemical composition of each component is given in TABLE 1. The 1%
solution of H3BO3 in distilled water with KOH additions (pH=8.0) was used as corrosive
environment.

TABLE 1. Chemical composition of three-layered material (% by wt).

FIGURE 1. Corrosion current density from the surface of cylindrical pit of variable radius r and
constant depth h=25 mm in three-layered material.

Based on the proposed methods of equivalent electrode for determination of electrochemical

current in corrosion pits [1], the numerical procedure for the determination of a corrosion current
density from the surface of deep cylindrical corrosion pit in three-layered material was developed.
Using the standard electrochemical parameters for each material in the given environment, the
corresponding values of the corrosion current density were calculated under different combinations
of pit’s depth and radius. The sample of a calculation is given in Fig. 1.
It has been shown that current density increases with increasing of pit’s radius and decreases
under increasing of its depth. From received results also follows that the zones of fusion have
992 I. M. Dmytrakh and V. V. Panasyuk

increased corrosion activity, especially it can be seen for zone of fusion between layer I and basic
material (Fig. 1).

FIGURE 2. Diagrams of fatigue crack growth resistance for three-layered material in air (a) and in
corrosive environment (b).

Based on developed fracture mechanics experimental methods [2], fatigue crack growth
behaviour in three-layered material was investigated both in air and corrosive environment. As
result the diagrams of fatigue crack growth resistance (fatigue crack growth rate da/dN as function
of stress intensity factor range 'K) were built (Fig. 2). These diagrams reflect an influence of
corrosive environment and provide the data for quantitative assessment of fatigue strength of
components and zones of fusion for given three-layered material.

References
1. Dmytrakh, I. M., Kolodii, B. I. and Bilyi, O. L. Physicochemical Mechanics of Materials.
International Scientific-Technical Journal, vol. 39, No 4, 14-18, 2003
2. Dmytrakh, I. M. and Panasyuk, V. V., Influence of Corrosive Environments on Deformation
and Fracture of Metals Near Stress Concentrators, National Academy of Sciences of
Ukraine, Lviv, Ukraine, 1999
 16. Environment Assisted Fracture 993

SIMULATION OF HYDROGEN ASSISTED STRESS CORROSION CRACKING

USING A TIME DEPENDENT COHESIVE MODEL

I. Scheider, M. Pfuff and W. Dietzel

GKSS Research Centre Geesthacht, Max-Planck-Str. 1, D – 21502 Geesthacht
ingo.scheider@gkss.de

Cohesive models are used for ductile tearing of metals since 1987 [1]. Nowadays, the numerical
implementation using finite element technique is such that interfaces elements are introduced,
which transfer cohesive stresses between continuum elements until they fail and form a crack. The
constitutive law of the cohesive model, the so-called traction-separation law, which defines the
dependence of the cohesive traction, V c, on the material separation, , contains two parameters, the
cohesive strength, V 0, and the critical separation, G 0. The integral of the function V c( G ) gives
the total energy dissipated by the cohesive element, called cohesive energy, * 0. During the last
decade, the traction-separation law has been extended to cover time-dependent processes. In the
case of stress corrosion cracking, the time dependence is caused by the differential equation of
hydrogen diffusion, which in its simplest form is

wC
Deff ’ 2 C
wt (1)
where C is the bulk hydrogen concentration and Deff the effective diffusivity. Implementations of
such differential equations have been introduced in cohesive modelling recently [2, 3]. The
approach in both investigations is based on the assumption that the hydrogen coverage of the
interface reduces the cohesive energy and thus, the cohesive strength V c,0 of the material.
The present paper addresses the development and implementation of a concentration
dependent cohesive model in the commercial finite element package ABAQUS. Originally, the
authors implemented interface elements for stable crack extension in ductile metals, [4]. This
implementation is now extended to cover a linear dependence of the cohesive strength on the
hydrogen concentration at the interface by

Vc V c ,0 (1  P C )
(2)
where P is an additional parameter (between 0 and 1) that controls the effect of hydrogen on
cohesive strength. The spatial hydrogen concentration gradient is considered in crack propagation
direction only. The boundary condition for the hydrogen is such that an environmental
concentration, Cenv, is defined at the moving crack tip; that is, when a cohesive element has failed,
it gets the value of Cenv. The initial condition is that the concentration of the cohesive elements is
zero, whereas Cenv is available at the crack flanks.
In a numerical study the effect of hydrogen diffusion has been studied using the time
dependent cohesive model. A C(T) specimen with W = 50 mm and an initial crack length a0/W =
0.5 has been analysed under different loading rates. The material is elastic-plastic with a yield
strength of 200 MPa and a hardening exponent of n = 15. It is assumed that the yield strength itself
does not depend on the hydrogen concentration, and the cohesive strength, which has an initial
value of V 0 = 560 MPa, is affected by hydrogen according Eq. (2) and using P Cenv = 0.9. -The
994 I. Scheider et al.

force load-line displacement curves for various loading rates are shown in Fig. 1. Of course, the
residual strength decreases with loading speed as expected, since the hydrogen needs time to
diffuse into the material. An experimental validation with a steel, which is already tested under
various loading speeds, is to be performed in the near future.

FIGURE 1: Numerical simulation of fracture tests in a corrosive environment under various

loading speeds.

References
1. Needleman, A., J. Appl. Mech., vol. 54, 525-531, 1987
2. Serebrinsky, S., Carter, E.A. and Ortiz, M., J. Mech. Phys. Solids, vol. 52, 2403-2430, 2004
3. Liang, Y. and Sofronis, P., J. Mech. Phys. Solids, vol. 51, 1509-1531, 2003
4. Scheider, I., In First M.I.T. Conf. on Comp. Fluid and Solid Mech., edited by K.J. Bathe,
Elsevier, Amsterdam, 2001, 460-462
 16. Environment Assisted Fracture 995

ENVIRONMENTAL STRESS CRACKING OF POLYETHYLENE PIPES IN

WATER DISTRIBUTION NETWORKS

J. P. Dear, N. S. Mason and M. Poulton

Imperial College London
South Kensington Campus, London, SW7 2AZ
j.dear@imperial.ac.uk

Water distribution networks are subject to both demanding operational and environmental stresses
and these vary, for example, with the proximity of pipes to booster pumps, operating valves and
other locations where the pipes are subjected to surge and other fluctuating stresses. Polyethylene
water pipes are often used as they have good damage resistance to these stresses. A process that
can shorten the life of polymer pipes is the removal of antioxidant from the material by chlorine
present in the water. The result can be to reduce the life of the pipe by early exposure of weak
features that can cause the pipe to fail. In assessing the average life of polymer water pipes, one
need is failure data related to the loss of antioxidants for different concentrations of chlorine in the
water. This study of the effects of chlorine in water pipes is part of researching the types and
distribution of failures of a typical town’s water distribution network.
For this study, laboratory experiments were performed using new polymer pipes subjected to
accelerated life assessments. Pipes were subjected to different concentrations of chlorinated water
and maintained at different pressures and for different elevated temperatures. Fig. 1(a) shows a plot
of log10(hoop stress) versus log10(time to failure) for MDPE pipes which failed when internally
pressurized with chlorinated water at 80oC for concentrations of 500 , 1000 , 3000 ,
5000 , 10000 , 15000 , 30000 , 45000 and 120000 mg litre-1. A
reference line is shown for Ductile and Brittle failures at 80oC when the pressurized water inside
the pipe is not chlorinated. Fig 1(b) shows multiple cracking of the inner surface of a failed MDPE
pipe exposed to chlorinated water.

FIGURE 1. MDPE pipe failure due to pressurized chlorinated water at 80oC: (a) Log10(hoop
stress) versus log10(time to failure); (b) Axial crack formation on inner surface.

Fig. 2 shows plots of antioxidant (OIT) profiles for failures in MDPE pipes (hoop stress of 4.6
MN m-2) versus distance from inner wall for a chlorine concentration of 15000 mg litre-1 and time
to failure of 552 hours (Fig. 2(a)) and for a chlorine concentration of 5000 mg litre-1 and time to
failure of 1176 hours (Fig. 2(b)). These are compared with concentration profiles from a model
simulation (Figs 2(c) & (d)) obtained by numerical solution of coupled differential equations for
antioxidant and chlorine diffusion/reaction. This modelling gives the normalised antioxidant
996 J. P. Dear et al.

concentration (CA/C0,A) and the normalised chlorine concentration (CCl/C0,Cl) as are shown across
the thickness of the pipe wall (inner pipe radius = a and outer pipe radius = b). The dotted lines in
Figs 2(a) and (b) are the initial antioxidant level prior to chlorine exposure.
The effect of chlorine is: (1) Chlorine is absorbed in the polymer, diffusing and consuming
antioxidant, (2) Oxidation of the polymer chains occurs leading to a reduction of molecular mass
through chain scission, (3) Reduction in tensile properties of the polymer lead to a brittle layer
forming and as the pipe expands due to creep, the brittle layer cracks and so provides an initiation
site for slow crack growth.

FIGURE 2. Antioxidant profile through pipe thickness: (a) and (b) experimental data from failed
MDPE pipes in chlorinated water at 80oC; (c) and (d) modelling data.

References
1. Dear, J.P. and Mason, N.S., Polymers and Polymer Composites, vol. 9, 1-13, 2001.
2. Smith G.D., Karlsson, K. and Gedde, U.W., Polymer Eng. and Science, vol. 32, 658-667,
1992.
 16. Environment Assisted Fracture 997

FATIGUE CRACK PROPAGATION IN 2XXX ALUMINIUM ALLOYS AT 223K

C. Gasqueres1, C. Sarrazin-Baudoux1, D. Dumont2 and J. Petit1

1LMPM/ENSMA, UMR CNRS 6617, Chasseneuil-Futuroscope, France
2PECHINEY/ALCAN, Centre de Voreppe, France

gasquere@graal.ensma.fr, baudoux@lmpm.ensma.fr, David.DUMONT@pechiney.com,

jean.petit@lmpm.ensma.fr

Aluminum alloys are widely used for constitutive parts of aircrafts and are consequently
confronted with a wide range of temperature depending on altitude: from 300K (ambient
temperature) on the ground down to 223K during a fly. Fatigue properties and particularly fatigue
crack growth resistance at low temperature have been poorly studied. The literature provides
results in cryogenic condition (77K) showing improved fatigue strength in Al-Li alloys [1,2]. This
paper deals with a study of the fatigue crack propagation behavior of two high strength 2xxx
Aluminum alloys elaborated in naturally-aged and peak-aged conditions.
Tests were performed at 223K in an environmental chamber with a controlled dry atmosphere
(dew point of 223K). Cooling of the specimen was performed by mean of two cooling blocks fixed
on both sides of the specimen near the back face and own to the circulation of cooled silicon oil.
Fatigue crack propagation tests were performed with an Instron servohydraulic testing machine
under load control. The loading signal was a sinusoidal waveform one with a frequency of 35 Hz
and a load ratio (R) of 0.1. The specimen used were of C.T. type (width W=50mm) with LT
orientation.
Each test began with a crack propagation in ambient air at K=12 MPa.m1/2. Then threshold
tests were performed in cold environment using a load shedding technique. Finally, K was
increased at constant load amplitude up to near-failure domain. Crack closure was detected using
the compliance method, and the crack was optically tracked.
The fatigue crack propagation mechanism in the naturally-aged alloys was shown strongly
modified by the cold environment with an abrupt transition from a stage II propagation regime at
room temperature to a highly retarded crystallographic stage-I like propagation regime at 223K.
An illustration of this transition is given in figure 1. The main consequence is an improvement in
fatigue crack growth resistance in cold environment as it can be seen in figure 2 for AA2024A
T351 tested in air at room temperature and at 223K. But no change was observed on the peak-aged
alloys and the fatigue crack propagation curves obtained in the cold environment are similar to
those in ambient air.
The propagation data after closure and Young modulus correction are analyzed in terms of a
previous modelling established for fatigue crack propagation at room temperature [3]. The
respective influence of microstructure, temperature and air dryness is discussed on the basis of
scanning and transmission microscopy observations.
998 C. Gasqueres et al.

FIGURE 1. Change in the fracture surface morphology of AA2024A T351 alloy induced by 223K
dry environment.

FIGURE 2. AA2024A T351 fatigue crack propagation curves in ambient air and 223K dry
environment.

References
1. Park, K.J. and C.S. Lee, Scripta Materialia, vol. 34(2), 215-220, 1996
2. Xu, Y.B., et al., Scripta Materialia, vol 33(2), 179-183, 1995
3. Petit, J., G. Henaff, and C. Sarrazin-Baudoux, Fatigue crack growth threshold, endurance
limits and design, Newman and Piascik, , American Society for testing and materials, West
Conshohocken, 4-30, 2000
 16. Environment Assisted Fracture 999

HYDROGEN ASSISTED CRACKING PATHS IN ORIENTED PEARLITIC

MICROSTRUCTURES

J. Toribio and E. Ovejero

Department of Materials Engineering, University of Salamanca
E.P.S., Campus Viriato, Avda. Requejo 33, 49022 Zamora, Spain
Tel: (34-980) 54 50 00; Fax: (34-980) 54 50 02, toribio@usal.es

Previous works [1-4] demonstrated that progressive cold drawing in eutectoid steels affects the
microstructure of the material at the two basic microstructural levels. Firstly [1,3], it produces a
preferential orientation and slenderising of the pearlitic colonies, so that they tend to become more
slender and to align in a direction quasi-parallel to the drawing direction. In addition [2,4], cold
drawing produces a preferential orientation and a densification of the pearlite lamellae, so that the
latter tend to align in a direction parallel to the wire axis, with a reduction at the same time of the
pearlitic interlamellar spacing.
This paper deals with the consequences of the afore-said microstructural evolution on the
posterior behaviour of the steels under hydrogen assisted cracking conditions. The experimental
results show that cold drawing induces strength anisotropy in the steel, and thus the resistance to
hydrogen embrittlement is a directional property that depends on the angle in relation to the
drawing direction. As a consequence, an initial transverse crack changes its propagation direction
to approach that of the wire axis, thus producing mixed mode propagation, the deflection angle (in
relation to the initial crack propagation direction) being an increasing function of the cold drawing
degree. This experimental result may be explained by micro-mechanical considerations on the
basis of the lamellar microstructure of the steels.
Special attention is paid to the hydrogen assisted cracking path as a function of the drawing
degree, analysing the possible fracture of the pearlitic lamellae or the delamination between ferrite
and cementite. While in the hot rolled steel (not cold drawn) the crack advances in a hydrogen
environment by breaking the pearlitic lamellae, in heavily drawn steels the predominant
mechanism of hydrogen assisted cracking is the delamination at the ferrirte/cementite interface (cf.
Fig. 1) or the decohesion between two adjacent pearlitic colonies, thus permitting a
micromechanical modelling as a micro-composite material.

FIGURE 1. Hydrogen assisted cracking path in heavily drawn steel.

1000 J. Toribio and E. Ovejero

References
1. Toribio, J. and Ovejero, E., Mater. Sci. Eng, vol. A234-236, 579, 1997.
2. Toribio, J. and Ovejero, E., Scripta Mater., vol. 39, 323, 1998.
3. Toribio, J. and Ovejero, E., J. Mater. Sci. Lett., vol. 17, 1037, 1998.
4. Toribio, J. and Ovejero, E., Mech. Time-Dependent Mater., vol. 1, 307, 1998.
 16. Environment Assisted Fracture 1001

EFFECT OF RESIDUAL STRESS-STRAIN PROFILE ON HYDROGEN

EMBRITTLEMENT SUSCEPTIBILITY OF PRESTRESSING STEEL WIRES

J. Toribio and V. Kharin

Department of Materials Engineering, University of Salamanca
E.P.S., Campus Viriato, Avda. Requejo 33, 49022 Zamora, Spain
Tel: (34-980) 54 50 00, Fax: (34-980) 54 50 02, toribio@usal.es

Cold drawn wires of eutectoid pearlitic steel are widely used for prestressing concrete structures
which usually work in hostile or aggressive environments, so that stress corrosion cracking of
prestressing steel is a problem of major technological concern. In addition, there is general
agreement that hydrogen embrittlement (HE) plays an important role in the environmental
cracking of such a steel due to particular working conditions or to the local electrochemistry in the
vicinity of a crack tip. In this framework, the Standard Test in Ammonium Thiocyanate was
proposed by the International Federation of Prestressing (FIP) as a suitable experimental method
for checking the susceptibility of high-strength prestressing steels to stress corrosion cracking in
general, and particularly to hydrogen embrittlement. In spite of some objections to this standard
corrosion test, it is the best suited for steel control and acceptance. However, the main
disadvantage of the FIP test is the scattering of the results, which increases as the externally
applied stress decreases. It can be caused by the distribution of residual stresses generated in the
vicinity of the wire surface during the manufacturing (cold drawing) process.
Previous research [1] established an important milestone by obtaining a quantitative
relationship between the level of residual stress (represented by theoretical residual stress laws)
and the fracture behaviour of prestressing steel wires under HE conditions, but a detailed analysis
of the influence of real residual stress profiles caused by different regimes of surface rolling on the
HE susceptibility of cold drawn prestressing steel wires has not been performed yet. This paper
goes further in the analysis, so that the earlier developed computer model [1] is advanced and
applied to analyze the influence of the residual stress-and-strain profiles after surface rolling on
the hydrogen embrittlement susceptibility of cold drawn prestressing steel wires in FIP tests. To
this end, a computer model was used to predict the wire life in the aggressive solution. The model
is based on hydrogen transport by diffusion assisted by stress and strain [2-5]. It allows a detailed
analysis of the influence on the wire life of the specific characteristics of the residual stress and
strain distributions in the wires, and particularly of the residual stress at the surface, the profile of
the residual stress law and the depth of the maximum hydrostatic stress point.
Profiles of hydrostatic residual stresses and plastic strains are presented in Fig. 1. It is seen that
a rolling process introduces severe inhomogeneity of residual stresses (samples C8 and C26) in
contrast to relatively uniform distributions after drawing only (sample C0). In addition, Fig. 2
shows an excellent agreement between the theoretical predictions of the models and real
experimental data obtained by means of the Ammonium Thiocyanate FIP test.
1002 J. Toribio and V. Kharin

FIGURE 1. Residual hydrostatic stress and plastic strain profiles in cold drawn wires.

FIGURE 2. Theoretical predictions of prestressing steel lifes and experimental data.

References
1. Toribio, J and Elices, M., Int. J. Solids Structures, vol. 28, 791-803, 1991.
2. Toribio, J. and Kharin, V., Fatigue Fract. Engng. Mater. Structures, vol. 20, 729-745, 1997.
3. Toribio, J. and Kharin, V., Nuclear Engng. Design, vol. 182, 149-163, 1998.
4. Toribio, J. and Kharin, V., Int. J. Fracture, vol. 88, 233-245, 1998.
5. Toribio, J. and Kharin, V., Int. J. Fracture, vol. 88, 247-258, 1998.
 16. Environment Assisted Fracture 1003

HYDROGEN EMBRITTLEMENT OF AUSTENITIC STAINLESS STEELS AT

LOW TEMPERATURES

L. Zhang, M. Wen, M. Imade, S. Fukuyama and K. Yokogawa

National Institute of Advanced Industrial Science and Technology
Central 5, 1-1-1 East, Tsukuba, 305-0046, JAPAN
yokogawa.kiyoshi@aist.go.jp

Stainless steels suffer from hydrogen embrittlement (HE) at low temperatures, thus the
improvement of the resistance to HE has been expected by alloying. The main elements of Ni and
Cr improve resistance to HE by stabilizing austenite phase with respect to martensitic
transformations and by increasing the stacking fault energy (SFE) to promote cross-slip. The
addition of N also stabilizes austenite phase and improves resistance to HE. Although N does not
change the SFE, N promotes planar-slip by inducing a short range order in the matrix [1,2]. Thus,
N and Ni play distinct roles in improving resistance to HE. It provides a way to distinguish HE due
to strain-induced alpha’ martensite from HE due to planar-slip.
The degree of austenite phase stability against strain-induced alpha’ martensitic transformation
and the tendency of forming austenite phase can be evaluated by an equation of Ni equivalent that
is a modified form of Hirayama’s equation [2]:
Nieq = Ni + 0.65Cr + 0.98Mo + 1.05Mn + 0.35Si + 12.6(C+N), (1)
where all elements are in weight fraction. Thus, HE can be correlated with Nieq. However, the
effect of Nieq on HE has not been studied until now.
In the present study, Ni content of the austenitic stainless steels of Fe(10-20)Ni17Cr2Mo alloy
and N content of those of Fe11Ni17Cr2Mo(0.001-0.25)N alloy, both based on type 316 stainless
steel, were changed to study the effect of Nieq on hydrogen environment embrittlement (HEE).
Tensile tests were conducted in hydrogen and helium at 1 MPa in the temperature range from 80 to
300 K to clarify the relation of HEE with Nieq. All tests were conducted with a strain rate of
4.2×10-5 /s. Alpha’ martensite content in austenite phase was measured as ferrite equivalent by the
magnetic method. The fracture surface of the specimen was analyzed by scanning electron
microscopy (SEM) after tensile tests. Martensitic transformation was analysed by transmission
electron microscopy (TEM).
HEE can be quantitatively described by the relative reduction of area (RRA) measured in
hydrogen at 1 MPa relative to in helium at 1 MPa (reduction of area in hydrogen / reduction of area
in helium). There is no HEE if RRA equals 1, and HEE increases with decreasing RRA. It was
observed that the RRA decreases with decreasing temperature, reaches a minimum at around 200
K and then increases with further decrease in the temperature. The most serioue HEE occured at
200 K, thus the effects of Nieq on ferrite equivalent and the RRA at 200 K are shown in Fig. 1. It is
clear that Nieq correctly predicts the austenite phase stability due to the same relation bewteen
ferrite equivalent and Nieq of both N and Ni added steels. Both N and Ni can stabilize austenite
phase by decreasing ferrite equivalent. However, these two kinds of steels have quite different
RRAs. In the case that Nieq is above 27 %, no HEE of Ni added steels is observed, while HEE of
N added steels is always shown although it decreases with increasing N.
1004 L. Zhang et al.

FIGURE 1. Effects of Nieq on alpha’ martensite content and the RRA at 200 K.

Fracture surfaces in hydrogen of Ni added steels revealed that transgranular fracture occurs
along alpha’ martensite laths and those of N added steels showed transgranular fracture togehter
with slip plane fracture. Hydrogen induces slip planarity in fcc structure, thus the difference in
RRA of N and Ni added steels is obviously due to planar-slip caused by N-induced a short range
order. TEM observations indicated that strain-induced alpha’ martensite is formed at the
intersections of the microscopic slip bands in the grains of both steels. The bcc structure is
believed to be inherently more susceptible to hydrogen-induced cracking than the fcc structure and
the bcc structure provides an effective short-circuit diffusion path for hydrogen due to the high
diffusivity of hydrogen in it [3], thus alpha’ martensite increases hydrogen content in the matrix
and promotes HEE. It is considered that strain-induced alpha’ martensite plays a key important
role in hydrogen-induced crack propagation in these steels.
It is thus concluded that in Ni added 316 series stainless steels HEE is completely controlled by
alpha’ martensite content. In N added stainless steels, strain-induced alpha’ martensite also plays
the primary role in HEE, but slip pnanarity plays more important role in inducing HEE when N is
increased.

References
1. Swann, P. R., Corrosion, vol. 19, 102, 1963.
2. V. Gerold, H.P. Karnthaler, Acta Metall. vol. 37, 2177, 1989.
3. Hirayama, T. and Ogirima, M., J. Jpn. Inst. Met., vol. 34, 507-510, 1970.
4. Han G., He J., Fukuyama S. and Yokogawa K., Acta Metall., vol. 46, 4559-4570, 1998.
 16. Environment Assisted Fracture 1005

HYDROGEN DIFFUSION AND EAC OF PIPELINE STEELS UNDER

CATHODIC PROTECTION

M. Cabrini and T. Pastore

Università degli studi di Bergamo – Dipartimento di Progettazione e Tecnologie
Viale Marconi, 5 24044 Dalmine (BG) Italy
marina.cabrini@unibg.it, tommaso.pastore@unibg.it

Pipeline steels can show Environmental Induced Cracking phenomena under slow straining with
Hydrogen Embrittlement mechanism under cathodic protection. Hydrogen evolution can take
place due to cathodic protection normally used in order to protect the pipeline against general
corrosion. The steel is polarized at cathodic potentials in the range –0.8 to –1.1 V vs SCE, but very
negative values could be reached in overprotected areas close to the impressed current anodes. The
hydrogen ions reduction reaction takes place on the metal surface, forming adsorbed hydrogen.
The adsorbed hydrogen diffuses into the metal owing its solubility in the metal lattice. Afterwards
it can concentrate in plastic deformation areas, and promote brittle growth. Figure 1 shows the
results of slow strain rate tests on a low alloyed steel. The loading curves obtained at –1050mV vs
SCE is very different from the curve obtained in air. The differences are related to the presence of
brittle areas on fracture surphace and secondary cracks (figure 1).

FIGURE 1. Stress vs strain curves obtained at 10–6 s–1 and different potentials and fracture surface
of the SSR specimen after the test at –1050 mV vs SCE.

Similar effect are also evident in Corrosion-Fatigue at low frequency: figure 2 shows da/dN vs
'K curves in air and in cathodic protection at –1050 and –900 mV vs SCE. The cathodic protection
increases the fatigue cracks growth for intermediate values of 'K. In this range the crack
propagation is dominated from hydrogen embrittlement. The EAC phenomena can take place at
the cracks tip, after the hydrogen concentration reachs a critical value. The hydrogen diffusivity
depends on microstructure of the steel. Table 1 summarizes the hydrogen diffusion coefficient of
pipeline steels. The normalised reduction of area after SSR tests and the average fatigue cracks rate
at 'K 20MPa—m at the same potential are also reported.
1006 M. Cabrini and T. Pastore

FIGURE 2. a) da/dN vs 'K curves in air and in artificial sea water at different potentials, b)
fracture surface of the specimens at the end of the CF tests in sea water at –1050 mV vs SCE

TABLE 1. Results of the hydrogen diffusion tests (transversal direction), SSR tests and CF tests at
E = –1050 mV vs SCE

All steels show at –1050 mV vs SCE hydrogen embrittlement under slow straining. The steels
with micrsotructures of bands of ferrite/pearlite have hydrogen diffusion coefficient higher than
the steels with babded microstructures of fine ferrite with martensite or bainite and of the steels
with temepred martensite. An increase of the hydrogen embrittlement susceptibility is observed as
the hydrogen diffusion coefficient increases.

References
1. Cabrini, M., Cogliati O., Maffi S., La metallurgia italiana, vol. 3, 13-20, 2003
2. Zucchi F., Grassi V., Frignani A., Monticelli C., Trabanelli G., In Atti del Convegno
Nazionale AIM, edited by AIM, Vicenza Novembre 17-19, 2004, CD-room .
 16. Environment Assisted Fracture 1007

INITIATION OF ENVIRONMENTALLY ASSISTED CRACKING IN LINE PIPE

STEEL

M. Elboujdaini
CANMET Materials Technology Laboratory - Natural Resources Canada,
568 Booth St., Ottawa, Ontario, Canada K1A OG1
melboujd@NRCan.gc.ca

Stress Corrosion Cracking Mechanism: Stress corrosion cracking (SCC) has been observed on the
soil side of buried, natural gas pipelines since the early 1960s. Transgranular SCC has caused
service as well as hydrotest failures, and cracks have been found associated with gouges and dents.
Transgranular cracking occurs in environments with pH about 6.5, and is referred to as near-
neutral pH cracking, as opposed to high pH cracking, which is intergranular in nature.
The stress corrosion cracking (SCC) results from multiple metallurgical, mechanical, and
environmental factors. Chemical composition of the steel, residual stress in the steel as well as
applied stress, water chemistry in the field, including CO2, oxygen, and ionic concentrations in the
groundwater near the pipe surface, may all have an effect on crack initiation and propagation [1,2].
Stress corrosion cracking in pipelines involves several steps: (i) The coating applied to the
pipeline during installation becomes degraded, an electrolyte comes into contact with the surface,
and the environment that causes SCC to develop; (ii) The initiation and growth of multiple cracks
that form colonies; (iii) These cracks may continue to grow and coalesce; and (iv) In the final step,
a dominant crack reaches a critical size for rapid growth to failure, producing either a leak or a
rupture. The time to failure depends on a number of factors, including the pipe material, stress
history, environment, and crack distribution.
Nearly all studies of SCC have been carried out without distinguishing the characteristics of
initiation from those of propagation. Many of the studies on propagation have focused on growth
of long cracks in pre-cracked specimens. Initiation of SCC is, however, studied using specimens
that are not pre-cracked. The definition of an initiated crack is not well defined, and there is no
clear mechanistic interpretation of the events that lead to initiation.
SCC has been observed to initiate from the base of localized corrosion sites (i.e. pits, crevices)
for a variety of metal-environment combinations [1,3]. There remains debate on whether the stress
intensification at the base of the pit or the enhanced electrochemical conditions within the pit is the
controlling factor in SCC [2,4]. Several investigators believe it is the localized environment that
plays the biggest role in crack initiation and not necessarily the stress concentration provided by
the localized corrosion site, whereas others believe that crack initiation in smooth samples requires
the presence of a stress raiser [3,4]. The most common way to establish such a stress raiser is
either through corrosion or mechanical damage.
Complexities in SCC Phenomena: SCC in steels for oil and gas pipelines is a very complex
and challenging phenomenon. The complexity of SCC is reflected in the changes, with time, of the
diverse parameters influencing the cracking phenomena, whereas the biggest challenge is in
obtaining field-relevant reproducible laboratory data. SCC encompasses major effects from
metallurgical, mechanical, and environmental parameters, all of which can be dominant under
specific conditions. Adding to the complexity are the loading conditions in operating pipelines
that define the mode of failure as SCC or corrosion fatigue (CF). While SCC and CF are
sometimes regarded as different modes of failure, the distinctions between them in mechanistic or
engineering terms are becoming less sharply defined [4].
1008 M. Elboujdaini

Figure 1: EDS spectra indicating chemical composition of three different inclusions as associated
with the initiation of pits in X-65 steel sample (before and after test at Fmax=90%YS in NS-4
solution saturated with N2/5%CO2). Note the rapid dissolution of CaS inclusions and the
formation of corrosion pits around the inclusions.

References
1. National Energy Board, Calgary, Alberta, Stress Corrosion Cracking on Canadian Oil and
Gas Pipelines, Report No. MH-2-95, 1996.
2. Parkins, R.N., et al., Corrosion, 50, 394-408, 1994.
3. Elboujdaini, M. et al,Corrosion/2000, Paper 00379, NACE, Houston, TX, 2000.
4. Parkins, R.N., Metals Science and Engineering, A103, 143, 1988.
 16. Environment Assisted Fracture 1009

FATIGUE CRACK GROWTH BEHAVIOUR DEPENDING ON ENVIRONMENT

IN MAGNESIUM ALLOYS

Masaki Nakajima1, Keiro Tokaji2, Yoshihiko Uematsu2 and Toshihiro Shimizu1

1Dept. of Mech. Eng., Toyota National College of Technology,
2-1 Eisei-cho, Toyota 471-8525, Japan
nakajima@toyota-ct.ac.jp
2Dept. of Mech. and Systems Eng., Gifu Univ., 1-1 Yanagido, Gifu 501-1193, Japan

Recently, magnesium (Mg) alloys have become of major interest as a light-weight structural
material because of their excellent specific strength. In extensive applications for structural
components, fatigue properties such as fatigue strength and fatigue crack propagation (FCP) are
critical, but studies on FCP behaviour are very limited. Kamakura et al. [1] have indicated that
FCP behaviour in wrought Mg alloys was influenced significantly by humidity in laboratory air.
This is due to very high electro-chemical activation resulting from poor corrosion resistance
(Eliezer et al. [2]). Therefore, it is necessary to further study the FCP behaviour and mechanisms in
corrosive environments. In the present paper, FCP tests have been performed using CT specimens
of AZ31 and AZ61 in dry air, in laboratory air and in distilled water, and the effect of environment
and mechanisms were discussed.
The materials used are rolled AZ31 and extruded AZ61 plates from which CT specimens with
L-T orientation were machined. FCP experiments were conducted at a stress ratio, R, of 0.05 in dry
air, in laboratory air and in distilled water using electro-hydraulic fatigue testing machine
operating at a frequency of 1 Hz. The dew point of dry air was –60qC. After experiment, fracture
surfaces were examined using a scanning electron microscope (SEM).

FIGURE 1. Relationships between FCP rate and stress intensity factor range: a) AZ31, (b) AZ61.

The relationships between FCP rate, da/dN, and stress intensity factor ranges, 'K, 'Keff in
AZ31 and AZ61 are shown in Fig.1(a) and (b), respectively. It can be seen in Fig.1 that both alloys
exhibits basically the similar FCP behaviour regardless of environment. FCP rates are nearly the
same in laboratory air and in distilled water, while approximately an order of magnitude slower in
dry air than in those environments. This indicates that humidity in laboratory air exerts a
significant influence on FCP behaviour, which acts in the same manner as distilled water. Such
1010 M. Nakajima et al.

tendency of FCP behaviour is maintained after allowing for crack closure, indicating the intrinsic
effect of environment on FCP behaviour.
Microscopic appearance of fracture surfaces was similar in laboratory air and in distilled water,
with many steps formed in each grain in high 'K region and extensive quasi-cleavage in low 'K
region. In dry air, such appearances were not recognized and fracture surfaces were ductile in the
entire 'K region.

FIGURE 2. Effect of cyclic frequency on FCP rate: (a) AZ31, (b) AZ61.

In order to further understand the effect of environment, constant K tests were performed at
K=3.5MPam1/2 in a wide range of frequencies of 0.01 Hz to 10 Hz. The obtained results are
represented in Fig.2. As described previously, the FCP rates in dry air are an order of magnitude
slower than those in laboratory air and in distilled water. It should be noted that the effect of cyclic
frequency is less remarkable in laboratory air and in dry air, while the FCP rates in distilled water
become faster with decreasing frequency, particularly remarkable in the low frequency region.
From the observation of fracture surfaces, corrosion products were not seen in laboratory air, while
extensively observed in distilled water. Therefore, it is believed that FCP rates would be enhanced
by a mechanism such as hydrogen embrittlement in laboratory air and a further enhancement of
FCP rate would result from anodic dissolution in distilled water.

References
1. Kamakura, M., Tokaji, K., Ishiizumi, Y. and Hasegawa, N., J. Soc. Mat. Sci., Japan, 53,
1371-1377, 2004.
2. Eliezer, A., Gutman, E.M., Abramov, E. and Aghion, E., Corrosion Review, 16, 1-26, 1998.
 16. Environment Assisted Fracture 1011

ASSESSMENT OF HIGH-TEMPERATURE HYDROGEN DEGRADATION OF

POWER EQUIPMENT STEELS

H. M. Nykyforchyn and O. Z. Student

Karpenko Physico-Mechanical Institute of National Academy of Sciences of Ukraine
5 Naukova Street, Lviv, 79601, Ukraine
nykyfor@ipm.lviv.ua, student@ipm.lviv.ua

Long term service of power equipment in corrosion and hydrogenated environments causes a
degradation of the physical, chemical and mechanical properties and correspondingly a decrease
its lifetime.
The method of the accelerated degradation of steels by thermocycling in the gaseous hydrogen
is developed, which allows to realize during some weeks degradation of metal like in service
during 10…20 years. The idea of the method has been based on the temperature dependence of
hydrogen solubility and a decrease of hydrogen mobility at lower temperatures. Student [1]
showed that this method causes changes of steel microstructures, similar to those that observed
after long term exploitation.
A relevancy of an usage of the effective range of stress intensity factor 'Kth eff as parameter
which detects mechanical state of the degraded material, is proved. This finding has been also
supported by relationship between the parameter Kth eff vs. the time of operation Wop and vs. the
number of thermocycles n, cf. Fig. 1.

FIGURE 1. Effect of the operation time W (1) and number of thermocycles in hydrogen n (2) on the
parameter 'Kth eff for the 12H1F (0.1% C; 1.1% Cr; 0.54% Mn; 0.26% Mo; 0.26% Si; 0.17% V;
0.019% S; 0.015% P) steel.

The fractography and surface roughness examination by Student et al. [2] indicated a high
probability of the premature contact of the corresponding fracture surfaces due to longitudinal shear in
the prefracture zone (mode III of crack loading) for hydrogen degraded steel. The obtained results
enabled to prove an existence of the fatigue crack closure mechanism due to mode III deformation and
to develop the well known model of crack closure, caused by fracture surface roughness.
An inversion of an effect of the absorbed by metal hydrogen on the 'Kth eff parameter in
dependence of metal state caused by ageing is shown (Fig. 2). The higher Kth eff for the hydrogen
charged metal in as received state (positive effect of hydrogen, which is typical for comparatively
1012 H. M. Nykyforchyn and O. Z. Student

low strength steels with high ductility, showed by Nykyforchyn [3]) changed to negative effect of
hydrogen on this parameter after long term exploitation. Fractography investigation of fracture
surfaces of the 12H1MF steel showed that the negative effect of hydrogen on the Kth eff level is
accounted by the susceptibility of steel to hydrogen cracking due to the material degradation. The
evidence of such hydrogen degradation is the fatigue striations situated perpendicular to the main
crack direction and decorated by secondary microcracks. On this base the criterion of a definition
of the limited state of metal as a result of its high temperature degradation is formulated: it is
achieved when hydrogen starts to influence negatively on the 'Kth eff parameter and material
becomes sensitive to hydrogen induced cracking.

FIGURE 2. Dependence 'Kth eff on- op for hydrogenated (1) and outgassed (2) 12H1MF steel.

An approach to the residual life time evaluation of the power plant steam pipe lines is
developed, which allows to take into account the negative hydrogen effect. It is based on the
criterion of a definition of the limited state of metal and the correlation between a time of service
and a number of cycles during the in-laboratory thermocycling in hydrogen.

References
1. Student, O. Z. Materials Science, V.34, N4, 490-496,1998.
2. Student, O.Z., Cichosz, P. and Szymkowski, J. Materials Science, N6, 1999.
3. Nykyforchyn, H.M. Materials Science, N4, 2002.
 16. Environment Assisted Fracture 1013

STRESS CORROSION CRACKING OF 18MN-4CR GENERATOR ROTOR

END-RETAINING RING STEEL

N. Mukhopadhyay and U. K. Chatterjee

General Manager, BHEL, Hyderabad, and Professor, IIT Kharagpur
Corporate R&D Division, BHEL, Hyderabad 500593, India, and
Department of Metallurgical & Materials Engineering, IIT Kharagpur 721302, India
nmuk@rnd.bhel.co.in, kc@metal.iitkgp.ernet.in

18Mn-4Cr austenitic steel has virtually been the standard material for the rotor end-retaining rings
in power generation plants during the period of 1960-1980, before it was replaced by the 18Mn-
18Cr variety. A good number of cases of failure of the 18Mn-4Cr steel ring by stress corrosion
cracking (SCC) led to the switchover to the higher chromium variety. The cases of failure have
been documented in the report of EPRI Workshop on Retaining Rings for Electrical Generators
(October 1982). SCC has been attributed to the presence of moisture, and though the sources of
humidity have been identified as imperfectly dried hydrogen, leakages from water-fed stator
winding, oil in the hydrogen-tightness circuit and moisture condensation during shut downs,
storage and transportation of the rings, no light was thrown on the mechanism of SCC or any
possible effect of the stress relieving temperature on SCC. The present study was undertaken with
these considerations in view.
A right circular cylindrical ring forging was used in the as-received and thermo-mechanically
treated conditions. The chemical composition of the steel was:0.52 C, 0.01 S, 0.02 P, 18.98 Mn,
3.87 Cr, o.035 Al. 0.05 V, 0.15 Ni, 1.04 Si, balance Fe. The thermo-mechanical treatment route
was in line with the fabrication and heat treatment method generally followed for the end-retaining
rings viz. solution annealing at 10400 C, water quenching, cold working to 15% reduction in area,
stress relieving at different temperatures, followed by furnace cooling. The temperatures for stress
relieving were chosen as 3500 C, 4000C, 4500C and 5000C.
Tensile specimens of 50 mm gage length and 3 mm diameter were prepared from both as-
received and thermo-mechanically treated steel. Stress corrosion tests were performed by constant
load technique using a Mayes constant load SCC device in a 50 ppm chloride solution. The tests
were carried out under free corroding as well as under polarization conditions using a Wenkin
potentiostat. Supplementary electrochemical tests were performed to study the localized corrosion
(pitting) behaviour under potentiostatic conditions.
Stress corrosion cracking was encountered in all the specimens. The as-received material has
shown predominantly intergranular cracking. The time to fracture has increased in specimens
stress relieved at 3500 C (ca. 540 hr), 4000 C (ca. 430 hr) and 4500 C (ca. 300 hr), compared to that
in the as-received specimens (ca. 150 hr), and the cracking mode is transgranular. The change in
SCC susceptibility as well as the mode of cracking can be directly correlated to the extent of
carbide precipitation. In the case of as-received specimens and specimens stress relieved at 5000 C,
the precipitated carbides are massive and are aligned mostly along the grain boundaries.
Potentiostatic etch tests were carried out by holding the sample potentiostatically at a series of
potentials ( -370 mV to –720 mV, OCP being –590 mV SCE), for 2 hr for localized attack to
initiate. Subsequent microscopic examination revealed localized attack at the sites of carbide
precipitation. Even under large cathodic potentials, the attack was observed at grain boundary
triple points indicating that the dissolution is chemically controlled. The corrosion product,
observed as mounds at these sites, induces crevice corrosion leading to the formation of grooves as
1014 N. Mukhopadhyay and U. K. Chatterjee

stress corrosion crack nucleation sites. However, in SCC tests, the time to fracture has been found
to be potential dependent, decreasing on anodic polarization and increasing on cathodic
polarization, indicating the electrochemical nature of the cracking process. The soluble carbides
that assist in pit initiation also help in crack propagation by providing an easy path for dissolution
and also by the formation of occluded cells at these sites. The latter process is electrochemical in
nature, and thus cathodic polarization has drastically slowed the process of crack propagation.
Fractographic studies have revealed the evidence of corrosion tunneling and ‘saw-tooth’ type
step formation in the transgranular fracture. Corrosion tunneling is a result of several slip steps
intersecting the surface exposed to a cracking environment [1]. In the present alloy where the
formation of a protective film is not indicated, the soluble carbides precipitated along the slip
planes may be ascribed a role in tunnel formation. The fine ‘saw-tooth’ steps are indicative of
fracture arising out of the joining of a number of microcracks in a single grain [2]. The formation
of this transgranular feature can be attributed to the presence of a large number of carbide particles
in the grain interior where microcracks will initiate through the process of dislocation build-up.
The exclusively intergranular fracture in the specimens stress relieved at 5000 C is an indication of
a shift in the heterogeneities from grain interior to grain boundaries, and the identifiable evidence
of massive carbide precipitation along the grain boundaries supports this.
Apart from the role of carbides and slip steps in providing paths for the ongoing crack front,
strain induced martensite could also play a role in crack propagation. Evidence has been obtained
that the present steel in as-received condition can be transformed to nonequillibrium bcc martensite
by pneumatically peening with hardened steel shots. At the advancing crack tip, the stress
concentration causes local plastic deformation, which is likely to produce martensite that promotes
faster dissolution of the crack tip due to its anodic nature.

References
1. Pickering, H.W. and Swann, P.R., Corrosion, vol. 19, 373-389, 1963
2. Scully, J.C., In The Theory of Stress Corrosion Cracking in Alloys, edited by J.C. Scully,
NATO Scientific Affairs Division, Brussels, 1971. 127-147
 17. SIM, Philosophy, Instrumentation and Analysis 1015

NON CONTACTING STRESS MONITORING

W. D. Dover, R. F. Kare and N. Stone

TSC Inspection Systems
6 Mill Square, Featherstone Road
Wolverton Mill, Milton Keynes, MK12 5RB
wddover@btinternet.com

Key input data for SIM calculations are the service stresses and the defect size. For some
applications only the service stresses are required. For well designed structures this could be for
stresses that arise during construction and early service life. For other situations it may be
necessary to monitor crack/defect size and the local stress.
For the offshore industry the Alternating Current Field Measurement (ACFM) technique, TSC
[1] was developed for subsea and topside inspection. ACFM has the capability to both detect and
size cracks. A recent innovation for ACFM has been the introduction of array probes. These can
collect crack depth information from various sites along the crack in one placement of the probe.
The array probe is an area inspection and hence can also be used for crack size monitoring if left in
place on the structure. An example of array probe monitoring is given in Karé and Tantbirojn [2].
More recently it has become possible to use a non contacting technique for measurement of the
service stress. Recent developments of the alternating current field measurement system have now
made non-contacting stress measurement and monitoring a possibility. It has been known for
some time that mechanical stress can influence the magnetic domain distribution in ferrous metals.
This feature, known as piezo-magnetism can be utilised to determine changes in the state of stress
on the surface of a metal. Magnetic permeability changes of this sort can be measured using AC
field measurement devices. These devices have been improved recently, by UCL and TSC
Inspection Systems Ltd, to the point where very small changes in permeability and hence stress are
detectable Zhou and Dover [3], Chen et al [4], Chen and Brennan [5], Brennan [6]. Thus changes
in stress, of the order of a few percent of the zero to yield stress range, have been measured in the
laboratory for structural steels. The very latest AC field measurement devices, known as the
Stressprobe [1] can now be used for non-contacting induced field stress measurements in service.
The Stressprobe has been used to monitor stress cycling from simple sine waves to broad band
variable amplitude loading typical of that recorded on North Sea platform structures. The
Stressprobe has a higher response for some materials, for example mild steels, and medium
strength steels. With high strength steels the response is smaller and hence the gain needs to be
larger. Despite this it is still possible to monitor the stress fluctuation and Fig. 1 shows a recent
example of monitoring on drill collar material subjected to stress cycling at 3Hz. These tests were
part of an attempt to use the Stressprobe to monitor the decay in residual stress during a fatigue
test. As can be seen in Fig. 1 the Stressprobe is closely following the strain gauge output.
Recent studies have included monitoring the loads in chains used in offshore mooring systems,
offshore risers/tendons, and for pipelines. Some of this work was reported in Dover et al [7] and
this ECF paper will report on more recent studies.
1016 W. D. Dover et al.

Figure 1 Monitoring with strain gauge and the Stressprobe on 4145H Drill collar steel

References
1. TSC Inspection Systems Ltd, 6 Mill Square, Featherstone Road, Wolverton Mill, Milton
Keynes MK12 5RB.
2. Kare, R. F. and Tantbirojn, N., SIMoNET Inaugural Seminar March 2000, UCL.
3. Zhou, J. and Dover, W. D., Journal of Applied Physics, Vol.83, No4, 1998, pp1694-1701.
4. Chen, K., Brennan, F.P., and Dover, W.D NDT&E International, 33, 2000, pp 317-323.
5. Chen, K. and Brennan, F.P., Journal of Strain Analysis, Vol 33, No 4, pp 291-303, 1998.
6. Brennan, F.P., Proceedings of the First International Conference on Non-Destructive
Evaluation in the Gulf, Doha, State of Qatar, November 1999, pp1-15.
7. W D Dover, F P Brennan and B de Leeuw ACSM Stressprobe; a new non contacting stress
measurement technique for the offshore industry
 17. SIM, Philosophy, Instrumentation and Analysis 1017

RAPID CALCULATION OF STRESS INTENSITY FACTORS

A. J. Love and F. P. Brennan

NDE Centre, Department of Mechanical Engineering, University College London,
Torrington Place London WC1E 7JE, UK
f.brennan@ucl.ac.uk

The demand to design evermore efficient, economic and safer structures continues and is set to
only increase throughout the twenty-first century and beyond. Thus, the challenges confronting the
engineer, concerned with ensuring a state of structural integrity prevails similarly grow evermore
demanding. The SIF is widely recognised as the fundamental parameter vital for the assessment of
defects, or cracks prone to linear elastic fracture behaviour. Difficulties in computing or
measuring SIF are widely accepted especially when the crack is situated in a complex geometry or
subject to a non-simple stress state. In addition, with the emergence of structural integrity
monitoring systems there is likely to be an increased demand for the rapid availability of accurate
SIF solutions for on-line defect assessment. This paper describes the development of a novel
weight function methodology, which will potentially permit the determination of SIF solutions for
such crack systems.
Recent years have seen the advent of readily available and greatly enhanced computer
processing ability leading to exciting developments in finite element and boundary element
approaches to complex fracture problems. These require considerable skill and insight from the
operator and are, therefore, likely to remain the preserve of specialists. Moreover, implementation
by a non-specialist could yield dangerously erroneous solutions. If realised, the weight function
methodology described below would present rapid, high quality solutions within the reach of
design engineers in a format conducive for incorporation into standards and design codes.
A general need exists in both industry and academia for SIF solutions for engineering
components for which currently published solutions represent an idealised and often unrealistic
approximation. Further to the limitations stated previously, solutions gained through numerical
techniques are developed for specific applications and are generally applicable within restrictive
limits of validity. Engineering optimisation and defect assessment of components in service
however, often require broad ranging solutions, which can be rapidly calculated. A weight
function approach meets these requirements and, while a specific application is cited in this paper
and solutions tailored accordingly, the methodologies developed will be of significant relevance to
the wider industrial and academic spheres.
The SIF weight function, ‘h(a,x)’, as defined by Rice and Bueckner, is widely recognised as a
powerful and efficient means of determining SIF solutions for cracks subject to complex loading
configurations. In essence it permits the influence of component geometry, which it represents
upon the SIF to be separated from that of applied loading. Once determined, it may be used in
conjunction with a crack-line stress, ‘V(x)’, arising from any loading mode to yield new SIF
solutions as shown below.
a
K ³ V ( x ) h ( a , x ) dx
0 (1)
By virtue of the definition of the weight function as a unique property of geometry, the
influence it represents can be isolated and combined. This characteristic is unique to the weight
function. A number of researchers developed similar composition approaches after Impellizzeri
1018 A. J. Love and F. P. Brennan

and Rich used what they called Geometry Correction Factors for the influence of geometric
anomalies on the weight function. These studies represented the beginning of the idea that
geometric influences could be analytically separated. Recognition of this and advancements in
weight function formulation led to the development of many subsequent engineering solutions by
the building of complex geometry weight functions from more simple constitutive geometry
weight functions.
Brennan and Teh, who envisaged the ‘library’ of solutions for numerous notch types,
conducted the most comprehensive study on the composition of weight functions. The ‘library’
refers to a generic set of constitutive solutions that systematically define a wide range of symmetric
notch parameters.
The weight function composition principle applied to symmetric notches proved the premise
that geometric influences, described as weight functions, may be combined, or composed, via a
suitable composition scheme to yield new SIF solutions. Extension of the principle to broaden the
library database and incorporate symmetric notches, of extreme geometric form and asymmetric
notches exposed limitations in the composition scheme in its presented form. Recent
developments have addressed these limitations to allow formulation of a universal composition
scheme applicable to all inter-related geometries shown above and application to the deepest point
of surface cracks.
SIF solutions obtained via the composition of weight functions have been shown to be of high
accuracy, having wide ranging limits of validity in a manner that is both rapid and of relative
mathematical simplicity. The composition of weight functions is therefore, ideally suited to
numerous engineering applications including that cited here, for which full numerical modelling
maybe unwarranted or inappropriate.
 17. SIM, Philosophy, Instrumentation and Analysis 1019

VARIABLE AMPLITUDE CORROSION FATIGUE OF HIGH STRENGTH

WELDABLE STEEL

S. S. Ngiam and F. P. Brennan

NDE Centre, Department of Mechanical Engineering, University College London,
Torrington Place London WC1E 7JE, UK
f.brennan@ucl.ac.uk

In recent years there has been considerable interest in the offshore industry in using high strength
weldable steels in the construction of offshore structures. This growing trend has seen high
strength steel widely used in the fabrication of Jack-up platforms for production purposes, largely
due to the optimisation of light weight design. This is particularly important due the competitive
Oil & Gas industry, as considerable reduction in weight can lead to a significant saving in both
manufacturing and operation costs.
High strength weldable steels with nominal yield strength of circa 450 to 700MPa have been
commonly used in the fabrication of jack ups, especially in the leg structures. Early realistic
loading fatigue studies that were conducted on offshore structures in the North Sea, focused only
on fixed platforms. For example, with the aid of the advancement of high-speed computer during
the last two decades, several working group such as COLOS (Common Load Sequence for the
European Coal and Steel Community Research Programme II), UKOSRP and Wave Action
Standard Load History (WASH) were established to standardise realistic variable amplitude
loading fatigue testing of offshore structures in the North Sea. These committees provided the
foundation of standard load histories for variable amplitude fatigue testing. These load histories
are not appropriate for Jack up structures as these are subjected to different loading conditions due
to the different dynamic response of these structures. In this paper, the use of a Jack up Offshore
Standard Load History (JOSH), which was developed under the framework of WASH, is reported.
Although many models, including those mentioned above, are able to simulate realistic load
sequences, the validity and suitability of the generated sequence are often overlooked. This paper
gives an insight to a stationary standardised load sequence, generated using the JOSH model, for
variable amplitude corrosion fatigue testing of jack-up structures. For clarity, the JOSH model,
together with its major components is presented.
Previous High Strength Steel projects have concentrated mainly on establishing the influence
of cathodic protection, and overprotection, on the fatigue life of welded joints as the HSE
Guidance has proposed the need for CP limiters when using High Strength Steels. A major finding
from the early studies was that at least some of the newer steels were not very susceptible to
overprotection (-1050mV) in a corrosive environment at modest lives. The impetus for this
research programme stems from concerns over the lack of data for fatigue of high strength steels in
the long life region. This study investigated the Long Life Corrosion Fatigue performance of
SUPERELSO 702 (SE 702), a widely used weldable High Strength Steel (made by Creusot-Loire
Industrie as a NiCrMo steel of similar strength).
The main objective of the investigation was to study the effect of cathodic over-protection on
the fatigue performance of high strength steel in the long life region. The corrosion tests were
conducted under simulated seawater environment. Two cathodic protection (CP) levels, -800mV
and -1050mV, were implemented by using CP limiters. Constant fatigue crack monitoring along
the weld toe of the specimens enabled the study of the influence of CP during long life fatigue.
The results obtained from this investigation are compared with previous Long Life Corrosion
Fatigue of high strength steel research programmes. The results are also compared with the fatigue
1020 S. S. Ngiam and F. P. Brennan

performance of conventional fixed platform steels such as BS EN 10025 Grade S355J2G3

(formerly known as BS4360 50D steel). Overall results have shown that there is a significant
enhancement in long life fatigue performance of SE 702 over the conventional fixed platform
steels. Although the cathodic over-protection specimens showed a significant reduction in fatigue
performance, the fatigue lives are well above the S-N design curves and the effect of hydrogen
embrittlement is less detrimental than expected.
 17. SIM, Philosophy, Instrumentation and Analysis 1021

CRACK MONITORING USING ACFM

R. F. Kare
TSC Inspection Systems
6 Mill Square, Featherstone Road
Wolverton Mill, Milton Keynes, MK12 5RB
ray@tscinspectionsystems.com

Key input data for SIM calculations are the service stresses and the crack/defect size. For the
offshore industry the Alternating Current Field Measurement (ACFM) technique [1] was initially
developed for subsea and topside inspection but is now used universally. ACFM has the capability
to both detect and size cracks. A recent innovation for ACFM has been the introduction of array
probes. These can collect crack depth information from various sites along the crack in one
placement of the probe. The array probe is an area inspection and hence can also be used for crack
size monitoring if left in place on the structure.
Previously, ACPD (Alternating Current Potential Drop), had been used for many years for
monitoring crack shape evolution in welded connections. In these cases the ACPD contacts were
usually spot-welded to the test component. The results from this work were of high quality and an
example of early fatigue crack growth on a tubular welded T joint tested under variable amplitude
corrosion fatigue [2] is shown in Figure 1. After the development of the non-contacting ACFM
technique from ACPD it became possible to monitor cracks without the need to attach probes.
In a recent series of fatigue tests [3] this new approach has been demonstrated using array
probes for monitoring. The array probes consisted of eight Bx and Bz coils arranged in a row at
10mm spacing. Four array probes were used to cover an area 160x20mm. The fatigue tests were
conducted on a high strength steel in four point bending at a frequency of 2Hz and a stress range of
200MPa. The specimen dimensions were 600 x 150 x 20mm and the weld had been ground to give
a smooth finish. Some detail of one of the tests is given here as an illustration of monitoring. The
test occupied a total of 1.1 million cycles and the majority of the crack data occurred over the last
night of the test.
Figure 2 shows the interpreted ACFM array data for crack depth growth rate for the latter part
of the test. The test ended in a fast fracture from a crack which was 10mm deep and 50mm in
length. It can be seen that the ACFM array probe showed a gradual increase in the crack growth
rate prior to the final fracture. The final crack size was confirmed from visual inspection after the
fracture event.

Figure 1: Early fatigue crack growth data, showing crack initiation in a tubular welded joint
1022 R. F. Kare

Figure 2: Interpreted ACFM array data for crack depth growth rate

References.
1. TSC Inspection Systems, 6 Mill Square, Featherstone Road, Wolverton Mill, Milton Keynes
MK12 5RB.
2. Myers, P. T., Corrosion fatigue and fracture mechanics of high strength Jack-up steels, Ph D
Thesis, London University, 1998.
3. Kare, R. F. and Tantbirojn, N., Monitoring Crack size using ACFM, SIMoNET Inaugural
Seminar March 2000, UCL.
 18. Fracture of Biomaterials 1023

FATIGUE BEHAVIOUR OF FIBER REINFORCED BONE CEMENT

B. Kumar and F. W. Cooke

National Institute of Aviation Research, Wichita State University
1845 N .Fairmount St Wichita KS USA
Othropaedic Research Institute Inc.,
Via Christi Regional Medical Center - St Francis Campus Wichita, KS, USA
brijesh.kumar@wichita.edu, Francis_Cooke@via-christi.org

The use of poly methyl methacrylate (PMMA) based bone cement as a grouting agent for the in-
vivo fixation of orthopaedic implants has been in practice for nearly fifty years. Fatigue failure of
the bone cement has been identified as the primary cause of cement failure. Implant loosening due
to the failure of the cement is one of the major reasons necessitating revision surgery. The need for
a more fatigue resistant bone cement is well documented in the literature [1, 2, 3, 4]. One method
of producing a more fatigue resistant bone cement is to reinforce it with short fibers [2].
In this investigation the impact of fiber reinforcement on the fatigue properties of the bone
cement was studied. The mechanism of fatigue in brittle matrices and short fiber reinforced
matrices is extremely complex. Therefore to during the course of this investigation only the
following two types of fiber reinforcements: short flexible Polyethylene Terephalate (PET) fibers
and stiff milled carbon fibers were used. To understand the mechanisms involved and their effect
on the fatigue life of the composite, a three pronged approach has been used in this research. The
key elements of the research can be summarized as follows:
• A test program was conducted to obtain fatigue data on fiber reinforced bone cement for a
wide range of test conditions.
• Rigorous statistical analyses of the fatigue data were performed.
• A fractographic study of the fracture surfaces of the fatigue specimens was performed to
develop a deeper insight into the micro-mechanics of the fatigue of reinforced bone
cement.
To optimize the fiber reinforcement of the cement the effect of the following parameters was
considered: (1) fiber types (stiff versus flexible), (2) fiber lengths, (3) fiber volume fraction and
(4) fiber surface treatments was also conducted. Since S-N type of testing for all the specimen
types would be very extremely time consuing, the comparison of all the specimen types was made
by testing at a maximum stress of 15 MPa, since this is the most widely used stress level quoted in
the literature on the subject [5]. The full S-N characterization of neat cement, one percent by
volume scoured PET fiber reinforced, one percent by volume as received carbon fiber reinforced,
one percent by volume treated carbon fiber reinforced, and two percent by volume as received
carbon fiber reinforced bone cement was conducted. It is not possible to provide all the results of
the testing conducted. Therefore only the S-N plot of different spedcimen types is shown on one
plot in figure 1. From the overall observation of the fatigue test data, the statistical analyses and
the fractographic analysis, it can be concluded that the reinforcement of the bone cement with the
stiffer carbon fibers was moderately beneficial to its fatigue life. Further, the interface between the
fiber and the bone cement matrix plays a crucial role in the energy dissipation process. The
fractographic evidence shows the lack of interfacial bonding between the PET resulting in poorer
reinforcement thus less positive influence on the fatigue resistance as compared to the carbon fiber
reinforcement.
1024 B. Kumar and F. W. Cooke

FIGURE 1. S-N Plot of the testing

References
1. Krause, W., and Mathis, R. S.," Fatigue Properties of Bone Cements: Review of the
Literature," Journal of Biomedical Materials Research, Vol. 22, No. A1, 37-53, 1988.
2. Reinforced Bone Cement in Orthopaedic Surgery," Journal of Biomedical Materials
Research, vol. 10, 893-906, 1976.
3. Topoleski, L. D. T., Ducheyne, P. and Cuckler, J. M., " Fracture Toughness of Titanium Fiber
Reinforced Bone Cement," Journal of Biomedical Materials Research, Vol. 26, 1599-1617,
1992.
4. Johnson, J. A., Provan, J.W., Krygier, J. J., Chan, K. H. and Miller, J., " Fatigue of Acrylic
Bone Cement – Effect of Frequency and Environment," Journal of Biomedical Materials
Research, Vol. 23, 819-831, 1989.
5. Rydell, N., "Forces in Hip Joint, Part II Intravital Measurements," Biomechanics and Related
Bio Engineering Topics, Ed. R. M. Kenedi, 351- 357, Pergammon Press 1965.
 18. Fracture of Biomaterials 1025

FRACTURE AND FATIGUE OF BONE AND BONE CEMENT: THE CRITICAL

DISTANCE APPROACH

David Taylor, David Hoey, Lorena Sanz and Peter O’Reilly

Trinity Centre for Bioengineering
Mechanical Engineering Dept., Trinity College, Dublin 2, Ireland
dtaylor@tcd.ie

This paper is concerned with the effect of stress concentrations on the failure of two materials of
major importance in the field of biomechanics: bone and bone cement. Stress concentrations such
as holes and notches are often introduced into our bones during orthopaedic surgery: examples are
holes drilled to accommodate screws in fracture fixation, and pieces of bone removed for biopsies.
Other workers have shown that the monotonic strength of whole bones is significantly reduced by
the presence of circular holes: two interesting observations are that the size of the hole has a
significant effect and that the size dependence is much greater for bones loaded in bending than
those loaded in torsion [1,2]. The present work aimed to predict these effects. Bone cement, which
is the polymeric material PMMA, is used widely in orthopaedic surgery to assist in the fixation of
implants such as the artificial hip and knee joints. Fatigue cracking of this material is responsible
for a large proportion of failures of these joint prostheses. Cement layers tend to contain defects,
including small spherical pores of entrapped air and larger defects which are essentially casting
defects. These features act as stress concentrations which can initiate fatigue cracking [3]. Much
effort has been expended to try to reduce the amount of porosity in bone cement: the work
described here attempted to quantify the effect of individual defect size and shape on high cycle
fatigue strength.
The theoretical model used in the work was the theory of critical distances (TCD). This
approach has been the subject of much recent investigation in our research group (see, for example,
[4,5]) and is also the subject of a special session at the present conference. The TCD can be used to
predict the effect of stress concentration features on both fatigue and monotonic brittle fracture.
The theory uses two material parameters: a critical stress and a critical distance. Failure is assumed
to occur when some function of the elastic stress in the vicinity of the feature becomes equal to the
critical stress. Various functions can be used, including the value of the stress at a fixed distance
from the feature (we call this the Point Method (PM) because we consider only the stresses at a
single point) and the average stress measured along a line drawn from the point of maximum stress
(we call this the Line Method). The TCD has not previously been used to predict the behaviour of
bone, though it has been extensively employed in studying fracture in fibre composite materials,
since the 1970’s [6]. We have previously shown that the TCD can be used to predict brittle fracture
in bone cement [5] and fatigue in metallic materials [4], but fatigue in bone cement has not
previously been studied using this approach.
Tests were conducted to measure the static strength (nominal stress to failure in monotonic
loading) for samples of bovine bone. The loading axis was either parallel or perpendicular to the
longitudinal axis of the bone. Some samples were plain (unnotched), others contained either a
central circular hole or a pair of sharp edge notches. Stress analysis was carried out using FEA.
Results showed that the TCD could be used, the critical stress being equal to the plain-specimen
tensile strength: critical distance values were found to be almost the same for transverse loading as
for longitudinal loading (L=1.5mm approx.), even though the tensile strengths in the two directions
were quite different. Applying these data, and using a multiaxial criterion of the critical-plane type,
we were able to predict other results on the effect of hole size; in particular we were able to
1026 D. Taylor et al.

demonstrate the much smaller effect of holes for bones loaded in torsion compared to bending
[1,2].
Bone cement samples were moulded and tested in fatigue up to one million cycles to failure.
Again we tested specimens which were either plain or contained holes or sharp notches, though in
this case even the plain samples contained small spherical pores as a result of the casting process.
Results showed that the TCD could be used also in this case, though it was found that the critical
stress was significantly higher than the plain-specimen fatigue strength, by a factor of slightly
more than 2. The consequence of this finding is that features having stress concentration factors
less than 2 (such as spheres) are predicted to have no effect in reducing fatigue strength (except for
any effect which they may have in reducing the load-bearing area). This prediction was confirmed
by the observation that, whilst the plain specimens contained spherical pores up to 2mm in
diameter, these pores were often not the source of failure: rather, cracks initiated from elsewhere in
the specimens.
In conclusion, the Theory of Critical Distances can be applied to both bone and bone cement,
to predict fracture and fatigue, using FEA to obtain information on local stress fields around
features and defects. The approach can usefully be employed to assess the likely reduction in
strength of a bone as a result of removal of material during an operation, and so to optimise the size
and shape of the removed portion. The theory can also be applied to analyse the effect of defects on
the long-term integrity of bone cement layers in artificial hip joints and other implants. In this
respect we showed that both the absolute size and the shape (especially the stress-concentration
factor) of the pores need to be considered when assessing the potential benefits of any procedure
for pore removal or reduction.

References
1. Specht, T.E., Miller, G.J. and Colahan, P.T. Amer.J.Vet Res. 51 1242-1246 1990.
2. Seltzer, K.L., Stover, S.M., Taylor, K.T. and Willits, N.H. Vet.Surg. 25 371-375 1996.
3. Culleton, T., Prendergast, P.J. and Taylor, D. Clinical Mater. 12 95-102 1993.
4. Taylor, D. Int.J.Fatigue 21 413-420 1999.
5. Taylor, D., Merlo, M., Pegley, R. and Cavatorta, M.P. Mater.Sci.Engng A 382 288-294 2004.
6. Whitney, J.M. and Nuismer, R.J. J.Comp.Mater. 8 253-265 1974.
 18. Fracture of Biomaterials 1027

FATIGUE FAILURE IN RECONSTRACTED ACETABULA – A HIP

SIMULATOR STUDY

J. Tong, N. P. Zant and P. Heaton-Adegbile1

Department of Mechanical and Design Engineering
University of Portsmouth, UK
1King Edward Hospital VII Hospital, UK

jie.tong@port.ac.uk

The long-term stability of total hip replacements (THRs) critically depends on the lasting integrity
of the bond between the implant and the bone. Late failure in the absence of infection is known as
“aseptic loosening”, a process characterised by the formation and progressive thickening of a
continuous layer of fibrous tissue at the interface between the prosthesis and the bone. Aseptic
loosening has been identified as the most common cause for long-term instability leading to gross
migration of the implants and failure of the THRs. There is clearly a need to study the failure
mechanisms in the acetabular fixation if the long-term stability of THRs is to be significantly
improved.
Retrieval studies based on revision operations at King Edward VII Hospital, UK, reveal that,
although microcracks develop in the cement mantle, it is the debonding between cement and bone
that often defines the final failure of cemented acetabular replacements. This was illustrated at the
revision surgeries by the easy removal of the acetabular cups with cement mostly attached to the
cup. It is felt that a fundamental understanding of the mechanisms that initiate and propagate the
interfacial failure at the bone-cement interface is the key towards solving the problem.
In this work, preliminary fatigue tests were carried out on cemented acetabular replacements
using a servo-hydraulic testing machine (Si-Plan Electronics Research Ltd., UK). Third-
generation of composite pelvic bones were used to which Charnley cups were implanted using the
bone cement, CMW, following the standard surgical procedures. The implanted hemi-pelvic bone
model was then constrained at the sacro-iliac and pubic joints to represent the anatomic constraint
conditions. To achieve the direction of the maximum hip contact force during gait, a milling vice
with two rotational degrees of freedom was used. The hemi-pelvis was loaded to the peak hip
contact force during normal walking (Bergmann et al., 2001), with a body weight of 70 kg
assumed. Sinusoidal waveforms were applied and the load was fully compressive from -0.2 to -1.6
kN at a frequency of 5 Hz. Damage was monitored using CT scanning and pronounced interfacial
fracture at the bone-cement interface was observed at 2 million cycles (Heaton-Adegbile et al.,
2005)
In vitro fatigue tests have been carried out utilising a new 4-station hip joint simulator (Simsol
Ltd, UK), specially designed for testing of implanted acetabula. Hip contact force during gait,
including both magnitude and direction (Bergmann et al., 2001), has been realised in the hip
simulator so that the effects of physiological loading on the damage development in the acetabular
replacement can be realistically assessed.
Damage development in the joints is monitored using CT scanning at regular intervals.
Permanent records are collected and the samples will be eventually sectioned and polished for
microscopic studies. Results from the conventional fatigue tests will be compared with those from
the hip simulator and recommendations with regard to pre- and post- clinical assessments of the
fixation will be made based on the findings of this work.
1028 J. Tong et al.

FIGURE 1. An implanted specimen under hip simulator testing.

References
1. Bergmann, G.; Deuretzbacher, G.; Heller, M.; Graichen, F.; Rohlmann, A.; Strauss, J., and
Duda, G. N. Hip contact forces and gait patterns from routine activities. Journal of
Biomechanics. 2001; 34(7): 859-872.
2. Heaton-Adegbile, P.; Zant, N. and Tong, J. In-vitro fatigue behaviour of a cemented
acetabular reconstruction. Submitted to J Biomechanics.
 18. Fracture of Biomaterials 1029

DEFORMATION AND FRACTURE OF BIOACTIVE PARTICULATE

COMPOSITES DEVELOPED FOR HARD TISSUE REPAIR

M. Wang
Department of Mechanical Engineering, The University of Hong Kong
Pokfulam Road, Hong Kong
memwang@hku.hk

Most human body tissues are composites in nature. Using natural tissues as templates, “designer”
bioactive ceramic-polymer composites are developed for tissue replacement and regeneration [1].
This biomimicking concept has now been extended into developing bioactive composites with
metals or ceramics as matrices. Furthermore, with the recent emergence of tissue engineering,
bioactive composite scaffolding materials are under active development [2]. The composite
approach is therefore being established as one of the most important and viable ways in developing
new biomaterials.
Hydroxyapatite (HA) reinforced polyethylene (PE) was the first bioactive ceramic-polymer
composite developed for bone substitution [3]. Considering the loading conditions of this
composite in its clinical applications situations, various types of mechanical testing have been
conducted [4-9]. The deformation behaviour of this material was different under different
deformation mode, which also depended on the HA content in the composite. In order to improve
the mechanical performance of HA/PE composite, two methods were investigated: chemical
coupling between HA particles and the polymer matrix [10], and hydrostatic extrusion [11]. The
deformation process and fracture of the improved HA/PE composite showed significant
differences from those of non-treated HA/PE composite [11, 12]. By carefully studying the
deformation process and fracture behaviour of HA/PE composite of different HA contents, insights
could be gained and lessons learnt for further developing various bioactive polymer-matrix
composites for human body tissue repair.

(a) Tensile behaviour and deformation (b) Compressive behaviour and deformation
FIGURE 1. Deformation and fracture of HA/PE composite (20vol.% of HA)

Generally, composite having HA of smaller mean particle size possessed higher Young’s,
compressive, torsional and flexural moduli. With an increase in HA amount in composite, modulus
values of the composite increased, i.e., the incorporation of hard HA particles in the soft PE matrix
stiffened the polymer. However, strength of the composite was not significantly increased by the
addition of HA into the polymer. During tensile testing, all specimens fractured, and necking was
noted before their fracture in composite having less than 20vol% of HA (Fig.1).
1030 M. Wang

2 3 4
FIGURE 2. HA/PE specimen during flexural testing
FIGURE 3. Flexural fracture surface of HA/PE composite
FIGURE 4. Deformation and fracture of HA/PE composite
During flexural testing, composite containing more than 40vol% of HA fractured but
composite of other HA volume percentages deformed without fracture after peak stresses had been
passed (Fig.2). During compression testing, only with a few exceptions, even composite containing
45% of HA did not fracture when compression tests were stopped at 70% compressive strain.
Composite specimens became barrel-shaped as compression tests proceeded and they went
through three main stages of deformation (Fig.1).
Debonding took place between HA particles and the polymer matrix during tensile and
bending tests as gaps were observed at the HA-polymer interface (Fig.3). Under tension and
bending loading conditions, the fracture mechanism of HA/PE composite was the same, which
involved dewetting, cavitation, voids-coalescence, and tearing and fracture of polymer fibrils
(Fig.4).
Tensile deformation of hydrostatically extruded HA/PE composite was significantly affected
by the polymer chain alignment [11]. Chemical coupling between HA particles and PE matrix
delayed the dewetting and cavitation process [12]. The shear stress component of the biaxial
fatigue stress could significantly shorten the fatigue life of HA/PE composite [9, 13].

References
1. Wang, M., Biomaterials, vol. 24, 2133-2151, 2003
2. Wang, M., et al., Journal of Materials Science: Materials in Medicine, vol.12, 855-860, 2001
3. Bonfield, W., Journal of Biomedical Engineering, vol.10, 522-526, 1988
4. Wang, M., et al., British Ceramic Transactions, vol.93, 91-95,1994
5. Wang, M., et al., Journal of Materials Science: Materials in Medicine, vol.9, 621-624, 1998
6. Wang, M., et al., Biomaterials, vol.19, 2357-2366, 1998
7. Wang, M., et al., Key Engineering Materials, vol.284, 693-696, 2005
8. Wang, M., et al., Key Engineering Materials, vol.254, 611-614, 2004
9. Wang, M., et al., In Proc. 10th Intern’l Conf Biomed. Engg, Singapore, 2000, 219-220
10. Wang, M., et al., Materials Letters, vol.44, 119-124, 2000
11. Wang, M., et al., Journal of Materials Science, vol.35, 1023-1030, 2000
12. Wang, M., and Bonfield, W., Biomaterials, vol.22, 1311-1320, 2001
13. F.H.-L.Tang, MEng Thesis, Nanyang Technological University, Singapore, 2001
 18. Fracture of Biomaterials 1031

FAILURE OF BIOMATERIALS IN IMPLANT FIXATION

Patrick J. Prendergast, John R. Britton, Paul T. Scannell and Alexander B. Lennon

Trinity Centre for Bioengineering
Department of Mechanical Engineering, Trinity College, Dublin, Ireland
pprender@tcd.ie

A surprisingly wide range of metals, polymers, and ceramics meet the requirements of
biocompatibility for use in orthopaedic implants [1]. Three metals are by far the most commonly
used: stainless steel, chromium cobalt alloys, and titanium and its alloys. Oxide ceramics (Al203,
ZrO2) are used for bearing surfaces and calcium phosphate bioceramics and glass ceramics are
used as coatings to invoke integration with host tissues. Regarding polymers, two are ubiquitous in
orthopaedic implants: ultra-high molecular weight polyethylene (UHMWPE) for bearings and
polymethylmethacrylate (PMMA) as a grouting to ‘cement’ implants into bones [2].
Unfortunately, in a very large number of cases, the biomaterials do not possess sufficient
mechanical durability – in consequence the implant fixation does not outlive the patient. Revision
operations are necessary in approximately 7% of hip and 5-10% of knee replacements at 10 years
[see the Swedish National Hip Arthroplasty Register at www.jru.orthop.gu.se and the Swedish
Knee Arthroplasty Register at www.ort.lu.se]. In the upper extremity, joint replacement has been
bedevilled by the challenges of creating durable implant/bone fixation between implant and bone
so that rather few operations are performed for these joints. Biomaterials can fail in their function
by:
• insufficient mechanical durability; failure arises by fatigue, damage accumulation, and
wear,
• provoking adverse biological responses; failure arises by bone loss or bone death due to
inappropriate stressing of the peri-prosthetic tissues, failure of bone ingrowth due to
relative motion between implant and tissues, or osteolysis due to wear particles, etc…
Therefore, despite the undoubted success of orthopaedic implants [2], great challenges remain
in improving the performance of the biomaterials if a satisfactory outcome for the individual
patient is to be guaranteed. How can this be done? We believe an important challenge lies in an
accurate understanding of how in-service loading regimes provoke various failure scenarios [3] in
the implant/bone complex in the individual patient.
In 1993, Huiskes [3] introduced this concept of ‘failure scenarios’. These scenarios are
interlinkings of (i) and (ii) above – interlinkings that are empirical and based on clinical evidence.
In this paper we will consider two failure scenarios for total hip replacement and we will present
computational simulations of how the biomaterial affects the failure scenarios. The first is the
damage accumulation failure scenario and the second is the failed ingrowth failure scenario. In
simulating the damage accumulation failure scenario, stochastic effects are introduced by
performing Monte Carlo simulations of damage accumulation in bone cement with randomly
generated distributions of porosity. Microcrack initiation and growth are modelled using a
multiaxial damage accumulation model based on bone cement fatigue strength data. This Monte
Carlo approach can simulate the same variability and mean fatigue life of bone cement specimens
as seen experimentally [4]. The algorithm can then be used to simulate the damage accumulation
failure scenario in a cemented joint reconstruction; a unique failure will be predicted for each
analysis – this is analogous therefore to the real situation. Data can then be combined from each of
the simulations to give a statistical estimate of variability and mean damage accumulation. The
1032 P. J. Prendergast et al.

failed ingrowth failure scenario occurs in non-cemented implants when bone fails to grow into the
implant surface; instead fibrous tissue persists at the bone-implant interface and, in the end, the
implant loosens. This process is simulated using finite element analysis in conjunction with an
algorithm that relates the rate of change of density to the applied strain and the accumulated
damage [5]. The influence of the Young’s modulus of the stem on the propensity for this failure
scenario can then be tested in a simulation.
To allow selection of the most appropriate implant for an individual patient, a model must be
generated of the patients specific bone. This is a difficult objective because a patient-specific
images and loading data must be imported - Fig. 1 gives an outline of the approach

FIGURE 1. A schematic diagram for the concept of patient-specific prosthetic analysis. Data from
the patient’s electronic record is used to construct a patient-specific finite element model. The
probability of failure (by damage accumulation or failed ingrowth) can then be computed.

References
1. Park, J.B., Lakes, R.S., Biomaterials. An Introduction, 2nd Ed., Plenum, New York, 1992
2. Prendergast, P.J., In Bone Mechanics Handbook, edited by S.C. Cowin, CRC Press, Boca
Raton, 2001, Chapter 33
3. Huiskes, R., Acta Orthop. Scand., vol. 64, 699-720, 1993
4. Lennon, A.B., Prendergast, P.J., Math. Proc. R. I. A. (www.ria.ie) vol. 104A, 155-171, 2004
5. Scannell, P.T., Prendergast, P.J., Engineer’s Journal, (in press) 2005
 19. Structural Integrity Assessment in Theory and Practice 1033

STRESS ANALISYS OF HIGH PRESSURE STEAMLINES IN THERMAL

POWER PLANTS

A. Jakovljevic
Electric Power Industry of Serbia, Head Department for Development and Investments, Belgrade,
Serbia
aleksandar.jakovljevic@dri.eps.co.yu

Steamlines present statically undefined pipe structures allocating between connection points on the
boiler and turbine; they consist of mutually welded elements connected to support structure by the
support system. For this reason, it is necessary to bear in mind during all stress analyses of steam
lines, an entire sequence of additional influences, which have considerable impact on the behavior
of steamlines, and are not directly related to the operating parameters of the fluid: pressure and
temperature. Overstressing by permanent, thermal and dynamic loads is primarily caused by
inadequate configuration of the steam line, irregular operation of support system and loads in
transintient states.
High pressure steam pipelines are critical components of thermal power plants, which have
significant influence on reliability and availability of plants as well as staff safety. Therefore,
during operation, it is necessary to monitor their actual condition and also to provide preconditions
for assessing high pressure steam pipeline’s behavior during further exploitation in order to
perform residual life assessment. In operation conditions, high pressure steam pipelines are
exposed to creep and low cyclic fatigue, which lead during the time, to microstructural degradation
as well as appearance and development of steel damage.
Methods for quantitative estimation of the remaining operation life are mainly based on the
assumption that component is exposed only to the activity of internal pressure, which simplifies
the problem. This is generally the reason why analyses are started with stress values, which do not
represent stress to which the steamline component is exposed to during operation.
Steamline has to be observed during the analysis as a complex structure, which is exposed,
besides the action of internal pressure of the operation fluid, also to the action of additional
external load. It greatly depends on configuration and support system of the steamline and it causes
the appearance of stress conditions that greatly differ from design ones. Their negative impact on
the operation life of the steamline and correct estimation of the remaining operation life require the
consideration and quantification not only the impact of internal pressure and temperature but also
all effecting loads. Due to that, usage of numerical methods for calculation and analisys of stress
condition is neccesary.
Most of numerical methods for the calculation of stress state are based on the application of
finite elements method, which can be used in two ways, from the aspect of steamline analysis: for
structural calculations of the complete structure, as well as for the calculation and analysis of
individual components.
The paper gives an example of the performed stress analysis of the main steamline at the unit
A1 – TPP Nikola Tesla in Obrenovac, which is used to present the importance of stress analyses
and the possibilities and restrictions of numerical methods for the performance of such analyses
(Fig. 1).
Considered steamlines were selected for the analysis because they with nearly 200.000
operation hours (designed for 100.000 hours) represent the oldest steamlines in Electric Power
Industry of Serbia plants.
1034 A. Jakovljevic

FIGURE 1. Stress condition of the main steam line in hot state

Structural analysis of stress state of these steamlines was performed according to ASME/ANSI
B31.1 standard. Then, two pipe elbows of the same size were chosen from steamline. Selection is
made so that one of them belongs to the group of pipe elbows with the highest load and the other to
the group with the lowest load on the steamline and that there is sufficient date from previously
performed investigations. Detailed analysis of the stress state of chosen elbows by the finite
elements method was performed using linear-elastic analysis, for three different load cases.
Application of finite elements method, although with some simplifications, enables to obtain
the clearer picture and better consideration of the behavior of analyzed steamline components.
Performed stress analyses enable the identification of critical points in which the stress is increased
in the steamline and its components. These points have to be taken into consideration during the
monitoring of steamline state. Impact of external load on the appearance of considerably different
stress state in some components with nominally the same geometry, operating on the same
pressure and temperature, as well as the possibility of appearance of the most unfavorable stress
state outside the anticipated zones was clearly demonstrated.
 19. Structural Integrity Assessment in Theory and Practice 1035

LAMINAR COMPOSITE MATERIALS DAMAGE MONITORING BY

EMBEDDED OPTICAL FIBERS

Aleksandar Kojovi, Irena Zivkovi1, Ljiljana Brajovi2, Dragan Mitrakovi

and Radoslav Aleksi
Faculty of Technology and Metallurgy
1Institute of Security
2Civil Engineering Faculty

Karnegijeva 4, Belgrade, Serbia and Montenegro

1Kraljice Ane bb, Belgrade, Serbia and Montenegro
2Bul Kralja Aleksandra 73, Belgrade, Serbia and Montenegro

koja@tmf.bg.ac.yu, zirena@tmf.bg.ac.yu, brajovic@grf.bg.ac.yu, dragan@tmf.bg.ac.yu,

aleksic@tmf.bg.ac.yu

This article the describes procedure of embedding fiberoptic sensors in laminar thermoplastic
composite material, as well as damage investigation after low energy impact, in real time.
Experimental testing was carried out in order to observe and analyze the response of material under
various load conditions. Different type of Kevlar® reinforced composite materials, and
combination of Kevlar® and metal mesh (thermoplastic or thermoreactive matrix) were made.
For that purpose intensity based optical fibers were built in specimens of composite materials.
Main advantages of fiber optic sensors are their shape and structure which result in easy
embedding while preserving mechanical properties of composite material, which make them a
reliable automatic system for structure health monitoring.

FIGURE 1. Specimen placeholder.

Impact toughness testing by Charpy impact pendulum with different loads was conducted in
order to determine method for comparative measurement of resulting damage in material.
Specimens were fixed in placeholder as shown in Fig. 1. Light from the light emitting diode (LED)
was launched to the embedded optical fiber and was propagated to the phototransistor based photo
detector. During each impact, level of signal, which is proportional to light intensity in optical
fiber, drops, and then slowly recovers
1036 A. Kojovi et al.

FIGURE 2. Characteristic signal change during impact.

Obtained signals were analyzed to find appropriate method for real time damage monitoring.
Major part of damage occurs during the impact. Damage reflects in local, temporary release of
strain in optical fiber, and raise of signal level as could be seen during the impact in Fig. 2.
Mathematical method was developed and applied on signals to monitor development of damage in
material and for analyzing quality of material.
Results show that intensity based optical fibers could be used for measuring damage in laminar
thermoplastic composite materials, and could be used as reliable automatic system for long-term
structure health monitoring. Light signal intensity drops in an optical fiber in response to applied
loading on composite material. Acquired optical fiber signals depend on type of material, but same
set of rules (relatively different, depending on type of material) could be specified. Existing
methods in most cases uses just intensity of signal before and after the impact, as measure of
damage. In this case using real time measurement of signal during impact and appropriate analysis
enables quantitative evaluation of impact damage in material and could be used to monitor damage
in real time, giving warnings before the fatal damage occurs.
 19. Structural Integrity Assessment in Theory and Practice 1037

SOL GEL SYNTHESIS AND STRUCTURE OF HYBRID NANOMATERIALS

WITH STRONG CHEMICAL BONDS

B. Samuneva, P. Djambaski, E. Kashchieva and G. Chernev

Department of Silicate Technology,
University of Chemical Technology and Metallurgy, Sofia, Bulgaria
I.M. Miranda Salvado, M.H.V. Fernandes, A. Wu
Department of Ceramic and Glass technology, CICECO, University of Aveiro, Portugal
samuneva@uctm.edu

Sol-gel synthesis at molecular level and low temperature gives us a possibility to obtain new
materials with non-traditional physical and chemical properties of special interest, highly
homogeneous and pure. Recently a great attention has been paid to this method because of
synthesizing new hybrid (organic-inorganic) materials having nanoscaled structure. The newly
synthesized hybrid coatings and bulk materials can be successfully applied in optics, electronics,
technics, medicine, and biotechnology. The chemical composition of hybrids and the nature of
chemical bonds between the organic and inorganic components in their structure are the most
important parameters determining physico-chemical behavior of these materials. The different
organic-inorganic hybrids can be classified into two broad families, according to the nature of
chemical bond between their organic and inorganic components: Class I – hybrid systems in which
one of components (organic, biologic or inorganic) is entrapped within a network of the other
component and Van der Walls Hydrogen bonding or electrostatic interactions are found; Class II –
hybrid materials in which the inorganic and organic parts are chemically bonded by a covalent or
ionic-covalent bond. The hybrids belonging to Class II are more appropriate for obtaining of
nanomaterials with high mechanical and corrosion resistance.
The main purpose of this paper is to present a review of our recent research results in sol-gel
synthesis and structure of multifunctional silica hybrid nanomaterials with strong chemical bonds
between their organic and inorganic components.
The silica hybrid materials have been synthesized via sol-gel technology at room temperature.
Four different types of silicon precursors have been used - tetraethylortosilane (TEOS),
tetramethylortosilane (TMOS), methyltriethoxysilane (MTES) and ethyltrimethoxysilane (ETMS).
The hybrids were synthesized by replacing a different quantity of inorganic precursors with 5, 10,
15 and 20 mol % of the following organic compounds, respectively: polyethylene glycol (PEG),
polyvinyl alcohol (PVA), acrilamide (AA), polyacrilamide gel (PAAG), methyl methacrylate
(MMA), calcium alginate and agar-agar. The processes of sol-gel synthesis and structure evolution
of gels have been studied by means of FT-IR (IR-MATSON 7000–FTIR spectrometer), XRD (X-
ray PW1730/10 diffractometer), BET-Analysis (Gemini 2370 V5.00), EDS (RONTEC EDS
System), SEM (Philips-515) and AFM (NanoScope IIIA Tapping ModeTM). Thin transparent
hybrid flakes and films have been obtained and no separation has been observed before and after
gelation point.
The results from the XRD - analysis prove that all the hybrids described in this study are in
amorphous state, which is in good correlation with some other published works. The intensity of
the XRD patterns depends on the type and quantity of the organic component. The data from BET
- analysis show that the pore diameter is about 2 nm and that is connected with an increase of the
organic component concentration; the surface area of such pores decreases from 587 m2/g to 125
m2/g.
1038 B. Samuneva et al.

FIGURE 1. AFM image of a hybrid sample (MTES) containing 20 % (MMA)

To establish the existing chemical bonds in the hybrids synthesized, the FT-IR Spectroscopy
has been used. It has been proved that the type of silica precursors influence significantly affects
the formation of different in-nature chemical bonds. The absorption band at 2975 cm-1 , 1255 cm-1
and 694 cm-1 , due to the presence of Si-O-R (CH3 and C2H5) and Si-C bonds have been
registrated in the IR spectra of the hybrids with ETMS. This fact directly proves the presence of
strong chemical bonds between the inorganic and organic parts of the synthesized materials. At the
same time it can be supposed that in the samples with TEOS only Van der Walls Hydrogen bonds
or electrostatic interactions exist. The presence of well-defined nano units and their aggregates in
the hybrid structure, formed by self-organizing processes, is clearly observed by AFM, Fig. 1. The
average size of nanoparticles on the sample surface is from several nanometers up to 30 nm; the
development of nanostructure evolution processes can be seen.
The synthesized multifunctional hybrid nanomaterials could be successfully applied as carriers
for cell immobilization and for corrosion and mechanical resistant coatings.
 19. Structural Integrity Assessment in Theory and Practice 1039

AN ALTERNATIVE APPROACH TO CONVENTIONAL DATA

PRESENTATION OF FATIGUE

D. Angelova
Donka Angelova – Professor, University of Chemical Technology and Metallurgy-Sofia (UCTM),
8 Kl. Ohridsky Blvd., 1756 Sofia, Bulgaria
dona_ang@techno-link.com

Although major advances have been made in fatigue modelling, the application of fatigue concepts
to different practical situations is highly individual and often involves empirical and semi-
empirical approaches including a large number of specifying constants. For most engineering
alloys the plot of crack growth rate, log da / dN against stress-intensity factor range,
log ' K exhibits a sigmoidal curve with three distinct regimes of crack growth, Suresh [1]: I of
threshold behaviour; II of Paris line log da / dN – log ' K ; and III of rapid increasing of da/
dN causing catastrophic failure. The Paris regime is described by an equation log da / dN 
C log ' K  , where C and m are scaling constants, and sometimes applied to all data
m

obtained: M(K) in Fig. 1. At elastic-plastic conditions, [1] modelling based on

da / dN  C log ' J m takes

FIGURE 1. Newly proposed alternative (Q) and conventional (M) data presentations of fatigue in
a high-strength spring steel (C 0.56, Mn 0.81, Si 1.85, Cr 0.21, Ni 0.15, P 0.026, Mo 0.025, S
0.024) under fully reversed torsion, Murtaza [2]

part: M(J) in Fig. 2. For the rest of ingineering materials the multitude of points M – M(K) or M(J)
– may be expressed by different types of curves or just as a cloud of points.
An alternative method is proposed comprising the fatigue testing of ingineering materials;
1040 D. Angelova

FIGURE 2. Alternative (Q) and conventional (M) presentations of fatigue in a E-Ti-6Al-4V alloy
(Al 6.52, V 4.00, Fe 0.16, O 0.182) under axial and combined loading, Hoshide [3]

measuring long-crack or main short-crack lengths a and the corresponding number of cycles
N ; calculating crack rates da / dN , stress-intensity factor range 'K or J-integral range
'J, and a newly introduced energy fatigue-function in its 4 versions W ( da / dN ) ' K or
W ( da / dN ) ' J ] or W ( da / dN ) a 1 / 2 or kW k ( da / dN ) ' K where k is a

normalizing constant from the type k N f a f K max a  , a f

f and N f are respectively the

final length of fatigue crack and the number of cycles at failure, and K max a f   is the stress
intensity factor at V max
from the applied stress range ' V V max V min ; kW is a non-
dimensional expression. The presentation log da / dN  log W or
log da / dN  log kW is an almost straight line, Q(K) or Q(J) in Figs. 1 and 2, which may be
termed the fatigue tendency of the material at a given stress range, mentioned for the first time and
only for 'K in Angelova [4]. The two figures show the equation of this line in Q(K) and Q(J) as a
thick line; each scatter band around such a line is indicated by two thin lines corresponding to ½
and 2 folds of da/dN.
A comparative analysis between M and Q presentations shows that at the same number of
crack-size measurements, the precision of the proposed method is significantly higher, expressed
quantitatively by the corresponding correlation coefficients fc shown in Figs. 1 and 2. This result
suggests a possible decrease of fatigue measurements and is approved for another 11 materials.

References

1. Suresh S., Fatigue of Materials, Cambridge University Press, Cambridge, UK, 1998.

2. Murtaza G., Ph.D. Thesis, University of Sheffield, UK, 1992.

3. Hoshide T., Hirota T., Inoue T., J. Materials Sci. Research Int.,Vol. 1, No 3, 169-174, 1995.
4. Angelova, D." In Proceedings of ECF13 on CD, Abstracts Volume, San Sebastian, September
2000, 128.
 19. Structural Integrity Assessment in Theory and Practice 1041

ABSORBERS OF SEISMIC ENERGY FOR DAMAGED MASONARY

STRUCTURES

D. Sumarac2, Z. Petraskovi1, S. Miladinovic1, M. Trajkovi3, M. Andjelkovic2

and N. Trisovic4
1system DC 90, Vele Negrinove 1,11000Belgrade, Serbia and Montenegro (SCG)
2Faculty of Civil Eng., Univ. of Belgrade, Bul. Kralja Aleksandra 73, 11000 Belgrade, (SCG)
3Faculty of Civil Eng. and Arch., Univ. of Nis, Aleksandra Medvedeva 14, 18000 Nis, SCG
4Faculty of Mechanical Eng. Univ. of Belgrade, Kraljice Marije 16, 11000 Belgrade, SCG

DC90@eunet.yu, sumi@eunet.yu, medabgd@sbb.co.yu, nrt2835070@eunet.yu

In the present paper problem of damage of masonry structures due to earthquake is considered. It is
known that masonry structures are very sensitive to earth shaking. Built from the bricks and bond
by mortar, they cannot transmit large tensile stresses. Due to small earthquake of the magnitude 4-
5 of the Richter scale those structures are usually damaged. Numerical calculation using FEM is
applied on typical example of two store building ( Fig. 3). Also in the paper procedure of retrofit of
damaged structures, using system DC 90, is shown. Results of laboratory testing of dampers of the
system DC 90 due to dynamic loading with frequency 1-10 Hz are presented. Results are obtained
in three institutes: IZIIS (Skopje), VTI (Belgrade) and IMS (Belgrade). Obtained results show high
performances of dampers, which after embedment increase stiffness and ductility of structures up
to 30% (Fig. 2). Besides that, results of model testing, obtained on vibro-platform for different type
of masonry structures, which have been used in the areas of tectonic faults, are shown. Special
attention have been made to testing in situ due to ambient vibrations (wind, earth shaking) and
induced vibrations due to eccentric mass (Fig. 1). This testing was performed on damaged
structures before and after embedment of dampers. Measurements are performed in Kolubara
region by institute IZIIS and IMS. All investigations show high performances of embedded
dampers (for which word prize was obtained on international exhibitions of patents and new
technologies, Brussels Eureka 97). In the present paper process and methodology of transfer of
technology for application of the system for large area heated by earthquakes is specially outlined.
On examples of large number of objects reconstructed using system DC 90 in Kolubara region
(Serbia), Algeria and Iran all advantages of the system DC 90 will be presented.

FIGURE 1. Equipment for generation of harmonic vibrations

1042 D. Sumarac et al.

FIGURE 2. Building before and after strengthening

FIGURE 3. FEM model of building

References
1. Petraskovic, Z., Miladinovic, J.S., Technology of the Seismic Strengthening of Masonry
Structures by Applying Vertical Ties and Diagonals with Seismic Energy Absorber, IMS
Institute, Conference, Belgrade, Serbia and Montenegro, 2000
2. Sumarac, D., Krajcinovic D., Elements of Fracture Mechanics, Naucna knjiga, Belgrade,
Serbia and Montenegro, 1990 (in Serbian)
3. Petrovic, B., Selected Topics of Seismic Engineering, Gradjevinska knjiga, Belgrade, Serbia
and Montenegro, 1989 (in Serbian)
 19. Structural Integrity Assessment in Theory and Practice 1043

NUMERICAL ANALYSIS OF TENSILE SPECIMEN FRACTURE WITH CRACK

IN HAZ

Gjorgji Adziev1, Aleksandar Sedmak2 and Todor Adziev1

1 Faculty of Mechanical Engineering - Skopje, FYROM
2 Mechanical Engineering Faculty, Belgrade, Serbia & Montenegro

gadziev@mf.ukim.edu.mk

The numerical solutions enable investigation of complex problems, including various effects,
particularly in the cases where the experimental investigations and mathematical solutions are
unappropriate due to high costs. This research deals with numerical modelling and behaviour
analysis of tensile specimens, with crack located in the heat-affected-zone (HAZ), with respect to
the effect of the weld strength mismatch on the stress-strain distribution around the crack tip and
the causes leading to failure. ANSYS package for structural analysis was used for preprocesing,
procesing and postprocesing. The welded joint is modeled as simplified multiregion material
system, consisting the typical root, fill and surface layers, as well the HAZ region consisting the
fine-grained (FG) HAZ and coarse-grained (CG) HAZ, Fig. 1. The crack tip with very small radius
of 0.05 mm i.e. the crack front in the 3D (Fig. 2) analysis is modeled with great refinement of the
mesh, thus enabling singularity of the stress-strain field around the crack tip. As material
properties, in the input file the module of elasticity, the Poason's factor and the multilinear V-H
curves are defined.
Numerical analysis consisted of 3D and 2D modeling. The behavior of a specimen is 3D, but
2D analysis is also beneficial since the plane stress and plane strain conditions limit the structure
behavior. Plane stress is dominant state on the free surface and is typical for structures of small
width. Plane strain state is dominant in the middle of the specimen and is typical for structures of
greater width. The three-dimensional condition is between these two extreme conditions, since the
width of the tested specimens is between the definition of small and great width.
Two cases are analysed, specimens with crack tip located in the fine-grained in the coarse-
grained HAZ, both made of microalloyed high strength steel for pressurezed equipment. The both
cases exhibited high fracture resistance, due to the barrier effect of the mismatching, shifting the
initial crack towards the more ductile parent metal, thus decreasing the crack propagation rate. As
proved both experimentally and numerically, the overmatching in welded joint has acted protective
in respect to crack propagation, regardless of the position of crack tip (FG HAZ or CG HAZ),
increasing significantly the fracture resistance. In both cases, after the initial propagation of the
crack in FG HAZ or CG HAZ, relatively small overmatching has changed crack propagation
direction toward weaker and more ductile parent metal. In the case of crack tip positioned in FG
HAZ, being between stronger CG HAZ and weaker parent metal, the effect of overmatching is
obvious. In the case of crack tip in CG HAZ, being between weaker FG HAZ and weaker weld
metal, this effect is due to combined influence of slightly lower yield strength of FG HAZ and its
significantly smaller size, and also due to the lower yield strength of the parent metal. Both pairs of
curves, J vs. CMOD and F vs. CMOD exhibited good agreement between the experimental and 3D
numerical results. Comparing with the 2D analysis, the experimental curves were between the two
extreme conditions, as expected. The partial mis-agreement between the numerical and
experimental results is due to: geometrical simplification of the welded joint, including its shape,
dimensions, and its composition, material behavior simplification, having in mind the analytical
determination - assessment of the V - H curves for the welded joint regions using the Ramberg-
Osgood law and, finally, the finite element analysis used here did not include crack extension.
1044 G. Adziev et al.

FIGURE 1. The simplified FEM model of the welded joint

FIGURE 2. The 3D model with enlarged crack tip region

 19. Structural Integrity Assessment in Theory and Practice 1045

DETERMINATION OF JR-CURVE BY TWO POINTS METHOD

I. Blacic and V. Grabulov

Military Technical Institute, Belgrade, Serbia and Montenegro
venciaiv@eunet.yu

Preliminary results of investigation on the application of new approach to determine resistance

curve based on two pairs of known J- 'a values are presented in the paper. [1,2]
Proposed procedure is based on the possibility to present the dependance of J-integral on crack
extension 'a as a polynomial function [1]. This relation, given in ASTM 1737 standard, imvolved
two constants:

C 1 ˜ ' a 
C2
J (1)
Constants C1 and C2 can be determined for two pairs of J integral and crack extension 'a, if
'a z 0, sincein that case J integral is zero for each value of C1 and C2. Statical conditions allow to
break the test and by heat tinting to mark crack extenision 'a. The Ji follows from

2 ˜U i
Ji
B ˜ b0 (2)
if no correction of J integral for crack extension is applied, or from recommendation ESIS P2

Ji
2 ˜U i §
¨¨ 1 
0 . 75 ˜ K  1 ˜ ' a ·
¸¸
B ˜ b0 © b0 ¹ (3)
which include correction of J integral for crack extension, with K = 2 for SENB type specimen.
Equation for J integral from ASTM 1737 is not convenient since it is of incremental character
for calculation of J integral plastic component (see A1.5.3. of this standard).
New approach for resistance curve determination is based on Equ. (1), with constants C1 and
C2 are defined based on the measurement of one pair values J- 'a at the end of the test and
selection of point P1 from the plot load - load line displacement (Fig. 1), belonging to the blunting
line, presented by relationship:

J M ˜V y ˜ 'a
(4).
In this way, with two pairs values of crack extension 'a and J integral, constants C1 and C2
are defined.
Anyhow, the practical application revealed that values of C1 and C2 strongly depends on
position of P1 point on P- f0, Fig. 1, but also on blunting line defining procedure. For that, criterion
of selection of point P1 must be defined.
The paper suggests criterion for selection of point P1 via relation deflection – load for point P1
and maximum load Pmax, characteristic point on load – deflection plot.
1046 I. Blacic and V. Grabulov

Preliminary investigation for defining criterion for point P1 selection are made with high
strength steels (seven strength levels in the range 700 to 1320 MPa). Based on found resistance
curves selection of point P1 is performed by compliance method, enabling best agreement for new
proposed method application. Obtained relationship load vs. displacement for load values
corresponding to points P1 i Pmax in Fig. 1 is given in Fig. 2 for steels of different strength levels.
The use of diagram from fig. 2 enables to define ratio f1/P1 from fmax/Pmax ratio, and thus the point
P1 position. In this way the pair of J- 'a is determined by Equ. (2) to (4) , which with another pair,
determined experimentally at the end of the test allows to determine constants C1 and C2 and
resistance curve.

FIGURE 1. Chracteristic points on load- deflection plot FIGURe 2. Dependance f1/P1 vs. fm

Performed preliminary investigation has shown that it is possible to define the criterion for
selection of point P1 on the load – deflection curve, but it is only one criterion, further investigation
is necessary.

Refferences
1. ASTM E 1737–96 “Standard Test Method for J-integral Characterization of Fracture
Toughnes”
2. ESIS P2-92 “ESIS Procedure for Determining the Fracture Behaviour of Materials”
3. Pod red V. Dalja, V. Antona “Statieskaja pronost i mehanika razrusenija stalej” Moskva
Metalurgija 1986 god.
4. M. Vnuk “Problemi stabilnosti zilavog loma” u monografiji “Uvod u mehaniku loma i
konstruisanje sa sigurnosu od loma” str. 141-168 TMF Beograd 1980 god.
5. G. A. Clarck i dr. “Single Specimen Test for JIc Determination” ASTM STP 590 ASTM
Philadelphia 1976 god.
6. G. Irwin “Fracture mechanics” u “Mechanical testing” ASM HANDBOOK vol. 8 ASM
Metals park
7. Dz. Dzojs “Ispitivanje J integrala i primena u elastoplastinoj mehanici loma” u monografiji
“Perspektive razvoja i primene mehanike loma” str. 27-60 TMF Beograd 1987 god.
 19. Structural Integrity Assessment in Theory and Practice 1047

MONITORING OF STRESS-STRAIN STATE OF BOILER DURING PESSURE

TEST

Jano Kurai, Zijah Burzic1, Nikola Garic, Milorad Zrilic2 and Bosko Aleksic
CertLab, Pancevo, Serbia and Montenegro
1Military Technical Institute, Belgrade, Serbia and Montenegro
2University of Belgrade, Faculty of Tehnology and Metalurgy, Serbia and Montenegro

certlab@panet.co.yu

In order to extend power capacity a new boiler BF-9602, steam production110 t/h has been
installed in Oil Refinery Panevo. Boiler producer is "Slovenske energetike strojarne A.S." Tlmae -
Slovakia. During the assembling of tube No 13, produced of St 35.8/I steel, missalignement
occurred, e.g. in welding process the was not properly positioned in the axis (Fig. 1).

FIGURE 1. Zone of misalignment of tube No 3.

During the assembling of tube No 52, also St 35.8/I steel, in the boiler upper part it has been
destorded and plastically deformed, Fig. 2.

FIGURE 2. The region of plastic deformation on the tube No 52.

In order to assess “fitness-for-purpose” of assembled tubes, Department of investment NIS Oil

Refinery Panevo in Inspection procedure for boiler acceptance has specified additional testing in
regions presented in Figs. 1 and 2, enabling to evaluate the stress and strain state of damaged zones
during proof pressure test by cold water. This testing should be performed from outer side,
applying appropriate methods, which will not affect the structural integrity of boiler. The final
decision about damaged zones repairing will be made after the analysis of stress state, assessed by
testing.
1048 J. Kurai et al.

Since new boiler has to be inspected after assembling, investor accepted the proposal to
perform also the tests for stress assessment on water mantle of boiler. Additional aim of this testing
was to define boiler state (so-called zero state) before service, only possible during first proof
water pressure test.
For accomplish this task, having in mind the location and orientation of damaged regions, as
well as medium and operating condition, following methods have been applied:
• Acoustic emission method, and
• Strain gauges measurement.
Obtained results and performed analysis allowed following conclusion:
Measured microstrains on strain gauges and calculated normal stresses in measuring regions
revealed that up to proof test of 78,8 bar all stresses have been in elastic range, indicating that no
plastic deformation in tubes and in boiler mantle.
This is confirmed by acoustic emission sensor, since during pressurizing no acoustic signal
was monitored, e.g. no one activity was plotted.
Based on performed testing and analysis it had been suggested to investor to accept the boiler
as “fit for purpose”, with no further action.
 19. Structural Integrity Assessment in Theory and Practice 1049

LOCAL VARIATION OF CRACK DRIVING FORCE IN A MISMATCHED WELD

Jozef Predan, Nenad Gubeljak and Otmar Kolednik1

University of Maribor, Faculty of Mechanical Engineering, Smetanova 17, Maribor, Slovenia,
jozef.predan@uni-mb.si
1Erich Schmid Institute of Materials Science, Austrian Academy of Sciences, Jahnstr. 12, A-8700

Leoben, Austria
kolednik@unileoben.ac.at

The paper deals with the assessment of the fracture resistance of an inhomogeneous welded joint.
The material inhomogeneities create a difference between the near-tip crack driving force, Jtip, and
the nominally applied far-field crack driving force, Jfar. This difference is quantified by the so-
called material inhomogeneity term, Cinh, which can be evaluated by a post-processing procedure
to a conventional finite element stress analysis. Figure 1 shows a welded joint, where the weld
metal is in half overmatched (OM) and half undermatched (UM) configuration. Such welds are
commonly used for repair welding or for welded joints where the possibility of hydrogen assisted
cracking exists. The crack is perpendicular to the mismatch interface. The material properties have
a jump at the mismatch interface, but are assumed constant in the regions above and below. In a
preliminary study it has been shown that the slant interfaces to the HAZ have no noticeable effect.
For this reason, the geometry is simplified to that of a CT specimen with a biomaterial interface
between the OM and UM weld metal. The material inhomogeneity term is given by

C inh  e˜³
6
>>I @@ I  ı ˜ >>grad u @@  ˜ n ds
(1)
where V is the Cauchy stress, u the displacement, I the identity matrix, e the unit vector in the
direction of crack growth, and n the unit normal to the mismatch interface 6.

FIGURE 1. Cross-section of the welded joint with notch position in under matched weld metal

A stationary crack is considered, but the distance between crack tip and interface is
systematically varied. To assess the influence of several material parameters on Cinh and Jtip,
analyses are performed for an inhomogeneity of the modulus of elasticity, the yield stress, and the
stress hardening exponent separately, as well as for several combinations of the material
parameters. The results demonstrate that the material inhomogeneity term is primarily determined
by the magnitude of the yield stress inhomogeneity.
1050 J. Predan et al.

FIGURE 2. The dependence of the material inhomogeneity term Cinh according to distance
between crack tip and interface L, is plotted, for different low values of Jfar and weld material
properties.

The material inhomogeneity term exhibits a singularity at the interface, (see Fig. 2), causing
infinitely large crack driving force (for an OM/UM transition) or zero crack driving force (for an
UM/OM transition), when the crack reaches the interface. This leads to an accelerated crack
growth or a pop-in in the case of an OM/UM transition and a reduced crack growth rate, or even a
crack arrest for an UM/OM transition. Alternatively, if the interface is not oriented perpendicularly
to the nominal crack extension direction, a crack deflection may occur.

References
1. Simha, N.K., Fischer, F. D., Kolednik, O. and Chen, C. R., J. Mech. Phys. Solids, vol. 51,
209-240, 2003.
2. Simha, N.K., Predan, J., Kolednik, O., Fischer, F. D., and Shan, G.X., J. Mech. Phys. Solids,
submitted.
3. Kolednik, O., Predan, J., Shan, G.X., Simha, N.K. and Fischer, F. D., Int. J. Solids Struct, in
press.
 19. Structural Integrity Assessment in Theory and Practice 1051

STRENGTH RECOVERY OF MACHINED ALUMINA BY SELF CRACK

HEALING

Kotoji Ando1, Koji Takahashi1, Wataru Nakao1, Toshio Osada1 and Shinji Sato2
1Department of Safety & Engineering, Yokohama National University

79-5 Tokiwadai, Hodogaya-ku, Yokohama-shi, Kanagawa-ken, Japan

andokoto@ynu.ac.jp
2NHK spring Co. Ltd. , 3-10 Hukuura, Kanazawa-ku, Yokohama-shi, Kanagawa-ken, Japan

In this study, the effect of crack-heal on cracks propagated during machining was studied
systematically. From the obtained results, we proposed a new method for increase in reliability of
machined alumina by using the crack-healing treatment.
Ceramics have various excellent mechanical properties. Thus, they are expected to apply to
various fields. However they have low fracture toughness, so that cracks propagated during
machining reduce strength and reliability of machined components significantly. To overcome
these problems, it is a good approach to heal the surface cracks completely.
As an example, we investigated the crack-healing effect on the cracks on the bottom of
semicircular groove, which initiated during machining, as shown in fig. 1. Machining operations
were performed using a diamond drill with the cutting depth by a pass of 10 Pm. These machined
specimens were crack-healed in air at 1100 oC- 1400 oC for 1-10 h. The bending tests of the as-
machined and crack-healed specimens were also performed at R.T. Using stress concentration
factor of 1.4, we evaluated the local fracture stress at the bottom of semicircular groove.
Local fracture stress, which was reduced by machining, was found to be completely recovered
by crack-healing at 1400oC for 10 h as shown in fig. 2. Moreover, as shown in Fig. 3, the
characteristic strength of crack-healed specimens at 1400oC-10 h and the heat-treated smooth
specimens were 1026 MPa and 1065 MPa, respectively, where the heat-treated smooth specimens
was confirmed to have no defect on the surface by the previous study [1-2]. Thus, it was found that
the fracture stress of crack-healed specimens at 1400oC-10 h became almost equal to that of heat-
treated smooth specimen and that the cracks were completely healed by these conditions. From the
Weibull distribution, failure probability, F(x), of the crack-healed specimens at 1400 °C for 10 h
was found to be less than 1% at the stress of 600 MPa which 90% as-machined specimens
fractured. Therefore, it was concluded that the crack-healing was effective to improve the integrity
of machined ceramics.

Schematic diagrams of machined specimens

FIGURE 1. Schematic diagrams of machined specimens.

1052 K. Ando et al.

FIGURE 2. Effect of healing temperature on Local fracture stress

FIGURE 3. Effect of crack-healing on weibull distribution of local fracture stress

References
1. K. Takahashi, M. Yokouchi, S.K. Lee, K. Ando, J. Am. Ceram. Soc., 86-12, 2003
2. S.K. Lee, K. Takahashi, M. Yokouchi, H. Suenaga and K. Ando, J. Am. Ceram. Soc., 87-7,
2004
 19. Structural Integrity Assessment in Theory and Practice 1053

CRACK INITIATION AND GROWTH IN HAZ OF MICROALLOYED STEEL

K. Geric and S. Sedmak

Faculty of Technical Sciences, Trg D.Obradovica 6, Novi Sad, S&CG, gerick@uns.ns.ac.yu
Faculty of Technology and Metalurgy, Karnegieva 4, Beograd, S&CG
gerick@uns.ns.ac.yu

Due to welded joint microstructure heterogeneity it is very complex to perform a its toughness test,
because it is difficult to position notch tip in the region of lowest toughness, and it is also difficult
to control growing crack path.
Microstructure of heat-affected-zone (HAZ) region in which fatigue crack tip in fracture
toughness testing could be located depend on thermal cycle, defined by passes number and heat
input. In order to get more insight in microstructure properties it is possible to simulate welding
process by programmed heating and cooling of samples, and that test the specimens made from the
samples.
Failure behaviour of heat - affected - zone by simulated specimens and real welded joint is the
subject of many papers, e.g. Chae et al. [1]. Low toughness in coarse grain HAZ region (CGHAZ)
was attributed to local microstructure M-A constituent (high carbon martensite with some retained
austenite), Davis and King [2].
Three microalloyed steels had been selected for the experimental analysis of crack-opening
displacement (COD) as fracture mechanics parameter of simulated samples and real welded joints.
The microstructure of HAZ of tested steels had been simulated on Smitweld LS1402 device. The
samples 60 mm long, 11 mm wide and thick, were heated to different temperatures (1350, 1100,
950 and 850°C, selected as typical), and programmed cooled. Welded samples are produced by
manual arc welding as multipass, without preheating due to small sample thickness. In performed 4
passes an average heat input was 1,5 kJ/mm.
Three point bend specimen is used to measure COD at room temperature and at -30°C.
Measured COD at -30°C on simulated samples is presented in Fig. 1.a, as depended on heating
temperature of simulated samples, and in Fig. 1.b is presented COD obtained with specimens
produced of real welded joint for different distances of notch tip from weld metal. Growing crack
path in the specimen with clear pop-in in test diagram is presented in Fig. 2.a, and corresponding
fracture surface, exhibiting ductile and cleavage regions is presented in Fig. 2.b.
Obtained results, presented in other papers, Neves [3] showed that in similar steels HAZ
toughness of simulated samples and of real welded joint differ. The explanation for high toughness
level in real welded joint at small cooling time 't8/5, can be testing insensitivity on local
heterogeneity region in multipass welded joint. One can suppose that that testing is more sensitive
on microstructural effects in the case of simulated HAZ, compared to real welded joint. If this is
the case, results can confirm that reduction in toughness is cause by brittle microphases (such as
M-A constituents) in beinite microstructure of simulated coarse grain HAZ region , at short
cooling time. This phenomenon is observed only in simulated HAZ, and not in HAZ of real welded
joint, where total effect could be twofold. Due to overlapping of thermal cycles M-a constituent
can be disolved, and no local brittle zone can occur.
1054 K. Geric and S. Sedmak

FIGURE 1. COD value in simulated (a) and real welded joints (b) on -30°C

FIGURE 2. The crack path, and SEM fractography

References
1. Chae, D., Young C.J., Gotto D.M.and Koss, D.A., Materials Transaction A, vol. 32A, 2001-
2229, 2001
2. Davis L. and King J.E., Mater.Sci.Technology, vol. 3, 563-573, 1994
3. Neves, J. and Loureiro A., J. Materials Processing Technology, vol. 32A, 1-7, 2004
 19. Structural Integrity Assessment in Theory and Practice 1055

STRUCTURAL INTEGRITY AT ELEVATED TEMPERATURES - RESIDUAL

SERVICE LIFE EVALUATION

L. Milovic and S. Sedmak

Faculty of Technology and Metallurgy
Karnegijeva 4, 11000 Beograd, Serbia and Montenegro
acibulj@tmf.bg.ac.yu
Society for Structural Integrity and Life (DIVK)
Bulevar vojvode Misica 43, room 258, 11000 Beograd, Serbia and Montenegro
sedmak@divk.org.yu

The essence of creep deformation is that the plastic strain is time dependent and can occur at
stresses which in an ordinary tensile test would be below the yield stress. The mechanism of creep
relies on the thermally activated passage of atoms over an activation barrier which is characterized
by an activation free energy.
The mechanisms of creep are well established. However, practical creep-resistant materials
tend to be extremely complex, because a large number and severe design criteria. It is important to
understand the long-term creep and rupture behavior of such materials for their safe use in safety-
critical applications such as power plant and aeroengines. For this reason there exist a large
number of empirical or semi-empirical methods which will be consider in this paper, in order to
understand well existing experimental methods for determining the creep parameters C(t) and Ct,
and to analyze posibility the determine experimentally the parameter C*.
Components in large-scale steam generating power plants are manufactured in thick section of
creep resistant low alloy steels, being exposed to high temperature and high pressure. The design
of high temperature pressure components is governed by national and international standards and
allowable stresses are generally based upon mean 1x105 h creep rupture data with appropriate
safety factors to take account of data scatter and departure from design conditions. Before
predicting remnant creep life, laboratory generated creep rupture data are required. The experience
has shown that in many cases components exposed to creep could successfully use well beyound
the design life.
Regarding weldments in thick walled components, as the most common location for creep
failures, the life management has remained on regularly performed inspection regimes optimized
through plant experience and through the analytical and experimental activities, despite the fact
that the major advances have been achieved in the modelling of their behaviour under creep
conditions. Thanks to the computer finite element analysis using continuum damage constitutive
equations, it is possible now to predict the weld failure locations and even the times. However,
valid formal design and assessment codes rely on more established procedures such as semi
empirical weld strength reduction or efficiency factors, and reference stress techniques.
The possibility of experimental determination of C* through J integral of weldments of steel
produced for high temperatures application will be also considered in this paper.
1056 L. Milovic and S. Sedmak

References
1. Bhadeshia, H. K. D. H., Comprehensive Structural Integrity, vol. 5, 1-23, 2003
2. Saxena, A., Comprehensive Structural Integrity, vol. 5, 201-240, 2003
3. Gooch, D. J., Comprehensive Structural Integrity, vol. 5, 309-360, 2003
4. Dogan, B. et al., In Abstract Book of the International Conference on Fracture ICF11, Turin
(Italy), March 20-25, 2005, pp. 607 and 608, available also on the disc
5. Saxena, A., et al., Fracture Mechanics, vol. 24, 510-526, 1995
6. Saxena, A., et al., In Proceedings of the Life Extension and Assessment of fossil Power Plants,
edited by B. Dooley, R. Viswanathan, EPRI, California, 1987, 575-605.
7. Berkovic, M., et al., In Proceedings of the Fifth International Fracture Mechanics Summer
School, edited by S.Sedmak, EMAS, United Kingdom, 1989, 71-88.
 19. Structural Integrity Assessment in Theory and Practice 1057

THE ANALYSIS OF SUPPORTING STRUCTURE OF PLANETARY GEAR

BOX SATELLITE

M. Arsi, V. Aleksi and Z. Anelkovi1

Institute for Material Testing, Bulevar vojvode Misia 43, Belgrade, Serbia and Montenegro,
1Institute GOSA, Milana Rakia 35, Belgrade, Serbia and Montenegro

miodrag.arsic@institutims.co.yu

Existing dynamical models of supporting structure, mechanisms and gear driving of rotary
excavator, with regard to models of external load caused by excavation reaction, do not allow
complex introspection of their influence to dynamical behavior of rotary excavator. In addition to
that, influence of own low-frequency oscillations and non-stationary loading regime to working
strength of supporting structures and reliable assemblies can not entirely be perceived in phase of
design and construction.
Premature damages and fractures of gear box elements frequently occurs in praxis, which can
be explained by application of inadequate design and construction methods, by the ignorance of
basic material properties and welded joints and by mistakes in technology of elements
manufacturing.
Analysis of damaged and fractured parts supplies important informations for improvement of
gear boxes design methods, existing material properties and technologies of their treatment and
also for new materials and technologies development. Damage and fracture analysis enable also
development of new technical solutions and examination methods in prototype phase.
Review of performed theoretical and experimental analysis of satellite supporting welded
structure of planetary gear box used for driving of excavator working wheel SRs 1300.26/
5.0+VR±10, with fracture occurred, is presented through block-chart in Fig. 1.
Satellite support on the planetary part of the gear box for driving of excavator working wheel is
exposed to driving momentum and transversal bending forces. Theoretical stress state analysis of
the satellite support is performed for different cases of loads:
• load of the satellite support caused by the operating force,
• load of the satellite support caused by the own weight and alternating bending,
• load of the satellite support caused by the alternating bending.
Safety factor to fatigue fracture of SD = 1.49 has been calculated by the application of TGL
19340 and stress spectrum has been obtained for the case of pure alternating variable bending
taking into account influence of notch according Stiler’s method.
Calculation of safety from fatigue fracture has been obtained on the basis of TGL 19340.
Specific properties of cast materials have been taken into account according TGL 19341, and the
influence of notch according TGL 14915. Taking into consideration that all mentioned regulations
are related with stress concentration, correction of stresses obtained in the calculation of satellite
support model has been performed. Calculation has been performed by using of software
application called “AUTRA”.
1058 M. Arsi et al.

Figure 1. Review of performed theoretical and experimental analysis of satellite support

Experimental assessment of stress spectrum has been performed by the tensometric measuring
of strains on the driving shaft of excavator working wheel by using of four gauges of XY-120-
HBM type, which are suitable for measuring of strains caused by the torsion momentum on the
shaft.
Measuring results stated by normal strain H, have been recalculated to tangent stress W through
elasticity modulus E and the Poisson’s coefficient Q, which together with polar resistance
momentum of cross-section Wp determines torsion momentum T on the output shaft of transporter
driving, which have been utilized for assessment of gear box operating life on the basis of
experimentally determined stress spectrum.
Comparison of stress spectrums determined by theoretical considerations and experimental
examinations, in relation to carrying capacity of satellite support has shown that working stresses
in critical sections have been close to yield point even without taking stress concentration into
consideration.
Design solution of satellite support elements and results of experimental examinations of
parent materials and welded joints have shown that all significant parameters for the conditions of
variable load have not been including during designing and manufacturing of welded structure and
welded joints as critical points.
 19. Structural Integrity Assessment in Theory and Practice 1059

FAILURE PROBABILITY OF GEAR TEETH WEAR

M. Ognjanovic
Faculty of Mechanical Engineering, University of Belgrade
Kraljice Marije 16, 11120 Belgrade, Serbia
ognjen@EUnet.yu, mognjanovic@mas.bg.ac.yu

Gear load capacity calculation is defined according to pitting of the teeth flanks. In service
conditions, a failure process is combined of a number of damage processes. Which of them will be
dominant depends on design parameters, technological and exploitation conditions. Periodically,
for some of gears, extremely difficult service conditions exist, which creates a possibility for
progressive teeth wear. Results of progressive wear are obtained by experiments. For these results,
failure boundaries which can be used for parameters of Weibull's function definition, for different
stress levels and for different stress cycles numbers (teeth mesh revolutions) have been defined.
The first feature of a progressive wear process is stochastic behavior combined of a few
elementary wear processes. The second one is that in service life these very strong working
conditions are not continual. Periodical service conditions may be presented by the probability of
these conditions p. By combining the failure probability PR of progressive teeth wear and the
probability of service conditions p, it is possible to obtain the complex probability Fp=pPR which
defines the probability of progressive wear in service life.
The failure probability is the result of gear wear testing by using the FZG gear tester or another
similar system for gear loading and long time testing. Figure 1 presents the results of gear testing in
planetary gear drive tested in a back to back system similar to the FZG gear tester. The inferior
boundary of failure probability distribution is defined by the visible flank failure beginning (10%
failure). The superior boundary is defined by the thickness of the layer of teeth flank wear of 0,3 m
(m – gear module ). Gear teeth is surface hardened and the hardened layer is eliminated by a
progressive wear process. Some of the points are obtained by the testing and some of them are
defined by approximation. For a more precise definition, it is necessary to perform a number of
tests, which will be done in the future. The results presented are compared with the gear endurance
limits available in the DIN 3990 for the surface hardened and not hardened steels.
By using the lines of inferior and superior failure boundaries, it is possible to obtain Weibull's
functions of the failure probability PR. The function of the stress cycles number (teeth mesh
revolution) PR(N) can be defined for every level of the stress VHN . For every stress cycle number
N (teeth mesh revolution) it is possible to define the following function PR(VHN)
E E
§N · §V ·
 ¨¨ ¸¸  ¨¨ HN ¸¸
PR  N  1  e ©K ¹
PR V HN  1 e © K ¹
; (1)
The parameters of the Weibull's function K and E are defined by using a coordinate of the
points from the boundary lines which include the failure probability 0.1 and 0.9.
1060 M. Ognjanovic

References
1. Floding A., Andersson S.: Simulation of Mild Wear in Spur Gears, Wear, Vol.207, pp 16-23.,
1997.
2. Hohn B.R.: Modern Gear Calculation, - Proceedings of the International Conference on
Gears, VDI-Berichte 1665, pp 23-43., 2002
3. Podgornik B., Vižintin J.: Wear Reaistance of Plasma and Pulse Plasma Nitrided Gears, -
Proceedings of the International Conference on Gears, VDI-Berichte 1665, pp 593-601, 2002
4. Weck M., Hurachy-Schonwerth O., Bugiel Ch.: Service Behaviour of PVD-Coated Gearing
Lubricated with Biodegradable Synthetic Ester Oils, - Proceedings of the International
Conference on Gears, VDI-Berichte 1665, pp 677-690, 2002
5. Bauer J., Andersson S.: Simulation of wear in gears with flank interference – a mixed FE and
analytical approach, Wear, vol. 254, pp 1216-1232., 2003.
6. Ding Y., Rieger F.N.: Spalling formation mechanism for gears, Wear, vol. 254, pp 1307-
1317., 2003.
7. Aslantas K., Tasgetiren S.: A study of spur gears pitting formation and life prediction, Wear,
vol. 257, pp1167-1175, 2004.
 19. Structural Integrity Assessment in Theory and Practice 1061

SOME ASPECTS OF ENGINEERING APPROACH TO STRUCTURAL

INTEGRITY ASSESSMENT

M. Kiric and A. Sedmak

Ministry of science and environmental protection of Republic Serbia
Mechanical engineering faculty
Njegoseva 12, Belgrade, Kraljice Marije 16
kmiodrag@eunet.yu

The structural integrity of nuclear steam supply systems in the U.S.A. was assured by designs that
adhered to the ASME Boiler and Pressure Vessel Code and many other regulatory standards. The
requirements of these codes and standards are based on linear elastic fracture mechanics (LEFM)
concepts. In much or all of the working temperature regime of nuclear systems, power and
chemical plants, as well aircraft propulsion, the material is being stressed above the transition
temperature, where the fracture response is ductile and the material capable of considerable plastic
deformation.
The engineering approach, developed by EPRI, permits fracture analysis in the elastic-plastic
regime by assuming the concept of J-controlled crack growth. When the conditions for J-controlled
growth are satisfied, the JR curve will be independent of crack configuration and stable crack
growth can be analyzed by the JR curve approach based on J. A rising JR curve is normally
associated with growth and coalescence of microvoids. The initial part of JR curve, nearly vertical,
can be analytically represented in various ways. The JR curve dependence on crack growth is
approximated by the polynomial approximation in this paper.
The engineering approach provides simple interpolation formulae for J integral and crack
mouth opening displacement (CMOD) which superpose fully plastic solutions to linear elastic
solutions. Fully plastic solutions and analyses are applied to cracked configurations which are
completely yielded. In small-scale yielding, the plastic contribution is small compared to the
elastic contribution and in other extreme in the fully plastic range, the plastic contribution is the
dominant term >1@.
In other words, the plastic component of J is negligible at low loads, but dominates at high
loads >2@. It is considered a cracked cylinder with a long axial crack on internal surface. In this
paper it is shown that some influencies should be taken into account. The contributions are
calculated and their dependence on strain hardening exponent (n) and cylinder geometry are
illustrated. The ratio of elastic J integral to the total J integral is less than 35% only at internal
pressure equal to or greater than about 50 Mpa, dependent on relative crack depth for assumed
cylinder corresponding to the size of a nuclear reactor pressure vessel. The ratio of crack depth to
wall thickness (a/H) has a strong inluence on similar ratio of elastic CMOD and the total CMOD.
The ratio of wall thickness to internal radius of cylinder (H/Ri) has opposite influences on the ratio
of elastic component to total value: its increasing, causes decreasing of the J values ratio, while in
the case of the ratio of CMOD values, it leads to their steep inrease. The partition of elastic and
fully plastic components for cylinder, can be analyzed in terms of the internal pressure for which
these components of J and CMOD are equal, Kiri >3@. The pressure value is dependent of crack
depth, geometry and n. The paper gives the influencies of some geometric parameters, (a/H) and
(H/Ri), on J and CMOD for assumed cylinder geometry, too. It is derived the parabolic relation
between J and CMOD. The relation is dependent of crack depth, H/Ri and n. Only for a/H=1/8, H/
Ri=1/20 and n=3 the dependence is polynomial of the third order. Therefore it is concluded that it
1062 M. Kiric and A. Sedmak

is possible, with some restrictions, to obtain J integral values from measurement of CMOD, as it
was proposed for cracked tensile plate. The number of data pairs increases with polynomial order.
To obtain predictions for crack initiation, stable growth and structure instability, the
engineering approach compares the crack driving force (CDF) in terms of applied J to the JR curve
in the CDF diagram. The J curve, tangent to the JR curve determines the point of instability given
by the pressure and the amount of crack growth at instability. These values are solutions to two
equations for the stable crack growth. They are found by numerical iterative calculation in this
paper. It is a convenient procedure because it quickly provides the conditions for instability of
various cylinder geometries and initial crack depths.
As an example was considered a cracked pressure vessel made of HSLA steel with an internal
axial crack. The analysis can be performed by using the formula for the hoop stress and the failure
assessment diagram. The estimation of internal pressure boundary value by using the condition
that the material begins to yield on inside surface of cylinder, provides too conservative value. It is
assumed a conservative approach that the crack is open to the internal surface and being part-
through thickness. Thus the engineering approach was applied and an analysis was performed. The
stable crack growth was calculated as well the full load carying cylinder´s capability.

References
1. V. Kumar, M.D. German, C.F. Shih, An engineering approach for elastic-plastic fracture
analysis, NP-1931, Research Project 1237-1, General Electric Company, New York 1981, p.
2-13.
2. T. L. Anderson, Fracture mechanics - fundamentals and applications, CRC press Inc. 1995.
Texas, p. 478.
3. Kiric, B.M., The Ph.D. thesis, The Faculty for mechanical engineering, Belgrade, 2000., p.
109.
 19. Structural Integrity Assessment in Theory and Practice 1063

STRUCTURAL INTEGRITY ASSESSMENT APPLYING ULTRASONIC

TESTING

M. Kiric
Ministry of science and environmental protection of Republic Serbia, Belgrade
kmiodrag@EUnet.yu

Nondestructive testing of materials and welded joints for detection and evaluation of cracks, as the
most dangereous discontinuities, includes different methods, like :
• Liquid penetrant testing (PT),
• Radiographic testing (RT),
• Electromagnetic (eddy current) testing (ET)
• Magnetic particle testing (MT)
• Ultrasonic testing (UT).
Each testing method has its features, thus the selection depends on the scope and range of
products and materials to be tested. Not all of them are suitable to evaluate cracks. This is the
reason why the UT is the subject of the consideration, taking in mind that there are supplementary
methods for detection of surface cracks like MT. The capabilities of UT for cracks detection are
discussed.
The experiences gained in pressure vessels testing have shown that UT by using pulse-echo
technique is the best method for discontinuities characterization because of its versatility and
possibility of quantification and memorization of results enabled by digital ultrasonic flaw
detectors.
The selection of UT should define the technique to be applied, also its sensitivity and
reliability. The initial cracks and other discontinuities, made during fabrication, may be not
detected before product use, but cracks can grow in time, dependent on service conditions, and be
detected only after a period of product service life. If such discontinuities are detected, two further
steps are possible, depending on crack size. The first is to leave them in the structure, but with their
repeat evaluation after given time intervals. The determination of crack features may include: basic
ultrasonic parameters, crack parameters and its proximity to the surface or to other discontinuities
as well its basic shape and orientation. The second measure is the complex task to assess the
structure integrity. For crack parameters evaluation by UT, estimated stress field and possible failure
type (brittle or ductile fracture, fatigue and stress corrosion) are to be taken into account.
The paper considers UT of welded joints in constructions, from the European standards point
of view. It is a multitask UT because it is aimed at detection of discontinuities, their
characterization and evaluation, by taking into account the geometry and welding technology.
Characterization of discontinuities involves the determination of those features which are
necessary for its evaluation with respect to known acceptable criteria. In order to correctly identify
the discontinuity types specified in the acceptance criteria, or to make a correct fitness-for-purpose
evaluation, it may be necessary to make a more detailed sizing and assessment of the shape of the
discontinuity. The attention is given to the classification of discontinuities as planar or non-planar
and criteria for the classification. One of the criteria considered are echodynamic patterns obtained
as transient echo shapes during probe movement. The difference between the quality control
concept and the fitness-for-purpose concept is made. Some examples are given to illustrate the
standard approaches for UT.
1064 M. Kiric

Evaluation of indications in accordance with standard, performed for detection by testings of

welded joints, is based on reference levels, established by european standard EN 1712. For method
1, the setting of sensitivity may be performed by 3 mm diameter side drilled holes. The diameter is
the same for different frequencies and parent material thickness, while for method 2, the distance
gain size system, ultrasonic probe frequencies and reference levels depend on parent material
thickness. On the other hand, evaluation level is determined by reference level. It is important from
two reasons: all indications equal to or exceeding evaluation level shall be evaluated and the other
is that the standard technique measures the lateral dimension of an indication over which the echo
is equal to or greater than the evaluation level. Similar question can be put regarding the reference
level for tandem examination. It is defined by 6 mm diameter side drilled hole for all thicknesses.
The evaluation of indications by using method 3 of sensitivity setting is discussed, too. It is
performed by using a rectangular notch with the depth of 1 mm if a probe angle is greater or equal
to 70o and for the thickness range from 8 mm to 15 mm. Evaluation level is defined via the
reference level regardless parent metal thickness and ultrasonic probe frequency. Because the
resolution power of a probe depends on its ultrasonic frequency, the detection of the notch with the
depth of 1 mm may be discutable. This question is considered from the point of double reflexion
from the right corner and transverse waves transformation.
The acceptance or rejection of welded joints with cracks is solved in two ways: crack are
unacceptable, or they are conditionally acceptable. The acceptance levels given by standards are
dependent on methods of sensitivity settings and thickness interval. They are analysed in respect to
the fracture mechanics relations derived by using stress intensity factor for internal cracks.
Acceptance levels allow greater relative echo amplitude for short indications, while for long
indications relative echo aplitude is lowered, as expected. The level number is maximum three,
while it could be greater, as it can be concluded from linear elastic fracture mechanics. As far as
cracks are considered, it seems reasonable to apply small number of acceptance levels and linear
elastic fracture mechanics, because UT is simplified in this way. Greater number of level would
requires a softver for acceptance/rejection of discontinuities in welded joints. This softver will
incorporate fracture mechanics criteria for evaluation of cracks and other discontinuities detected
by UT.
 19. Structural Integrity Assessment in Theory and Practice 1065

(CRACK-HEALING + PROOF-TEST): METHODOLOGY TO GUARANTEE

THE RELIABILITY OF CERAMICS

Masato Ono, Wataru Nakao, Koji Takahashi, Kotoji Ando and Masahiko Nakatani1
Department of Material science & Engineering, Yokohama National University
79-5 Tokiwadai, Hodogaya-ku, Yokohama, 240-8501, Japan
Fax: 81-45-339-4024, d05sa401@ynu.ac.jp (Masato Ono)
1NHK SPRING Co., Ltd, Fukuura, Kanazawa-ku, Yokohama, 236-0004, Japan

Structural ceramics have excellent heat corrosion and wear resistance. However, fracture
toughness is low. Thus, they are brittle and sensitive to flaws. The flaws in ceramics are classified
into propagated surface cracks in machining and embedded flaws.
Ando et al. have progressed in development of the structural ceramics attached crack-healing
ability by using oxidation of SiC. When ceramics admixed with SiC are kept in air at high
temperature, SiC located on the crack surface reacts with O2 in air. Then, crack is completely
restored by the products and the heat of the reaction. Moreover, the restored part is mechanically
stronger than the other parts. However, the embedded flaws cannot be healed.
Therefore, the present authors expect the combination of crack-healing and proof-test as a new
methodology to guarantee of the reliability of a ceramics component. For this purpose, the
minimum guarantee fracture, VGT, stress at high temperature for the specimens proof-tested at R.T.
after crack-healing was estimated by using the measured temperature dependence of fracture stress
and toughness. The fractures stress of the proof-tested specimens was also measured at elevated
temperatures. From the obtained results, the validity of the estimated minimum fracture stress was
discussed.
The mixture of alumina powder and 20 mass% SiC powder was blended well in alcohol for 48
h. Then, the slurry was dried. Rectangular plates of 90 mm x 90 mm x 7 mm of alumina/SiC
particle composite were sintered in N2 gas for 2 h via hot press under 35 MPa at 1923 K. The
sintered plates were cut into the 3 mm x 4 mm x 36 mm rectangular specimens bar. This study
includes investigation of effect of proof test on the embedded flaws as one of the purposes. Thus,
many embedded flaws were consciously introduced to the test specimens.
A semi-elliptical surface crack of 100 Pm in surface length was made at the center of the
tensile surface of specimens with a Vickers indenter. The ratio of depth (a) to half the surface
length (c) of the crack (aspect ratio) was a/c = 0.9. The specimens were crack-healed at 1573 K for
1 h in air. Crack-healed specimens, that are crack-healed above condition, were subjected to the
proof-test at R.T. The proof test stress was 435 MPa.
In the previous study, Ando et al. have proposed the estimation method of VGT [1]. In most
ceramics, the fracture strength has negative temperature dependence. Thus, the minimum
guarantee fracture stress at high temperature, VGT, can be lower than proof test stress at R.T. VGT
can be expressed as following,

2 2 1
2V 0T ­°§ K T · § V 0R · °­ § SV p
R
· °½ ½°
V T
arccos ®¨¨ ICR ¸¸ ¨¨ T ¸¸ ®sec ¨ ¸  1¾  1¾
G
S ¨ R ¸ °
°̄© K IC ¹ ©V0 ¹ °̄ © 2V 0 ¹ ¿ °¿ (1)
1066 M. Ono et al.

where KICT and V0T indicate the plane strain fracture toughness and the fracture stress of plane
specimen at high temperature, respectively. For the estimation of VGT, it is necessary to obtain the
temperature dependences of KICT and V0T.

Figure 1 shows the comparison between VGT and experimental results as a function of test
temperature. The open and closed diamonds indicate the fracture stress of the crack-healed
specimens and the proof-tested specimens, respectively. The minimum values of the measured
fracture stress of the proof-tested specimens at several temperatures are smaller than the proof test
stress at R.T. In shorthand, one can observe the temperature dependence of the fracture stress of
the proof-tested specimen. Double circle in Fig. 1 indicates the estimated value of the minimum
guarantee fracture stress. All specimens except one tested at 1373 K fractured with higher strength
than VGT at the same temperature. Thus, this estimation of VGT was accepted as correct. From the
obtained results, it is concluded that the combination of crack-healing and proof-test is the
effective techniques to guarantee of the fracture stress at high temperature of the ceramics
component.

FIGURE 1. The comparison between minimum guarantee fracture stress and experimental results
as a function of test temperature.

Reference
1. K. Ando, Y. Shirai, M. Nakatani, Y. Kobayashi and S. Sato, J. Euro. Ceram. Soc., 22, 121-
128, 2002.
 19. Structural Integrity Assessment in Theory and Practice 1067

RISK BASED INTEGRITY ASSESSMENT OF CONCRETE STRUCTURES

M. Pavisic
PM Lucas Enterprises Ltd. (Cyprus)
Belgrade, Internacionalnih brigada 25a, Serbia and Montenegro
m.pavisic@pmlucas.com

From the very beginning of its service life the concrete structure is exposed to the various loading
conditions and excessive and adverse events – hazards (overloading, fire, explosions, floods,
earthquakes and so on). As a consequence, the overall integrity of structure is jeopardized and
some structural limit states could be reached.
Risk is a probability that a particular adverse event (limit states) occurs during a stated period
of time multiplied by the consequences.
Damage initiation and development in concrete structures is time dependant process
determined by the many factors as: aggressive environment, quality of the construction, loading
history and the quality of maintenance. In the same time, damage progress affects the structural
resistance, safety and reliability decreasing them during the time (Pavisic [1]).
Having in mind mentioned, the right question concerning the concrete structure integrity
assessment is: what degree of structural damage may be permitted and tolerable with no taking any
protective actions to the structure or what is the limit of acceptable damage? The response to this
question is very important not only from the point of view of the structural integrity assessment,
but for making decision about the needs for structural repair and its remaining service life
prediction (Fig. 1.).

FIGURE 1. Damage function

Where is:
D – damage (deterioration)
T – time
D0 ( t0 ) – damage (deterioration) acceptibility limit
D1 ( t1 ) – limit state
Ts – predicted structural service life
TsTse – extended structure service life
1068 M. Pavisic

t0 – time when acceptable damage limit is overcome

t1 – time when repair has to be carried out
D0D1 – integrity assessment (inspecting, monitoring, testing)
In fact, integrity assessment is an evaluation of the remaining structural resistance taking in
account damage created and identified on the structure.
The basic equation of the structure limit state is:

g RS t0 (1)
where, R is structural resistance and S loading, both time dependant functions:

R R(D,t) , S S (t ) (2)
It is already well known that the parameters characterizing the structural properties and degree
of structural damage are uncertain and random values by its nature and may be defined as the
statistical variables having its probability density distribution (Hart [2]). Thus, we may count on
the probability of some limit states to take place on the structure after some time being in service.
The probability may be expressed by the following equation (Stewart and Rosowsky [3]):

f
Pf t  Pr >R d S @ Pr [ R  S d 0] Pr [ g ( R, S ) d 0] ³f R (t ) f S (t )dt
0
(3)
where is:
g(R,S)- limit state function
fR(t) – probability density distribution of resistance
fS(t) – probability density distribution of loading
In the paper, the overall procedure, from integrity assessment and risk analysis, to the
calculation of probability of limit state, is presented and the basic elements for decision making
discussed.

References
1. Pavisic, M., In Proceedings of 7th Internationl Conference on Mechanical Behaviour of
Materials, Hague, The Netherlands, 1995.
2. Hart, G.C., Uncertainty Analysis, Loads, and Safety in Structural Engineering, Prentice-Hall,
Englewood Cliffs, New Jersey, 1982.
3. Stewart, G.M, Rosowsky, V.D., J. Struct. Safety, vol. 20, 91-109, 1998.
 19. Structural Integrity Assessment in Theory and Practice 1069

STRUCTURAL INTEGRITY ASSESSMENT BY LOCAL APPROACH TO

FRACTURE

M. Zrilic, M. Rakin, Z. Cvijovic, A. Sedmak1 and S. Sedmak2

Faculty of Technology and Metallurgy,Karnegijeva 4, 11000 Belgrade, Serbia & Montenegro
1Mechanical Engineering Faculty, Kraljice Marije 16, 11200 Belgrade, Serbia & Montenegro
2Society for Structural Integrity and Life (DIVK), Bulevar vojvode Misica 43, 11000 Belgrade,

Serbia & Montenegro

zrilic@eunet.yu

Local approach to fracture has been developed for complete understanding of fracture mechanism
[1,2], including the material degradation process. This approach combines theoretical,
experimental and numerical solutions, enabling less conservative assessment of crack significance
and residual stress.
Steam lines, in service exposed to high temperature and high pressure, has been designed for a
life of 100 000 hours. In order to assess residual life, extended experimental investigation of
samples, taken from virgin 14MoV6 3 steel, Fig. 1 (left), not used for steam line construction, and
an old steel of the same class, taken after exhausted nominal life, after 117 000 operating hours [3].
Five specimens of virgin and of used steel for three geometries (notch root radius 10 mm; 4
mm and 2 mm) and each geometry had been tested for determination of critical void growth rate R/
R0 values and void growth rate R/R0, corresponding to ultimate tensile strength. Critical value of
void growth rate, corresponding to final fracture, is designed by subscript c, and the value of void
growth rate, corresponding to maximum loading, e.g. to ultimate tensile strength, is designed by
subscript m.
Tearing fracture (ductile fracture, nucleation and coalescence of voids) have been analysed in
local approach. For that, mechanical properties and fracture mechanics parameters are necessary.
Substantial link in local approach is numerical simulation by finite elements method (FEM),
applied through elastic-plastic analysis for ductile fracture solution using 3D elements of an
axisymetric model.
FEM calculation is performed by elastic-plastic analysis in NASTRAN software, using 3D
elements [4]. Upper specimen half is modeled as a wedge element, with angle of 5o, corresponding
to 1/72 part of upper specimen part, Fig. 2 (right). This affects the shape of finite elements mesh,
producing wedge type elements around axes, and the other elements are of brick type.
Calculated local criterion is material characteristic, representing local stress state, which
corresponds to critical void growth and next unstable fracture. This material characteristic is a
convenient form, directly applicable in FEM calculation. That means, when high stress level is
found in a structure and void growth rate R/R0 is verified, at the critical (R/R0)c level, failure of
structure can be expected. From engineering point of view this verification should be performed
for maximum load, e.g. for ultimate tensile strength, (R/R0)m [5]. This should be the indication for
critical behaviour of a structure.
1070 M. Zrilic et al.

FIGURE 1. Notched tensile specimen (left), notched tensile specimen mesh (right).

References
1. ESIS P1-92, Recommendations for Determining the Fracture Resistance of Ductile Materials,
ESIS, 1992.
2. Rakin, M., at al, Strength of Materials, Kluwer online, vol. 36, 1, 33-36, 2004.
3. Zrilic, M., Aleksic, R., Thermal Science - International Journal, vol. 7, 1, 35-49, 2003.
4. Zrilic, M., Burzic, Z., Cvijovic, Z., Strength of Materials, Kluwer online, vol. 36, 1, 47-58,
2004.
5. Zrilic, M., Doctorate of Philosophy in Metallurgy Science, Belgrade, Serbia & Montenegro,
2004.
 19. Structural Integrity Assessment in Theory and Practice 1071

BRITTLE AND DUCTILE FRACTURE IN SERVICE OF PRESSURE VESSELS

N. Filipovic and K. Geric

Zavod za zavarivanje – Beograd
Fakultet tehnickih nauka –Novi Sad
zzz@bitsyu.net

Underground storage tank for liquefied natural gas, 9.96 m3 in volume, fractured during pressure
proof test by liquid nitrogen (Fig. 1a). Cracks initiated in circumferential welded joints in lower
part of the vessel mantle. Direction of crack growth is perpendicular to circumferential welded
joints and parallel to vessel axis, in the direction of maximum tensile stress (Fig. 1b).
The analysis of fractured surface on the left and on the right side along the heat-affected-zone
(HAZ) of circumferential welded joint revealed that crack surface is typical for brittle fracture,
normal to the plate, with coarse grains and typical V marks, directed to the initiation point.
Vessel mantle was produced of structural steel S355J2G3 according to EN10025. Although
microstructure shows expressed lamination and secondary structure, impact testing results in
rolling and cross rolling direction at -20C, obtained from two plates have not exhibited significant
difference (average energy values of 34 J and 39 J), with individual values greater for specimens in
longitudinal direction. Impact energy for HAZ and weld metal is 20 to 25% lower compared to
parent metal, but all of them are well above specified values in standard (13,5 J for KV150/5).
Liquid nitrogen during pressurizing produced the temperature bellow nil ductility transition
temperature of the steel and welded joint, which caused brittle fracture
In the mobile storage tank for transportation of ammonia, due to overfilling for about 20 kg and
steadily increasing temperature (month of July) the inner was completely filled by liquid phase,
resulting in increase of pressure, which caused plastic deformation, bulging, wall thickness
reduction in this position from 8 to 5.3 mm, and to final failure of tank (Fig. 2a).
Crack initiated in the heat-affected-zone (HAZ) of fillet weld between mantle plate and
support reinforcement, because plastic deformation was constrained in this position (Fig. 2b).
Initial cause of cracking was the undercut 1 mm deep, at the beginning of fillet weld, which
introduced local stress concentration Crack developed in brittle manner through HAZ, containing
microstructural interphases of quenching with hardness of 320 HV10 up to 5 mm in depth, and
then through reduced wall thickness of parent metal by shearing.
Applied steel Nioval 470 MPa of nominal yield strength 470 MPa exhibited satisfactory
impact toughness (120 J to 140 J at OC), what explains crack growth through HAZ of
circumferential welded joint with coarse grain microstructure and increased hardness of 260HV10
/1/, and not through parent metal.
1072 N. Filipovic and K. Geric

FIGURE 1. Brittle fracture of pessure vessel FIGURE 2. Plastic collapse of pressu

References
1. Katarina Geric, Cracks in welded joints (in Serbian), Faculty of Technical Sciences, Novi
Sad, 2005.
 19. Structural Integrity Assessment in Theory and Practice 1073

MECHANISMS OF FRACTURE IN MEDIUM CARBON VANADIUM

MICROALLOYED STEELS

N. Radovic, Dj. Drobnjak and H. Hraam

Department of Metallurgical Engineering, Faculty of Technology and Metallurgy,
University of Belgrade, PO BOX 35-03, 11120 Belgrade, Serbia&Montenegro
nenrad@tmf.bg.ac.yu

The relationship between microstructural parameters and cleavage fracture has been studied in
(950-1300°C) air cooled vanadium microalloyed steels by means of impact testing, light
microscopy and scanning electron microscopy. Large variations in impact energy are obtained as a
function of test temperature and microstructure. The results show that accicular ferrite (AF) and a
multiphase structure consisting of ferrite-perlite (FP) and 30-70% AF posses room temperature
toughness superior to that of classical bainitic sheaves (BS) as revealed by impact energy level.
However, AF is superior to FP-AF in term of energy transition temperature.
At liquid nitrogen temperature, all steel grades show similar behavior. Transgranular cleavage
is exclusive mode of fracture. Primary brittle nuclei, which control the critical tensile strength for
fracture, F, are found to be brittle TiN particles of diameter >2m. Large TiN particles are friendly,
because the cracks are initiated by stress much smaller than that required for crack propagation.
This means that cracks ill become blunted out, what will make them inactive before the stress for
crack propagation is achieved. Carbides and martensite/austenite/carbide (MAC) constituent are
tentatively identified as secondary brittle fracture nuclei, but they are of less significance. As the
critical stage of brittle fracture is crack propagation through particle/matrix interface, the
morphology and ferrite grain size play little role. Calculation of maximum tensile stress below the
notch, max, and critical tensile strength for fracture, F, under the assumption that diameter of TiN
particle, which are the primary nucli, are equal to penny shaped crack size, have shown that the
requirements for crack propagation through particle/matrix interface, max > F, is satisfied from the
beginning of fracture process at liquid nitrogen temperature in all steels studied in this work.
Room temperature behavior of these steels is considerably different. The dominant fracture
mechanism is still transgranular cleavage, but this is preceded by a lower (BS steels) or higher (AF
steels) degree of ductile fracture. Calculation have shown that at the beginning of fracture , max < F.
This means that the brittle crack initiated on brittle particles can not propagate; instead, the ductile
crack will be initiated and propagated. During propagation, ductile crack is assumed to accelerate,
what, in turn, increases the strain rate, and consequently, Y and max. At a critical ductile crack
length, critical condition, max > F , is achieved, and brittle cracks, initiated in brittle particles ahead
of the ductile crack tip are activated. These propagate across particle/matrix interface and cause
fracture. Higher toughness of steels with AF structure requires a longer ductility crack to be
formed, than in steels with BS structure, before requirement for brittle fracture, max = Y· n > F, is
attained, because the strain hardening exponent (n) of the former steel is much lower and product
Y· n is smaller in spite of Y is higher. The overall contribution of ductile fracture to the toughness is
relatively small, because the shear decohesion mechanism, which dominates ductile fracture in
both steels, is characterized by a low expenditure of energy. In spite of TiN inclusions are present,
the primary nuclei are assumed to be carbides, smaller than 1µm. The TiN-cracks are blunted out
before the conditions for cleavage are attained. This can be ascribed to influence of a large plastic
zone which is produced ahead of the ductile crack. In addition to being more resistant to brittle
crack propagation across particle/matrix interface, the steels with AF structure show the
susceptibility to cracks being arrested at the grain boundaries, presumably AF plate boundaries.
1074 N. Radovic et al.

This feature is observed only in steels with AF and not with BS structure, providing thus additional
barriers to crack propagation in former.

References
1. Drobnjak, Dj. and Koprivica, A., Proceedings of theInt.ernational Conference Fundamentals
and Applications of Microalloyed forging Steels, edited by C.V.Tyne, G.Krauss and
D.K.Matlock, TMS, Warrendale PA, 1996, 93-106.
2. Bhadeshia, H.K.D.W., Bainite in Steels, Institute of Materials, London, England 1992.
3. Hraam, H., Mechanism of Brittle Fracture in Medium Carbon V-microalloyed Steels, MSc
Thesis, Faculty of Technology and Metallurgy, University of Belgrade, 1998.
 19. Structural Integrity Assessment in Theory and Practice 1075

COMPUTATION AND EXPERIMENTAL INVESTIGATIONS OF NOTCHED

COMPONENTS FATIGUE LIFE ESTIMATION

S. Maksimovic, Z. Burzic and K. Maksimovic

VTI- Aeronautical Institute, Kataniceva 15, 11000 Belgrade, Serbia and Montenegro
s.maksimovic@net.yu

This work considers the analytical methods and procedures for obtaining the stress intensity factors
and for predicting the fatigue crack growth life for cracks in notched structural components. Many
failures in aircraft structures are due to fatigue cracks initiating and developing from fastener holes
at which there is large stress concentration. Stress intensity factor (SIF) solutions are required for
fracture strength and residual fatigue life assessment for defects in structures, or for damage
tolerance analysis [1] recommended to be performed at the stage of aerospace structure design.
Many efforts have been made during the past two decade to evaluate the stress intensity factor for
corner cracks and for through-the-thickness cracks emanating from fastener holes. A variety of
methods have been used to estimate the SIF values, such as approximate analytical analysis, finite
element (FE), finite element alternating, weight function, photoelasticity and fatigue tests. In this
paper the analytic methods and procedures for obtaining the SIF and predicting the fatigue crack
growth life for cracks at attachment lugs. Single through crack and single corner cracks in the
attachment lug analysis are considered. For this purpose analytic expressions [2] are evaluated for
SIF of cracked lug structures. For validation of the analytic stress intensity factors FEM with 3D
cracked finite elements is used. Typical finite element model of cracked lug is presented in Fig. 1.

FIGURE 1. Finite Element Model of cracked lug with stress distribution

The stress intensity factors of cracked lugs, analytic and finite elements, for through-the-
thickness cracks are shown in Table 1. Analytic results are obtained using relations derived in this
paper. Good agreement between finite element and analytic results is obtained. It is very important
because we can to use analytic derived expresions in crack growth analyses.
1076 S. Maksimovic et al.

TABLE 1. Comparisons analytic and FE results for SIF, KI

Lug No.
a >mm@ K IMKE
max
K IANAL
max
.

2 5.00 68.784 65.621

6 5.33 68.124 70.246
7 4.16 94.72 93.64

FIGURE 2. Analytic and test results for crack propagation in lug 2 – H = 44.4 mm; kt = 2.8

Figure 2 shows a comparison between the experimentally determined crack propagation curves
and the load cycles calculates to Elber law [3] for several crack lengths. Relatively close agreement
between test and presented computation results is obtained. The analytic computation methods
presented in this work can to satisfy requirements for damage tolerance analyses of notched
structural components such as lugs-type joints.

References
1. MIL-A-83444, Airplane Damage Tolerance Requirements.
2. Maksimovic, K., Nikolic-Stanojevic, V., and Maksimovic, S., Efficient Computation Method
in Fatigue Life Estimation of Damaged Structural Components, FACTA UNIVERSITATIS,
Vol. 4, No. 16, 2004.
3. Walker, K., The effect of stress ratio during crack propagation and fatigue for 2024T3
Aluminum, ASTM STP 462, Effects of environment and complex loading history on fatigue
life, 1970.
 19. Structural Integrity Assessment in Theory and Practice 1077

FAILURE ANALYSIS OF LAYERED COMPOSITE STRUCTURES:

COMPUTATION AND EXPERIMENTAL INVESTIGATION

S. Maksimovic
VTI- Aeronautical Institute, Kataniceva 15, 11000 Belgrade, Serbia and Montenegro
s.maksimovic@net.yu

A geometric nonlinear finite element method based on the von Karman - High Order Shear
Deformation Theory (HOST) is used to study the first-ply failure and the postbuckling behavior of
laminated type composite structures. For this purpose and for the investigation of the failure
responses improved 4-node layered shell finite elements are used. The finite element formulation
is based on the third order shear deformation theory with four-node shell finite elements having
eight degrees of freedom per node. The first-ply failure of laminates and the onset of delaminating
process in first-ply failure computation are some of the features of geometric nonlinear
formulation. The load-displacement curves for different types of graphite/epoxy laminates are
obtained. Stresses are computed in order to determine the first-ply failure of the mentioned axially
compressed laminated composite structure based on the maximal strain failure criterion. In this
procedure postbuckling and failure behavior of axially compressed flat and curved composite
panels are investigated numerically and experimentally. Computational results using linear and
geometrically nonlinear analyzes are compared with experiments. The effects of stacking
sequences on initial failure load are investigated. The HOST used here assumed the parabolic
distribution of the transverse shear stresses across the laminate thickness. The displacement field
for the parabolic transverse shear deformation through the shell thickness is

2
ª ww 4§ z · § ww ·º
u1 x, y, z  u  z« a  b<1  c ¨ ¸ ¨ <1  ¸»
«¬ wx 3©h¹ © wx ¹»¼
2
ª ww 4§ z · § ww ·º
u2  x, y, z  v  z« a b<2 c ¨ ¸ ¨ <2  ¸» u3x, y, z  w
¬« wy 3© h¹ © wx ¹¼»
(1)

FIGURE 1. Axially compressed curved composite panel.

1078 S. Maksimovic

FIGURE 2 – Postbuckling response. Axial load – radial displacement (P-w).

The buckling, postbuckling and failure behavior of axially compressed flat and curved
composite panels, Fig. 1, are investigated numerically and by experimentally obtained results,
which are presented in Fig. 2 and Table 1.
The buckling and postbuckling behavior of axially compressed composite panels is
investigated. The results of finite element analyzes are compared with experimental results. A
good agreement between numerical and experimental results is obtained. The comparison of
numerically obtained results with the experimental data shows that the proposed improved 4-node
shell finite element can be successfully applied for the prebuckling, buckling and postbuckling
analysis as well as for the initial failure analysis and for the prediction of the location of initial
failure analysis of compressed layered composite structures.

TABLE 1. Experimental and FE results of axial compressed curved pane

l
ucr–critical axial displacement (correspond to Pcr); Pcr,l–buckling load (linear “eigen” analysis);
Pcr,n–buckling load (geometrical nonlinear analysis); Pf-applied load at failure for maximum
deformation

References
1. Reddy,J.N.,(1984)Energy and Variation Methods in Applied Mechanics,John Wiley,New
York
2. Maksimovic, S., (1994) Improved geometrically nonlinear finite element analysis and failure
of fiber reinforced composite laminates, Proc. Int. Symp. Brittle Matrix Composites 4,
Warsaw, Sept. 13-15, Eds. A.M. Brandt, V.C. Li and I.H. Marshall, Woodhead Publ.
3. Norman, F., Knight Jr., James, H., Starnes Jr., and William Allen Waters, Jr, (1986)
Postbuckling behavior of selected graphite-epoxy cylindrical panels in axial compression,
27th Structures, Structural dynamics and material conference, San Antonio.
 19. Structural Integrity Assessment in Theory and Practice 1079

LOADING RATE EFFECT ON HSLA STEEL WELDED JOINTS FRACTURE

RESISTANCE

V. Grabulov, I. Blai, A. Radovi1 and S. Sedmak1

Military Technological Institute, Belgrade, Serbia and Montenegro
1Faculty of Technology and Metallurgy, Belgrade, Serbia and Montenegro

venciaiv@eunet.yu

One of the most important requirements for service safety of welded structures, produced of high
strength steel, is to achieve corresponding level of toughness in all three weldment constituents :
base metal (), weld metal (WM) and heat-affected-zone (). The evaluation of weldment toughness
is very complex, because of microstructures heterogeneity and mechanical properties of WM and .
Charpy test, although very old method is generally accepted for the evaluation of the impact
toughness. Developed instrumentation of Charpy test significantly extended its capacity, enabling
not only the separation of energy portions required for crack initiation and crack propagation, but
also the evaluation of loading level during the fracture process. Specifications for heavy loaded
welded structures normally include impact energy values for and WM, and transition temperature
for service at low temperature. However, there is still the problem how to evaluate toughness of ,
since it is difficult to position notch root precisely in region of lowest toughness.
In order to establish more severe testing loading, explosion crack starter test had been
introduced [1]. Fast loading rate and notched brittle bead, welded n the plate specimen assured
severe testing conditions. Applied to welded joint specimens [2], this test enables to determine the
most critical region for fracture in weldment.
Further improvement in crack resistance testing is offered by fracture mechanics tests, that
involved pre-cracked specimens. The application to welded joints allows for convenient
determination of crack resistance of BM and WM, but it is followed again by uncertainty in
defining of critical crack tip position in HAZ [3], since in prescribed preparing method fatigue
crack would follow the path of notch root rather than direction of critical HAZ region.
The application of all three above described methods for the evaluation of brittle behaviour of
welded joints produced by manual arc welding, performed of high strength armour, ship buildings
and pressure vessels steels by manual arc welding, are presented in the paper: HY 100 type steel
(A), an armour steel (B) and SUMITEN 80P steel (C), produced in Japan.
Basic properties of weldments for steel A and C are shown in Table1. Transition temperatures
i.e. 50% Charpy impact energy and nil ductility drop weight test temperature (table 2) were
evaluated too as a part of weldability testing. The results, obtained for , WM and are analyzed and
compared.
It was concluded that applied testing methods do not exclude each other, since they produce
complementary results, helping to understand better brittle fracture behaviour of welded joints as a
important part of global weldability assessment.
1080 V. Grabulov et al.

Table1. Results of tensile and bending tests of welded joints and hardness of welded joints

Table 2: Nil-ductility-transition temperatures, oC

FIGURE 1. Results of explosion bulge test (L- notch in cross-rolling, C- in rolling direction, base
metal –BM, weld metal-WM)

References
1. W.S.Pellini, Guidelines for fracture-safe and fatigue-reliable design of steel structures,TWI,
Abington, 1983.
2. Matthews, J. R. , Porter, J. F. , Church, J. and Macecek, M. , Welding J. 36, 1991, pp 35-41.
3. S.Sedmak, Welded joint properties evaluation by cracked specimens testing, in Fracture
Mechanics of Weldments, Gosa-TMF, Belgrade, 1985, pp 281-306. (in Serbian)
 19. Structural Integrity Assessment in Theory and Practice 1081

CASE STUDY OF SUPPORTING TUBES FAILURE

Vera Sijacki Zeravcic, Milos Djukic, Gordana Bakic, Biljana Andjelic1

and Bratislav Rajicic
Faculty of Mechanical Engineering, University of Belgrade,
Kraljice Marije 16, 11120 Belgrade 35, Serbia,
1Technical Faculty Cacak, University of Kragujevac, Svetog Save 65, 32000 Cacak, Serbia

vsijacki@mas.bg.ac.yu

This paper deals with determination of the failures causes of supporting tubes in a 350 MW fossil
fuel power plant. Due to frequent fractures of supporting pipes occurring always at the same
location and the ensuing reduction of the plant availability as well as substation financial losses,
there was an urgent need to resolve the problem [K. Takahashi, M. Yokouchi, S.K. Lee, K. Ando,
J. Am. Ceram. Soc., 86-12, 2003]- [3]. This work describes the test results and the conclusions
drawn from the results.
Supporting pipes with dimensions ‡30x5 were made of steel 12H1MF (GOST). The critical
zone is located at 40.5 m corresponding to the position of the vapour preheater 3. The characteristic
feature of the failure was a form of – ”mouth” in the immediate vicinity of the pipe bending and in
the pipe inner side, Fig. 1. Although the pipe wall thickness was reduced in the fractured zone, the
fracture itself falls into the category characteristic of thick wall failures (degree of plastic
deformation is < 57%).
Thick sintered deposits were found on the pipe outer surface propagating up to the root of the
fracture but not in the fracture zone. The pipe surface at the opposite side of the fracture was
relatively smooth with many parallel cuts 3-5 mm long. The pipe inner side showed no signs of
macro corrosion damage and was covered with thick, burnt deposit of black color. Many parallel
cuts were observed on the inner side of the fracture zone.

FIGURE 1. Macroview of failured tube.

Chemical composition of the tests pieces (outer side) is given in Table 1 and is in accordance
with the (GOST, TU 14-3-460-75), bug with somewhat lower carbon content.

TABLE 1. Chemical composition of investigated sample.

1082 V. S. Zeravcic et al.

The results of mechanical testing carried out on the samples prepared from the part of the pipe
outside the fracture zone are given in Table 2. The results show that none of the measured
parameters was satisfactory indicating that the material was in the state of an increased fragility.

TABLE 2. Mechanical properties of investigated samples.

The microstructure of test samples corresponds to that of overheating. Due to substantially

higher and non-permissible temperatures, decarburization and grain growth occurred at the pipe
outer side and characteristic features of overheating at the inner side. The presences of numerous
thermal fatigue cracks were observed at the pipe outer side at the locations where the deposit
fractured transversally and where thermal and corrosion fatigue occurred from the pipe inner side.
Based on the test results, plant operating history and other relevant indications, it was
concluded that the main cause for the supporting pipe failure was due to the disturbed flow and
boiling of the two-phase water-vapour mixture. This provoked sporadic contact of thermally
weaker conducting vapour with the pipe metallic wall. This process was occurred during relative
short-term exploitation but was very intensive due to: 1-disturbance in the heat transfer from the
inner side to the working fluid; 2-increase in temperature of the pipe metallic wall; 3-thermal
cycling; 4-damage of the protective magnetite layer; 5-appearance of cracks and corrosion
damages on the unprotected metallic surface; 6-increased intensity of deposit formation; 7-action
of increased and unfavourable stress field in the pipe metallic wall. On the vapour side the
formation of thermally induced fatigue cracks occurred due to: 1-intensive formation of very
compact and hard deposits; 2-overall increase in the pipe temperature; 3-thermal cycling caused by
difference in the thermal expansion of base metal and deposits.
The above described processes occurring in the flow and boiling of two-phase mixture and
deposit formation at the pipe inner and outer side was due to disturbance in the flame propagation
side and height-wise in the burner. In this particular case, due to: • flame buildup, unstable and
generally reduced fluid flow in the pipe; • increased content of noncombustible mineral
components in the smoke gases; • deviations in the medium quality; • inadequate quality of the
pipe inner side; • flow barriers, in this particular case, the presence of a welded joint with poorly
made weld root in the vicinity of fracture.

References
1. Sijacki Zeravcic, V., Bakic, G., Djukic, M., Andjelic, B., Case study of supporting tubes
failure, Report 12-02-12.04/2003, Fac.of Mech.Eng., Belgrade, 2004
2. Sijacki Zeravcic, V., Bakic, G., Djukic, M., Andjelic, B., Rajicic, B., Case study of supporting
tubes failure, Report 12-02-12.04/2004, Fac.of Mech.Eng., Belgrade, 2004
3. Sijacki Zeravcic, V., Bakic, G., Djukic, M., Andjelic, B., Rajicic, B., Case study of supporting
tubes failure, Report 12-06-12.04/2004, Fac.of Mech.Eng., Belgrade, 2004
 19. Structural Integrity Assessment in Theory and Practice 1083

STRUCTURE INTEGRITY OF PRESSURE VESELS REPAIR WELDING

JOINTS

Vera Sijacki Zeravcic, Gordana Bakic, Milos Djukic and Bratislav Rajicic
Faculty of Mechanical Engineering, University of Belgrade,
Kraljice Marije 16, 11120 Belgrade 35, Serbia,
vsijacki@mas.bg.ac.yu

Vertical cylindrical vessel-chambers as a part of coal – drying plants, whose purpose is to collect
wastewater, are supported at 3 points in upper dish head area and are made of fine-grained Mn
steel plates, joined by welding. Significant thinning and leaking in upper dish head area of the
vessel occurred due to original design provoking an intensive abrasion, cracking and rupture. After
reconstruction, in the upper zone of cylindrical shell, two new joints were made in sity by manual
arc welding, with subsequent, local post-weld heat treatment. However, cracks appeared firstly in
new welded zones in radial and axial joints, and then in zones of openings. Later, the similar
failure features began to appear in the area of original welded joints. All of these cracks were
repaired by properly specified technology. Unfortunately, after some period of exploitation the
initiation of new cracks was observed, at first by the visual inspection. This problem was detected
in the repaired areas in all (16) vertical cylindrical vessels. Cracks have propagated in different
directions with various penetration depths, up to 3 mm. It is interesting to note that crack appeared
in the HAZ of vertical joints, while in the area of radial welded joints the cracks were randomly
distributed in a larger zone. For cracks up to 3 mm, deep grinding was applied whereas for greater
crack depths repair welding with local post-weld heat treatment was used. Since welding with
subsequent heat treatment could only be used twice to repair the welded joints [1, 2], it is clear that
in the areas repaired several times (up to 7) proper mechanical characteristics of the material is
rather difficult to maintain.
In order to solve the described problem of cracking, an extensive examinations were applied
including: temperature measuring of the shell walls under working conditions; measurement of the
stress-state in all three phases of thermal and dynamic exploitation cycle (temperature uploading,
loading and downloading); determination of mechanical and chemical properties of material,
microstructural examination of base material and welded zone by replications, as well as residual
stress measurements in the critical zone. According to manufacturer specification, the investigated
material belongs to the class of manganese steels with max 0,23% C; 0,4-0,6% Si; 1,0-1,7% Mn
and max 0,04% S. Chemical analysis conducted on two plates indicate that the carbon content in
one plate is higher (C = 0,301%) than specified, while the content of other elements fall within
specified ranges. One of the parameters indicating changes in the material properties during
several repeated reparation is the hardness of the welded zones. Hardness values, according to
manufacturer specification for this type of steel, are in range of 186-244HB. However, measured
hardness values of base material in the vicinity of welded joint of inner vessel side were at the low
limit or lower. In certain area, significant low hardness of material was measured (120HB - in weld
metal zone inside the vessel; 140HB - in the same zone outside the vessel; in the base material
118HB and 146HB inside and outside, respectively). This could significantly decrease the loading
capability of welded joint. Furthermore, since material in the weakened areas has no adequate
mechanical properties, especially yield strength, these areas are beneficial sites for the appearance
of cracks. Optical microscopy examination of replicas and specimens prepared in the standard
manner indicate the presence of several different micro structural features. Base material belongs
to the class of fine grained ferrite-pearlite steel with significant striation features. In the areas
repaired several times, zones with martensite and Widmannstätten structure mostly in the
1084 V. S. Zeravcic et al.

overheating zone in HAZ were observed (Fig. 1). A large amount of MnS inclusions with
decohesion, which also could be the beneficial sites for crack nucleation in inclusions-matrix
interface were observed in the base material as well as welded joints, as shown in Fig. 2.

The stress fields in the critical zones of welded joints were determined using the magneto-
elastic devices for structural stress measurements. Structural stress states representing cumulative
influence of residual stresses in welded joints; residual stresses induced by manufacturing; and
stresses originated due to vessel weight were determined at the outer and inner side of vessel,
mostly in the areas of longitudinal and transversal welded joints. The distribution of detrimental
stresses along one of four longitudinal welded joints in the upper shell inner zone is presented in
Fig. 3, line 1 (maximal measured value of residual stresses is 34.8 kN/cm2).
The causes of pressure vessel damages could be considered from the three points of view:
construction design and loading, material and enviromental influence. A beneficial distribution of
residual stresses can assure a better operating reliability of pressure vessels. It is often the practice,
like in this particular case, that the residual stresses were not considered as an important parameter
in the design and strength calculation of pressure vessel that could exert significant influence on
the construction reliability. The contribution of residual stresses to the total stress state could be
very significant and detrimental thus endangering the vessel operating reliability. In this paper, a
possible redistribution of detrimental residual stresses by shoot penning as well as methods for
improving the integrity of installation is described.

References
1. V. Sijaki Zeravi: Residual stresses – Monography, Fac. of Mech. Eng. Belgrade, 1999
2. W.Mitter, R.Scheierl, et al: Investigation on the Relaxation of Stresses during Annealing of
Heat Treatable Steel, Residual Stresses in Science and Technology, Vol. 2, pp. 629,
Informationsgesellschaft, Verlag, 1987.
 19. Structural Integrity Assessment in Theory and Practice 1085

EFFECT OF MICROALLOYED STEEL WELDING PROCEDURE ON FATIGUE

CRACK GROWTH

Zijah Burzic, Vencislav Grabulov, Meri Burzic1, Mersida Manjgo2, Vladimir Gliha3
and Tomaz Vuherer3
Military Technical Institute, Belgrade, S&CG, 1NIC Uzice, S&CG
2University of Mostar, Faculty of Mechanical Engineering, Mostar, BiH
3University of Maribor, Faculty of Mechanical Engineering, Maribor, Slovenia

zburzic@eunet.yu

The application of high strength steel, e.g. NIOMOL 490K, developed for highly loaded welded
structures and low temperatures, depends on properties of welded joint critical regions. Strength,
toughness and crack resistance of welding joints have to be tested. Fatigue crack growth rate is
defined by relation of stress intensity factor range, 'K, induced by variable loading and crack
extension for one cycle, da/dN. For that precraked specimen has to be tested [1]. Samples for
investigation had been prepared by metal manual arc welding (MMAW) and MAG welding. Two
plates of 16 mm thick NIOMOL 490K steel of nominal yield stress 490 MPa [2], had been
prepared for asymmetric 2/3 X welded joint by edge machiningInitial notch for fatigue crack
growth rate testing was machined in weld metal and HAZ [3].
The chemical composition and mechanical properties are shown in Tab. 1 and 2, respectively.

TABLE 1. Chemical composition of NIOMOL 490K steel, %

TABLE 2. Mechanical properties of NIOMOL 490K steel

The fatigue crack growth tests had been performed on SE(B) specimens at room temperature,
with bending moment composed of a dynamic and a static part of variable amplitude in wave form,
under the ratio R = 0.1. Number of cycles for every 0.1 mm crack growth has been automatically
registered, enabling the design of relationship crack length a vs. number of cycles N. From this
curve it is possible to derive data, necessary for Paris law. It allows to calculate the number of
cycles to failure Nf for the initial, ai and critical, ac, crack sizes. Here, da/dN is the growth in crack
size a per unit cycle, N is cycle number, C and m are constants obtained from experiments, 'K =
Kmax - Kmin is stress intensity factor range in the loading cycle. Obtained relationship da/dN vs. 'K
for welded joint (WM and HAZ) is given in Fig. 1. Parameters C and n in Paris law (Eq. 1),
together with fatigue threshold 'Kth value are tabulated in Tab. 3.

da
C (' K ) m
dN (1)
1086 Z. Burzic et al.

TABLE 3. Parameters C and m and fatigue threshold 'Kth value for WM and HAZ

FIGURE 1. Diagram da/dN - 'K for welded joint, a) MMAW, b) MAG

References
1. Burzic Z., Sedmak S., Sedmak A., Grabulov G.,: Characterization of different crack
parameters in high strength steel welded joint, IIW International Congress on Welding and
Allied Processes, Vol. 1, A6, Nov. 29 - Dec. 1, Cairo, Egypt, 2004.
2. High Strength Low Alloyed (HSLA) Steels, ACRONI Jesenice, 2002.
3. Sedmak S., Hossin M. M., Burzi Z., Grabulov V., Sedmak A.,: Fatigue of high strength steel
weldment in pressure vessels application, Symposium Modern welding techniques and its
effects on industrial products quality improvement, Tripoli, p. 221, 2003.
 19. Structural Integrity Assessment in Theory and Practice 1087

FRACTURE RESISTANCE OF HIGH-STRENGTH 7000 FORGING ALLOYS

Z. Cvijovic, M. Rakin and M. Vratnica1

Faculty of Technology and Metallurgy, University of Belgrade, Karnegijeva 4, 11120 Belgrade,
Serbia and Montenegro
1Faculty of Metallurgy and Technology, University of Montenegro, Cetinjski put b.b., 81000

Podgorica, Serbia and Montenegro

zocvij@tmf.bg.ac.yu, mrakin@eunet.yu, majav@cg.ac.yu

Fracture resistance of high-strength 7000 aluminum alloys is strongly influenced by a number

parameters, including a range of microstructural features [1,2]. To obtain suitable microstructures
for the required fracture properties, the alloy composition and its processing history have to be
optimized. However, the modeling of the fracture toughness in wrought alloys requires a method
of incorporating a multiple fracture micromechanisms for validation of the toughness predictions
[3-5].
In the present work an attempt to model plane strain fracture toughness, KIc, of overaged 7000
alloy forgings with a multimechanistic approach have been made. Extensive microstructural and
fractographic analysis along with fracture toughness tests are carried out using three alloys with
different (Fe+Si) impurity levels. The fracture modes are identified and area fractions of coarse
voiding at intermetallic (IM) particles, AAp, intergranular fracture regions, AAi, and transgranular
fracture regions, AAt, are determined. The data are then quantitatively correlated to the KIc values
and the bulk microstructural attributes associated with the coarse IM particles and precipitates
estimated by image analysis.
The obtained results are compared with existing models and a new model, which is presented
by the following equation:

K Icp ª § § n ˜ w PFZ ·1 / 2 · º
K Ic ˜ exp « ( A At )1 / 2 ˜ exp ¨  ¨ ¸ ¸¸ »
( A Ap ) m
«¬ ¨ © O ¹ ¹ »¼
©
(1)
where KIcp denotes fracture toughness, evaluated using equation (2) [1]

1
1 2
ª §S · 3 º 1
K Icp | « 2V y E ¨ ¸ D » fv 6

«¬ ©2¹ »¼
(2)
is developed.
The proposed model appears to work resonably well, Fig. 1. The predicted values are lower
than the experimental ones, however, deviation ranges from 8 to 12.5%.
1088 Z. Cvijovic et al.

FIGURE 1. The KIc values as a function of the impurities contents.

Hence, the model developed in this work accurately describes the dependence of KIc on basic
tensile properties and microstructural parameters such as volume fraction of coarse IM particles, fv,
their size, D, and spacing, , density of intergranular precipitates, number, n, and width of
precipitate free zones, wPFZ.

References
1. Morere, B., Ehrstrom, J-C., Gregson, P.J. and Sinclair, I., Metall. Mater. Trans. A, vol. 31A,
2503-2515, 2000.
2. 2. Deshpande, N.U., Gokhale, A.M., Denzer, D.K. and Liu, J., Metall. Mater. Trans. A, vol.
29A, 1191-1201, 1998.
3. 3. Dumont, D., Deschamps, A. and Brechet, Y., Mater. Sci. Eng., vol. A356, 326-336, 2003.
4. 4. Gokhale, A.M., Deshpande, N.U., Denzer, D.K. and Liu, J., Metall. Mater. Trans. A, vol.
29A, 1203-1210, 1998.
5. 5. Kamp, N., Sinclair, I. and Starink, M.J., Metall. Mater. Trans. A, vol. 33A, 1125-1136,
2002.
 20. Critical Distance Theories of Fracture 1089

DOES A CHARACTERISTIC CRACK-TIP DISTANCE IMPLY

DISCONTINUOUS CRACK PROPAGATION?

A. P. Kfouri
University of Sheffield,
Department of Mechanical Engineering, ex S.I.R.I.U.S.
Mappin Street, Sheffield S1 3JD, U. K.
a.kfouri@bluewin.ch

Fracture Mechanics came into prominence as a legitimate study and contributor to structural
integrity mainly after the U.S.A. became a participant in the second world war. When the
construction of Keiser Liberty Ships designed according to the accepted rules of the day took place
at a phenomenally rapid rate it became apparent that designs based purely on the strength of the
material used, measured by its resistance to yielding – the yield stress – provided no satisfactory
insurance against catastrophic structural failure. As is often the case there were some forebodings
in the form of theoretical findings dating back several decades. Using the mathematical theory of
elasticity, elastic analyses revealed the existence of a stress-strain singularity at the tip of a sharp
crack or wedge and this rendered irrelevant conventional designs based solely on never exceeding
the yield stress anywhere in the structure. New safety criteria based on the strength of the
singularity, namely the stress intensity factor K or equivalently on Griffith’s total potential energy
release rate G were evolved. (For the ideal linear elastic material considered the two parameters are
uniquely related by the equation K2 = EG where E is the modulus of elasticity) For some time it
was hoped that these parameters were adequate to describe fracture events. In particular an applied
stress parallel to the crack has no effect on the values of K or G. However a singularity is a
mathematical concept. A real material cannot sustain an infinite stress and when an unsustainable
stress level is reached in the vicinity of the crack tip a mechanism such as plastic yielding (or the
formation of micro-cracks in rock-like materials) is invariably activated and its effect is to
eliminate the singularity. Within the crack-tip plastic zone of size rp there is a (usually) much
smaller finite region of size d in which fracture events take place and its material properties bear no
relation to those of the bulk material. Several models that attempt to describe stable or unstable and
fatigue crack growth found in the literature incorporate a finite crack tip characteristic distance d.
More particularly the elimination of the crack tip singularity implies that K and G vanish during
continuous crack propagation, i.e., when the ratio of the growth step 'a (= d)/ rp is infinitesimal.
But finite element analyses have revealed that when ('a / rp) is finite there is a surplus energy rate
G' to sustain crack growth. This leads one to conclude that crack propagation may be
discontinuous, in finite growth steps 'a.
The present paper describes a possible mechanism for discontinuous crack advance in which
surface separation occurs initially within (at the centre of) the crack tip characteristic distance d (=
'a) and not at the crack tip itself, spreading towards (and away from) the crack tip. The crack
extension occurs when the crack tip is reached. Finite element analyses show that this causes the
formation of a rippled crack face surface in elastic-plastic materials. Some pictorial evidence of
void formation ahead of the crack tip and of ripples during propagation, found in the literature, is
presented. The implication on the brittle-ductile transition of the dependence of the energy release
rate, GÄ, on the ratio ('a / rp) is also discussed.
 20. Critical Distance Theories of Fracture 1091

A MULTIAXIAL CRITERION FOR NOTCH FATIGUE USING A CRITICAL-

DISTANCE METHOD

Andrea Carpinteri, Andrea Spagnoli, Sabrina Vantatori and Danilo Viappiani

Dept. of Civil and Environmental Engineering & Architecture
Parco Area delle Scienze 181/A, 43100 Parma, Italy
andrea.carpinteri@unipr.it, spagnoli@unipr.it, sabrina.vantadori@unipr.it

According to the original work of Taylor [1] (see also the independent work of Tanaka [2]), the
endurance limit condition in a notched specimen occurs either when the range of a relevant stress
component (e.g. the normal stress perpendicular to the notch bisector), calculated from a linear
elastic analysis at a distance from the notch tip (point method), is equal to 'V0 ('V0 = fatigue limit
measured from a smooth specimen), or when the averaged value of such a stress range over a line
(line method) or a semi-circular area (area method) ahead of the notch tip is equal to 'V0. Such so-
called critical-distance methods assume that the distance from the notch tip, the length of the line
and the radius of the area ahead of the notch tip are equal to L/2, 2L and L, respectively, where L is
a material constant equal to the ElHaddad ‘intrinsic’ crack length [3].
Taylor has shown that these methods can successfully be applied to predict the endurance limit
of components containing either sharp, blunt or short notches [1,4] subjected to uniaxial cyclic
loading. The predicting capability of the line method seems to be connected to a theoretical
argument in the case of sharp notches, which are characterised by non-propagating cracks at the
endurance limit [5]. As a matter of fact, the stress intensity factor range of a non-propagating
crack of length 2L originating from the tip of the notch, due to a stress field characterised by an
averaged value of the normal stress range along the crack line in the uncracked body equal to 'V0,
attains the threshold stress intensity factor range of the material [6].
When a notched component is subjected to multiaxial cyclic loading, appropriate stress
components ahead of the notch tip must be selected if one wants to exploit the above critical-
distance methods. For this purpose, a multiaxial fatigue criterion (based on the critical plane
approach) has been used together with the point method by Susmel & Taylor [7] and Susmel [8]
with reference to the case of proportional multiaxial loading.
Recently, the present authors have proposed a high-cycle critical plane-based multiaxial
fatigue criterion [9-14] suitable for plain components. Accordingly, the critical plane orientation is
correlated to some averaged directions of the principal stresses, whereas other criteria available in
the literature determine this orientation by maximizing the amplitude and/or the maximum value of
some stress components. Then, the endurance limit estimation is performed by considering a
quadratic combination of the shear stress amplitude and the maximum value of the normal stress
acting on the critical plane. The application of the proposed criterion for evaluating the endurance
limit requires the knowledge of three material parameters: the fatigue limit under fully reversed
normal stress, the fatigue limit under fully reversed shear stress, the slope of the S-N curve in the
high-cycle regime for fully reversed normal stress.
In the present paper, the proposed criterion is applied to notched components under fatigue by
employing the point method. Emphasis is placed to the case of non-proportional multiaxial
loadings, for which the appropriate location of the critical point (at a distance L/2 from the notch
free surface) is discussed.
1092 A. Carpinteri et al.

References
1. Taylor, D., Int. J. Fatigue, vol. 21, 413-420, 1999.
2. Tanaka, K., Int. J. Fract., vol. 22, R39-R45, 1983.
3. ElHaddad, M.H., Dowling N.F., Topper, T.H., Smith, K.N., Int. J. Fract., vol. 16, 15-24,
1980.
4. Taylor, D., Wang, G., Fatigue Fract. Engng Mater. Struct., vol. 23, 387-394, 2000.
5. Smith, R.A., Miller, K.J., Int. J. Mech. Sci., vol. 20, 201-206, 1978.
6. Taylor, D., Fatigue Fract. Engng Mater. Struct., vol. 24, 215-224, 2001.
7. Susmel, L., Taylor, D., Fatigue Fract. Engng Mater. Struct., vol. 26, 821-833, 2003.
8. Susmel, L., Fract. Engng Mater. Struct., vol. 27, 391-411, 2004.
9. Carpinteri, A., Brighenti, R., Macha, E., Spagnoli, A., Int. J. Fatigue, vol. 21, 83-88, 1999.
10. Carpinteri, A., Brighenti, R., Macha, E., Spagnoli, A., Int. J. Fatigue, vol. 21, 89-96, 1999.
11. Carpinteri, A., Brighenti, R., Spagnoli, A., Fatigue Fract. Engng Mater. Struct., vol. 23, 355-
364, 2000.
12. Carpinteri, A., Spagnoli, A., Int. J. Fatigue, vol. 23, 135-145, 2001.
13. Spagnoli, A., Int. J. Mech. Sci., vol. 43, 2581-2595, 2001.
14. Carpinteri, A., Spagnoli, A., Vantatori, S., Fatigue Fract. Engng Mater. Struct., vol. 26, 515-
522, 2003.
 20. Critical Distance Theories of Fracture 1093

SIZE EFFECTS FOR CRACK INITIATION AT BLUNT NOTCHES OR

CAVITIES IN BRITTLE MATERIALS

D. Leguillon, E. Martin1, D. Picard2 and C. Putot2

LMM, CNRS UMR 7607, Université P. et M. Curie, Paris, France.
1LCTS, CNRS UMR 5801, Université Bordeaux 1, Pessac, France.
2IFP, Direction Mécanique Appliquée, Rueil Malmaison, France.

LMM, Université P. et M. Curie, case 162, 4 place Jussieu, 75252 PARIS Cedex 05, France.
Phone : (33)144 275 322, Fax : (33) 144 275 259,
dol@ccr.jussieu.fr

This work relies on an Irwin-like criterion able to predict brittle crack initiation at corners, v-
notches and other situations such as interfaces breaking a free surface (delamination initiation). It
is based simultaneously on an energy and a maximum stress criteria. The reason of this dual
formulation can be found for instance in Parvizi et al.1 experiments as analysed in a paper by one
of the author.2 It is shown that, if the singular exponent of the stress elastic field at the
concentration point is not ½ (a pure crack), the crack initiation is a brutal process (unstable) at least
on a short initiation length " .2,3 This distance being used as a small parameter in matched
asymptotics, it leads to the following expansion of the energy release rate G and next to the Irwin-
like criterion:

1 O
§ Gc ·
G k A " 2 O  1  ... ; k t kc ¨ ¸ V c2 O 1 (1)
© A ¹
where A is a scaling coefficient, k is the generalised intensity factor of the singularity with
exponent O . Here, Gc and V c are two failure parameters: the toughness and the strength of the
material (the use of two parameters is a usual feature in Cohesive Zone Models). The above
criterion matches with the Griffith’s one for a crack ( O 1 / 2 ) and with the stress one for a
straight edge without stress concentration ( O 1 ).
The matched asymptotic procedures allow analysing neighbouring problems dealing with
blunt cracks (existence of a notch tip radius, a soft layer ahead of the crack tip in a bi- or laminated
material), small cavities…
The main difference is that there are now two competing small parameters. In the examples
cited above it is respectively the notch tip radius, the layer thickness or the cavity diameter. In
these cases, the matched asymptotic procedures involve the structural small parameter d instead
of the crack extension length " , and the energy release rate writes:

G k A ( P ) d 2 O 1  ... (2)

where P " / d is the ratio of the two small parameters. The scaling coefficient A is a function
of P and the criterion is no longer explicit since the stress field cannot be asymptotically
expanded. Nevertheless it is possible to give a numerical expression that requires simply the
knowledge of the stress field ahead of the primary crack. A FE computation can be carried out
1094 D. Leguillon et al.

once for all on a simplified geometry so-called "inner domain". The role of the characteristic
length d of the microstructure at the origin of the size effect is clear in (2).
Applications are proposed on the influence of the notch tip radius on the apparent toughness;
4,5 the step-over mechanism in bedded sediments; 6 the role of anadhesive layer between two steel
7
plates; the crack path in ceramic laminates.8

To illustrate our purpose let us consider a small circular hole in a plate submitted to a remote
tension. The two small parameters are the crack increment length and the diameter of the hole. It is
known that the stress concentration factor of such a structure is 3 prior to any crack onset.
Following a maximum stress criterion, it means that failure will occur for V f V c / 3 whatever
the diameter of the cavity ( V f is the remote tension and V c the strength of the material). On the
contrary experiments show that if the cavity becomes smaller and smaller, the applied load leading
to failure increases from V c / 3 to V c . The above approach, taking into account an initiation
length, is able to render (imperfectly but this study is in progress) this effect as shown in figure 1
(the solid line is the theoretical prediction and the triangles correspond to experiments on PMMA
9).

Figure 1: Failure stress variation with the hole diameter on PMMA specimens. Solid line:
prediction, triangles: experiments.9

References
1. Parvizi A., Garett K.W., Bailey J.E., J. Mater. Sci., 1978, 13, 195-201.
2. Leguillon D., Eur. J. of Mechanics – A/Solids, 2002, 21, 61-72.
3. Martin E., Leguillon D., Int. J. Solids Structures, 2004, 41, 6937-6948.
4. Leguillon D., Yosibash Z., Int. J. of Fracture, 2003, 122, 1-21.
5. Picard D., Leguillon D., Putot C., J. Eur. Ceram. Soc., 2005, to appear, available on line.
6. Picard D., PhD thesis, 2005, Paris.
7. Leguillon D., Laurencin J., Dupeux M., Eur. J. of Mech. A/Solids, 2003, 22, 509-524.
8. Leguillon D., Tariolle S., Martin E., Chartier T., Besson J..L., J. Eur. Ceram. Soc., 2005, to
appear, available on line.
9. Li J., Zhang X.B., 11th Int. Conf. on Fracture, Conference CD Rom, Turin, Italy, March 2005.
 20. Critical Distance Theories of Fracture 1095

THE THEORY OF CRITICAL DISTANCES

D. Taylor
Department of Mechanical Engineering
Trinity College Dublin, Ireland
dtaylor@tcd.ie

This paper summarises the keynote talk opening the session on ‘Critical Distance Theories of
Fracture’; my aim is to describe the methodology used, to demonstrate its applicability in
predicting experimental data on fracture and fatigue, to discuss the fundamental basis of the theory
and, to suggest areas for future work in this field. The Theory of Critical Distances (TCD) is the
name which I have given to a group of theories used for predicting the effects of notches and other
stress-concentration features. These theories have a long history, but currently they are being
investigated by only a small number of researchers worldwide. This is a pity, because the TCD is
capable of accurate predictions in a wide variety of situations. The essential elements are an elastic
stress analysis and a material-dependent critical distance, a parameter which we call L. The value
of L can be written in terms of other material parameters as:
2
1 § Kc ·
L ¨ ¸¸
S ¨© V o ¹ (1)
Here Kc is a critical stress intensity value (e.g. the fracture toughness in static loading or the
crack propagation threshold in fatigue) and Vo is a material strength parameter (in static or cyclic
loading) which may or may not correspond to the plain-specimen strength. Four different versions
of the TCD will be considered: (a) the Point Method (PM) in which failure occurs when the stress
is equal to Vo at a distance L/2 from the notch root; (b) the Line Method (LM), in which the
average stress over a distance r=0 to 2L is used; (c) the Imaginary Crack Method (ICM) in which
an imaginary crack of length L is placed at the notch root and LEFM conditions assumed, and; (d)
Finite Fracture Mechanics (FFM) in which fracture conditions are derived by assuming a finite
increment of crack growth, equal to 2L.
These methods give predictions which are usually similar, but not always identical and which
are, in some particular cases, highly divergent. The LM and PM were first proposed for use in
fatigue, by Neuber [1] and Peterson [2] respectively. They have since largely dropped out of use,
except in the form of empirical equations: recent work shows that they can be used with high
accuracy by employing them in conjunction with FEA [3]. The ICM has been used in various
forms, both in fatigue (e.g. ElHaddad’s method) and in predicting fracture in brittle materials. The
FFM method is relatively new [4]; it may provide important insights into the reason for the success
of the other methods. The PM and LM were also proposed by Whitney and Nuismer [5] in the
1970s, for the prediction of fracture in fibre composite materials. This initiated a large body of
literature – in fact this is the main area in which the TCD is currently being used industrially, and
actively researched. Kinloch, Williams and co-workers [6] proposed using the PM for polymers
but their initiative was not pursued by others. Recently we have shown that the TCD can be used to
predict brittle fracture not only in polymers [7] but also in ceramics [8] and metals [9]. Contact
problems such as fretting fatigue can also be addressed [10], and the theory can be applied to
multiaxial loading situations by combining it with an appropriate mixed-mode failure criterion.
A major challenge for continued study is the theoretical basis of the TCD: why do these
methods work so well? Certainly their success is not limited to one particular micromechanism of
1096 D. Taylor

failure (for example they work equally well for crack propagation by cleavage and by microvoid
coalescence modes) nor do they require a realistic analysis of notch-root stresses (modelling of
plastic deformation is not necessary). Thus they cannot be rationalised by reference to any physical
mechanism (in contrast to, for example, the cracked-carbide model of cleavage fracture). They
share many features of LEFM in being linear-elastic, continuum-based models, but avoid some of
LEFM’s limitations – for example they can be applied to very short cracks and small notches. N
many cases the value of L is similar to a microstructural parameter (e.g. grain size) but in some
materials, such as fibre-composites and concretes, L values are much larger and may equate to
process zone sizes. Some light may be thrown on these issues by our recent work on FFM, in
which we recalculated strain-energy release rates using finite (instead of infinitesimal) crack
extensions. The assumption here is that, during a fracture process, crack growth is not smooth and
continuous but rather discontinuous, occurring in a step-like manner. This observation can
certainly be applied to some fracture processes (e.g. near-threshold fatigue crack growth and
‘stick-slip’ brittle fracture in polymers) and may be true in all cases. This provides a possible
explanation for the success of the other TCD methods such as the PM and LM, which may in fact
be approximations of FFM.
In conclusion, the Theory of Critical Distances is a theory which has not received as much
attention as it deserves. One barrier to its general acceptance is the lack of a clear theoretical basis
to explain its success. This is a challenge for future work; however, we should remember that the
first and most important test of any scientific theory is its ability to predict the experimental data:
this is a test which the TCD passes with flying colours.

References
1. Neuber, H. Theory of Notch Stresses Springer, Berlin, Germany 1958.
2. Peterson, R.E. In Metal Fatigue edited by G.Sines and J.L.Waisman, McGraw Hill, New
York, 1959, 293-306.
3. Taylor, D. Int.J.Fatigue 21 413-420, 1999.
4. Taylor, D., Cornetti, P. and Pugno, N. Engng Fract.Mech. 72 1021-1038, 2005.
5. Whitney, J.M. and Nuismer, R.J. J.Comp.Mater. 8 253-265, 1974.
6. Kinloch, A.J. and Williams, J.G. J.Mater.Sci. 15 987-996, 1980.
7. Taylor, D., Merlo, M., Pegley, R. and Cavatorta, M.P. Mater.Sci.Engng A 382 288-294 2004.
8. Taylor, D. Engng Fract.Mech. 71 2407-2416, 2004.
9. Taylor, D. In. proc.ECF11, Stockholm, Sweden 2004.
10. Vallellano, C., Dominguez, J. and Navarro, A. Engineering Failure Analysis 11 727-736,
2004
 20. Critical Distance Theories of Fracture 1097

STRENGTH ANALYSIS OF COMPOSITE PINNED JOINTS

H. A. Whitworth, O. Aluko and N. Tomlinson

Department of Mechanical Engineering, College of Engineering, Architecture &
Computer Sciences, Howard University, Washington, D.C. 20059
hwhitworth@howard.edu

Joining by mechanical fasteners is common practice in the assembly of structures. For advanced
composites, attachment by means of bolts or screws becomes more attractive, not only in view of
assembly and disassembly, but also because of fabrication simplicity and structural efficiency,
Theo De Jong [1]. In the analysis of mechanically fastened joints in orthotropic materials, the
prediction of the stress distribution around the fastener hole is of fundamental interest for the
prediction of strength. Since joints can lead to premature failure of the structure, joint strength is an
important property in any design.
The applications of composite materials in many structural designs have led to extensive
research aimed at developing an understanding of the sensitivity of composite structures to the
presence of holes and inclusions [2-5]. Lekhnitski [2] and later Savin [3] analyzed problems
related to the determination of the stress distribution in anisotropic plates weakened by an opening
and deformed by forces applied to the mid plane. Green and Zerna [4] also independently
proposed a formulation for the deformation of anisotropic elastic materials containing stress
raisers.
In order to investigate the influence of hole size on composite strength, Konish and Whitney
[5] proposed an approximate solution to the stress distribution in the vicinity of flaws. To evaluate
the strength of composite plates containing through the thickness discontinuities, Whitney and
Nuismer [6] introduced two related failure criteria: the point stress and the average stress failure
criterion. Both criterion evaluate failure based on a characteristic dimension and indicate that
failure occurs when the stress at a characteristic distance from the edge of the hole reaches the
unnotched tensile strength of the material. Although Whitney and Nuismer [6] suggested the
characteristic length to be a material constant, it is shown by Whitworth, et al. [7] that this value
depends on specimen geometry. In this analysis, based on the Chang et al. [8] characteristic curve
model, the point stress failure criterion will be used to evaluate the characteristic lengths in tension
and compression and a two-dimensional stress analysis used to evaluate the stress distribution in
the vicinity of the joint.
The present analysis of the bearing strength of pin-loaded composite joints involved using the
Chang-Scott-Springer characteristic curve model and a two-dimensional analysis to evaluate the
stress distribution around the fastener hole. Stress calculations are based on the analytical method
of complex stress functions and a numerical approach used for satisfying the displacement
boundary conditions of the contact area between the pin and the hole. For the characteristic curve,
characteristic lengths in tension and compression were evaluated by applying the point stress
failure criterion to a center notch plate subjected to tensile loading and a plate containing a circular
inclusion subjected to compressive loading. The Yamada-Sun [9] failure criterion was used to
evaluate joint failure.
In the present analysis, in order to evaluate the strength of composite pinned joints, the stress
distribution along a characteristic dimension around the hole must first be evaluated. The
conditions for failure can then be predicted with the aid of an appropriate failure criterion. The
Yamada-Sun failure criterion was used for this analysis. This criterion is given by the relationship
[9]
1098 H. A. Whitworth et al.

2 2
§ V 1 · § W 12 ·
¨¨ ¸¸  ¨ ¸ e2
© XL ¹ © S ¹ (1)
where V1 and W12 are the longitudinal and shear stresses, XL the ply longitudinal strength and S the
ply shear strength. In this model, failure is expected to occur when the value of e is greater than or
equal to unity.
The characteristic curve model proposed by Chang et al. [8] can be expressed as

rc T  R  Rt   Rc  Rt  cosT (2)
where Rt and Rc are the characteristic dimensions in tension and compression, and the angle T
ranges in value from -S/2 to S/2. The parameters Rt and Rc are evaluated by applying the point
stress failure criterion [6] to a plate with an open hole loaded in tension and a plate with an
inclusion loaded in compression. Material property for AS4/3501-6 graphite/epoxy laminates is
used for this analysis and the results compared with available experimental data.

References
1. De Jong, Theo, J. Composite Materials, vol. 11, 313-331, 1977.
2. Lekhnitskii, S. G., Anisotropic Plates, Gordon and Beach, London, 1968.
3. Savin, G.N., Stress Concentration Around Holes, Pergamon Press, New York, 1968.
4. Green, A.E. and Zerna, W., Theoretical Elasticity, Dover Publications, Inc., New York, 1968.
5. Konish, H. J. and Whitney, J.M., J. Composite Material, vol. 9, 157-166, 1975.
6. Whitney, J.M., and Nuismer, R. J., J. of Composite Materials, vol. 8, 253-265, 1974.
7. Whitworth, H.A., Othieno, M. and Barton, O, Composite Structures, vol. 59, 261-266, 2003.
8. Chang, F.K., Scott, R.A. and Springer, G.S., J. of Composite Materials, vol. 16, 470-494,
1982.
9. Yamada, S. E. and Sun, C. T., J. of Composite Materials, vol. 12, 275-284, 1978.
 20. Critical Distance Theories of Fracture 1099

APPLICATION OF THE THEORY OF CRITICAL DISTANCE TO FRETTING

FATIGUE

J. A. Araujo, L. Susmel1, D. Taylor1 and L. H. M. Lopes

Department of Mechanical Engineering, University of Brasília
CEP 70910-900 – Brasília – DF – Brazil
alex07@unb.br
Department of Mechanical Engineering
Trinity College, Dublin 2, Ireland

Fretting fatigue occurs at the contact interface of mechanical joints which are designed to be at rest
but that nevertheless experience some sort of relative movement due to vibration. This work aims
to develop a crack initiation threshold condition for fretting fatigue. The methodology considers
the use of Taylor’s stress point method [1] associated with multiaxial models, which divide the
stress space in safe and damage zones [2-4].
Experimental data
The analysis will be validated considering experimental data for a cylinder on plane contact
configuration (Fig. 1) reported in [5]. These tests revealed an effect of the pad radius (or contact
size, a) on fretting life. It was observed that for small contact sizes fretting tests last infinitelly
while for large contacts specimens broke within a finite number of cycles. The range defined by
the largest contact to show infinite life and the smallest contact to show finite life was termed the
critical contact size range, acrit. Twenty nine tests were performed and divided within four data
series.
Results
The first step in the analysis was to define the cyclic stress tensor at the center of the structural
volume for the Al 4% Cu tested under fretting. Algorithms were implemented to compute the
equivalent shear stress amplitude defined according to the multiaxial models studied. Figure 2
shows some of the results provided by the analysis considering the Modified Wohler Curve
Method (MWCM) and a set of experimental data. Such diagram make it evident that our method
was successful in predicting high-cycle fatigue strength under fretting, allowing run-out data to fall
within an error interval of about ±20%. Finally, Fig. 2 shows also that those data characterised by
Nf<107 cycles to failure were always above the threshold condition predicted by the method
(straight lines characterised by SU=0%), confirming again the soundness of the proposed
approach.
Conclusion
The Theory of Critical Distances, when used in conjunction with multiaxial stress based
models as the MWCM, is capable of predicting the results of fretting fatigue experiments with a
high degree of accuracy ( r20%). Compared with other notch analogue methodologies proposed
by the authors [6] for fretting fatigue, this approach has the advantage of defining the critical
distance as a material parameter. Hence, if the basic fatigue parameters are appropriatelly defined
for an specific alloy the crack initiation risk can be directly computed without the need to carry out
further fretting fatigue calibration tests to define the size of the structural volume. The
methodology proved simple to implement and the fact that it requires only the linear-elastic stress
state calculated at the centre of the structural volume to perform an accurate high-cycle fatigue
assessment makes it extremelly appealing from a engineering point of view.
1100 J. A. Araujo et al.

FIGURE 1. Cylinder on plane contact configuration tested.

FIGURE 2: High-cycle fretting fatigue predictions plotted in Wa vs. U diagrams.

References
1. Taylor, D., Bologna, P., Bel Knani, K. (2000) Prediction of fatigue failure location on a
component using a critical distance method. Int. J. Fatigue 22, 735-742.
2. Susmel, L., Lazzarin, P. (2002) A bi-parametric Wöhler curve for high cycle multiaxial
fatigue assessment. Fatigue Fract. Engng. Mater. Struct. 25, 63-78.
3. Dang Van, K., Papadopoulos, I.V. (1987), Multiaxial fatigue failure criterion: a new
approach, Proc. of the Third International Conference on Fatigue and Fatigue Thresholds.
4. Mamiya, E. N., Araújo, J. A. & Gonçalves, C. A. (2005), Multiaxial Fatigue: a stress based
criterion for hard metals, International Journal of Fatigue, pp .177-187.
5. Nowell, D. (1988), An analysis of fretting fatigue, Doctor of Philosophy thesis, Oxford
University.
6. Araùjo J. A., Vivacqua R. C., Bernardo A. T. S., Mamiya E. N., (2005) A crack initiation
threshold methodology in fretting fatigue. Submitted to: Journal of Strain Analysis
 20. Critical Distance Theories of Fracture 1101

THE THEORY OF CRITICAL DISTANCES: APPLICATIONS IN FATIGUE

L. Susmel
Department of Mechanical & Manufacturing Engineering – Trinity College, Dublin (Ireland)
Department of Engineering - University of Ferrara, Via Saragat, 1 – 44100 Ferrara (Italy)
lsusmel@ing.unife.it

This papers reviews the use of the Theory of Critical Distances (TCD) to predict high-cycle fatigue
strength of real components when the complexity of their geometry causes stress concentration
phenomena.
The TCD applied to fatigue problems assumes that fatigue damage depends on the stress field
distribution in the vicinity of the stress concentrator. In other words, it assumes that fatigue
damage can correctly be estimated only if the entire stress field damaging the fatigue process zone
is correctly accounted for.
According to the TCD, notched components are in their fatigue limit condition when an
effective stress, 'Veff, which depends on the entire stress field distribution ahead of the tip of the
stress concentrator, equals the material plain fatigue limit, 'V0:

' V eff 'V 0

(1)
The TCD can be formalised in different ways resulting in different definitions of the effective
stress, 'Veff. For instance, the point method (PM) postulates that the controlling parameter is the
stress at a given distance from the notch tip. On the contrary, the line method (LM) assumes that
the effective stress must be calculated averaging the stress along a straight line emanating from the
notch tip, and the integration path length is a function of the material fatigue properties. Finally,
according to the area method (AM), the effective stress can also be calculated by averaging the
maximum principal stress over an area.
The point method was first proposed by Peterson [1], whereas the line method was introduced
by Neuber [2]. Both of them were subsequently revisited by Tanaka [3], Lazzarin et al. [4] and,
finally, by Taylor [5].
It is interesting to observe that the most modern formulations of the TCD are based on the
assumption that the critical distance, that is, the material characteristic length, has to be calculated
according to El Haddad’s definition [6]:
2
1 § ' K th ·
L ¨ ¸¸
S ¨© ' V 0 ¹ (2)
The previous equation makes it evident that L depends on two material fatigue properties (the
plain fatigue limit, 'V0, and the threshold value of the stress intensity factor, 'Kth) for this reason
it is in turn a material property which is different for different material and load ratios, R.
These methods were seen to be successful in estimating high-cycle fatigue strength of notched
components, giving predictions falling within an error interval of about 20% [7, 8].
The main advantage of the above formalisations of the TCD is that these methods are based on
the use of linear elastic stresses: this makes them suitable for being used in situation of practical
interest by post-processing linear-elastic FE analysis.
1102 L. Susmel

Another interesting feature of the TCD is that it can successfully be used in conjunction with
the critical plane approaches [9, 10, 11] to predict the high-cycle fatigue strength of notched
components subjected to multiaxial fatigue loading. In particular, such an extension takes as its
starting point the idea that, in order to correctly predict fatigue damage in the presence of complex
stress states, two different aspects must be taken into account: stress gradients and degree of
multiaxiality of the stress field damaging the fatigue process zone. The first problem is directly
accounted for by the TCD, whereas the latter by the multiaxial fatigue criteria based on the critical
plane approach.
Finally, it is interesting to observe even that the TCD was successfully employed to predict
fretting fatigue damage in the high-cycle fatigue regime, using both uniaxial methodologies [12]
and multiaxial approaches [13].

References
1. Peterson, R. E., Notch sensitivity. In: Metal Fatigue (Edited by G. Sines and J. L. Waisman),
MacGraw-Hill, New York, 293-306, 1959.
2. Neuber, H., Theory of notch stresses. Springer Publishers, Berlin, 1958.
3. Tanaka, K., Int. J Fracture 22, R39-R45, 1983.
4. Lazzarin, P., Tovo, R. and Meneghetti, G., Int. J Fatigue 19, 647-657, 1997.
5. Taylor, D., Int. J. Fatigue 21, 413-420, 1999.
6. El Haddad, M. H., Topper, T. H. and Smith, K. N., J. Engng. Mater. Tech. (ASME Trans.)
101, 42-45, 1979.
7. Susmel L., Taylor D., Int. J. of Strain Analysis 38, 443-452, 2003.
8. Taylor D., Bologna P., Bel Knani K., Int. J. Fatigue 22, 735-742, 2000.
9. Susmel L., Taylor D., Fatigue Fract. Engng. Mater. Struct. 26, 821-833, 2003.
10. Susmel L., Fatigue Fract. Engng. Mater. Struct. 27, 391-411, 2004.
11. Susmel L., Taylor D., Can the conventional high-cycle multiaxial fatigue criteria be re-
interpreted in terms of the theory of critical distances? Submitted to Engineering Fracture
Mechanics, 2005.
12. Vallelano, C., Dominguez, J., Navarro, A., Engineering Failure Analysis 11, 727-736, 2004.
13. Araújo J. A., Susmel L., Taylor D., Ferro J. C. T. & Mamiya E. N., On the use of the theory of
critical distances and the modified Wöhler curve method to estimate fretting fatigue strength
of cylindrical contacts. Submitted to International Journal of Fatigue, 2005.
 20. Critical Distance Theories of Fracture 1103

FATIGUE ASSESSMENT USING AN INTEGRATED THRESHOLD CURVE

METHOD - APPLICATIONS

M. D. Chapetti
INTEMA – Research Institute for Materials Science and Technology
CONICET - University of Mar del Plata
J.B. Justo 4302 (B7608FDQ) Mar del Plata, Argentina
mchapetti@fi.mdp.edu.ar

In this work the fatigue resistance is study by using a threshold curve for fatigue crack propagation
that includes the short crack regime and thus is a function of crack length, a (see Figure 1). The
approach considers the effective driving force applied to the crack as the difference between the
total applied driving force defined by the applied stress distribution corresponding to a given
geometrical and loading configuration, 'K(a), and the threshold for crack propagation, 'Kth(a)
[1,2].
The Chapetti’s model [3] is applied to estimate the threshold for fatigue crack propagation by
using the smooth fatigue limit, 'VeR, the threshold for long cracks, 'KthR, and the microstructural
characteristic dimension d (e.g., grain size):

' K th ' K dR  ' K thR  ' K dR  >1  e  k ( a  d ) @ at d

(1)
where 'KdR is given by:

' K dR Y 'V eR S d
(2)
where Y is the geometrical correction factor. Figure 1 shows schematically the threshold curve
given by expression (1). For a crack length a = d, 'Kth = 'KdR, and 'Kth tend to 'KthR for long
cracks.

FIGURE 1. Threshold curve given by expression (1)

Applications to the analysis of the fatigue resistance of notched and damaged specimens, but-
welded joints and a microporous material were carried out. Figure (2) shows some results.
1104 M. D. Chapetti

FIGURE 2. (a) Fatigue behavior of short cracks at notches. (b) Influence of toe angle and initial
crack length on fatigue limit of butt-welded joints. (c) Short cracks from foreign object damage in
Ti alloy. (d) Fatigue of austempered ductile iron (ADI).

References
1. [1] Miller, K.J., Fat. Engng. Mater. Struct., Vol.5, No.3, pp.223-232, 1999.
2. [2] Chapetti, M.D., International J. of Fatigue, Vol.25, No12, pp.1319-1326, 2003.
 20. Critical Distance Theories of Fracture 1105

ANAYTICAL APPROACHES VS ATOMISTIC SIMULATIONS IN FRACTURE

N. Pugno, A. Carpinteri, M. Ippolito1, A. Mattoni1 and L. Colombo1

Department of Structural Engineering and Geotechnics, Politecnico di Torino, Corso Duca degli
Abruzzi 24, 10129, Italy
1INFM-SLACS Sardinian Laboratory for Computational Material Science and Dipartimento di

Fisica Università di Cagliari, Cittadella Universitaria, I-09042 Monserrato (Ca), Italy

nicola.pugno@polito.it

In this paper two different research groups have merged to try to compare their methods, i.e.,
analytical approaches (i) or atomistic simulations (ii), in the study of the fracture phenomenon.
Simple different case studies are considered.
(i) Due to the current limited computational capability, only tiny structures (composed by less
than ~104-6 atoms) can be numerically investigated with the atomistic simulations. On the other
hand, in treating such small scales classical Linear Elastic Fracture Mechanics could reveal limits.
In contrast, the energy based Quantized Fracture Mechanics [1] seems to be able to derive correct
predictions for the strength of the structure also at the nanoscale. A more classical Neuber-
Novozhilov stress based criterion, could also be of interest.
(i) We carried out damped-dynamics atomistic simulations with the aim of reproducing quasi-
static (and adiabatic) crack propagations at 0K, Fig. 1. We focus our work on cubic beta SiC [2]
since it is a prototype of an ideal brittle material up to extremes value of strain, strain rate and
temperature, and because of its technological relevance as a structural and nuclear material.
Atomic forces were calculated according to the Tersoff’s model. External loading was represented
in terms of tractions. Atomic positions were relaxed according to the local values of the forces and
by constantly damping velocities to zero, until the maximum force was less than a threshold value
(10-4 eV/Å). Accordingly, the fracture strength is computed.
Considering the same case studies analysed by the atomistic simulations, we derive the
corresponding analytical predictions for the fracture strength. Both atomistic and analytical
approaches show R-curve behaviours, Fig. 2, revealing a noticeable agreement.

FIGURE 1. Mode I crack in SiC crystal: atomistic simulations

1106 N. Pugno et al.

FIGURE 2. R-curve behaviour; dimensionless fracture energy vs crack length according to the
atomistic simulations (points) and to Quantized Fracture Mechanics (solid line).

References
1. 1. Pugno, N., Ruoff, R., Philosophical Magazine, vol. 84/27, 2829-2845, 2004.
2. 2. Mattoni, A., Colombo, L., Cleri, F., Physical Review B, vol. 70, 094108-1, 2004.
 20. Critical Distance Theories of Fracture 1107

A COUPLED STRESS AND ENERGY CRITERION WITHIN FINITE

FRACTURE MECHANICS

Pietro Cornetti, Nicola Pugno, Alberto Carpinteri and David Taylor1

Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, Turin, Italy
1Department of Mechanical Engineering, Trinity College, Dublin 2, Ireland

pietro.cornetti@polito.it, nicola.pugno@polito.it, alberto.carpinteri@polito.it, dtaylor@tcd.ie

Aim of the present paper is to introduce a new failure criterion in the framework of Finite Fracture
Mechanics. Criteria assuming that failure of quasi-brittle materials is affected by the stresses acting
at a finite distance from the crack tip are widely used inside the Scientific Community. These
approaches, in addition to some others, can be grouped together under the general term of Theory
of Critical Distances (Taylor [1]), in which linear-elastic analysis is combined with a material-
dependent critical length parameter. The relevant length parameter is assumed to be a material
constant: it generally approximates a microstructural length in the material such as the grain size.
On the other hand, the novelty of the present approach relies on the assumption that this finite
distance (i.e. the finite crack extension) is not a material constant but a structural parameter. Its
value is determined by a condition of consistency of both energy and stress approaches. In this
case, the relevant length parameter can be associated with the size of the damage zone at failure.
In formulae, the proposed criterion reads:

­a  '
° ³ V y  x  dx V u '
° a
®a  '
° K 2 a  da K 2 '
°³ I Ic
¯ a (1)
where Vu and KIc are material strength and toughness, a the crack length, Vy(x) the stress field
ahead of the crack tip, KI(a) the stress intensity factor and ' the finite crack extension.
In order to check its soundness, an application of the present criterion to the strength prediction
for three point bending tests (Carpinteri [2]) of various relative crack depths (a/b) and of different
sizes is performed. It is seen that the present model predicts the same trend of the Multi-Fractal
Scaling Law (Carpinteri [3]) for the size effect upon strength of un-notched structures. Finally, a
comparison with experimental data by Karihaloo et al. [4] on high strength concrete three point
bending specimens is shown in the bi-logarithmic plot of Figure 1, where Vf is the nominal stress at
failure and E the dimensionless structural size. An excellent agreement is found. It is interesting to
point out that the proposed criterion (1) is able to catch the concave-convex transition when
passing from notched to un-notched specimens in the log-log plot.
Applications of the theory in predicting the behaviour of other materials, and other fracture
processes, will also be discussed.
1108 P. Cornetti et al.

FIGURE 1. Bi-logarithmic plot of nominal stress at failure vs. specimen size for various relative
crack depths: comparison with experimental data on high strength concrete specimens obtained by
Karihaloo et al. [4].

References
1. Taylor, D., Engineering Fracture Mechanics, vol. 71, 2407-2416, 2004.
2. Carpinteri, Al., Engineering Fracture Mechanics, vol. 16, 467-481, 1982.
3. Carpinteri, Al., Mechanics of Materials, vol. 18, 89-101, 1994.
4. Karihaloo, B.L., Abdalla, H.M., Xiao, Q.Z., Engineering Fracture Mechanics, vol. 70, 979-
993, 2003.
 20. Critical Distance Theories of Fracture 1109

LOCAL STRAIN ENERGY DENSITY ANDFATIGUE STRENGTH OF WELDED

JOINTS

P. Lazzarin, P. Livieri1 and F. Berto

1University of Padova, Dept. Management and Engineering, St.S.Nicola 3 – 3100 Vicenza (Italy)
1University of Ferrara, Dept. of Engineering, Via Saragat, 1 – 44100 Ferrara (Italy)

plazzarin@gest.unipd.it, plivieri@ing.unife.it, berto@gest.unipd.it

Weld bead geometry cannot, by its nature, be precisely defined. Parameters such as bead shape and
toe radius vary from joint to joint even in well-controlled manufacturing operations (Taylor et al.,
[1]). In the notch stress intensity approach to the fatigue assessment of welded joints, the weld toe
is modelled as a sharp V-notch and local stress distributions in plane configurations are given on
the basis of the relevant mode I and mode II notch stress intensity factors (NSIFs). These factors
quantify the magnitude of asymptotic stress distribution obeying Williams’ solution [2]. When the
constancy of the weld toe angle is assured and the angle is large enough to make mode II
contribution non-singular, the mode I NSIF can directly be used to describe the fatigue strength of
fillet welded joints having different geometries (Lazzarin and Tovo [3], Lazzarin and Livieri [4]).
As an example, Fig.1 summarised fatigue strength data related to fillet welded joints, with an angle
at the weld toe equal to 135 degrees. In those welded joints, all fatigue failures originated at the
weld toe.
NIFs are endowed by an odd dimensionality, which depends on the V-notch angle. Varying the
angle, the comparability can be re-established by using the strain energy density averaged over an
ad hoc control volume surrounding the weld root or the weld toe as fatigue relevant parameter
(Lazzarin and Zambardi [6]). Recently, the radius RC of the averaging zone has been

FIGURE 1. Fatigue strength of welded joints as a function of the mode I NSIF. Scatter bands
defined by mean values r 2 standard deviations (from Livieri and Lazzarin [5]).
identified with reference to the conventional arc welding processes. For welded joints made of
structural steels and aluminium alloys RC was found to be 0.28 mm and 0.12 mm, respectively [5].
Figure 2 shows about 650 fatigue data, mainly from transverse fillet welded joints with failures
originated from the weld toe or the weld root. The main plate thickness ranged from 3 to 100 mm,
the transverse plate thickness from 3 to 200 mm. Due to the use of energy , not only the variability
of 'W increases with respect to stress-based curves (from about 4.0 MJ/m3 at 104 cycle to about
0.1 MJ/m3 at 2·106 cycles), but also the scatter increases. However, as soon as one reconverts the
1110 P. Lazzarin et al.

T'W value (3.3) into the more usual TV value, referred to a 10-90% stress-based band, the result
would be TV  3 . 3 / 1 . 21 1 . 50 This value matches exactly the TV value of the Haibach 10-
90% Normalised S-N Scatter Band for steel welded joints.
The aim of this contribution is to summarise the fundamentals of the method based on the local
strain energy density and to apply it to a number of fatigue strength data reported in the literature.
Analogies and differences with respect to Taylor’s stress-based Volume Method (Taylor [7]) will
be also highlighted.

FIGURE 2. Strain energy-based fatigue strength of cruciform welded joints subjected to tension or
bending loads (root and toe failures, load-carrying and non-load-carrying welds) [5].

References
1. Taylor, D., Barrett, N., Lucano, G., Int. J. Fatigue, vol. 24, 509-518, 2002
2. Williams, M.L., J. Appl. Mech., vol. 19, 526-528, 1952
3. Lazzarin, P., Tovo, R., Fatigue Fract Engng Mater Struct, vol. 21, 1089-1104, 1998
4. Lazzarin, P., Livieri, P., Int. J. Fatigue, vol. 23, 225-232, 2001
5. Livieri, P., Lazzarin, P., Int. J. Fracture, 2005, in press
6. Lazzarin, P., Zambardi, R., Int. J. Fracture, vol. 112, 275-298, 2001
7. Taylor, D., Int. J. Fatigue, vol. 21, 413-420, 1999
 20. Critical Distance Theories of Fracture 1111

AN IMPLICIT GRADIENT APPLICATION TO FATIGUE OF NOTCHES AND

WELDMENTS

R. Tovo and P. Livieri

Department of Engineering, University of Ferrara, Via Saragat 1, 44100 Ferrara (Italy)
rtovo@ing.unife.it, plivieri@ing.unife.it

Accurate predictions of the fatigue behaviour of structures in the presence of high stress
concentrations is of great interest in mechanics.
In the case of welded structures, as well as other sharply notched structures, local stress
methods could overestimate the effective stress at notch, by predicting strength values lower than
the experimentally measured one. To overcome this drawback, volume averages of the local stress
components have been often adopted to investigate fracture and fatigue damage at weld toes and
notch tips [1, 2].

Let us consider a body with volume V and define a field representing a local stress scalar ]
over V . According to Edelen [3] Eringen [4], a non-local stress scalar ] can be defined from the
average of its local counterpart ] weighted by the Gauss function a(x,y):

1
] ( x) ³ D ( x, y )] ( y ) dy in V
Vr ( x) V
(1)

In Eq.(1), the symbol Vr ( x) denotes the reference volume and is calculated as

Vr ( x) ³ D ( x, y ) dy . In Eq.(1) the symbol D denotes a weight function, such as the Gaussian
V
function, that is a distance-dependent function that vanishes at a certain distance from the actual
point, which hence denotes a kind of intrinsic length l of the model.

In this contribution, a non-local stress scalar ] is defined. After some calculations, the idea of
non-locality can be rewritten as an implicit gradient expansion:

] ( x) # ] ( x)  c 2 ’ 2 ] ( x) in V (2)
Eq. (1) has been first proposed by Peerlings et al. [5] and contains the Laplacian of the non-
local equivalent stress where c is a diffusive length.
In order to obtain the solution of eq. (2) in a two-dimensional domain with geometric
discontinuity (a sharp open corner), the Finite Element technique is adopted. The local equivalent
stress is calculated by assuming the elastic analytical Williams’ stress field. Boundary conditions
of Neumann type are assumed. For simplicity, in this contribution, we report only the results
obtained by imposing the local equivalent stress equal to the maximum principal stress.
On the basis of eq. (2), welded joint geometries taken from literature are analysed and
experimental fatigue life is compared to our theoretical estimation. A scatter band was obtained in
terms of maximum ] range by analysing steel cruciform welded joint with main plate thickness
range between 6 and 100 mm [6,7] (see figure 1). The welded joints were characterised by a toe
failure with an opening angle of 135°. In this paper, a numerical procedure is proposed in order to
obtain the fatigue life estimation of a generic welded joint geometry under traction or bending
1112 R. Tovo and P. Livieri

loading. Figure 1 shows the fatigue life prediction of a T welded joint under traction with a
opening angle 2D=118° in terms of the maximum ] range. Experimental points of the T-joint
fatigue failure fall into the predicted scatter band obtained on the basis of previous analysis.
In the final paper several examples will be given also concerning geometrical notches and it
will be shown how the proposed approach is much more efficient than nominal stress and
structural stress (hot spot) approaches for fatigue strength estimation of welded joints.

FIGURE 1: Fatigue life prediction of a welded T-joints made in steel, with an opening angle of
118° under traction

References
1. P. Livieri, R. Tovo. Fatigue limit evaluation of notches, small cracks and defects: an
engineering approach, Fatigue Fract. Engng Mater. Struct., vol. 27, n. 11, pp. 1037-1050,
2004.
2. L. Susmel, D. Taylor Fatigue Design in the Presence of Stress Concentrations. Int. J. of Strain
Analysis 38, 443-45, 2003
3. D.G.B. Edelen. Continuum Physics. Vol. IV, 75-204, Academic Press, 1976, New York
4. R. de Borst, H.B. Mühlhaus. Gradient dependent plasticity: formulation and algorithmic
aspect. Int. J. Num. Meth. Engng., 35, 521-539, 1992
5. R.H.J. Peerlings, R. de Borst, W.A.M. Brekelmans, J.H.P. de Vree. Gradient enhanced
damage for quasi-brittle material. International Journal of Numerical Methods in Engineering,
39, 3391-3403, 1996
6. P. Lazzarin, R. Tovo. A Notch Intensity Approach to the Stress Analysis of Welds. Fatigue
and Fracture of Engineering Materials and Structures 21,1089-1104, 1998
 20. Critical Distance Theories of Fracture 1113

USE OF JV5 TO PREDICT STATIC FAILURES IN NOTCHED COMPONENTS

P. Livieri
University of Ferrara, Dept. of Engin., via Saragat, 1 – 44100 Ferrara (Italy)
plivieri@ing.unife.it

The tangent stress VT over the free border of U-notch under mode I loadings may be evaluated by
means of analytical expression proposed in reference [1]. If VT,max and VT,tg are the values of stress
at the apex and tangent point respectively, we have:

>
V T V T , max K  (1 K ) cos(I ) s @ (1)

K being the ratios VT,tg to VT,max. The s and K parameters influence the distribution of VT along the
free edge and depend essentially on ratio between the depth a and the notch radius U. The accuracy
of equation (1) is very high and the mean square errors from the notch tip to tangent point range
between 1 and 2% when the a/U ratio is greater then 8. This error reaches the maximum value of
4.7% when a/U equals unity. As shown in figure 1, the J-integral (namely JVU) for a path between
two points A and B outside the tangent point may be calculated from eq. (1). After some
calculation, JVU assumes the form [1]:

2
S /2 V T , max 2
JVU
U
E'
V T , max 2 ³ >(1 K ) cos(I )  K @ cos(I ) dI
s
U
E'
Z
0 (2)
VT,max is the peak stress and E’ is equal to the Young modulus E for plane stress or to E/(1-Q) for
plane strain, respectively.
It is well known that failure analyses can be performed in two-dimensional brittle components
even by using the J-integral parameter [2,3]: the critical value of the applied load is reached when
J-integral equals a critical value depending on the material. In this paper J-integral will be used
even as failure criterion for brittle material in the case of rounded U-notches with linear flank. A
critical path radius U* is calculated from the knowledge of two static parameters: Fracture
toughness KIC and ultimate strength Vuts of the material:

U* 1  Q  KV 2 IC
2
uts (3)
The JVU is now evaluated from the notch tip between the two point T1 and T2 from the
intersection of the critical path U* centred at the notch tip and the notch border (see figure 1).

Figure 1: path for JVU evaluation

1114 P. Livieri

The limit case of a crack is obtained when U* >> U while for U >> U* the JVU method agrees
with the classic peak stress approach.
In order to confirm this analytical failure criterion, two type of brittle material have been
investigated. In figures 2 and 3 experimental failure nominal stress Vn against the stress
concentration factor Kt (gross area) are reported. Additionally, in the same figures the failure
prediction with the JVU parameter is reported. The two materials present different value of U*. The
PMMA material shows a fully notch behaviour because the U* is greater then the smaller tested U
value. On the contrary, the PVC material has the critical U* of the same order of the smaller tested
U. Note that the nominal stress prediction with JVU gives a plateau for higher Kt value despite of the
classic peak stress approach.

FIGURE 2: Experimental and analytical prediction for PMMA material for different U notch
values

FIGURE 3: Experimental and analytical prediction for PVC material for different U notch values

References
1. Livieri, P. A new path independent integral applied to notched components under mode I
loadings. International Journal of Fracture, 123, 107-125, 2003
2. Rice J.R., A path independent integral and the approximate analysis of strain concentration by
notches and cracks. ASME-Journal of Applied Mechanics, Vol. 35, 379-386, 1968
3. Anderson, T.L. Fracture Mechanics, fundamentals and applications. CRC Press, 1991
 20. Critical Distance Theories of Fracture 1115

STANDARDIZATION OF STRENGTH EVALUATION METHODS USING

CRITICAL DISTANCE STRESS

Toshio Hattori, Naoya Nishimura and Minoru Yamashita

Gifu Univ., Yanagido, Gifu City, Gifu, Japan
hattori@cc.gifu-u.ac.jp

In previous paper[1][2], the authors present a new strength evaluation methods using stress
singularity parameters H(intensity of stress singularity) and (order of stress singularity) for
adhesive bonding edges and contact edges. Using these two parameters the stress distributions near
these edges can be expressed as follows (see Fig. 1).

H
V (1)
rO

Fig. 1 Stress distributions near in stress singularity fields

The difficulty with this method was in obtaining the critical value of intensity of stress
singularity parameter Hc for each order of stress singularity .So, in this paper we present new
method for formularizing Hc by based on critical distance stress (point stress) , which can be
obtained typical material strength parameters such as the fatigue limit of smooth specimens w0 and
the threshold stress intensity factor range Kth of the cracked specimens. The estimated example of
these critical value of intensity of stress singularity parameter Hc for each order of stress
singularity for Ni-Cr-Mo-V Steel are shown in Fig. 2, in this case the critical distance is 0.012mm
and critical distance stress is 360Mpa. Using these easily obtained formularized critical value Hc(),
we can estimate the fretting fatigue crack initiation criteria for each contact edge angle, and thus
optimize the contact edge geometry. Finally, we discuss the development of these strength criteria
in terms of stress singularity fields for general stress concentration fields.
1116 T. Hattori et al.

Fig. 2 Formularized critical value of intensity of stress singularity Hc for Ni-Cr-Mo-V Steel

References
1. Hattori, T., et. al, J. Electronic Packaging, Trans ASME, vol. 1, 243-248, 1989.
2. Hattori, T., et. al, Developments in Fracture Mechanics for the New Century, JSMS, Tokyo,
Japan, 2001.
 20. Critical Distance Theories of Fracture 1117

APPLICATION OF POINT STRESS METHOD TO HYDRO-FRACTURING

TECTONIC STRESS MEASUREMENT

T. Ito
Institute of Fluid Science, Tohoku University
2-1-1 Katahira, Aoba-ku, Sendai, Japan
ito@ifs.tohoku.ac.jp

Knowledge of tectonic stress magnitudes and orientation is essential for understanding the crustal
dynamics. The stress at the depths in concern can be measured from the results of in-situ tests
carried out at those depths in drilled boreholes. Especially for the measurement at depth more than
1 km, hydraulic fracturing has been used generally, since, compared with the other methods, its
procedure and equipment of in-situ test are quite simple and appropriate for operating in such long
and narrow space as boreholes. With this technique, an interval of a borehole which is free of
natural fissures is sealed off with a straddle packer system and then pressurized by injection of
fluid to generate a tensile circumferential stress around the borehole. When this tensile stress
exceeds the strength of a rock and the concentration of tectonic stresses by the borehole, fracture
initiation occurs on the borehole wall. Then, the so-called breakdown pressure Pf is observed as
the borehole pressure at the fracture initiation. In addition to this pressure, several characteristic
pressure related to the tectonic stresses are observed in the borehole pressure – time history during
the test. Consequently, the tectonic stresses are computed from those observed pressures by using
the formulae of theoretical relations between the tectonic stresses and the pressures which are
derived theoretically in advance.
The classical theory for interpreting the breakdown pressure Pf assumes that the fracture
initiation occurs, when the maximum tensile stress reaches the tensile strength of the rock, Vf, on
the borehole wall. This condition is expressed assuming vertical borehole as follows;

ST P Pf
V f , ST S H  3S h  P (1)

where P is the borehole pressure, ST is the circumferential stress on the borehole wall, and SH and
Sh are the maximum and minimum tectonic stresses in a horizontal plane.
On the other hand, there are infinite number of small pores inside almost rocks, while the
pore’s shape and connectivity are dependent on the type of rock. The existence of such pores
brings special feature in mechanical behavior of rocks, different to cases of other materials. First,
the borehole pressure raised by fluid injection in hydraulic fracturing, penetrates through the pores
into the rock from the borehole wall, and the fluid penetration causes an additional circumferential
stress in compression at the borehole wall. Secondly, the pore fluid pressure has the effect to loose
strength in failure, where this phenomenon is known as the Terzaghi’s effective stress law. Taking
account both of the additionally-induced stress and the effect of the pore fluid pressure on tensile
failure, Eq. (1) is modified as follows;

­S
® T
¯ P Pf
 
 A Pf  p0 ½¾  Pf
¿
Vf (2)

where p0 is the initial value of the pore fluid pressure and A is a poro-elastic constant. Then we
come up with the following theoretical relation between the tectonic stresses and the breakdown
pressure Pf , which has been originally derived by Haimson and Fairhurst [1];
1118 T. Ito

3S h  S H  V f  Ap0
Pf
2 A (3)
However, this breakdown equation is still not so sufficient that it cannot explain the
dependency of Pf on increasing rate of the borehole pressure as shown in Fig. 1. To correct Eq. (3)
taking account of the dependency on pressurization rate, we have proposed a theoretical model
based on a criterion taking account of a length scale which characterizes the failure process [2].
Namely this model assumes that fracture initiation to occur when the maximum effective stress in
tension reaches the tensile strength of rock at a point that is not on the borehole wall but is inside
the rock by a radial distance of d. The d is a material constant of few millimeters. The solid line in
Fig. 1 shows the breakdown pressure predicted by our model, and it is actually consistent well with
the experimental results. Note that the gradient of pore fluid pressure developed around the
borehole becomes steeper as the pressurization rate becomes larger, and as a result, both of the
pore fluid pressure and the stress state at the distance d from the borehole wall change with the
pressurization rate. This is the reason why the breakdown pressure predicted by our model varies
with the pressurization rate. Furthermore, our model shows that if borehole diameter is sufficiently
large compared with the characteristic length d, the breakdown pressure will be close to the
breakdown pressure obtained by the conventional equation of Eq. (3), where the borehole diameter
prepared for the laboratory experiments shown in Fig. 1 is 3 mm.

FIGURE 1. Comparison of observations obtained by laboratory experiments using boreholes of 3

mm in diameter [3], and theoretical predictions [1, 2].

References
1. Haimson, B. and Fairhurst, C., Soc. Pet. Eng. J., vol. 7, 310-318, 1967.
2. Ito, T. and Hayashi, K., Int. J. Rock Mech. Min. Sci. & Geomech. Abstr., vol. 28, 285-293,
1991.
3. Zoback, M.D., Rummel, R., Jung, R. and Raleigh, C.B., Int. J. Rock Mech. Min. Sci. &
Geomech. Abstr., vol. 14, 49-58, 1977.
 20. Critical Distance Theories of Fracture 1119

A UNIFIED FAILURE CRITERION FOR BRITTLE OR QUASI-BRITTLE

MATERIALS UNDER ARBITRARY STRESS CONCENTRATION

J. Li and X. B. Zhang
University of Paris XIII, LPMTM, CNRS UPR 9001, France
Blaise Pascal University of Clermont II, LaMI, France
zang@moniut.univ-bpclermont.fr

The stress concentrations are of many types and of different levels. For brittle and quasi-brittle
materials, failure criteria proposed by different authors are established to study different situations.
For example, when a stress distribution is not uniform but does not present a singularity, the
maximum stress criterion ( V V c ) is used. For the case of a crack, the criteria are based on the
Linear Fracture Mechanics. When a stress concentration presents a singularity but weaker than a
crack one, such as sharp notches, the failure criteria are kinds of combination of the criteria for the
two above cases. In fact, each criterion only functions with a particular stress concentration case
for which it was developed. Its extension to a more general case often induces important errors.
Therefore, the purpose of the present work is to establish a unified crack onset criterion for brittle
and quasi-brittle materials that enables to count for all types of stress concentrations.
The study is based on two concepts: The first one is described by Leguillon [1] who combined
consistently the material strengthVc and the material toughness for crack propagation Gc in the
prediction of the crack onset from a sharp notch in brittle material. The second one is the stress
concentration dependency of the energy dissipation in a new surface creation, issued from direct
observation of the fracture surfaces in brittle materials. According to the observations on the
fracture surfaces of specimens made of PMMA obtained from our experiments, we propose a three
parameter criterion in order to predict the crack initiation from a smooth surface. A part from the
two material parameters used in the Leguillon criterion, we suppose that there exists a third
material parameter Gu which represents the material toughness under uniform tensile stress. This
parameter can be regarded as the energy dissipation by unit nominal area when the material is
subjected to uniform uni-axial tension. An empirical expression to determine this parameter is
proposed in this work.
Proposed criterion
The critical energy release rate for any local stress concentration level, the so-called specific
fracture energy, can be estimated by interpolation rule as follows:

Gs (D ) D mGc  (1  D ) n Gu (1)
where m and n are two real constants to be adjusted by experimental data; D t 0 is a real parameter
describing the local stress concentration level. D depends on the critical height of the crack band
formed just before the main crack initiation. It is assumed that the micro-cracks develops within
the zone where the maximal principal stress exceeds the material strength, i.e., V1 > Vc. The height
of this zone h can be regarded as the height of the pseudo crack band. Therefore, D can be defined
as follows:

hcrack
D
h (2)
1120 J. Li and X. B. Zhang

where hcrack is the height of the zone near a crack tip in which V1 > Vc when the crack grows. It
can be calculated according to the near tip stress field. It is clear that D = 1 for the case of a crack
and D = 0 when the material is under uniform tension.
The proposed criterion with three material parameters is as follows:
The crack initiation occurs at the direction where the tensile stress is maximal and if both the
following conditions are satisfied: (a) In the non-cracked configuration, the maximum principal
stress of the point at the characteristic distance V 1 (l ) reaches the material strength V c ; (b) In the
virtual cracked configuration, the incremental energy release rate at the virtual crack tip situated at
the same point G (l ) reaches the critical toughness Gs (D ) . Where the critical energy release rate
Gs (D ) should be evaluated using Equations (1) and (2).
Results and conclusions
In order to verify the proposed criterion, some experimental studies are carried out based on
the experimental results performed by authors or found out in the literature. These experimental
data are used to verify the accuracy of the proposed criterion. The predictions obtained by using
the Leguillon criterion are also presented for comparison. The material selected in this study is the
PMMA (polymethyl metacrylate), for its brittle characteristics. Three kinds of plane specimens are
considered: plates with double V-notches under uni-axial tension, plates with a central hole under
uni-axial tension and plates with double U-notches under uni-axial tension. Figures 1 and 2 show
some results of the comparison. The experimental results have shown that the proposed criterion is
physically reasonable, highly accurate and easy to apply. It can be used in crack initiation
prediction of engineering structures made of brittle or quasi-brittle materials.

FIGURE 1. Critical loads for holed plates FIGURE 2. Critical loads for U-notched plates
versus hole diameter versus notch radius

Reference
1. Leguillon D., Strength or toughness? A criterion for crack onset at a notch, European J.
Mech. A/Solids, 21, 61-72, 2002
 22. New Investigations on Very High Cycle Fatigue of Materials 1121

MORPHOLOGY OF STEP-WISE S-N CURVES DEPENDING ON NOTCH AND

SURFACE ROUGHNESS IN HIGH STRENGTH STEEL

H. Itoga, K. Tokaji1, M. Nakajima2 and H. N. Ko

Nakanihon Automotive College,
Sakahogi-cho, Kamo-gun, Gifu Prefecture 505-0077, Japan
1Dept. of Mech. and Systems Eng., Gifu Univ., 1-1 Yanagido, Gifu 501-1193, Japan
2Dept. of Mech. Eng., Toyota National College of Technology,

2-1 Eisei-cho, Toyota 471-8525, Japan

itoga@tcp-ip.or.jp

It has been indicated that subsurface fracture occurred at stress levels below the conventional
fatigue limit in long life region more than 107 cycles, particularly in high strength steels (Dahlberg
[1], Naito et al. [2], Nakajima et al. [3]). A notable feature of subsurface fracture is the presence of
a fish-eye with an inclusion near its centre, and the formation mechanism of fish-eye has not been
fully understood. High strength steels usually possess very high sensitivity to notch and surface
defect. Therefore, it is very important to understand the effects of notch and surface roughness on
crack initiation and associated S–N characteristics. In the present study, rotary bending fatigue tests
have been conducted in laboratory air at ambient temperature using notched specimens with
different stress concentration factors and specimens with roughened surfaces in high strength steel.
The effects of notch and surface roughness on step-wise S–N characteristics, particularly on the
transition stress, were discussed.
The material used is a Ni–Cr–Mo steel, which was quenched at 880qC followed by tempered at
200qC. Specimens with hourglass-shape (smooth specimen), three types of notched specimens
(stress concentration factor Kt=1.16, 1.51 and 2.0), and three surface-roughened specimens
(maximum height Rz=10 m, 16 m and 19 m) were prepared. Fatigue tests were performed using
cantilever-type rotary bending fatigue testing machine operating at a frequency of 52.5 Hz in
laboratory air at ambient temperature. After experiment, fracture surfaces were examined in detail
using a scanning electron microscope (SEM).

The S–N diagram of the notched specimens is shown in Fig.1, where the open and solid
symbols represent surface-related crack initiation and subsurface crack initiation, respectively. The
smooth specimen and the notched specimen with Kt=1.16 show subsurface fracture in long life
region, while subsurface fracture is not seen in the notched specimens with Kt=1.51 and 2.0. In the
1122 H. Itoga et al.

surface fracture region, fatigue strength decreases with increasing stress concentration factor, but
in the subsurface fracture region, there was no discernible difference in fatigue strength. It was
found that the high strength steel studied had very high notch sensitivity. Figure 2 represents the S-
N diagram for the surface-roughened specimens. All surface-roughened specimens exhibit
subsurface fracture, where the transition stress at which fracture mode changed decreases in the
specimens with rougher surface. In the surface fracture region, fatigue strength decreases with
increasing surface roughness, while in the subsurface fracture region, it does not depend on surface
roughness.
A fish-eye was always seen on fracture surfaces in subsurface fracture region, with a
nonmetallic inclusion near its centre from which the crack initiated. It was found that fish-eyes
were nearly circular and their periphery always touched the surface. Such morphologies of fish-
eyes were not affected by the presence of notch and surface roughness.
The transition stresses, i.e. the conventional fatigue limits, for the surface-roughened
specimens were predicted by using the —area parameter model (Murakami [4]).

(1)

where Vwp: predicted fatigue limit (MPa), HV: Vickers hardness (kgf/mm2), R: stress ratio, D:
0.226+HVu10-4. The equivalent defect sizes for surface roughness were employed as the —area
values (Murakami et al. [5]). The obtained results were listed in Table 1. It can be seen that the
predicted fatigue limits are in good agreement with the experimental ones when the arithmetical
mean deviation, Ra, was used as the height of surface roughness.

TABLE 1. Fatigue limit predicted by the —area parameter model

References
1. Dahlberg, E.P., Trans ASM, 58, 46–53, 1965.
2. Naito, T., Ueda, H. and Kikuchi, M., Metal Trans, 15A, 1431–1436, 1984.
3. Nakajima, M., Tokaji, K., Itoga, H. and H.-N. Ko, Fatigue Fract. Engng Mater. Struct., 26,
1113–1118, 2003.
4. Murakami, Y., Effects of Small Defects and Non-metallic Inclusions, Yokendo Ltd., Tokyo,
Japan, 1993.
5. Murakami, Y., Takahashi, K. and Yamashita, A., Trans. Jpn Soc. Mech. Eng., A63, 1612-
1619, 1997.
 22. New Investigations on Very High Cycle Fatigue of Materials 1123

VERY HIGH CYCLE FATIGUE BEHAVIOUR UNDER CYCLIC TORSION

LOADING

H. Mayer and S. Stanzl-Tschegg

BOKU, Institute of Physics and Materials Science, Peter-Jordan Str. 82, 1190 Vienna, Austria
herwig.mayer@boku.ac.at, stefanie.tschegg@boku.ac.at

Several technical components are stressed in cyclic torsion. Valve springs, for example or torque
transmission shafts must withstand high numbers of torsion load cycles in service, and fatigue
properties of materials under these loading conditions are of great interest therefore.
The present investigation serves to determine cyclic torsion fatigue strength at high and very
high numbers of cycles and to compare the measured data with results of axial loading
experiments. Fatigue tests are performed using ultrasonic fatigue testing equipment. Specimens are
stimulated to resonance vibrations, and the cycling frequency of about 20 kHz allows measuring
high cycle fatigue data within relatively short testing times. Ultrasonic fatigue testing of materials
is mainly performed in cyclic tension-compression with or without superimposed static loading.
Similar electronic control equipment and power generators are used to perform cyclic torsion tests.
Measurement and control of the vibration amplitude serves to control the magnitude of loading,
and control of resonance frequency serves to automatically detect crack initiation and specimen
failure. The mechanical equipment, however, had to be re-designed to account for the shorter
wavelength of shear waves compared with tension-compression waves.
Experiments are performed with the aluminium alloy 2024-T351 (ultimate tensile strength
Rm=460 MPa, yield strength Rp0.2=352 MPa, fracture strain A5=18%) and with Ck15 (carbon steel
with 0.15% C in normalized condition, Rm=834 MPa, Rp0.2=368 MPa, A5=9%). Axial and torsion
fatigue properties are investigated under fully reversed loading conditions in the regime from 105-
106 to 109-1010 cycles.
Results of cyclic torsion and cyclic tension compression S-N tests are shown in Fig. 1. Mises
equivalent stresses are used to compare axial and torsion fatigue data, i.e. cyclic torsion stress
amplitudes are multiplied with factor 3. Specimen failure is defined in cyclic torsion tests, when a
fatigue crack of minimum 300 Pm length has formed.
The aluminium alloy does not show a fatigue limit below 109 cycles in both loading
conditions. S-N data may well be approximated using power law functions of stress amplitudes
and cycles to failure. At the same equivalent stresses, cycles to failure in torsion tests are about
factor 2-10 higher than lifetimes measured in axial loading tests.
Cyclic tension compression tests of carbon steel deliver an S-N curve, which is approximately
parallel to the abscissa at numbers of cycles greater 107. Fatigue data produced by cyclic torsion
are comparable to the tension compression data using Mises equivalent stress. However, the
limited data available until now prohibits further analysis.
1124 H. Mayer and S. Stanzl-Tschegg

FIGURE 1. S-N data of Ck15 (circles) and 2024-T351 (squares) in cyclic torsion (closed symbols)
and cyclic tension compression (open symbols) at fully reversed loading condition.

Fatigue crack initiation is at the surfaces of the specimens in both materials in both loading
conditions whereas the crack paths are different. High cycle torsion fatigue of 2024-T351
produced fatigue cracks oriented in the directions of maximum shear stresses, i.e. in specimen's
length and circumferential direction, and cracks in length direction appeared first. Thus crack
propagation is in the direction of maximum shear stresses and the crack tip is loaded in cyclic
mode II. In contrast, cyclic axial loading of the aluminium alloy produced flat fracture surfaces at
the crack initiation places, which are perpendicular to the applied principal stress.
High cycle torsion fatigue of Ck15 produced crack paths, which are inclined by about 45° to
the specimen's axis, i.e. the fracture surface is perpendicular to the principal stresses. Within single
grains, however, the crack path follows one of the maximum shear stress directions. The
macroscopically inclined crack path is caused by the alternative growth of the crack in specimen's
length and circumferential direction. Fracture surface produced in high cycle tension compression
fatigue testing of Ck15 is macroscopically oriented perpendicular to the principal stress.
 22. New Investigations on Very High Cycle Fatigue of Materials 1125

MODELLING OF FATIGUE CRACK GROWTH FROM EXFOLIATION AND

PITTING CORROSION

G. Clark, P. K. Sharp and R. Jones1,2

Air Vehicles Division, Defence Science and Technology Organisation, 506 Lorimer Street,
Fishermans Bend 3207, Australia.graham.clark@dsto.defence.gov.au
1DSTO Centre for Structural Mechanics, Department of Mechanical Engineering, Monash

University, P.O. Box 31, Monash University, Victoria, 3800, Australia

2Cooperative Research Centre for Integrated Assett Management (CIEAM), Department of

Mechanical Engineering, Monash University, P.O. Box 31, Monash University, Victoria, 3800,
Australia
rhys.jones@eng.monash.edu.au

When structures are operated in corrosive environments, fatigue failures can occur from the
formation and propagation of cracks from exfoliation and pitting corrosion. Corrosion-nucleated
fatigue can be particularly problematic in structures that are not thought to be fatigue critical, and
are therefore not inspected, but can become critical in the presence of corrosion. Although the risk
factor of corrosion-nucleated fatigue should never be ignored, the most demanding corrosion-
related issue is the escalating maintenance burden caused by use of a “Find and Fix” corrosion
management policy. This “Find and Fix” policy exists largely because tools do not exist to
accurately assess the structural significance of corrosion when it is detected. Hence corrosion must
be treated as an immediate threat. The development of analytical tools capable of accurately
assessing the effect of corrosion on the durability of a structure would be a major benefit in that the
management philosophy could begin to transition to an “Assess and Monitor” framework. The
success of such a philosophy would greatly reduce unnecessary maintenance and defer other
actions to a more convenient time, such as when an aircraft is due for a heavy maintenance check.
Pitting and exfoliation are the most significant types of corrosion being considered by DSTO
for analytical tool development. In this paper, as in [1], pitting and exfoliation will be addressed
together because from a fatigue perspective, the mechanisms driving structural life for these two
damage types are very similar, see [1]. DSTO has developed a modelling approach for fatigue
from exfoliation based corrosion [1] based on the concept that the exfoliation process requires
development of pit-like intrusions into the substrate material, and that these can then be
represented in the fatigue model as if they are pits. This common feature then allows the use of a
single modelling approach to modelling cracking from both pitting and exfoliation (in aluminium
alloys).
In this context it should be noted that the work of C. G. Schmidt et al [2], and R. S. Piascik and
S. A. Willard [3] found that the growth of small corrosion fatigue cracks could be described by the
Frost Dugdale law [4], viz

L n a  EN  Ln(ao ) a a o e EN
or (1)
where N is the “fatigue life”, E is a parameter that is geometry, material and load dependent, a is
the crack depth and time N, a0 is the initial size of the defect. However, recent research [5] has
shown that the Frost Dugdale law [4], which can also be viewed as a fractal based growth law, can
be applied to a wide class of engineering problems resulting in an approximately exponential
relationship between crack size and number of cycles. Examining the open literature, we show
that the approximation presented in [3, 4] appears to be a reasonable fit to data on corrosion-
1126 G. Clark et al.

initiated fatigue crack growth. Fig. 1 presents data from [5], showing that there is near linear
relationship between the number of cycles and the log of the crack size. As a result a methodology
is presented that links NASTRAN with FASTRAN to automatically assess the effect of corrosion
on the structural performance of a component.

FIGURE 1 Experimental data presented in [5] for corrosion cracking

References
1. P. K. Sharp, T. B. Mills, and G. Clark, “Modeling of Fatigue Crack Growth from Pitting and
Exfoliation Corrosion”, International Committee on Aeronautical Fatigue, Toulouse, France,
(2001).
2. C. G. Schmidt, J. E. Crocker, J. H. Giovanola, C. H. Kanazawa, D. A. Shockey,
“Characterization of Early Stages of Corrosion Fatigue in Aircraft Skin”, DOT/FAA/AR-
951108, February 1996.
3. R. S. Piascik and S. A. Willard, "The Growth of Small Corrosion Fatigue Cracks in Alloy
2024," NASA Technical Memorandum 107755, NASA Langley Research Center, Hampton,
VA (April 1993).
4. N. E. Frost and D. S. Dugdale, “The propagation of fatigue cracks in test specimens”, Journal
Mechanics and Physics of Solids, 6, pp 92-110, 1958.
5. Barter S., Molent L., Goldsmith N. and Jones R., 2004, "An Experimental Evaluation Of
Fatigue Crack Growth", Journal Engineering Fracture Mechanics, 2004.
6. A.T. Kermanidis, P.V. Petroyiannis, S.G. Pantelakis, “Fatigue and damage tolerance
behaviour of corroded 2024 T351 aircraft aluminum alloy”, Theoretical and Applied Fracture
Mechanics, (2005).
 22. New Investigations on Very High Cycle Fatigue of Materials 1127

DOES COPPER UNDERGO SURFACE ROUGHENING DURING FATIGUE IN

THE VH REGIME?

Stefanie Stanzl-Tschegg, Hael Mughrabi1 and Reinhard Schuller

Institute of Physics and Materials Science
BOKU, University of Natural Resources and Applied Life Sciences,
Peter-Jordan-Strasse 82, A-1190 Vienna, Austria
1Institute of Materials Sciences, University of Erlangen-Nuernberg

Martensstrasse 5, D-91058 Erlangen, Germany

Stefanie.Tschegg@boku.ac.at, Reinhard.Schuller@boku.ac.at, Mughrabi@ww.uni-erlangen.de

Extensive studies have been performed on nature and formation of persistent slip bands (PSBs) in
the past. They have been identified as sites of localized cyclic plastic deformation, which are
typically found in fcc metals, like copper, nickel or solid solutions of these metals. These sites of
plastic deformation become visible as extrusions and microcracks (intrusions) on the specimen
surface and as „ladder“, “vein“ or cell structures of the dislocations in the interior of the specimen.
The stresses and plastic strains for their formation are determined by a plateau stress in the cyclic
stress strain curve (CSS curves) >1@ with a typical plateau stress at approximately 28 MPa and a
resolved shear strain between approximately 10-4 and 8x10-3 for copper single crystals. Additional
investigations have been performed at stresses below the plateau regime at stresses between 3 and
26 MPa and number of cycles up to several 107 in order to study the evolution of sliding and
dislocation characteristics >2@
In more recent time, the question on the fatigue behaviour and damage in the very high cycle
regime (VHCR) obtained increased attention, after it was detected that failure is possible at cyclic
stresses below the fatigue limit in steels after 109 or more cycles >3@. Mughrabi >4@, in a systematic
survey, discussed possible mechanisms for the existence or non-existence of a fatigue limit not
only in steels but other materials, like fcc metals. While inclusions were identified as sites of
internal crack initiation below the conventional fatigue limit of high strength steels (named type II
materials), shear band cracking in PSBs or grain boundary cracking by PSBs are the dominant
failure mechanisms at intermediate and lower amplitudes in the HCF regime for copper or nickel
or their solid solutions. A plastic strain fatigue limit has been defined as PSB threshold in a Coffin-
Manson diagram and has been transformed into a stress fatigue limit of an S-N diagram.
Below the PSB threshold, PSBs do not form and cyclic slip is reported to be more or less
homogeneously distributed. Mughrabi proposed that finite slip irreversibility remains, although
this irreversibility is much smaller at lower amplitudes. He assumed in a new model that failure
occurs, if the irreversible slip steps at the surface lead to a critical surface roughness, whereby
much smaller rates of its formation can be expected than for PSBs >5@ . The local plastic strain
amplitudes at stresses slightly below the PSB threshold value may be expected to be an order of
magnitude lower than in the PSBs and two or three orders of magnitudes lower than in the so-
called matrix structure in single-phase materials. Therefore, it may be expected that the number of
cycles required to develop the critical surface roughness in the absence of PSBs just below the
PSB-threshold will be one to three decades larger than those required for failure in the presence of
PSBs. Further reduction of the amplitude will finally lead to the so-called irreversibility limit,
below which cyclic slip will be reversible and therefore non-damaging. Thus, a multi-stage S-N
diagram has been predicted by Mughrabi for type I materials with the true fatigue limit at the
irreversibility limit, after extremely high numbers of cycles, like 109 or higher.
1128 S. Stanzl-Tschegg et al.

It is the aim of this study to characterize the response of copper on fatigue loading at cyclic
strain amplitudes below the threshold PSB limit at very high numbers of cycles. Polycrystalline
copper specimens with a diameter of 8 mm and a grain size of approximately 80 Pm were used
after mechanical plus electrolytic polishing. They were ramp loaded up to a total strain and stress
amplitude being slightly below the PSB threshold stress. At this stress, the specimen was then
fatigued with the ultrasonic technique at a testing frequency of 20 kHz up to 1010 cycles
The experiment was interrupted several times in order to investigate the resulting surface
appearance in an SEM and AFM. Owing to the sine strain distribution along the specimen axis in
the ultrasonic loading experiment, lower strains and stresses are present along the specimen length
in one specimen, so that systematic studies of all desired stress amplitudes are possible with one
specimen. Thus the induced surface appearance and roughness was determined in the centre of the
specimens, where maximum strains and stresses were selected such that they were slightly below
the PSB threshold, and in some specified distances apart from this, where defined smaller
amplitudes are present.
As first result, the distribution and extent of slip activity was determined for several defined
strain and stress values below the PSB threshold. Secondly, the strain and stress limit has been
determined, below which no cyclic slip traces are visible at the surface, i.e. where cyclic slip was
reversible and therefore non-damaging. Thirdly, it could be shown that the portion of areas
showing slip features depends on the applied number of cycles. Finally, surface roughness and
profiles have been determined for several specified strains and stresses.

References
1. H. Mughrabi, Mat. Sci. Eng., 33, 1978, 207-223)
2. L. Buchinger, S. Stanzl and C. Laird, Phil. Mag. A, 1984, No. 2, 275-298
3. T. Naita, H. Nueda and M. Kikuchi, Metall. Trans. 15A, 1431-1436
4. H. Mughrabi, Fatigue Fract. Engng. Mater. Struct. Vol. 22, No. 7, 1999, 633-641
5. H. Mughrabi, VHCF-3, Proc. Ehird Int. Conf. on VHCF, eds. T. Sakai and Y. Ochi, The Soc.
of Mat. Sci., Japan, 2004, 14-18
 22. New Investigations on Very High Cycle Fatigue of Materials 1129

CRACK INITIATION MECHANISM OF BEARING STEEL IN HIGH CYCLE

FATIGUE

T. Sakai
College of Science and Engineering, Ritsumeikan University
1-1-1 Nojihigashi, Kusatsu, Shiga, 525-8577 Japan
Hisashi Harada and Noriyasu Oguma
R&D Center, Koyo Seiko Co., Ltd.
24-1 Kokubu Higanjyo-cho, Kashiwara, 582-8588 Japan
sakai@se.ritsumei.ac.jp

In the last decade, fatigue behavior of metallic materials in very high cycle regime has been
coming out as an important subject to guarantee the safety of mechanical structures during the long
term service. The authors had developed special fatigue testing machines to perform the fatigue
tests effectively in both rotating bending and axial loading in gigacycle region. By using these
testing machines, long life fatigue tests were carried out for several kinds of metallic materials
including the high carbon chromium bearing steel(JIS: SUJ2).
Since a round-robin tests for this steel were performed as a common research project in the
Research Group for Statistical Aspects of Materials Strength[RGSAMS], a number of fatigue test
results were accumulated and analyzed. Among these results, typical examples of S-N property are
presented in this paper together with the X-ray diffraction pattern for the microstructures at the
crack initiation sites in the interior inclusion-induced fracture having the fish-eye.
Specimens and Experimental Procedure

Materials used in this study is the high carbon chromium bearing steel (JIS: SUJ2). Chemical
composition of this steel is provided in TABLE 1, and shape of the specimen is hourglass type with
the stress concentration factor of D 1.06 . These specimens were quenched (1108 u K 40min +
Oil quenching) and tempered (45 u 3K 120min + Air cooling). Hardness distribution over the
cross section of this material was almost flat and the average value is given as HV=730. A multi-
type fatigue testing machine in rotating bending was used in the present project. By using this
machine, four specimens can be simultaneously tested at the frequency of 3150rpm in the ordinary
room atmosphere without control of the temperature and humidity. S-N property of this steel in
gigacycle regime was first examined and fracture surface of every failed specimen was then
observed by means of SEM and the microstructures at the crack origins were observed by TEM. In
addition, the X-ray diffraction pattern was also examined for the microstructure at the crack
initiation site in the interior inclusion-induced fracture.
Experimental Results and Discussions
S-N diagram obtained in this program is shown in Fig.1, in which fatigue data for surface-
induced fracture are plotted by hollow marks and the data for interior inclusion-induced fracture
are plotted by solid marks, respectively.
S-N characteristics of this steel can be well explained as a duplex S-N curves consisting of S-N
curves for the respective fracture modes of surface-induced fracture and interior inclusion-induced
1130 T. Sakai

fracture. In this figure, the S-N curve for surface-induced fracture has the fatigue limit at V w
1270MPa, but another S-N curve for interior inclusion-induced fracture has no fatigue limit until
gigacycle regime.

FIGURE 1 S-N diagram for SUJ2 steel FIGURE 2 SEM image of FGA at center of the fish-
eye

Fine granular area(FGA) was necessarily found around the inclusion at the center of the fish-
eye as indicated in Fig.2. Microstructure of the longitudinal section at position in Fig.2 was
observed by TEM. The microstructure in thin layer of the FGA region was significantly
polygonized into the fine granular structure, although the microstructure at deep portion kept the
original microstructure. Image of SADP in such a thin layer with the fine granular structure was
examined in order to reconfirm such a polygonization. Thus, we obtained the typical diffraction
pattern forming some circular rings in the SADP image. But the pattern in the deep portion at FGA
is isolated spots governed by the Bragg’s condition. These results indicate that the microstructure
of thin layer inside the FGA becomes fine granular through the extensive polygonization of the
original microstructure. In other words, fatigue crack in the very high cycle regime takes place
through this polygonization and debonding of fine grain boundaries. Based on this mechanism,
morphology of the characteristic area of FGA becomes fine granular. In addition, it was found that
stress intensity factor range at the FGA front keeps almost constant of 4-6MPa m and this is
corresponding well to the threshold value for the fatigue crack propagation.
Conclusions
1 S-N characteristics of SUJ2 steel were well explained as duplex S-N property consisting of
S-N curves for the respective curves for surface induced fracture and interior inclusion-
induced fracture.
2 Fatigue cracks took place along the fine granular microstructure inside the FGA around the
inclusion at center of the fish-eye.
 22. New Investigations on Very High Cycle Fatigue of Materials 1131

VERY HIGH CYCLE FATIGUE BEHAVIOR OF HIGH STRENGTH STEELS

Yoshiaki Akiniwa, Nobuyuki Miyamoto1, Hirotaka Tsuru and Keisuke Tanaka

Mechanical Engineering, Nagoya University
Furo-cho, Chikusa-ku, Nagoya 464-8603 Japan
akiniwa@mech.nagoya-u.ac.jp
h044117m@mbox.nagoya-u.ac.jp
ktanaka@mech.nagoya-u.ac.jp
1Materials R&D Department, DENSO CORPORATION

1-1 Showa-cho, Kariya, 448-8661 Japan

NOBUYUKI_MIYAMOTO@denso.co.jp

Life extension of engineering plants and high-speed operation of engineering machines are
required in various fields. In these situations, conventional fatigue strength determined as the
strength for 107 cycles is not sufficient. It is necessary to adopt the data up to giga cycle regime for
design and maintenance.
In the present study, ultrasonic fatigue tests were conducted with smooth specimens of bearing
steels (JIS SUJ2, 746HV and 777HV), alloy tool steels (JIS SKD11, 707 HV and 710HV) and
martensitic stainless steels (661HV and 678HV). The fracture surface was examined with scanning
electron microscopy to identify the crack nucleation site.

FIGURE 1. S-N curves

Figure 1(a) shows the S-N curves for two batches of SUJ2. The open and solid marks indicate
the data of surface fracture and internal fracture, respectively. The fine solid and dotted lines
indicate the published data of the electro-polished smooth specimens obtained by rotating bending
fatigue operated at a frequency of 50 Hz [1,2,3]. The data of ultrasonic fatigue are close to those of
rotating bending fatigue, although they do not show a distinct two-step. For SUJ2-A, it is
interesting to note that the surface fracture takes place not only at short fatigue lives, but also at
very long lives. Figure 1(b) shows the data for SKD11. The S-N curve can be approximated by
linear line.
In our previous paper, the propagation process of fatigue cracks initiated from internal
inclusions has been divided into three regions [4]. The first region surrounding inclusions is a fine
granular facet area. The crack propagation in the facet region is called Stage A and the crack
1132 Y. Akiniwa et al.

propagation outside the facet region is Stage B. In these regions, the cracks propagate without
environmental effects of atmospheric air. When the fatigue cracks reaches the specimen surface,
the fatigue cracks propagates as surface cracks (Stage C). Since most of the fatigue life for the
VHCF regime is consumed in Stage A, the threshold condition of facet crack is very important.
Figure 2 shows the relation between facet size and inclusion size for various steels [5,6]. The facet
size increases with the inclusion size. When the fatigue limit is related to the facet size, the size can
be evaluated from the distribution of the inclusion size.

FIGURE 2. Relation between facet size and inclusion size

References
1. Shiozawa K, Lu L., Fatigue Fract Engng Mater Struct, Vol. 25, 813-822, 2002.
2. Goto M, Yamamoto T, Nisitani H, Sakai T, Gawagoishi N., J Soci Mat Sci Japan, vol. 49,
786-792, 2000.
3. Ochi Y, Matsumura T, Masaki K, Yoshida S., Fatigue Fract Engng Mater Struct, vol. 25,
823-822, 2002.
4. Tanaka K, Akiniwa Y., Fatigue Fract Engng Mater Struct; vol. 25, 775-784, 2002.
5. Miyamoto N., Asai H., Miyakawa S., Akiniwa Y. and Tanaka K., Trans Japan Soci Mech
Eng A , vol. 70,1080-1085, 2004.
6. Tanaka K, Akiniwa Y, Miyamoto N, Proc. 3rd VHCF,2004, 56-67.
 23. Deformation and Fracture of Engineering Materials 1133

FRACTURE TOUGHNESS OF HYDRIDED ZIRCALOY-4 EXPERIMENTAL

AND NUMERICAL STUDY

C. Langlade, P. Bouffioux1 and M. Clavel

MSSMat laboratory, Ecole Centrale Paris, Grande Voie des Vignes,
92295 Chatenay Malabry, France.
1EDF R&D/MMC, Site des Renardières, 77818 Moret s/ Loing Cedex, France.

langlade@mssmat.ecp.fr, clavel@mssmat.ecp.fr, pol.bouffioux@edf.fr

Zircaloy-4 is used as cladding tubes in pressurized water reactors (PWR). Their oxidization by
primary circuit water produces hydrogen. The tube picks up a fraction of hydrogen during
irradiation. When the solubility limit of hydrogen into zirconium is reached, zirconium hydride
platelets precipitation takes place. Hydrided stress-free platelets are oriented in the circumferential
direction of cladding (Fig. 2) thanks to its texture. If a sufficiently high tensile stress is applied in
the circumferential direction of cladding during hydriding process, ZrHx platelets are oriented in
the radial direction (Fig. 2). This process is called reorientation. In order to study hydride platelets
orientation effect on cladding fracture toughness, the pin loading test previously (PLT, Fig. 1)
proposed by Gregoriev et al. ([1]) has been involved. Furthermore a better understanding of this
test was allowed by a 3D finite element simulation.
Tests made on Zircaloy-4 samples show that pin loading tests are reproducible. Zircaloy-4
samples exhibit a steady crack growth. Fracture surface consists in a ductile tearing 45° tilted in the
sample thickness in agreement with the plane stress conditions.
Zircaloy-4 with circumferential hydrides and an amount of hydrogen around 300 ppm exhibits
very little change compared to Zircaloy-4 samples hydrogen free. Fracture surface consists in a
ductile tearing, roughly plane in the sample thickness, with smooth areas corresponding to broken
hydride platelets.
Zircaloy-4 with circumferential and radial hydrides and an amount of hydrogen around 150
ppm exhibits a brittle fracture. Fracture surface consists in smooth and plane areas which have the
size than radial hydride platelets clusters through which damage probably takes place ahead of the
crack tip.
Pin loading tests show that radial hydrides are much more deleterious than circumferential
ones. In order to understand this effect, a mesoscopic plasticity model implemented in Abaqus
finite element software by Hoc et al. ([2]) has been used for modelling hydrided microstructure.
This one is based on continuous dislocation density evolution.
In parallel, direct identification of parameters of the modelling have been done using TEM
observations.
1134 C. Langlade et al.

FIGURE 1. Pin Loading Test principle

FIGURE 2. Pin Loading Test principle

References
1. Grigoriev V., Josefsson B., Lind A. and Rosborg B., Scripta Met. et Mat., vol. 33, 109-114,
1995
2. Hoc T., Rey C. and Raphanel J. L., Acta materialia, vol. 49, 1835-1846, 2001
 23. Deformation and Fracture of Engineering Materials 1135

CRACK GROWTH BEHAVIOR IN A HIGHLY FILLED ELASTOMER

C. T. Liu, R. Neviere1 and G. Ravichandran2

AFRL/PRSM, 10 E. Saturn Blvd.
Edwards AFB CA 93524-7680
1SNPE, 91710 Vert-Le-Petit, France
2Graduate Aeronautical Laboratories, California Institute of Technology, Pasadena, CA 91125

Chi.liu@edwards.af.mil

An important engineering problem in structural design is evaluating its integrity and reliability. It
is well known that structural strength may be degraded during its design life due to mechanical or
chemical aging, or a combination of these two aging mechanisms. Depending on the structural
design, material type, service loading, and environmental condition, the cause and degree of
strength degradation due to the different aging mechanisms differs. One of the common causes of
strength degradation is the result of damage and crack development in the structure. Therefore, to
effectively use the material in structural applications one needs to understand the damage initiation
and evolution processes, the effects of damage and crack development on the material’s response,
and the remaining strength and life of the structures.
In recent years, a considerable amount of work has been done studying crack growth behavior
in particulate composite materials under different loading conditions at ambient pressure [1-4].
Experimental findings indicate that power law relationships exist between the crack growth rate,
da/dt, and the Mode I stress intensity factor, KI. These experimental findings support the theory
developed by Knauss [5] and Schapery [6] in their studies of crack growth behavior in linear
viscoelastic materials. It is known that classical fracture mechanics principles, especially linear
elastic fracture mechanics, are well established for single-phase materials. Experimental data
indicate that linear fracture mechanics theories can be applied to the particulate composite
materials with varying degree of success.
In this study, a series of experiments were conducted on uniaxial specimens with and without
pre-crack to investigate the constitutive and crack growth behavior in a highly filled elastomer,
hard particles embedded in a rubbery matrix. The specimens without pre-crack were tested at three
different displacement rates (5, 50, and 500 mm/min.) in a dilatometer. For the pre-cracked
specimens, a 6 mm crack was cut in the middle at one of the long edges of the specimen with a
razor blade, and they were tested at a displacement rate of 25 mm/min. During the crack
propagation test, a high-speed camera was used to monitor the test. In addition, the load and time
were also recorded. The raw data was used to determine stress, strain, dilatation, and Mode I stress
intensity factor for the onset of crack growth.
In addition to the experimental study, stress-strain response and crack initiation at room
temperature in the highly filled elastomer were analyzed using a modified version of a
phenomenological damage model developed earlier [7] and accounting for time dependence of the
homogenized material. Crack initiation is simulated in specimens with edge cracks subjected to
prescribed displacement rates at the boundaries. The relaxation and damage parameters determined
from the uniaxial stress-strain curves are used in the finite element calculations to update the local
moduli and the damage parameter. Crack initiation is determined by the attainment of critical
dilatation at a critical distance from the crack tip and is used as the failure criteria [8]. The
predicted and the measured critical value for the stress intensity factor (KIc) for the onset of crack
growth are 0.046 MPa m1/2 and 0.045 MPa m1/2, respectively. The good correlation between the
1136 C. T. Liu et al.

measured and the predicted KIc validates the critical damage based simulations on the onset of
crack growth.
In conclusion, experimental and numerical modeling results reveal that a good correlation
exists between the predicted and the measured KIC for the onset of crack growth.

References
1. Beckwith, S. W., and Wang, D. T., 1978, “Crack Propagation in Double-Base Propellants,”
AIAA, 78-170.
2. Liu, C. T., 1990, sCrack Growth Behavior in a Composite Propellant with Strain Gradients –
Part II,” Journal of Spacecraft and Rockets, 27, pp. 647-659.
3. Liu, C. T., 1990, “Crack Propagation in a Composite Solid Propellant,” Proceedings of the
Society of Experimental Mechanics, Spring Conference, pp. 614-620.
4. Liu, C. T., and Smith, C.W., 1996, “Temperature and Rate Effects on Stable Crack Growth in
a Particulate Composite Material,” Experimental Mechanics, 36(3), pp. 290-295.
5. Knauss, W. G., 1970, “Delayed Failure – The Griffith Problem for Linearly Viscoelastic
Materials,” International Journal of Fracture Mechanics, 6, pp. 7-20.
6. Schapery, R. A., 1973, On a Theory of Crack Growth in Viscoelastic Media, Report MM
2765-73-1, Texas A&M University.
7. Ravichandran, G., and Liu, C. T., 1995, “Modeling Constitutive Behavior of Particulate
Composites Undergoing Damage,” Int. J. Solids and Structures, 32, pp. 979-990.
8. Ravichandran, G., and Liu, C. T., 1998, “Crack-Tip Shielding in Particulate Composites
Undergoing Damage,” Engineering Fracture Mechanics, 59, 713-723.
 23. Deformation and Fracture of Engineering Materials 1137

CRACK TIP BEHAVIOR IN TIAL WHEN APPROACHING GRAIN BOUNDARY

Fu-Pen Chiang, Sheng Chang and Kai Wang

Mechanical Engineering Department
Stony Brook University
Stony Brook, NY 11794-2300, USA
Fu-Pen.Chiang@stonybrook.edu
Sheng.Chang@stonybrook.edu

Lamellar titanium-aluminum alloys (TiAl) show high stiffness, high density-normalized strength
and high fatigue resistance at elevated temperature. This is a promising substitute material for high
performance jet engines. TiAl has two phases: the J(TiAl) phase and the D2 (Ti3Al) phase. Grains
(or colonies) of TiAl are in lamellar form with alternating Jand D2 lamellae, but not necessarily in
equal numbers. The mechanical properties of this material depend on the grain size, lamella
orientation, etc. Intensive investigations have been carried out on the measurement of mechanical
properties of TiAl (Maruyama et al. [1], Appel et al. [2], and Cao et al. [3]). However, most of
them focus on the global mechanical properties of the material with either many grains together or
just one large single grain because there is a lack of measurement techniques capable of mapping
the local deformation at micro/nano scales. In this paper we introduce the SIEM (Speckle
Interferometry with Electron Microscopy) technique (Chiang et al. [4]) to map the full field
deformation around a crack tip in TiAl specimens under uniaxial tension.
SIEM is a micro/nano experimental mechanics technique that is able to perform full field
displacement mapping over a region of only several microns in size. The basic principle of SIEM
is as followings. A speckle pattern consisting of micro or nano sized random particles is first
deposited onto the specimen surface which is to be loaded inside the chamber of a SEM (Scanning
Electron Microscope). Load is applied to the specimen with the speckles following the
deformation. The speckle patterns are digitally recorded sequentially under incrementally applied
loads and compared using a specially designed algorithm called CASI (Computer Aided Speckle
Interferometry) (Chen and Chiang [5], Chiang et al. [6]). The result is a distribution of
displacement vectors (or displacement contours) representing the deformation. Appropriate
displacement-strain relations are used for the calculation of the strain tensor distribution. Length
scale and the technique’s sensitivity are determined by the speckle size.
Fig. 1 shows the shape and dimension of the specimen. We cut a notch on the side of the
specimen. Then a fatigue machine was used to initiate the crack tip. The specimen was under
uniaxial tension inside the SEM. The crack propagated either in a translamellar mode, cutting
across several Jand D2 lamellae, or along the D2/Jinterface depending on the lamellar orientation.
Fig. 2 shows that at the beginning of the load, the crack is almost perpendicular to the lamellar
orientation and it adapts the translamellar mode. Later, after the crack advances into another grain
whose orientation is about the same of the crack, the crack switches to mode II propagation as
shown in Fig. 3.
1138 F.-P. Chiang et al.

FIGURE 1. Uniaxial tension specimen with an initial crack.

FIGURE 2. Crack in translamellar mode. FIGURE 3. Crack along D2/Jinterface.

In a separate study we have shown that the stiffness of the interface can be as much as 7 times
stronger than that of the grains themselves. There is a noticeable resistance to crack growth across
grain boundaries. There are new cracks initiated in the next grain across the boundary before the
boundary is broken. The deformation fields around the crack tip are obtained by SIEM.
Acknowledgement
The authors acknowledge gratefully the financial support provided through the AFOSR grant
#FA95500410303 managed by Dr. Craig Hartley. The authors would like also to thank Dr.
Hartley and Dr. James M. Larsen of AFRL/MLLMN for their intellectual support for this project.

References
1. Maruyama, K., Suzuki, G., Kim, H.Y., Suzuki, M., and Sato, H., Materials Science and
Engineering A329-331,190-195, 2002.
2. Appel, F. and Wagner, R., Materials Science and Engineering R22, 187-268, 1998.
3. Cao, G. Fu, L., Lin, J., Zhang, Y., and Chen, C., Intermetallics, Vol. 8, 647-653, 2000.
4. Chiang, F.P., Wang, Q., and Lehman, F., In Nontraditional Methods of Sensing Stress, Strain
and Damage in Materials and Structures, ASTM STP 1318, edited by Lucas, G.F. and Stubbs,
D.A., American Society for Testing and Materials, 156-169, 1997.
5. Chen, D.J. and Chiang, F.P., Applied Optics 3(2), 225-236, 1993.
6. Chen, D.J., Chiang, F.P., Tan, Y.S. and Don, H.S., Applied Optics, 32(11), 1839-1849, 1993.
 23. Deformation and Fracture of Engineering Materials 1139

EFFECT OF LOADING RATE ON THE ENERGY RELEASE RATE IN A

CONSTRAINED ELASTOMERIC DISK

H. K. Ching, C. T. Liu1 and S. C. Yen

Materials Technology Center
Southern Illinois University, Carbondale, IL 62901, USA
1AFRL/PRSM, 10 E. Saturn Blvd., Edwards AFB, CA 93524-7680

hching@siu.edu

Defects such as flaws and cracks may form in elastomeric materials due to the manufacturing,
handling or ageing. To ensure the integrity and reliability for such structural components, fracture
toughness should be ascertained so that the onset of the crack growth can be determined based on
the fracture resistance of the material. The energy release rate is a measure of the fracture
toughness, and commonly used as a criterion to determine the maximum operating loads for a
given pre-existing defect.
We have performed finite element analyses to find out the variation of energy release rate with
the crack length for an elastomeric disc, which has a star-shaped hole located at the center,
subjected to the pressure and isothermal loads (Fig.1). The deformations of the disc are constrained
by a circular steel ring enclosing the disk. Numerical studies [1] have shown that that the value of
energy release rate increases with the increase of the crack length, reaches a maximum value at 1
in. of crack length, and then decreases gradually. It was also found that the value of energy release
rate strongly depends upon the material compressibility, and decreases sharply with the increase in
Poisson’s ratio for the pressure loading case. In this study, we delineate the effect of the loading
rate on the energy release rate. Due to the viscoelastic response of the elastomeric material, the
stress-strain relation highly depends on the applied loading rate. Here, three different uniaxial
stress-strain relations obtained from loading rates of 200ipm, 50ipm and 2ipm, respectively are
used in the present analysis (Fig. 2). The disk is simulated either as a linear elastic material under
the Hook’s law or hyperelastic material described by the Ogden strain energy potential. The
present investigation shows that the energy release rate increases with an increase in the loading
rates for the thermal loading case regardless of the material compressibility. For the pressure
loading case, the energy release rate increases with the decrease in the loading rate when the
material becomes more compressible while it decreases when the material approaches the
incompressible limit. In addition, with the presence of the crack surface pressure, the energy
release rate needs to be modified by including an additional line integral as shown in Eq. (1). The
contribution to the value of the energy release rate from the work done by the crack surface
pressure can thus be examined by plotting the crack opening deformation. Crack opening profiles
for different loading rates will be drawn.

§ wu i · wu
2
J ³ * ¨ Wn1  Tij n j wx ds ¸  ³ * p wx dx1
© 1 ¹ c
1
(1)
1140 H. K. Ching et al.

FIGURE 1. Specimen geometry

FIGURE 2. The stress-strain curves of the disk mateiral in tension under different loading rates

References
1. H. K. Ching, C. T. Liu and S. C. Max Yen, “FE calculations of J-integrals on Crack Length in
a Constrained Elastomeric Disc with Crack Surface Pressures and Isothermal Loads”, the
ASME International Mechanical Engineering Congress, Anaheim, California, November,
2004
 23. Deformation and Fracture of Engineering Materials 1141

ANALYSES OF PROGRESSIVE DAMAGE AND FRACTURE OF

PARTICULATE COMPOSITE MATERIALS USING S-FEM TECHNIQUE

Hiroshi Okada, Satoyuki Tanaka, Yasuyoshi Fukui and Noriyoshi Kumazawa

Kagoshima University
1-21-40 Korimoto, Kagoshima 890-0065, Japan
okada@mech.kagoshima-u.ac.jp

In this paper, the s-fem (s-version FEM; an element overlay technique [1,2,3,4]) is applied to the
damage analyses of particulate composite materials (Fig. 1 (a)). We conduct a series of unit cell
analyses in which we assume many (~40) distributed reinforcing particles. We assume two kinds
of damage modes in our analyses. One is due to microvoid formation in matrix material that is
accounted for by a continuum damage model [5]. The damage constitutive model is extended so
that the modes of damage can be split to dilatational and deviatoric parts. Another damage mode is
“dewetting” between the reinforcing particles and matrix material. The cohesive zone model [6,7]
is adopted to appropriately model the separation between particles and matrix material.
Present authors have been developing the s-FEM technique that can be applied to the analyses
of particulate composite materials [3]. In the s-FEM modelling for particulate composite material
[3], finite element models for structure or the region of unit cell as whole and a particle and its
immediate vicinity are built separately. Then, as shown in Fig. 1 (b) for a 2-D schematics, the finite
element model for a particle and its immediate surrounding is placed in the analysis region
repeatedly. Thus, a number of particles can be distributed without any difficulties even for 3-D
cases.
The s-FEM program [3] has been extended so that it can cope with elastic-viscoplasticity,
elastic damage [5], dilatational/deviatoric elastic damage and dewetting between particles and
matrix material using the cohesive zone model [6,7]. In the presentation, we discuss about some of
our recent results.

FIGURE 1. The schematics of particulate (a) composite material and (b) s-FEM model.

The formulation of s-FEM for elastic problems can be summarized as follows. The finite
element model representing the structure or the unit cell as whole is called “global mesh” or
“global model”. Ones that are for particles and their immediate vicinities are called “local mesh” or
“local model”. The displacements ui are expressed by the sum of those based on the global model
G LI LJ
and the local models where they overlap. For example, let ui , ui and ui be the displacements
that are expressed by the global and the I-th and J-th local models, respectively, and the
displacements at a point which is inside the I-th and J-th local models are written to be:
1142 H. Okada et al.

ui uiG  uiLI  uiLJ

(1)
Then, the displacements that are expressed like equation (1) are used in the statement of virtual
work principle.

 
³: Gui , j Eijk"uk ,"d: ³: Guibi d:  ³w: t Guiti d w:
t
(2)
Thus, we get discretized equation. If we replace the virtual work principle (2) by its rate
formulation, we can perform nonlinear analyses.
The cohesive zone elements are implemented in the local models only. We do not need to give any
special treatments to the cohesive zone elements, even they overlap with the global and the other
local models.
In Fig. 2, the distributions of stress in a unit cell model when an isotropic damage and dilatational
damage cases are employed and are compared to each other. They are the typical results that have
been obtained in present investigation. When the dilatational damage mode dominates, the damage
progresses much faster than the case of isotropic damage.

FIFURE 2. The distributions of damage parameters (a) Isotropic damage and (b) Dilatational
damage dominates.

References
1. Fish, J., Computers & Structures, vol. 43, 539-547, 1992.
2. Fish, J., Markolefas, S., Guttal, R., Nayak, P, Applied Numerical Mathematics, vol. 14, 135-
164, 1994.
3. Okada, H., Liu, C.T., Ninomiya, T., Fukui, Y. and Kumazawa, N., CMES: Computer
Modeling in Engineering & Sciences, vol. 6, 333-347, 2004
4. Okada, H., Endoh, S. and Kikuchi, M., Engrg. Fracture Mech., vol. 72, 773-789, 2005.
5. Simo, J.C., Ju, J.W., Int. J. Solids Structures, vol. 23, 821-840, 1987.
6. Needleman, A., J. Appl. Mech., vol. 54, 525-531.
7. Chandra, N., Li, H., Shet, C. and Ghonem, H. Int. J. Solids Structures, vol. 39, 2827-2855,
2002.
 23. Deformation and Fracture of Engineering Materials 1143

FRACTURE MECHANICS ON PVDF POLYMERIC MATERIAL : SPECIMEN

GEOMETRY EFFECTS

L. Laiarinandrasana and G. Hochstetter

Centre des Matériaux – Ecole Nationale Supérieure des Mines de Paris- UMR CNRS 7633
BP87 – F 91003 Evry Cedex – France
ARKEMA - Centre d'Etude de Recherche et de Développement (CERDATO)
F-17470 Serquigny - FRANCE
lucien.laiarinandrasana@ensmp.fr, gilles.hochstetter@arkemagroup.com

Polyvinylidene fluoride (PVDF) is a semi-crystalline polymer that has been widely studied for
structural applications, because it exhibits good mechanical properties and chemical resistance.
Mechanical properties investigations of the PVDF under study have been already published
elsewhere [1] where the macroscopic tensile and creep behaviour over several strain rate decades,
and over a large range of temperatures have been investigated. During viscoplastic deformation,
the material whitens after the onset of necking due to nucleation and growth of voids. In reference
[1] notched specimens and cracked specimens were used in order to investigate damage
development over a wide range of loading conditions. A numerical simulation of these tests is
performed using a modified Gurson-Tveergard-Needlman model [2-4], which was adapted to the
present material. The yield surface is expressed as:
2 def .V *
V eq

) V ,V * , f  V *2
§q V ·
 2 q1. f . cosh ¨¨ 2 kk 2
¸¸  1  q1 f
© 2V * ¹
2
0
(1)

V eq is the Von Mises stress, and V kk the trace of the stress tensor. f is the porosity, q1 and
q2 are model parameters that were introduced to improve the model predictions for periodic
arrays of cylindrical and spherical voids. q2 was expressed as a function of the maximum
principal plastic strain p1 .
The constitutive model was validated on Double Edge Notch Tensile (DENT) and Single Edge
Notch Bending (SENB) specimens. The numerical simulation well reproduces the observed
instability on DENT specimens and the stable crack growth on SENB specimens.
In the global approach of non linear fracture mechanics framework, the two-parameters
approach indicates that according to the opening stress state in the remaining ligament, the crack
growth can be more or less stable. Namely, tensile crack specimen such as DENT favours
instabilities whereas bending specimens (like SENB) are proved to present stable cracking. Even
the effect of specimen length of Single Edge Notch Tensile (SENT) specimens has been
investigated for metallic materials [5-6].
The aforementioned two-parameters approach is not often used on polymers. Nevertheless, it
seems that the same effects have been encountered on the present PVDF material. This work deals
with comparing the already used local approach of fracture on this PVDF material with the global
approach of fracture mechanics with single parameter J-integral, which is generally unable to
predict the in fracture plane stress effects. Note that K and G load parameters will be ignored since
the PVDF material under study does not exhibit any linear elastic mechanical response.
Q-stresses on DENT and SENB specimens are computed with the help of FE modelling and a
calculated Hutchinson Rice and Rosengren (HRR) field as reference stress. The material toughness
1144 L. Laiarinandrasana and G. Hochstetter

JIC is then determined by means of experimental data gathered with both specimens. If JIC depends
on the geometry, then, we have to find out whether a unique JIC-Q curve exists for the PVDF
material. In addition, the effects of specimen geometry on the resistance curve (crack propagation)
will be analysed.

References
1. Challier, M., Besson, J., Laiarinandrasana, L. and Piques, R. to be published in Engng. Fract.
Mech., 2005
2. Gurson, A. L., J Eng Mat Tech. 99:2-15. 1977
3. Tvergaard, V. J Mech Phys Solids. 30:399-425, 1982;.
4. Tvergaard, V. and Needleman, A., Acta Metall. 1984; 32:157-169.
5. Paris, P.C., Tada, H., Zahoor, A. and Ernst, H. Elastic-plastic fracture, ed J.D. Landes, J.A.
Begley and G.A. Clarke, ASTM STP 668, 1979, 5-36.
6. Paris, P.C. and Hutchinson, J. W. Elastic-plastic fracture, ed J.D. Landes, J.A. Begley and
G.A. Clarke, ASTM STP 668, 1979, 37-64.
 23. Deformation and Fracture of Engineering Materials 1145

FRACTURE TOUGHNESS OF ALLOYED AUSTEMPERED DUCTILE IRON

(ADI)

Olivera Eric, Dragan Rajnovic1, Zijah Burzic2, Leposava Sidjanin1 and Milan T. Jovanovic
Institute of Nuclear Sciences “Vinca”, Belgrade, Serbia and Montenegro
1University of Novi Sad, Faculty of Technical Sciences, Novi Sad, Serbia and Montenegro
2Military Technical Institute, Belgrade, Serbia and Montenegro

oliverae@vin.bg.ac.yu

Austempered Ductile Iron (ADI) has been established as an advanced material because of its
excellent mechanical properties based on a good combination of wear resistance, toughness and
very high strength combined with a relatively high ductility. It has a unique acicular matrix
structure that consists of high–carbon austenite and ferrite with graphite nodules dispersed in it >1@.
Some previous studies >1-4@ have been carried out on the mechanical properties of alloyed ADI.
However, very little information is available regarding the influence of microstructure on the
tensile properties and fracture toughness of alloyed ADI. The present investigation was undertaken
to examine the influence of microstructure on the tensile properties and the plane strain fracture
toughness (KIC) of copper and copper nickel alloyed austempered ductile iron (ADI). Two ductile
irons with chemical compositions (in wt.%): (a) 3.6C; 2.5Si; 0.28Mn; 0.04Cr; 0.45Cu; 0.01P;
0.014S; 0.066Mg, and (b) 3.07C; 2.15.Si; 0.26Mn; 0.04Cr; 1.6Cu; 1.5Ni; in both alloys balance
was Fe, were produced in a commercial electro-induction foundry furnace. Specimens for
mechanical testing were machined from the test section of the Y-block. Machined specimens
austenitized in a protective argon atmosphere at 900oC for 2h were rapidly transferred to a salt bath
at austempering temperature 350oC, held between 1 and 6h and then air-cooled to room
temperature. The reported results represent the average values of three measurements. Fracture
surfaces of specimens after tensile and fracture toughness tests were examined by a JOEL JSM-
6460LV scanning electron microscope (SEM) operated at 25kV. Metallographic investigations
were carried out by light microscope. Change in the volume fraction of retained austenite during
austempering was determined by the X-ray diffraction technique on diffractometer “Siemens D-
500” with Ni filtered CuK radiation.
The microstructure of as-cast Cu alloyed ductile iron consisted of graphite nodules surrounded
by ferrite in a pearlitic matrix. The as-cast microstructure of copper nickel alloyed ductile iron was
predominantly (over 95%) pearlitic. The nodules count in the lower section of the keel block of
both ductile irons was 100mm2. The microstructure of ADI alloyed with copper and copper nickel
austempered at 350oC for 2h revealed the characteristic microstructure of ausferrite. The
austempered microstructure was composed of ferrite (dark) and retained austenite (white) with
graphite nodules dispersed in the matrix. Figure 1a,b shows typical fractographs of copper ADI
and copper nickel ADI. The fracture surfaces of these specimens were mostly ductile showing
extensive micro ductility.
1146 O. Eric et al.

FIGURE 1 Fracture morphology after austempering at 350oC for 2h (SEM)

(a) copper ADI and (b) copper nickel ADI

Table 1 reports the fracture toughness, hardness and volume fraction of ferrite and retained
austenite of copper and copper nickel alloyed ADI. Values of fracture toughness satisfied all the
necessary conditions for ASTM E 399 and therefore are valid KIC values. The test results show
that the fracture toughness of copper and copper nickel alloyed ADI is lower than that of the
unalloyed ADI >1@ at the same hardness level. This indicates that the addition of alloying elements
like copper and nickel does not improve in ADI the fracture toughness. Bartosiewicz et al. >2@
studied the fracture toughness of two alloyed ductile irons austempered at four different
temperatures and found that fracture toughness was only a function of the volume fraction of
ferrite and retained austenite. Fracture toughness reached the maximum at a retained austenite
content 30vol.% in the matrix. Results of this paper support the literature data, i.e. copper ADI
with high amount of ferrite and retained austenite exhibits higher fracture toughness.

TABLE 1. Mechanical properties and volume fraction of ferrite and retained austenite after
austempering at 350oC

References
1. O.Eric, L.Sidjanin, Z. Miskovic, Zec, M.T. Jovanovic,Mat. Let, Vol.58, 2707-2711, 2004
2. Bartosiewicz, L., Krause, A. R., Kovacs, B. V., and Putatunda, S. K., AFS Transactions,
American Foundry Society, vol.92, 135-142, 1992
3. Bartosiewicz, L., Krause, A. R., Singh, I, and Putatunda, S. K., Materials Characterization,
vol.30, 221-234, 1993
4. Greech, M., Bowen, P., and Young, J.M., In Proceedings of The Second World Conference on
ADI, edited by Ann Arbor, MI, B. Kovacs, 1992, 135-148
 23. Deformation and Fracture of Engineering Materials 1147

PREDICTION OF CRACK GROWTH UNDER RANDOM LOAD IN RAILWAY

WHEEL

Rami Hamam, Sylvie Pommier and Frederic Bumbieler1

Lab. of Mechanics and Technology (LMT), 61 av. du Président Wilson, 94235 Cachan, France
1Agence d’Essai ferroviaire (AEF), 21 av. du Président Allende, 94407 Vitry sur Seine, France

hamam@lmt.ens-cachan.fr

Three aspects are indispensable to master railway systems: design, manufacture and maintenance.
Maintenance should define surveillance intervals which are defined at present by experience. The
aim of the French Railway Agency (SNCF) is to use fracture mechanic in order to satisfy to
European interoperability imperative of railway stock. One study here crack growth in wheel disk
under biaxial random load. An example of radial disk stress is shown in Figure 1; one can see the
random aspect of the load even during straight line. Classic crack growth models are not able to
predict propagation under this complex load.

FIGURE 1. Example of radial wheel stress the speed of the train is 100 km/h

An incremental crack growth model is employed for this problem [2]; this model is developed
to predict crack growth under any random load. We implement this model and we extend it to
biaxiality load [3]. Incremental model is divided in two laws:
• A cracking law, which is an instantaneous relationship between crack growth and crack tip

blunting: da D d U . This law contains only one adjustable material parameter D to be

dt dt
adjusted using one fatigue crack growth experiment.
• The second one is the blunting law, it gives the evolution of the crack tip blunting
according the Stress Intensity Factor (SIF), the material behavior and the geometry of the
dU
cracked structure : f  applied SIF, m aterial, geom etry  .
dt
This second law (blunting law) was established using finite element analysis. Figure 2 shows
for instance the evolution of the crack tip blunting for a Griffith crack. The blunting law describes
the evolution of these curves by system of time derivative equations; it is an elastic-plastic
constitutive behavior for the cracked structure which contains a set of material parameters that are
determined using the finite element method trough an automated identification protocol. One push-
pull experiment is necessary to identify the constitutive behavior of the material.
1148 R. Hamam et al.

FIGURE 2. Evolution of the crack tip blunting versus applied load

The model was integrated and applied on a few cases; it reproduces the main features of
fatigue crack growth, namely the fact that the crack growth rate obeys the Paris law in constant
amplitude fatigue (figure 3), the overload retardation phenomena and the sensitivity of the
retardation behavior to the overload ratio.

FIGURE 3. Crack growth rate for 3 different R ratio

References
1. Laird C, “The influence of metallurgical structure on the mechanisms of fatigue crack
propagation”, In Fatigue crack propagation, STP 415, (1967), pp. 131-68.
2. Pommier S and Risbet M, “Partial-derivative equations for fatigue crack growth in metal”, Int
Jal of Fracture. Vol 131. Num 1. Pages 79-106. 2005.
3. Hamam R., Pommier S., Bumbieler F., “Mode I fatigue crack growth under biaxial loading”
accepted in International Journal of Fatigue.
 24. Materials Damage Prognosis and Life Cycle Engineering 1149

PREDICTING THE EVOLUTION OF STRESS CORROSION CRACKS FROM

PITS

A. Turnbull, L. N. McCartney and S. Zhou

National Physical Laboratory
Queens Road, Teddington, Middlesex, TW11 0LW, UK
alan.turnbull@npl.co.uk

Predicting the evolution of stress corrosion cracks from pits and their subsequent growth has been
a major challenge because of the need to address the statistics of pit growth, the transition to a
stress corrosion crack, and the subsequent growth in the short and long crack domain. A purely
statistical approach has constraints in its predictive capability but several deterministic models for
pit and crack growth have been developed [1-4]. These all adopt the phenomenological
requirements proposed by Kondo [5], that the pit depth must be greater than a threshold depth and
that the crack growth rate should exceed the pit growth rate. Turnbull et al [5] developed a
deterministic model for the evolution of pits and the pit-to-crack transition and applied the
approach sucessfully to steam turbine disc cracking in simulated condensates with excellent
prediction of the time evolution of the pit distribution and of the percentage of pits that transform
to cracks at different pit sizes. The extension of this model to include the growth of cracks
following the pit-to-crack transition is now described.
It can be shown that the evolution of the pit size distribution function P(x,t) from an initial
stable population is governed by the partial differential equation

wP ( x, t ) w
 [ g ( x) P ( x, t )]
wt wx , (1)
where g(x) is the growth rate of the pit described by

dx
g ( x) ED 1 / E x (11 / E )
dt (2)
The crack growth rate is expressed by

dx
C Vp xq
dt (3)
for short cracks and is considered independent of crack size for deep cracks. The transition to the
experimentally measured growth rate for deep cracks is treated approximately by adopting a
critical crack size, the value of which is determined by fitting.
An example of the predictions from this model is given by Fig. 1.
1150 A. Turnbull et al.

FIGURE 1. Comparison of model prediction and experimental measurement of crack-size

distribution for turbine disc steel in 1.5 ppm Cl- at 90 qC after 9187 h exposure. The total depth
refers to the distance from the surface and includes the crack depth and the extension of the crack
beyond the pit base.

The ability to predict the crack size distribution so well is quite remarkable and very
encouraging. Nevertheless, in view of the sparseness of experimental data, extension to other
exposure periods is important to enhance confidence. There are limitations. The adoption of a
sharp transition in growth rates at some critical crack size is artificial; a more gradual transition is
to be expected. Furthermore, there is no independent measurement of crack growth rates in the
short crack regime. The latter presents a challenge in measurement.

References
1. Wei, R.P., In Proceedings of conference on Environmentally Assisted Cracking: Predictive
methods for risk assessment and evaluation of material, equipment and structures, edited by.
R.D. Kane, ASTM STP1401, ASTM, Pa, 2000, 3-19.
2. Engelhardt, G., Macdonald, D.D., Zhang, Y. and Dooley, B., Power Plant Chemistry, vol 6,
647-661, 2004
3. Engelhardt, G. and Macdonald, D.D., Corrosion Science, vol. 46, 2755-2780, 2004
4. Turnbull, A., McCartney L.N. and Zhou, S. ‘A deterministic model of pitting corrosion and
the pit-to-crack transition, submitted to Corrosion Science’, 2004.
5. Kondo, Y. Prediction of fatigue crack initiation life based on pit growth, Corrosion, vol. 45, 7-
11, 1989.
 24. Materials Damage Prognosis and Life Cycle Engineering 1151

CORROSION PROBLEMS IN NUCLEAR INDUSTRY : LESSONS LEARNED

AND PERSPECTIVES

J. M. Boursier, F. Foct, F. Vaillant and E. Walle

Electricité de France - R&D
Materials and Mechanics of Components Department
Site des Renardières
77818 Moret sur Loing Cedex.
jean-marie.boursier@edf.fr

In France, most of the electricity produced is of nuclear origin. To produce this energy, Electricité de France
has been operating for many years 56 pressurized water reactors with a permanent concern for safety efficiency
and availability. These reactors consist of a main primary system which includes the reactor vessel, the steam
generators, the pressurizer, and the reactor coolant pumps for flowing the coolant : the water. The heat
produced by the fission reactions is collected through the steam generator by the secondary system where the
steam produced is transformed into mechanical energy by a turbine which drives a generator producing the
electrical energy.
The components of nuclear power plants operate in harsh conditions. High operating temperature and
neutron radiation lead to a decrease of mechanical characteristics. The chemical media encountered inside the
installations can lead to corrosion phenomena. These phenomena occur through time, this is also called ageing.
Operating reactors safely is of major concern for Electricité de France. Therefore it is essential to have a
good knowledge of the scale, the seriousness and the development of identified damages. Some phenomena are
taken into account in the design: for example the embrittlement of the reactor vessel steel due to neutron
radiation is controlled through regular check to ensure that the evolution of damage remains within the initially
anticipated limits. On the other hand, others are unexpected and appears after several thousands hours of
operation: this is the case for many corrosion processes. In some cases, it is necessary to replace the damaged
component and to improve materials with a better resistance than those used previously. This is possible only
with a good understanding of the mechanisms which cause the damage. This understanding is gained thanks to
laboratory studies associated with the analysis of nuclear reactor feedback collected every year during controls
on site or assessment of faulty component.
For example, in the case of vessel head penetration stress corrosion cracking, the laboratory studies of
alloy 600 cracking revealed that this phenomenon has 3 successive phases : an initiation period which lasts a
few hundred hours and depends on the stress level, a slow propagation phase which corresponds to the longest
part of the test, and finally a 10 time faster propagation phase before the final failure. This last phase is not
always observed. When it is observed, only a few cracks, which have a critical size, reach this stage of
propagation. The “slow to fast” transition is still under investigation. It may be due to a stress intensity factor
threshold KISCC . From an industrial point of view, in the case of the adapters, as well as for other thick
products made of alloy 600, the first two phases : initiation and slow propagation are considered as one
“visible” initiation phase. The propagation is of interest when the cracks are deeper than a critical depth, which
corresponds to the shift to the fast propagation phase (typically a few hundred microns)
Stress corrosion susceptibility was studied in laboratory with various types of corrosion test in autoclave.
The main tests are constant load tests on tensile test specimens and constant deformation tests carried out on U-
bend type, or reverse U-bend type specimens. A classification relies on the definition of a stress corrosion
sensitivity index (inversely proportional to the cracking time) containing the following elementary indexes :
• A stress index is = k1s4 : in this equation deduced from laboratory tests, an evaluation of the stress
level of the component is introduced. The detrimental effect of surface strain hardening is taken into
account by using an effective stress which is a function of the depth of the strain hardened layer, of the
yield stress at the surface and inside and of the stress applied using the following formula.
1152 J. M. Boursier et al.

1

ª S 1 ,1 6
ac
 1, 9
º 3 ,16
V eff « 2
¬« K IS C C
V ap pliqu ée ³
0
LI (a) da »
¼»

with LI the integral width of the X-ray diffraction peak measuring the surface strain hardening, ac is the crack
critical depth depending on KISCC.
• A temperature index iT = k2.exp(-180 000 /RT), the activation energy is also determined from
laboratory tests.
• A material index iM defined during laboratory corrosion tests, either starting from the time to cracking
of 10% of the test specimens taken from the same material as the component, or, in a more
conservative way, starting from the initiation time of the first crack observed.
The cracking time (tf) of the component in operation conditions is determined using the equation:

10000
t f (h)
iV i T i M

The propagation kinetics of stress corrosion cracks in thick alloy 600 has been studied in laboratory in
order to determine the effect of the main parameters (temperature, stress intensity factor, material parameters).
Knowing their influence can help to improve the maintenance policy of the various components made of alloy
600 by optimizing their inspection time interval. These studies have been the subject of coordinate research
programmes between European laboratories. Propagation rates of stress corrosion cracks were determined in
primary system conditions using fracture mechanics test specimen, precracked by fatigue and in some cases
equipped with a DCP monitoring of the crack. It depends on the temperature with an activation energy of 130 ±
20 kJ/mol. The fast propagation rate can be expressed as a function of the stress intensity factor K using a law
da/dt = b (KTinitial – KISCC)g For thick products like adapters, g is equal to 0,3 and the threshold KISCC is close
to 9 MPa÷m. The rates obtained are consistent with the CGRs measured in service using NDE techniques (4
mm/cycle).
Another subject, which illustrates the contribution of R&D to the understanding and the fight against
corrosion, is in nuclear waste storage. Today in France, no long-lived high-level radioactive waste storage
site is defined. There is a big difference between the maintenance of industrial facilities over a few decades (30
to 60 years) and storage where the time scale is a few thousand years. The container properties and the
surrounding medium are essential parameters which lead to an original approach to the fight against corrosion.
It is important to emphasize that between laboratory tests (which are most of the time limited in scale and
time) and the reality of millennial storage, corrosion test can be and have been carried out over intermediate
time periods and scales. They have been carried out in underground laboratories under conditions
representative of those expected during storage. The representative is not only at the level of the geological
environment, but also of the temperature and the irradiation by including in the system heating facilities and
irradiation sources. This is done in Mol (SCK-CEN, Belgium) where in situ exposure tests are being carried out
for several years. The time duration of the test can be as long as 5 years. These tests enable the validation of
laboratory tests as well as the models developed, despite the fact that these durations are still short in
comparison with the thousand years required.
Finally futures challenges for the Nuclear industry will be highlighted. Particularly, Very High Temperature Reactors
(VHTR) have been selected in the framework of the Generation IV roadmap for the capability to reach high yields for energy
production and for the possibility to produce hydrogen by mean of a water decomposition process. The high levels of
temperatures reached in the reactor - i.e. more than 950 °C - arises the question of structural materials of this kind of concepts.
Nickel based alloys are the most promising materials for the conception of the reactor. Apart from very important issues like
thermal ageing or mechanical resistance of these kind of materials, their compatibility with the impure helium environment of
the primary circuit – i.e. helium with traces of CH4, CO, H2O - remains to be demonstrated.
That’s the reason why we defined an R&D program, with the aim to define :
• Chemical specifications of the helium atmosphere to minimise the interaction with materials.
• The maximum temperature of the materials
• The nature and the kinetics of the degradation process.
 24. Materials Damage Prognosis and Life Cycle Engineering 1153

ALUMINIUM ALLOYS FATIGUE EVALUATION METHOD

S. Rymkiewicz
Gdansk University of Technology
Ul. G. Narutowicza 11/12, 80-952 Gdansk, Poland
srymkiew@pg.gda.pl

Fatigue and corrosion-fatigue investigations are usually very expensive and time-consuming.
Energetic criterion implemented in calculations makes it much cheaper and quicker to obtain the
results. They are based on the statement of possible implications between fatigue and non-
reversible dissipated energy during test of static loading. Assumption of accumulated fatigue
deformation energy equal to static tensile test cracking energy, even approximately is in many
cases sufficient [1]. Dependence between the total dissipated energy and the number of cycles to
failure can be expressed by the equation:

Nf
¦ Di const (1)
1

where Di demonstrates dissipation at the i cycle, and Nf is the number of cycles to failure. Identical
to equation (1) is the expression:

Nf
W ¦ Di (2)
1

where W is the critical failure energy. C.Feltner and J.Morrow [4] accepted, that energy W is equal
to the critical strain energy U in the static tension:

2kN f m 1
W U Va m (3)
m 1
where k and m are the material constants calculated from the static tensile test, and Va is the
calculated amplitude of cyclic loading. Calculated fatigue strength in comparison with
experimental data is usually compatible with the 103 – 106 number of cycles range [1].
W.T.Troszczenko and L.A.Hamaza [1] assumed, that total dissipated energy is equal to:

V rzz 4
W U( )
Va (4)
and after transformation they achieved:

1
log V a log V rzz  log 2 N f
4m (5)
Coefficient m for aluminium alloys shows equation [1]:

m 15  4  0.675 log 2 N f
(6)
The final mathematical expression is:
1154 S. Rymkiewicz

1
log V a log V rzz  log 2 N f
15  0,675 log 2 N f (7)

where: Vrzz - intensity of stress at the time of breaking the sample, Nf - number of cycles to failure
and Va - amplitude of cyclic loading.
The W.T.Troszczenko and L.A.Hamaza calculation procedure has been implemented in the
experimentally obtained data calculations.
The AlZn5Mg alloy has been selected for the research [2]. Standardized samples have been
prepared for the tension tests. The parameters like fracture energy, time to failure, reduction in area
in fracture zone and maximal strength to rupture have been determined in slow strain rate tests.
The results are shown in the Table 1.

TABLE 1. Slow strain rate test results (mean values).

Fatigue limit and corrosion fatigue limit have been calculated according to the formula (7).
The obtained results can be considered as mere estimates due to the method of calculation applied
for fatigue and corrosion fatigue limit based on energy criterion expressed in the formula by W.T.
Troszczenko and L.A. Hamaza.

References
1. Kocanda, S., Szala, J., Podstawy obliczen zmeczeniowych, PWN, Warszawa, Poland, 1997.
2. Zielinski, A., Evaluation of operating properties of aluminium alloys for marine applications,
Polish Academy of Sciences, Branch in Gdansk Marine Technology Transactions, Vol. 8,
233, 1997.
 25. Mixed-Mode Fracture 1155

SINGULAR STRESS FIELDS SITUATIONS IN MODE-II AND MIXED-MODE

LOADED CRACKS

Daniel Fernández-Zúñiga, Jörg F. Kalthoff1, Antonio Blázquez2 and Alfonso Fernández-Canteli

EPSIG, University Oviedo, Campus de Viesques, 33203 Gijón, Spain.
afc@uniovi.es
1Dept. of Experimental Mechanics, Ruhr-University Bochum. Universitätsstr. 150,

44780 Bochum, Germany. Joerg.F.Kalthoff@ruhr-uni-bochum.de

2Dept. of Civil Engineering, Materials and Construction, University of Málaga, Plaza El Ejido s/n,

29013 Málaga, Spain. abg@uma.es

The stress fields at initiation of a kinked (daughter) crack starting from a (mother) crack subjected
to mode-II loading conditions and the influence of these stress fields on the instability condition is
investigated. In the literature, often the stress distribution along the prospective propagation
direction of the daughter crack (before the daughter crack is actually formed), is considered and
postulated to be the actual driving force for the crack formation [1,2]. Special emphasis is given in
this paper to the changes the stress fields experience in the transition phase from before to after the
kinked daughter crack is initiated; in addition, the situation is studied for growing lengths of the
daughter crack, from very small crack lengths approaching zero till medium sized crack lengths. It
is found that, the conventional consideration of the stress distribution along the prospective path of
the daughter crack represents a pseudo-mode-I stress distribution.
This pseudo-mode-I situation shows a typical circumferential stress singularity but not a radial
stress singularity. Numerical calculations are performed to investigate this discrepancy. The
calculations show that as soon as the daughter crack is formed, a singular crack tip stress field
according to the regular mode-I loading situation results, showing a radial stress singularity in
addition to the circumferential stress singularity (see Fig. 1 and Table 1). This is the case even for
vanishingly short lengths of the daughter crack. These different stress characteristics and the
consequences are discussed with the aim of finding an appropriate model for describing
realistically the instability process of mode-II loaded cracks for the stage from before to after the
kinked crack is formed. The stress field that builds up at the corner notch after the kinked daughter
crack initiated is also analyzed for comparison. The BEM, due to its inherent high spatial
resolution [3,4], is used as an adequate technique for this kind of numerical investigation. The
validity of the numerical data is confirmed by experimental results using shadow optical caustic
and photoelastic techniques [5].

FIGURE 1. Stress intensity factors a) for the daughter crack, b) for the crack noctch and c) energy
release rate as a function of the length of the daughter crack, aD
1156 D. Fernandez-Zuniga et al.

TABLE 1. Stress intensity factors, K IDa Da

- and K Ir of the daughter crack, after its formation,
normalized to the corresponding data of the mother crack.

* analytical results from Amestoy and Leblond [6].

References
1. Erdogan, F., Sih, G.C., J. of Basic Engineering.Transactions of the American Society of
Mechanical Engineering, ASME, 519-527, 1963.
2. Richard, H.A., Materialprüfung, Vol. 45, 513-517, 2003.
3. Blázquez, A., Kalthoff, J.F., Palomino, I., Fernández-Canteli, A., In Proceedings of the
ICCES 04, Edited by S.N. Atluri and A.J.B. Tadeu, Madeira, 1783-1790, 2004.
4. Blázquez, A., Manti, V., París, F., Cañas, J., Int. J. of Solids and Structures, Vol. 35, 24,
3259-3278, 1998.
5. Podleschny, R., Kalthoff, J.F., In Proceedings of the Tenth European Conference on
Fracture, Structural Integrity: Experiments, Models and Applications, 211-221, 1994.
6. Amestoy,M., Leblond, J.B., Int. J. of Solids Structures, Vol. 29, No. 4, 465-501, 1992.
 25. Mixed-Mode Fracture 1157

EVALUATION OF M-INTEGRAL FOR RUBBERY MATERIAL PROBLEMS

CONTAINING MULTIPLE CRACKS

J.-H. Chang and D.-J. Peng

Department of Civil Engineering, National Central University, Chungli, Taiwan
t320001@cc.ncu.edu.tw

An energy parameter Mc based on the concept of the M-integral ([1]) is proposed for describing
the fracture behavior corresponding to a 2-D multi-cracked rubbery mechanical system under the
action of large deformation. The parameter is defined by originating the coordinate system at the
geometric center of the all the cracks, and then conducting the integration along a contour
enclosing all the cracks. With such a definition, Mc appears to be a problem-invariant parameter.
For problems containing rubbery materials, the aforementioned contour integral is applied with
state variables reinterpreted with respect to this reference configuration when the body is subjected
to large deformation. By definition, the integration is performed by taking the limiting case in
which the portions of contours are shrunk onto the crack tips and lying along the crack surfaces.
Note that, while the value of M varies with respect to different selections of origin O, the result of
Mc appears to be invariant for a given problem. Attention is addressed to discussion of the physical
meaning of Mc. It is shown that Mc can alternatively be related to (twice) the surface energy
corresponding to formation of the cracks ([2]).
In the special situation where the crack in an infinite body is subjected to a far-field uniform
loading system such that the nominal stress field is homogeneous, it is observed that the M-integral
then becomes equivalent to Mc. In such a case, the value of M appears to be dependent upon the
stress state and the crack length, but remains unchanged for different selections of the origin O of
the coordinate system.
Due to path-independence (i.e., the integration contour can be arbitrarily chosen, as long as
they contain the same set of cracks and no other singularity), the complicated singular stress field
in the near-tip areas is not involved in the calculation of the Mc-integral. Based on this
characteristic, it is therefore suggested that Mc be practically used as a fracture parameter for
describing the degradation of material and/or structural integrity caused by irreversible evolution
of cracks and/or other defects in rubbery media.
The hyperelastic material behavior in this proposed research is modeled with an Ogden-type
isotropic incompressible strain energy density function. In the numerical calculations, quadratic
finite elements are to be used for interpolation of the displacement field. Also, non-conforming
linear elements with reduced integration are to be used for the pressure interpolation in order to
satisfy the discretized LBB condition.
The applicability of Mc is illustrated in the following example, where a plane strain test
specimen modeled in terms of Ogden strain energy density function ([3]) and containing N parallel
cracks is considered. The deformed finite element mesh is shown in Figure 1, where the
undeformed configuration is depicted in dashed lines. By arbitrarily choosing a remote contour
(e.g. Figure 2), the results of the M-integral with respect to various coordinate origins are shown in
Tables I .
1158 J.-H. Chang and D.-J. Peng

Figure 1 The deformed mesh of the specimen.

Figure 2 The integration contour.

Table I The results of M wrt different D's(Pa-m2).

References
1. Budiansky, B. and Rice, J. R.. ASME, J. Appl. Mech., vol. 40, 201-203, 1973.
2. Chang, J.H. and Chien, A.J. Int. J. Fract. vol. 114, 3, 267-289, 2002.
3. Chang, J.H. and Chen, C.B. Fin. Elem. Anal. Des., vol. 28, 151-163, 1997.
 25. Mixed-Mode Fracture 1159

USE OF A CRACK BOX TECHNIQUE FOR CRACK BIFURCATION IN

DUCTILE MATERIAL

D. Lebaillif, X. B. Zhang1 and N. Recho1

GIAT, Bourges, France
1Blaise Pascal University of Clermont II, LaMI, France

d.lebaillif@giat-industries.fr

In this paper, an automatic Crack Box Technique (CBT) [1] is used to perform fine fracture
mechanic calculations with iterative remeshings in elastic-plastic materials found in literature.
Two types of experiments are analysed i.e. experiments carried out by Tohgo and Ishii [2] and
those done by Li, Zhang and Recho [3].
In reference [2], experiments are based on single-edge-cracked specimens subjected to bending
moment and shearing force, on aluminium alloy 6061-T651 (hardening exponent n|7). 2 three-
point-bending and 3 four-point-bending specimens have been tested, from pure mode I (beam A)
to pure mode II (beam E) (see table 1). Experimental results give two critical fracture toughnesses
: JIC=14N/mm and JIIC=46N/mm. Furthermore they show that cracks in beam A to C seem to
initiate for mode I predominant loading (T: tensile type fracture) whereas cracks in beam D and E
initiate for mode II predominant loading (S: Shear type fracture).

TABLE 1 : mixed mode in elastic calculation

Plasticity calculations using CBT are based on small strain assumption and were carried out
using ABAQUS [4]. The refined crack tip mesh is shown in fig. 1.

FIGURE 1 : crack box mesh inserted in automatic Delaunay triangulation

The two associated J-integrals J*I and J*II are then computed accurately to determine the
plastic mixity Mp parameter. It’s transition value between T and S type fracture is obtained from
the experimental results which give the critical mixity parameter MPc = 0.75. Using ABAQUS [4],
the calculated Mp is compared to MPc in order to obtain the bifurcation angle T for each beam.

The tensile or shear type fracture prediction for all 5 beams are presented in table 2. Mp are
given for two first steps of propagation to check that the crack keeps following either maximum
circumferential stress criterion or slip band criterion. These predictions appear to be in good
agreement with experimental observations.
1160 D. Lebaillif et al.

TABLE 1 : MP results and tensile/shear prediction

Finally, in reference [3], the static tests were conducted on 4mm thickness and 90 mm width
CTS (Compact Tension Shear) specimens. A fatigue pre-crack was introduced up to a/w|0.6. The
specimens were tested under 0 to 90° loading with respect to the crack axis. The specimen is made
of the fine grained structural steel StE550 with a critical mixity parameter Mpc about 0.77.
The numerical results of T-type crack growth are close to the experimental data. On the other
hand, the S-type crack growth angles calculated numerically follow a different trend, with respect
to the experimental values, as opposed to Togho’s specimen.

References
1. Lebaillif D., Recho N. Crack Box Technique Associated to Brittle and Ductile Crack
Propagation and Bifurcation Criteria. ICF11-the 11th International Conference of Fracture.
March 20-25, 2005; Turin, Italy
2. Tohgo K. and Ishii H. Elastic-plastic fracture toughness test under mixed mode I-II loading.
Eng. Fracture Mech. 41, 529-540, 1992
3. Li J., Zhang X.B. and Recho N. J-Mp based criteria for bifurcation assessment of a crack in
elastic-plastic materials under mixed mode I-II loading. Engineering Fracture Mechanics.
2004; 71: 329-343
4. ABAQUS Software Version 6.4.1. www.hks.com.
 25. Mixed-Mode Fracture 1161

MIXED MODE FRACTURE OF LINEAR ELASTIC MATERIALS WITH CUBIC

SYMMETRY

D. E. Lempidaki, N. P. O’Dowd and E. P. Busso

Department of Mechanical Engineering, Imperial College London,
South Kensington Campus, London SW7 2AZ, UK
{d.lempidaki; n.odowd; e.busso}@imperial.ac.uk

In order to analyse flawed structures, it is necessary to understand the behaviour of cracks under
mixed-mode loading. A crack experiences mixed mode conditions (combined Mode I and Mode
II) when it is subjected to remote combined tension and shear or when the crack is tilted with
respect to one of the material axes or with respect to the loading axis. Under mixed mode loading,
it has generally been observed that a crack in a homogeneous material will branch out of its initial
plane. A number of criteria have been proposed to explain this behaviour, the most common of
these being the maximum hoop stress criterion.
Suresh et al. [1] provided the mixed mode fracture locus for isotropic materials. However, in
order to obtain the corresponding result for anisotropic materials, the anisotropic crack tip fields
are needed. The general anisotropic problem has been studied by Sih and Liebowitz [2]. More
recently analytical solutions for orthotropic materials have become available, Liu et al. [3]. These
solutions were phrased in terms of two dimensionless factors, originally introduced by Suo [4]. For
the special case of materials with cubic symmetry (three perpendicular axes of symmetry) the
crack tip fields depend only on one parameter, namely ,

2 s12  s66
U ,
2 s11s22 (1)

where sij are the usual terms in the linear elastic material stiffness matrix.
A common definition of mode mixity is,

2 §K ·
M tan 1¨¨ II ¸
¸
S © KI ¹ (2)

with M = 0 corresponding to pure Mode I loading and M = 1, pure Mode II loading. To obtain the
mixed mode fracture toughness behaviour using the hoop stress criterion, the angle,, ,at which the
maximum hoop stress is attained must be determined, i.e.,

wV TT df I (T , U ) df (T , U )
0 Ÿ KI  K II II 0
wT dT dT (3)

where fI and fII are the angular crack tip distributions which depend on U. The mixed mode locus
can then be phrased in terms of Uand the Mode I fracture toughness, KIC, as

KI K
f I (T , U )  II f II (T , U ) 1
K IC K IC (4)

The dependence of the mixed mode crack tip fields for a cubic material on the parameter has
been studied in Lempidaki et al. [5]. The analytical crack tip field solutions from [3] have been
compared with finite element solutions for a range of -values and excellent agreement was
1162 D. E. Lempidaki et al.

obtained. Based on these solutions and using the maximum hoop stress criterion, the predicted
crack initiation angle, as a function of mode mixity, and the corresponding mixed mode fracture
toughness locus have been obtained. The results are shown in Fig. 1, which illustrates the
dependence of the crack angle and mixed mode fracture locus on . It can be seen in Fig. 1(a) that
for M = 0 (Mode I) T 0$ regardless of the value of , whereas for M =1 (Mode II) the value of
increases with increasing . From Fig. 1(b) it can be seen that as the value of decreases, starting
from the isotropic case ( = 1), the predicted fracture toughness locus expands from the value of KII/
KIC =0.87 to a value of KII/KIC approximately equal to 1.02 for = –0.5.

FIGURE 1. (a) Predicted crack initiation angle as a function of crack tip mixity parameter, M, and
(b) predicted mixed mode fracture toughness locus using the maximum hoop stress theory.

References
1. Suresh S., Shih, C.F., Morrone, A. and O’Dowd, N.P., J. Am. Ceram. Soc., vol. 73, 1257-
1267, 1990
2. Sih G.C. and Liebowitz, H., In Fracture: an Advanced Treatise, edited by H. Liebowitz, Vol.
2, Academic Press, New York, 1968, 67-190.
3. Liu, C., Rosakis, A.J., Ellis, R.W. and Stout, M.G., Int. J. Fracture, vol. 90, 355-382, 1998
4. Suo Z., J. Appl. Mech., vol. 57, 627-634, 1990
5. Lempidaki, D.E., O’Dowd, N.P. and Busso, E.P. in Proceedings of the Fifteenth European
Conference of Fracture, edited by F. Nilsson, ESIS, Stockholm, 2004
 25. Mixed-Mode Fracture 1163

THREE-DIMENSIONAL EXPERIMENTAL AND NUMERICAL SIFS AND

CRACK GROWTH

Dan Mihai Constantinescu, Bogdan Bocaneala and Liviu Marsavina1

Department of Strength of Materials, University “Politehnica” of Bucharest,
Splaiul Independentei nr. 313, Bucharest, Romania
1Department of Strength of Materials, University “Politehnica” of Timisoara,

B-dul Mihai Viteazul nr.1, Timisoara, Romania

dancon1@form.resist.pub.ro, bbocaneala@form.resist.pub.ro, msvina@mec.utt.ro

The desire to establish a three-dimensional framework for use in analyzing problems of stable
mixed-mode crack propagation has received lately a considerable attention. The study of
predicting crack paths under the most general possible hypotheses (three dimensions, arbitrary
geometry of the body and of the crack, arbitrary loading) is an ambitious objective. Recently,
Leblond [1] and Leblond et al. [2] established formulae which specify the general functional form
of the successive terms of the expansions of the SIFs along the front of the extended crack and, as
they underline, the formulation of the propagation criterion is an open problem only in the
presence of mode III. Therefore, in a mode I+II situation, the widely accepted “principle of local
symmetry” of Goldstein and Salganik [3] receives a general recognition. In numerical experiments
the FRANC3D code uses new concepts with a model that allows for the implementation of 3D
crack growth mechanics [4] with the support of both finite and boundary elements. Keeping in
mind such beneficial developments, one should emphasize that for a specific geometry and loading
we may obtain various experimental crack paths which are dependent on the local position of the
initial crack, as any deviation from “symmetry” influences the future crack trajectory.

FIGURE 1. Tested model geometry.

Results obtained by Smith et al. [5,6] from experimental tests using frozen stress
photoelasticity gave very interesting insights on the complicated processes of crack growth
trajectories. The geometry of the model studied here is shown in Fig. 1 and represents a motor
grain configuration. A model contained two starter cracks, a symmetric and an off-axis crack
separated by an uncracked fin to avoid any interference. The off-axis crack (located at the
coalescence of the two radii of 1,3 mm and 11 mm ) can be inclined (as shown in Fig. 1) or
straight-in – parallel with the fin axis. After inserting the starter cracks by striking a shaft with a
blade at the end held normal to the inner fin surface, the models were capped with RTV rubber
caps which were glued with PMC-1 adhesive and were subjected to the stress freezing cycle under
internal pressure. Thus optical data (isochromatic fringes) can be converted into Mode I and Mode
1164 D. M. Constantinescu et al.

II SIFs, if the last one exists. All symmetric cracks remained in the plane of the axis of symmetry
and grew as semi-elliptic cracks. Studies of 11 tests on the motor grain geometry [5,6] covered
cracks in different locations with projected a/c values (crack depth/half length of crack in fin tip
surface) of 0,5 to 0,9 and a/t values from 0,2 to 0,6 (t = 37,08 mm is the cylinder wall thickness at
fin tip). It appears that both SIF values and crack geometry during growth are quite variable due to
shear modes for the off-axis inclined cracks. On the other hand, the off-axis straight-in cracks
situated parallel to the fin axis are quite predictable in their growth, as they tended to grow much
more readily than the off-axis inlined cracks.
Numerical 3D analyses were done by using the FRANC3D code [4], modelling half of the
geometry (Fig. 2) and applying the internal pressure at which stresses were frozen in the
experiment. FRANC2D/L results obtained under a plane strain assumption, and 2D and 3D track
trajectories are compared and discussed. Only as an example, results for a symmetric crack which
propagated in depth up to a/t = 0,28 are shown in Fig. 3 by comparing experimental and
numerical normalized SIFs in the middle slice and at 20º from the surface of the fin.

FIGURE 2. Numerical 3D model FIGURE 3. Normalized SIFs for a/t = 0,28

Off-axis cracks are retarded in growth by shear modes (II or II and III) and eventually gain
their local and global symmetry.

References
1. Leblond J.B., Int. J. Solids and Structures, vol. 36, 79-103, 1999.
2. Leblond J.B., Lazarus V. and S. Mouchrif S., Int. J. Solids and Structures, vol. 36, 105-142,
1999.
3. Goldstein R.V. and Salganik R.L., Int. J. Fracture, vol. 10, 507-523, 1974.
4. Carter B.J., P.A. Wawrzynek P.A. and Ingraffea A.R., Int. J. Numer. Meth. Eng., vol. 47, 229-
253, 2000.
5. Smith C.W., Constantinescu D.M. and Liu C.T., In Proceedings of the 2002 SEM Annual
Conference& Exposition on Experimental and Applied Mechanics, Session 36, Experiments
in Fatigue & Fracture, SEM, June 10-13, Milwaukee, Wisconsin, 2002, Paper 3.
6. Smith C.W., Constantinescu D.M. and Liu C.T., In Proceedings of the 2002
International Conference on Computational Engineering & Sciences, Chapter 7,
Technical Science Press, Irvine, Ca., 2002, Paper 5.
 25. Mixed-Mode Fracture 1165

AN ARBITRARILY ORIENTED CRACK NEAR A COATED FIBER

H. M. Shodja and F. Ojaghnezhad

Sharif University of Technology
Department of Civil Engineering, P.O. Box 11365-9313, Tehran, Iran
shodja@sharif.edu

The stress intensity factors (SIFs) for an arbitrarily oriented plane crack in the vicinity of a coated
circular fiber is being sought. The method of solution is based on Shodja and Sarvestani's [1]
equivalent inclusion method (EIM) for multi-inhomogeneity systems, which is an extension of
Eshelby's [2] theory for a single ellipsoidal inhomogeneity. The proposed approach is very robust
in the sense that it can effectively and systematically be applied to wide variety of fundamental
problems, which are essential for micromechanical studies of composite materials, for example,
see, Shodja et al. [3], Shodja and Roumi [4,5].
To the best of the authors' knowledge, the present problem has not been solved in the literature.
Atkinson [6] considers the symmetric problem of the interaction between a crack and a circular
inhomogeneity, in which the crack plane is normal to the uniform far-field applied stress and
coincides with the plane of symmetry of the circular inhomogeneity, Fig.1. Erdogan et al. [7]
consider the same problem, except that the crack plane is parallel to a plane of symmetry of the
inhomogeneity and is situated at a distance c away from it. In the current work, it is proposed that
the fiber be coated and the crack be arbitrarily oriented as shown in Fig. 2.

FIGURE 1. A crack near a circular fiber

For the sake of comparison and verification of the accuracy of the proposed approach, the
special case considered by [6] and [7] are readily re-examined by the present methodology. For
example, the SIFs for the problem considered by [6] are verified to be in good agreement with the
results obtained using the present theory, Table 1.

TABLE 1. Mode I SIFs corresponding to Fig.1

1166 H. M. Shodja and F. Ojaghnezhad

In Table 1, µ m and µ f are the shear modulus for the matrix and fiber, respectively. The
Poisson's ratio is assumed to be 1/4, a c 1/2, ı$ 1, and a f 1 . The values of the SIFs are
normalized in accordance to [6]. Note that the results of [6], which are presented in Table 1 are
read visually from their plots.

FIGURE 2. An arbitrarily oriented crack near a circular coated fiber

References
1. Shodja, H.M., Sarvestani A.S., Journal of Applied Mechanics, Vol. 68, 3-10, 2001
2. Eshelby, J. D., Progress in Solid Mechanics, Vol. 2, Edited by I. N. Sneddon and R. Hill,
North-Holland, Amesterdam, 1961, 89-140.
3. Shodja, H.M., Rad, I.Z., Soheilifard, R., J. Mech. Phys. Solids, vol. 51, 945-960, 2003
4. Shodja, H.M., Roumi, F., Mechanics of Materials, Vol. 37, 343-353, 2005.
5. Shodja, H.M., Roumi, F., In XXI Int. Congress of Theoretical and Applied Mechanics,
Warsaw, Poland, August 15-21, 2004.
6. Atkinson, C., Int. J. Engng. Sci., vol. 10, 127-136, 1972
7. Erdogan, F., Gupta, G.D., Ratwani, M., Journal of Applied Mechanics, vol. 41, 1007-1013,
1974
 25. Mixed-Mode Fracture 1167

SIMULATION OF THE MIXED MODE FRACTURE OF CONCRETE WITH

COHESIVE MODELS

J. C. Gálvez, D. A. Cendón, E. Reyes1, J. M. Sancho2 and J.Planas

E.T.S. Ingenieros Caminos, Universidad Politécnica Madrid
c/ Profesor Aranguren s/n, 28040 Madrid, Spain
1E.T.S. Ingenieros Caminos, Universidad Castilla La Mancha

Av. Camilo José Cela s/n, 13071 Ciudad Real, Spain

2E.T.S. Arquitectura, Universidad Politécnica Madrid

c/ Juan de Herrera 4, 28040 Madrid, Spain

jcgalvez@caminos.upm.es

Considerable effort has been devoted to developing numerical models to simulate the mixed mode
fracture of quasi-brittle materials. Traditionally, the numerical methods based on the Finite
Element Method were classified into two groups [1]: smeared crack approach and discrete crack
approach.
In the smeared crack approach the fracture is represented in a smeared manner: an infinite
number of parallel cracks of infinitely small opening are (theoretically) distributed (smeared) over
the finite element [2]. The cracks are usually modelled on a fixed finite element mesh. Their
propagation is simulated by the reduction of the stiffness and strength of the material. The
constitutive laws, defined by stress-strain relations, are non-linear and show a strain softening.
This approach was pioneered with fixed-crack orthotropic secant models and rotating crack
models.
However, strain softening introduces some difficulties in the analysis. The system of equations
may become ill-posed, localization instabilities and spurious mesh sensitivity of finite element
calculations may appear. These difficulties can be tackled by supplementing the material model
with some mathematical condition. Other strategies are the non-local continuum models, the
gradient models, and the micropolar continuum. These procedures are suited to specific problems,
but none gives a general solution of the problem.
The discrete approach is preferred when there is one crack, or a finite number of cracks, in the
structure. The cohesive crack model, developed by Hillerborg and co-workers [3] for mode I
fracture of concrete, was shown to be efficient to model the fracture process of quasi-brittle
materials. It has been extended to mixed mode fracture (modes I and II) and incorporated into
finite element codes [4-8] and into boundary element codes [9]. One of the difficulties associated
with these codes is that they require the remeshing and/or refinement of the finite element mesh
when the crack grows, and some of them also require an input of material properties that are
difficult to evaluate.
In recent years, a new methodology based on the so-called strong discontinuity approach
(SDA) has been proposed [10]. The SDA complements the classical approaches, the smeared crack
and the discrete crack, and has been successful in the analysis of the fracture of brittle materials. In
contrast to the smeared crack model, in the SDA the fracture zone is represented as a discontinuous
displacement surface. In contrast to the discrete crack approach, in the SDA the crack geometry is
not restricted to interelement lines, as the displacement jumps are embedded in the corresponding
finite element displacement field. A comparative review of the various approaches to the
embedded crack concept is presented by Jirárek [11].
1168 J. C. Galvez et al.

Based on the above exposed we may conclude that there are two main procedures to
incorporate the cohesive model in the discrete crack approach. One of them is based on the
previous knowledge of the crack path, which is incorporated on the finite element mesh, and the
cohesive model is implemented over the crack trajectory. The other one proposes a finite element
model with a cohesive crack embedded. The paper compares both approaches, establishes the
advantages and disadvantages of every one and compares the numerical results for different testing
geometries. The experimental results from several researchers are modelled, and the differences
between the numerical results are commented.
Especial attention is devoted to the experimental results supplied by the authors, based on the
TPB geometry, under proportional and non-proportional loading, obtained for three sizes of the
specimens. These tests include a complete series of results, like the stresses in the crack during the
test, very useful for the comparison of the numerical procedures.

References
1. Elices M., Planas J. in: Fracture Mechanics of Concrete Structures,. Chapman & Hall,
London, 16-66, 1989.
2. Bazant Z.P., Planas J. Fracture and Size Effect in Concrete and Other Quasibrittle Materials,
CRC Press, New York, 1998.
3. Hillerborg A., Modéer M., Petersson P., Cement and Concrete Research, vol. 6, 773-782,
1976.
4. Cervenka J. Discrete Crack Modelling in Concrete Structures, Ph.D. Thesis, University of
Colorado, 1994.
5. Xie M., Gerstle W., Journal of Engineering Mechanics, vol. 121, 1349-1358, 1995.
6. Valente S., in: Fracture of Brittle Disordered Materials, Concrete, Rock and Ceramics, E&FN
Spon, 66-80, 1995.
7. Cendón D.A., Gálvez J.C., Elices M., Planas J., International Journal of Fracture, vol. 103,
293-310, 2000.
8. Gálvez J.C., Cervenka J., Cendón D.A., Saouma V., Cement and Concrete Research, vol. 32,
1567-1585, 2002.
9. Saleh A., Aliabadi M., Engineering Fracture Mechanics, vol. 51, 533-545, 1995.
10. Simo J., Oliver J., Armero F., Computational Mechanics, vol. 12, 277-296, 1993.
11. Jirásek M., Computer Methods in Applied Mechanics and Engineering, vol. 188, 307–330,
2000.
12. Gálvez J.C., Elices M., Guinea G.V., Planas J., International Journal of Fracture, vol. 94, 267-
284, 1998.
13. Galvez, J.C. Cendón, D.A., Planas J , International Journal of Fracture, vol 118, 163-189,
2002.
 25. Mixed-Mode Fracture 1169

MICROMECHANICAL ANALYSIS OF RUPTURE MECHANISMS IN MIXED

MODE DUCTILE FRACTURE

I. Barsoum and J. Faleskog

Department of Solid Mechanics, Royal Institute of Technology
Osquars backe 1, 100 44 Stockholm, Sweden
imad@hallf.kth.se , jonasf@hallf.kth.se

The fracture toughness of ductile materials may differ considerably under mode I and mode II/III
loading. The main reason for this is that the fracture mechanisms differ in important aspects
between the opening loading mode (mode I) and the shear loading mode (mode II/III).
Experimental studies show that the mode I failure mechanism involves void nucleation, growth
and coalescence. In mode II/III however, intense plastic straining ahead of the crack tip promotes
nucleation of micro voids which typically experience limited growth before linking up under
intense shear deformation. Hence, under general mixed mode loading, there will be at least two
failure mechanisms that may co-operate or even compete. At lower stress triaxialities (mode II/III),
the shear dimple rupture mode will be favoured, whereas at higher stress trixailities (mode I) the
flat dimple rupture mode will be favoured. In the flat dimple rupture mode the final link up
between voids occurs through necking of void ligaments until impingement. By contrast, in the
shear dimple rupture mode, final link up between voids occurs through shear localization of plastic
strain in the ligaments between voids. In the latter case, voids rarely grow until impingement. In
this study we are focusing on the conditions that govern the transition between the two
mechanisms. This is carried out by micromechanical modelling using finite element analysis.
The localized type of deformation characterizing both failure modes fits well into the general
theoretical framework of plastic localization introduced by Rice (1976). In that study the
conditions for having a localized band in which the deformation varies with the position across the
band were investigated. Figure 1 illustrates the band. Here we make use of the kinematical
conditions for the deformation across the band to analyse the transition between the two failure
mechanisms for ductile mixed mode fracture. The deformation gradient is homogeneous out side
the band in the Rice model, whereas it varies in a continuous manner with position across the band.
This facilitates localization into a symmetric mode, a shear mode or a combination of both modes.
The micromechanical model employed consists of an array of equally sized cells located within
the planar band shown in Figure 1. Each cell has the same thickness as the band and contains one
spherical void located at its centre. Thus, nucleation of the void is not considered here. The
periodic arrangement of the cells allows the study of a single cell as the representative volume
element (RVE). To simplify matters within the framework discussed above, a state of generalized
tension is used to represent pure mode I type of loading (symmetric) and a state of generalized
shear is used to represent mode II/III type of loading (shear). Mixed mode loading can then be
accomplished by a combination of the two generalized cases where the macro stresses (average

stresses) acting on a cell are: 61 60  V , V t0, 62 63 60 , 612 W t 0 and

613 6 23 0 , see Figure 1. Generalized tension corresponds to W 0 and generalized shear
corresponds to V 0 . The mean stress and the effective stress can be expressed as

Vm 60  V 3 and Ve V 2  3W 2 , respectively, with stress triaxiality defined as

2
T V m V e . The mode mixture is here quantified by the parameter K (2 S ) arctan(3W 2 V ) .
1170 I. Barsoum and J. Faleskog

The RVE is loaded in such a way that the macro stresses is kept constant, i.e. proportional stressing
prevails throughout the loading. Geometrically, the RVE (see Figure 1) is characterized by the

height aspect ratio defined as O S x / S y , and the void spacing aspect ratio defined as F R / Sy .

The width and the depth of the RVE are set to be equal, S z Sy .

With the model described above the transition between the flat dimple rupture mode and the
the shear dimple rupture mode will depend on stress triaxiality, T, mode mixture, K , and
geometry, O and F . Results will be given for an elastic ideal plastic material and an elastic
moderately hardening plastic material.

FIGURE 1. The representative volume element containing a spherical void, loaded in generalized
tension and/or simple shear with a superposed hydrostatic stress.

References
1. Rice, J.R., In Theoretical and Applied Mechanics 14th IUTAM Congress, edited by W.T.
Koiter, North-Holland, Amsterdam, 1976, pp. 207-220.
 25. Mixed-Mode Fracture 1171

MODE I PRELOADING-MODE II FRACTURE IN WARM PRE-STRESSING

M. R. Ayatollahi and M. Mostafavi

Fatigue and Fracture Laboratory, Mechanical Engineering Department,
Iran University of Science and Technology
IUST, Narmak, Tehran, 16844, IRAN
m.ayat@iust.ac.ir

The loading history of cracked specimens plays an important role in their in-use behavior. Warm
Pre-stressing (WPS) can be mentioned as a pre-loading procedure that significantly changes the
apparent fracture toughness in some steels. In this procedure the specimen is subjected to a preload
in high temperature and is used in a lower service temperature. When the cracked specimen is
preloaded and fractured both in mode I, it is well known that WPS improves the load bearing
capacity of specimens [1]. Meanwhile, very little work has been conducted for cases where
preloading and fracture take place in different modes of loading [2].
In this paper the finite element method is used to study the effects of mode I preloading on
mode I and mode II fracture. The finite element analysis is conducted on a biaxially loaded center
crack plate in which the crack makes an angle 45o relative to the loading direction (see Fig. 1). For
O=1, the specimen is under pure mode I loading and for O= -1 it is loaded in pure mode II. The
finite element code ABAQUS is employed for simulation of warm pre-stressing and fracture. It is
assumed that the final fracture in reloading (whether in mode I or mode II) is always due to
cleavage fracture. The RKR model [3] is used as fracture criterion. According to this criterion
cleavage fracture initiates whenever the hoop stress reaches a critical value in a critical distance
from the crack tip. Cleavage fracture is assumed to be initiated along the crack line in mode I
fracture and along the direction of maximum hoop stress in mode II fracture.

FIGURE 1. Cracked specimen used for WPS simulation

Four different crack lengths (a/W= 0.1, 0.2, 0.3 and 0.4) are considered in the specimen. For
each crack length, the specimen is preloaded in pure mode I (i.e. O=1) with five different
preloading levels corresponding to 20%, 40%, 60%, 80% and 95% of fracture J. After the
unloading and cooling stages, the specimen is reloaded either in mode I (O=1) or in mode II (O= -1)
till the hoop stress attains its critical value. The fracture loads with and without warm prestressing
are then compared. Figs 2 and 3 show the improvement in fracture load when the specimen is
reloaded in pure mode I and pure mode II, respectively. While a significant improvement in
fracture load is seen for mode I preloading-mode I fracture, there is almost no improvement for
mode I preloading-mode II fracture. The negligible effect of mode I preloading on mode II fracture
can be attributed to the distribution of residual stresses around the crack tip after unloading.
1172 M. R. Ayatollahi and M. Mostafavi

FIGURE 2.- Improvement in Mode I fracture load due to Mode I preloading.

FIGURE 3. Improvement in Mode II fracture load due to Mode I preloading.

References
1. Smith, D.J., Hadidimoud, S., and Fowler, H., Engineering Fracture Mechanics, 2004.
71:2015-2032.
2. Swankie, T.D. and Smith, D. J., Engineering Fracture Mechanics, 1998. 61: 387-405.
3. Ritchie, R.O., Knott, J.F., and Rice, J.R., Journal of the Mechanics and Physics of Solids,
1973. 21: 395-410.
 25. Mixed-Mode Fracture 1173

PREDICTIONS OF MIXED MODE I/II FRACTURE TOUGHNESS FOR SOFT

ROCKS

M. R. Ayatollahi and M. R. M. Aliha

Fatigue and Fracture Lab., Department of Mechanical Engineering
Iran University of Science and Technology
Narmak, Tehran, 16844, Iran
m.ayat@iust.ac.ir
m_aliha@yahoo.com

Cracked rock masses are usually subjected to complex loading conditions. Therefore, it is among
favorite concepts for rock fracture researchers to investigate the fracture of rocks under mixed
mode I/II loading. There are numerous experimental studies on mixed mode I/II fracture of rocks
using different test specimens. Semi-circular bend (SCB) specimen is one of the favorite test
configurations for conducting mixed mode fracture toughness experiments in rock materials. This
specimen is simple and its manufacturing is inexpensive. By changing the parameters a/R, S/R and
E (see Fig. 1), the SCB specimen can provide pure mode I, pure mode II and any combination of
modes I and II.
In addition to the experimental studies, there are several fracture criteria for mixed mode
fracture behavior of brittle materials. However, the results obtained from fracture tests on a
synthetic soft rock called Johnstone [1] by using the SCB specimens can not be appropriately
interpreted when the available fracture criteria are used. For example, the maximum tangential
stress (MTS) criterion [2] fails to predict the test results; especially for mode II dominated loading
conditions. As shown in Fig. 2, a great discrepancy exists between the experimental results [1] and
predictions by the MTS criterion for the SCB specimen.

FIGURE 1. Cracked semi-circular bend (SCB) specimen.

In this paper, improved predictions are achieved for fracture toughness of Johnstone by using a
more accurate description of tangential stress in the MTS criterion. The tangential stress around the
crack tip is outlined by the Williams [3] series expansion as:

V TT
1
2Sr
cos
Tª
«
2¬
T 3
2 2
º
¼
 
K , cos 2  K ,, sin T »  T sin 2 T  O r1 2
(1)
1174 M. R. Ayatollahi and M. R. M. Aliha

where r and T are the conventional crack tip co-ordinates. KI and KII are the mode I and mode II
stress intensity factors, respectively and T is the non-singular stress term. While the stress intensity
factors (KI, KII) describe the severity of stress singularity around the crack tip the magnitude and
the sign of T-stress can also influence significantly the onset of fracture in cracked specimens
subjected to mixed mode I/II loading [4].
It is shown in this paper that a modified MTS criterion which takes into account the effects of
both singular term and the T-term of stress around the crack tip can predict very well the results
obtained from mixed mode I/II fracture tests on Johnstone [1]. In the modified MTS criterion crack
tip parameters (KI, KII, T ) are used for evaluating the fracture load of cracked specimens.

FIGURE 2. Mixed mode I/II fracture toughness envelop for SCB specimens with (a/R= 0.35 and S/
R = 0.5) made of a soft rock.

References
1. Lim, I.L., Johnston, I.W.,. Choi, S.K, Boland, J.N., Int. J. Rock Mech. Min. Sci. Geomech.
Abstr., vol 31(3), 199-212, 1994
2. Erdogan, F., Sih, G.C., J. Basic Eng., Trans. ASME 85, 519-525, 1963
3. Williams, M.L., J. Appl. Mech., vol, 24,109-114, 1957
4. Smith, D.J., Ayatollahi, M.R., Pavier, M.J., Fatig. Fract. Eng. Mater. Struct., vol. 24, 137-150,
2001
 25. Mixed-Mode Fracture 1175

AN INTERFACE MODEL FOR MIXED-MODE, BUCKLING-DRIVEN

DECOHESION OF SUPERFICIAL LAYERS

S. Bennati and P. S. Valvo

Department of Structural Engineering, University of Pisa
Via Diotisalvi 2, 56126 Pisa, Italy
s.bennati@ing.unipi.it, p.valvo@ing.unipi.it

There are several situations in which the local phenomena acting in structural elements can lead to
the partial detachment of a superficial layer from its underlying substrate. In the case of fibre-
reinforced composite laminates, for example, delamination is one of the most insidious events that
can compromise their mechanical performance. Analogous decohesion phenomena can be
observed, for instance, in film coatings, sandwich plates, etc. (Hutchinson and Suo [1]).
When a structural element containing a partially detached layer is subjected to compression,
local instability phenomena can easily arise (Kachanov [2]). These may in turn provoke the growth
of the decohered area itself, as they result in high values of the interfacial stresses. In the
framework of Fracture Mechanics, an adequate description of the process by which the detached
region extends requires accounting for the so-called “mode mixity”, i.e., the ratio between the
contributions (generally varying during the process) of the conventional fracture modes I, II and III
(respectively, opening, sliding and tearing modes). For plane problems, where only modes I and II
are present, according to Hutchinson [3], the mode mixity can be measured by the angle

kII
\ tan 1
kI (1)

where kI and kII are the stress intensity factors at the delamination front, corresponding to modes I
and II, respectively.

FIGURE 1. Interlaminar stresses at the interface.

Despite the complexity of the problem, a simple mechanical model, where separation is
assumed to occur along an elastic-brittle interface joining the decohered layer and the substrate,
has proven to be effective in describing the decohesion phenomenon (Bennati and Valvo [4]).
Actually, an explicit solution to the problem has been deduced, allowing us to evaluate the
evolution of the peak stresses at the delamination front (Fig. 1). Since these stresses attain finite
values, the expression of the mode mixity angle (1) is conveniently modified as follows,
1176 S. Bennati and P. S. Valvo

W XZ
\ tan 1
VZ (2)

where WXZ and VZ are the interfacial stresses evaluated at the delamination front. Thus, the energy
variations involved in its further crack extension can be deduced through direct calculations. In
particular, both the mode mixity angle and the critical strain energy release rate can be obtained
(Fig. 2). Finally, through the choice of appropriate growth criteria, it is possible to achieve an
analytic description of the entire process of buckling-driven decohesion growth under both
monotonic and cyclic loading conditions. In the authors’ opinion, the proposed model can help
explaining many apparently strange and unsolved experimental results reported in the literature.

FIGURE 2. Mode mixity angle and critical strain energy release rate.

References
1. Hutchinson, J.W., Suo, Z., Adv. Appl. Mech., vol. 29, 63-191, 1992.
2. Kachanov, L.M., Delamination buckling of composite materials, Kluwer Academic
Publishers, Dordrecht-Boston-London, 1988.
3. Hutchinson, J.W., In Metal-ceramic interfaces, edited by M. Rühle, A.G. Evans, M.F. Ashby
and J.P. Hirth, Pergamon Press, New York, 1990, 295-306.
4. Bennati, S., Valvo, P.S., Key Eng. Mater., vol. 221-222, 293-306, 2002.
 25. Mixed-Mode Fracture 1177

MXED-MODE FRACTURE ANALYSS OF ORTHOTROPC FUNCTONALLY

GRADED MATERALS

Serkan Dag, Bora Yildirim1, Duygu Sarikaya

Department of Mechanical Engineering, Middle East Technical University, Ankara, Turkey
1Department of Mechanical Engineering, Hacettepe University, Ankara, Turkey

sdag@metu.edu.tr, boray@hacettepe.edu.tr, duygusarikaya@gmail.com

Functionally graded materials (FGMs) are multiphase composites that have spatial variations in the
composition and microstructure. These variations are intentionally introduced in order to take
advantage of different thermomechanical properties of the constituent materials. The fracture
mechanics theory and analysis of functionally graded materials is based on the assumption of
continuous variations in the related thermomechanical properties (see for example Yildirim et al.
[1] and Kim and Paulino [2]). Fracture mechanics problems that occur in FGMs are generally
observed to be either due to mode I edge cracks that are perpendicular to the free boundaries or due
to the embedded cracks that are parallel to the material surface. These observations could be
attributed to the fact that some of the processing methods used to create graded layers induce an
oriented microstructure. For example, as shown by Sampath et al. [3], FGMs that are processed by
the plasma spray technique have a lamellar structure which has weak fracture planes parallel to the
boundary. Embedded cracks that can initiate at the weak cleavage planes are inherently under
mixed-mode mechanical or thermal loading. One of the approaches to examine fracture mechanics
problems in this type of structures is to model the functionally graded medium as orthotropic with
principal directions of orthotropy parallel and perpendicular to the free surface (Ozturk and
Erdogan [4]).
The objective in the present study is to develop analytical and computational methods to examine
embedded crack problems in orthotropic FGMs under mixed-mode loading. The geometry of the
crack problem considered is depicted in Fig. 1. x1 and x2 in this figure are the principal directions
of orthotropy. The layer is graded in x2-direction and contains an embedded crack of length 2a at

x2 0 . The crack is loaded by normal and/or shear tractions which are applied at x2 0 and
x2 0 for x1  a . The problem is formulated using the averaged constants of plane
orthotropic elasticity which are first introduced by Krenk [5]. The governing partial differential
equations are obtained in terms of the displacement components and then they are reduced to a pair
of singular integral equations using Fourier transforms. The integral equations are solved
numerically using an expansion-collocation technique to compute the modes I and II stress
intensity factors and the energy release rate at the crack tips.
Enriched finite element formulation for mixed-mode fracture analysis of orthotropic FGMs is
developed by Dag et al. [6]. This method is used in the second part of the present study to solve the
problem shown in Fig. 1. Enriched and graded finite elements are used to model the orthotropic
functionally graded medium. Fig. 2 shows the deformed shape of the finite element model of an
embedded crack subjected to uniform normal loading. Enriched crack tip elements contain the
theoretical asymptotic displacement field and it is possible to compute the mixed-mode stress
intensity factors directly from the solution of a linear equation system. The numerical results
calculated using the analytical formulation and enriched finite elements technique are shown to be
in good agreement. In the final part of the study, the effects of boundary conditions, material
nonhomogeneity, orthotropy parameters and loading conditions on the mixed-mode stress intensity
1178 S. Dag et al.

factors and the energy release are explored by carrying out parametric analyses and the results are
briefly discussed.

FIGURE 1. An embedded crack in an orthotropic functionally graded layer.

FIGURE 2. Deformed shape of the finite element model of the crack shown in Fig. 1.

References
1. Yildirim, B., Dag, S. and Erdogan, F., Int. J. Fracture, vol. 132, 369-395, 2005.
2. Kim, J.-H. and Paulino, G.H., Int. J. Numer. Meth. Eng., vol. 53, 1903-1935, 2002.
3. Sampath, S., Herman, H., Shimoda, N. and Saito, T., MRS Bulletin, vol. 20, 27-31, 1995.
4. Ozturk, M. and Erdogan, F., Int. J. Fracture, vol. 98, 243-261, 1999.
5. Krenk, S., J. Compos. Mater., vol. 13, 108-116, 1979.
6. Dag, S., Yildirim, B. and Erdogan, F., Int. J. Fracture, vol. 130, 471-496, 2004.
 25. Mixed-Mode Fracture 1179

NEW SCHEME FOR FEA OF MIXED MODE STABLE CRACK GROWTH

S. K. Maiti, S. Namdeo and A. H. I. Mourad1

Mechanical Engineering Department, Indian Institute of Technology Bombay, Powai, Mumbai
400076, India
1Mechanical Engineering Department, Faculty of Engineering, United Arab Emirates University,

Al-Ain, POB 17555, UAE

skmaiti@me.iitb.ac.in, s_namdeo@iitb.ac.in, ahmourad@uaeu.ac.ae

A scheme for elastic-plastic finite element analysis (FEA) has been proposed for the study
of stable crack growth (SCG) from initiation to instability in both mode I and mixed mode (I and
II). In the analysis the condition for crack extension at every stage of the SCG is considered to be
CTOA/CTOD reaching a critical value (Newman [1], Maiti and Mahanty [2], Maiti and Mourad
[3]). The scheme permits predictions of load variation with load line displacement (LLD), crack tip
current plastic zone and crack edge profile.
For the study of the whole span of SCG in mode I a single discretization is employed.
For a similar study of a mixed mode problem a single discretization is carefully made basing on the
assumption that the whole span of the SCG takes place along a straight line inclined with the main
crack at an angle corresponding to the direction of initial crack extension (Fig.1). The direction of
initial crack growth is predicted using a full field elastic analysis and the maximum tangential
principal stress (MTPS) criterion. Most of the elements in a discretization are 8-noded
quadrilaterals. Ahead of the crack tip, elements are squares of size 0.4 mm. External loads applied
at node H and two nodes on its either side; simultaneously node G and two nodes on its either side
are constrained (Fig.1). The predictions for the initial direction of crack growth for a few cases are
compared with the experimental data reported by Mourad et al. [4] on AISI 4340 steel.

0
FIGURE 1. Typical discretization for study of crack growth for 75 loading angle.

The proposed scheme of analysis of the whole SCG is done in a few stages using the ANSYS
(version 8.0) software. The scheme is different from the schemes proposed by Newman [1], Maiti
and Mahanty [2], and Miller and Kfouri [5]. Each stage is analyzed separately using a
discretization, which is related very closely to that of the previous stage, and considering the
domain to be free of any residual stresses and strains. In each stage crack is allowed to extend by
0.4 mm. The detail of the scheme is presented in the paper. Both mode I (a0/w = 0.42 and 0.43) and
mixed mode problems (loading angle =75o, 65o and 60o) have been studied. This new scheme
helps to determine the variation of fracture load with crack extension without requiring much
computer storage and time. A typical variation of load with LLD is shown in Fig.2. Predictions for
1180 S. K. Maiti et al.

the direction of initial crack extension in a mixed mode, initiation and maximum loads in both
mode I and mixed mode compare very closely with the experimental results [4] in the case
AISI 4340 steel.

FIGURE 2. Comparison of theoretical and experimental variation of load with load line
displacement for mixed mode for 75o loading angle.

References
1. Newman, J. C., Jr., Fracture Mechanics: Fifteenth Symposium, ASTM STP 833,
edited by R. J. Sanford, American Society for Testing and Materials, 1984, 93-117.
2. Maiti, S. K., and Mahanty, D. K., Engineering Fracture Mechanics, vol. 37, 1251-
1275, 1990.
3. Maiti, S. K., and Mourad, A.H.-I., Engineering Fracture Mechanics, vol. 52, 349-
378, 1995.
4. Mourad, A.H.-I., Alghafri, M.J., Abu Zeid, O.A., and Maiti, S.K., Nuclear Engineering and
Design, vol. 235, 637-647, 2005.
5. Miller, K.J., and Kfouri, A.P., Elastic-Plastic Fracture, ASTM STP 668, edited by
J. D. Landes, J. A. Begley, and G. A. Clarke, American Society for Testing and Materials,
1979, 214-228.
 25. Mixed-Mode Fracture 1181

NUMERICAL SIMULATION OF NONLINEAR CRACK PROPAGATION

UNDER MIXED-MODE IMPACT LOADING

T. Fujimoto and T. Nishioka

Kobe University
5-1-1 Fukaeminamimachi, Higashinada-ku, Kobe 658-0022 Japan
takehiro@maritime.kobe-u.ac.jp, nishioka@maritime.kobe-u.ac.jp

In some of the accidents of industrial structure, which have been happened by collision, earthquake
and etc., fast crack propagations caused in the parts of the structure. In some cases of these
fractures, large deformation with plastic strain accompany the crack propagation. Prediction and
control of crack propagation direction have been required to prevent critical destruction of the
structure.
In order to simulate and measure the crack propagation phenomena, some numerical methods
have been proposed. The nodal release method [1] in FEM is one of very popular technique for
non-linear crack propagation. However, because the crack propagation path has to be expressed by
the boundaries of finite elements in the nodal release method, the nodal release method is not suit
to predict non-straight crack propagation path. The element release method is adopted in some
commercial code of FEM. In this method, an element in singular strain area near the propagating
crack tip is removed and energy balance is not satisfied during crack propagation.
In this study, accurate measurement of nonlinear propagating crack tip condition is considered
by using the moving finite element method, which has been proposed by Nishioka and co-workers
[2]. In the cases of some nonlinear fractures, distinct wake zone is observed near the crack
propagation path. These plastic strain singularities are estimated by the moving finite element
technique as shown in Fig. 1. Non-straight crack propagation path and/or crack bifurcation paths
can be reproduced by the moving element technique [3].
In the moving FEM, the mappings of deformation parameters have to be operated at each time step.
In order to estimate the solution at next time step t+'t, Nishioka [4] derived the variational
principle to satisfy the equilibrium of total force based on the mapping parameters ()(t0).
t 
³V S ij ' t G u i , j dV  ³V U ui 0 ' t G u i dV  t    t  t 
³ S t t i 0  t i ' t G u i dS  ³V V ij 0 G u i , j dV  ³V U ui 0 G u i dV (1)

FIGURE 1. Fine mesh subdivision near propagating path

where () means the time derivative of ( ) . S ij and V ij mean the Lagrange stress and the

Cauchy stress, respectively. ui and ti are displacement and traction force, respectively. U
denotes the mass density.
In this study, the crack propagation path is predicted numerically. Some fracture path
prediction theories have been proposed, and popular theories are summarized in Ref. [5]. Some of
1182 T. Fujimoto and T. Nishioka

the prediction theories are introduced into the moving finite element method. In order to discuss
the validity of the numerical simulation and the numerical fracture path prediction algorism, the
numerical results are compared with the experimental result for the fracture of pure aluminium
(A1050) under mix mode load.

FIGURE 2. Pure aluminium specimen under mix-mode load

FIGURE 3. Progress of nonlinear crack propagation in pure aluminium

The one of experimental set-up is shown in Fig. 2. The ultra-high speed camera records the
history of deformation and crack propagation. Figure 3 shows the progress of crack propagation.
Framing rate is 8000 frame/sec. Average of crack kinking angle is about 6 degree, and crack
propagation velocity is 4.44m/s.

References
1. Kobayashi A.S., Mall S., Urabe Y. and Emery A.F., Numerical Methods in Fracture
Mechanics, Univ. College, Swansea, 709-720, 1978.
2. Nishioka, T., JSME International Journal, Series A, Vol.37, No.4, 313-333, 1994.
3. Tchouikov, S., Nishioka, T. and Fujimoto, T., Computers, Materials and Continua, Tech
Science Press, Vol.1, No.2, 191-204, 2004.
4. Nishioka, T. and Wang, Z.M., Constitutive and Damage Modeling of Inelastic Deformation
and Phase Transformation, Neat Press, Fulton, Maryland, 741-744, 1999.
5. Nishioka, T., "Computational Aspects on Dynamic Fracture", Comprehensive Structural
Integrity, Volume 3: Computational Methods (R, de Borst and H.A. Mang), Chapter 4,
Elsevier Science.
 25. Mixed-Mode Fracture 1183

ELASTIC-PLASTIC BEHAVIOUR OF CRACK PROPAGATION UNDER

BIAXIAL CYCLIC LOADING

T. Hoshide
Department of Energy Conversion Science, Kyoto University
Yoshida-Honmachi, Sakyo-ku, Kyoto 606-8501, Japan
hoshide@energy.kyoto-u.ac.jp

The fatigue crack growth behaviour under biaxial stress state is affected by non-singular stress
under elastic-plastic deformation. The effects of non-singular stress with plasticity on biaxial
crack growth should be clarified for the integrity design of machineries under practical service
loading. In this work, the aforementioned subjects were discussed by reconsidering previous
works by Hoshide et al. [1, 2]. In previous studies, fatigue crack propagation tests were conducted
by using cruciform specimens of low carbon steel under uniaxial, equi-biaxial and shear loading
modes, and also by using thin-walled tubular specimens of pure copper under combined in-phase
axial-torsional loading modes. The yield strength of the steel and the copper is 228MPa and
24.8MPa respectively.
In cruciform specimens of steel, it was confirmed that the crack propagation rate correlated
with the stress intensity range 'K was significantly affected by the non-singular stress in the
direction parallel to the crack. The opening displacement at the centre of crack was monitored, and
it was revealed that the above result was mainly caused by the crack closure behaviour influenced
by the non-singular stress. The crack propagation rate was almost uniquely correlated with the
effective stress intensity factor 'Keff. However, in the case of shear loading, significant plasticity
was detected and the crack growth rate for the same 'Keff -value shifted toward higher rate region
compared with that in other loading cases.
In tubular specimens of copper, it was clarified that the cracking morphology was a bent-type
in the combined mode and a branch-type in the torsional case. Cracks were found to propagate in
such direction that the Mode II component of the stress intensity factor was minimized. By using a
devise to detect the opening and sliding displacements at the centre of crack, the deformation
behaviour of bent- or branch-cracked specimens was also examined to evaluate the crack opening
point. It was remarked that tails appeared near both tips of the observed hysteresis loop in the case
of reversed torsion. In this situation, any good correlation was not seen even if the effective stress
intensity factor 'KIeff of Mode I was employed to correlate the growth rate. It was found that this
behaviour was mainly associated with gross plasticity.
A simple method to estimate the J integral for a centre-cracked plate under biaxial stresses was
proposed based on the finite element analysis, and a tentative procedure to estimate the J integral
was also developed for the bent crack. In both situations, it was cleared that the centre opening
displacement should be used in evaluating the J integral range 'J. The growth rate of crack in
cruciform specimen and bent crack in tubular specimen were correlated with 'J evaluated by using
the proposed methods. Independently of biaxiality and plasticity, the crack propagation rate da/dN
was found to be correlated with 'J by a unique power function as

da
C ('J ) m (1)
dN
1184 T. Hoshide

Consequently, it is suggested that the parameter governing the crack propagation under biaxial
stress state is the range of J integral which is evaluated by using the hysteresis loop of applied
force versus opening displacement at the centre of crack.

References
1. Hoshide, T., Yamada A. and Tanaka, K., J. Soc. Mater. Sci., Japan, vol. 32, 528-534, 1983
2. Hoshide, T., Kawabata K., Yamakawa, N. and Inoue, T., J. Soc. Mater. Sci., Japan, vol. 38,
280-286, 1989
 25. Mixed-Mode Fracture 1185

NUMERICAL ANALYSIS OF MIXED-MODE CRACKING IN CONCRETE

DAMS

Z. Shi
Research and Development Center, Nippon Koei Co., Ltd.
2304 Inarihara, Tsukuba, Ibaraki, 300-1259, Japan
a4739@n-koei.co.jp

Since the establishment of the fictitious crack model (FCM) for crack analysis of the mode-I type
in concrete by Hillerborg et al. [1], continuous research effort has been made to apply the concept
to model mixed-mode fracture because most of the practical fracture problems in concrete are of
the mixed-mode nature, involving mode-I and mode-II. This paper presents a numerical study of
the mixed-mode fracture in concrete dams, using the extended FCM by Shi et al. [2-4]. For crack
propagation, the frequently adopted maximum principal stress criterion is used, which is based on
the notion that the mode-I condition is dominant at the tip of a mixed-mode crack. As the tip stress
reaches the tensile strength of concrete a mixed-mode crack propagates. A main feature of this
study is that normal and tangential tractions are applied directly to the crack surface, following
specific tension-softening and shear-transfer laws.
The scale-model test of a concrete gravity dam subjected to equivalent hydraulic loads by
Carpinteri et al. [5] is selected as a mixed-mode fracture problem to be solved. Two 1:40 scale
models of a gravity dam were prepared, and each contained a horizontal notch on the upstream
face located at a quarter of the dam height. Let W represent the dam thickness at the height of the
notch. For brevity of presentation, only one of the three tests is studied in the following under the
mixed-mode fracture condition, i.e., the second specimen with the notch size of 0.2W (Test 3). In
numerical studies, the bi-linear tension-softening relation of Fig. 1a is applied to the normal
traction and the crack opening displacement (COD). As for the shear-transfer law two types of bi-
linear shear-COD relation are assumed to exist between the tangential traction and the COD; see
Fig. 1b and Fig. 1c.

FIGURE 1. Assumed tension-softening and shear-transfer relations of concrete

As the influence of the shear force on the maximum load is found to be small and negligible,
the following discussion is focused on the crack path, which may be the foremost concern for the
crack analysis of concrete dams. Fig. 2 shows the obtained crack paths under the mode-I and the
mixed-mode conditions, as well as the experimental observations. As seen, during the steady crack
growth the mode-I path follows a trajectory with a larger downturn angle (measured from the
horizontal plane of the notch) than that of the observed crack path. Under the mixed-mode
condition of case 1 in which only a small shear strength fs = 0.1ft is assumed, the crack propagates
1186 Z. Shi

along the mode-I path initially; as the shear transfer takes place after the COD reaches Ws1, it
diverges to a new path that lies in between the mode-I and the actual crack trajectories with a much
smaller downturn angle. Increasing the shear strength in case 2 to fs = 0.2ft, the crack propagates
briefly along the mode-I path and then diverges to another path that merges with the observed
crack trajectories. These cracking behaviors obtained are easy to comprehend. Under the mode-I
condition, the discrepancy is caused by ignoring the frictional force or shear from the sliding
surfaces of the crack, which is tantamount to omitting a local moment that acts against the surface
deformation allowed under the mode-I condition. Here, the local moment is presumably formed by
a pair of opposite shear forces acting on the two crack surfaces at the distance of a COD. Under the
actions of these local moment forces in cases 1 and 2, the crack paths eventually diverge from the
mode-I path and move upwards. As shown in Fig. 2, in the later stage of the unsteady crack
propagation totally different fracturing behaviors are predicted by the numerical analyses that do
not agree with the experimental observations. This is believed to be caused by a transition of the
loading conditions that may have occurred during the experiments and is not reflected in the
numerical studies [3].

FIGURE 2. Experimental results and numerical predictions of crack trajectories under mode-I and
mixed-mode conditions with two types of shear-COD relations

References
1. Hillerborg, A., Modeer, M. and Peterson, P. E., Cement and concrete research, vol. 6, 773-
782, 1976.
2. Shi, Z., Ohtsu, M., Suzuki, M. and Hibino, Y., J. Struct. Engrg., vol. 127, 1085-1091, 2001.
3. Shi, Z., Suzuki, M. and Nakano, M., J. Struct. Engrg., vol. 129, 324-336, 2003.
4. Shi, Z., J. Struct. Engrg., vol. 130, 1738-1747, 2004.
5. Carpinteri, A., Valente, S., Ferrara, G. and Imperato, L., In Fracture mechanics of concrete
structures, edited by Z. P. Bazant, Elsevier Applied Science, New York, 1992, 351-360.
 26. Fracture Mechanics Characterization of Wood 1187

SPECIES AND OTHER PHYSICAL EFFECTS ON PARAMETERS

DESCRIBING A WOOD TOUGHNESS TEST.

B. Thibaut and J. Beauchene

CNRS French Guiana, Cirad French Guiana
L3MA, IESG campus Saint Denis, 97300 Cayenne, France
Cirad, ZI Pariacabo 973 Kourou, France
bernard.thibaut@guyane.cnrs.fr, jacques.beauchene@cirad.fr

Fracture mechanics is strongly dependant on material microstructure and defect occurrence in the
specimen. Wood microstructure is rather complex at different levels. In order to investigate the
influence of wood structural features, the Nordtest [1] specimen with 40mm height and 20mm
thickness (Schatz [2]) was performed on ten homogeneous tropical species, at green state, in order
to avoid drying checks, at room temperature and after 1 hour boiling in hot water at different
temperatures up to 80°C (Bardet [3]). The test was performed with the LT plane as rupture plane
and the L direction as rupture propagation direction.

Table1 Values of the structural parameters for the ten species

Species were selected in order to have a rather wide range of structural parameters: vessel
diameter: VD, number of vessels per mm2: VN, number of rays per mm: RN, fibre length: FL,
fibre width: FW, double cell wall thickness: WT, WT/FW ratio, parenchyma abundance: PA,
together with a wide range of basic density: BD (Table 1).

Table 2 Values of the mechanical parameters for the ten species

Indicators of fracture toughness test were: equivalent rupture stress Sr and its value Sr/D after
dividing by BD raised to the best statistical power at each temperature, fracture energy Gf and span
at rupture Fr, the ratio between the values at 25°C and 80°C, Gf 25/80, Fr 80/25 and Sr 25/80 are
indexes of temperature sensitivity of these parameters (Table 2).
1188 B. Thibaut and J. Beauchene

Table 3 Correlation coefficients between structural and mechanical parameters

Not surprisingly Sr has significant correlations, positive with WT/FW or density and negative
with FW (table 3), these parameters being highly inter correlated. Sr/D is positively influenced by
vessel diameter and parenchyma abundance. The same tendency was observed for Gf but with
lower significant values. PA has a positive significant correlation with Gf at high temperature
when density influence is low. Span at rupture is positively influenced by fibre length except for
high temperature where density becomes influent.
Temperature sensitivity is influenced for Gf by vessel density (+) and by parenchyma
abundance (-), for Sr by vessel density (+), basic density and WT/FW (+), for Fr by density (+)
and fibre width (-).
Other parameters as chemical composition should be looked at because there are also strong
variations between these species for that. Moreover, much more species should be investigated in
order to examine how vessel porosity, parenchyma occurrence, fibre length and fibre geometry
have a significant influence in conjunction with or explaining density role.

References
1. Nord Test (1993), NT Build, 422. Wood: fracture energy perpendicular to the grain Nordtest,
Espoo, Finland, From Gustafsson P.J., (1988): a study of strength of noched beams. CIB
W18A paper 21-10-1 meeting Vancouver
2. Schatz, T. (1995). "Determination de l'energie de rupture dans le bois." Holz als Roh- und
Werkstoff: 171-176.
3. Bardet S., Beauchene J., Thibaut B. (2003) Influence of basic density and temperature on
mechanical properties perpendicular to grain of ten wood tropical species, Annals of Forest
Science, vol. 60, 49-59.
 26. Fracture Mechanics Characterization of Wood 1189

YEW AND SPRUCE WOOD: MECHANICAL PROPERTIES AND FRACTURE

SURFACE STUDIES

Daniel Keunecke, Christoph Marki and Peter Niemz

Institute for Building Materials (Wood Physics Group), Swiss Federal Institute of Technology
Schafmattstrasse 6, CH-8093 Zurich, Switzerland
keunecke@ifb.baug.ethz.ch

Generally, there is a close connection between the anatomy of wood on the microscopic/
macroscopic level and its mechanical characteristics. Yew (Taxus baccata L.) and spruce (Picea
abies [L.] Karst.) are gymnosperms with wide differences in their microscopic structure (fig. 1)
and therefore in their fracture behavior and mechanical properties. Having been object of countless
studies, the individual mechanical properties of spruce wood are best-known. In contrast to spruce
wood, for the high dense yew wood only few reference values are available, even though it is
known for its extraordinary toughness and its strength.

FIGURE 1. SEM-images of yew (left) and spruce (right) wood.

The aim of this work was to compare selected mechanical properties of yew and spruce wood.
The determined properties are
• static and dynamic modulus of elasticity (MOE),
• fracture toughness KIC (ASTM E 399-90),
• work to ultimate load (as defined by Bodig and Jayne [1]) and
• impact bending strength.
Standard static test methods like the fracture toughness test (fig. 2) as well as micro tensile
tests and nondestructive test methods like sound velocity measurement and resonance frequency
measurement were applied. The portion of the elastic and plastic part of failure was examined,
wherever applicable. Furthermore the fracture surfaces have been analyzed and compared in order
to link the wood structure to the failure process.

FIGURE 2. Compact tension specimen: Dimensions [mm].

1190 D. Keunecke et al.

The MOE which was determined by sound velocity and resonance frequency was within the
known range for both species. The work to ultimate load of yew wood which was determined by 3-
point-bending was more than twice as high compared to spruce wood. The same applies for
fracture toughness KIC, determined in the RL and TL crack propagation system. The impact
bending strength of yew wood was also significantly higher. Table 1 gives a brief overview of
selected mechanical results.

TABLE 1. Material properties.

Reference
1. Bodig, J. and Jayne, B.A., Mechanics of Wood and Wood Composites, Van Nostrand
Reinhold, New York, USA, 1982.
 26. Fracture Mechanics Characterization of Wood 1191

CRITICAL CRACK LENGTHS IN FRP REINFORCED GLULAM BEAMS

Justin Desjarlais, William G. Davids and Eric N. Landis

University of Maine
Center for Advanced Engineered Wood Composites
Orono, Maine 04469 USA
justin.dejarlais@umit.maine.edu, william.davids@umit.maine.edu, landis@maine.edu

Tension reinforcement of glued-laminated (glulam) timber beams has been shown to be an

effective way to utilize lower grade wood in structural applications. Fibre reinforced polymer
(FRP) composites are a particularly good reinforcement because they can be fabricated to be
compatible with the compliance properties of the wood substrate. While the reinforcement has
been shown to be effective at improving strength and to some degree stiffness, there is a concern
that long-term exposure to moisture and freeze-thaw cycles could lead to some delamination
between the wood and the reinforcement. This is a particularly important issue when reinforced
beams are used as bridge girders, as environmental exposure can be severe. In order to make
quantitative assessments of the effects of these potential delaminations, a laboratory study of
mixed mode fracture toughness was combined with numerical simulations of in-service load
conditions. The goal was to make reasonable estimates of allowable delamination sizes for in-
service conditions.
Numerical Simulation of Strain Energy Release Rate
3D numerical simulations were conducted in order to gain an understanding of the stress states
at the bond line of a reinforced glulam. Fig. 1 illustrates the geometry of the beam considered. This
beam geometry was also used in an extensive laboratory study of fatigue properties. Strain energy
release rate, G, was determined for different load conditions by releasing pairs of adjacent nodes to
simulate the crack. G is determined by evaluating the work required to reconnect the series of
nodal pairs. In this manner, the contribution from mode I, mode II, and mode III can all be isolated.
An example plot of G versus crack length is shown in Fig. 2 for a delamination starting from the
end of the FRP. It is important to note that both mode I and mode II show significant contributions
to the total G.

FIGURE 1. Geometry of simulated and tested glulam beam.

1192 J. Desjarlais et al.

FIGURE 2. Plot of G versus crack length for FIGURE 3. Experimental configuration for
reinforced glulam beam. mixed mode fracture measurement.

Mixed-Mode Fracture Energy

Because of the contribution of both mode I and mode II in the calculated G for the beams, it
was clear that the fracture toughness of the wood-FRP bond required a mixed mode analysis.
Using a loading configuration as illustrated in Fig. 3, the contribution of mode II relative to mode I
fracture can varied by adjusting the position of the load, P. As the position, x, increases, the
contribution of mode II increases.
Small laboratory fracture specimens were prepared according to the same material and bond
specifications as the glulam beams, and were tested for fracture energy using the configuration of
Fig. 3. Fracture energy was determined by measuring the change in compliance that accompanied a
change in crack length. Fig. 4 shows the results of the fracture energy measurements plotted as in
terms of mode I – mode II interaction. The results indicate a moderate amount of interaction,
although scatter is significant. Furthermore, the results indicate the glulam beams can
accommodate a significant level of delamination before crack lengths become critical.

FIGURE 4. Interaction of mode I and mode II fracture in mixed mode configuration.

 26. Fracture Mechanics Characterization of Wood 1193

FAILURE ANALYSIS OF ENGINEERING WOOD PRODUCTS

Ian Smith, Monica Snow and Andi Asiz

University of New Brunswick
Faculty of Forestry and Environmental Management, PO Box 44555, Fredericton,
New Brunswick, E3B 6C2, Canada
ismith@unb.ca

This paper synthesizes information in the literature and results of studies at the University of New
Brunswick (UNB) to provide guidance on failure analysis of engineering wood products (EWP).
EWP are either solid materials like lumber and glued-laminated-timber, or composites
manufactured by hot-pressing veneers, wafers or strands of wood that were pre-coated with
synthetic adhesives. Important issues include size of specimen effects on apparent strength and
how to handle situations where stress concentrations arise. Consideration is given to both bodies
that are nominally undamaged and those that are pre-damaged (cracked) at the time of loading.
This elucidates the extent to which one might validly apply common continuum failure theories.
Limitations of continuum theories, whether or not they assume existence of a crack(s), are that they
can not always reliably predict the load level at which damage initiates, nor the location of failures.
Also, most theories cannot predict how failures propagate once initiated. This reflects limitations
in theoretical postulates when applied to EWP (Smith et al 2003). Even when they have a
physically correct basis fracture mechanics models still depend on an analyst’s ability to correctly
locate cracks. An advanced ‘bridging fiber’ model has been developed to include closing forces to
crack surfaces in a zone behind the crack tip (Smith and Vasic 2003). The closing forces replicate
the effect of partially peeled wood cells that toughen any naturally formed crack. This means that
the bridging crack model is capable of capturing a transition from an artificial crack to a ‘natural’
crack condition, meaning that it can be used to model crack propagation processes. Other fracture
models cannot represent the artificial to natural crack transition so are unsuitable for predicting
crack growth. Ability to predict crack growth equals ability to predict collapse loads for initially
stable systems. Whatever fracture model is employed, one still requires a reliable empirical
knowledge or algorithm of how crack fronts will advance. Usually, but not always, cracks in solid
wood develop in planes that lie parallel to the grain direction. For EWP composites prediction of
how cracks, or other damage, will develop is a much less certain undertaking! A very promising
generalized failure prediction tool is discrete element (lattice network) based models (Smith et al
2003). Potentially discrete element models are the most powerful tools for failure analysis of EWP
materials, because they can mimic effects of various morphological structures. Lattices can
account for global and local structured variations in properties, and localized random variations
using simulation techniques. This said, much remains to be done on the subject, particularly in
regard to modeling strategies and element property calibrations for composite EWP. Modern
composite EWP exhibit morphological characteristics at multiple scales. Researchers at UNB and
elsewhere are addressing these issues.
The remainder of the document presents a typical example of work in progress to predict
failure processes in EWP materials. The chosen example is assessment of how members of solid
wood (Pine lumber) an Laminated Strand Lumber (a modern composite made from wood strands)
fail under loading from steel bolts that passes through them, to create a connection in which the
bolts load the EWP perpendicular to the strong material axis, Fig. 1. LSL like lumber is a distinctly
anisotropic material, but because of cross laminating between bonded strands the former is
relatively resistant to splitting. The test arrangement loads the bolts in double shear. Transparent
high strength side-plates were used instead of steel, as would be normal, so that fracturing patterns
1194 I. Smith et al.

could be observed. A continuum finite element model was created to analyze the arrangement and
assist in interpretation of the failure development. Predicted element stresses were used in
conjunction with the Tsai-Wu failure criterion (Tsai and Wu 1971) to predict likely locations of
damage, i.e. the potential sites for initiation of global failure, Fig. 2. As can be seen different
failure initiation sites are predicted for Pine and LSL members, and different stress components
govern depending on the zone within the member. These predictions match experimental
observations well. The most likely failure for a Pine member is fracturing between the bolts, while
for a LSL member the most likely failure process is excessive tension on the bottom face. This
reflects the effects of the different stiffness and strength properties of the two materials.

FIGURE 1. Test with failed pine center FIGURE 2. Predicted likely failure initiation
member sites: a) Pine member, b) LSL member

References
1. Smith, I., Landis, E. and Gong, M. 2003. “Fracture and fatigue in wood”, Wiley & Sons,
Chichester, UK.
2. Smith, I. and Vasic, S. 2003. ”Fracture behaviour of softwood”, Mechanics of Materials,
35(2003): 803-815.
3. Tsai, S. W. and. Wu, E. M. 1971. “A general theory of strength for anisotropic materials”,
Journal of Composite Materials, (1): 58-80.
 26. Fracture Mechanics Characterization of Wood 1195

MODELIZATION OF SLOW CRACK GROWTH IN WOOD CONSIDERED AS A

DAMAGE VISCOELASTIC MATERIAL

M. Chaplain and G. Valentin

Laboratoire de Rheologie du Bois de Bordeaux - France
69 route d’Arcachon 33612 Cestas Cedex
chaplain@lrbb3.pierroton.inra.fr

Prediction of delayed failure of wooden beams is an important problem in timber engineering

especially for beams containing stress concentrations (as notches or tapered end-notches or
openings). Several models have been developed to provide duration of load (DOL) of timber
structures. The damage theory is probably the most employed one to predict the time to failure of
wood elements. However, the fracture mechanics approach developed to provide the crack growth
can be performed to obtain the time to failure as well. A correct modelling of delayed fracture
needs to model two durations: duration before the apparition of the crack (incubation time) and
duration of the crack growth, that is the time to reach a critical length [1].
The incubation time may be predicted by a damage approach using a damage model. The
second model of Barrett and Foschi [2], with non-linear damage evolution and non-linear
cumulative damage, has been chosen (1):

­ b
° dD § F (t)  F o ·
a ˜ ¨¨ ¸¸  O .D(t) if F (t) ! F o
° dt © Fs ¹
®
° dD
° 0 if F (t)  F o
¯ dt (1)
Where F is the applied load, Fs the static strength and Fo a threshold load. a, b and O are parameters
depending on the material properties.The Fo and the parameters used in the damage model is
calculated from the results of short term tests on notched beams. In this approach, damage appears
as a characteristic parameter D ranging from 0 at the beginning of loading to Di at the crack
initiation. The critical damage parameter Di is evaluated from stiffness variation finite elements
calculations for a crack length equal to 5 mm.
The viscoelastic crack model (VCM) used in this study is based on the Schapery theory [3].
The crack grows in an orthotropic viscoelastic medium and a damage area is assumed to exist at
the crack tip. This fracture mechanics model introduces a viscoelastic compliance. Mixed mode
crack propagation can occur in notched beams. However the model has only been established for
the growth of a crack along the grain direction in opening mode. The reduced creep compliance in
mode I, N 22v (t ) can be assumed to be represented by a power law (2):
N 22v ( t ) A o  A 2 .t n (2)
Ao, A2 and n are material parameters.
1196 M. Chaplain and G. Valentin

1/ m
ª º
« »
da S « C2 » K I2 (11 / m )
O1m/ m
The crack velocity isgiven by [4] : dt 2 « § 2 ·» V m I 1 2 (3)
« 8* ¨1  K I ¸»
« ¨© 2
K Ig ¸»
¹¼
¬

with K Ic 2 G I / A o is the critical stress intensity factor

For finite dimension beams, an expression of the stress intensity factor KI depending on the
shape of the beam and of the type of loading and on the elastic stiffness properties of the material is
introduced in VCM. It can be obtained by finite element calculations. The duration of load
predicted by the model are compared to experimental results of bending tests up to three months on
notched Laminated Veneer Lumber beams (LVL). The figure 1 illustrates the evolution of the
crack length and of the propagation time obtained experimentally and by simulation under a step
by step loading.

FIGURE 1. Step by step loading on LVL notched beams : propagation time (a) - crack evolution
(b) : experiments and simulations

The predictions of this damage model are compared to the relation proposed by Nielsen [5].
Compared to experiments on notched LVL beams, both models give fair predictions of the time to
failure.

References
1. Chaplain, M., Valentin, G., In Proceeding of the Third international Conference of the
ESWM, Villa Real, Portugal, 2004,135-142.
2. Barrett, J.D., Foschi, R.O., Can. J. Eng. Vol 5,. 505-514, 1978
3. Schapery, R. A., International Journal of Fracture, vol 11: 141-159, 1975
4. Valentin, G., Chaplain, M., In Proceeding of the first international Conference of the ESWM,
Lausanne, Switzerland, 2001, 243-253
5. Nielsen, L. F., In Proceeding of the International Workshop on Duration of Load in Lumber
and Wood Products, Richmond, Canada, 1985, 67-78
 26. Fracture Mechanics Characterization of Wood 1197

MODE I CRACK PROPAGATION IN SOFTWOOD, MICROANALYSES AND

MODELING

P. Navi and M. Sedighi-Gilani

Institute of Material Science
Ecole polytechnique Federale de Lausanne
MX-G Station 12, 1015 Lausanne, Switzerland
parviz.navi@epfl.ch
marjan.sedighigilani@epfl.ch

A method was developed for microscopic observation of cracks in wood which provides
instructive information on crack initiation and crack propagation. This method has been used on
softwoods to understand their fracture behavior at cellular level during cracking in mode I and
mode II. It has been shown that different mechanisms would be activated depending on the
orientation of the growth rings with the plane of fracture.
In order to observe clearly crack extend and network of microcracks in the complex
microstructure of wood, the cracked part was cut out of the test specimen and impregnated with
epoxy resin[1]. A fluorescent dye was added to the resin to give a good contrast during
microscopic observation. This resin has high fluidity and penetrates easily into the wood. To
prepare the slices for microscopic observations, the impregnated sample containing the crack could
be cut at any directions. In RL and TL crack propagations most information can be extracted from
perpendicular to the grain direction cuts. A confocal microscope with a 488 nm laser was used for
observation. To stop the fracture process of the specimen before complete cracking during test,
fracture test has to be conducted with any specimen giving stable crack propagation.
Our microscopic observations haven’t showed the classically defined “fracture process zone”
as a wide network of micro-cracks was formed in front of the crack. Initial damage that could be
observed optically is localized on a few cells and expands until a continuous crack is formed. In
this sense a fracture process zone within certain distance along the specimen width could be
defined. In typical RL crack propagation, the crack develops through earlywood ring as this is the
weakest part of the structure. The first micro-cracking, that has been observed is located along a
line and consist both on intercellular separation of cells and intercellular break of cell walls. All
micro cracks seems to be located within one or two cells width and no other disperse micro-cracks
are observed. When a crack crossing the late wood ring, the crack can be wide open with the block
of latewood cells bounded to both sides of the crack making a bridge. Such bridge can transmit
forces both in vertical and lateral direction of the crack.
In the TL crack propagation the cracking process is quite different. As the late wood ring has a
higher modulus of elasticity than the earlywood ring, it supports most of the stress and as the
interface between the cells or between the layers S1 and S2 of the cells are the same, the cracks
begin into the latewood ring and then traverse the early wood. This is what was usually observed
during the microscopic observation. In the case where the crack propagation occurs at angle
between TL and RL, either one or the other mechanism typical for TL or RL will dominate. If the
angle of the annual ring with crack plane is more than 30 – 40, then the TL mechanism will
dominate and the first damage will occur in the latewood.
To better understand the role of the micromechanical characteristics of wood and fibers in the
wood fracture mechanics a fracture model is being developed. Morphological based models, i.e.
lattice model, was used, as comparing to continuum fracture mechanics, it has the capacity to
represent the specific features of wood microstructure and predict the damage pattern and force-
1198 P. Navi and M. Sedighi-Gilani

displacement curve of wooden samples[2, 3]. However if lattice is represented in cellular level
(each elements as a wood fiber), the computational needs to model the large-scale specimens
becomes quite large. As the crack path and fracture process zone prior to crack propagation in
wood is approximately predictable, to reduce the computing time and storage requirements, only
the neighborhood of fracture process zone was modeled with lattice network.
As the first step, to model the fracture propagation, a 3D lattice of beam elements was defined.
The beam elements, depends to representing the earlywood or latewood fibers, have different
properties and cross sectional area. To consider the effect of bordered pits and non-uniformities of
microfibril angle in earlywood fibers[4], each fiber was modeled by three different beam segments
with different properties. As the microfibril angle along latewood fiber is approximately uniform,
each latewood fiber was modeled by one beam element. Each fiber was connected to its neighbors
by diagonal and vertical beam elements representing the bounding between fibers (intercellular
layers). The effective properties of lattice are estimated through the behavior of two parts, solid
and lattice, by homogenization method. The lattice is placed in fracture process zone of the 3D
continuum body which has common nodes in its boundary interfaces with lattice part. Removing
the beam elements with limit strain criteria during the loading process shows the crack propagation
paths in cellular level of microstructure.

References
1. Job, L. and P. Navi. Microscopic analysis of crack propagation in softwood, model I and II. in
International COST 508 wood mechanics conference. 1996. Stuttgart, Germany.
2. Landis, E.N., et al., Coupled Experiments and Simulations of Microstructural Damage In
Wood. Experimental mechanics, 2002. 42(4): p. 1-6.
3. VanMier, J.G.M. and M.R.A. VanVliet, Influence of micostructure of concrete on size/scale
effects in tensile fracture. Engineering fracture mechanics, 2003. 70: p. 2281-2306.
4. Sedighi-Gilani, M., H. Sunderland, and P. Navi, Within-fiber nonuniformities of microfibril
angle. accepted to be published in Wood and fiber science.
 26. Fracture Mechanics Characterization of Wood 1199

FRACTURE PROPERTIES OF PINE AND SPRUCE IN MODE I

N. Dourado1, S. Morel2, M. F. S. F. de Moura3, G. Valentin2 and J. Morais1

1
CETAV/UTAD, Depart. de Engenharias, Quinta de Prados, 5001-911 Vila Real, Portugal
2
LRBB, UMR 5103 (CNRS/INRA/Univ. Bx1), 69 route d'Arcachon, 33612 Cestas, France
3
Fac. de Engenharia da Universidade do Porto, R. Dr. Roberto Frias, 4200-465, Porto, Portugal
ndourado@utad.pt, morel@lrbb.u-bordeaux.fr, mfmoura@fe.up.pt

The fracture behaviour of a quasibrittle material such as wood, characterized by the development
of a large fracture process zone (FPZ), is nowadays well-known to be efficiently described by
cohesive crack models. The most common applications of fictitious or cohesive crack models used
to simulate nonlinear fracture mechanics of quasibrittle materials are considered as variations [1]
of a model proposed by Hillerborg and co-workers [2]. These applications need to introduce a
cohesive zone at the crack tip, i.e., a fictitious line crack transmitting normal stress V dependent on
the corresponding opening displacement w. Cohesive crack models are typically applied through
finite elements calculations. Interface elements or springs [3] with prearranged strain-softening
properties are integrated into the structural model along the most probable crack path. Since the
pioneering application of cohesive crack models to wood due to Boström [4], a so called bilinear
strain-softening model was applied by Stanzl-Tschegg et al. [5] to obtain wood load-deflection
curves according to a developed wedge-splitting test protocol.
In this study, from Three Point-Bending (TPB) fracture tests inducing a mode I failure in two
species of wood: Maritime pine (Pinus pinaster Ait., Lin.) and Norway spruce (picea abies L.), the
Load-Deflection curves and the corresponding resistance curves (R-curve) (Lawn, 1993[6]) are
estimated, then simulated through FE calculations using a cohesive crack model. In a first step of
this study, from the Load-Deflection curves, the corresponding resistance curves are estimated in
the framework of ‘equivalent linear elastic fracture mechanics (LEFM)’ in which the increase of
the specimen compliance due to the FPZ development is attributed to the propagation of a linearly
equivalent elastic crack (Morel, 2005 [7]). On this basis, typical rising R-curve, common among
materials that exhibit toughening mechanism, are observed. After a characteristic equivalent crack
length ac, the resistance to crack growth becomes independent of the equivalent crack length
defining a plateau value of the resistance to crack growth also called critical resistance and noted
here as GRC. Note that such a levelling off of the R-curve at longer crack lengths might indicate
that the influence of the toughening mechanism is not infinite, especially for wood, where fibre
bridging requires sufficient deformation to produce closing force (Smith et al., 2003 [1]). The
second step consists on simulating the Load-Deflection curves on the basis of a cohesive crack
model. First, the fracture energy GF characterizing the area under the softening function is fixed to
the plateau value GRC of the experimental resistance curve, as suggested in (Coureau and Morel,
2004 [8]). We show that this assumption can be easily verified through preliminary simulations.
On this basis, a developed Genetic Algorithm is used to estimate the characteristic values of the
softening function, such as the break point coordinates (wb, fb), and the tensile strength ft or the
critical opening wc (Figure 1), enabling to obtain a good agreement between simulated and
experimental Load-Deflection curves (Figure 2). The classical bilinear softening function used in
these simulations allows discussing the obtained results in terms of fracture energy associated to
microcracking and to fibre bridging. Finally, it is shown that the characteristic values of the
bilinear constitutive law estimated for both wood species induce different critical sizes of the
softening zones. The implications of these FPZ sizes on the fracture properties of both wood
species are discussed.
1200 N. Dourado et al.

FIGURE 1. Bilinear softening model. GRC=Gf1 + Gf2.

(a) (b)
FIGURE 2. Load - displacement curves (a) Pinus pinaster (b) Picea abies. CCM: Cohesive Crack
Model.

References
1. Smith, I., Landis, E. and Gong, M., Fracture and Fatigue in Wood, John Wiley & Sons, 2003.
2. Hillerborg, A., Modeer, M. and Petersson, P.E., Cement and Concrete Research vol. 6, 773-
782, 1976; Hillerborg, A., Int. J. Frac. vol. 51, 95-102, 1991.
3. de Moura MFSF, Gonçalves JPM, Marques AT, Castro PMST. Modelling compression
failure after low velocity impact on laminated composites using interface elements. Journal
Composite Materials 1997; 31:1462-79.
4. Bostrom, L., Method for determination of the softening behaviour of wood and the
applicability of a nonlinear fracture mechanics model. Universitatis Gothorum, CODEN:
LUTVDG/(TVBM-1012) 1: 132, 1992.
5. Stanzl-Tschegg, S.E., Tan, D.M. and Tschegg, E.K., Wood Sc. Tech. vol. 29, 31-50, 1995.
6. Lawn, B.R., Fracture of Brittle Solids, 2nd Ed., Cambridge University Press, 1993.
7. Morel, S., Dourado, N., Valentin, G. and Morais, J., Int. J. Fract. vol. 131, 385-400, 2005.
8. Coureau, J.-L., Morel, S., In Proceedings of the Third Conference of the European Society of
Wood Mechanics, ESWM2004, Vila Real, Portugal, September 6-8, 2004.
 26. Fracture Mechanics Characterization of Wood 1201

INFLUENCE OF THE SPECIMEN GEOMETRY ON R-CURVE: NUMERICAL

INVESTIGATIONS.

Christophe Lespine, Stephane Morel, Jean-Luc Coureau and Gerard Valentin

Lab. de Rhéologie du Bois de Bordeaux, UMR 5103 (CNRS/INRA/Univ. Bordeaux. 1)
Domaine de l’Hermitage, 69 route d’Arcachon, 33612 Cestas Cedex, France
morel@lrbb.u-bordeaux.fr

During fracture test of quasibrittle materials such as concrete, rocks, or wood, a fracture process
zone (FPZ) is observed ahead of the crack tip. Various toughening mechanism take place in this
FPZ such as microcracking, crack branching or crack bridging. If Linear Elastic Fracture
Mechanics (LEFM) cannot be directly applied to such quasibrittle fractures, an adaptation of
LEFM, which consists in considering an equivalent linear elastic problem, provides useful
approximations of fracture properties [1,2,3]. Within the framework of ‘equivalent LEFM’, the
increase of the specimen compliance due to the FPZ development is attributed to the propagation
of an elastically equivalent crack. On this basis, the fracture behaviour of quasibrittle materials is
characterised by a more or less pronounced rising resistance curve, briefly called R-curve [4]. For
instance, R-curve obtained from a Norway spruce (Picea Abies L.) Double Cantilever Beam (DCB)
and two kinds of Tapered DCB (TDCB) specimens extracted from [3] are plotted in Fig. 1. Such
rising R-curves are common among materials that exhibit toughening mechanism. In fracture
perpendicular to grain of wood, it has been shown that the main toughening mechanism is crack
bridging [5,6]. After a characteristic equivalent crack length ac, the resistance to crack growth
becomes independent of the equivalent crack length defining a plateau value of the resistance to
crack growth also called critical resistance and noted here as GRC.

Fig 1.Experimental R-curves obtained from Norway Spruce DCB, TDCB30 and TDCB60
specimen [3]

Several experiments have evidenced that the R-curve behaviour is not an intrinsic material
property but appears geometry dependent. As shown in Fig. 1, the dependence on the geometry is
usually observed on the critical resistance GRC but also on the crack length ac for which this
plateau value is reached.
On the other hand, cohesive crack models are nowadays well-known to successfully describe
quasibrittle fracture behaviour and especially for wood [7] where crack bridging induces strong
toughening.
1202 C. Lespine et al.

In this study, numerical investigations using cohesive crack model implemented in joint-
elements are used to study the link between specimen geometry and R-curve behaviour. In a
previous study [8], it has been suggested that the fracture energy GF characterizing the softening
behaviour must equal the crack growth resistance GRC, hence, the fracture energy GF can not be
considered as an intrinsic material property, GRC being geometry dependent as shown in Fig. 1. On
this basis, we propose to consider the fracture energy GF as GF =GD * eFPZ where GD is a critical
damage energy release rate per unit volume of damaged material (i.e. per unit volume of FPZ) and
eFPZ corresponds to the width of FPZ which is proportional to the characteristic size of the
damaged zone. The critical damage energy GD is assumed to be an intrinsic material property and
hence the width of the FPZ eFPZ should be dependent on the specimen geometry. Moreover, from
a classical bilinear softening behaviour, we show that the tensile strength of the interface can be
considered as an intrinsic material property while the ultimate opening of the interface can be
assumed dependent on the width of the FPZ eFPZ.
We show that the assumptions concerning the parameters characterizing the softening
behaviour allow us to obtain simulated Load-Deflection curves in agreement with those obtained
from DCB and the two kinds of TDCB specimens extracted from [3]. Finally the implications of
such a softening behaviour on the fracture properties of wood are discussed.

References
1. Bazant, Z.P., Eng. Fract. Mech. vol. 69, 165-205, 2002.
2. Morel, S., Dourado, N., Valentin, G. and Morais, J., Int. J Fract. vol.131, 385-400, 2005.
3. Morel, S., Mourot, G. and Schmittbuhl, J., Int. J. Fract. vol. 121, 23-42, 2003.
4. Lawn, B.R., Fracture of Brittle Solids, Cambridge University Press, 2nd ed, 1993.
5. Vasic, S. and Smith, I., Eng. Fract. Mech. vol. 69, 745-760, 2002.
6. Smith, I., Landis, E. and Gong, M., Fracture and Fatigue in Wood, John Wiley and Sons Ltd,
2003.
7. Stanzl-Tschegg, S.E., Tan, D.M. and Tschegg, E.K., Wood Sc. Tech. vol. 29, 31-50, 1995.
8. Coureau, J.L. and Morel, S., In Proceedings of the Third Conference of the European Society
of Wood Mechanics, ESWM2004, Vila Real, Portugal, September 6-8, 2004.
 26. Fracture Mechanics Characterization of Wood 1203

FRACTURE BEHAVIOUR AND CUTTING OF SMALL WOOD SPECIMENS IN

RT-DIRECTION

S. Koponen and P. Tukiainen

Senior Research Scientist, Research Scientist
Helsinki University of Technology
Simo.Koponen@tkk.fi, Pekka.Tukiainen@tkk.fi

Wood is strongly anisotropic and inhomogeneous material. The total number of different fracture
propagation systems is 8. Most of the earlier tests in Mode I are performed in LR- or LT-system
and only few in RT system. RT crack propagation system is important in rotary peeling of veneer
logs, in defibration process, Persson [1], in the case of stresses perpendicular to grain in curved
glulam beams, Dill-Langer et al. [2], and in the case of drying stresses, Boström [3], for example.
The aim of this work is to study fracture in RT-system, and the main emphasis is in pure Mode I
loading, but limited number of cutting tests with knife and razor blade has been performed to
understand veneer-manufacturing process. The quality of the veneers depends on the mechanism
of the crack propagation in cutting process, the material property variation of wood (density e.g.),
moisture content of wood and the temperature of log.
Fracture properties and crack propagation under pure Mode I loading of wood in RT crack
propagation system are studied. The main test series and theoretical analysis were made for birch
hardwood. Limited number of tests for spruce has also been performed. The moisture content
(MC) of test pieces was either 14 % or the tests were done using green wood (birch MC=63 %).
The oven density Ug based on green volume of tested birch was 525 kg/m3.
In tests small specimens were used and the tests were performed with micro-testing device.
The size of the specimen was 37.5x36x15mm3 and the use of micro-testing device makes it
possible to observe the behaviour of process zone for example under light microscope or in situ
tests in the chamber of Environmental Scanning Electron Microscope (ESEM). In these tests the
digital camera and light microscope were used. The crack mouth opening (CMOD) displacement
was measured using dial gauge, and deformation of the specimen and crack propagation were
calculated from digital images using image correlation technique.
The specimen size, the characteristic length of the specimen and the ligament length (15 mm)
were small compared to the process zone size (1…5 mm). Thus, Linear Elastic Fracture Mechanics
(LEFM) is not valid and clear size effect on fracture energy is expected. The analysis is done using
Nonlinear Fracture Mechanics (NFM). Fictitious crack model (FCM) and Finite Element Method
(FEM) were used in the numerical simulations. The micro cracking and bridging is taken into
account using ‘cohesive zone’ with closing stress as a function of crack opening displacement.
In this paper the crack mouth opening displacement (CMOD) is the increase of the dimension c
at the line between loading bolts (Fig 1.). The initial crack was made with small band saw and
razor blade. The initial length of crack, a, is 15 mm. Fig. 1 shows the CT-tests specimen and load-
displacement curves obtained for wet birch in RT propagation system.
1204 S. Koponen and P. Tukiainen

FIGURE 1. Test specimen (w=30mm, b=14.5 mm, t=15mm, a=15mm, c=6.3mm) and loading
curve of wet CT-specimen (RT004, RT006 and RT007).

The digital images were used to measure the increase of the distance of lines located at the
positions ±c/2 from the symmetry line of the specimen. The increase of this distance is called crack
opening displacement (COD) in this paper and it was compared to calculated values. Theoretical
stresses were also analyzed and stresses along the ligament are shown in fig. 2.

FIGURE 2. Calculated stress distributions along the ligament (FCM: ft=3N/mm2, Gf=480N/m)

The measured fracture energy Gf for wet birch specimen was 697 N/m and it was 850…1100
N/m at 14 % MC. The calculated crack propagation and the opening deformation were in good
agreement with test results. The process zone length was 3…4 mm.

References
1. Dill-Langer, G. and Lütze, S. and Aicher, S., Wood Sci. Technol., vol. 36, 487-499, 2002
2. Persson, K., Micromechanical modelling of wood and fibre properties, Doctoral thesis, Publ.
TVSM-1013, Div. of Struc. Mech., Lund University, Sweden, 2000.
3. Bostrom, L., Method for Determination of the Softening Behaviour of Wood and the
Applicability of a Nonlinear Fracture Mechanics Model, Doctoral thesis, Report TVSM-
1012, Div. of Struc. Mech., Lund, Sweden, 1992.
 26. Fracture Mechanics Characterization of Wood 1205

FRACTURING OF WOOD UNDER TORSIONAL LOADING: FRACTURE

MECHANISMS AND MECHANICS

E. K. Tschegg and S. E. Stanzl-Tschegg

Vienna Technical University, Institute of Solid State Physics
Karlsplatz 13, A-1040 Vienna, Austria
BOKU, University of Natural Resources and Applied Life Sciences,
Institute of Physics and Materials Science
Peter-Jordan Strasse 82, A-1190 Vienna, Austria
Tschegg@ifp.tuwien.ac.at, Stefanie.Tschegg@boku.ac.at

Damage of wood under mechanical loading is mostly induced by some kind of mixed mode
loading, like for example, superposition of a crack opening mode (mode I) plus torsional load
(mode III). In addition, the high anisotropy of wood itself, with directions and paths for easy crack
propagation in the neighbourhood of areas with high crack resistance, leads to a mixed-mode
loading condition. In order to characterize superimposed loading, systematic studies have been
performed of loading under pure mode I, and it was tried, to also realise pure mode II or mode III
testing conditions. It could be shown, however, that sliding modes cannot be realized in practice
without any mode I component being present, as fracture surface roughness, causes crack tip
opening. Therefore, Ehart et al. >1@ and Frühmann et al. >2@ have performed studies on this fracture
surface interaction and resulting mixed mode fracture, as well as on damage and fracture
mechanisms during nominal mode III loading of wood and wood composites. In addition, a testing
technique has been developed >3@ to perform mixed mode (mode I plus II) fracture experiments,
with different defined mode II and, respectively, mode I components.
The characterization of fracture of wood as a highly anisotropic material was concentrated on
concepts of linear elastic fracture mechanics (LEFM) mostly in the past, before Schniewind and
Pozniak >4@, Aicher and Reinhardt >5@, Boatright and Garrett >6@, Holmberg et al. >7@ and Tschegg
et al. >8@, among others, introduced and used non-linear elastic fracture mechanics parameters, like
the fracture energy as a characterizing parameter. To obtain this, complete load-displacement
diagrams have to be detected, which is made possible by testing procedures that allow stable crack
propagation during the whole experiment, until final fracturing takes place. As a useful technique,
the wedge splitting technique has been developed >9@.
This technique has been further developed to perform superimposed mode I plus mode II load
experiments >3@, and investigations by systematically varying the portions of mode II and I,
respectively, demonstrated that the contributions are not additive, but coupled >10@. To the authors´
knowledge, no systematic experimental investigations exist on superposition of mode III and mode
I loading on wood. As has been pointed out by Ehart et al. >1@ and Frühmann et al. >2@, remarkable
portions of crack opening forces and thus mode I loads are present under nominal mode III loading
owing to sliding and friction of the already formed fracture surfaces.
As a consequence, in a new study, experiments were performed by varying the tension (mode
I) and torsion (mode III) components and detecting the resulting load-displacement curves. The
influence of mode III and mode I can be studied separately, as they can be changed independently
with the used tension-torsion machine. With this, wood of spruce and beech was tested with a
special testing set-up in different orientations. Load-displacement diagrams have been recorded,
which allow to determine the specific fracture energy and other characteristic values, like
maximum load and stress intensity or initial slope of the load-displacement diagram, which
characterizes the material stiffness. The results allow discussing the different mechanisms of
1206 E. K. Tschegg and S. E. Stanzl-Tschegg

cracking and demonstrate the importance of quantifying the components of mode III and mode I in
order to better understand their influence on damage of wood. Fracture mechanical values, as well
as fractographic features strongly depend on wood species, its structure and orientation, as well as
its humidity and other influencing parameters.

References
1. R.J.A. Ehart, S.E. Stanzl-Tschegg and E.K. Tschegg, Eng. Fracture Mechanics 61 (1998)
253-278
2. K. Fruhmann, A. Reiterer, E.K. Tschegg and S.E. Stanzl-Tschegg, Phil. Mag. A, Vol. 82
(2002) 3289-3298
3. E.K. Tschegg and S.E. Stanzl-Tschegg, Patent No. AT 409.038, 7.2.2002
4. A.P. Schniewind and R.A. Pozniak, Eng. Fracture Mechanics 2 (1971) 223
5. S.W. Boatright and G.G. Garrett, J. Mat. Science 18 (1983) 2181
6. S. Aicher and H.W. Reinhardt, Holz als Roh- und Werkstoff 51 (1993) 215
7. S. Holmberg, K. Persson and H. Petersson, Comp. Struct. 72 (1999) 459
8. S.E. Stanzl-Tschegg, D.M. Tan and E.K. Tschegg, Wood Sci. and Technology 29 (1995) 31-
50
9. E.K. Tschegg, Patent No. AT 233/96, 390 328, 1986
10. E.K. Tschegg, A. Reiterer, T. Pleschberger and S.E. Stanzl-Tschegg, J. of Materials Science
36 (2001) 3531-3537
 26. Fracture Mechanics Characterization of Wood 1207

ON THE INFLUENCE OF HUMIDITY CYCLING ON FRACTURE PROPERTIES

OF WOOD

S. Vasic and S. Tschegg

Marie Curie EU Fellow, Professor
Institute for physics and materials science, BOKU University,
Peter Jordan Strasse 82, 1190 Vienna, Austria
svetlana.vasic@boku.ac.at
stefanie.tschegg@boku.ac.at

The importance of environmental effects on the wood structural performance has been known for
ages, moisture being the most pronounced degradative effect that needs to be taken into account at
different levels of wooden structures durability. Moisture content has an effect on many
mechanical properties bellow the fibre saturation point, with the increase in the property with the
decrease in moisture content.
In-service, wood is exposed to both long-term (seasonal) and short-term (daily) changes in the
relative humidity and temperature. These changes are usually gradual, and short-term fluctuations
tend to influence only the wood surface in large cross-section wooden components. It is a common
wood technology practice to have the drying targeted moisture content as close as possible to the
one wood will experience in the in-service conditions. This minimizes the seasonal variation in
moisture content and the dimensional changes after the installation, avoiding the problems such as
permanent deformations and cracking. However, this strategy has its limits, and it was deemed
necessary to perform the fracture mechanics study on the influence of humidity cycling on the
fracture properties. The results of the experimental and numerical investigations are reported
herein.
Four different wood species were chosen for this study, namely spruce, pine, oak and beech, all
cut in green state in the Lower Austria. The specimens prepared for opening mode I tests were
conditioned from the green state to the relative humidity of 65 percent, at 20oC. The wood samples
were then subjected to varying humidity 30-70 percent and for the harsher environment 40-90
percent relative humidity. The number of humidity cycles was a variable, ranging from 1 to 5 to 10
cycles. The length of a humidity cycle was chosen to be 1 week, which was considered sufficient
for small fracture specimens to reach equilibrium through each cycle, resulting in steady-state
moisture conditions.
In order to investigate the physical mechanisms of woods fracture when different humidity
cycles are imposed, in-situ real-time Environmental Scanning Electron Microscope (ESEM)
experiments were performed. For this purpose, wedge-splitting loading stage has been built to
enable specimen loading and fracture propagation observations at the same time (Fig. 1, Frühmann
et al. [1]).
The in-situ real-time ESEM experiments enable evaluation of complete stable load-
deformation curves, from which it is possible to derive common fracture mechanics properties.
1208 S. Vasic and S. Tschegg

FIGURE 1. Schematics of wedge-splitting loading set-up in ESEM

Fracture toughness is obtained using Finite Element Method with the orthotropic material
model, while total fracture energy and specific fracture energy can be obtained directly form the
load-deformation curves.
Different physical mechanisms of fracture initiation and propagation were evidenced
depending on the magnitude of humidity, the number of cycles and wood species. This implies the
interplay of the effect of moisture content and its history with the wood ultrastructure. Also,
fracture parameters such as fracture toughness, total fracture energy and the specific fracture
energy were affected by the humidity conditions and the wood species. In softwoods, bridging of
fibres behind the crack tip was evidenced as the main toughening mechanism, similar to previous
findings (Vasic and Smith [2]). The number of humidity cycles tends to decrease the fracture
parameters such as fracture toughness and fracture energy, implying the development of damage
through the humidity cycling.
Numerical modeling of the moisture transfer was performed using the commercial code
Abaqus 6.2 [3]. The moisture transfer analysis was coupled to stress analysis of the wooden
specimens, with the adjustment of elastic properties depending on the moisture content and wood
species and their anisotropy. The numerical analysis resulted in the computation of stress intensity
factors employing the quarter-point Barsoum finite elements. It was shown that the obtained
numerical values correlate well with the experimental data. Further numerical efforts will
concentrate on incorporating the moisture transfer analysis into lattice fracture finite element
model, in order to simulate complete load-deformation curves and damage patterns throughout the
crack propagation process.

References
1. Fruhmann, K., Burgert I., Stanzl-Tschegg S.E., Tschegg E.K., Mode I Fracture Behaviour on
the Growth Ring Scale and Cellular Level of Spruce (Picea abies [L] Karst.) and Beech
(Fagus sylvatica L.) Loaded in the TR Crack Propagation System, Holzforschung, 57, 653-
660, 2003
2. Vasic S., Smith I., Bridging Crack Model for Fracture of Spruce, Engineering Fracture
Mechanics, Vol. 69, 745-760, 2002
3. ABAQUS/CAE, User’s Manual, Version 6.2, Hibbit, Karlsson & Sorensen, Inc. 2001
 26. Fracture Mechanics Characterization of Wood 1209

DETERMINATION OF COHESIVE FRACTURE PARAMETERS FOR WOOD

T. Astrup1, J. F. Olesen1, L. Damkilde2 and P. Hoffmeyer1

1
Department of Civil Engineering, Technical University of Denmark, Denmark
2Esbjerg Institute of Technology, Aalborg University, Denmark

tha@byg.dtu.dk, jfo@byg.dtu.dk, ld@aaue.dk, ph@byg.dtu.dk

The application of non-linear fracture mechanics to wood is a relatively new topic in the area of
wood science, however, linear elastic fracture mechanics (LEFM) was first applied to wood in the
nineteen sixties. The fracture propagation in wood is mainly governed by two aspects, namely the
direction of the principal stresses, and the microstructure. There are six principal crack propagation
systems in wood, which are illustrated in Fig 2, see eg. Smith et al. [1] or Reiterer and Tschegg [2].
The usual presumption is that wood is perfectly brittle–elastic (LEFM). When a crack propagates
perpendicular to grain, the observed fracture behaviour is quasi-brittle, i.e. a tensile softening
branch exists [2]. This indicates that a fracture process zone of some length exists. The LEFM
approach is not an adequate approximation if the specimen size is of the same order of magnitude
as the length of fracture process zone.

FIGURE 1. Wedge splitting test. Specimen and experimental setup. From [6]

In recent years this quasi-brittle fracture behaviour has been studied by e.g. Reiterer and
Tschegg [2] ,Stefansson [3] ,Gustafson [4] and several references given in [1].
In this paper the quasi-brittle fracture behaviour of Norway spruce (picea abies) is studied. The
wedge splitting test (WST), shown in Fig. 1 and further described by Rossi et al. [5] is used to
produce stable crack growth, i.e. to obtain a softening branch, as shown in Fig. 3. The inverse
analysis, as described by Østergaard [6], is performed in order to determine the stress-crack
opening relationship from the global load-crack mouth opening displacement curve (CMOD
curve), i.e. the softening behaviour. Several different test specimens conditioned at varying
climates are used, in order to investigate the influence of moisture content on the fracture
behaviour.
Similar tests have been performed by Reiterer and Tschegg [2]. However, only the fracture
energy and the relationship between the load and CMOD were presented, whereas the softening
behaviour in terms of the stress-crack opening relationship was not extracted.
1210 T. Astrup et al.

FIGURE 2. Principal crack propagation FIGURE 3. Typical load–CMOD curves for

systems of wood. From [2]. crack propagation perpendicular to grain.
From [2].

The stress-crack opening relationship can be used to model crack propagation e.g. in end-
notched beams or around mechanical joints.

References
1. Smith, I., Landis, E. and Gong, M., Fracture and Fatigue in Wood, WILEY, England, 2003.
2. Reiterer, A. and Tschegg, S., Journal of Materials Science, vol. 37, 4487-4491, 2002
3. Stefansson, F., Fracture Analysis of Orthotropic Beams, Lund University., Licentiate
Dissertation, 2001.
4. Gustafsson, P.J., Fracture Perpendicular to Grain – Structural Applications, in
Thelanderson, S. and Larsen, H. J., Timber Engineering, WILEY, England, 2002
5. Rossi, P., Brühwiler, E., Chhuy, S., Jenq, Y.-S. & Shah, S. P., Fracture properties of concrete
as determined by means of wedge splitting tests and tapered double cantilever beam tests, in
S. Shah & A. Carpinteri, eds, `Fracture Mechanics Test Methods for Concrete', Chapmann &
Hall, 1991
6. Østergaard, L., Early-Age Fracture Mechanics and Cracking of Concrete, Technical
University of Denmark, Ph.D Thesis, 2003
 26. Fracture Mechanics Characterization of Wood 1211

THE ROLE OF FRACTURE TOUGHNESS IN THE CUTTING OF WOOD

T. Atkins
Department of Engineering, University of Reading,
Reading, RG6 6AY, England
a.g.atkins@rdg.ac.uk

When chips in wood cutting are formed by through thickness shear along a major shear plane
emanating from the tip of the tool to the free surface of the timber, the Ernst-Merchant theory (e.g.
Ernst and Merchant, 1944), originally propounded for continuous-chip cutting in ductile metals, is
often employed for analysis of wood cutting forces. The original rigid-plastic theory considers that
the forces in cutting are determined by two work components, viz: (i) work required permanently
to deform the chip and (ii) work against friction. Despite all the improvements to the basic Ernst-
Merchant model (secondary shear giving chip curl, the effect of work hardening, temperature and
rate effects, and so on) and improved modelling of friction, it remains a fact that such algebraic
analyses are not able to predict some well-established features of the mechanics of cutting. For
example, the theory predicts that the primary shear plane angle is independent of the metal being
cut whereas experiments show that the angles vary not only with friction but also with the metal
and mechanical properties. Again, when experimental cutting forces are plotted against the uncut
chip thickness (depth of cut), there is invariably a positive force intercept at zero uncut chip
thickness: the theory predicts that force vs uncut chip thickness plots should pass through the
origin of coordinates. The same experimental facts, not agreeing with the theory, are also found
when cutting other materials such as plastics and wood.
An explanation, for the failure of all algebraic theories of cutting that employ plasticity and
friction only in the modelling, was given by Atkins (2003). Shortcomings of the old theory have
been largely removed. The improved model has been assessed against experimental data for
various materials, and has also been employed to consider discontinuous chip formation and tear
(split) chip formation, Atkins (2004a,b).
The new model also explains why, in FEM simulations of cutting, the tool will not move
appreciably unless an additional criterion (a so-called separation criterion) is invoked at the tip of
the tool. (The same FEM codes are quite able to simulate other metal forming processes such as
forging, extrusion etc, without the additional criterion).
The key to the re-examination is the inclusion of significant work of surface separation in the
algebraic model as well as the usual components of plasticity and friction. Inclusion of fracture
toughnesses at the kJ/m2 level typical of ductile metals, contrasts with the analysis by Shaw and
co-workers (Cook et al, 1954) which used ‘surface free energy’ values of a few J/m2, from which
study it was concluded that the work of creating new surfaces in cutting would be insignificant,
which has been the received wisdom ever since. But inspection of FEM simulations revealed that
the specific energy associated with separation criteria were at the kJ/m2 levels typical of the
fracture toughnesses of ductile metals. The new theory has been able to resolve the question why,
if a separation criterion is essential to FEM simulations of metalcutting, is it not required in Ernst-
Merchant type algebraic models.
The present paper applies the new analysis to experiments on 30 mm diameter beech, oak and
Douglas fir cylinders. The samples were cut both along, and across, the axes of the cylinders in
order to investigate anisotropy. The tool employed had a small rake angle so that chips were
formed by shear so that the modified Ernst-merchant model was appropriate for analysis. (Note
1212 T. Atkins

that the model is inappropriate for those types of tools having large positive rake angles as
employed in the production of veneer. Veneer sheets are removed by a bending and splitting action
rather than by shear through the thickness to the free surface, Wyeth & Atkins, 2005). For both
beech and oak, fracture toughnesses of some 4 kJ/m2 were determined for cutting in the axial
direction and some 2 kJ/m2 in the radial direction. For Douglas fir, the corresponding toughnesses
were 7kJ/m2 and 0.2 kJ/m2. The relationship between these average values and the cyclically-
changing grain structure as the cylindrical testpieces are cut, and the appearance of the cut
surfaces, is discussed.

References
1. A G Atkins “Modelling Metal Cutting using Modern Ductile fracture mechanics: Quantitative
Explanations for some Longstanding Problems” Int J Mech Sci 2003 45 373-396
2. A G Atkins “Rosenhain and Sturney Revisited: the ‘Tear’ Chip in Cutting” Proc I MechE Part
C (J Mech Engr Sci) 2004a 1181-94
3. A G Atkins “Toughness and Cutting: a New Way of Simultaneously Determining Ductile
Fracture Toughness and Strength” Engr Fract Mech 2004b
4. N H Cook, I Finnie and M C Shaw Discontinuous Chip Formation Trans ASME 1954 76 153-
62
5. M E Merchant Basic mechanics of the metal Cutting Process J Appl Mech 1944 11 A168-75
6. V Piispannnen Eripaines Teknillisesla Aikakauslehdesla 1937 27 315-22 (see also J App Phys
1948 19 876-95)
7. D J Wyeth and A G Atkins “Cutting Forces for Wood Determined on an Instrumented Sledge
Microtome” Proc COST Action E35 University of Technology, Laapeentranta, Finland 2005
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1213

SHORT FATIGUE CRACKS OF IN-SERVICE FATIGUED TURBINE BLADES

A. A. Shanyavskiy, M. A. Artamonov, A. L. Tushentsov and Yu. A. Potapenko

State Centre for Safety Flight of Civil Aviation, Moscow, Russia
124340, Moscow, Airport Sheremetievo, P.O. Box 54, SCSFCA
shana@flysafety.msk.ru

It will be discussed features of the in-service short crack growth (SCG) through the foil base of the
turbine blades manufactured from the superalloy GS6K, Fig.1. There were several cases of in-
flight fatigue failures turbine blades at the airplane have flown (500-1500) hours. The blade
frequency during flight under the biaxial cyclic loads of bending-torsion is approx. 4 kHz. So, the
material in-flight fatigue failure took place in VHCF (very-high-cycle-fatigue) area –
(4000x3600x[500-1500]) = (0.7-2.0) x1010 cycles. There was during flight temperature neat to the
5000C around the blade volume where the fatigue fracture process was performed.
The paper reviewed cases studied of the blade fatigue failures and discussed the SCG rate in a
number of flights. The short and long crack growth rates dependences on the crack length were
joined, Fig.2. The stress equivalent for the crack resistance in VCHF area was also calculated on
the basis of the well-known Murakami’s equation.

FIGURE 1. Overview of the origin area of in-service fatigued turbine blade at the number of
2x1010 cycles.

FIGURE 2. Spacing of the meso-line, h, and number of cycles, Np, of the short (less than 1mm)
and long crack growth against the crack length, cr, in the right hand of the origin.
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1215

SHORT CRACK GROWTH UNDER CYCLIC TORSION WITH STATIC

TENSION

Isao Ohkawa1, Shun Hirano2, Toyokazu Negishi2 and Masaaki Misumi3

1Department of Mechanical Engineering, Faculty of Engineering, Hosei University

3-7-2 Kajino-cho, Koganei, Tokyo, Japan 184-8584

2Division of Engineering, Graduate School, Hosei University
3Department of Electrical and Mechanical Engineering,

Faculty of Science and Technology, Seikei University

ohkawa@k.hosei.ac.jp

The superposition of static loading on cyclic torsion has been recognized to affect the crack growth
behavior and fatigue life of materials, due to the mean stress acting on the crack planes. However,
the effect of static loading on torsional fatigue is different depending on the material type. This
appears to be related to the difference in microstructure and cyclic hardening/softening
characteristics of the materials.
This paper presents the results of an extensive experimental study on the effect of static tension
on short crack growth and fatigue life under torsional loading. Load- and stress-controlled tests
were performed at room temperature in an intermediate cycle regime on smooth tubular specimens
of four different materials, structural carbon steel S45C, carbon steel for pressure vessels SGV410,
austenitic stainless steel SUS316NG and aluminum alloy A6N01. In the stress-controlled test, load
and torque were corrected intermittently taking account of the influence of cyclic ratcheting. The
distribution of microcracks and development of dominant crack were observed in details.
Furthermore, stable cyclic shear stress-strain response and cyclic ratcheting strain were measured
under various tensile loading conditions.
Majority of microcracks initiated in the vicinity of the maximum shear stress amplitude planes
rather than the planes of the maximum shear stress, irrespective of the materials and stress
conditions.However, growth behavior of the dominant crack was different depending on the
microstructure and existence of the static tension. In S45C steel having banded ferrite and pearlite
microstructure along the specimen axes, the dominant crack in pure torsion grew accompanying
the coalescence of many shear microcracks in axial direction and branched along the principal
stress planes. Cycle ratio at the branching was 0.5 to 0.8. For application of static tension, the
crack tends to grow in circumferential direction due to mean tensile stress acting on the crack
plane. SGV410 steel also has banded ferrite and pearlite microstructure but distances between the
bands on the surface vary continuously along the circumferential direction.The shear crack in pure
torsion developed along the axial direction. However, with superposition of static tension, the
cracks around the region with wide spacing between the bands joined together and grew
preferentially in circumferential direction. The crack branching occurs in a later stage of the life
and thus majority of the life is spent for shear crack growth. In SUS316NG stainless steel,
dominant crack developed along axial or circumferential direction for pure torsion, while grew
only in the latter direction with application of tension. Except for lower stress levels, the shear
crack growth occupies more than 90% of the life. For A6N01 aluminum alloy, a few cracks grew
only in circumferential direction, irrespective of tensile stress levels. The cracks initiated later than
half life and branched close to final failure. For S45C, SGV410 and SUS316 steels, a marked
influence of the microstructure on the crack growth rate, alternating deceleration and acceleration,
was observed in the region up to about 300 m in length but no significant deceleration was
revealed in A6N01 alloy.
1216 I. Ohkawa et al.

Although a little effect of static tension on fatigue life was revealed in S45C, the lifetime of
SGV410 and A6N01 reduced evidently than in pure torsion. On the contrary, addition of static
tension increased the torsional fatigue life of SUS316. The difference in effect of static tension
appears to be related to deformation behavior of the materials. While cyclic stress-strain relation in
SGV410 and A6N01 remained unchanged with superposition of tension, plastic shear strain in
S45C and SUS316NG reduced due to an additional hardening. Although mean tensile stress acting
on the crack planes promotes the crack growth, the additional hardening can reduce the mean stress
effect.
Based on an empirical small crack growth law proposed by Hobson and Brown, a model which
can express the difference in effect of static tension on torsional fatigue of the materials was
investigated. Crack growth rate was assumed to vary as a power function of equivalent stresses
incorporating the effect of mean tensile stress on Mode and growth planes. Regarding decrease of
the plastic shear strain as a measure of hardening with superposition of tension, effect of the
additional hardening was taken into consideration. For load-controlled tests, effect of stress
increase induced by cyclic ratcheting was incorporated in the model. It has shown that the model
can explain the influence of static tension on the crack growth and the lifetime in torsional fatigue
of the present materials.
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1217

RESISTANCE-CURVE METHOD FOR PREDICTING FATIGUE THRESHOLDS

UNDER COMBINED LOADING

Keisuke Tanaka, Yoshiaki Akiniwa and Masami Wakita

Department of Mechanical Engineering, Nagoya University, Nagoya 464-8603, Japan
ktanaka@mech.nagoya-u.ac.jp

The fatigue threshold of materials with small defects or sharp notches is not controlled by the
nucleation of fatigue cracks, but by its propagation. After nucleation, the fatigue crack first
decelerates and then stops when the applied stress amplitude is below the fatigue threshold.
Tanaka et al have shown that the development of crack closure with crack extension is primarily
responsible for crack deceleration and stoppage. They have proposed the R-curve method for
predicting the fatigue thresholds of notched components and have shown a good agreement with
the experimental results for uniaxial loading cases. In this paper, this method is applied to the
fatigue thresholds under combined loading.
The effective stress intensity factor range, Keffth, at the crack stoppage takes a constant value
irrespective of the notch geometry or stress amplitudes. The threshold value of the maximum
stress intensity factor, Kmaxth, is defined by

K m a x th ' K e ffth  K op th
(1)
where Kopth is the stress intensity factor at the crack-tip opening at the threshold. The value of
Keffth is a material constant independent of the defect geometry. Once the change of Kmaxth with
crack extension is known, the fatigue limit and the nonpropagating crack length can be determined
on the basis of the R-curve method. The R-curve method is illustrated in Fig. 1, where the Kmax
value is taken as the ordinate. The R-curve is drawn with the solid line and the applied Kmax value
at a constant stress amplitude with the dashed lines. The fatigue thrreshold for crack initiation, w1,
is the stress amplitude corresponding to the applied Kmax value equal to the Keffth at the Stage I
crack length c1. The fatigue threshold for fracture, w2, is the stress amplitude where the applied
Kmax curve is the tangent of the R-curve. At stress amplitudes between w1 and w2, the length of
nonpropagating cracks is determined as the intersection of two curves.
The effect of the stress multi-axiality on the fatigue threshold may be evaluated simply by
changing the applied Kmax curve, because nonpropagating cracks are mode I crack in Stage II even
under combined loading. Thin-walled tubular specimens made of a medium-carbon steels (JIS
S45C) with a hole of 0.2 mm diameter were fatigued under cyclic torsion (case A), cyclic torsion
superposed on in-phase cyclic tension compression (case B), and cyclic tension compression (case
D). The stress ratio was R = -1 for all cases. The stress intensity factor for a crack from hole in the
biaxial stress filed was calculated by the boundary element method. The change of Kopth
( MPa m ) with crack length cnp (m) for S45C is expressed by

K o p th 9 8 .8 c n p  c1 fo r c1 d c n p d c 2
(2)

K o p th K opthf fo r c 2 d c np (3)

where c2 is the crack length at K op th K o pth f , c1 is given by

1218 K. Tanaka et al.

2
c1  ' K e ff th 1 .1 2 2 V w0  S (4)

and K m ax th f 2 .9 4 M P a m , ' K effth 2 .9 4 M P a m and V w0 22 3M P a .

Figure 2 shows the prediction of the fatigue thresholds under combined loading of cyclic
torsion and tension-compression, where the dotted and solid lines indicate the fatigue thresholds
for crack initiation and for fracture, respectively. The experimental data obtained for cases A, C
and D agree fairly well with the prediction. The length of the nonpropagating cracks predicted also
agree with the value observed experimentally.

Fig. 1. Resistance-curve method. Fig. 2. Fatigue threshold under combined

loading.
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1219

THE GROWTH OF SHORT CRACKS FROM DEFECTS UNDER MULTI-AXIAL

LOADING

M. Endo and A. J. McEvily

Dept. of Mech. Engng., Fukuoka Univ., Fukuoka 814-0180, Japan, endo@fukuoka-u.ac.jp
Dept. Mats. Sci. and Engng. and Inst. Mats. Sci, Univ. of Connecticut, Storrs, CT, USA
mcevily@mail.ims.uconn.edu

Multi-axially loaded components may contain defects which under cyclic loading can facilitate the
initiation of fatigue cracks. This paper describes a means for predicting the threshold levels and the
fatigue lifetimes of such components.
The basic equation used in the analysis is (McEvily et al. [1]):

da
A ' K eff  ' K effth 2 (1)
dN
where a is the fatigue crack length, N is the number of cycles, A is a material constant, 'Keff is the
effective range of the stress intensity factor, and 'Keffth is the effective range at the threshold
level. In order to make use of this relationship in the short crack range, three modifications are
needed. These are: (1) a modification to take into account the elastic-plastic behavior of short
cracks, (2) consideration of the Kitagawa effect, and (3) a modification for crack closure
development from zero up to the macroscopic level as a newly developed crack extends. With
these modifications the stress intensity factor for a Mode I crack is expressed as
da
dN
A > 2 S re F  Y 
S aF ' V  (1  e  O k )( K op max  K min )  ' K effth @
2
(2)

where re is a material constant of the order of 1 Pm, F is an elastic-plastic correction term equal to
1 SV max
F (1  sec ) , Vmax is the maximum stress in a loading cycle, VY is the yield stress, Y
2 2V Y
is a shape factor, equal to 0.73 for a semi-circular surface crack, k is a material constant, O is the
length of a newly formed crack, Kopmax is the crack opening level for a macroscopic crack, and
Kmin is the minimum stress intensity factor in a loading cycle. Equation 2 is modified for the cases
of torsion and equi-biaxial loading. It is assumed that under biaxial loading fatigue cracks grow
from defects as Mode I cracks. On this basis it is possible to predict the fatigue strength of part
containing surface defects for various multi-axial loading conditions.
Fatigue tests were carried out under in-phase combined tension and torsion loading at R of -1.
Materials tested were an annealed 0.37 % carbon steel with VY of 328 MPa and a heat-treated Cr-
Mo low alloy steel with VY of 851 MPa. To simulate a small defect, an artificial small defect was
introduced into the surface of bar specimens. Geometries of specimen and defects are shown in
Fig. 1. Both the major axes of in-line holes and the crack faces were perpendicular to the maximum
principal stress. A servo-hydraulic combined tension and torsion testing machine was used at
operating speed of 30-50 Hz.
A relationship between the initial defect size, a0, and number of cycles to failure, Nf, as a
function of stress amplitude was obtained by integrating Eq. 2 between the limits a0 and the final
crack length, af, with the value of af taken to be 5 mm. Figure 2 shows the S-N curves, in which a
comparison of calculated results (solid lines) and experimental data is made. The initial size of
1220 M. Endo and A. J. McEvily

holes was converted into the depth, a0, of an equivalent semi-circular crack by use of the relation:
a0 = 2 / S area (Murakami [2]). The principal stress ratio, V3/V1, is 0 for tension, -1 for
torsion and -0.382 for combined tension and torsion with the same stress amplitude. It is seen in
Fig. 2 that the finite life and the fatigue limit stress are well predicted. It is noted that at the same
value of a0, the influence of load types is greater for annealed steel with a low yield stress than for
the heat-treated steel with a high yield stress. In this paper, crack growth curves and design charts
of fatigue strength are given. Experimental data obtained using cruciform specimens, which
include crack growth data for biaxial tension, are also investigated.

FIGURE 1. Shapes and dimensions of specimen and defects.

FIGURE 2. S-N curves.

References
1. McEvily, A. J., Eifler, D. and Macherauch, E., Engng Fract Mech, vol. 40, 571-584, 1991.
2. Murakami, Y., Metal Fatigue: Effects of Small Defects and Nonmetallic Inclusions, Elsevier,
Oxford, 2002.
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1221

SHORT FATIGUE CRACKS IN NOTCHED AND UNNOTCHED SPECIMENS

UNDER NON-PROPORTIONAL LOADING

Olaf Hertel1, Timm Seeger1, Michael Vormwald1, Ralph Doring2 and Jens Hoffmeyer3
1Technische Universität Darmstadt, Fachgebiet Werkstoffmechanik

Petersenstraße 12, D-64287 Darmstadt, Germany

2IAMT mbH, Morgenbergstr. 19, D-08525 Plauen, Germany
3Volkswagen AG, D-38436 Wolfsburg, Germany

hertel@wm.tu-darmstadt.de, seeger@wm.tu-darmstadt.de, vormwald@wm.tu-darmstadt.de,

ralph.doering@iamt.de, jens.hoffmeyer@volkswagen.de

A dominant part of fatigue life is spent in short crack growth. Modelling this process using fracture
mechanics based methods has led to both increased insight into fatigue and improved prediction
accuracy. A short crack growth model for the prediction of constant amplitude fatigue life to
technical crack initiation under multiaxial nonproportional loading has been proposed by Döring
et. al. [1]. Here, results of further work are presented which are intended to improve and extend the
mentioned short crack model.
The model is based on the integration of a Paris type crack growth equation expressed in terms
of the effective range of the cyclic J-integral 'J eff . The basis for using this approach is the
information on the complete local stress and strain histories at the critical location. These histories
are calculated applying a plasticity model also developed by Döring et. al. [2]. This plasticity
model allows for a realistic description of cyclic hardening and softening, Ratchetting, mean stress
relaxation and also nonproportional hardening.
A very important aspect is crack closure, which is taken into account. Sound estimates on the
crack opening behaviour are fundamental for predicting realistic crack growth rates and fatigue
lives. According to [1] an extended Newman formula [3] is used, applicable for multiaxial
nonproportional loading. Additionally to Mode I crack closure, Mode II crack closure is taken into
account to obtain effective shear stress and shear strain ranges considering roughness of crack
surfaces in combination with normal stress dependent shielding effects.
Modelling of short crack growth follows the critical plane concept, where the angle of the
critical plane is determined by the criterion of maximum crack growth rate and minimum fatigue
life, respectively.
In [1] the influence of microstructure and stage I crack growth is mirrored only in a starter
crack length a0, which is determined by backward integration of the crack growth law (based on
'J eff ) from uniaxial experimental fatigue life data. The starter crack length a0 is substituted by an
intrinsic crack length ã0 of the same dimension (26µm steel S460N, 36µm Al5083). Introducing an
intrinsic crack length in spite of a starter crack length, the damage variable (e.g. crack length) now
starts at zero. Therefore, measured crack growth curves are described more precisely, except for
the rare cases of broken inclusions, where real starter cracks can be found. Crack coalescence as
well as environmental influences on crack growth are not considered within the framework of the
short crack model.
The short crack model [1] is based on the assumption of a semi-circular surface crack growing
with constant shape in a homogenous stress field. Within the current work the model is extended to
1222 O. Hertel et al.

semi-elliptical surface crack with variable shape in linearly varying stress fields. Therefore, the
size effect of notches according to stress gradients can be taken into account.
A satisfying accuracy of calculated fatigue lives is achieved. This is verified by experimental
data (Hoffmeyer et. al. [4]) for notched shafts and unnotched hollow tube specimen of two
different materials (S460N, Al5083) under a variety of multiaxial nonproportional loading
sequences. The correlation between the perception as stated in the model and real damage
evolution is shown comparing calculated and experimentally determined short crack growth curves
and initiation angels, measured with a replica technique (Fig. 1).

n=17011 n=21507 n=26007

FIGURE 1. REM pictures of a short crack taken by replica technique (Al5083, 90° out-of-phase
loading, Ja=0.4%, Ha=0.231%).

References
1. Doring, R., Hoffmeyer, J., Seeger, T. and Vormwald, M., In Proceedings of the Seventh
International Conference on Biaxial/Multiaxial Fatigue and Fracture, DVM, Berlin, 2004,
253-258.
2. Doring, R., Hoffmeyer, J., Vormwald, M. and Seeger, T., Comp. Mat. Sci., vol. 28, 587-596,
2004.
3. Newman, J.C., Int. J. Fract., vol. 24, R131-R135, 1984.
4. Hoffmeyer, J., Döring, R., Seeger, T. and Vormwald, M., In Proceedings of the Seventh
International Conference on Biaxial/Multiaxial Fatigue and Fracture, DVM, Berlin, 2004,
223-228.
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1223

MICROCRACKS GROWTH IN PUSH-PULL AND REVERSED TORSION IN

STAINLESS STEEL.

V. Doquet and G. Bertolino

Laboratory of Solids Mechanics. CNRS.
Ecole Polytechnique. 91128 Palaiseau cedex. France
doquet@lms.polytechnique.fr

In many metals, like stainless steel, the arrest of Stage I cracks on grain boundaries (G.Bs.)
determines the endurance limit. The mechanisms of crack growth beyond a grain boundary in a
globally elastic polycrystal have thus to be understood and properly modelled.
In most models, the resistance of a GB to crack growth is considered overcome as soon as
plasticity has been activated in the next grain, which leads to a critical resolved shear stress
criterion on a potential dislocation source ahead of the blocked slip band. Either this condition is
fulfilled and the crack is considered to grow immediately into the next grain or it is not and the
crack is considered arrested once and for all. However, microcracks are often observed to stay
arrested for a large number of cycles at GBs and then resume propagation (see below). This
suggests that crack growth beyond a GB requires an incubation period after the activation of a slip
system, during which a microcrack could initiate in the next grain, due to local cyclic plasticity and
link with the arrested one, thus allowing further propagation.
Such an incubation period was modelled by Morris et al [1] and the present authors [2]. In both
cases, it was evaluated from the resolved shear stress on a potential slip system in the next grain.
However, the comparison of arrest periods at GBs in push-pull and reversed torsion reported below
suggest that it should rather depend on a combination of shear and opening stress on this slip
system.
Stress-controlled push-pull and reversed torsion tests were performed on 316LN steel
specimens in the high-cycle range (2.105 to 2.106 cycles). The tests were periodically interrupted
for preparation of replicas, which allowed tracing back of microcracks development (Fig.1).
Microcracks initiated early (in some cases, the main crack was already present at 15% of the
fatigue life) and propagated along slip bands over 1 to 5 grains, depending on the stress range and
loading mode: longer Stage I cracks were observed for smaller stress ranges and push-pull led to
earlier transition to stage II than reversed torsion for equivalent stress ranges.
Stage I transgranular crack growth rates were less than 10 Burgers vectors/cycle and showed
no clear correlation with the stress range or loading mode. Microcracks were frequently arrested
for tens to hundreds thousands cycles at grain boundaries and then resumed propagation. The
smaller the applied stress range, the longer the arrest periods (Fig. 2). The scatter in arrest periods
increased as the stress range decreased. In reversed torsion under r260MPa Tresca equivalent
stress range, arrest periods up to 105 cycles were observed at some G.Bs, while for a smaller
equivalent stress range (r215MPa), arrest periods do not exceed 4.104 cycles, in push-pull. Further
tests in torsion are in progress to document this point.
1224 V. Doquet and G. Bertolino

FIGURE 1. Crack growth kinetics: a-c) reversed torsion, r130MPa d-e) push-pull, r215MPa

FIGURE 2 Fatigue life and arrest periods as a function of VeqTresca.

References
1. Morris W.L, James M.R. and Buck O., Metall Trans., vol.12A:57-64, 1981.
2. Bertolino G., Doquet V. and Sauzay M., Int Journ Fatigue, vol 27/5, 471-480, 2005.
 28. Short Fatigue Crack Growth under Multi-axial Loading Conditions 1225

HYDROGEN AND NOTCH EFFECTS ON TORSIONAL FATIGUE OF

STAINLESS STEEL

Y. Kondo, M. Kubota and K. Ohguma

Dept. of Intelligent Machinery and Systems, Kyushu University
744, Moto-oka, Nishi-ku, Fukuoka, 819-0395, Japan
ykondo@mech.kyushu-u.ac.jp
1. Introduction
It has been pointed out that absorbed hydrogen in metal has detrimental effect in the case of
fatigue as well as in hydrogen embrittlement. Austenitic stainless steel is regarded as one of the
most promising materials for hydrogen utilization machines. The effects of hydrogen and small
notch on the high cycle torsional fatigue of SUS304, SUS316 and SUS316L were studied.
2. Effect of hydrogen on austenitic stainless steels with different hardness
The tensile strength was varied by cold work, hot drawing and solution treatment. Torsional
fatigue test was done using buff-finished bar specimen at a stress ratio of R=-1. The absorption of
hydrogen was done continuously during fatigue test by cathodic charge in dilute sulfuric acid
(pH=2.0). S-N curves of SUS316 are shown in Fig.1 as an example. The fatigue strength of
hardened materials substantially decreased in cathodic charge condition (Fig.1(a),(b)). In the case
of solution heat treated material, however, no significant reduction was observed (Fig.1(c)). This
tendency was observed irrespective of chemical composition (Fig.1(d)).

(a)SUS316-WPA cold worked material (b) SUS316 drawn material

(c) SUS316 solution treated material (d)Reduction of fatigue limit under

cathodic charge
FIGURE 1. Effect of tensile strength on the fatigue strength under cathodic charge

A small shear crack was observed at the crack origin (Fig.2). The crack nucleated in shear
mode and formed a triangular facet on the fracture surface. It changed into a crack perpendicular to
the principal stress and continued to propagate to the final fracture as a mode I crack.
1226 Y. Kondo et al.

FIGURE 2. Crack origin of buff-finished specimen under cathodic charge (SUS316-WPA cold
worked, a = 178MPa, Nf =7.24×106)
3. Effect of small notch introduced by pre-corrosion
In order to simulate small surface notches, a buff-finished specimen was pre-corroded by
anodic polarization. Grain boundaries, slips and twins were etched and very small defects were
introduced. The pre-corroded specimen was torsional fatigue tested under cathodic charge
condition. The test result is shown by triangular symbols in Fig.3. The pre-corrosion gave a
reduction of fatigue strength. The crack origin was quite different from those of buff-finished
specimen. No distinct shear mode crack was found at the crack origin in the case of pre-corroded
specimen. Corrosion traces of slip and grain boundaries were found near the crack origin. These
corrosion trace and cathodic charge enabled the crack initiation without the formation of shear
mode crack and resulted in lower fatigue strength.

FIGURE 3. Torsional fatigue of pre-corroded specimen in cathodic charge (SUS316 cold

worked)

4. Conclusion
1 The absorbed hydrogen gave a detrimental effect even on the torsional high cycle fatigue
of work hardened austenitic stainless steels irrespective of chemical compositions. On the
contrary to this, no significant effect was seen in the case of solution heat treated material.
2 High hardness caused by work hardening and small notch were detrimental factors for the
torsional fatigue of austenitic stainless steel used in hydrogen environment. Care should be
taken in the usage for the hydrogen utilization equipment.
 30. Integrity of gears 1227

INFLUENCE OF MOVING TOOTH LOAD ON GEAR FATIGUE BEHAVIOUR

D. T. Jelaska and S. Podrug

University of Split, FESB, R. Boskovica b.b., HR 21000 Split, Croatia
Tel: + 385 21 305874, Fax: + 385 21 463877
djelaska@marjan.fesb.hr

A computational model for determination of service life of gears in regard to bending fatigue in a
gear tooth root is presented. Two cases are being explored, first in which gear tooth was loaded
with normal pulsating force acting at the highest point of single tooth contact, and second in which
the fact that in actual gear operation the magnitude as well as the position of the force changes as
the gear rotates through the mesh is taken into account. A quasi static numerical simulation method
is presented in which the gear tooth engagement is broken down into multiple load steps and
analyzed separately, Lewicki et al. [1], Spievak et al. [2].
The fatigue process leading to tooth breakage is divided into crack initiation and crack
propagation period. The critical plane damage model, Socie and Bannantine [3], has been used to
determine the number of stress cycles required for the fatigue crack initiation. Critical plane
approaches are based upon the physical observation that fatigue cracks initiate and grow on certain
material planes, called critical planes, the orientation of which is determined by both stresses and
strains at the critical location. Depending upon strain amplitude, material type and state of stress,
materials generally form one of two types of cracks – shear cracks or tensile cracks. Consequently,
the critical plane methods predict not only fatigue crack initiation life, but also the initiated crack
direction, which makes a good starting point for further fatigue crack propagation studies.
Finite element method and linear elastic fracture mechanics theories are then used for the
further simulation of the fatigue crack growth under a moving load. Moving load produces a non-
proportional load history in a gear's tooth root (the ratio of KII and KI changes during the load
cycle). Consequently, the maximum tangential stress theory will predict a unique kink angle for
each load increment, but herein is refined procedure given in [2] which compute crack’s trajectory
at the end of the load cycle.
An approach that accounts for fatigue crack closure effects is developed to propagate crack
under non-proportional load. Effective range of FIN is calculated using Budiansky and Hutchinson
model, Budiansky and Hutchinson [4], Newman et al. [5], for plasticity induced crack closure, and
influence of oxide induced crack closure, and roughness induced crack closure is taken into
account by the concept of partial crack closure, Kujawski [6].
The total number of stress cycles for the final failure to occur is then a sum of stress cycles
required for the fatigue crack initiation and number of loading cycles for a crack propagation from
the initial to the critical length. The computational results are compared with other researchers’
numerical results and with service lives of real gears. The fatigue lives (Fig. 2) and crack paths
(Fig. 1) determined in this paper exhibits a substantial agreement with experimental results and
significant improvement compared with the existing numerical models.
1228 D. T. Jelaska and S. Podrug

FIGURE 1. Comparison of crack paths: A – for a load in HPSTC, B – for a moving force.

FIGURE 2. Loading cycles for a crack propagation from the initial to the critical length.

References
1. Lewicki, D.G., Spievak, L.E., Handschuh, R.F., Consideration of Moving Tooth Load in Gear
Crack Propagation Predictions, NASA/TM-2000-210227, 2000.
2. Spievak, L.E., Wawrzynek, P.A., Ingraffea, A.R., Lewicki, D.G., Simulating Fatigue Crack
Growth in Spiral Bevel Gears, Engineering Fracture Mechanics, 68, 53-76, 2001.
3. Socie, D., Bannantine, J., Bulk Deformation Damage Models, Materials Science and
Engineering, A103, 3-13, 1988.
4. Budiansky, B., Hutchinson, J.W., Analysis of Closure in Fatigue Crack Growth, Journal of
Applied Mechanics, 45, 267-276, 1978.
5. Newman, J.A., Riddell, W.T., Piascik, R.S., A Threshold Fatigue Crack Closure Model: Part I
– Model Development, Fatigue Fract Engng Mater Struct, 26, 603-614, 2003.
6. Kujawski, D., Enhanced Model of Partial Crack Closure for Correlation of R – Ratio Effects
in Aluminum Alloys, International Journal of Fatigue, 23, 95-102, 2001.
 30. Integrity of gears 1229

COMPARISON OF SOLID SPUR GEAR FACE LOAD FACTORS

G. Marunic
University of Rijeka Faculty of Engineering
Vukovarska 58, 51000 Rijeka, Croatia
gmarunic@riteh.hr

The research into stress state of standard spur gear has enabled the comparison of analytically and
numerically established effect of a load distribution over the facewidth, upon the stresses at the
tooth-root.
Numerous investigations of spur gear based upon 3D approach accomplished by Curti et al.
[1], Flašker et al. [2], Glodež and Ren [3] and Baret et al. [4], have underlined the difference
between the actual state of stress at tooth-root and the stress results obtained by analytical
procedure proposed by the standard ISO [5]. The comparison of numerical results accomplished by
use of FEM simulation, with the corresponding ones resulting from the procedures according to the
standards, points to some advantages of numerical over standard methods.
The stress field at the tooth-root was determined by means of 3D FEM and the developed
single gear model. The different linear shape load distributions, from constant load to the triangular
shape (Fig. 1), were applied at the outer point of single pair tooth contact, as critical engagement
position. The calculations were performed for the gear of the chosen geometrical parameters,
covering the range of facewidth values. These facewidths were subjected to equal specific load.
The ratio U = 'w/w0 [4] was established, relating the difference between maximum and minimum
value of the applied specific load 'w, and the mean value of the applied load w0.
The procedure proposed by the standard ISO 6336, introduces the face load factor for tooth-
root stress calculation that takes into account uneven distribution of load over the facewidth due
the mesh-misalignment caused by inaccuracies in manufacture, elastic deformations, etc.
Analytically determined face load factors are directly compared to maximum tooth-root stress
increment caused by non-uniform load distribution. The local load increase related to its average
value in the case of uniform load distribution, i.e. the ratio V1max/Vm0, is taken into account.

FIGURE 1. Linear shape load distribution.

Analytical and the corresponding numerical values of the face load factors are presented on
Fig.2 for the facewidth value b/mn =8 (mn – module) and the chosen load distribution cases
expressed by the ratio U.
1230 G. Marunic

FIGURE 2. Comparison of analytical and corresponding numerical results of face load factors.

The obtained results proved the actual non-uniform stress distribution along the facewidth for
the case of uniform load distribution. Maximum stress always appears approximately at the middle
of the facewidth, and the stress decrement at the tooth edges strongly depends on the tooth actual
facewidth, as well as the achieved agreement of analytically and numerically determined face load
factors for different linear shape load distribution.
In the case of the adopted uniform load distribution, non-uniform tooth-root stress distribution
is taken into account for numerical data that are compared with analytical values of the face load
factors provided by the standard ISO, and this fact has initially determined the magnitude of the
differences. For the cases of non-uniform load distribution, numerical values of the face load
factors are in good agreement with the analytical ones for the facewidth b/mn =6, and U >1.

References
1. Curti, G., Raffa, F.A., Garavelli, D., Baret, C., In Proceedings of the JSME International
Conference on Motion and Powertransmissions, JSME, Hiroshima, 1991, 787-794.
2. Flasker, J., Glodez, S., Pehan, S., J. of Mech. Engineering, vol.39, 299-308, 1993
3. Glodez, S., Ren, Z., In Proceedings of International Design Conference-Design ´98, edited
by D. Marjanovi, FSB, WDK, CTT, Dubrovnik, 1998, 133-138.
4. Baret, C., Coccolo, G., Raffa, F.A., In Proceedings of International Gearing Conference,
edited by J.N. Fawcett, UK National Gear Metrology, Univ. of Newcastle Upon Tyne, British
Gear Assn., Newcastle Upon Tyne, 1994, 149-154.
5. ISO 6336-1, Calculation of load capacity of spur and helical gears, Part 1: Basic principles,
introduction and general influence factors, 1996.
 30. Integrity of gears 1231

PREDICTION OF CONTACT FATIGUE INTERNAL CRACK PROPAGATION

IN HYPOID GEARS

M. Vimercati1, M. Guagliano1, L. Vergani1 and A. Piazza2

1Politecnico di Milano, Mechanical Engineering Dept. - Via La Masa 34, 20158 Milan (ITALY)
2Centro Ricerche FIAT ScpA - Strada Torino 50, 10043 Orbassano (TO) (ITALY)

martino.vimercati@polimi.it

Rolling contact fatigue is becoming more and more important in the design of many mechanical
systems like gears and bearing. This is due to the ever more severe load conditions that
components in contact must undergo if lightness is a primary goal of the design process. This
objective can be achieved only by the application of a “design by analysis” procedure. This means
that the definition of a more refined design approach requires the ability to develop very accurate
simulation procedures to reproduce accurately the actual working condition of the element in
contact.
One of the damage type occurring in RCF of gears is called “spalling” and consists in the
removal of the surface layer of material as a consequence of the stable propagation of a crack
initiated at some distance from the surface, most of time in correspondence of an internal defect
sited near the position of maximum shear stress, according to the Hertzian stress distribution.
Since this type of damage can be tolerated till the material is macroscopically removed, from
the designer point of view it is of great importance to predict the number of cycles necessary for
this to happen. In particular, since crack propagation begins after a very small fraction of the total
life of the components, the prediction of the crack growth rate is a fundamental importance for an
accurate sizing of contact elements.
This result cannot be obtained without an accurate knowledge of the contact pressure
distribution and without knowing the values of the stress intensity factors concerning Modes I, II
and III of propagation.
With this aim it is very useful to develop a numerical approach able to predict both the contact
pattern and the stress intensity factor values during the whole meshing cycle. However this
objective is not easy to achieve when the geometry of the gear tooth is complex and the deflection
of the tooth shares the load over more than one tooth.
In this paper an approach for calculating the SIF of internal cracks in hypoid gears,
characterized by high spiral angle values and complex meshing conditions, is discussed: it derives
from a previously developed approach aimed to study cracked railway wheels [1] and it is based on
two different steps, the one concerning the uncraked tooth and the second one regarding the
cracked part of the tooth.
Firstly, starting from an accurate geometry description of the gear tooth [2], a contact-FEM
analysis carried out by means of an advanced contact solver [3] allowed to obtain the contact
pressure distribution and the stress state over the entire meshing cycle [4]. In Fig. 1 it is shown the
studied hypoid gear transmission; Fig. 2 reports the pinion teeth with the contact pressure
distribution calculated in an instant of meshing.
1232 M. Vimercati et al.

FIGURE 1. Finite Element model of the FIGURE 2. Contact pressure distribution

studied hypoid gear transmission. over studied pinion teeth.

Once the stress and displacements state in the uncracked tooth are known, they are applied as
boundary conditions to a second finite element model of the cracked zone, being the aim the stress
intensity factor calculation for the mode I, II, III along the crack front. It was verified that the
displacement field induced by the presence of the crack does not affect the boundary of the model.
The model includes the contact between the crack faces, with or without friction. By using the
proposed approach it is possible to easily considerate different crack dimensions, positions and
applied loads, without constructing different and expensive FE models.
In the present paper the results obtained by considering circular and elliptical cracks are
shown. By opportunely making dimensionless the results it was possible to give them a more
general interpretation.
The results, together with the knowledge of the threshold value of range of 'K of the material,
allows to evaluate the conditions of propagation of internal defects.

References
1. Guagliano M. and Vergani L., “Experimental and numerical analysis of sub-surface cracks in
railway wheels”, Eng. Fr. Mech., vol. 72, 255-269, 2005.
2. Vimercati M., “Mathematical Model for Tooth Surfaces Representation of Face-Hobbed
Hypoid Gears”, submitted to Comp. Meth. App. Mech. Eng..
3. Vijayakar S.M., “A Combined Surface Integral and Finite Element Solution for a Three-
Dimensional Contact Problem”, Int. J. Numer. Methods Eng. vol 31, 525–545, 1991.
4. Vimercati M. and Piazza A., “Computerized Design of Face Hobbed Hypoid Gears: Tooth
Surfaces Generation, Contact Analysis and Stress Calculation”, submitted to AGMA Fall
Technical Meeting, Detroit, Michigan, USA, 2005.
 30. Integrity of gears 1233

FATIGUE CRACK INITIATION ALONG INCLUSION INTERFACES OF

CONTACTING MECHANICAL ELEMENTS

S. Glodez, M. Ulbin and J. Flasker

University of Maribor, Faculty of Mechanical Engineering
Smetanova 17, 2000 Maribor, Slovenia
srecko.glodez@uni-mb.si, ulbin@uni-mb.si, joze.flasker@uni-mb.si

This paper utilises the theory of fatigue crack initiation along inclusion interfaces of contacting
mechanical elements. The assumption is warranted by the fact that hard inclusions in form of
carbides are often present in high strength steels and have a significant impact on local strength
reduction. If the inclusion and the interface are strong enough to withstand breaking during the
initial loading, the plastic deformation is accumulated in the grain matrix after some cyclic loading.
The motion of dislocations in the matrix is blocked by inclusion. Fig. 1 illustrates the theoretical
model with inclusion of radius R residing within the slip band zone (Tanaka and Mura [1]). The
slip band zone is elliptic with semi-major axis l1 and semi-minor axis l2. The inclusion is assumed
to be much smaller than the slip band zone.

FIGURE 1. Inclusion in the zone of multiple slip bands

If the plastic zone spreads over the whole circular grain, i.e., l1=l2, the number of stress cycles
Ni required for the fatigue crack initiation can be determent with following equation [1]:

Gv  G G ˜J 2 f
Ni ˜ ˜
R ' W  2W 
2
Gv f
f (1)
where G is the shear modulus of the material matrix, Gv is the shear modulus of the inclusion, J is
the fracture energy per unit interfacial area, 'W is the grain stress range, Wf is the frictional stress
and f is the irreversibility factor of dislocations. Experimentally determined values of
irreversibility factor f are small and range from 10-4 for small to 10-1 for large plastic strain
amplitudes (Zhou et al [2]). The grain stress loading range 'W can be represented either with the
shearing stresses or with the equivalent Mises stress (Mura and Nakasone [3]).
The presented model is used for the numerical determination of crack initiation period Ni in a
contact area of a spur gear pair made of flame hardened steel AISI 4130. The maximum equivalent
von Mises stress (Veq)max=800 N/mm2 and its position (depth H=0.126 mm under the contact
surface) in the contact area of gear flanks have been determined numerically using FEM. For the
case of fatigue crack initiation along inclusion interfaces the number of stress cycles Ni required
for the crack initiation can then be determined with eq. (1), where the computed maximum
1234 S. Glodez et al.

equivalent stress (Veq)max=800 N/mm2 is used as the applied stress amplitude 'W to the grain. The
frictional stress Wf is material resistance to the motion of dislocations and is for metals equal to
Wf|25 N/mm2 [3]. The dislocation pileup length 2l can be equalled to the grain diameter (2l=D),
where the average value of the grain diameter D|50 Pm has been determined previously (Glodež
[4]). For the treated material with relatively small plastic strain amplitudes, the irreversibility
factor is taken to be in the range f=10-4 to 10-3 and the crack surface energy to be in the range J=1
to 10 N/m [2, 3]. Using this data set the number of stress cycles Ni required for the fatigue crack
initiation along inclusion interfaces is computed according to eq. (1) for various combinations of
the shear modulus of the inclusion Gv and inclusion radius R. The results are presented in Fig. 2.
From these computations it can be concluded, that the number of stress cycles Ni required for the
fatigue crack initiation decreases with increase of the inclusion radius R and its shear modulus Gv.
Taking into account that the usual values for 2R/D lie in the interval from 0.01 to 0.1, the computed
number of stress cycles Ni required for the fatigue crack initiation along inclusion interfaces is
between 1.372˜102 and 2.286˜105 load cycles.

FIGURE 2. Stress cycles Ni for the fatigue crack initiation

References
1. Tanaka, K. and Mura, T., Metallurgical Transactions, vol. 13A, 117-123, 1982.
2. Zhou, R.S., Cheng H.S. and Mura, T., ASME Journal of Tribology, vol. 111, 605-613,1989.
3. Mura, T. and Nakasone, Y., ASME Journal of Applied Mechanics, vol. 57, 1-6, 1990.
4. Glodez, S., PhD thesis, Faculty of Mechanical Engineering, University of Maribor,1995.
 30. Integrity of gears 1235

ENERGY BASED GEAR FAULT DIAGNOSTICS

S. J. Loutridis
Technological Education Institute of Larissa
Department of Electrical Engineering
GR 41-110 Larissa, Greece
loutridi@teilar.gr

Gear mechanisms are widely used in rotating machinery. For this reason, gear health monitoring
has been the subject of intensive investigation and research. Among several other methods,
vibration measurement and analysis is considered as the most general basis for fault detection.
Practical condition monitoring systems need quantities that can be used as features for the
diagnostic procedure. In this work energy-based features are proposed. The instantaneous energy
density calculated using advanced signal processing techniques, is shown to obtain high values
when defected teeth are engaged and consequently can be directly related to damage magnitude.
The experimental rig consists of two electrical machines, a pair of spur gears, a power supply
unit with the necessary speed control electronics and the data acquisition system. A DC machine of
1.5 kW rotates the pinion and the load is provided by an AC asynchronous machine, which is
configured as a brake. The transmission ratio is 35/19 = 1.842. The vibration signal generated by
the gearbox was picked up by an accelerometer bolted to the pinion body. and the electrical signal
was transferred to an external charge amplifier through slip rings and recorded by a PCMCIA
acquisition card with a sampling frequency of 20 kHz.
The Wigner-Ville distribution [1], is a very important quadratic form time-frequency
distribution defined as

f
W xx t , f  ³ x §¨ t  W ·¸ x §¨ t  W ·¸ e  j 2 S fW d W .
f© 2¹ © 2¹ (1)

FIGURE 1. Prediction based on Wigner distribution.

The instantaneous energy per unit time (power) is calculated by considering the absolute value.
f2

E t  ³ W t , f  df .
xx
f1
(2)
1236 S. J. Loutridis

In practice, integration is approximated by a summation over all frequency bins and the result
is divided by the number of bins for normalization. A prediction for a range of crack magnitude
from 15% up to 75% of tooth root is shown in Fig.1.
The empirical mode decomposition (EMD) pioneered by Huang et al [2], decomposes a time-
series into a finite set of oscillatory functions called the intrinsic mode functions (IMF). An IMF is
a function that satisfies two conditions: (1) the number of extrema and the number of zero
crossings must either equal or differ at most by one; (2) the running mean value of the envelope
defined by the local maxima and the envelope defined by the local minima is zero. The intrinsic
modes represent the embedded time scales in the signal. Instantaneous energy density can be
calculated directly from the IMFs by application of the Hilbert transform. A prediction is shown in
fig. 2.

FIGURE 2. Prediction based on EMD

References
1. Claasen, T.A.M. and Mecklenbrauker, W.F.G., Philips Journal of Research, vol. 35, 217-250,
1980.
2. Huang, N.E., Shen, Z., Long, S.R., Wu, M.C., Shih, H.H., Zheng, Q., Yen, N., Tung C.C. and
Liu H.H.,1998 Proc. Royal Society of London Series, vol. A454, 903-995, 1998.
 30. Integrity of gears 1237

CRACK PROPAGATION IN GEAR TOOTH ROOT

S. Pehan, B. Zafosnik and J. Kramberger

University of Maribor
Slovenia, 2000 Maribor, Smetanova 17
stanislav.pehan@uni-mb.si

The paper describes the problem of crack propagation. The object of consideration are the common
gearbox gears. From the measurements it is obvious that the crack can appear at the very different
locations in tooth root and then propagate in their own direction, Pehan [1]. Theoretically it can
happened in very narrow field on the tooth root surface where the principal stresses are the highest.
Few reasons are discussed that can influence on the start position of crack initiation: the loading
distribution over the tooth flank, the undercutting of the tooth and the tracks that remains from the
cutting tool. It is confirmed that all counted phenomena can cause wide scatter of crack
propagation in tooth root, Pehan [2].
The service life of the gearbox already exceed one and half millions kilometers what means
that all components should be designed very carefully in order not to become a critical component.
One of the most exposed elements in the gearbox are gears, because they are carried the high
specific loading all the service time. For that reasons the material and heat treatment also of gear
are chosen very carefully. Due to the good thermal treatment of the gear the tiny surface layers are
very hard and have high resistant to the wear but consequently such gears are sensitive to cracks in
the tooth root. And to understand the reasons for starting crack at the appointed location is the
problem that is discussed in presented paper.
Many things about the crack initiation and the crack propagation are known quite well. Theory,
which is mostly based on the linear fracture mechanic, was developed in previous century already.
For example, in theory the initial crack should appears on the surface where the tension or
principal stresses are the highest. Theoretically it is quite easy to find this location in tooth root.
But the problem still exist because the experimental results deviate from the theory quite a lot. The
experts explain this phenomenon by not understanding all the influences about the crack
propagation procedure. It is surely truth. So in order to improve the theoretical models and to make
it more efficiency a lot of scientist try to modify the theoretical equations by adding additional
members. In our case it will be shown that theoretical accessories maybe works correctly if the
new edge conditions are considered.
The loading spectra that is characteristic for delivery truck is calculated by on propeller shaft
measured torques.. However, the final failure of gearbox gears is not the wear but the fracture. A
few results of crack path measurement is presented, where the big scatter or dissemination of crack
path is obvious. The wideness of the scatter is grater than one millimeter, for what is to be
concerned. This wide scatter of fractures means namely that also the service life of such gear can
be shorter as it is expected, because of uncontrolled influences.
In the continuation the few possible reasons for resulting the wide scatter of fractures are listed.
From the gearbox design it is known that the loaded shafts are never parallel. Due to the radial
forces the shaft is deformed and the gear mounted on it is positioned in such way that the load over
the tooth flank is not distributed equal.
The position of the Mises pick stress in tooth root is not always at the exactly same position. It
is discussable where is actually the point of the highest stress. It seems to be considered also the
influence of the FEM mesh density. The gear model was meshed on different ways and it become
clear that mesh indeed have quite big influence on appearance of initial crack. Especially important
1238 S. Pehan et al.

becomes the FEM mesh in case of accounting different effects that are results of preliminary
treatments. The final treatment for vehicle gears are normally grinding, preliminary treatment are
usually milling, which left on tooth root great tracks.
According to the valid laws of linear elastic fracture mechanics the stress field at the tip of a
crack is fully described by the stress intensity factor. Due to the complex load distribution and
complex gear tooth geometry the stress intensity factor can be determined by numerical methods
only. The gear thickness allows plain strain conditions to be assumed, but the validity of linear
fracture mechanics should b permanently monitored by checking the size of the yield zone at the
crack tip.
In our case for each crack length the all three stress intensity factors: KI , KII and KIII and T-
stresses are calculated. In order to get confidence in obtained results and to get the better filling the
2-dimensional and 3-dimensional analysis were done. The biggest importance on crack
propagation have stress intensity factor KI. Other influences including T-stresses can be neglected.
The observation of T-stresses was primary introduced in order to better explain the crack
propagation in tooth root.
In case of the exactly known position of the initial crack the crack path and the distribution of
so called effective stress intensity factor along the crack can be determined easily. When the
theoretic value of Keff touches the critical material value called KIc (core material) then crack gets
the sound velocity and the tooth root is broken in the moment.
Due to the preliminary treatment that easily cause the surface defects on the tooth root the
initial crack can appear in many places. The path of crack propagation strongly depends on initial
crack. Presumably the direction of crack propagation is determined by the position of highest
virtual stress intensity factor in each crack tip. Consequently also the distribution of the stress
intensity factor along the length depend from the crack propagation path.
It was clearly shown the existence of few obvious influences that could have decisional effect
on appearance of initial crack. Before the designers would calculate the service life of the gears
they must carefully investigate the actually state of tooth root in order to be sure that the initial
crack is positioned on the proper place. The more research is in progress in order to evaluate the
effects of separate influences.

References
1. Pehan S, Helen T H, Flasker J, Applaying Numerical Methods for Determining the Service
Life of Gears, Fat. Frac, Engng., Mater. Struct., 18(9), pp.971-979, 1995
2. Pehan S, Helen T H, Flasker J, Glodez S., Numerical Methods for Determining Stress
Intensity Factors vs Crack Depth in Gear Tooth Roots, Int. J. Fatigue, 19(10), pp.677-685,
1997
 30. Integrity of gears 1239

EXPERIMENTAL EVALUATION OF STRESS INTENSITY FACTORS IN SPUR

GEAR TEETH

V. Spitas1, G. Papadopoulos2, Th. Costopoulos3 and C. Spitas3

1Laboratory of Applied Mechanics, Technical University of Crete

University Campus, 74100, Chania, GREECE

immg@otenet.gr
2Laboratory of Strength of Materials, National Technical University of Athens

Iroon Polytechniou 7, Zografou, GREECE

gpad@central.ntua.gr
3Laboratory of Machine Elements, National Technical University of Athens

Iroon Polytechniou 7, Zografou, GREECE

cost@central.ntua.gr

In most engineering applications, gears are the most widely used machine elements for the
transmission of power and motion from one shaft to another. Hence it is evident that there is a need
for reliability and longer service-life, which requires precise knowledge of the stress field
developed in the gear tooth. However this stress analysis becomes more complex when a small
crack appears after overloading or fatigue conditions. It is therefore critical to calculate either the
remaining life of the cracked gear under the same loading conditions, or the new maximum fail-
safe operating load to cover the initially calculated service time.
Any attempt to approach the problem of the estimation of the service life of gears must pass
through crack analysis. The majority of the catastrophic gear-tooth failures usually results from
overloading of a single tooth in which one or multiple fatigue cracks have already appeared and
propagated. As even a standard gear tooth undergoes highly variable loading in terms of both load
magnitude, direction and position on the tooth flank, it becomes evident that the analysis of such a
phenomenon is quite complicated and difficult.
Up to now a number of works have been carried out investigating gear tooth cracks using
numerical tools for the calculation of the Stress Intensity Factors (SIFs) (McAldener and Olsson
[1], Spievak et al. [2], Guagliano and Vergani [3], Pehan et al. [4], Sfakiotakis and Anifantis [5]).
Few of them used also experimental techniques (such as strain gauges or potential drop methods
Glodez et al. [6]) to verify the numerical predictions but with controversial and difficult to
interpret results. The classic experimental techniques for stress-analysis fail to focus at the
extremely small dimensions of the plastic zone at the crack tip with the high stress gradients and
also need excessive time to post-process the results particularly due to the tooth / crack geometry
dependent stress field. Photoelasticity in particular is unable to assess the stress condition at
singular points such as crack tips or contact points since the density of the isochromatic fringes at
the vicinity of the stress singularity is so high that it is impractical to measure the fringe order in
this region.
In this work the crack path is determined through static fracture experiments of several gear
teeth. Then the stress concentration and the stress intensity factors in modes I and II are determined
numerically using FEM for various crack lengths along the geometry dependent fracture curve.
The findings of the analysis are verified with photoelasticity regarding the stress field and with the
stress-optical method of caustics on specially machined cracked tooth specimens regarding the
SIFs.
1240 V. Spitas et al.

The results indicate that there is a good agreement between numerically predicted and
measured values. Moreover they indicate that the mode of crack propagation is mixed and not
uniform as gear tooth cracks tend to be almost pure mode I cracks near the surface (root fillet of the
tooth) whereas mode II becomes considerable as the cracks propagate towards the interior of the
tooth. Finally it should be noted that this is the first time that the stress optical method of caustics is
applied for measuring the SIFs in cracked gear teeth.

Reference
1. McAldener, M., Olsson, M., Eng. Fracture Mechanics, vol. 69, 2147-2162, 2002.
2. Spievak, L.E., Wawrzynek, P.A., Ingraffea, A.R. and Lewicki, D.G., Eng. Fracture
Mechanics, vol. 68, 53-76, 2001.
3. Guagliano, M., Vergani, L., Int. J. of Fatigue, vol. 23, pp. 65-73, 2001.
4. Pehan, S., Trevor, H.K., Flasker, J. and Glodez, S., Int. J. of Fatigue, vol. 19 (10), 677-685,
1997.
5. Sfakiotakis, V.G. and Anifantis, N.K., Fin. El. in Analysis and Design, vol. 39, 79-92, 2002.
6. Glodez, S., Pehan, S. and Flasker, J., Int. J. of Fatigue, vol. 20 (9), 669-675, 1998.
 35. High Temperature and Thermomechanical Fatigue 1241

ISOTHERMAL AND THERMOMECHANICAL FATIGUE BEHAVIOR OF THE

ODS SUPERALLOY PM1000

W. O. Ngala, G. Biallas and H. J. Maier

Lehrstuhl für Werkstoffkunde, Universität Paderborn
Pohlweg 47-49, D-33098 Paderborn, Germany
hmaier@zitmail.uni-paderborn.de

PM1000 is an oxide dispersion strengthened (ODS) nickel based superalloy with applications in
the aerospace and glass processing industries. It contains about 0.6 weight percent of incoherent
finely dispersed yttria particles. Earlier work has addressed the high- temperature isothermal low-
cycle fatigue (LCF) [1-2], creep [3] and oxidation [4] behaviour of PM1000. However, no known
work has been done to characterize the thermomechanical fatigue (TMF) performance of this
superalloy. The present work therefore involved a study of the high-temperature cyclic stress-
strain response of PM1000 under isothermal fatigue (IF) and TMF loading conditions. Both IF and
TMF tests were conducted in the temperature range from 450 to 950 °C, with mechanical strain
amplitudes varying from 0.35 to 0.5% in laboratory air conditions. The maximum mechanical
strain and maximum temperature occurs simultaneously in in-phase (IP) TMF, while out-of-phase
(OP) TMF has the maximum strain coinciding with minimum temperature.
The experiments were done on a servohydraulic Material Testing System (MTS 810). All
fatigue tests were conducted in fully reversed symmetrical push-pull using a triangular wave shape
as the strain command signal. Axial strains were measured with a high-temperature MTS
extensometer with 12 mm gauge length, attached to the specimens. The specimens were heated
with a high frequency induction heater. Temperature gradients along the specimen gauge length
were typically less than 4 °C, as monitored by a pyrometer.
The effects of the applied strain amplitudes, test temperatures and strain rates on the cyclic
deformation behaviour and fatigue life were examined. Figure 1 shows IP TMF hysteresis loops
recorded in the first cycle and at half life during a 450 to 850 °C temperature range test. As
expected, a compressive mean stress was observed in IP TMF, while OP TMF tests resulted in a
tensile mean stress. Slight hardening characterized the first few cycles of the IP TMF test, which
was followed by pronounced cyclic saturation. The stress-strain response observed compares quite
well with the work of Linde [5], where the TMF behaviour of MA 754, a British analogue of
PM1000, was investigated.
The initial hardening observed in the IP TMF is attributed to the yttria particles blocking
dislocation slip in the low temperature part of the cycle, thus resulting in an increase in dislocation
density. By contrast, initial softening was observed during the first few cycles of the IF tests,
followed by stable cyclic stress-strain response until macro crack growth set in.
1242 W. O. Ngala et al.

FIGURE 1: IP TMF hysteresis loop for the 450-850 °C temperature range.

The saturated isothermal stress amplitudes compare quite well with the corresponding stresses
at the maximum temperature of the TMF cycle, an indication of good microstructural stability.
Despite the similarity in stress amplitudes, it was found that fatigue lives are strongly dependent on
the actual loading conditions. IP TMF presents the most damaging test mode followed by IF, while
OP TMF resulted in the longest fatigue life.
The ramifications of the observations made with respect to modeling of cyclic stress-strain
response and damage evolution will be discussed.

References
1. Müller, F. E. H., Heilmaier, M. and Schultz, L., Mater. Sci. Eng., vol. A234-236, 509-512,
1997.
2. Heilmaier, M., Maier, H.J., Jung, A., Nganbe M., Müller F.E.H. and Christ H.-J., Mater. Sci.
Eng., vol. A281, 37-44, 2000.
3. Estrin, Y., Heilmaier M. and Drew, G., Mater. Sci. Eng., vol. A272, 163-173, 1999.
4. Weinbruch, S., Anastassiadis A., Ortner, H.M., Martinz, H.P. and Wilhartitz, P., Oxidation of
Metals, vol. 51, 111-128, 1999.
5. Linde, L. and Henderson, P.J., Scripta Metall. Mater., vol. 26, 1687-1692, 1992.
 35. High Temperature and Thermomechanical Fatigue 1243

FATIGUE-CREEP-ENVIRONMENT INTERACTIONS IN A DIRECTIONALLY-

SOLIDIFIED NI-BASE SUPERALLOY

A. P. Gordon1, M. M. Shenoy2, R. W. Neu3 and D. L. McDowell4

1George W. Woodruff School of Mechanical Engineering, Atlanta GA 30332-0405
2School of Materials Science and Engineering, Atlanta GA 30332-0245
1alipgordon@yahoo.com, 2gt6898a@mail.gatech.edu,
3rick.neu@me.gatech.edu, 4david.mcdowell@me.gatech.edu

Directionally solidified (DS) GTD-111 is a Ni-base superalloy designed to withstand creep

damage occurring in the first and second stage blades of gas-powered turbines. The service
conditions in these components, which generally exceed 600ºC, facilitate the onset of one or more
damage mechanisms via fatigue, creep, and environmental corrosion. Under these conditions
microstructural damage mechanisms operate interactively and independently to initiate cracks that
lead to the eventual reduction of service life. Because of the distinctive microstructure of DS
GTD-111, the manner in which these mechanisms interact to initiate cracks is related to grain
structure and chemical composition. In addition to fatigue cycling, certain sections of these high
temperature components are subjected to sustained dwell periods (i.e., creep) either in tension or
compression. Experiments have been carried out to simulate a variety of thermal, mechanical, and
environmental operating conditions endured by longitudinally (L) and transversely (T) oriented DS
GTD-111. In some case, tests in extreme environments/temperatures were needed to isolate one or
at most two of the damage mechanisms. Assuming a unique relationship between the damage
fraction and cycle fraction with respect to cycles to crack initiation for each damage mode, the total
crack initiation life has been represented in terms of the individual damage components (fatigue,
creep-fatigue, creep, and oxidation-fatigue respectively), and is based on that developed by Neu
and Sehitoglu [1] and modified by Gordon et al. [2]:

1 1 1 1 1
 c  f  cr  ox
Nitot Nifat
Ni Ni Ni (1)

Observations from micrographs (Fig. 1) of sections of oxidized, fatigued, and crept samples have
been incorporated to develop damage mechanism maps (DMMs). The DMMs, which quantify the
transitions between damage mechanism regimes, have been applied to develop physically-based
relationships for each of components of a crack initiation model.
A crystal plasticity based model has been formulated by Shenoy et al. [3] to capture the stress
strain behavior in the material. Fatigue (Fig. 2) and creep data were used to calibrate the model
which was implemented as an UMAT in ABAQUS. Using numerical simulations of experiments,
damage and crack initiation life predictions have been made. It was determined that under high
frequency and isothermal conditions with large plastic strain ranges and low temperatures, fatigue
damage dominates. The contribution of the coupled creep-fatigue mechanism appears by
increasing either the temperature or cycle time. Tests with small plastic strain range including
those with superimposed thermal cycling, are subject to environmental-fatigue damage since the
surface-related mechanisms are active at long exposure times.
1244 A. P. Gordon et al.

FIGURE 1: Oxide spiking observed in L-oriented DS GTD-111 with 982qC, 'H=0.5%, R=-
1, and 2min compressive holds.

FIGURE 2: Simulated and actual experimental responses of DS GTD-111 under

isothermal LCF in (a) L-orientation and (b) T-orientation at two strain rates. For each case
982qC (1800qF), RH=-1, and 'H = 1.0%.

References
1. Neu, R. W., and Sehitoglu, H., Metallurgical Transactions A, vol. 20, 1769-1783, 1989.
2. Gordon, A. P., Shenoy, M. M., and Neu, R. W. (2005). In Proceedings of 11th International
Congress of Fracture (ICF11), Turin, Italy, Elsevier Science.
3. Shenoy, M. M., Gordon, A. P., Neu, R. W., and McDowell, D. L., In press for publication in
the Journal of Engineering Materials and Technology, 2005.
 35. High Temperature and Thermomechanical Fatigue 1245

THE EFFECTS OF MICROSTRUCTURE, DEFORMATION MODE AND

ENVIRONMENT ON FATIGUE

S. D. Antolovich and B. F. Antolovich

School of MSE, Georgia Tech, Atlanta Georgia, 30332, USA Centre de Recherche de Trappes,
Trappes, France 78193
steve.antolovich@mse.gatech.edu, bruce.antolovich@crt-trappes.fr

So-called laws for predicting the lives of components at high temperatures have been proposed
(e.g. strain range partitioning, frequency modified fatigue life, time and cycle fraction etc.). These
“laws” usually work quite well for certain situations but are less successful in predicting life for
other conditions and materials that are quite similar. These laws are based on generic notions of
damage with several adjustable parameters. Such laws are useful from an engineering perspective
when considerable data base is available. However, they are far less successful in predicting
behaviour of new materials or existing materials under different external conditions. In essence,
most of these models have evolved into very sophisticated curve fits and generally fail to take into
account the actual behaviour of the material. For example, interactions between deformation
mechanisms in the material and environmental attack and the changing nature of the material are
not explicitly considered. In this paper, the physical damage processes that take place in high
temperature fatigue and fatigue crack propagation (FCP) of different commercial high temperature
alloys is presented. This information is used to develop fatigue physically-based models.
High Temperature Low Cycle Fatigue (LCF). In one set of studies, the LCF behaviour of a
series of Ni-base alloys was studied by Antolovich and co-workers [1,2,3,4]. Fatigue testing was
done over a range of temperatures, strain rates and strain ranges in air. Even after a very small
number of cycles significant microstructural changes which can have profound effects upon the
life were seen. Prominent changes that were observed included the following: rapid coarsening of

the J cprecipitates, extensive oxidation along grain boundaries and slip bands and precipitation of
carbides precipitates on slip bands and grain boundaries. These microstructural changes interacted
with the basic cyclic deformation mechanisms to cause damage. One example is shown in the
micrograph of Fig. 1. Slip bands are impinging on oxidized grain boundaries. Microcracks are
seen quite clearly at the tips of these slip bands where the stresses are very high. Furthermore,
detailed examination by selected area diffraction and dark field TEM conclusively demonstrated
that carbides precipitated on the slip bands and were themselves at least partially oxidized. The
behaviour depends sensitively on the nature of the cycle, the kinetics of stress-assisted
precipitation and the deformation mode; it is not captured in conventional life prediction
methodologies. Such information is used to develop a set of physics-based TMF models.

FIGURE 1. Slip-band-induced cracking of oxidized grain boundaries in Waspaloy tested in LCF at

1073K . Microcracks are indicated by the dark arrows.
1246 S. D. Antolovich and B. F. Antolovich

FCP in Single Crystalline Ni-base Alloys. Fatigue crack propagation studies were carried out on
CMSX-2, a typical superalloy used in blade applications which have been partially reported
elsewhere by Antolovich et al. [5]. Duplicate specimens were tested as a function of temperature
(298K and 973K), environments (air and 10-8 torr) and crystallographic orientation ([010] or [110]
with a common projected (100) crack plane). A detailed FEA analysis including the elastic
anisotropy was carried out to understand the different cracking modes that were observed for the
different conditions. It was found that the morphology of the crack surface was either “shearing”
or “precipitate avoidance” depending upon the ratio of the normal stress to the shear stress at a
given temperature and environment. Basically precipitate shearing and large crystallographic
facets which were not normal to the plane of loading were favored by low temperature, low stress
ratio, and vacuum. The two cracking modes are shown in Fig. 2.

FIGURE 2. Typical FCP fracture surface morphologies of CMSX-2. Precipitate shearing (a) and
avoidance (b). Morphology depended on temperature, orientation and stress ratio.

Detailed calculations for LCF and FCP that are based on fundamental physical processes are
presented in the full paper.

References
1. Lerch, B. A., Antolovich, S.D., and Jayaraman, N., J. Mat. Sci. and Engr., 151-165, 1984
2. Domas, P. D. and Antolovich, S.D., J. Engrg. Fract. Mech., Mechanics, 21, 203-214, 1984
3. Antolovich, S.D., Rosa, E and Pineau, A., J. Mat. Sci. and Engr., 47, 47-57, 1981.
4. Antolovich, S.D., Liu, S., and Baur, R., Met. Trans., 12A, 473-481, 1981.
5. Antolovich, B.F., Saxena,A., and Antolovich, S.D., In Superalloys 1992, Edited by S.D.
Antolovich, R.W. Stusrud, R.A. MacKay, D.L. Anton, T. Khan, R.D. Kissinger, D.L.
Klarstrom , AIME, Warrendale, PA, 1992, 727-736.
 35. High Temperature and Thermomechanical Fatigue 1247

COMPARING FATIGUE BEHAVIOUR OF TI6242 AND NOVEL TIAL

INTERMETALLICS

T. K. Heckel, A. Guerrero-Tovar and H. J. Christ

Universität Siegen, Institut für Werkstofftechnik
57068 Siegen, Germany
thomas.heckel@uni-siegen.de

In the present study the isothermal and thermomechanical fatigue behaviour of the commercial
near-alpha titanium alloy Ti6242 and the novel gamma-TiAl alloys TNB-V2 and TNB-V5a was
investigated and compared. The comparison was carried out in order to define the suitable
temperature range of application for each of these alloys and to illustrate the similarities and
differences in the cyclic deformation behaviour and damage evolution.
Ti6242 is the most widely used near-alpha high-temperature titanium alloy often applied in the
high-pressure compressor of civil jet engines. Although near-alpha titanium alloys feature a good
mechanical capability up to temperatures of 650°C, which represents the current compressor outlet
temperature, the maximum service temperature is restricted to 540°C because of the potential risk
of titanium fire. Therefore, it is still necessary to revert to heavy Ni-base superalloys in the last
stages of high-pressure compressors.
TNB-V2 and TNB-V5a belong to the latest developed near-gamma TiAl alloys (third
generation TiAl alloys), which possess superior mechanical properties compared to those of
previous generations (Paul et al. [1], Appel et al. [2], Appel et al. [3]). The increase in strength,
ductility and oxidation resistance is caused by a relatively high content of niobium (5-8 wt.%).
This class of near-gamma TiAl alloys holds the potential to fill the gap to the design of an all
titanium-based compressor, which would reduce weight and thus increase efficiency of jet engines.
Therefore it is essential to determine the fatigue properties of these novel alloys and compare the
observations to the rather well-known behaviour of commercial near alpha high-temperature Ti-
based alloys.
Isothermal as well as thermomechanical tests were conducted in both, laboratory air and
vacuum environment. The main purpose of the vacuum tests was to eliminate the effect of gas-
metal interaction on cyclic life, in order to characterize and quantify the extent of environmentally-
caused life reduction by means of a comparison with the cyclic life data obtained in air. All tests
were conducted in total strain control using a triangular command signal at a strain range of 1.4%
and a R-ratio of -1. The thermomechanical fatigue tests were carried out at a phase shift of 0° or
180° between strain and temperature signal (i.e., in-phase and out-of-phase loading, respectively).
Temperature ranged from 350 to 650°C for Ti6242 and from 550°C to 850°C for TNB alloys. The
microstructural changes were studied applying optical and transmission electron microscopy.
Crack initiation and propagation were investigated by means of scanning electron microscopy
often in combination with orientation imaging (OIM).
Figure 1 shows the stress amplitude and the corresponding cyclic life of Ti6242 observed in
isothermal fatigue tests. The materials was studied in the bi-modal condition, which is
recommended for applications in jet engines, since the bi-modal microstructure exhibits a good
balance between creep and fatigue properties (Eylon et al. [4], Lütjering and Williams [5]). It is
clearly seen in Fig. 1 that the cyclic strength of the material decreases continuously with increasing
temperature. More important, the cyclic life is tremendously reduced limiting strongly the
temperature range of application.
1248 T. K. Heckel et al.

The deformation behaviour of the TNB alloys, which were produced and heat-treated by
GKSS Research Centre, Geesthacht, Germany and which were tested in the lamellar condition, is
characterized by a rather pronounced ductile to brittle transition (DBT). As a consequence of the
high Nb content, the DBT temperature lies in the range of 750°C-800°C, i.e. well above the DBT
temperature of previous TiAl alloy generations. The cyclic stress-strain response and the
microstructure were found to be very stable both under isothermal and thermomechanical
conditions. However the materials are very susceptible to thermomechanical out-of-phase loading,
since this testing mode is connected with a tensile mean stress, oxidation in the high-temperature
part of the cycle, and a brittle behaviour (i.e. low damage tolerance) in the tensile-going low-
temperature cycle part.

FIGURE 1. Stress response and fatigue lifetime vs. temperature of Ti6242.

References
1. Paul, J.D.H., Appel, F. and Wagner, R., Acta Mater., vol. 46, 1075-1085, 1998.
2. Appel, F., Oehring, M. and Wagner, R., Intermetallics, vol. 8, 1283-1312, 2000.
3. Appel, F., Oehring, M., Paul, J.D.H. and Lorenz, U., In: Structural Intermetallics 2001,
Proceedings of the Third International Symposium on Structural Intermetallics, edited by
K.J. Hemker, D.M. Dimiduk, H. Clemens, R. Darolia, H. Inui, J.M. Larsen, V.K. Sikka, M.
Thomas and J.D. Whittenberger, TMS, Warrendale, 2001, 63-72.
4. Eylon, D., Fujishiro, S., Postans, P.J. and Froes, F.H., Journal of Metals, vol. 36, 55-62, 1984.
5. Lütjering, G. and Williams, J.C., Titanium, Springer, Berlin, Germany, 2003.
 35. High Temperature and Thermomechanical Fatigue 1249

A TBC FAILURE MODEL BASED ON CRACK NUMBER DENSITY

Xijia Wu1, Zhong Zhang2 and Rong Liu2

1Structures and Materials Performance Laboratory

National Research Council of Canada

1200 Montreal Road, Ottawa, ON, Canada K1A 0R6
Tel: 1-613-990-5051; Fax: 1-613-990-7444
2Department of Aerospace and Mechanical Engineering

Carleton University
1125 Colonel By Drive, Ottawa, Ontario, Canada K1S 5B6
xijia.wu@nrc-cnrc.gc.ca

Thermal barrier coatings (TBC) are used as protective coatings for hot-section components in
advanced gas turbine engines. When exposed to the hot gas, a layer of thermally grown oxides
(TGO) forms at the interface between the ceramic topcoat and the metallic bond coat in TBC,
which often causes cracking and eventually leads to TBC spallation. The damage accumulation in
TBC appears as the collective results of nucleation and growth of numerous microcracks. A
description of the evolution of such a process is the key to assess TBC durability.
Based on the crack number theory, the evolution of the crack number is given as [1]:

wn w
 [ c n ] nN
wt wc (1)

where n is the number of cracks at time t, c is the crack length and c is the crack growth rate, and
nN is the crack nucleation rate.
Generally, Eq. (1) determines the crack number distribution against the crack size as function
of time, at given nucleation and growth rates.
An air-plasma sprayed TBC was examined. A comparison of the above model with
experimentally established crack number distribution is presented, as shown in Fig. 1. The
mechanism of crack nucleation and growth will also be discussed.

FIGURE 1. Crack number function for 24 hour exposure at 1200oC.

References
1250 X. Wu et al.

1. Fang, S B., Hong, Y.S., and Bai, Y.L. “Experimental and theoretical study on numerical
density evolution of short fatigue cracks”, Acta Mechanica Sinica (English Edition) 11, 144-
152.
 36. Impact Failure of Laminated and Sandwich Composite Structures 1251

IMPACT INDUCED COMPOSITE DELAMINATION: STATE AND PARAMETER

IDENTIFICATION VIA UNSCENTED KALMAN FILTER

Alberto Corigliano, Aldo Ghisi and Stefano Mariani

Dipartimento di Ingegneria Strutturale, Politecnico di Milano
Piazza Leonardo da Vinci 32, 20133 – Milano, ITALY
alberto.corigliano@polimi.it, aldo.ghisi@polimi.it, stefano.mariani@polimi.it

The problem of impact induced delamination in layered composites (Abrate [1]) is considered and
numerically modelled on the basis of the following assumptions: the layers are elastic, interface
softening laws are used to simulate the progressive interlaminar decohesion due to delamination
(Allix and Corigliano [2], Corigliano [3]). An explicit dynamic finite element code is used for the
step by step analysis.
The main purpose of the present study is to evaluate the performance of the recently proposed
sigma-point Kalman filter, also termed unscented Kalman filter (Wan and van der Merwe [4],
Ljung [5]), for real-time identification of the unknown interface properties and of the debonding
surface(s) under impact loading. Only free-surface measurements are used to drive the
identification procedure.
Among the most important features of the unscented Kalman filter, it is important to mention
that: it does not require the computation of the gradient (Jacobian) of the equation of motion; the
statistics of the system state variables (model parameters and nodal displacements-velocities-
accelerations) are accurately propagated in time up to the third-order. Hence, the sigma-point
transformation turns out to be superior to the first-order approximation featured by the extended
Kalman filter (see Corigliano and Mariani [6], Mariani and Corigliano [7]).

FIGURE 1. Impact on a SiC specimen. (a): space-time diagram; (b): free-surface velocity record.

The filtering procedure is here applied to the experimental results of Dandekar and Bartkowski [8],
concerning Silicon Carbide specimens subject to shock loading. In these tests, spalling takes place
across a material interface whose location is determined by means of an elastic-brittle analysis; the
free-surface velocity is considered as the only available experimental datum (see Fig. 1). Key
constitutive parameters, like the fracture energy, are identified and the performances of the
extended and of the unscented Kalman filters are critically compared (see Fig. 2).
1252 A. Corigliano et al.

FIGURE 2. Impact on a SiC specimen: current estimated value of the fracture energy.

References
1. S. Abrate. Impact on composite structures. Cambridge University Press, 1998.
2. O. Allix and A. Corigliano. International Journal of Solids and Structures, vol. 36, 2189-
2216, 1999.
3. A. Corigliano. Comprehensive Structural Integrity, vol. 3, chapter 9. Elsevier Science, 2003.
4. E.A. Wan and R. van der Merwe. In S. Haykin, editor, Kalman filtering and neural networks,
221-280. John Wiley & Sons, Inc., 2001.
5. L. Ljung. System identification. Theory for the user. Prentice Hall, 1999.
6. A. Corigliano and S. Mariani. Computer Methods in Applied Mechanics and Engineering,
vol. 193, 3807-383, 2004.
7. S. Mariani and A. Corigliano. Computer Methods in Applied Mechanics and Engineering, in
press, 2005.
8. D.P. Dandekar and P.T. Bartkowski. Technical Report ARL-TR-2430, Army Research
Laboratory, Aberdeen Proving Ground, 2001.
 36. Impact Failure of Laminated and Sandwich Composite Structures 1253

MODELLING IMPACT DAMAGE IN SANDWICH CONCEPT STRUCTURES

A. Johnson and N. Pentecote

German Aerospace Center (DLR), Institute of Structures and Design
Pfaffenwaldring 38-40, 70569 Stuttgart, Germany
alastair.johnson@dlr.de, nathalie.pentecote@dlr.de

To reduce development and certification costs for composite aircraft structures, efficient
computational methods are required by the industry to predict structural integrity and failure under
dynamic loads, such as crash and impact. By using meso-scale models based on continuum
damage mechanics (CDM), proposed by Ladevèze and co-workers [1], [2], it is possible to define
materials models for FE codes at the structural macro level which embody the salient
micromechanics failure behaviour. CDM provides a framework within which in-ply and
delamination failures may be modelled. In previous work [1], [3] ply failure models were
developed for unidirectional (UD) fibre and fabric reinforced plies with three scalar damage
parameters representing in ply microdamage and damage evolution equations introduced relating
the damage parameters to strain energy release rates in the ply. Delamination models for interply
failure were obtained by applying the CDM framework to the ply interface, as described in [2].
Failure at the interface is modelled by degrading stresses using two interface damage parameters
corresponding to interfacial tension and shear failures, whilst fracture mechanics concepts are
introduced by relating the total energy absorbed in the damaging process to the interfacial fracture
energy. The ply CDM and delamination models have been implemented into a commercial explicit
FE crash and impact code [3], which uses a numerical approach for delamination modelling based
on stacked shell elements with contact interface conditions, which may separate when the interface
failure condition is reached.
The paper describes the application of these simulation methods to design concept studies on a
novel form of double-walled composite panels with energy absorbing cores currently being
assessed for use in aircraft structures. These are non-standard sandwich structures, in which a main
load-bearing composite laminate is protected from impact damage by an energy absorbing core
and a second cover laminate. Core materials being considered include folded composite plate
structures, polymer foams and Nomex honeycomb. Impact load cases of interest include high
velocity impacts from steel impactors and deformable, soft body impactors such as ice and rubber.
Representative structures are modelled and FE simulation results are presented, which simulate
numerically the observed impact failure modes and failure progression under medium to high
impact velocities representative of civil aircraft applications. Of particular interest is to determine
the impact damage threshold in the inner load-bearing laminate for impacts on the cover laminate,
as impactor type and impact energy is varied. Effective delamination models are required in this
inner laminate to determine damage levels for the cases when the projectile penetrates the cover
laminate and core. FE simulation results are compared with gas gun impact test data on idealised
double-wall panel structures at impact velocities in the range 60 – 250 m/s.
The sandwich skin composite laminate is modelled by layered shell elements or stacked shells
with a contact interface which may fail by delamination. The shells are composed of composite
plies which are modelled as a homogeneous orthotropic elastic damaging material whose
properties may be degraded on loading by microcracking prior to ultimate failure. The simulation
results presented here are based on the ‘bi-phase’ ply model for UD and fabric plies in which it is
supposed that damage evolution is dependent on strain invariants. In the layered shell element the
stiffness properties of the plies are degraded as the shear strain invariant increases until eventually
a damaged shell element is eliminated from the computation when the shear strain invariant
1254 A. Johnson and N. Pentecote

reaches a pre-defined critical value. A delamination model [3] is implemented in the PAM-
CRASHTM code, with the laminate modelled as a stack of shell elements. Each ply or sublaminate
ply group is represented by a set of layered shell elements and the individual sublaminate shells are
connected together using a contact interface with an interface traction-displacement law. Contact
may be broken when the interface energy dissipated reaches the mixed mode delamination energy
criteria. This ‘stacked shell’ approach is an efficient way of modelling delamination, with the
advantage that the critical integration timestep is relatively large since it depends on the area size
of the shell elements not on the interply thickness.

FIGURE 1. Impact model of foldcore sandwich plate (M = 22 gram, V0 = 105)

There is current interest in sandwich panels with carbon composite skins and a folded aramid paper
composite core (foldcore), with a complex periodic folded geometry. The core is modelled in detail
by shell elements with the bi-phase materials model, together with layered shells or stacked shell
elements and delamination interfaces for the sandwich skins. Fig. 1 shows simulation results for
impact penetration from a 22 gram rigid projectile at 105 m/s normal impact. These conditions
model stone impact and gas gun tests at the DLR led to penetration of the outer skin, which was
also predicted in the FE model. Thus the methodology developed here may be used further to
evaluate monolithic, double shell and sandwich design concepts for aircraft fuselage and wing
structures subjected to high velocity impact loads.

References
1. Ladevèze, P., Inelastic strains and damage, Chapt. 4 in Damage Mechanics of Composite
Materials, R. Talreja (ed), Composite Materials Series, Vol 9, Elsevier, 1994.
2. Allix, O., Ladevèze, P., Interlaminar interface modelling for the prediction of delamination,
Composites Structures, 22, 235-242, 1992.
3. Johnson, A.F., Pickett, A.K., Rozycki, P., Computational methods for predicting impact
damage in composite structures, Composites Science and Technology, 61, 2183–2192, 2001.
 36. Impact Failure of Laminated and Sandwich Composite Structures 1255

PUNCH SHEAR BEHAVIOR OF COMPOSITES AT LOW AND HIGH RATES

B. A. Gama and J. W. Gillespie Jr.

University of Delaware Center for Composite Materials (UD-CCM)
209 Composite Center, UD, Newark, DE 19716, USA
gama@ccm.udel.edu, gillespie@ccm.udel.edu

A quasi-static penetration model of ballistic penetration of thick section composites is proposed.

Quasi-static punch shear tests (QS-PST) and ballistic punch shear tests (B-PST) are conducted to
identify five different phases of ballistic penetration, i.e., i) impact-contact, ii) hydrostatic
compression, iii) compression-shear, iv) tension-shear, and v) structural vibration. It is well
known that the energy absorption in QS-PST is much lower than the B-PST e.g., Mines et al [1];
and the physics of ballistic penetration is difficult to model with quasi-static models. In order to
bridge the gap between QS-PST and B-PST energy absorption, conservation of momentum and
energy principles is used to predict the rest kinetic energy of the projectile-laminate system in case
of a partial penetrating projectile. Combining the QS-Penetration model and the analytical rest
kinetic energy model, a good prediction of ballistic energy absorption is obtained.

FIGURE 1. Quasi-Static Punch Shear Test: Load-Deflection and Damage.

The authors [2, 3] have shown that QS-PST at different span to punch ratio (SPR) can be used to
simulate different phases of ballistic damage mechanisms e.g., hydrostatic compression (SPR =
0.0), compression-shear (SPR = 1.1 & 2.0), and tension-shear (SPR = 4.0 & 8.0). Fig. 1 shows the
QS-PST behavior of plain-weave (814 g/m2) S-2 glass/SC15 composites as a function of SPR.
The energy absorption in QS- and B-PST is summarized in Table 1. Clearly, the energy absorption
increases with SPR pointing to the fact that an increase in SPR involves more volume of material
and the plate goes under large deformation. At SPR = 8, the ratio of ballistic to QS-PST is about
1.82, a well documented phenomenon in Ref. [1]. Comparing the ballistic penetration phases and
QS-PST load-deflections as a function of SPR, a QS-Penetration model of ballistic penetration is
proposed by the authors [3].
C C
P Pmax 0  G  G max
§G GF · C
P0  P1 ¨ ¸ G max  G  GF (2)
© G ¹
1256 B. A. Gama and J. W. Gillespie Jr.

Where, the impact-contact force is assumed to be constant in the hydrostatic compression

1
phase and to have a G behavior in the other penetration phases. This QS-Penetration model
provided the concept of a QS-PST-Envelope [3] curve, which represents the total energy absorbed
by material damage, and is about 80% of the total ballistic energy.

TABLE 1. Energy Absorption in QS- & B-PST.

It is also known that at ballistic limit, the projectile is arrested, and the total energy of impact is
transferred to the composite plate. If one assumes that the projectile forms a moving cone of the
plate with a velocity profile, 9 r  >V C / 1  * [ @1  [ / r ,the conservation of momentum
and energy principles yields the following relation for the rest kinetic energy of the projectile-
PL
laminate system, E KE .

P -L
E KE ª ­° *[2 *[ ½° º
: PL / 2C «1  * 0 ® ln *[   ¾»
°̄ *[  1 
2
EI «¬ *[  1 °¿ »
¼ (3)

Where, / C VC / V I >1  * 0 / 2 *[  1 @ , V is the maximum cone velocity. * 0 is the

1
C

areal-density ratio of the laminate to projectile, and * [ is the cone to projectile diameter ratio.
The energy absorption at ballistic limit then can be expressed in terms of the QS-PST-Envelope
Envelope
energy as: E I
Damage
E Material / 1  :  . A good correlation between QS-PST and B-PST
QS  PST
experiments is obtained using these models.

References
1. Mines, R.A.W., Roach, A.M., Jones, N., Int. J. Impact Engng, vol. 22, 561-588, 1999.
2. Gama, B.A., Rahman, M., and Gillespie Jr., J.W., In CD Proceedings of SAMPE 2004, Long
Beach, California, 2004, 909-921.
3. Gama, B.A., Islam, S.M.W., Rahman, M., Gillespie Jr., J.W., Bogetti, T.A., Cheeseman,
B.A., Yen, C-F., and Hoppel, C.P.R., In CD Proceeding of SAMPE 2005, Long Beach,
California, 2005.
 36. Impact Failure of Laminated and Sandwich Composite Structures 1257

REPEATED IMPACT BEHAVIOUR AND DAMAGE PROGRESSION OF

GLASS REINFORCED PLASTICS

G. Belingardi, M. P. Cavatorta and D. S. Paolino

Politecnico di Torino, Dipartimento di Meccanica
Corso Duca degli Abruzzi 24 – Torino – Italy
giovanni.belingardi@polito.it, maria.cavatorta@polito.it, davide.paolino@polito.it

The paper investigates the response of glass fibre reinforced plastics subjected to repeated low
velocity impact loading. The loading condition of repeated impacts is of particular relevance for
industrial applications, especially when intended for motor vehicle or naval applications. Recently
few papers addressed the problem [1-3].
Experimental tests are performed according to ASTM 3029 standard using an instrumented
free-fall drop dart testing machine. Specimens are 100 x 100 mm square plates, mechanically
constrained over a circular 76.2 mm diameter edge. The impactor has a total mass of 20 kg, its
head is hemispherical with a radius of 10 mm. Static indentation and dropweight impact tests were
performed; the comparison of quasi-static and impact curves shows that the material, under the
stated loading conditions, has no sensitivity to strain-rate effects.
Four falling heights were considered (125 mm, 250 mm, 500 mm, 750 mm) and four
specimens for each height were subjected to ten repeated impacts. By means of a piezoelectric load
cell, force-time curves were acquired and, with a double integration, force-displacement and
deformation energy-displacement curves were obtained [4]. From the force-displacement curves,
the actual value of the bending stiffness was calculated. The variation of the plate bending stiffness
was chosen as the indicator for the progression of the structural damage within the laminate.
Fig.1 shows the mean peak load as a function of the impact number. As visible, for the 125 mm
and the 250 mm specimens the peak load slightly increases in the first impacts as compaction of
the different layers is achieved. Analysis of the impacted surfaces shows delamination damage
from the first impact; however, no penetration is achieved even at the tenth impact. For the 500
mm and the 750 mm specimens, the peak force reaches a maximum and then decreases. The
maximum in the peak force seems to correspond to penetration of the first unidirectional ply. For
these specimens, delamination and fibre breakage are observed. For the 750 mm specimens,
perforation of the laminate plate occurs at the ninth impact.
The peak force at the first impact increases almost linearly with the falling height (i.e. the
impact energy). The linear relationship continues to exist at higher impact numbers only for the
125 mm and the 250 mm specimens, for which the penetration threshold is not reached even at the
tenth impact [5].
1258 G. Belingardi et al.

FIGURE 1. Peak impact load as a function of impact number.

Fig. 2 on the left end side plots the mean stiffness against the impact number. As visible, the
stiffness-impact plot shows a monotone decreasing trend, with the maximum stiffness loss
concentrated in the first impact. The overall stiffness reduction increases for increasing falling
heights. Fig.2 on the right end side reports both the mean stiffness and the total delamination area
plotted against the impact number for the 500 mm specimens. Correlation between the area of
delamination and the stiffness reduction is evident.

FIGURE 2. Stiffness vs. number of impact (left) Correlation between stiffness reduction and
delamination area - h=500 mm- (right).

References
1. Azouaoui, K., Rechak, S., Azari, Z., Benmedakhene, S., Laksimi, A. and Pluvinage, G., Int J
Fatigue, vol. 23, 877-885, 2001.
2. Roy, R., Sarkar, B.H. and Bose, N.R., Composites: Part A, vol. 32, 871-876, 2001.
3. Baucom, J.N. and Zikry, M.A., Composites: Part A, vol. 36, 658-664, 2005.
4. Belingardi, G. and Vadori, R., Composite Structures, vol. 61, 27-38, 2003.
5. Liu, D., J. Composite Materials, vol. 38, 1425-1442, 2004.
 36. Impact Failure of Laminated and Sandwich Composite Structures 1259

IMPACT BEHAVIOUR MODELLING OF A COMPOSITE LEADING EDGE

STRUCTURE

G. Labeas and Th. Kermanidis

Laboratory of Technology and Strength of Materials
Department of Mechanical Engineering and Aeronautics,
University of Patras, 26500 Rion, Greece
labeas@mech.upatras.gr

A methodology for the numerical simulation of bird impact on a novel Leading Edge (LE)
structure of a Horizontal Tail Plane is presented. The innovative LE design is based on a ‘tensor
skin’ structure, which is an efficient impact resistant design, comprising one or more folded
composite sub-laminates, which unfold during the bird impact, providing the required high-energy
absorption efficiency. The simulation technique is based on the non-linear dynamic Finite Element
PAM-CRASH code and is performed in two steps.
The first step deals with the development of suitable material damage models for the
composite fabric materials of the LE skin. The developed damage models are capable to describe
the material behaviour at the high strain rate conditions, which occurr during the high velocity bird
strike event. Different material damage modelling approaches were considered, resulting to a
successful and innovative solution for the representation of the composite fabric materials stress –
strain behaviour.
The second step of the numerical methodology deals with the numerical simulation of bird
strike experiments on representative LE structures. Critical issues of the modelling, such as, the
damage model of the bird impactor, which is considered as flexible in the current analysis, the
mesh density of the highly impacted areas and the methods to represent the actual contact
phenomenon of the structural interfaces, are presented. The validation of the methodology is
performed by comparisons between numerical and experimental results. Assessing the optimal set
of parameters in the numerical model, such as, mesh density, contact thickness and anti-crossing
force parameter of the Smooth Particle Hydrodynamics (SPH) bird model, required a thorough
examination of the way each parameter affects the numerical results.
A comparison between the experimentally and numerically predicted bird strike sequence of
the Leading Edge structure is presented in figure 1. Comparisons between measured and calculated
test rig reaction force histories were also performed, in order to evaluate the model efficiency (see
fig. 2). The numerical predictions for an 80 m/s bird strike indicate that the structure could survive
the impact with minimum laminate damage. These simulations are found to be in good agreement
with test results. The measured and calculated reaction loads are also found to be in a good
agreement, signalising that the numerical model with the optimised material damage model and the
bird impactor modelled as a flexible body reaches a high level of success in predicting the loads
and damage of the impacted structure.
It can be concluded that a simulation methodology for the bird strike on a composite LE was
developed, which comprises a useful tool for the analysis of the high velocity impact of a large
object on an anisotropic thin shell structure with complex geometry. The work is part of BRITE
EURAM Project “CRAHVI” (“CRashworthiness of Aircraft for High Velocity Impact”) supported
by European Union. The authors wish to acknowledge financial support from the CEC FW5
Research Program and the CRAHVI Partners for their scientific contributions.
1260 G. Labeas and Th. Kermanidis

FIGURE 1: Comparison of experimentally and numerically predicted bird strike sequence of a

Leading Edge structure

FIGURE 2: Comparison between experimentally measured and numerically predicted test rig
reaction forces during bird strike

References
1. Gallard JP. Report on Composite Commuter Leading Edge Bird Strike Tests. CRAHVI
internal report, CEAT, 2002
2. PAM-CRASHTM Solver notes and reference. ESI Group, 2001
3. Johnson AF, Pentecôte N, Kraft H and Weissinger H. Measurement of Dyneema/epoxy
composite Mechanical Properties, CRAHVI internal report, DLR IB 435-2002/15, 2002
 36. Impact Failure of Laminated and Sandwich Composite Structures 1261

BENDING STRENGTH OF SANDWICH PANELS WITH DIFFERENT CORES

AFTER IMPACT

W. Goettner and H. G. Reimerdes

Institut für Leichtbau, Aachen University
Wüllnerstr. 7, D-52062 Aachen, Germany
hg_reim@ilb.rwth-aachen.de

In this paper results of low velocity impact tests on sandwich panels are presented. All investigated
panels have the same quasi isotropic CFRP-face sheets. Goal of this work was to investigate the
impact behaviour of panels having different cores. Besides Nomex honeycomb core, slotted
Nomex honeycomb core, and slotted Aramid honeycomb core, foam cores with embedded CFRP-
pins (X-core and K-core, see Fig. 1 from [1]) are investigated. The pins are connected to the top
layers to reinforce the panels mechanical properties.

FIGURE 1. X-core and K-core structures [1].

The tests are performed using a drop tower with an impact body which has a ball-shaped head of
one inch diameter. Various impact energies are used to generate different damage mechanisms in
the panels. During the tests the reaction force, the displacement and the impact velocity are
measured.
After impact, the achieved damage sizes in the upper face sheets and in the core are measured
by means of ultrasonic scanning. Selected specimens with impact damage are cut out of the panels
and analyzed by use of a scanning-electron microscope.

FIGURE 2. Wrinkling failure of sandwich panel in the four-point bending tests.

The bending strength of undamaged and of impacted panels was measured in four-point bending
tests. In most cases the specimen failed in wrinkling of the upper face sheets (FIG. 2). This was
confirmed by simple analytical investigations of the problem.
1262 W. Goettner and H. G. Reimerdes

In parallel, the impacts were simulated numerically using LS-DYNA (FIG. 3). By use of
material data achieved by measurement a good agreement between tests and numerical
investigations was reached (FIG.4).

FIGURE 3. FE-model of the X-core panel.

FIGURE 4. Impact force versus time.

References
1. AZTEX Inc. http://www.zfiber.com/x-cor.php
 36. Impact Failure of Laminated and Sandwich Composite Structures 1263

ENERGY ABSORBING ABILITY OF SANDWICH COMPOSITE STRUCTURES

J. P. Dear, W. Maruszewska, S. T. Oh and H. Lee

Imperial College London
South Kensington Campus, London, SW7 2AZ
j.dear@imperial.ac.uk

There is increasing interest in the impact energy absorbing ability of lightweight sandwich and
similar materials. This is for reducing the unladen weight of transporters to achieve improved
operational performance with better fuel economy and lower other direct operating costs. These
requirements very much relate to the increasing demand for more transportation systems and the
need for increased safety as traffic congestion rises. Two safety needs are to reduce the risk of
penetration into the cabin space when impacts occur and to lower the deceleration rate for the
occupants. The aim of this study was to investigate the impact energy absorbing ability of
sandwich materials that have different skin and honeycomb cores. This was for different impact
conditions as simulated by drop-weight and constant velocity impact experiments. These studies
assess the energy absorbing damage at all stages of the impact and relate this to the impact
conditions in terms of size and shape of impact contact surface and closing velocities. Also,
determined were the retained integrity of the sandwich materials at different stages of the impact
up to and after it has been penetrated.
Examples of some of the sandwich materials evaluated [1-4] are given in Table 1.

TABLE 1. Examples of sandwich material.

For the drop-weight impact experiments, an impactor mass of 1.55 kg was used with drop heights
of 0.2 to 1.0 m (up to 4.4 m s-1). Honeycomb sandwich material specimens were 150 mm square
with support diameter of 100 mm. Force-time and energy absorption data were correlated to high-
speed photography as well as c-scan and sectional damage assessments. Fig. 1 shows the behaviour
of a Fibrelam sandwich panel for 0.2, 0.6 and 1.0 m drops with the force-time trace for a 1.0 m
drop annotated with frame numbers from a high-speed sequence. Fig. 2 shows that Fibrelam panels
exhibit the highest peak loads and this is for the lowest maximum displacement of the impactor.
The c-scan damage data show that the Fibrelam and H620 front skins spread the impact
deformation process more than the skins of H220 and H640 reducing the depth of sectional
damage.
1264 J. P. Dear et al.

FIGURE 1. Fibrelam: (a) Force-time and (b) Energy absorption for 0.2, 0.6 and 1.0 m drops; (c)
Force-time and (d) High-speed sequence for 1.0 m drop.

FIGURE 2. Impact response and damage: (a) For different sandwich panels; (b) Example of c-scan
related to sectional damage for Fibrelam panel for a 1.0 m drop.

References
1. Dear, J.P., Lee H. and Brown, S.A., Int. J. Impact Eng., In Press for 2005.
2. Lee, H., Drop-weight and ballistic impact of honeycomb sandwich structures, PhD Thesis,
Imperial College London, University of London, 2004.
3. Maruszewska, W., Failure processes in composite sandwich structures for automotive and
similar applications, PhD Thesis, Imperial College London, University of London, 2005.
4. Oh, S.T., Impact response and damage to composite pultruded beams, PhD Thesis, Imperial
College London, University of London, 2005.
 36. Impact Failure of Laminated and Sandwich Composite Structures 1265

IMPACT BEHAVIOUR OF METAL FOAM CORED SANDWICH BEAMS

S. McKown and R. A. W. Mines

University of Liverpool,
Department of Engineering, Brownlow Street, Liverpool, L69 3GH, UK
smckown@liv.ac.uk, r.mines@liv.ac.uk

Previous work on the drop weight impact behaviour of polymer composite sandwich beams subject
to three point bending has been concerned with beams with polymeric cores (Divinycell H100,
H200) and pre-preg UD glass epoxy skins (SP Systems SE84) laid up in a cross ply form [1,2]. In
the work, a mass of 1.49 kg was dropped from a height of 2 metres giving an impact velocity of 5.9
m/s and an impact energy of 26 J. The beam span was 200mm and the core depth was 10mm. The
impact energy was enough to cause failure of the upper skin, and some progressive collapse of the
core. The failure of the upper skin was due to in-plane compression.
In references [1,2], the experimental tests were simulated using the implicit finite element
code, ABAQUS Standard. In the simulation, the failure of the upper skin was modelled using a
user defined field, in which the stiffness of the skin was degraded as a function of strain and the
skin then completely failed at a given strain. The multi-axial crush behaviour of the core was
modelled using critical state theory (i.e. the *FOAM model), and both uniaxial and hydrostatic
foam crush data was used as input into the simulation. Good agreement was shown between
experiment and simulation. It was shown that strain rate effects for skin and core damage were
secondary, and so these effects were not included in the analyses.
This work has now been extended to three point bend beams with metal foam cores (Alporas)
and UD glass reinforced polypropylene skins (Plytron). The skin lay up was similar, i.e. crossply.
This paper describes some experimental tests, in which a mass of 1.83 kg is dropped from a
height of 2 m, giving an impact velocity of 5.9 m/s and an impact energy of 32 J. The beam span
was 200mm, and the core depth was between 20 and 40mm. It is shown that the mode of beam
failure is similar to the polymeric foam cored case, i.e. upper skin compression failure and stable
crushing of the core. However, an additional complication is core shear failure at large beam
deflections, after the upper skin has failed.
Experimental tests are simulated using the explicit finite element code, DYNA. As the focus
of this study was the progressive collapse of the metal foam core, a simple damage model was
employed for the skin. This was based on the Hashin criterion [3] for initiation and a constant
damage stress to a specific failure strain for post failure behaviour (Mat54 in DYNA).
The simulation of the crush behaviour of the core was achieved with a continuum model
(Mat63 in DYNA). A comparison of the available continuum metal foam crush models in DYNA
has been given by Hanssen et. Al. [4], and they conclude that all the available models have their
advantages and disadvantages. It was decided here to use the simplest continuum model in
DYNA. Hydrostatic core crush and core and skin strain rate effects were not included in the model,
as these were thought to be secondary.
However, the shear failure of the metal foam core after upper skin failure was modelled, and a
maximum principal strain criterion was used. This criterion will be compared with other failure
criteria in the literature.
A problem with using an explicit finite element code is the modelling of quasi static tests.
Quasi-static conditions were modelled by using increased materials densities and adjusted force
time input.
1266 S. McKown and R. A. W. Mines

The static and impact numerical simulations compare well with experiment, from the points of
view of remote force and beam displacement data, as well as overall beam deformation profiles.
The beam numerical simulation raises some interesting issues relating to the modelling of the
progressive collapse of composite sandwich structures. Firstly, the applicability of a continuum
model to the foam core is discussed. Typically, cell sizes in the Alporas foam are of the order of
3mm, and this size can vary by up to 100%. This means that the core depth in the sandwich
structure needs to be at least 20mm, and preferably 40mm, for a continuum approach in foam core
modelling to be valid.
Also, continuum models become inaccurate for shear stress cases and for progress damage
cases. These issues were investigated using an Arcan testing technique for foams [5], in which
simultaneous tensile and shear stresses can be imposed on a block of foam. Experimental data
from foam Arcan tests were simulated using DYNA, and the limitations in continuum modelling
defined.
This work goes some way to defining the limitations of continuum modelling for the
progressive collapse of composite sandwich beams. The paper briefly discusses the alternative
approach of micro mechanical modeling for the metal foam, as a way of more accurately
modelling shear and progressive metal foam failure, in the context of the given structure.

References
1. Mines, R.A.W. and Alias, A., Comp. Part A, vol 33,11-26,2002
2. Mines, R.A.W. and Alias, A. In Progress in Structural Mechanics, edited by F Paris and J
Morton, Seville University Press, 2000, 11-21.
3. Hashin, Z., J. Appl. Mech., vol. 47, 329-334, 1980
4. Hanssen, A.G., Hopperstad, O.S., Langseth, M., and Ilstad, H., Int. J. Mech. Sc., vol 44, 359-
406, 2002
5. Chen, C., and Fleck, N.A., J. Mech. Phys. Sol., vol. 50, 955-977, 2002
 37. Mesofracture and transferability 1267

STRESS GRADIENT AT NOTCH ROOTS USING VOLUMETRIC METHOD

H. Adib and G. Pluvinage

Post-doctorant, Laboratoire de Fiabilité Mécanique (LFM)
Full professor, Laboratoire de Fiabilité Mécanique (LFM)
Université de Metz-ENIM, Ile du Saulcy, 57045 Metz, France
hradib_2000@yahoo.com

The hot spot methods exhibit few defects to appraise flawless fatigue life prediction. The mentioned
methods suffer from a real maximum stress location and magnitude, elaborated plastic zone features at
notch tips or welded toes [G. Pluvinage, Fracture and Fatigue Emanating from stress concentrators,
Dordrecht: Kluwer Academic Publishers, 2003. ] and stress gradient [A. Buch, Analytical approach to
size and notch effects in fatigue of material specimens, Material Science and Engineering, Vol. 15, 75-
85, 1974.][A. Brandt, Calcul des pièces à la fatigue par la méthode du gradient, Editor CETIM, 1980.].
According to the finite element elastic-plastic stress distribution along notch tips, the peak stress does
not appear at notch head and it requires certain distance to attain maximum stress value. Moreover, the
fatigue mechanism needs to be considered as a physical volume according to fatigue tests large scatters
based on Weibull‘s distribution method, specimen size and stress gradient dependence of experimental
fatigue results [G. Pluvinage, Fracture and Fatigue Emanating from stress concentrators, Dordrecht:
Kluwer Academic Publishers, 2003. ]. The stress gradients around notch tips are also significant
according to the weight influence on extracted stresses along notch tips [G. Qylafku, Z. Azari, N. Kadi,
M. Gjonaj, G. Pluvinage, Application of a new model proposal for fatigue life prediction on the notches
and key-seats, International Journal of Fatigue,21, 753-760, 1999.][H. Adib, G. Pluvinage, Theoretical
and numerical aspects of volumetric approach for fatigue life prediction in notched components,
International Journal of Fatigue, 25(1), 67-76, 2003.]. In the present study, the role of relative stress
gradient for fatigue life assessment using volumetric method is highlighted. The polynomial volumetric
method point of view has been proposed which satisfies weight function conditions, removes numerical
derivation errors, elucidates effective stress terms as subtraction of average stress and relative stress
gradient based phrase. According to the elastic-plastic finite element outcomes for considered notched
specimens, Stress Peak Trajectory Path “SPTP” and Effective Stress Trajectory Path “ESTP” have been
proposed. The “PSTP” represents linear characteristics in logarithmic diagram and it can be applied to
avoid more excessive mesh density generation refinement around notch roots for low applied loading
magnitudes. The “ESTP” yields the effective distance-effective stress curve and effective stress-applied
loading magnitude curves which can facilitate the calculation of effective stress and effective distance
due to the parabolic shape configuration of mentioned curves.

FIGURE 1. a) Maximum principal stress distribution including Peak Stress Trajectory Path b) Effective Stress
Trajectory Path versus applied loading magnitude and effective stress.
1268 H. Adib and G. Pluvinage

The volumetric method and Stress Field Intensity method are compared for the weight functions
and corresponding stresses at notch roots. The comparisons exhibit that polynomial volumetric
weight function completely satisfies all necessary requirements and the volumetric effective stress
is lightly greater than Stress Field Intensity stress. The current phenomenon concludes less number
of cycles to failure relative to the Stress Field Intensity method which overestimates it.
In the present study, volumetric method is investigated and compared to the stress field
intensity method. The main outcomes can be summarized as:
1 The stress gradients near notch roots play an important role in fatigue life prediction of notched bodies
and the non-gradient based methods can not perfectly assess fatigue life duration of multi-notched
specimens.
2 The stress concentration values near notch ahead are not solely main object and it is necessary to be
accompanied by stress gradient considerations.
3 The polynomial volumetric method notation for effective stress, effective distance resolves the
numerical derivation errors which arise due to truncation and round off errors.
4 The polynomial notation significantly prepares more complete physical ideas about the average stress
and role of stress gradient in weight function and effective stress computations.
5 The volumetric weight function displays interactive response as a function of distance along notch
roots and it has not been restricted to the notch ahead like Yao’s weight function [Yao Weixing, Xia
Kaiquan, Gu Yi, On the fatigue notch factor kf, International Journal of Fatigue, 17(4), 245-
251,1995.].
6 The polynomial weight function completely satisfies smooth specimen consideration fact, whereas,
the Yao’s weight function does not exactly fulfill this requirement [Yao Weixing, Xia Kaiquan, Gu
Yi, On the fatigue notch factor kf, International Journal of Fatigue, 17(4), 245-251,1995.].
7 The proposal of Stress Peak Trajectory Path “SPTP” as one useful tool to avoid mesh density variation
along notch roots and corresponding linear behavior in logarithmic diagrams.
8 The proposal of effective distance-effective stress curve and effective distance-applied loading
magnitude curve based on Effective Stress Trajectory Path “ESTP” concept for evaluating of effective
distance and effective stress in parabolic shape graphs.

References
1. G. Pluvinage, Fracture and Fatigue Emanating from stress concentrators, Dordrecht: Kluwer
Academic Publishers, 2003.
2. A. Buch, Analytical approach to size and notch effects in fatigue of material specimens,
Material Science and Engineering, Vol. 15, 75-85, 1974.
3. A. Brandt, Calcul des pièces à la fatigue par la méthode du gradient, Editor CETIM, 1980.
4. G. Qylafku, Z. Azari, N. Kadi, M. Gjonaj, G. Pluvinage, Application of a new model
proposal for fatigue life prediction on the notches and key-seats, International Journal of
Fatigue,21, 753-760, 1999.
5. H. Adib, G. Pluvinage, Theoretical and numerical aspects of volumetric approach for fatigue
life prediction in notched components, International Journal of Fatigue, 25(1), 67-76, 2003.
6. Yao Weixing, Xia Kaiquan, Gu Yi, On the fatigue notch factor kf, International Journal of
Fatigue, 17(4), 245-251,1995.
 37. Mesofracture and transferability 1269

LOCAL APPROACH USE AT SOLUTION OF FRACTURE PARAMETERS

TRANSFERABILITY

L. Jurasek, M. Holzmann and I. Dlouhy

Institute of Applied Mechanics, Veveí 95, 611 00 Brno
Institute of Physics of Materials ASCR, Žižkova 22, 61662 Brno, Czech Republic,
idlouhy@ipm.cz

A number of works have been published demonstrating that the pre-cracked Charpy (PC)
specimens can be, after meeting the validity conditions and/or adjusting the raw data, successfully
used for specification of fracture toughness characteristics corresponding to standard (1T)
specimens. The procedures showing how to adjust the fracture toughness data determined from PC
specimens for fracture resistance characterisation of full scale component include the probabilistic
methods, toughness scaling models, introduction of other methods for constraint loss correction,
master curve methodology etc. There are boundary cases when no direct fracture toughness
characteristics adjustment is necessary and only the validity conditions have to be carefully
checked. Typically this is the case of pearlitic steel EN7 (commonly used for wheelset production)
having Lüders plateau and higher strength properties. In order to explain the deformation and
fracture processes 3D FE simulation of the PC specimen quasistaticaly loaded in three point
bending has been performed.
The aim of the paper can be seen in explanation of the local behaviour at the crack tip of PC
specimens being affected by Lüders flow. Relation of the measured and calculated characteristics
of the material behaviour should be analysed giving corresponding explanation.
The standard EN7 steel cut from exactly defined locations of railway wheelsets was used for
the investigation. Deformation and strength properties and true stress/true strain curves were
generated from standard tensile tests. Standard fracture toughness values were determined by using
CT specimens, as prescribed by EN 13262. PC specimens were tested at room temperature and at 0
°C. The results are summarized in rank probability diagrams shown in Fig. 1.

FIGURE 1. Rank probability diagram a) left: PC specimens at 0 and 20 °C,

b) right: CT specimens at 21.5 °C
1270 L. Jurasek et al.

Two types of fracture behaviour have been observed: Several specimens having fracture force Ffr
at the point of instable cleavage fracture less than net section yielding force Flim and the other part
of specimens with the fracture force at onset of cleavage greater than Flim. For specimen with Ffr <
Flim the fracture toughness was calculated using Linear-Elastic Fracture Mechanics but the initial
crack length was adjusted for crack tip plasticity effects (Irwins approach, Kec fracture toughness
values). For specimens with Ffr > Flim elastic plastic fracture toughness KJc was calculated from Jc
evaluated at the point of cleavage.
There is a gap in fracture toughness data of PC specimens between the value of 60 and 90
MPam1/2 (Fig. 1a). The phenomenon that is caused by great increase of deflection, which occurred
due to plastic hinge in net section of “small” PC specimens loaded by bending to the point when
the applied load attains closely Flim forces. At the same, time this phenomenon is also connected to
several times steeper increase of crack driving force J; following the Flim force small increment of
applied force F above Flim causes great increase of J.
3D FEM simulation have shown that the distribution of maximum principal stress 1 ahead of
the crack tip in PC specimen loaded into fully plastic regime (J = 72.4 kN/m) have been identical
with that of reference stress-strain field in plane-strain SSY condition (Fig 2a). There is no
constraint loss. It is believed that this deformation behaviour of PC specimen is related to the real
deformation properties of steel EN7 used as input into numerical study (true stress/true strain curve
with Lüders flow). Fig. 2b shows the plot of local peak, 1max, of maximum principal stress near the
crack tip for PC specimen and the reference stress-strain SSY field as a function of applied J, for
two radii (R) simulating the crack tip in numerical computation. It is again observed the agreement
of the magnitude of 1max for the two stress-strain states.

FIGURE 2. Stress distribution below the crack tip (having radii R of 0.01) for PC specimens (a)
and development of the maximum stress with J values compared to SSY field (b).

It has been found that for steel having at onset of yielding the Lüders plateau the deformation of
PC specimen is different from those having continuous onset of plastic flow. Hence, when
evaluating the fracture toughness values for the purposes of fracture resistance characterisation
(e.g. for determination of reference temperature T0) this phenomenon has to be accounted for.
Authors gratefully acknowledge to Grant Agency of the Academy of Sciences (GAAV
200410502 and 1QS200410502) for their financial support.
 37. Mesofracture and transferability 1271

DAMAGE IN RUBBER-MODIFIED POLYMERS : EXPERIMENTAL,

MODELLING AND COMPUTATIONAL ASPECTS

Naïma Belayachi, Fahmi Zaïri, Noureddine Benseddiq and Moussa Naït Abdelaziz
Laboratoire de Mécanique de Lille (UMR CNRS 1441), Université de Lille 1
Polytech’Lille, Cité Scientifique, 59655 Villeneuve d’Ascq cedex
Moussa.nait-abdelaziz@polytech-lille.fr

Microstructural modification of materials is a well known technique which is used to enhance their
mechanical properties. Examples of such materials are polymer blends, polymer composites, metal
matrix composites, alloy systems and most of the natural materials, such as bone and wood. This
heterogeneous aspect acts as an important factor on their macroscopic mechanical behaviour
response. Relationships between the micro structural phenomena and the macroscopic deformation
behaviour are required when predicting macroscopic properties from the microstructure of a
material.
Homogenisation methods are one of the way which can be followed to obtain a quantitative
relation between the two scales. Here, the heterogeneous material is effectively replaced by an
equivalent homogeneous one that represents the real material in an averaged sense. The
fundamental assumption in these methods is the existence of a unit cell that is representative for
the microstructure of the material under consideration, the so called representative volume element
(RVE).
In this work, a numerical homogenisation method has been adopted to model the mechanical
behaviour of an heterogeneous solid material of which the microstructure typically consists of a
continuous matrix material (here polymetylemethacrylate) with distributed rubber particles. In
general, macroscopic quantities are formulated as averages of the corresponding microscopic state
variables. The average of a quantity is taken over the region occupied by the unit cell [1, 2].
Furthermore, attention will be focused on the modelling of the mechanical behaviour of solid
polymers.
To account for both the large strains, time and history dependent material behaviour of the
studied material, an hyper-elastic-viscoplastic description was adopted. The behaviour of rubber
particles is described by a Mooney-Rivlin energy density function as formulated by Peric et al [3] ,
while the perfectly viscoplastic behaviour of the matrix is described by the model of Perzyna [4].
The Mooney Rivlin strain energy density function is written as :

1
W c10 ( I 1  3)  c 01 ( I 2  3)  c1 1 ( I 1  3)( I 2  3)  ( J  1) 2 (1)
d
where I1and I2 are the invariants of the right Cauchy-Green deformation tensor C, J the Jacobian,
cij and d are material constants
The history and time dependent nature of polymethylemethacrylate has been taken into
account by Perzyna’s viscoplastic yield stress model [4] in which the viscoplastic strain rate is
expressed as:
1

vp §V ·m
D J ¨  1¸ (2)
©V0 ¹
1272 N. Belayachi et al.

In equation 1, m and J defines respectively the strain hardening and the viscosity parameters (i.e.
the sensitivity strain rate parameters). V is the material yield stress, and V 0 the static yield
4 1
stress of the material ( i.e. when strain rate does not exceed 10 s ).
In order to perform numerical simulations, the constitutive model has been implemented in a
finite element code. A cell containing a cavitated particle has been modelled and the influence of
surface tension has been analysed.
The method has been validated by comparing results of homogenised simulations with
experimental data. Figure 1 shows as an example, a comparison between experimental and finite
elements results. A good agreement is highlighted.

Figure1 : Influence of strain rate (a) and particle volume fraction (b) on the strain-stress
relationship of an RT-PMMA

References
1. R. Hill, (1963), J. Mech. Phys. solids, 11, p. 357, 1963
2. O. Van Der Sluis, P. J. Schreurs, H. E. H. Meijer, Mech. Mater. , 33, p. 499, 1999
3. D. Peric, D. R. J. Owen, M. E. Honnor, Comput. Methods Appl. Mech. Engrg. , 94, p.35,
1992
4. P.Perzyna, , plasticity today-modelling, methods and applications, A. Sawczuk (Eds) p. 657-
679, 1985
 37. Mesofracture and transferability 1273

FAILURE ASSESSMENT DIAGRAMS BASED ON THE CRITERION OF

AVERAGE STRESS

Y. G. Matvienko
Mechanical Engineering Research Institute of the Russian Academy of Sciences
4 M. Kharitonievsky Per., 101990 Moscow, Russia
matvienko7@yahoo.com

The cohesive zone model and the failure criterion averaging the local stress over the cohesive zone
ahead of the crack tip have been suggested for solids with a crack or a notch to develop the failure
assessment diagram approach.
A simple Dugdale-Barenblatt type cohesive zone model is employed to describe the
deformation and failure process ahead of the crack (or notch) tip. The fracture criterion has been
employed in the form of the average stress limitation in the cohesive zone ahead of the crack tip,
i.e. the normal elastic stress ahead of the crack tip is averaged over the cohesive zone length. The
stress distribution on the crack extension line has been described for a body with a finite crack by
exact elastic solution according to the well-known form. In contrast to a classical cohesive model,
the cohesive stress V coh is treated by Matvienko [1] according to von Mises yield criterion as a
property of both the material and the applied stress.
Failure curves have been obtained for an infinite plate with a finite crack under two simple
mode of loading, namely, an uniform remote tensile stress V and an uniform internal stress
applied to crack’s faces (Matvienko [2, 3]). The results clearly show that the failure curves are
dependent on mode of loading and stress conditions (plane strain or plane stress). It is seen that
opposite to the case of uniform remote loading the failure assessment curve decreases with the
decreases of normalized applied stress (Fig. 1). Moreover, the failure assessment diagram becomes
raised above the failure assessment diagram for a crack under remote uniform loading.
The cohesive model and the local failure criterion of the average stress have been successfully
adopted for a solid with a notch as well as a crack. In this case, the stress distribution at the notch
tip is simplified considerably on the continuation of the notch (e.g. Creager and Paris [4]). The
fracture criterion for a solid with a finite notch is presented as
2 2 1 / 2
§ V · ª § V coh · 1 º
K 1notch K 1C 1  ¨¨ ¸¸ «1  ¨ ¸ 2 »
© V coh ¹ «¬ © V ¹ K t ¼» (1)
The failure curve of a solid with a crack falls below the failure curve of a solid with a notch. The
difference between these failure curves increases with decreasing the elastic stress concentration
factors K t (Fig. 2).
The present approach has been spread to describe failure curves in terms of the CTOD and the
J-integral. When the requirements of the standard SSY state, or the J-dominant stress state are met,
the J-integral and the CTOD are equally valid crack tip characterizing parameters. When the crack
tip conditions are no longer characterized by the value of J, the J-CTOD relation becomes
dependent on the applied critical stress, i.e. the crack tip constraint effect appears.
1274 Y. G. Matvienko

FIGURE 1. Failure assessment diagrams for a crack under uniform remote and internal tensile
stresses (plane stress).

FIGURE 2. Failure assessment diagrams for specific elastic stress concentration factors K t (an
infinite plate with a notch under an uniform remote tensile stress).

References
1. Matvienko, Yu.G., Int. J. Fracture, vol. 98, L53-L58, 1999
2. Matvienko, Yu. G., Int. J. Fracture, vol. 124, 107-112, 2003
3. Matvienko, Yu. G., Int. J. Fracture, vol. 131, 309, 2005
4. Creager, M. and Paris, P.C., Int. J. Fracture, vol. 3, 247-252, 1967
 38. Damage in Composites 1275

MATERIAL MODELS FOR DAMAGED COMPOSITE LAMINATES

J. Varna
Dept of Applied Physics and Mechanical Engineering
Lulea University of Technology, SE 97187, Lulea, Sweden
Janis.Varna@ltu.se

In long fiber composites with unidirectional or multidirectional orientation of the reinforcement

several damage modes develop during the service life. The most typical and first damage mode is
related to intralaminar crack formation in off-axis layers of prepreg laminates or in off-axis
bundles in case of Non-Crimp Fabric (NCF) or woven composites. Accumulation of individual
fiber fractures in layers oriented parallel to the main loading direction is also typical. More unusual
are breaks of fiber bundles oriented in the loading direction observed recently in NCF composites.
Part of the bundle interface can be debonded from the rest of the composite. Some examples are
given in Fig.1. Common for these damage entities is that each damage mode is usually localized in
a certain phase (in a fiber, matrix, at the interface, inside a bundle of a given orientation, inside a
layer) and is characterized by a distinct orientation and size. This simplifies the damage
characterization and model development.

a) intralaminar cracks in cross-ply laminate

b) broken bundle and intralaminar cracks in NCF composite

FIGURE 1. Typical damage modes in long fiber composites.

In this paper a unified approach for thermo-elastic properties reduction due to the described
damage is presented. Exact closed form expressions are derived which contain constituent
properties, composite architecture parameters and density of damage entities belonging to
particular mode. Very robust local characteristics of the micro-damage – normalized average crack
face opening (COD) and crack sliding (CSD) displacements also enter these expressions. Their
dependence on constraining material stiffness and geometry is described by simple power laws
which are obtaining in result of numerical parametric analysis using FEM.
For example, stiffness matrix of the damaged composite may be expressed in the following
form
1276 J. Varna

U kn
>U @k >Q @k >H @k >T @Tk  >S 0 @RVE > @
§ 1 ·
>Q @RVE ¨ >I @  ¦
¨
© k E Tk
V k >Q @ >T @
k
T
k
¸ Q
¸
¹
0 RVE

(1)

Here > Q @ k is a layer stiffness matrix in symmetry axes, > T @k is the stress transformation matrix,
V k is the volume fraction of the constituent (bundle, layer, fibers, matrix etc.), matrixes with
index 0 represent undamaged composite compliance and stiffness, U kn is normalized crack
density in the damaged constituent, >U @ k is the matrix of normalized crack face displacements
described by power law, ETk is transverse modulus of the constituent.

The > H @ k -matrix defines the relationship between k-th element boundary averaged strains
and the strains applied to the RVE. In this work the relationships for > H @ k are developed based
on results of FEM parametric analysis. In the local system of coordinates coupled with the material
symmetry or with the local geometry the relationship is

ba
­° H L ½° ª H 11 º ­ HL ½°
» °
H 12 0
«
® HT ¾ « H 21 H 22 0
» ® HT ¾
°̄J LT °¿ k ¬« 0 0 » k °̄J
H 66 ¼ °
LT ¿ RVE (2)
The H-matrix dependence on geometrical and mechanical parameters of constituents has been
analyzed using FEM and dependence is described by power laws. In case of laminate the
constituent is a unidirectional layer with a certain orientation and the H-matrix is identity matrix.
The accuracy of the developed predictive tool is high and the model has been validated by
comparing with direct calculations using FEM as well as with numerous experimental data.

References
1. Joffe, R., Mattsson D. and J. Varna, Proceedings of CANCOM 2005 conference, Vancouver,
2005, 10p.
2. Highsmith, A.L. and Reifsnider, K.L., Damage in Compos. Mater., ASTM STP 775,
American Society for Testing and Materials, 1982, 103-117.
3. Varna, J. and Berglund, L.A., J. Compos. Technol. Res., 16, 1994, 77-87.
4. R. Joffe, A.Krasnikovs and J. Varna, Composites Science and Technology, 61(2001), 637-
656.
5. J.Varna, Proceedings of ECCM-10, Brugge, 2002, 7 p.
6. Lundmark P. and Varna J., International Journal of Damage Mechanics in press 2005.
 38. Damage in Composites 1277

RAMAN SPECTROSCOPY ASSESSMENT OF STIFFNESS REDUCTION AND

RESIDUAL STRAINS DUE TO MATRIX CRACKING IN ANGLE – PLY
LAMINATES

P. Lundmark1, D. G. Katerelos2, J. Varna1 and C. Galiotis2,3

1Department of Applied Physics and Mechanical Engineering, Lulea University of Technology,

SE 971 87, Lulea, Sweden

2Foundation Of Research and Technology Hellas/Institute of Chemical Engineering and High

Temperature Processes, Stadiou str, Platani, Patras, PO Box 1414, GR-265 04, Greece
3Materials Science Department, University of Patras, Rio University Campus, Patras,

GR–265 04, Greece

Janis.Varna@ltu.se

Off – axis ply cracking in composite laminates and especially its effect on composites thermo-
mechanical behaviour has been extensively studied in the case of cross – ply stacking sequence;
see Katerelos and Galiotis [1], Joffe and Varna [2], and Katerelos et al. [3]. During the last decade,
the extension of this research to more complex composite systems used in practice has been
addressed and is still under development; see Crocker et al. [4], Katerelos et al. [5], Lundmark et
al [6]. In the presented work a detailed analysis of the effects of evolved damage on the behaviour
of a [0/45]s laminate under uniaxial tension is presented. Laser Raman Spectroscopy has been
applied for the experimental measurements of the strain state variations due to damage within the
specimens. Residual strain and stiffness reduction were derived using experimental Raman data, as
a function of crack density increase and the results were compared with theoretical predictions
using the model presented in [6]. Indicative conclusions on viscoelastic phenomena affecting the
behaviour of angle – ply laminates were extracted.
A glass fibre reinforced epoxy system was used for experimental investigation of the damage
developed in a [0/45]s composite laminate. It was selected due to the special mechanical properties
that it exhibits in addition to the ability of manufacturing translucent specimens useful for the
Raman spectroscopy inspection. The use of an embedded Aramid fibre – sensor within the 0° ply
and near the 0°/45° interface was necessary due to the poor Raman signal for a glass. This
technique has been extensively applied in the case of cross – ply laminates. The derivation of
macroscopic quantities from Raman data is explained in [1] and [2].
One of the main problems arising when dealing with model laminates different than cross – ply
is that the common shear – lag analyses and variational models are not applicable, for example, if
the laminate is non – balanced (i.e. [0/45]s laminate) . The model used in this work does not have
this problem. It gives exact closed form expressions for all thermo-mechanical properties of a
general symmetric laminate with cracks in arbitrary layers. The properties are expressed in terms
of laminate lay-up, layer properties and crack density. Parameters characterizing the crack shape:
average crack face opening and crack face sliding displacements also enter the stiffness
expressions. For a given laminate configuration these robust parameters can be calculated using
simple power laws. The derived constitutive relationships can be used also to determine the
dimensional changes due to evolution of the damage state. For this task they are applied for a zero
mechanical load case. The residual strain development is caused by release of thermal tensile
stresses in the cracked layer due to crack development.
1278 P. Lundmark et al.

Residual strain and stiffness reduction as a function of crack density derived from
experimental Raman data as well as the theoretical model predictions are presented in the
following comparison charts of Fig. 1a and b.

(a) (b)
FIGURE 1 (a) Residual strain variation with increasing crack density. (b) Longitudinal modulus
reduction as a function of increasing crack density in 45-layer

The stiffness reduction is predicted with a high accuracy whereas the measured residual strains
are larger than predicted. The good agreement of elastic modulus reduction proves that the power
law expression for crack face displacements, which is the only approximation in the model, is
accurate. However, the same values of displacements used in residual strain do not fit the
experiment. We explain the difference by viscoelastic – viscoplastic behavior of the off axis layer
in shear which increase the “apparent” residual strain. The two curves in Fig 1.b correspond to two
possible boundary conditions in the load application area.

References
1. Katerelos, D.G., McCartney, L.N. and Galiotis, C., Acta Mater, accepted for publication
(DOI: 10.1016/j.actamat.2005.03.045).
2. Joffe, R. and Varna, J., Comp Sci Tech, vol. 59, 1641 – 1652, 1999.
3. Katerelos, D.G. and Galiotis, C., App Phys Lett, vol. 85, 3752 – 3754, 2004.
4. Crocker L.E., Ogin S.L., Smith P.A. and Hill P.S., Comp A: App Sci Man, vol. 28A, 839 –
846, 1997.
5. Katerelos D.G., Galiotis C., Ogin S.L. and Whattingham R.D., In Proceedings of the ECCM –
9 Composites – From Fundamentals to Exploitation, Proceedings on CD – ROM, 2000.
6. Lundmark P. and Varna J., Int Jrnl Dam Mech in press, 2005.
 38. Damage in Composites 1279

PHYSICAL MODELLING OF FAILURE PROCESSES IN COMPOSITE

MATERIALS

P. W. R. Beaumont
University of Cambridge
Department of Engineering,
Trumpington Street, Cambridge CB2 1PZ, United Kingdom
pwrb@eng.cam.ac.uk

Despite the acquisition of vast collections of mechanical property data, our ability to predict the
structural integrity of a damaged composite subjected to mechanical and thermal stresses and
hostile environment remains restricted. This is because our understanding of problems of
composite failure is based almost entirely on this store of information being empirical in nature
with limited knowledge of the structural changes taking place in the material over time.
Lack of mastery in combining the design of the architectural features of composite materials at
the atomistic and micron level of size, and elements of the engineering structure metres in length,
has led to the opening of a gap in our knowledge of composite failure. This weakness can be
traced to the changing nature of fracture as size increases from the sub-micron level of structure to
the metre level of component failure. This size (or length) scale, which spans several orders of
magnitude, provides a framework for understanding the failure characteristics of material on the
one hand and performance limitation of the component on the other (Fig. 1). Coming to terms with
these differences in behaviour as size changes appears to be a key source of design difficulty
because at some point on this lengthscale the material problem becomes a structural one.
Difficulty arises, however, when conditions become stringent, so that more properties are
involved in the design process. We need constitutive equations for design that include all of the
design variables including the material properties that in turn depend upon the geometry of the
laminate and microstructure of the individual ply. A description of the material's response to a new
set of circumstances requires a completely new set of experiments from which new continuum
rules must be distilled.
Modelling techniques are described, which quantify the accumulation of damage in the
composite material in terms of the important structural features of the composite material, applied
stress, time, temperature, and environment. Particular emphasis is placed on the internal state
variable method. This is to take the path of “physical modelling”.
Examples of physical models will be given. One case study is of fatigue cracking in carbon
fibre-strengthened polymers. The physical model is based on the coupling mechanisms of splitting
and delamination, microscopic processes controlled essentially by the properties of the matrix
material. Another physical model is based on a micro-mechanical theory of stress-corrosion
cracking in glass fibre-strengthened polymers. In this case study, the time-dependence of
composite strength is explained in terms of the structural features of the composite, applied stress
and environment. This physical model is based on stable crack growth by fibre fracture and matrix
bridging in the crack wake.
1280 P. W. R. Beaumont

FIGURE 1 Hierarchy of structural scales ranging from the micron to the metre (and greater) level
of size, from the single ply to the final structure, and discrete methods of analysis ranging from
micro-mechanics to the continuum levels of modelling.
 38. Damage in Composites 1281

NCF CROSS-PLY LAMINATES: DAMAGE ACCUMULATION AND

DEGRADATION OF ELASTIC PROPERTIES

R. Joffe and D. Mattsson

Division of Polymer Engineering
Lulea University of Technology
SE-971 87 Lulea, Sweden
Roberts.Joffe@ltu.se, David.Mattsson@ltu.se

Traditionally high performance composite parts (for instance structural parts used in aerospace
applications) are made by lamination of prepregs. However, multi-axial warp knitted fabric (non-
crimp fabric (NCF)) proved to be a good alternative. NCF composites combine fairly good
mechanical properties with low production costs, high deposition rates and unlimited shelf life.
Additionally to these benefits, NCF composites have also been reported to show increasing out-of
plane fracture toughness and damage tolerance [1,2]. NCF composites are manufactured from
preforms with multiple layers of straight fiber bundles with different orientations stitched together
by a warp knitting procedure [3] (see Figure 1a). The main difference between prepreg-tape based
laminates and NCF composites is that the former has an internal structure consisting of continuous
fibers that are rather homogeneously dispersed; whereas NCF composite has continuous fibers
combined in fiber bundles with well defined geometry (see Figure 1b).

(a) (b)
FIGURE 1. NCF: (a) schematic picture (from Vectroply) and (b) micrograph.

In this study, the response of Non-crimp fabric cross-ply composites to tensile loading is
investigated. Tensile tests revealed that carbon fiber based NCF cross-ply laminates with different
lay-up demonstrated large difference in damage tolerance. It was found that laminates of type A
([0/90/0/90]S) showed 40% reduction of Young’s modulus before final failure, whereas a laminate
of type B ([90/0/90/0]S) showed a reduction in modulus of only 5%. These results are presented in
Figure 2.
1282 R. Joffe and D. Mattsson

FIGURE 2. Degradation of elastic modulus as a function of applied strain for carbon fiber based
NCF cross-ply laminates with different lay-ups.

Analysis of micrographs showed that damage in form of broken and delaminated bundles
oriented in the loading direction were present in the laminates of type A, whereas laminates of type
B only contained cracks in layers transverse to applied load. Numerical modeling (FE calculations)
and ply discount model predicted results within the experimentally measured decrease of modulus
(40%) due to a combined effect of both failure and delamination of 0°-bundles and transverse
cracks in the 90°-bundles. Results presented in Figure 3 show normalized elastic modulus
(normalized with respect to the modulus of undamaged laminate) as a function of extent of damage
in 0°-layer (number of failed bundles).

FIGURE 3. Modulus reduction in a cross-ply NCF composite (due to failure and delamination of
0°-bundles) as a function of number of failed bundles.

References
1. Dransfield, K., Jain, L. and Mai, Y-W., Compos Sci & Techno, vol. 58, 815-827, 1998
2. Jain, L., Dransfield, K. and Mai, Y-W., Compos Sci & Techno, vol. 58, 829-837, 1998
3. Steggall, P., In Proceedings 31 st International SAMPE Technical Conference, October 26-
30, 1999, Chicago, IL, USA
 38. Damage in Composites 1283

MATRIX CRACK INITIATION AND PROPAGATION IN LAMINATES WITH

OFF-AXIS PLIES

N. Vrellos, S. L. Ogin and P. A. Smith

School of Engineering, Guildford, Surrey, GU2 7XH
n.vrellos@surrey.ac.uk, s.ogin@surrey.ac.uk, p.smith@surrey.ac.uk

Matrix cracking damage in composite laminates is a generic type of damage, common to carbon,
glass and aramid fibre composites fabricated using unidirectional plies, as well as more complex
fibre architectures (e.g. textile reinforced composites). For such a common type of damage, it is
perhaps surprising that currently proposed failure criteria are unable to predict matrix crack
development, especially for off-axis plies. Previous work on GFRP laminates by Crocker et al. [1]
suggested that matrix crack propagation in off-axis plies is governed by the transverse normal
stress. The overall aim of the present work is to provide validated constitutive relations for crack
accumulation in off-axis plies of CFRP under mixed mode loading. In this paper, results on crack
initiation and crack propagation are presented for a range of multi-layer angle-ply laminates.
A number of laminate configurations have been tested, providing both different off-axis ply
angles and different off-axis ply thicknesses. The laminates were (a) four unbalanced laminates of
the type (02/4)s, where is 45o, 60o, 75o and 90o; (b) a (0/90)4s laminate, which also provides
(±45)4s coupons; and (c) quasi-isotropic laminates with configurations (0/90/r45)s, (02/902/r452)s
and (02/r602)s. For the (02/4)s specimens with off-axis plies at angles of 45o, 60o or 75o, oblique
end-tabs have been used to accommodate extension-shear coupling at the ends of the specimen.
The angle for the end-tabs was calculated using a modification of a procedure developed initially
by Sun et al. [2] in work on unidirectional composites. Coupons have been tested both with
polished edges and with machined-in defects. The edges of all coupons were polished to a 1 µm
finish using a Struers Pedemax-2 grinding/polishing machine. For the crack initiation
experiments, coupons were tested with polished edges, whereas crack propagation experiments,
which have been carried out only on the (02/4)s coupons, required the off-axis central plies to be
notched. A 0.8 mm drill bit was used to drill holes, 3 mm deep, parallel to the direction of the
fibres in the (off-axis) centre plies, without damaging the outer 0o plies. The tests were carried out
to progressively higher strains by loading the coupons to a specific strain level, unloading, and
then re-loading to a higher strain. Edge microscopy and X-radiography was used to monitor the
crack accumulation during the strain increments.
An example of crack development in a (02/454)s coupon containing notches is shown in Fig.1.
The figure show six notches drilled into each edge of the 45-ply; matrix cracks can be seen which
have developed from the notches after an applied strain of 1%. Data were obtained for crack
density as a function of applied stress/strain and the reduction in laminate properties (Young’s
modulus, Poisson’s ratio). Fig. 2 shows results for both unnotched and notched (02/4)s coupons.
The transverse normal stress 2 value corresponding to initial cracking from notches is similar for
the (02/904)s, (02/754)s and (02/604)s samples. Similarly, the stress for crack development in
unnotched coupons is similar for these three laminate configurations, and about 20 MPa higher
than for crack propagation from notches. The behaviour of the (02/454)s laminates is somewhat
different, since there appears to be a significantly reduced value of 2, suggests that there is an
interaction between the transverse normal and shear stress components in this case.
1284 N. Vrellos et al.

FIGURE 1. Dye-penetrant X-radiograph of a notched (02/454)s coupon subjected to a strain of 1.0

%, showing cracks which have developed from the notches

FIGURE 2. Transverse normal stress as a function of shear stress for crack development in
notched ( ) and unnotched ( ) coupons

Data from the other multi-layer angle-ply laminates show consistent behaviour with the data
from the (02/4)s laminates when the effects of ply thickness and neighbouring ply interactions are
considered.

References
1. Crocker, L.E., Ogin, S.L., Smith, P.A. and Hill P.S., J Compos Mater, 28A, 839-846, 1997
2. Sun, C.T. and Chung, I., J Compos Mater, 24, 619-623, 1993.
 38. Damage in Composites 1285

STRESS OSCILLATION AND INSTABILITY OF YIELDING IN POLYMERS

AND NANOCOMPOSITES

D. E. Mouzakis, G. Kandilioti, S. Tzavalas and V. Gregoriou

Foundation for Research and Technology-Hellas, Institute for Chemical Engineering
and High Temperature Chemical Processes, (FORTH-ICE-HT)
Patras, GR-26504, Hellas. Tel : +30 2610 965205, Fax: +30 2610 965223
mouzakis@iceht.forth.gr

Stress Oscillation (SO) is a yielding related-phenomenon known for polymers since the 1970’s.
Some polymers exhibit periodic fluctuations of yield stress in the post yield phase of neck
propagation. These periodic fluctuations correspond to the formation of opaque zones in the
specimen necking area. The overall result is a striation pattern of alternating opaque and
transpanent zones perpendicular to the drawing direction. SO has been confirmed for several types
of polymers like amorphous copolyesters and poly(ethylene terephtalate) (PET)[1], high density
polyethylene [2] and syndiotactic polypropylene (sPP) [3]. It has been also shown that it can
appear at both semi-static [1] and impact loading conditions [2]. The appearance of this
phenomenon and initiation mechanisms are still somewhat vague. Though, it has been successfully
shown that the result of the phenomenon, are local changes in the polymer crystallinity and
conformational order [3]. The purpose of our study was to study this phenomenon in
nanocomposites based on sPP and a nanoclay, namely montmorrilonite. The phenomenon
appeared in several compositions of the nanocomposites in montmorrilonite when the strain rate of
tensile loading was changed. Structural changes in the SO zones of the nanocomposites samples
were studied by means of scanning electron microscopy. Fourier transformed infrared
spectroscopy (FT-IR) was employed in order to determine the exact changes in the
nanocomposites microstructure and matrix conformational order.
Nanocomposite thin films (200 m) films were manufactured by melt mixing of an organically
modified montmorillonite (Cloisite®20A) with the sPP matrix and subsequent hot pressing and
quenching at 0 oC. sPP nanocomposites with several compositions of montmorillonite, 0,1,3,5,7
and 10wt% were manufactured in this way. Tensile testing was performed on a H20K-W
Hounsefield frame. Tests were run at different crosshead speeds varying from 5 to 40 mm/min, in
order to determine the SO envelope of the nanocomposites. Specimens were cut by scissors in
dumbbell shape. Finally, Fourier Transform Infrared Spectroscopy was performed with a Nicolet
850 FT-IR spectrometer equipped with an MCT/A detector. 128 scans were accumulated for each
spectrum at 2 cm-1 spectral resolution.
Figure 1 shows the Force-Displacement (F-X) curve of a sPP matrix specimen. SO was
obtained at two different crosshead speed transitions 5 to 40 mm/min and 40 to 30 mm/min
respectively. SO transition A was unstable whereas SO transition B held up to specimen rupture.
So, it should be concluded here that the SO phenomenon is indeed dependent on the loading
conditions and its stability is a function of the testing speed.
1286 D. E. Mouzakis et al.

FIGURE 1. Stress Oscillations in the sPP MATRIX. A:5 to 40 mm/min and B:40 to 30 mm/min
crosshead speed transitions respectively

FIGURE 2. SEM micrographs showing the stress oscillated surface (A) and fracture surface (B) of
a pure s-PP specimen

As seen in Figure 2 SEM scans, the SO phenomenon produced a surface undulation on the sPP
matrix seen in frame A on the left. Also, different fracture morpfologies are seen for the SO
exterior and interior of the same specimen in frame B resulting in a skin-core structure of the SO
zone. FT-IR scans have showed [3] a significant difference in the sPP matrix at the SO area: the
helical form I is transformed to the metastable trans-planar from III marking a crystal crystal
transformation for the SO bands.

References
1. Karger-Kocsis, J. Benevolenski, O.I. Moskala, E.J., J Mater. Sci., vol 36, 3365 – 3371, 2001
2. Mouzakis, D. E. Karger-Kocsis, J., J App. Polym. Sci., vol. 68, 561–569, 1998
3. Gregoriou, V.G., Kandilioti, G. and Gatos, K.G., Vibr, Spectroscopy, vol 34, 47–53, 2004
 38. Damage in Composites 1287

PREDICTION OF CYCLIC DURABILITY OF WOVEN COMPOSITE

LAMINATES

V. Tamuzs and K. Reifsnider

Istitute of Polymer Mechanics
Pratt&Whitney Chair for Reliability and Design
Aizkraukles 23 St, Riga LV-1006, Latvia, 244 Weaver Road Unit 5233, Storrs,
CT 06269-5233, USA
tamuzs@pmi.lv, reifsnid@engr.uconn.edu

It is well known that nonhomogeneous materials (such as composites), under cyclic loadings,
change their properties because of damage accumulation.
The classical approach to the lifetime prediction based on the damage concept involves an
assumption that, at the end of durability, the damage level reaches a critical value. The term
"damage" actually is not physically defined and obtains a physical meaning only through the
parameters measured during a test. More often, the damage is identified with the density and
orientation of microcracks or with the change in the stiffness or damping properties of a material
[1]. For a woven composite, under a cyclic loading in the principal direction of orthotropy, the
lifetime prediction by monitoring the damage kinetics through the change in compliance and the
linearly increasing selfheating temperature was verified experimentally a long time ago [2].
In the present investigation [3] the behavior of a wowen orthotropic composite laminate in
fatigue off-axis loadings was studied.
In our program, the monitoring of changes in the modulus and hysteresis loop was carried out
in all tests. It was found that the final critical damage level before failure, measured as a reduction
in the cyclic modulus, did not depend on the durability and was invariant with respect to the
loading direction too. Such an invariance of reduction in the final modulus allows one to predict
the cyclic durability of materials with a reasonable accuracy for any testing angle.
A satin-weave glass fiber-epoxy matrix composite was tested. A fiberite 934 epoxy was used
for the matrix. A four-harness satin woven glass-fiber cloth was used for the reinforcement. The
panels were cured for 2h at 180qC. The plate consisted of six plies — four plies in the 0º direction
and two in the 90º direction. The thickness of the plate was 0.58-0.63 mm. The samples were cut
out from plates in directions 0º, 15º, 30º, 45º, and 90º to the material symmetry axes. The width of
the specimens was 18-20 mm, and the length between tabs was 100-120 mm.
The tests were carried out on a MTS test rig. The strain was measured by an attached strain clip
gage, and for data collection and processing of measurements, a Spider 8 device having eight
channels and a maximal frequency of 9600 records per second was used.
In static loading, at least two samples were tested for each direction. The V – H curves obtained
were shown on Fig. 1, and revealed an essential nonlinearity.
The cyclic loading was performed at a constant stress level. The loading frequency was 10 Hz,
with an R-ratio of r = 0.1. The strain and stress values were registered at 120 points during one
cycle, repeating these measurements 15-20 times during the experiment. These measurements
allowed us to obtain the change in the cyclic modulus during the test.
In Fig. 2 the S–N curves for all tested directions are shown. The experimental values are
marked by dots; the continuous straight lines are approximations by the leastsquares method. The
dotted lines are results of prediction, which is discussed below.
1288 V. Tamuzs and K. Reifsnider

The mean damage rate <D> during the lifetime is defined as the final (registered in the last
measurement) change in the modulus E divided by the corresponding number of cycles to failure
N:  D !  'E N , which is a function of durability, the stress applied, and the loading
direction.

FIGURE 1. The stress-strain curves for FIGURE 2. S-N curves for all the directions
specimens cut in five directions tested

FIGURE 3. Mean damage rate as a function FIGURE 4. Changes in the elastic modulus
of durability for all test angles before failure for all test angles and
durabilities
It is remarkable that the graph of the mean damage rate as a function of durability does not
depend upon the angle of specimen orientation to the symmetry axis (Fig. 3). Independence of this
curve on the angle T means that the final reduction of modulus 'E does not depend on the
orientation of specimen and the final reduction of modulus does not depend on the durability N.
In Fig. 4, the measured final change in the modulus as a function of durability for all test angles
is shown in logarithmic coordinates. Having at our disposal the damage accumulation rate as a
function of applied stress and the invariant isotropic value of the final damage level expressed in
terms of the final reduction in the modulus, the cyclic durability of the woven composite
investigated can be calculated.
The predicted S-N curves are shown in Fig. 2 by dotted lines for all test angles.
In the second part of paper the results for other loading asymmetries are reported.

References
1. Kuksenko, V., Tamuzh, V., Martinus Nijhoff Publ., 1981.
2. Oldyrev, P., Tamuzh, V., Polymer Mechanics, vol. 3(5), 571-576, 1967.
3. Tamuzs, V., Dzelzitis, K., Reifsnider, K., Applied Composite Materials vol. 1(5), 259-293,
2004.
 39. Aging Aerostructures 1289

REPAIR OF CORRODED AEROSPACE ALUMINIUM PANELS USING

ULTRASONIC IMPACT TREATMENT

C. A. Rodopoulos1, S. Pantelakis2, M. Liao3 and E. Statnikov4

1Structural Materials and Integrity Research Centre, Materials and Engineering Research Institute,

Sheffield Hallam University, City Campus, Howard Street, S1 1WB, Sheffield, UK,
2Laboratory of Technology and Strength of Materials, Department of Mechanical Engineering &

Aeronautics, University of Patras, Patras 26 500, Greece.

3Structures, Materials and Propulsion Laboratory, Institute for Aerospace Research, National

Research Council Canada, M14, 1200 Montreal Road, Ottawa, ON. K1A 0R6, CANADA.
4Applied Ultrasonics, P.O. Box 100422, Birmingham, Al. 35210, USA.

C.Rodopoulos@shu.ac.uk, pantelak@mech.upatras.gr, Min.Liao@nrc-cnrc.gc.ca,

estatnikov@appliedultrasonics.com

The practise of identifying corrosion damage, air-blasting to remove loose material, grinding to
remove corrosion, shot peening to increase the fatigue properties and finally back assembly onto
the aircraft for many years has governed aircraft operators under the term "find and fix". This
empirical practise inevitably causes thickness and hence load bearing reduction while can locally
overstress components. To avoid that a rule of thumb is implemented making sure that no more
than 10% of the initial thickness of the material has been lost. Otherwise, the component should be
replaced. It is not difficult to understand that such practise primarily lies within the expertise of the
person performing the repair as well as the geometric complexity of the damage. If the corrosion
damage is situated at a location which is difficult to grind or, and to peen then most certainly the
quality of the repair will vary. The problem can have significant implication under cyclic loading
considering that the residual stresses induced by shot peening will undergo continuous
redistribution as part of their relaxation process, while irregular thicknesses can transform the
design philosophy used for that component from safe life to fatigue damage tolerance. Herein, such
repairs can start fatigue cracks following local stress raisers, due to stiffness loss especially when
close to stiffeners or by redistributing shear strains when close to joints. The Ultrasonic Impact
Treatments process is employing continuous ultrasonic vibrations at the ultrasonic transducer
output end strengthened with hard materials (carbide-containing alloys, artificial diamonds etc.)
and being in direct and generally continuous contact with the treated surface. During impact the
near surface of the material experiences high strain rates as well as heating. The first is responsible
for plastically deforming the material. The induced compressive residual stresses are significantly
more stable than those induced by shot peening due to the fact that high rate straining of
aluminium alloys generates energetically stable dislocation cells. Heating, on the other hand is
responsible for reimbursing the loose material by melting the corrosion oxides. Hence, the final
repair does not produce thickness reductions.
The table below shows the results from exposing aluminium alloy 2024-T351 under ASTM
exfoliation conditions for 36,48 and 72 hours prior to static testing. The results indicate that after
exfoliation, all the measured mechanical properties experience degradation with time. Yet, after
UIT the yield and ultimate tensile strength showed significant value recovery while in the case of
the 36h registered values higher than the original uncorroded material. Microhardenss
measurements, prior and after UIT, exhibited that the process can fully recover the corroded layer.
1290 C. A. Rodopoulos et al.

TABLE 1. Experimental Results

Material: Al 2024 T351
Direction/Number of Corrosion 0.2% Ultimate Elongation to Energy
specimens exposure prior Yield tensile failure density
to test strength strength A50 [%] W [MJ/
Rp0.2 Rm [MPa] m3]
[MPa]
L2 None 376.12 488.90 15.35 71.90
LT2 None 349.80 484.50 15.20 70.30
L2 Exfoliation 318.30 392.65 6.34 24.60
corrosion 36h
LT2 Exfoliation 300.15 369.15 3.95 14.70
corrosion 36h
L1 EXCO 36h + 407.30 479.00 6.85 29.00
Ultrasonic
treatment
LT1 EXCO 36h + 413.70 462.60 4.65 18.65
Ultrasonic
treatment
L2 Exfoliation 330.80 403.50 5.95 23.75
corrosion 48h
LT2 Exfoliation 314.90 375.25 2.90 11.30
corrosion 48h
L1 EXCO 48h + 391.75 459.65 5.10 28.40
Ultrasonic
treatment
LT1 EXCO 48h + 383.15 426.50 2.95 24.90
Ultrasonic
treatment
L2 Exfoliation 299.60 354.00 4.20 14.90
corrosion 72h
LT2 Exfoliation 313.05 372.80 2.80 10.90
corrosion 72h
L1 EXCO 72h + 362.50 433.45 4.40 16.40
Ultrasonic
treatment
LT1 EXCO 72h + 417.10 461.70 3.25 16.90
Ultrasonic
treatment
 39. Aging Aerostructures 1291

FATIGUE CRACK INITIATION IN STRESS CONCENTRATION AREAS

C. Schwob, F. Ronde-Oustau and L. Chambon

EADS CCR/ENSTIMAC, ENSTIMAC, EADS CCR
12 rue Pasteur, 92152 Suresnes Cedex, FRANCE
cyrille.schwob@eads.net

The fatigue life of structural components is determined by the sum of the elapsed cycles required
to initiate a fatigue crack and to propagate the crack up to the critical dimension. The relative
importance of these two phases depends on the level of loading. In the case of aeronautical
structure, the loads are supposed to be far enough below the elasticity limit to consider that most of
the fatigue life is expanded initiating the crack.
Prediction of fatigue crack initiation is commonly achieved through the use of criteria.
Although numerous criteria exist in the literature, few of them take into account stress-gradient
effect, due to their local formulations. As a consequence they do not predict the well known
difference which is observed in all metallic materials between traction and bending fatigue
experiments, and are therefore of limited use for the analysis of structures where stress
concentrations exist.
To alleviate this limitation the basic idea is to use a non local formulation of these criteria. This
is usually done either by introducing gradient dependent terms in their formulation (see
Papadopoulos and Panoskaltsis [1]) or by averaging their values over a given volume (see Fouvry
et al. [2], Banvillet et al.[3]). The latter approach has been retained in our model, since it appears to
be consistent with experimental observations (Vivensang and Gannier [4]): before specimen
failure, a damaged volume is observed where micro cracks develop. This fact is well illustrated by
Fig. 1 which represents an observation of a specimen subjected to cyclic loading. The observation
has been captured at 3.104 cycles before the specimen’s failure which occurred around 105 cycles.
It can be seen on this figure that fatigue damage occurs in a widespread area.

FIGURE 1. Damaged area on the edge of the central hole of the specimen (Cracks are highlighted
in red on the right)

Since these micro cracks are initiated by micro plasticity at the level of the grain, a mesoscopic
criterion has to be used in order to define this damaged volume:

Vd ^M , mesoscopic _ criterion( M ) t 0` (1)

To define this mesoscopic criterion, we propose to follow the developments of Papadopoulos
[5], who introduced a material constant which will be noted G. The new criterion is then defined
by the following equation:
1292 C. Schwob et al.

1
Vd V
³³³  Ta
2

 DPmax dM d E
(2)
d
where and are material parameters which has to be identified on fatigue tests. Although G might
have a physical meaning, we will identify it by using an additional fatigue test. This identification
has been carried out for an aeronautical steel - for which enough data were available in the
literature - and for an aluminium alloy for which a dedicated test campaign has been done.
The predictive capabilities of the criterion has been evaluated by computing Haigh’s diagram
for several kind of loadings and confronting the results with experimental data. The influence of
the choice of the tests chosen to identify the three parameters of the model has been studied, and
found to be reasonably limited.

References
1. Papadopoulos, I.V., Panoskaltsis, V.P., Multiaxial Fatigue and Design, Mechanical
Engineering Publications, London, UK, 1996.
2. Fouvry, S., Elleuch, K., Simeon, G., The Journal of Strain Analysis for Engineering Design,
vol. 6, 549-564,2002
3. Banvillet, A., Palin-Luc, T., Lasserre, S., International Journal of Fatigue, vol. 25, 755-769,
2003
4. Vivensang, M., Gannier, A., Recent Advances in Experimental Mechanics, Balkema,
Rotterdam, Netherlands, 1994.
5. Papadopoulos, I.V., High-Cycle Metal Fatigue, Springer, Berlin, Germany, 1999.
 39. Aging Aerostructures 1293

HYDROGEN TRAPPING: DEFORMATION AND HEAT TREATMENT

EFFECTS IN 2024 ALLOY

H. Kamoutsi, G. N. Haidemenopoulos, V. Bontozoglou, P. V. Petroyiannis and Sp.G. Pantelakis

Department of Mechanical & Industrial Engineering
University of Thessaly, Volos, Greece, hgreg@mie.uth.gr
Departmernt of Mechanical and Aeronautical Engineering
University of Patras, Rio, Greece.
hgreg@mie.uth.gr

Corrosion is a major concern to the structural integrity of aging aircraft structures. The effect of
corrosion on the damage tolerance ability of advanced aluminum alloys calls for consideration of
the problems associated with the combined effect of corrosion and embrittling mechanisms. In
recent work [1,2] the authors have shown evidence of corrosion-induced hydrogen embrittlement
in aluminium alloy 2024. Hydrogen is produced during the corrosion process and is being trapped
in distinct energy states, which correspond to different microstructural traps. These traps are
activated and liberate hydrogen at different temperatures. In alloy 2024, four traps T1 to T4 were
identified. Trap T1 is considered to be a reversible trap, which liberates hydrogen continuously at
low temperatures. Traps T2, T3 and T4 saturate with exposure time and are considered to be
irreversible with critical evolution temperatures of 200, 410 and 500oC respectively. The hydrogen
front advances with the corrosion front, so hydrogen penetrates in the material through the
intergranular paths generated by the corrosion process. Then hydrogen diffuses further in the
material establishing a hydrogen affected zone beneath the corrosion zone.

FIGURE 1. Total quantity of hydrogen for 2024 alloy in the solution treated and quenched
condition and in the aged condition. (in EXCO for 6 and 12 hours).

Removal of the corrosion layer (equal to the depth of attack) leads to complete restoration of
yield strength and partial restoration of ductility. Removal of the corrosion layer and heating above
the T4 activation range for hydrogen desorption (to activate all traps) leads not only to complete
restoration of strength but also to complete restoration of ductility.
Although some indications of the trapping states exists [3], the exact nature of trapping sites is
not known. The present work makes a contribution towards the identification of the
microstructural characteristics of trapping sites. Two series of experiments were performed. In the
first series the 2024 alloy was solution treated and quenched, in order to obtain a supersaturated
solid solution by dissolving the S phase. Immediatelly after quenching, the alloy was exposed in
1294 H. Kamoutsi et al.

the EXCO solution for 6 and 12 hours. The hydrogen trapped was measured by a thermal
desorption technique described in [3]. The results are shown in Fig.1. A significant decrease of
trapped hydrogen is documented for the solution treated material relative to the standard aged
material of the T351 temper. This result indicated that the T4 state is related to hydrogen trapping
by the S phase. In the second series of experiments the alloy was subjected to plastic deformation
prior to corrosion exposure in order to alter the dislocation structure of the material. Tensile tests
were interrupted at specified plastic strains. Then the specimens were exposed in the EXCO
solution for 24 hours and the hydrogen trapped was measured.

FIGURE 2. Total quantity of hydrogen versus plastic deformation.

The total quantity of hydrogen trapped is shown as a function of prior plastic strain in Fig.2.
The results indicate a decrease of trapped hydrogen at high plastic strains, indicating a potential
"dislocation shielding" of the S phase to hydrogen.

References
1. Kamoutsi, H., Haidemenopoulos, G.N., Bontozoglou, V., and Sp.G. Pantelakis, In
Proceedings of the 11th International Conference on Fracture (ICF11), Turin, Italy, 2005.
2. Kamoutsi, H., Haidemenopoulos, G.N., Bontozoglou, V., and Sp.G. Pantelakis, accepted for
publication in Corrosion Science, 2005
3. Charitidou, E., Papapolymerou, G., Haidemenopoulos, G.N., Hasiotis, N., and Bontozoglou,
V., Scripta Materialia, vol. 41, 1327-1332, 1999.
 39. Aging Aerostructures 1295

AN INTEGRATED METHODOLOGY ASSESSING THE AGING BEHAVIOUR

OF AIRCRAFT STRUCTURES

G. Labeas and I. Diamantakos

Laboratory of Technology and Strength of Materials,
Department of Mechanical Engineering and Aeronautics,
University of Patras, 26500, Rion, Greece
labeas@mech.upatras.gr
Institute of Structures and Advanced Materials
Patron-Athinon 57, 26441 Patras, Greece
diamond@istram.gr

The term ‘Ageing aircraft’ indicates an aircraft structure that is about to reach its original design
goal. At this stage, the light alloy structures used in commercial aircraft are susceptible to
Widespread Fatigue Damage (WFD) and possibly other deteriorating effects, such as corrosion
damage. One typical form of WFD is Multiple Site Damage (MSD), which refers to the
simultaneous existence of multiple interacting fatigue cracks at different sites of the same
structural component (Fig.1). In the presence of MSD, critical crack sizes are greatly reduced, thus
decreasing the residual strength of the structure below critical levels [1, 2] and leading to
catastrophic failures due to the sudden cohesion of such interacting cracks [e.g. 3].

FIGURE 1. Typical aircraft joint under MSD condition.

Assessment of the aging behaviour aircraft structures under MSD condition is a very
demanding task to fulfil, as it requires suitable tools for the estimation of all different phenomena
involved in MSD. Firstly, models for the prediction of crack initiation and the calculation of crack
propagation are necessary. In addition, link-up criteria for the prediction of the coalesce of
interacting cracks and the assessment of the residual strength of the structure are required. On the
other hand, the physical background of the WFD and MSD processes needs deeper understanding.
Furthermore, the computational effort for calculating MSD fatigue behaviour of an aircraft
structure is usually huge and depends on the complexity of the structure and the number of
interacting propagating fatigue cracks.
In the present work an integrated methodology for assessing the aging behaviour of aircraft
structures in the presence of MSD is presented. The methodology combines several modules for
the prediction of crack initiation, crack growth, crack link-up and residual strength of the
structures, as well as an effective technique for the necessary stress analysis. In order to keep the
amount of computing effort low, an incremental approach for calculating crack initiation and crack
propagation has been adopted.
1296 G. Labeas and I. Diamantakos

Estimation of crack initiation scenarios is based on a probabilistic approach, as the

phenomenon exhibits a quite stochastic nature. The number of fatigue loading cycles required for
the evolution of a new crack has been experimentally observed to follow a normal distribution, the
variables of which are depended on the local stress level at the initiation site. These variables and
their relationship with stress are calculated from simple experiments and are defined as a local
material property.
Coalesce of interacting cracks and residual strength of the structure is predicted using a strain
energy based criterion. This criterion is based on the quantity of the ‘specific’ strain energy
increase, i.e. the strain energy change (because of the ligament failure) divided by the ligament
area and has a similar physical background to the energy release rate (J-integral) concept. The
‘specific’ strain energy change, which is the critical quantity for ligament fracture is dependent on
both the plastic deformation of the ligament, as well as, on its fracture toughness.
For stress analysis and computation of Stress Intensity Factors a numerical model utilizing the
sub-structuring technique of the Finite Element Method is used, in order to reduce the computing
effort due to the large model size and the FE discretization difficulties appearing in crack
propagation calculations. The applied methodology is based on the development of suitable super-
elements used for the numerical analysis of aircraft structures susceptible to MSD.
All above modules are combined in an integrated methodology capable to assess the evolution
of MSD at aging aircraft structures from the stage of cracks development to the final failure of the
structure. Numerical results obtained by the proposed methodology are compared to experimental
results concerning typical aluminium joints used in aircraft structures under the presence of MSD,
and very good agreement is observed.

References
1. Schijve, J, Fat & Fract of Eng Mat & Struct, vol. 18, 329-344, 1995.
2. Smith, B.L., Saville, P.A., Mouak, A. and Myose, R.Y., J. of Aircraft, vol. 37, 325-331, 2000.
3. Hendricks, W.R., In Structural Integrity of Aging Airplanes, edited by Atluri, S.N., Sampath
S.G. and Tong, P., Springer-Verlag Berlin, Heidelberg, 1991, 153-165.
 39. Aging Aerostructures 1297

NUMERICAL INVESTIGATION ON THE TENSILE BEHAVIOUR OF PRE-

CORRODED 2024 ALUMINIUM ALLOY

P. V. Petroyiannis1, G. Labeas1, Sp. G. Pantelakis1, E. Kamoutsi2, V. Bontozoglou2

and G. N. Haidemenopoulos2
1Laboratory of Technology and Strength of Materials

Department of Mechanical Engineering & Aeronautics

University of Patras
Panepistimioupolis Rion, 26 500 Patras, Greece
2Laboratory of Materials

Department of Mechanical & Industrial Engineering

University of Thessaly
Pedion Areos, 38 334 Volos, Greece
3Transport Processes & Process Equipment Laboratory

Department of Mechanical & Industrial Engineering

University of Thessaly
Pedion Areos, 38 334 Volos, Greece
hgreg@mie.uth.gr

The synergetic effect of corrosion and corrosion induced hydrogen embrittlement damage
processes which occur at local scale has been found to result in a dramatic macroscopic tensile
ductility loss of the 2024 aluminum alloy [1,2].
To make a realistic simulation of the mechanical behaviour of the corroded 2024 alloy
manageable, a diligent consideration of the underlying corrosion and hydrogen embrittlement
mechanisms is required. From the microstructural viewpoint the corrosion attack of the alloy is not
uniform. The corrosion damage evolution and the development of a hydrogen diffusion zone
associated to it are extensively discussed in [2-4]. As shown in [3,4], corrosion in the 2024 alloy
starts in the form of pits mainly located at intersections of cracks in the protective surface oxide
layer. With increasing exposure time pits become deeper and connect through channels of
intergranular corrosion paths. This process leads to clustering and coalescence of pits. From that
point on, corrosion does not penetrate much deeper but instead spreads beneath the surface and
causes exfoliation of surface layers. The described process facilitates the transport of corrosion
solution deep into the material. Thus, the corrosive solution attacks the uncorroded material at the
front of the corrosion layer (e.g. at the bottom of pits or corrosion notches) and produces hydrogen.
The hydrogen generated at the front of the corrosion layer spreads to the adjacent unaffected
material creating a hydrogen diffusion zone below the corrosion zone. Recall that the described
damage processes occur in atomic scale.
To model these complex damage processes in atomic scale in order to estimate their effect on
the macroscopic tensile behaviour of the material, is extremely difficult. In the present work it is
proposed to simulate the tensile behaviour of the degraded specimen due to corrosion and
hydrogen embrittlement with the tensile behaviour of an uncorroded specimen involving
microscopic cracks of proper length.
In the present work, the tensile behaviour of corroded 2024 T3 tensile specimens has been
estimated on the basis of FE analysis by taking into account the local material properties in the
damaged areas. A parametric study is involved to account for the effect of thickness and maximum
depth of attack in the results.
1298 P. V. Petroyiannis et al.

To simulate the above mentioned behaviour of the pre-corroded specimen, uniform corroded
layers of 200 up to 350 m width evolving from the free specimen edges, as shown in Fig. 1(a), have
been assumed. Below the corroded layers lie uniform embrittled zones of 150m thickness and
below them undamaged material. Since the corroded zones are expected to have a much lower
tensile strength compared to the tensile strength of the undamaged material, they are expected to
fail already at low tensile loads. Taking the above into consideration, the initial model may be
replaced by a specimen out of undamaged 2024 T3 material with one edge crack or two symmetric
edge cracks, with the latter being the most severe case and hence representing a conservative
prediction, Fig. 1(b).
Calculated tensile properties obtained with the analysis agree well with experimental data. The
FE analysis results are supported by metallographic and fractographic analyses.

(a) (b)
FIGURE 1. (a)Tensile specimen configuration and (b) proposed model of the pre-corroded
specimen.

References
1. Pantelakis, Sp.G., Vassilas, N.I., and Daglaras, P.G., Metall., vol. 47, 135-141, 1993.
2. Petroyiannis, P.V., Kermanidis, Al.Th., Kamoutsi, H., Pantelakis, Sp.G., Bontozoglou, V.,
and Haidemenopoulos, G.N., Fatigue Fract. Engng. Mater. Struct., vol. 28, 565-574, 2005.
3. Kamoutsi, H., Haidemenopoulos, G.N.. Bontozoglou, V., and Pantelakis, Sp.G., accepted for
publication in Corrosion Science, 2005.
4. Kamoutsi, H., Doctoral Thesis, Dept. of Mechanical & Industrial Engineering, University of
Thessaly, Volos, Greece, 2004.
 40. Residual Stress and its Effects on Fatigue and Fracture 1299

ASSESSMENT OF DEFECTS UNDER COMBINED PRIMARY AND

RESIDUAL STRESSES

A. H. Sherry and M. R. Goldthorpe

Materials Performance Centre
University of Manchester
Manchester M60 1QD
United Kingdom
andrew.sherry@manchester.ac.uk

Residual stresses can provide a significant element of the crack driving force for defects in welded
components. Structural integrity assessment methods are available, such as the R6 defect
assessment procedure [1], which provide detailed guidance for the assessment of such defects
under the combined influence of primary and residual stresses. However, in some circumstances
these methods may be unduly conservative due, in part, to an over-estimation of the crack driving
force due to the residual stress, KJs. This over-estimation can lead to a pessimistic view of actual
safety margins for welded components and premature replacement or repair strategies.
This paper describes a programme of experimental and analytical work undertaken to better
characterise the influence of residual stress levels on KJs for fatigue cracks in a high strength, low
toughness aluminium alloy, AL2024-T351. A compact specimen has been designed in which a
highly tensile residual stress field is mechanically induced through a compressive preload prior to
fatigue pre-cracking. The residual stress field generated by this method was characterised using a
combination of surface strain mapping using Electronic Speckle Pattern Interferometry (ESPI),
and 3-D elastic-plastic finite element modelling, Figure 1.

FIGURE 1. Comparison between ESPI and 3D FEA surface strains in a preloaded notched
compact tension specimen
1300 A. H. Sherry and M. R. Goldthorpe

FIGURE 2. Comparison of crack opening displacement data with numerical data for simultaneous
(solid lines) and progressive (dashed lines) cracks

Following pre-cracking, a combination of metallographic examination and modelling has

provided an accurate characterisation of the crack-opening displacement close to the crack-tip at
the centre and surface of the specimen. This gives a set of reference data against which both
instantaneous and progressive approaches for introducing cracks into finite element models can be
critically assessed. As shown in Figure 2, for the fatigue pre-cracks introduced into the residual
stress fields, the progressive approach yields good agreement with experimental data, particularly
close to the crack tip.
Based on the results of this work, and using enhanced approaches for calculating the J-integral
for cracks located in regions of residual plastic strain [2], practical guidance is provided regarding
the suitability of alternative approaches for characterising KJs with greater accuracy. The
influence of these approaches on the reliability of defect assessments is discussed.

References
1. R6: Assessment of the integrity of structures containing defects, British Energy Generation
Limited, Revision 4, 2001.
2. D W Beardsmore and A H Sherry, “Allowance for residual stresses and material interfaces
when calculating J in and close to welded joints”, ASME Pressure Vessels and Piping, Vol.
464, pp. 11-21, 2003.
 40. Residual Stress and its Effects on Fatigue and Fracture 1301

EFFECT OF RESIDUAL STRESSES ON THE CRACK GROWTH IN

ALUMINUM

B. Kumar and J. E. Locke

National Institute of Aviation Research & Wichita State University
1845. N Fairmount St, Wichita KS 67260, USA
brijesh.kumar@wichita.edu
James.locke@wichita.edu

Shot peening is a cold working process in which the surface of a part is bombarded with small
spherical media called shots. This process results in a hemisphere of cold-worked material that is
highly stressed in compression. Fuchs [1] and several other researchers have reported that the
values of the compressive stresses are at least as high as 50% of the ultimate strength of the
material. In the open literature there is little information of crack growth data through the residual
stress fields generated by shot peening. Kocada et al [2] suggested that this is probably due to the
fact that the observation of short crack initiation and propagation in a heavily deformed surface
layer is very difficult to monitor. The focus of this investigation is on the fatigue crack growth
rates of short cracks in the shot peened aluminum alloys.
To assess the effects of residual stresses on fatigue crack growth, testing is being conducted on
specimens fabricated from the 7050-T7451 and 7075-T7351 aluminum alloys. The material was
obtained from Alcoa in the form of 0.25-inch thick sheets. To produce specimens with short
cracks several specimen geometries proposed by Suresh & Ritchie [3] were used. After several
specimens were tested we found that the double edge notched specimen geometry used by Everett
et al [4], was most suitable for the present investigation. The short crack specimens are then shot
peened at two locations. The edge containing the short crack is shot peened so that the crack is
embedded in the residual stress field. The specimens are also shot peened on the front and back
surfaces such that there is a zone on the surface of the specimens which is unpeened. Figure 1
show the region described.
Understanding the influence of compressive residual stresses and how it affects the fracture
mode is central to any investigation on how it may be utilized to improve fatigue resistance of the
material. Further the peening process can modify the topography of the fracture surface at the
crack initiation site. Turnbull et al [5] suggest that this is most likely caused by the distortion that
the grain structure experienced due to the surface deformation as a result of shot peening.
Therefore detailed fractographic analysis has also been undertaken. The results of fractography
showed there were numerous crack initiation sites, before the main fatigue crack formed. There
was evidence of ratcheting also observed on the specimens. Due to file size constraint’s we are
unable to submit photomicrographs of our findings.
1302 B. Kumar and J. E. Locke

Figure 1 & 2: Residual Stress Profile of the Shot Peened 7050–T7451 Aluminum with 100% and
200% coverage

The primary fatigue enhancing effect of any surface treatment is the residual compressive
stress field. Thus, quantifying and measuring the residual stress field is necessary to understand its
affect on crack growth. We are using the X-ray diffraction technique to determine the residual
stresses and results on Al 7050-T7451 with 100 % and 200% coverage’s are shown in figures 1
and Fig. 2. Currently we are analyzing the Al 7075T-7351 alloy.
We are currently testing the double edge notch coupons using the ACPD probes and an optical
microscope. Detailed results will be provided for the paper.

References
1. Kocada, D., Kocada, S., and Tomazek, H. “Description of Short Crack Growth in Shot-
Peened Medium Carbon Steel,” Fatigue & Fracture of Engineering Materials & Structures,
vol 21, pp 977-985, 1998.
2. Suresh, S., and Ritchie, R. O., “Propagation of Short Fatigue Cracks”, International Metals
Review, vol. 29, No.6, pp 445-476, 1984.
3. Everett Jr., R. A., Mathews, W.T., Prabhakaran, R., Newman, Jr., J. C., and Dubberly, M.J.,
“The Effects of Shot and Laser Peening on the Crack Growth and Fatigue Life in 2024
Aluminum Alloy and 4340 Steel” NASA/TM-2001-210843, ARL-TR-2363.
4. Turnbull, A., De Ls Rios, E. R., Tait, R. B., Laurant, C., and Boabaid, J.,S., “Improving the
Fatigue Resistance of Waspaloy by Shot Peening” Fatigue & Fracture of Engineering
Materials & Structures, vol 21:pp 1513-1524, 1998
 40. Residual Stress and its Effects on Fatigue and Fracture 1303

EFFECT OF THE CRYOGENIC WIRE BRUSHING ON THE SURFACE

INTEGRITY AND THE FATIGUE LIFE IMPROVEMENT OF THE AISI 304
STAINLESS STEEL GROUND COMPONENTS

N. B. Fredja, H. Sidhoma and C. Brahamb

a Laboratoire de Mecanique, Materiaux et procedes (LMMP ; LAB-STI-03). ESSTT, 5 Av. Taha

Hussein Montfleury 1008 Tunis-Tunisie.

b Laboratoire Microstructure et Mecanique des Materiaux (LM3 ; CNRS UMR 8006) ENSAM,

151 Boulevard de l’Hopital, 75013 Paris, France.

chedly.braham@paris.ensam.fr

In this investigation, ground surface integrity and fatigue behavior improvements of the AISI 304
SS resulting from the application of wire brushing at ambient and low temperatures were
investigated.
Quality of surfaces generated under cryogenic wire brushing
The formation of plastic induced martensite D’ is favored by the cold work hardening
generated by the cryogenic wire brushing. This can be clearly grasped from the X-ray diffraction
peaks. On the other hand the wire brushing carried out under dry condition induces an increase of
the interface temperature between the steel wires and the brushed surface (about 85°), which is
enough high to inhibit any austenitic plastic transformation to martensite. Wire brushing of the
AISI 304 SS surfaces induces superficial cold work hardening (depth less 100µm) with a level
which is as important as the brushing temperature is low. Brushing under low temperatures results
in a higher resistance of the work material to the plastic deformation. Therefore, the material
removal by the successive passes of the brush wires on the work surface is minimized under these
conditions. Consequently, the initially preexisting grinding groves are partially erased by the steel
wires and only a slight total roughness differences between the ground (Ra=2.2µm, Rt=16.5µm)
and the cryogenic brushed (Ra=1.95µm, Rt=11.5µm) states are noticed. However, brushing under
dry condition significantly deforms plastically the brushed surfaces and therefore, generates a
completely different finished surface morphology than the ground one.
Wire brushing redistributes the residual stresses fields generated by the grinding process and
put in compression the upper layers of the brushed surfaces (figure 1). This redistribution is
thought to be very beneficial to the fatigue life of mechanical components subjected to cyclic
loading.
Fatigue behavior of cryogenic wire brushed surfaces
Near surface cold work hardening associated to the compressive residual stress fields
generated under cryogenic wire brushing, improve substantially the resistance of ground
components to fatigue crack nucleation and propagation (figure 2). Moreover, the formed plastic
induced martensite by cryogenic wire brushing is found to contribute significantly to increase the
levels of the compressive residual stresses and therefore, delays the fatigue crack nucleation. On
the other hand, the formed plastic induced martensite by the cyclic loading at the tips of the
nucleated fatigue crack, contributes to slow down the crack propagation speed and to stop it in
some cases.
1304 N. B. Fredj et al.

Figure 1 Residual stress profiles measured using the incremental hole method

Figure 2 Fatigue life improvements by wire brushing under dry and cryogenic cooling of the AISI
304 SS ground surface

Substantial improvement of the fatigue life of the AISI 304 SS ground surfaces could be
realized by the application of cryogenic wire brushing (72%). This improvement is thought to be,
mainly, the consequence of the high level of the compressive residual stresses induced by both
clod work hardening at low temperature and volume change associated to the formation of the
plastic induced martensite. These stresses delay the nucleation of fatigue cracks. On the other hand
the formation of the plastic induced martensite at the tips of the fatigue cracks was observed to
block the propagation of the cracks and therefore, increases the propagation fatigue life.
 40. Residual Stress and its Effects on Fatigue and Fracture 1305

INTERACTION OF RESIDUAL STRESS WITH MECHANICAL LOADING IN

FERRITIC STEELS

A. Mirzaee-Sisan, C. E. Truman and D. J. Smith

Department of Mechanical Engineering
University of Bristol
Queen’s Building, University Walk
Bristol, BS8 1TR, UK
c.e.truman@bris.ac.uk

The paper will present recent results obtained as part of the EU FP5 project “Enpower –
Management of Nuclear Plant Operation by Optimised Weld Repair”1 concerning the effect of
residual stress load-history on fracture. This paper concentrates on the behaviour of A533B low
alloy ferritic steel. A companion paper concentrates on type 316H austenitic stainless steel. The
rationale of the research was to validate advanced numerical modelling techniques with an
experimental fracture programme. In order to achieve this, a laboratory test specimen was required
which contained a well-defined residual stress field. This would enable complementary numerical
and experimental fracture programmes to proceed in parallel. The paper therefore begins with a
description of the mechanism developed to generate residual stress fields in beam specimens. The
paper will then show numerical predictions of the residual stress state in the beam specimens,
details of the numerical fracture modelling, the experimental fracture test results, and will conclude
by discussing the comparison of experimental results and predictions.

Figure 1. Schematic diagram of in-plane loading. All diamensions are in mm

An in-plane compression procedure was developed to generate residual stress fields in A533B
ferritic steel single edge notch bend, SEN(B), specimens [1]. An in-plane compressive load was
applied to two ‘V’ notches at the two ends of the beam causing a localised zone of plasticity to
develop at the root of a shallow notch at mid-length. A residual stress field was subsequently
generated when the load was removed. A schematic diagram of the procedure, and specimen
dimensions is shown in Fig. 1.
ABAQUS was used to model the in-plane compression procedure using room temperature
material properties. A notch was then instantaneously introduced into specimens containing a
residual stress field generated by in-plane compression. The material properties were then set to

1.ENPOWER was co-sponsored by the Nuclear Fission Safety Programme of the European Commission
and the project partners: Institut de Soudure (co-ordinator, F), British Energy Generation Ltd (UK), Mitsui
Babcock Energy Limited (UK), Framatome ANP (D), Joint Research Centre (NL), University of Bristol
(UK) and Industeel Arcelor (formerly Usinor Industeel) under contract No FIKS-CT-2001-00167.
1306 A. Mirzaee-Sisan et al.

those at -150°C, and the specimens then loaded to failure in what was termed a Compress, Unload,
Cool, Facture (CUCF) cycle. Two complementary approaches were followed. A ‘global’ approach
used a modified J-integral routine to determine KJmod values for test specimens containing
combined primary loads and residual stresses. These KJmod values were then used in a Weibull
probability model to reflect the observed statistical scatter in the fracture load. In parallel, a ‘local’
approach [2] was also used to predict the cleavage fracture behaviour of the A533B SEN(B)s
following in-plane compression. Weibull parameters were obtained by fitting a Weibull
probability distribution to as-received experimental fracture data. The same parameters were then
used in a local model, along with a Weibull stress component determined by integrating the
maximum crack tip principal stress over the crack tip plastic zone.

Figure 2. Probability of failure of SEN(B) ferritic steel specimens in the as-received state and after
CUCF

Fracture tests were performed on 8 SEN(B) specimens in the as-received state, and 5 SEN(B)
specimens following the CUCF cycle. These results are shown in Fig. 2., along with predictions
based on the global approach, detailed above. Two predictions are shown. One assumed that the
total influence of the in-plane compression procedure could be encapsulated in the parameter Kres,
the magnitude of the equivalent stress intensity factor of the residual stress field without primary
load. The second set of predictions are based on the values of KJmod determined from the modified
J-integral routine.

References
1. Mirzaee-Sisan, A., Mahmoudi, A. H, Truman, C. E. and Smith, D. J., Proc. ASME PVP
conference, Denver, Colorado, USA, 2005.
2. Hadidi-Moud, S., Mirzaee-Sisan, A., Truman, C. E. and Smith, D. J., Fat. and Frac. Eng.
Mat. Struct., 27, 931-942, 2004.
 40. Residual Stress and its Effects on Fatigue and Fracture 1307

EVALUATION OF NOVEL POST WELD HEAT TREATMENT IN FERRITIC

STEEL REPAIR WELDS BASED ON NEUTRON DIFFRACTION

C. Ohms*, D. Neov*, R. C. Wimpory** and A. G. Youtsos*

*EC-JRC-IE, High Flux Reactor Unit, PO Box 2, 1755 ZG Petten, Netherlands
**Hahn-Meitner-Institute, Glienicker Str. 100, 14109 Berlin, Germany

carsten.ohms@jrc.nl

The occurrence of cracks in – normally welded – components with safety relevance in, e.g. nuclear
installations or in the (petro-)chemical industry, is not an unusual event. In almost all cases such
cracking is detected in periodic inspections prior to complete failure of the component. Sometimes
a detected defect necessitates repair of the damaged component to facilitate its further operation.
Repairing of a crack would in most cases be facilitated by excavation of the material
surrounding the crack and subsequent filling of the excavation by welding. However, such a repair
welding process leaves the component in a sensitive state in that it generates a complicated residual
stress pattern and that the heat affected zone of the weld might become very susceptible to the
formation of new cracking [1].
Post weld heat treatment of a repaired component can be an option to mitigate the damaging
impact of the welding process. Through heat treatments residual stresses can be severely reduced
or redistributed to obtain stress fields around the weld deemed less detrimental. At the same time a
heat treatment process could positively influence the HAZ sensitivity for further cracking. In any
case, a thorough assessment of the welding process is necessary to ensure a safe continued
operation of the repaired component.
In this context letterbox repair welds applied to thin ferritic steel plates to simulate repair of
postulated shallow cracks have been manufactured. A typical excavation shape is shown in Fig. 1
below [2]. Such an excavation would typically be filled with 20 to 30 welding passes. Components
have been made available in the as welded state and after the application of a PWHT. Two
different heat treatment processes were analysed: a. a full scale treatment, where the entire
component has been subjected to an elevated temperature for several hours in order to significantly
reduce the residual stresses, and b. an alternative treatment whereby the heat is applied locally for a
short period of time in order to redistribute the stresses in a controlled manner.
In this paper the experimental determination of these residual stresses in the as welded and tin
the heat-treated states is presented. Such measurements have been performed by neutron
diffraction at the High Flux Reactor of the Joint Research Centre of the European Commission in
Petten, the Netherlands.
The principle of residual stress measurements by neutron diffraction is introduced [3] and the
particular considerations for performing such measurements in multi-pass butt welds are briefly
outlined [4]. The experimental approach is presented and explained and an outline is given on the
data analyses. Results are depicted in the form of comparison between the as received and the heat
treated stress states. The derived data facilitate conclusions on the effects and effectiveness of the
applied heat treatments and they also demonstrate that neutron diffraction is a very suitable tool for
non-destructive analysis of internal residual stress fields in such welded components of
considerable thickness. In addition, the method is well suited for the validation of predictive
numerical models.
1308 C. Ohms et al.

FIGURE 1. Typical repair excavation geometry.

References
1. Boucher, C., Bourchard, P.J., Brown, B., Smith, D., Lawrjaniec, D., Hein, H., Truman, C.,
Smith, M., Ohms, C., Dauda, T.A., Cardamone, D. and Youtsos, A.G., submitted for:
Proceedings of ASME Pressure Vessels and Piping Conference 2005, Denver, July 17-21,
2005
2. Ohms, C., Wimpory, R.C., Neov, D., Lawrjaniec, D. and Youtsos, A.G., submitted for:
Proceedings of ASME Pressure Vessels and Piping Conference 2005, Denver, July 17-21
3. Hutchings, M.T., Krawitz, A.D., editors, Measurement of residual and applied stress using
neutron diffraction, Kluwer Academic Publishers, Dordrecht, Boston, London, 1992.
4. Ohms, C., Youtsos, A.G. and van den Idsert, P., in: Proceedings of Baltica V – International
Symposium on Condition and Life Management for Power Plants, edited by S. Hietanen & P.
Auerkari, VTT Technical Research Centre, Espoo, Finland, June 2001, Vol. 2, 487-497
 40. Residual Stress and its Effects on Fatigue and Fracture 1309

SURFACE CRACK DEVELOPMENT IN TRANSFORMATION INDUCED

FATIGUE OF SMA ACTUATORS

Dimitris C. Lagoudas1, Olivier W. Bertacchini1 and Etienne Patoor2

1Aerospace Engineering Department

Texas A&M University, College Station, TX 77843-3141

d-lagoudas@tamu.edu, obertacchini@tamu.edu
2Laboratoire de Physique et Mecanique des Materiaux UMR CNRS 7554/ENSAM Metz

4 rue Augustin Fresnel

57078 Metz, France
patoor@lpmm.univ-metz.fr

Fatigue properties of Shape Memory Alloys (SMAs) have been primarily studied based on the
pseudoelastic response of SMAs with some reference work done by Tobushi et al. [1] and
Miyazaki et al. [2]. However, thermally induced transformation fatigue is a more recent subject,
where the applied level of stress has a major influence on the developed plastic strains and
therefore on the fatigue performance of SMA actuators (Bigeon and Morin [3], De Araujo et al.
[4]). Recent work on the surface aspects has been carried out and has been demonstrated to be
relevant to the fatigue properties (Horbogen and Eggeler [5]).
This paper is based on the study of a post mortem analysis of NiTiCu shape memory alloy
actuators (SMA) undergoing thermally induced martensitic phase transformation fatigue under
various stress levels. In a previous study (Bertacchini et al. [6]), fatigue life results were obtained
for both complete and partial phase transformations of SMA wire actuators. In addition to inducing
cyclic loading with the transformation cycles, forced fluid convection cooling was utilized to
increase the cycling frequency (up to 1Hz), leading to an important corrosion phenomenon due to
the temperature variations. These experimental conditions led to the formation of a microcracked
structure with a uniform distribution of surface cracks (Fig. 1).

FIGURE 1. Periodic cracks in post mortem SMA wire (crack spacing | 200Pm).

In order to understand the stress state in the SMA wires leading to the observed microcracks, a
modified shear lag model is developed that accounts for eigenstrains due to corrosion, plasticity
and due to the martensitic transformation. Such inelastic eigenstrains were added to an elastic
solution previously derived by McCartney [7]. Some results are shown in Figs. 2-3.
1310 D. C. Lagoudas et al.

FIGURE 2.Axial stress profile at failure vs. crack spacing: 200Pm in complete transformation (4%
strain) External load = 154MPa Thickness of cracked brittle layer = 55Pm.
Eigenstrain mismatch = 0.5%

FIGURE 3. Axial stress profile at failure vs. crack spacing: 400Pm in partial transformation (2%
strain) External load = 154MPa Thickness of cracked brittle layer = 70Pm.
Eigenstrain mismatch = 0.5%

Critical values of crack spacing are reached at fatigue failure: spacing of 200Pm for complete
cycles and 400Pm for partial transformation cycles. The amount of transformation strain (and
therefore the plastic strain accumulation) seems to drive the ultimate crack spacing at failure.

References
1. Tobushi, H., Hachisuka, T., Hashimoto, T., Yamada, S., 1998. Journal of Engineering
Materials and Technology 120, 64-70.
2. Miyazaki, S., Mizukoshi, K., Ueki, T., Sakuma, T., Liu, Y., 1999. Material Science and
Engineering A 273-275, 658-663.
3. Bigeon, M., Morin, M., 1996. Scripta Materialia Vol.35 (N°12), 1373-1378.
4. C.J. De Araujo, M. Morin, G. Guénin, Material Science and Engineering. A 273-275 (1999)
305.
5. E. Hornbogen, G. Eggeler. Materialwissenschaft und Werkstofftechnik, Volume 35, Issue 5 ,
255-259.
6. Bertacchini, O., Lagoudas, D., Patoor, E. In: Proceedings of SPIE. Editors: Dimitris C.
Lagoudas, San Diego, CA, 2003, Vol. 5053, 612-624.
7. L.N. McCartney. In: Local Mechanics Concepts for Composite Material Systems. Eds., J. N.
Reddy and K. L Reifsnider, Proc. IUTAM Symposium, Blacksburg, VA, 1991, 251–282.
 40. Residual Stress and its Effects on Fatigue and Fracture 1311

FINITE ELEMENT SIMULATION OF WELDING IN PIPES: A SENSITIVITY

ANALYSIS

Dimitrios Elias Katsareas1, Carsten Ohms2 and Anastasios George Youtsos3

1Machine Design Lab., Mech. Engg & Aeronautics Dept, University of Patras

GR-26010 Rion, GREECE

katsareas@hotmail.com
2High Flux Reactor Unit, Institute for Energy, EC-JRC

PO2, 1755 ZG Petten, The Netherlands

carsten.ohms@jrc.nl
3High Flux Reactor Unit, Institute for Energy, EC-JRC

PO2, 1755 ZG Petten, The Netherlands

Anastasius.YOUTSOS@cec.eu.int

Thermal cycling, high heating rates, high temperature peaks and inter-pass and post weld cooling
are parameters that largely affect residual stress generation in and around welds. Residual stresses
influence considerably nuclear power plant component integrity, by affecting service-induced
crack initiation and even crack propagation. This influence can be even more severe under the
presence of corrosion mechanisms, like inter-granular stress corrosion cracking in austenitic steel
pipes. Such failure mechanisms are common in stainless steel piping used in pressurized water
reactors. As early as 1979 [1] residual stresses in pipes due to welding have been investigated
experimentally. Faure and Leggatt [2] utilized destructive test methods, like the centre-hole and
layering methods, to determine the residual stress fields in austenitic-ferritic pipe welded joints.
The common trend in industry, when developing a welding procedure specification, is to base it on
a large number of costly and complicated experiments. Computer simulation of welding and finite
element prediction of residual stresses present a cost effective and more efficient alternatives to the
engineer/designer of welds, as long as these methods have been validated and proven in the field.
Finite element simulations are in general supplementing the experimental tests for the evaluation
of welding residual stresses. As early as 1978 [3-8] two and three-dimensional models of welds
have been used for pipe welded joints. Lindgren [9] has shown that over the last 25 years of
recorded weld simulation efforts, FE model sizes have increased by a factor of 104. This dramatic
increase was allowed, off course, by a similar increase in computer power and clearly illustrates
the continuous effort for model refinement and increase in accuracy. This effort reflects the trend
in industry to shift the welding residual stress evaluation from an experimental to a computational
based procedure. Fricke et al [10] have developed FERESA, a finite element code based on
ABAQUS, dedicated to welding simulation and residual stress predictions and validated it by
analysing an austenitic pipe weld using 2-D and 3-D models. It is common practice among
researchers to develop in-house finite element codes for weld simulation. Such codes, though, lack
the universality of commercially available software, which is favoured by the industry. This is due
to the fact that, residual stress analysis procedures based on them can be readily transferred to
industrial applications.
In the present paper, a multi-pass weld joining two pipes made of different materials and a
single-pass benchmark pipe weld, are simulated using 2-D axi-symmetric finite element models.
The proposed methodology for weld simulation incorporates the well-known birth of elements
technique and follows the prescribed temperature approach for heat input modelling. Predicted
residual strains are compared to neutron diffraction measurements. The effect of various aspects of
modelling, on the accuracy of predicted residual strains, is investigated through a series of
1312 D. E. Katsareas et al.

sensitivity tests. After examination of the impact of free convection, as the most common cooling
method between weld passes, on predicted residual strains, it is derived that the heat transfer
coefficient has no significant effect, as long as its value remains within the free convection regime.
Radiation also contributes to cooling of a welded structure, but incorporating it into the FE model
shows that its effect on resulting residual strains is negligible. The same conclusion is reached
when applying two different methods of heat input modelling. The prescribed temperature and the
heat generation rate technique are based on entirely different approaches, concerning the FE
idealization of weld-induced thermal load. Numerical tests using both these approaches illustrate
that the end result of the simulation is insensitive to the choice. Comparison of computed residual
strains with and without incorporating phase change (solidification) in the model indicates that its
impact is also insignificant. The effect of non-linear material model choice and creep model on the
accuracy of results is also investigated, leading to the conclusion that they have to be accounted for
during welding simulation. Numerical tests are also performed concerning mesh optimisation as
well as procedure optimisation. In the later detailed bead-by-bead simulation results are compared
to layer-by-layer and lump-by-lump simulation results. Lumping is simply grouping of weld beads
in order to reduce computational costs. The ultimate goal of the present analysis is to determine the
level of simulation detail, which can be implemented under a realistic computational cost, but at
the same time achieving the industrial requirement of accuracy, regarding residual stress
predictions. The proposed simulation methodology is evaluated as a weld-induced residual stress
predictive tool and future steps towards this direction are outlined.

References
1. Ellingson, W.A. and Shack, W.J., Experiment. Mech., vol. 19, 317-323, 1979
2. Faure, F. and Leggatt, R.H., Int. J. Pressure Vessels Piping, vol. 65, 265-275, 1996
3. Rybicki, E. and Stonesifer, R., ASME J. Press. Vessel Technol., vol. 101, 149-154, 1979
4. Brust, F.W. and Rybicki, E., ASME J. Press. Vessel Technol., vol. 103, 226-232, 1981
5. Rybicki, E. and McGuire, P.A., ASME J. Press. Vessel Technol., vol. 103, 294-299, 1981
6. Koch, R., Rybicki, E. and Strttan, R., J. Engng. Mater. Technol., vol. 107, 148-153, 1985
7. Josefson, B.L. and Karlsson, C.T., Int. J. Pressure Vessels Piping, vol. 38, 227-243, 1989
8. Karlsson, R. and Josefson, B., ASME J. Press. Vessel Technol., vol. 112, 76-84, 1990
9. Lindgren, L.E., J. Therm. Stress, vol. 24, 141-192, 2001
10. Fricke, S., Keim, E. and Schmidt, J., Nucl. Engng. Design, vol. 206, 139-150, 2001
 40. Residual Stress and its Effects on Fatigue and Fracture 1313

RESIDUAL STRESS PREDICTION IN LETTERBOX-TYPE REPAIR WELDS

Loukas Keppas1, Nikolaos Konstantinos Anifantis2, Dimitrios Elias Katsareas3

and Anastasios George Youtsos4
1Machine Design Lab., Mech. Engg & Aeronautics Dept, University of Patras

GR-26010 Rion, GREECE

loukaskeppas@yahoo.gr
2Machine Design Lab., Mech. Engg & Aeronautics Dept, University of Patras

GR-26010 Rion, GREECE

nanif@mech.upatras.gr
3Machine Design Lab., Mech. Engg & Aeronautics Dept, University of Patras

GR-26010 Rion, GREECE

katsareas@hotmail.com
4High Flux Reactor Unit, Institute for Energy, EC-JRC

PO2, 1755 ZG Petten, The Netherlands

Anastasius.YOUTSOS@cec.eu.int

Many researchers over the last two decades have paid attention to predict satisfactorily the residual
stresses in weldments. Dong et al [1] inferred that residual stresses in weld repairs typically exhibit
strong three-dimensional features, depending on both component and repair geometry. Repair
welds are a common way in industry of repairing cracks or other forms of defects in steel
components and structures. Part of the metal is excavated through machining and with it the crack.
The groove, which has usually a letterbox geometry, is filled with weld metal of the same
composition as the parent material. Since the welding procedure is of a multi-pass type, a repair
weld is considered an extreme case (many passes) of a multi-pass weld. Robinson et al [2]
concentrated on practical weld repair procedures for low alloy steels. During the design phase of
structures and their components or during evaluation of a potential crack initiation and growth, it is
important to have a complete description of the residual stress distribution. Moreover, there is the
potential for stress distributions to become even more complicated, when weld repairs are
undertaken within regions of components that have been fabricated previously by welding. Indeed,
in the construction of new plants and for their continued operation, local repair welds are
undertaken, so it is necessary to be able to underwrite these for safe operation. As a consequence,
knowledge of the residual stresses and their distribution is an important input to an overall
structural integrity assessment. Several researchers have conducted numerical analyses
incorporating PWHT and creep effects in their models. The main scope of these studies is the
examination of hot-cracking risk during the PWHT or during the operation under real conditions.
Hyde et al [3] using a lumped bead 3D finite element model to simulate a pressurized CrMoV pipe.
A full PWHT was carried out and the analyses covered a number of initial damage levels,
magnitudes of axial loads and repair excavation depths. Weld simulation involves complicated
aspects of modelling like metallurgical phase transformation, temperature dependent material
properties, creep, phase change, radiation, heat input models, etc. The impact of these on the
accuracy of the predicted residual stress has attracted researchers for some time now. Lindgren in
his review [4] demonstrates the complexity of weld simulation models if aspects such as solid-state
phase transformations and hot cracking are involved in order to achieve a more accurate analysis.
The scope of the present work is the determination of the residual stresses in a 2¼CrMo plate,
containing a multi-pass repair weld of the same material, including the effects of different post
weld heat treatments. The weld simulation procedure is incremental in nature as is the real process
1314 L. Keppas et al.

of weld bead deposition. Each weld bead is descretized in a number of increments and these are
“deposited” sequentially. “Deposited” means, in numerical terms, activated and that refers to the
elements that constitute the weld bead. The deactivation and activation of elements during the
simulation is achieved through the very well known “birth & death of elements” technique, a
feature common to many commercial FE codes. Element deactivation or “death”, as it is called, is
not achieved by actual removal of "killed" elements, but by multiplying their stiffness,
conductivity, etc, by a severe reduction factor. When an element is reactivated, its stiffness,
conductivity, etc. return to their full original values. The proposed methodology, for multi-pass
weld simulation follows the prescribed temperature approach for heat input modeling.
Metallurgical phase transformation effects are not included in the model, although it is general
knowledge that its role in the formation of a residual stress field might be quite significant. Such
transformations might be introduced in a future model, in an appropriate FE code. Experimental
results are used to validate predictions obtained by finite element computer simulation.
Specifically, computed temperature and thermal strain histories are compared to measured data,
whereas predicted residual strains are compared to neutron diffraction measurements. Results
obtained using the aforementioned methodology will be compared with simplified models, in the
frame of the sensitivity analysis, in order to establish a balance between acceptable accuracy and
computational costs. The overall objective of the present paper is the development of a residual
stress predictive tool for repair welds, based on the evaluation of currently developed finite
element techniques, as encountered in industrial applications. Emphasis is to be given on the
applicability such a tool will have for the industry, in respect of the computational cost of its
implementation.

References
1. Dong, P., Zhang, J. and Bouchard, P., ASME J Press. Vessel Technol., vol. 124, 74-80, 2002
2. Lant, T., Robinson, D., Spafford, B. and Storesund, J., Int J. Press. Vessels Piping, vol. 78,
813-818, 2001
3. Hyde, T., Sun, W., Becker, A. and Williams, J., Int J. Press. Vessels Piping, vol. 81, 1-12,
2004
4. Lindgren, L.E., J. Therm. Stress., vol. 24, 141-192, 2001
 40. Residual Stress and its Effects on Fatigue and Fracture 1315

EFFECT OF REFLECTION SHOT PEENING AND FINE GRAIN SIZE ON

IMPROVEMENT OF FATIGUE STRENGTH FOR METAL BELLOWS

H. Okada1 , A. Tange1 and K. Ando2

1NHK SPRING CO.,LTD. ,Kanagawa-Pref, JAPAN,
2Yokohama National University, Kanagawa-Pref, JAPAN

h.okada@nhkspg.co.jp, tange@nhkspg.co.jp, andokoto@ynu.ac.jp

Generally metal bellows (it is hereafter described as a bellows) are known for a seal to defend a
leak being elastic1)The appearance photograph of bellows is shown in Fig.1.Recently small size
bore bellows are getting to expect to a seal of pump which was small and can use under the
circumstances of high pressureHowever, until now, it would be difficult to achieve the fatigue
strength for required specifications in most cases. Meanwhile, in order to increase the fatigue
strength of automobile parts, the author proposed the new improvement method on fatigue
strength, through the clarification of the process of fatigue fracture and defensive factors2-4). That
is: (a)To increase the Vickers Hardness of the component as high as possible.(b)To introduce high
compressive residual stress as high as possible.(c)To fine a grain size as fine as possible. In this
study, these methods were applied to bellows.The experiment was carried out by using bellows of
SUS631 steel and SUS304 steel. Grain diameter was adjusted by changing bright annealing
temperature after cold working. It could be realized that fatigue strength increases in proportion
with decreasing grain diameter. It was known for that shot peening had an effect on improving
fatigue strengthIt can be expected that it is difficult to obtain a large compressive residual stress on
the inner surface of small bore size bellows by shot peening processIn order to obtain the effect of
shot peening on the inner surface of bellows, a new shot peening processes by using a air peening
machine and a reflective plate were developedThe reflection plate shown in Fig.2 installed to the
nozzle, in order to apply the shots effectively to the inside surface of bellows5). It was found that
this method could apply the shots to the inner surface of bellows effectively, to improve the fatigue
life of bellows, comparing with non-shot peening one. The optimum shot peening conditions were
decided by the relationship between residual stress, pressure and glass beads size in experiment.
Fatigue limit of 107 cycles of SUS631 bellows was improved by 140% from SUS304 bellows
when the optimum condition is applied. The reason why was improved fatigue strength by shot
peening was considered from the viewpoint of the residual stress, work hardening, surface
roughness and grain size.

Fig.1 Appearance photograph of bellows

1316 H. Okada et al.

References
1. T.Mitsushiba, a vacuum, 26-10, (1983), 757.
2. H.Ishigami,K.Matsui,A.Tange and K.Ando,JHPI,Vol.38 No.4,(2000),205-215.
3. A.Tange and K.Ando,JHPI,Vol.38 No.4,(2000),216-223.
4. K.Matsui, H.Eto,K.Kawasaki,Y.Misaka and K.Ando, The collection of the Japan Society of
Mechanical Engineers papers(A piecies),65-637(1999),1942-4947.
5. H.Ishigami, K.Matsui, Y.Jin and K.Ando Fatigue and Fracture of Engineering Materials &
Structures,23(2000),959-963.
 40. Residual Stress and its Effects on Fatigue and Fracture 1317

VISCOSITY EFFECT ON DISPLACEMENTS AND STRESSES OF A TWO-

PASS WELDING PLAT

W. El Ahmar and J. F. Jullien

LaMCoS, CNRS UMR 5514, INSA-Lyon
INSA-Lyon, 20 Avenue Albert Einstein, 69621 Villeurbanne Cedex, France,
Walid.El-Ahmar@insa-lyon.fr, jean-francois.jullien@insa-lyon.fr

The highly localized transient heat and strongly nonlinear temperature fields in both heating and
cooling processes cause nonuniform thermal expansion and contraction, and thus result in plastic
deformation in the weld and surrounding areas. Consequently, residual stress, strain and distortion
are permanently produced in the welded structures. High tensile residual stresses are known to
promote fracture and fatigue, while compressive residual stresses may induce undesired, and often
unpredictable, global or local buckling during or after the welding. It is particularly evident with
large and thick panels, as used in the construction of nuclear building. These adversely affect the
fabrication assembly and service life of the structures. Therefore, prediction and control of residual
stresses and distortion from the welding process are extremely important for the nuclear
installation’s security.
The aim of this study is to investigate the effect of each temperature-dependent material
property on numerical results. A three-thermo-mechanical behaviour of 316L stainless steel,
during a TIG welding process, with 316L material filler, is adopted and based on an experimental
study, which is carried out on an industrial two pass weld benchmark [1]. The proposed simulation
analyses, using finite element method, are performed with Code_Aster from EDF.
A two-pass weld using the TIG process with 316L material filler is made along the chamfer of
the plate (270 x 200 x 30mm) in the long axis direction. The weld begins on the appendix and ends
10mm from the plate edges. The welding parameters used for the trial are U = 9V, I = 155A, and a
travelling speed of 0.667mm.s-1. The plate is lying on three points in its lower face. Temperatures
are continuously recorded during the welding, using thermocouples (kind K (±15%)) in some
points of the plate surface. Welding parameters (Tension, Intensity, travelling speed of the torch)
are also continuously recorded during the test. After cooling of the second pass, the residual
stresses in the middle section perpendicular to the welding direction of the plates are measured
with X-rays (±50MPa).
A three-dimensional numerical simulation is adopted. Due to the symmetry of the plate (the
role of the thermocouple T4 and the captor D6 is to verify that the symmetry is conserved), only
one half was modelled. For the modelling of the heat source, it is of course possible to consider
different kinds of mathematical models, from surface, like a Gaussian heat source [2], to
volumetric, like the double ellipsoid from Goldak. Different kinds of modelling of the heat source
have been considered for the thermal steady state calculation, with an efficiency parameter K fitted
so as to adjust the simulated temperatures considering the measured ones (Fig.2). If the net total
heat flux KUI is the same, it appeared that, the way of the heat flux geometry density has little
effect on the macroscopic simulated temperature field. For that raison, the chosen heat source
modelling for the 3-Dimensionnal transient thermal calculation was rather simple: it presents a
volumetric heat flux density, with a triangular aspect (Fig.1).
1318 W. El Ahmar and J. F. Jullien

FIGURE 1. Heat source modelling FIGURE 2. Thermal adjust

FIGURE 3. Transient displacement FIGURE 4. Longitudinal residual stress

The main results and conclusions are summarized as follows:

• The transient displacement (Fig.3) is mainly sensitive to the thermal expansion coefficient.
• The residual stress (Fig.4) is mainly sensitive to the yield stress and hardening model.

References
1. D.Ayrault, O.Blanchot, A.Fontes, Two-Passes GTA Welding Instrumented Tests as references
for both the Round Robins on numerical analysis and measurements of residual stresses, 56th
Annual assembly of International Institute of welding, July 6-11, 2003, Bucharest, Romania,
paper IIW-X/XIII/XV-RSDP-82-03
2. L.Depradeux, J-F.Jullien, Experimental and numerical simulation of thermomechanical
phenomena during a TIG welding process, J.Phys.IV France 120 (2004) 697-704
3. Y.Vincent, J-F.Jullien, P.Gilles, Thermo-mechanical consequences of phase transformations
in the heat-affected zone using a cyclic uniaxial test, International Journal of Solids and
Structures, 42 (2005) 4077-4098.
 40. Residual Stress and its Effects on Fatigue and Fracture 1319

SURFACE INTEGRITY IN HIGH SPEED MACHINING OF

TI-6WT.%AL-4WT.%V ALLOY

Juan David Puerta Velasquez1*, Bernard Bolle1, Pierre Chevrier2 and Albert Tidu1
1 Laboratoire d’Etude des Textures et Applications aux Materiaux (LETAM), UMR CNRS 7078,

ISGMP, F-57045 Metz Cedex 01, France

2 Laboratoire de Physique et Mecanique des Materiaux (LPMM), UMR CNRS 7554, ISGMP, F-

57012 Metz Cedex 01, France

CEPGV – Ecole Nationale D’Ingenieurs de Metz (ENIM), F-57045 Metz Cedex 01, France
*velasquez@letam.univ-metz.fr

Titanium alloys are very interesting materials for industrial applications because of their elevated
mechanical resistance having a low density and their excellent corrosion resistance, even at high
temperatures. Despite these features, utilisation of titanium alloys is still limited because of their
poor machinability, intimately bound to their thermal and chemical properties [1]. In fact, low
thermal conductivity of titanium hinds the evacuation of heat generated during cutting process,
leading to temperature rise of the work-piece. Furthermore, the high chemical reactivity of
titanium, which increases with high temperature, produces an early damage of the cutting tools,
affecting the final quality of obtained surface as well as increases in production costs.
High speed machining is widely appreciated in industry for its high quality and surface
finishing in the obtained parts. Nevertheless, erroneous selection of cutting parameters can
generate poor surface finishing. Surface integrity can be defined as a measure of the quality of a
machined work-piece and it includes roughness, crystallographic texture, residual stress and
metallurgy of the obtained surface. The presence of residual stresses and residual stress gradients
in surface of parts under fatigue conditions are recognised as important factor affecting their life.
Generally it is accepted that the presence of compressive residual stresses in surface is beneficial
for work-piece lifetime and the residual tensile stresses can lead to crack propagation and final
failure of the work-piece. The aim of this work is to study the surface integrity in high speed
machining of Ti-6wt.%Al-4wt.%V alloy.
Machined material was an (D+E) titanium alloy (tube diameter 77.98 mm, average thickness 5
mm). High speed turning experiments were carried out using orthogonal cutting geometry in a
computer numerically controlled lathe (Ernault HS-400). All tests were carried out without
lubrication. Machining parameters are given in Table 1. Diamond coated flat top inserts (Sandvik
TMCW cutting angle of 0° and a clearance angle of 7°, Balzers Balinit Diamond coating) were
used. A new insert was used for each test, in order to avoid negative cumulated wear effects.

TABLE 1. Machining parameters

Cutting speed (m.min-1): 20, 40, 100, 140, 180, 220,

260, 300, 420, 540, 660
Cutting depth (mm): 0.6
Cutting width (mm): 5
Radial cut depth (mm.rev-1): 0.12
1320 J. D. P. Velasquez et al.

Machined surfaces were characterised by X-ray diffraction (XRD). Measurements were

carried out on a texture goniometer equipped with a curved position sensitive detector (CPS120,
Inel). The XRD patterns, crystallographic texture and residual stresses were obtained using Cu KD
(O = 1.5418 Å) radiation emitted by a rotating anode (Rigaku RU300). An isotropic model, so-
called sin2\ method, was used to determine residual stress. Detailed explanation of this method
can be founded in [2]. Samples were also observed by cross sectional scanning electron
microscopy (SEM), using 6500F JEOL FEG-SEM.
Microstructural images obtained by SEM shows a turning of grains at the machined surface, in
the direction of the cutting tool passage (Fig. 1a). Samples were also evaluated by XRD
techniques. An analysis of XRD patterns leads to determinate the effect of cutting tool passage in
the obtained surface. Changes in peaks intensity ratios and peaks broadening, depending on the
cutting speed, evidences a high plastic deformation and a modification of crystallographic texture
in the obtained surface (Fig. 1b).

(a) (b)
Figure 1. Microstructure modification of surface after cutting tool passage (a). XRD patterns of as-
received material and machined surfaces at different cutting speeds (b). Theoretical XRD patterns
of Ti D and Ti E phases are given on the top of the figure.

Linear correlation between the cutting speed and the residual stress was revealed. Lower
cutting speed leads to compressive residual stress. Higher cutting speed causes the residual stress
to change from compression to tensile.
Surface integrity study of parts obtained by high speed machining was carried out. SEM and
XRD techniques reveal microstructural modifications of surface after cutting process. Major
influence of cutting speed on crystallographic texture and residual stress has been noticed.

References
1. Ezugwu, E.O., Wang, Z.M., J. Mater. Process Tech., vol 68, 262-274, 1997
2. Chevrier, P., Tidu, A., Bolle, B., Cezard, P. and Tinnes, J.P., Int. J. Mach. Tools & Manu.,
vol. 43, 1135-1142, 2003
 40. Residual Stress and its Effects on Fatigue and Fracture 1321

PHASE TRANSFORMATION AND DAMAGE ELASTOPLASTIC

MULTIPHASE MODEL FOR WELDING SIMULATION

T. Wu, M. Coret and A. Combescure

LaMCoS, CNRS UMR 5514, Institut National de Sciences Appliquees INSA-Lyon
Bat. Coulomb, 20 avenue A. Einstein 69621 Villeurbanne Cedex, France
John.wintl@twi.co.uk, Michel.coret@insa-lyon.fr

The article is dedicated to the study of the introduction of damage concept in the multiphasic
behaviour which occurs during by welding process. Attention is paid to coupling between ductile
damage, small strain elasticity, finite visco-plasticity and phase transformation. Based on the
theory of thermodynamics and continuum damage mechanics (CDM), we built constitutive
equations to describe damage growth and crack appearance during and after welding. The
thermodynamics of irreversible processes with state variables is used as a framework to develop
the phase coupling model. The related numerical aspects concern both the local integration scheme
of the constitutive equations and the global resolution strategies. The models are implemented in
CASTEM 3M finite element code. Based on the finite element program, some examples and
applications are presented to forecast the damage induced by welding and validate the models.
In the filed of metal hot process, virtual process is more and more used in many institutes and
enterprises in order to forecast the results and adjust the process parameters before implement of
the real experiments and manufacture processes. Since welding is an important method of
manufacture process, many researchers studied welding numerical simulation and one want to
prevent cracks or moreover damage induced by welding process. Simulation should help a lot in
order to avoid long test periods. However, now there is no evidence of a model which is able to
predict the damage induced by welding process. One observes high spatial and temporal gradients
as well as phase transformations. Such situations imply to cope with a diversity of damage models
and add complexity to standard constitutive equations. The model contains three main ingredients,
continuum damage mechanics, transformation plasticity and multiphase behavior.
To model the damage, the strains are assumed small, and the damage isotropic. The article
focuses on the coupling of damage with thermal strains as well as visco-elastoplasticity including
the mixed isotropic and kinematic nonlinear hardening. The framework used is the
thermodynamics of irreversible processes with state variables. The basic work is presented in
 
Lemaitre & Chaboche [1-2]. The two observable state variables are H , V . The internal variables

  
e & &

are H , V , q , g grad (T ) , and their conjugate forces are r , R , D , X  . D, Y  are the
isotropic damage and its conjugate force.

For multiphase situation, the volumetric proportion of each phase, denoted zi , is calculated
from continuous cooling transformation (CCT) diagram obtained by experiments and some models
for calculating phase fraction are available and chosen depending on different temperature history
such as Waeckel’s model, Koistinen-Marburger’s law and Leblond’s transformation plasticity
model. M. Coret and A. Combescure have proposed a model for plastic transformation and
behavior of multiphase materials [3-5].
1322 T. Wu et al.

e
The Helmholtz free energy < (H , r , D , D , T , z i ) is considered to be the addition of free

energy of each phase i of the n phases considered and is described by the following equation
(1):

e n e e n p tr
U< (H , r , D , D , T , zi ) ¦ U< (H , D , T , zi )  ¦ U< ( r , D , D , T , zi )  U< ( D , T , z )
i 1 i 1 (1)

e p tr
In this equation < is the elastic potential, < the inelastic one and < the
transformation plasticity one. By introducing the above state potential into the Clausius-Duhem

basic inequality, the dual variables V , R, X , Y  can be derived together with the residual or
th
dissipation inequality. In the inequality thermal dissipations ) and mechanical dissipations
m
) are uncoupled.

m th
) ) ) t0 (2)

p  &
Analysing dissipations ) and potential of dissipations M , flux variables (H , D , D , q T )
will be presented. The idea in this model is to characterise rupture by single common parameter to
all phases. This approximation is rather crude but is a first step to a more realistic model which
should have a damage evaluation for each phase: this type of model necessitates new models to
describe the relationship between these damages (history effects).
The constitutive equations and formulae are implemented in CAST3M finite element code.
Application examples shall be presented.

References
1. Lemaitre, J. and Chaboche, J. L., Mechanics of Materials, Cambridge University Press,
Cambridge, 1994.
2. Lemaitre, J., A Course on Damage Mechanics, Springer-Verlag, 1992.
3. Coret, M., Etude Experimentale et Simulation de la Plasticite de Transformation et du
Comportement Multiphase de l’acier de Cuve 16MND5 sous Chargement Multiaxial
Anisotherme, These, LMT-Cachan, Paris, France, 2001.
4. Coret, M., Calloch, S. and Combescure, A., Experimental Study of the Phase
Transformation Plasticity of 16MND5 Low Carbon Steel Induced by Proportional
and Nonproportional Biaxial Loading Paths, European Journal of Mechanics - A/Solids,
vol. 23, 823-842, 2004.
5. Coret, M., and Combescure, A., A Mesomodel for the Numerical Simulation of the
Multiphasic Behavior of Materials under Anisothermal Loading, International Journal
of Mechanical Sciences, vol. 44, 1947-1963, 2002.
 40. Residual Stress and its Effects on Fatigue and Fracture 1323

THE PRESENT SANS INSTRUMENT AND THE NEW HFR-PETTEN SANS

FACILITY BASED ON A COLD NEUTRON SOURCE

O. Ucaa,b, C. Ohmsa, D. Neova and A. G. Youtsosa

aHigh Flux Reactor Unit, Institute for Energy,

EC-JRC, POs2, 1755 ZG Petten, The Netherlands

boktay.uca@jrc.nl

A great deal of the properties of materials is influenced by phenomena taking place in the sub-
micron region. Scattering techniques play an important role for obtaining structural information.
Small-Angle Neutron-Scattering (SANS) is one such scattering technique by which one can obtain
structural information of the material being studied. Structural information here means size and
form of the object under investigation. In a SANS experiment one collects data as a function of the
momentum transfer which is proportional to the scattering angle. The probed length scales by
SANS varies from a few nanometer to 600 nanometers.
The Small-Angle Neutron-Scattering facility at the 45 MW High-Flux Reactor in Petten, the
Netherlands, is constructed in the late eighties. It is a medium size SANS instrument covering a q
range of 5x10-3 Å-1 to 0.4 Å-1. The instrument has been constructed with a double crystal
monochromator consisting of six pairs of Pyrolytic Graphite crystals giving a monochromatic
beam of 4.75 Å. The present SANS machine uses the HB3 radial beam tube. The flux at the sample
position is for the current case ~104 n/cm2/s. The instrument has a length of 10 meters. The two
dimensional position sensitive detector can be moved from 90 to 425 cm, Alf and Zurita [1] and
Vlak et al. [2]. After the construction of the facility the scientific activities were stopped due to
several non-scientific reasons. In recent years there has been effort to make the facility become
operational. This article consists of two parts:
a) In first instance, we will show the first preliminary results obtained at the HFR SANS
facility from measurement of two different samples. The first sample is a solution of silica particles
(SiO2) solved in cyclohexane. The volume fraction of the silica particles is 0.02. The second
measured material is a ceramics sample. In both cases one can determine the radius of gyration.
Actually these samples are old samples and aging effects cannot be excluded. However, at this
stage we are interested in obtaining a signal from the instrument and the question of absolute
calibration is not important. Moreover, determination of several characteristics in SANS such as
the radius of gyration is independent of absolute calibration. In future we plan to do experiments
with accurately defined mono-disperse micro-spheres to absolutely calibrate the instrument. After
this, the facility can be used to do scientific experiments on an absolute scale.
b) Secondly, we will discuss the possibility of increasing the flux at the sample position of the
SANS facility. The main components/characteristics of the instrument are:
1 The initial flux which is extracted from the reactor pool.
2 The moderator.
3 The in-pile collimator
4 The neutron guide
5 Monochromatization of the beam
6 Sample collimator
7 Sample station
1324 O. Uca et al.

8 Flight tube
9 The two-dimensional position sensitive detector.
For the purpose of upgrading only points 1, 2, 4 and 5 are important. It is evident that a higher
initial flux is favorable. The flux at the entrance of the HB10 radial beam tube is two times higher
than the HB3 radial beam tube, which is the present tube for the facility. So moving the facility to
the new beam tube will double the flux. At the moment, the instrument uses neutrons of 4.75 Å at
room temperature. The neutrons from the reactor follow a Maxwell-Boltzamn distribution.
Therefore it is favorable to use a cold source for increasing the flux.
The monochromatization of a white beam can be done by several methods. These are: double-
crystal monochromator, velocity selector and polycrystalline BeO filter Aswal, [3]. Although, it is
used in SANS facilities, the BeO filter is not a good choice because of the asymmetric wavelength
distribution around the mean wavelength. The velocity selector gives a slightly higher flux
compared to the double monochromator. However, with a velocity selector one has the choice to
continuously tune the wavelength form 2 Å to 25 Å Rosta [4].
The neutron guide will have a transmission of 1 for neutrons above the cutoff wavelength, Oc
and for neutrons whose paths are inclined at angles less than the critical angle T c to the nominal
beam direction. For 4.75 A and 58Ni, Tc=9.64 10-3 rad Crawford [5] and Baruchel et al. [6].
Therefore, the divergency of the in-pile collimator has to match the divergency of the guides.

References
1. Ahlf, J. and Zurita, A. (editors), High Flux reactor (HFR) Petten- characterization of the
installation and the irradiation facilities, Report No. EUR 15151 EN, 1992
2. Vlak, W.A.H.M., Dijk C. , Slagter, S., The ECN Small- Angle Neutron Scattering facility,
Report No. ECN-PB-89-3
3. Aswal, V. K.,, Journal of Applied Crystallography 33, 118-125, 2000
4. Rosta L., Physica B 174, 562-565, 1991
5. Crawford, R. K. and Carpenter, J.M ., J. Appl. Cryst,. 2, 589-601,1988
6. Baruchel. J. et al. (editors), Neutron and synchrotron radiation for condensed matter studies,
HERCULES, volume 1, p. 125-126
 40. Residual Stress and its Effects on Fatigue and Fracture 1325

RESIDUAL STRESS NUMERICAL SIMULATION OF TWO DISSIMILAR

MATERIAL WELD JUNCTIONS

Philippe Gilles, Ludovic Nouet and Pascal Duranton

Framatome-ANP
Tour AREVA Paris la Défense 92084, France
Philippe.gilles@framatome-anp.com
Pascal Duranton
ESI-France
Le Discover, 84 Bd. Vivier Merle 69485 Lyon France

In nuclear reactors such as Pressurized Water Reactors (PWR), heavy section components made in
low alloy steel are connected with stainless steel piping systems. The dissimilar material weld
(DMW) junctions are performed following a special manufacturing procedure to ensure a good
resistance of the joint. However, several experiences from the field confirm sensitivity to fatigue,
corrosion or low toughness areas in this type of junction.
In the framework of the European Community Research and Development Programme two
projects (DG-RTD programmes BIMET, C. Faidy et al. [1] and ADIMEW, C. Faidy [2]) have
been sponsored on the fracture behaviour of cracked stainless steel/ferritic steel bimetallic welds.
In each of these projects, a task was carried out on of residual stresses evaluation. Residual stress
measurements were performed using the neutron diffraction technique (C. Ohms [3]) across the
piping thickness in the buttering, weld and the HAZ of the base material. In the ADIMEW project,
residual stress measurements were also carried out on the surface by the Hole Drilling method and
verifications were also made later by the Cut-Compliance method (H. Schindler [4]).
The BIMET tubular mock-ups had a length of 393 mm and outer and inner diameters of 168
and 118 mm (25 mm thick). Two mock-ups were used for fracture testing and one for material
characterisation and residual stress measurements. The 308/309SS DMW connects a A508 ferritic
steel pipe section to a 304 stainless steel pipe section. A 308/309SS buttering is applied on the
ferritic pipe.
For the ADIMEW project, two DMW specimens (Fig. 1) have been manufactured of which
one has been tested in 4-point bending and one being utilised for the experimental determination of
welding residual stresses and for the generation of material property data for input to the analysis
procedures. The ADIMEW mock-ups were much larger than the BIMET ones, the diameter and
thickness being respectively 445 and: 51 mm. To guarantee the quality of the DMW's, a
considerable experience and high degree of quality control has been required within the weld
manufacturing process, in accordance with FRAMATOME-ANP basic nuclear specification.
This paper presents numerical simulations of the welding process of the BIMET 6” and the 16”
ADIMEW Dissimilar Metal Weld mock-ups. The calculations were performed by ESI and
FRAMATOME-ANP using different levels of modelling:
A simplified analysis, which consists in a cooling down calculation and assumes a stress-free
state after the stress-relief heat treatment,
A detailed analysis, which simulates each elementary step of the mock-up manufacturing
procedure using a lumped element technique.
A refined analysis, which simulates each elementary step of the mock-up manufacturing
procedure by modelling each of the passes
1326 P. Gilles et al.

FIGURE 1. ADIMEW Dissimilar Weld Metal Junction mock-up

The paper compares numerical results with the residual stress measurements to discriminate
the different numerical simulation techniques. All o these simulations assume axisymmetry, which
is shown to be realistic by reference to another test on butt welded pipes. Comparisons are also
made on residual stress fields obtained on the two geometries to put in evidence of their similarity.
It is shown that the simplified approach based on a cooling down simulation underestimates the
level of residual stresses. The lumped technique gives satisfactory results except in the root weld
area. The pass by pass simulation may be considered as the most reliable which is supported by
comparison to other simulations of the same type. It should be underlined that for the ADIMEW
simulation, the phase transformation process has been modelled in the ferritic part, allowing
predicting size and hardness of the Heat Affected Zone.

References
1. Faidy C. et al., "BIMET: Structural Integrity of Bi-Metallic Components", In Proceedings of
FISA 1999 Conference, Luxembourg, 1999.
2. Faidy C., “Structural integrity of dissimilar welds – ADIMEW project overview” In
Proceedings of PVP 2004, ASME Pressure Vessel and Piping Conference, 2004, Ed ASME
USA.
3. Ohms C., Katsareas D. E., Wimpory R. C., Hornak P., Youtsos A. G., “Residual stress
analysis in a thick dissimilar metal weld based on neutron diffraction”, PVP Vol. 479, ASME
Pressure Vessel and Piping Conference, July 2004, USA.
4. Schindler, H. J., 2003, 'Residual stress effects on crack growth mechanisms and structural
integrity', 9th Conf. on Mechanical Behaviour of Materials, Geneva, Switzerland, 25-29 May.
 40. Residual Stress and its Effects on Fatigue and Fracture 1327

IDENTIFICATION OF WELD RESIDUAL STRESS LENGTH SCALES FOR

FRACTURE ASSESSMENT

P. J. Bouchard and P. J. Withers

British Energy
Barnwood, Gloucester GL4 3RS, UK
p.john.bouchard@british-energy.com
Manchester Materials Science Centre, University of Manchester
Grosvenor Street, Manchester M1 7HS, UK
philip.withers@manchester.ac.uk

It is unlikely that any engineering component is entirely free from residual stress because of the
material processing, fabrication and service load history. Residual stresses originate from the
elastic accommodation of misfits between different regions in a structure. The interaction between
the misfit and the restraint of the surrounding material and structure determines the magnitude of
the resultant residual stress and its length scale. In order to assess the influence of residual stress on
fracture, it is essential to quantify the residual stress field over the length scales of concern. This
paper identifies the residual stress length scales that must be considered in fracture mechanics
analyses for welded joints in engineering structures and proposes a new approach for the treatment
of short length scale stresses. The findings are illustrated by results from recent weld residual stress
finite element simulations and measurement studies.
Simplified fracture assessment methods, such as the R6 procedure [1], are widely used by
industry to determine the structural integrity significance of postulated cracks, manufacturing
flaws, service-induced cracking or suspected degradation of engineering components. The R6
procedure was originally developed for the assessment of macro-cracks in pressurised steel vessels
and pipework. For this type of application the length scale (through-thickness extent) of macro-
cracks often lies within the range 0.5 – 5mm. Such macro-cracks might be based on the largest
defect that could have been missed by the inspection method, the size of a weld bead (a lack of
fusion defect for example), or an assumed “engineering crack” initiation size. Of course much
larger defects in structures are assessed when discovered by service inspection. More generally, the
length scale of macro-cracks can be defined as being an order of magnitude greater than the
material grain size.
Multi-pass welding introduces a very complex residual stress field in a structure. The stresses
vary in 3-dimensions over a range of macro length scales [2]. However, for simplified engineering
fracture assessments a representative through-wall line stress profile is required that applies across
the width of the crack. The R6 procedure provides upper bound (Level 2) residual profiles for
various classes of welded joint that can be used, but these are very conservative. Recently, more
realistic (Level 3) profiles for pipe butt welds have been introduced into R6. These are based on
detailed residual stress measurements and finite element predictions for several welds and it is
claimed that they capture the underlying residual stress distribution [3]. However, some marked
differences between measurements and the idealised profiles have been noted and this undermines
confidence in their use.
Rapid advances in the capability of measurement techniques, such as the Contour method and
Synchrotron diffraction, now readily allow residual stresses and strains in welded structures to be
mapped. Recent experimental and analytical studies have revealed the importance of short length
scale residual stress variations arising from weld bead pass sequence effects, weld bead start/stops
and stress concentrating features. Measurements at a limited number of points in a structure will
1328 P. J. Bouchard and P. J. Withers

quantify the local stresses at those points but this may not reveal the underlying stress field
characterising the structural behaviour of the joint (i.e. the Level 3 profile).
When Level 3 profile stress modulations are significant it is important that they are accounted
for in fracture assessments. A criterion is proposed for filtering benign short-range residual stress
perturbations from fracture assessments based on the size of the defect relative to the length of the
residual stress tensile zone. In addition, it is shown how significant stress modulations might be
conservatively treated in fracture assessments by using classical stress intensity factor solutions for
self-equilibrating stress fields having appropriate length scales and magnitudes.

References
1. R6 Revision 4, 2004, Assessment of the Integrity of Structures Containing Defects, British
Energy Generation Ltd., Gloucester UK.
1. Bouchard, P.J., Withers P.J., J Neutron Research,12(1-3):81-91, 2004.
1. Bouchard, P.J., Bradford R.A.W., In Proceedings of ASME Pressure Vessels & Piping
Conference, PVP-Vol. 423, ASME, 2001, 93-99.
 40. Residual Stress and its Effects on Fatigue and Fracture 1329

HIGH-RESOLUTION NEUTRON DIFFRACTION FOR PHASE AND

RESIDUAL STRESS INVESTIGATIONS

P. Mikula and M. Vrana

Nuclear Physics Institute and Research Centre Rez, plc.
250 68 Rez, Czech Republic
mikula@ujf.cas.cz

In this contribution the authors present attractive properties of an unconventional neutron

diffractometer employing a dispersive multiple-reflection monochromator which permits
ultrahigh-resolution macro- and microstrain scanning of bulk polycrystalline materials as well as to
distinguish individual phases of a multiple phase material having very close structure properties.
The latest instrumentation achievements exploiting neutron Bragg diffraction optics permits
considerably to improve the resolution properties of the future generation neutron diffractometers.
The effects of multiple Bragg reflections in a deformed single crystal can be observed when
more than one set of planes are simultaneously operative for a given wavelength i.e. when more
than two reciprocal lattice points are at the Ewald sphere. Using bent prefect crystals we
determined several strong double reflection processes on several pairs of lattice planes which are
mutually in the dispersive geometry (see upper part of Fig. 1). In relation to the value of bending
radius, the obtained doubly reflected beam has however a narrow band-width 'OO of 10-4 - 10-3
and 'T-collimation of the orders of minutes of arc. It is clear that in comparison with the
conventional single reflect-ion monochromators the monochromatic neutron current is lower
proportionally to a smaller 'O and 'T spread. New experimental studies of the multiple reflection
monochromator proved its possibility for using for high-resolution mono-chromatization. For our
test we used the double-reflection process based on two pairs of 153/1-3-1 and -31-1/513
reflections at O= 0.156 nm which was realized in a cylindrically bent Si-crystal set for diffraction
in symmetric transmission geo-metry [1]. Fig. 1 shows D-Fe diffraction profiles taken with a
diffractometer performance employ-ing this multiple reflection monochromator (without using any
Soller collimators and with a 2 mm wide sample) which demonstrates its re-solution abilities. The
profiles were taken by a PSD detector with 1.5 mm spatial resolution and situated at the distance of
1.42 m from the sample. Using only FWHM of the diffraction profiles, the resolution in 'd/d
calculated from our results is of about 8x10-4 in the vicinity for 2TS 90o. But an estimation of the
contribution coming from the width of the sample together with the detector spatial resolution
gives a comparable uncertainty value. Therefore, it is expected that the intrinsic instrument
resolution would be better than 5x10-4. Thus, it follows from the obtained results that at the high-
flux sources the strains can be measured with the sensitivity of 10-5 in a large range of scattering
angles.
Then, as a first step, the diffractometer equipped with this monochromator we used for
investigation of Fe-reflections in an induction hardened S45C steel rod ()=20 mm) having
different phase composition at different distances from the rod axis. The gauge volume was
determined by 2 mm wide slits in the incident as well as diffracted beam. Fig. 2 displays the
diffraction profile obtained at the distance 8 mm from the axis. Similarly, Fig. 3 displays the
diffraction profile obtained at the distance 6 mm from the axis. Thanks to the used high-resolution
monochromatic beam, after a fitting procedure we could reliably determined contributions of the
individual phases.
1330 P. Mikula and M. Vrana

Finally, it can be stated that diffractometers employing multiple reflection mono-chromator

can provide very high resolution at low monochromator take-off angles. The resolution can be
even comparable to back-scattering instruments. They can be efficiently used namely at high-flux
neutron sources. For adjustment of optimum parameters (the crystal cut, the thickness and the
bending radius) of the monochromator and of the per-formance of the whole scattering device
Monte Carlo simulations would be desirable [2].
These studies were supported by the projects AV0Z10480505 and MSM2672244501.

References
1. Mikula P., Vrana M. and Wagner V., Physica B, vol. 350 , e667-e670, 2004.
2. Saroun, J. and Kulda, J., 1997, Physica B, vol. 234-236, 1102-1103, 1997.
 40. Residual Stress and its Effects on Fatigue and Fracture 1331

SENSITIVITY OF PREDICTED RESIDUAL STRESSES TO MODELLING

ASSUMPTIONS

S. K. Bate1, R. Charles1, D. Everett2, D. O’Gara1, A. Warren1 and S. Yellowlees1

1Serco Assurance, 2Rolls-Royce
1Walton House, 404 The Quadrant, Birchwood Park, Warrington, Cheshire, WA3 6AT,

United Kingdom
steve.bate@sercoassurance.com

The treatment of residual stresses in welded components is considered in defect assessment

procedures such as R6 [1], BS7910 [2] and API579 [3]. R6 considers three approaches for
determining the as-welded residual stress distribution. The first approach, Level 1, is a simple
estimate of stresses which enable an initial conservative assessment of a defect to be made. The
second approach, Level 2, uses a published compendia that characterises bounding profiles for a
range of structures. The third approach, Level 3, entails the use of analysis coupled with
experimental measurements to define the detailed spatial distribution of residual stress. The
assessor should first try to apply Levels 1 and 2, however these may lead to insufficient margins in
the assessment. There may also be cases where a more comprehensive understanding of the
residual stress field is required and the Level 3 approach has to be used.
The use of analytical, mainly finite element, and experimental approaches to characterise weld
residual stresses is fairly widespread but the limitations to these methods are not always known
and therefore needs to be recognised. However, when the two approaches are combined, and the
results corroborate each other sufficiently well, the resulting residual stress distribution can be
confidently used for assessment.
The use of numerical techniques to simulate the welding process is not new and the increase in
computing power has seen the size and complexity of the models increase. Such features mean that
in some cases simplifications and assumptions have to be utilised to approximate residual stresses.
Thus there is an uncertainty in the accuracy of these methods which has meant that conservative
approximations, i.e. upper bound residual stress profiles, have still to be relied upon when carrying
out structural integrity assessments.
A programme of work is now underway to develop a procedure for residual stress prediction
which will account for how the various simplifications and assumptions affect the magnitude of
predicted stresses, and to identify the limitations of the various modelling techniques. This will
include heat source representation, material behaviour in terms of high temperature annealing,
cyclic hardening or softening, creep, and phase transformations. Supporting experimental data is
required in order to validate the predictions and a series of mock-ups will be manufactured to
enable measurements to be carried out.
Use will also be made of benchmarks which are the subject of round-robin exercises which
combine analytical predictions with various measurement techniques. Initial work has considered
an austentic single weld bead-on-plate specimen, see Fig. 1 below, which is one of the benchmarks
being studied within the NET European network.
1332 S. K. Bate et al.

FIGURE 1. Schematic of NET Benchmark Plate (17mm Thick)

This paper describes the analytical work that has been carried out by Serco Assurance which
looks at how various assumptions on modelling the bead on plate specimen affect the predicted
residual stresses. In each of the cases considered, a 3-dimenensional finite element representation
of the plate has been used for the analysis. The cases presented in this paper are:
Varying the geometric parameters defining the Goldak [4] heat source model.
Varying the weld efficiency.
The effect of different material hardening models and annealing.

References
1. R6 - Revision 4, Assessment of the integrity of structures containing defects, British Energy,
September 2000.
2. BS7910:1999, Guide on methods for assessing the acceptability of flaws in metallic
structures, BSI 10-2000.
3. Fitness-for-Service, API Recommended Practice 579, First edition , January 2000.
4. Goldak J., Chakravarti, A. and Bibby, M., Metallurgical Transactions, vol. 15B, 229–305,
June 1984.
 40. Residual Stress and its Effects on Fatigue and Fracture 1333

WELDING EFFECTS ON THIN STIFFENED PANELS

T. T. Chau
TECHNICATOME (Aix-en-Provence, France)
Centre Jean-Louis Andrieu BP34000
13791 Aix-en-Provence, FRANCE
Commission SNS / AFM (Paris, France)
CHAU.TUNGTHIEN@technicatome.com

The major problems due to welding effects are the residual stresses and deformations of which the
levels affect more or less the resistance and lifetime of welded structure.
In steel industry and particularly in shipbuilding, during these last few decades, thin plates are
used more and more in ship construction in order to lighten the structure weight. Unfortunately,
excessive distortions occurred on these thin stiffened panels and straightening works must be
executed in respecting the limit tolerance fixed by the Quality Standard of Ship Construction.
These futil works reduce Productivity and Quality, increase Construction Cost and get longer
Fabrication Delay. Thus, it is necessary to evaluate, control and minimize the distortion and stress
levels of thin welded panels before welding assembly operations.

FIGURE 1 : A Metallurgical Concept for Numerical Simulation of Arc Welding.

Carried out from RD studies in France (1991-1995), subsidized by the French Ministry of
Reseearch and Industry with the cooperation of French shipyards, a simplified methodology of
numerical simulation of arc welding has been developed and validated successively by
measurements on 2D and 3D samples [1, 2, 3]. Based on a metallurgical concept (Fig. 1), the
Methodology presents interesting advantages for uses: simplicity in modelling, rapidity on
computer time within reliable results.
In this paper, a short presentation of the methodology is described and its industrial
applications, in 1996, on two 3D FE models of thin deck and bulkhead panels of superstructure of
a Chemical parcel tanker in fabrication stage are presented.
Another application of the methodology has been realized in 2002 for a RD project with the
cooperation of the “Chantiers de l’Atlantique” shipyard (CAT). The methodology was successfully
applied and validated on a very large thin stiffened “Testing” Panel in full scale (17m x 6m x
6mm).
1334 T. T. Chau

The "Testing"Panel has been realized under real conditions of fabrication (by automation
welding chain) in the shipyard. A 3D FE model was created for calculation according to the
Methodology, representing the welded assembly of more than five hundred thousand of volume-
elements within more than two million of degrees of freedom .
The numerical results due to welding effects so obtained within short computer-time (three
hours and half) on a linear FEM software were verified with measured stress values and identified
with the buckling state of the “Testing” Panel before and after welding operations by
photographies (Photo 1).

PHOTO 1 : Deformed shape of the “Testing” Panel after assembly welding operations.

References
1. Chau, T.T. & Masson, J.C., Une méthode d’evaluation des contraintes et des deformations
residuelles dans les assemblages d’elements métalliques soudes d’epaisseurs moyennes,
Proc., ATMA 1992, 207-233.
2. Chau, T.T., Paradis, A., Masson J.C., A simple method for evaluating the 3D welding effects
on thin stiffened panel assemblies in shipbuilding, Proc., 3rd International Conference on
Marine Technology, ODRA’99, Szczecin, Poland, Marine Technology III, 1999, 485-530.
WIT Press: witpress@witpress.com
3. Chau, T.T., A metallurgical Concept for Numerical Simulation of Arc Welding, Proc. ASME
PVP 2005, 17-21 July 2005, Denver, Colorado, USA., paper 71654.
 40. Residual Stress and its Effects on Fatigue and Fracture 1335

EVALUATION OF RESIDUAL STRESSES IN CERAMIC POLYMER MATRIX

COMPOSITES USING FINITE ELEMENT METHOD

K. Babski, T. Boguszewski, A. Boczkowska, M. Lewandowska,

W. Swieszkowski* and K.J. Kurzydlowski
Division of Materials Design
Faculty of Materials Science and Engineering
Warsaw University of Technology
Woloska 141, 02-507 Warsaw, POLAND
Wojciech.Swieszkowski@materials.pl

Composites are complex materials which comprise of multiple phases (components) of different
thermal and mechanical properties. By combining such components, especially metals, ceramics,
glasses, and polymers, one can produce new functional and construction materials with superior
properties which could be tailored for the specific applications. However, due to mismatch in to the
properties of their constituents, composite are prone to build-up of residual stresses. This is in
particular true to the thermal stresses arising due to mismatch in thermal expansion coefficients.
These thermal stresses might be generated either during manufacturing process, more precisely
during the cooling from fabrication temperature, or due to thermal cycling in in-service conditions.
They, in general, can improve the properties of the composites [1]. However, they can also have
detrimental effect on their performance. This calls for their control over manufacturing and in-
service conditions.
The aim of the study was to evaluate the residual stresses in two types of composites polymer –
ceramic composites using the finite element method (FEM). The effect of the composite structure,
defined in terms of the volume fraction, size and shape of constituents, on the residual stresses
built up was also analyzed.
The ceramic - polymeric matrix composites were made of porous SiO2 ceramic infiltrated with
urea – urethane elastomer [2]. In the infiltration process liquid mixture of the substrates is
incorporated into ceramic pores using the vacuum pressure and temperature of 120oC. Since the
thermal expansions of the elastomer and ceramics are different upon cooling to ambient
temperature thermal stresses are generated. Moreover, the elastomer shrinks as a consequence of
its transformation from the mixture of substrates in the liquid to the solid states. These two
phenomena result in buildup of considerable residual stresses.
The bis-glycidylmethylmethacrylate (bis-GMA) polymeric matrix with the ceramic fillers,
which is used for dental restoration [3], was also investigated. As during the restoration the matrix
polymerizes, it shrinks and for large cavities the composite material might be pulled away from the
walls. This in turn, leads to restoration failure by de-bonding at the composite-tooth interface. The
fillers are added to the polymeric matrix to reduce the material shrinkage. The effect of the fillers
size, shape and arrangement on the residual stresses at composite-tooth and resin-ceramic filler
interfaces has been evaluated in the present study.
Both composite materials were analyzed by the finite element method in the Ansys software.
The models were 3D axi-symmetric. The Solid92 elements (which have 10 nodes) were used in the
FE analyses. A linear and isotropic properties have been assumed for the ceramic matrix composite
[2] as well as for the dental restoration [4]. For infiltrated composite (CMC) the simulations of
both thermal and tensile loading of material were carried out. The models were subjected to
thermal load simulating the cooling from fabrication (120oC) to room temperature (20oC),
1336 K. Babski et al.

followed by a tensile straining. For the dental restoration (polymer matrix composite (PMC)) the
polymerization shrinkage of the composite was modeled using temperature-dependent expansion.
The residual stresses were investigated in the composite itself as well as in the restoration-tooth
system.
The analysis of distribution of principal stresses in the CMC shows that change of temperature
leads to buildup of high tensile stresses in elastomeric phase and both tensile and compressive ones
stresses in the ceramic pre-form (Fig.1a). It was found that the thermal stresses present in
composite mostly reduce the maximum values of tensile stresses in ceramics. It can be
advantageous from the mechanical point of view and result in an increase of toughness.
The FE results obtained for the PMC show the effect of the resin shrinkage on the residual
stresses at the resin-ceramic filler interface, which can cause de-bonding (Fig.1b). Various fillers
have been examined in terms of the efficiency in the reduction of these residual stresses (Fig.1c).

a) b) c)
FIGURE 1. Residual stresses in: a) ceramic-elastomer composite, b) resin- ceramic composite and
c) composite – tooth system

In more general sense, the present studies show the potential application of the finite element
method in investigation of the residual stresses in different types of the composite materials. The
future work will be concentrated on the experimental validation of the numerical results.

References
1. Antheunis, V. et al. Dental Materials, 20, 554-564, 2004.
2. Boczkowska, A. et al. Procced. E-MRS 2004 Fall Meeting, Warsaw, 6-10. X. 2004, p. 225.
3. Boguszewski, T. et al. e-Polymers, 032, 2005.
4. Barink, P.C.P. et al . Biomaterials, 24, 1427-1435, 2003.
 41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) 1337

TOWARDS DATA-DRIVEN MODELING AND SIMULATION OF

MULTIPHYSICS DEGRADING SYSTEMS

J. G. Michopoulos and C. Farhat

Computational Multiphysics Systems Lab
Special Projects Group, Center for Computational Material Science, Code 6390.2
Naval Research Laboratory, Washington DC 20375 USA
john.michopoulos@nrl.navy.mil
Department of Mechanical Engineering &
Institute for Computational and Mathematical Engineering
Stanford University, Stanford CA 94305 USA
cfarhat@stanford.edu

Recently it is becoming more and more apparent that research and development (R&D) activities
in developed economies are driven by motivations asserted by the various stake holders involved
in the production and consumption processes. Producers are interested in tailoring R&D processes
to drive various economic metrics such as total cost of ownership and return on investment towards
their benefit. Consumers are interested in optimized utility-based metrics such as functionality,
reusability, safety and maintainability, thus forcing producers to pay attention into building these
properties into their R&D products. Furthermore, today’s cradle-to-grave engineering
requirements for validated, safe, economic and maximally functional design, flexible
manufacturing, qualification, certification, utilization and maintenance of system products have
significantly raised the demand for validated, efficient and quick simulation of the behaviour of
complex whole systems. In the particularly complex category of degrading systems that exhibit a
time-varying behaviour for the large time scale where aging and maintenance are critical, life
extension for usability purposes has become a focal area of interest. On the other hand,
simulations inherit all of their utility and economic properties from those of the analytical and
computational models they are based upon. Research conducted at NRL for the past forty years
within the context of the material constitutive behaviour characterization has been both
anticipating and exploiting the computational technologies evolution in a manner that is consistent
with the previously mentioned motivational drivers.
The present paper focuses on describing the utilization of these ever evolving computational
and mechatronic technologies, to automate data-driven model generation for multiphysics
degrading systems with emphasis on multi-field excitation of composites [1,2].
A brief exposure of NRL’s data-driven and mechatronically automated method [3] is
presented. This method is based on the concept of identifying the constitutive behaviour of
degrading composite material via parameter estimation associated with the dissipated energy
density function. Its extension to include temperature dependence in a multiphysics context is also
presented. The continuum multiphysics formulation for multi-field and multi-domain problems is
based upon the experimental determination of an energy density dissipation function that depends
on the multiphysics state variables of the system under investigation. The integrated over the
volume of the specimen is matched by the experimentally measured dissipated energy of the
system computed as the difference between the total energy imparted into the specimen system
minus the recoverable energy. This matching is implemented as an inverse problem solved via
standard global optimization techniques. The experimental values of dissipated energy are
computed from the displacements and tractions measured from the multi-degree of freedom testing
1338 J. G. Michopoulos and C. Farhat

m
machines created at NRL. The dissipated energy density function I for a given material m is
constructed according one of three possible combinations,
n
n
I m (T  H jk ) ¦c
i 1
i
m
(T ) F i ( H jk ) I m (T  H jk ) ¦ c im F i ( T , H jk )
, or i 1 , or
n
I m (T  H jk ) ¦c
i 1
i
m
\ i (T ) F i (H jk )
,
that encode the decomposition combinations between temperature and strain field variables
dependence of the corresponding basis function(s). The corresponding energy balance equation
used for each loading step of the pre-programmed loading paths is now extended from the previous
forms to be:

ur 1
³ tu q v d q v  tsu tu v  Q ³ I m ( T ( x j )  H ik ( x j )) d x j
0 2 wV
,

where tu ,uv , are the traction and displacement resultants along the pre-programmed path and Q
is the thermal energy imparted into the system.
Finally, the field equations for the multi-field and multi-domain problem of aero-thermo-
structural fluid-structure interaction are given and simulation results resulting from their space and
time integration by use of Stanford University’s AERO suite of codes [4] are presented as an
example for simulation-based design of a multiphysics degrading system such that of an aircraft.

References
1. Michopoulos J., Tsompanopoulou P., Houstis E., Farhat C., Lesoinne M., Rice J., Joshi A.,
“On a Data Driven Environment for Multiphysics Applications”, Future Generation
Computer Systems Journal special issue on PSEs, vol 21/6 pp 953-968, 2005.
2. Michopoulos J., “Mechatronically Automated Characterization of Material Constitutive
Respone”, Proceedings of the 6th World Congress on Computational Mechanics (WCCM-
VI), September 5-10 2004, Beijing China, Tsinghua University Press and Springer, pp. 486-
491, 2004.
3. Mast, P.W., Nash, G.E., Michopoulos, J.G., Thomas, R., Badaliance, R., Wolock, I.,
Theoretical and Applied Fracture Mechanics, vol. 22, 71-125, 1995.
4. Farhat C., Geuzaine, Brown R and G., “Application of a Three-Field Nonlinear Fluid-
Structure Formulation to the Prediction of the Aeroelastic Parameters of an F-16 Fighter,”
Computers and Fluids, vol. 32, pp. 3-29, 2003.
 41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) 1339

MATHEMATICAL MODELLING OF PIEZOCERAMIC TRANSDUCER

PERFORMANCE IN THE PRESENCE OF MATERIAL DEFECTS

Torben Amby Christensen1,2, Niels Lervad Andersen1, and Morten Willatzen2

1
Center for Product Development and 2Mads Clausen Institute of Product Innovation, University of Southern
Denmark, Grundtvigs Alle 150, DK-6400 Sønderborg, Denmark
torben@amby.org

We model the influence of air-void imperfections inside a piezoceramic PZT-5H material. Results show that
significant modifications of the frequency characteristics are observed with air voids increasing from 0 to 1 mm
corresponding to half of the piezoelectric material’s height. Similar conclusions are reached when the width is
increased for given height.
IMechatronic products become increasingly advanced involving smart materials and smart structures.
Piezo-materials are often used as a smart material for sensing and actuating. In addition, piezomaterials are
show fast response characteristics but suffer from limited range of motion. This means that practical
applications require sensors and actuators to perform near materials-strength limits with the possibility of
imperfection formation. In the present work, we examine the influence of air-void imperfections on the signal
characteristics of piezoelectric media which can be useful in the quality assessment of such materials.
Piezoelectric model:
The domain ȍp is modeled as PZT-5H ceramic including the coupling between mechanical and electric
effects. Stress equilibrium is assumed:  ’ ˜ T  B 0 where T is the stress matrix and B is mass times
acceleration. Furthermore, assuming zero free charge density, Gauss’s law states that ’ ˜ D 0 , where D is
the electric displacement. Consider the piezoelectric constitutive laws in stress-charge
E S
form: T kl c klmn S mn  e ikl E i and D i H ik  e ikl S kl , where S, E, c, e, and is strain, electric field,
stiffness, piezoelectric stress constant, and permittivity. Using these, the governing equations can be expressed
as functions of strain and electric potential. Neglecting magnetic effects implies that the electric field is
conservative: ’ u E = 0, thus it is possible to express the electric field as the gradient of a electric potential:
E ’V . Assuming small displacements, the normal and shear strain can be expressed as
S ii w u i w x i and S ji 
½ wu i wx j  wu j wxi  where ui and xi is the displacement and spatial
variable in the ith-direction. It is assumed that stresses and the electric field in the z direction can be neglected
& & iZ t
(i.e., Tzz = Tyz = Txz = Ez 0). Using a monofrequency excitation source : V  x , t  V  x e all other
2
variables are modulated with the same frequency in a linear approach. Hence, B i  U p Z u i , where U p
is the density for the piezoelectric material. Using the symmetries of the hexagonal system class 6mm which
PZT-5H belongs to the governing equations become:

c~13E v y  c~33E u x  e~33 V x x  c 55E u y  c 55E v x  e15 V y y U p Z 2 u

 ’ ˜T
’ ˜D 0
B½
¾
¿
Ÿ c 55E u y  c 55E v x  e15 V y x  c~11E v y  c~13E u x  e~31 V x y U p Z 2 v
e~31 v y  e~33 u x  H~3S V x x  e15 u y  e15 v x  H 1S V y y 0
where:
E
c~11
§¨ c E
© 11
 
E 2
 c12
E
c 22
·¸ ,
¹
E
c~13 c13E  c12E c 23E E
 E
c 22 , c~33
§¨ c E
© 33
 E 2
 c 23
E
c 22
·¸
¹
e~31 e 31  c12E e 32 E

c 22 , e~33 e 33  c 23E e 32 E

c 22 , H~3
S
H S
 
3  e 32
2 E
c 22  .
1340 T. A. Christensen et al.

The domain ̛ f is modeled as a fluid with no dissipative effects while assuming acoustic processes with
2
UfZ u ’p
small amplitude. Consider the Linear Euler equation: where U f u, and p is the fluid density,
displacement vector and the acoustic pressure respectively. Assuming irrotational flow, the displacement
2
vector can be expressed as the gradient of a potential: u ’I implying I p Z U f . In addition, the

2
 2
Linear wave equation for fluids: ’ p  U f c Z p , where c is the adiabatic bulk modulus, is employed
in the mathematical model framework. The latter equation can be reformulated in terms of the scalar potential
2 2
as follows: c ’ I  U f Z I 0 .

Boundary conditions

Numerical values:

Numerical Results:
We have modeled the situation where the left-most (right-most) interface faces water (air). In Figure 1, we
plot the center velocity of the aperture facing water as a function of frequency for five different heights of the
air void inside the piezoelectric medium. The width of the void is 50m for all height cases. The solid line shows
the frequency plot corresponding to the case without an air void in the piezoelectric medium. One pronounced
resonance peak is observed close to 0.75 MHz. If an air void with height 50m is present, the overall
characteristics mimic those of the perfect piezoelectric medium except that two small satellite resonances are
found at frequencies 0.94 and 1.4MHz. As the height of the air void increases these satellite peaks increase
substantially in their maximum values (note, however, that the maximum value decreases somewhat when the
height is approaching 1 mm). Moreover, the main resonance becomes increasingly red-shifted with air-void
height.
In Figure 2, we show the frequency plot of the center aperture velocity as a function of air-void width all
corresponding to the air-void height 0.2mm. Evidently, the presence of a 25 m width air-void does not lead to
significant disturbance of the main peak but, again, two satellite peaks show up at higher frequencies. As the
width increases further, the frequency of the main peak becomes red-shifted in agreement with the observations
in Figure 1.

Figure 1 Figure 2
 41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) 1341

A CONTINUUM APPROACH FOR IDENTIFYING ELASTIC MODULI OF

COMPOSITES

J. G. Michopoulos and T. Furukawa

Center of Computational Material Science
Naval Research Laboratory, Washington DC 20375 USA
john.michopoulos@nrl.navy.mil
School of Mechanical and Manufacturing Engineering
The University of New South Wales, UNSW 2052 Australia,
t.furukawa@unsw.edu.au

Composite materials can be tailored to meet specific needs such as high strength and stiffness
combined with low weight [1], which are not achievable by metallic and other materials. Recent
past decades have seen the accelerated use of composite materials in structural components due to
cost reductions. The performance of a composite structure is, however, highly sensitive to the
mechanical behaviour of the composite material. As the expectation of structural performance
becomes more demanding, the need for reliable prediction of the mechanical behaviour of
composite materials has become ever more important [2].
Despite a growing need, a basic obstacle to the comprehensive understanding of the
mechanical behaviour of composite materials is the complexity of their observed behaviour.
Unlike metallic and other isotropic materials where plastic deformation occurs by the slippage
between planes of atoms in the crystal grains of the material, the nonlinear mechanical behaviour
of composites is caused by a local decrease in stiffness, which results from an accumulation of
different types of microscopic damage [3]. Microscopic damage modes such as microcracks in the
matrix and debonding between the matrix and the reinforcement are common in all composites. In
continuous fibre-reinforced composites, failure can be also caused by the breakage and
microbuckling of the fibre and, due to fibre alignment, their mechanical behaviour is direction-
dependent, i.e. anisotropic. The nonlinear behaviour of laminar composites can further result from
delamination. As a consequence, composites fail in an extremely complicated manner [4] (in
contrast to homogeneous isotropic materials, the fracture of which occurs by the propagation of a
single macroscopic crack [5]), so that classical fracture mechanics is not applicable to modelling
failure behaviour in composites. Various macroscopic and microscopic approaches have been
proposed for failure modelling [6,7 and references therein] but, due to the complexity of the
systems, they mainly focus on simply structured laminates such as unidirectional or cross-ply
laminates under a specified set of uni-axial tests, leaving the prediction of failure behaviour of
complicated laminates such as angle-ply composites under various loading conditions as a further
issue.
The approach proposed by the authors, which is to be used for failure prediction of various
composites, quantifies the deformation, including damage, developed in a test specimen on a
continuum basis. Provided that experimental data from various loading tests, which describe the
full-field deformation of a continuum, and associated techniques to handle such a wide variety of
experimental data are available, failure behaviour of composites can be predicted irrelevant to the
type of composite and design of experiments. As the first step to the prediction of the failure
behaviour, this paper presents a technique to determine the elastic moduli of composites from
various full-field deformation tests of a continuum. The multi-DoF material testing machine used
by the authors enables the measurements of boundary displacement and force of a test specimen.
The elastic moduli are identified such that the total energy calculated from the boundary
displacement and force is equated with that computed from the stress-strain analysis by finite
1342 J. G. Michopoulos and T. Furukawa

element method. Figure 1 shows some results that compare the total energy simulated with the
identified moduli and the corresponding total energy by experiments. Detailed discussion of the
numerical analyses will be presented in the final paper.

FIGURE 1. Pareto-optimal solutions in parameter space (left: K  D , right: n  H ).

References
1. Dowling, N.E., Mechanical Behavior of Materials: Engineering Methods for Deformation,
Fracture, and Fatigue, Prentice-Hall, 1993.
2. Mast, P.W., Nash, G.E., Michopoulos, J.G., Thomas, R., Badaliance, R., Wolock, I.,
Theoretical and Applied Fracture Mechanics, vol. 22, 71-125, 1995.
3. Staab, G.H., Laminar Composites, Butterworth-Heinemann, Boston, 1999.
4. Badaliance R. and Hill, H.D., STP775 Damage in Composite Materials, edited by K. L.
Reifsnider, Philadelphia, PA., 229-242, 1982.
5. Matthews, F.L., Davies, G.A.O., Hitchings D. and Soutis, C., Finite Element Modelling of
Composite Materials and Structures, Woodhead Publishing, Cambridge, England, 2000.
6. Vinogradov, V. and Hashin, Z., Int. J. Solids Struct., vol. 42, 365-392, 2005.
7. Rikards, R., Abramovich, H., Green, T., Auzins J. and Chate, A., Mechanics of Advanced
Materials and Structures, vol. 10, 335-352, 2003.
 41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) 1343

REGULARIZED IDENTIFICATION OF MATERIAL CONSTANTS USING

MULTI-OBJECTIVE GRADIENT-BASED METHOD

T. Furukawa and J. G. Michopoulos

School of Mechanical and Manufacturing Engineering
The University of New South Wales, UNSW 2052, Australia
t.furukawa@unsw.edu.au
Center of Computational Material Science
Naval Research Laboratory, Washington DC 20375, USA
john.michopoulos@nrl.navy.mil

The past decades have seen an increasing complexity of nonlinear material models proposed as
scientific interests in materials grow and high performance materials are introduced. One of the
core processes in the material modelling is the parameter identification, where constants in a
material model are identified such that the model can describe material behaviour accurately. Due
to the non-linearity of the material model, the parameter identification problem is most commonly
converted into minimisation of an objective function describing errors between the measured and
computed outputs using an optimisation method. However, the problem of this iterative technique
is that the solution often cannot be obtained if measurement data and/or the model contain large
errors [Bard, 1974] as this makes the objective function complex.
The most common approach to tackle this problem is to add a regularisation term, which
normally consists of a function multiplied with weighting factors such as regularisation
parameters, to the objective function. This approach additionally requires a technique to find the
best set of weighting factors since the solution obtained depends upon the weighting factors. Such
techniques conventionally adopted include Morozov discrepancy principle [Morozov, 1984] and
the generalised cross validation [Groetsche, 1984]. Later, Reginska [1996] considered the
maximizer of the L-curve as the optimal parameter. Kubo, et al. [1998] also proposed a technique
using Singular Value Decomposition, while Zhuang and Zhu [1998] proposed a multi-time-step
method for inverse problems involving systems consisting of partial differential equations.
Despite their performance to some degree, the fundamental question common in all the techniques
is whether the automatic determination of a single solution is necessary, as the solution of the
inverse analysis will be never known in nature unlike the forward analysis. In addition, an
additional parameter must be often introduced to find the best regularisation parameter, the
solution being again dependent on the additional parameter.
Meanwhile, multi-objective optimisation methods have been proposed for solving multi-
objective design optimisation problems [Coello, 1999]. These methods allow the design
parameters to be optimised without weighting factors on design criteria such as weight and energy
consumption. The solution of this vector functional formulation is henceforth represented as a
space, namely the solution space, rather than a point, and the methods find a set of admissible
solutions in the solution space. Due to the derivation of multiple solutions and the possible
complexity of the objective functions, the methods are mostly based on the evolutionary
algorithms (EAs), which execute robust search from multiple search points for single objective
optimisation. Many EAs are, however, very robust at the expense of efficiency in contrast to the
conventional calculus-based methods, so that they are not inefficient for parameter identification
problems of concern, which have a relatively simple formulation. Moreover, since these
algorithms find only a fixed number of solutions in the solution space, the solutions are sparse and
not well distributed in the solution space.
1344 T. Furukawa and J. G. Michopoulos

In this paper, a technique for solving a regularised parameter identification problem without
weighting factors is first proposed. In this technique, regularisation terms are each formulated as
another objective function, and the multi-objective optimisation problem is solved by a multi-
objective optimisation method. Furthermore, Multi-Objective Gradient-based Method (MOGM) is
proposed as a multi-objective optimisation method to find the solutions for this class of problems
efficiently. The algorithm is also formulated such that its solutions can describe the solution space
to be derived.
Numerical and experimental results will be detailed in the final paper, but FIGURE 1 shows
the resulting Pareto-optimal solutions in some parameter spaces when parameters of Chaboche
model were identified. The well-distributed solutions obtained by the proposed technique, the
results indicate the total solution space of the identification problem. In addition, the resulting
Pareto-optimal solutions can be advantageously used to further investigate the sensitivity of the
solution space.

FIGURE 1. Pareto-optimal solutions in parameter space (left: K  D , right: n  H ).

References
1. Bard, Y., Nonlinear Parameter Estimation, Academic Press, New York, 1974.
2. Morozov, V.A., Methods for Solving Incorrectly Posed Problems, Springer-Verlag, New
York, 1984.
3. Groetsche, C.W., The Theory of Tikhonov Regularization for Fredholm Integral Equation of
the First Kind, Pitman, Boston, 1984.
4. Reginska, T., SIAM Journal of Scientific Computation, vol. 17, 740-749, 1996.
5. Hansen, P.C., SIAM Review, vol. 34(4), 561-580, 1992.
6. Kubo, S., Takahashi, T. and Ohji, K., In Inverse Problem in Engineering Mechanics, edited
by M. Tanaka and G.S. Dulikravich, 337-344, 1998.
7. Zhang, X. and Zhu, J., Inverse Problems in Engineering Mechanics, edited by M. Tanaka and
G.S. Dulikravich, Elsevier Science, 299-308, 1998.
8. Coello, C.A., Int. J. Knowl. Inform. Sys., vol. 1(1), 1999.
 41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) 1345

LOADING AND MATERIAL FEATURES INFLUENCE ON PIEZOELECTRIC

MATERIAL PERFORMANCE

V. G. DeGiorgi and S. A. Wimmer

System Design and Integration Section, Code 6353
Naval Research Laboratory, Washington, DC 20375, USA
degiorgi@anvil.nrl.navy.mil

Even though much research has been completed on constitutive relationships for piezoelectric
ceramics, there is generally the assumption that material and loading axes are aligned. Current
actuator designs focus on alignment of these axes. Natural flaws may occur at any orientation to
material and loading axes. The criticality of a flaw may depend directly on its relationship to the
primary loading direction. Understanding the interrelationship between flaw orientation, material
orientation and loading axes will have multiple benefits. It will allow for evaluation of flawed
components based on scientific understanding and it will allow for more diverse and novel designs.
The work presented examines the interaction between geometric flaws and misalignment
between load and material axes. We examine the relative merit of different geometric features, the
effects of material property axes rotation about the main loading axes (3-direction) and the effects
of material property axes rotation about the out of plane axes (2-direction). The ASTM E8 test
specimen has been customized in order to establish a complex stress state and to increase the
amount of data that can be obtained from a single test specimen [1]. A close up of the gage region
of the test specimen is shown in Fig. 1. Three types of loads are considered; electrical, mechanical
and mechanical followed by electrical. The current work is a computational study using finite
element methodology. Follow-on work planned includes experimental verification. The material
used in this study is PZT-5A, a commercially available piezoelectric ceramic.

FIGURE 1. Close up of gage region of test specimen.

Results of interest are the stress levels and displacement in the test specimen for each loading
condition. Detailed examination of these results will begin to provide insight into the
interrelationship between flaw size, flaw location, loading axis and material axes. Fig. 2 shows the
Von Mises stress levels for electrical, mechanical and combined electrical and mechanical loading.
As can be seen from the contours, there are significant differences between the responses for
electrical and mechanical loads.
1346 V. G. DeGiorgi and S. A. Wimmer

(a) (b) (c)

FIGURE 2. Von Mises stress contour plots for (a) mechanical, (b) electrical and (c) combined
mechanical and electrical loadings.
The next step is to rotate the material axes: the material is rotated about the 2-axis resulting in
a material axes that are rotated in the 1-3 plane with respect to the specimen geometry. The 2-
direction axis for material and loading path are identical in this case. The second rotation
considered rotates the material axes about the 3-axis so that the material is rotated in the 1-2 plane
with respect to the specimen geometry. The 3-direction axis for material and loading path are
identical in this case. In both cases only the material orientation is rotated, there is no change in
specimen geometry or load application. Variations in stress and deformation values are evaluted
to determine the impact of the different conditions.

References
1. Wimmer, S.A. and DeGiorgi, V.G. “Component Based Test Specimen Design”, Small
Specimen Test Techniques: Fourth Volume, ASTM STP 1418, p. 251-266, 2002.
 41. Computational Modeling of Multiphysics Degrading Systems (CMMDS) 1347

MODELING OF PLASMA CHEMICAL DEPOSITION AND DEGRADATION OF

SILICON THIN FILMS

Valeria V. Krzhizhanovskaya1,2, Peter M. A. Sloot1 and Yuriy E. Gorbachev2

1Section Computational Science, University of Amsterdam

Kruislaan 403, NL-1098 SJ Amsterdam, The Netherlands.

{valeria, sloot}@science.uva.nl
2St. Petersburg State Polytechnic University

Polytechnicheskaya 29, St. Petersburg 195251, Russia.

{lera, gorbachev}@csa.ru

Thin silicon films are important semiconductor material widely used in modern microelectronics
and solar cells. One of the technologies employed for industrial production of these films is plasma
enhanced chemical vapor deposition (PECVD). Modeling and simulation of this industrially
important technology is essential for optimizing physical and chemical parameters, reactive
chamber geometry and operating regimes of the installation, in order to reduce the costs of film
production and to provide a better growth rate and film quality in terms of layer composition and
homogeneity. For the latter, spatial processes (convection, diffusion, heat transfer, chemical
transformations and plasma non-uniformity), as well as the variation of system behavior in time,
play a decisive role. We have developed a 3D transient multiphysics multiscale model taking into
account all relevant chemical kinetics, plasma physics and transport processes that occur in the
bulk of a PECVD reactor and on the surface of the growing film [Krzhizhanovskaya, V.V.,
Zatevakhin, M.A., Ignatiev, A.A., Gorbachev, Yu.E., Goedheer, W.J. and Sloot, P.M.A., In
Proceedings of the 5th International Bi-Annual ASME/JSME Symposium on Computational
Technologies for Fluid/Thermal/Structural/ Chemical Systems with Industrial Applications,
ASME PVP-vol. 491-2, 59-68, 2004]. In addition to that we have built an efficient problem
solving environment for scientists studying PECVD processes and end-users working in chemical
industry [Krzhizhanovskaya, V.V., Sloot, P.M.A. and Gorbachev, Yu.E., Simulation: Transactions
of the Society for Modeling and Simulation International, vol. 81, No. 1, 77-85, 2005].
One of the most challenging and least studied topics in modeling a time-dependent film
deposition is the process of film degradation at the end of the production cycle, when the plasma is
‘switched off’ and the inflow of the source gases is stopped. Experimental analysis of processes
occurring at this stage is very difficult because the processes of plasma recombination, radicals'
diffusion towards the surface and their attachment to the film, are much faster than the time needed
for film characterization. Moreover, the measurements of the composition and homogeneity are
usually done ‘off-line’, after the film is taken out of the reactive chamber. In order to study the film
degradation processes computationally, we carried out a series of simulation experiments varying
the basic physical and chemical parameters (pressure, temperature, inflow rate, raw mixture
composition and chemical reaction rates), as well as the moments in time when the finishing stages
occur (plasma switch-off, gas inflow and heating shut-down).
In this paper we briefly introduce the previously developed 3D model of plasma chemical
deposition emphasizing the dissipative nature of the processes; then we present the latest
development for modeling the film degradation processes; describe the problem solving
environment created to provide distributed parallel multitask simulation and visualization on
computer Grids; and finally present the simulation results.
The model presented is an extension of the 1D [Gorbachev, Yu.E., Zatevakhin, M.A.,
Krzhizhanovskaya, V.V. and Shveigert, V.A., Technical Physics, vol. 45, No 8, 1032-1041, 2000],
1348 V. V. Krzhizhanovskaya et al.

2D [Krzhizhanovskaya, V.V., Zatevakhin, M.A., Ignatiev, A.A., Gorbachev, Y.E. and Sloot,
P.M.A., Lecture Notes in Computer Science, vol. 2328, 879-888, 2002] and 3D
[Krzhizhanovskaya, V.V., Zatevakhin, M.A., Ignatiev, A.A., Gorbachev, Yu.E., Goedheer, W.J.
and Sloot, P.M.A., In Proceedings of the 5th International Bi-Annual ASME/JSME Symposium on
Computational Technologies for Fluid/Thermal/Structural/ Chemical Systems with Industrial
Applications, ASME PVP-vol. 491-2, 59-68, 2004] models previously developed by the authors,
with a new description of chemical processes taking into account oligomer formation [Gorbachev,
Yu.E., Technical Physics, in press, 2005.]. The model is based on the numerical solution of the full
Navier-Stokes equations for transient laminar flows of viscous compressible multi-component
mixtures of chemically reacting gases. The boundary conditions take into account temperature
jumps, as well as molecular slipping and sticking processes on the walls. For simulation of
capacitively coupled plasma discharge, a fluid model [Nienhuis, G.J. and Goedheer, W.J., Plasma
Sources Sci. Technol., v. 8, 295-298, 1999] was used. The electron and ion continuity equations
were solved consistently with the Poisson equation for the electric field distribution.
We developed a “Virtual Reactor”, generic problem solving environment [Krzhizhanovskaya, V.V., Sloot,
P.M.A. and Gorbachev, Yu.E., Simulation: Transactions of the Society for Modeling and Simulation
International, vol. 81, No. 1, 77-85, 2005] with an advanced interaction capabilities integrating the basic
modules for reactor geometry design, computational mesh generation, plasma, flow and chemistry simulation,
as well as editors of chemical reactions and gas properties, databases, pre- and postprocessors, advanced
multimodal visualization modules, Web-interfaces and a Grid portal. The first steps in integrating distributed
resources through Grid technology promise efficient utilization of the computing power.
The model developed has shown a good ability to predict the main characteristics of PECVD processes
[Krzhizhanovskaya, V.V., Zatevakhin, M.A., Ignatiev, A.A., Gorbachev, Yu.E., Goedheer, W.J. and Sloot,
P.M.A., In Proceedings of the 5th International Bi-Annual ASME/JSME Symposium on Computational
Technologies for Fluid/Thermal/Structural/ Chemical Systems with Industrial Applications, ASME PVP-vol.
491-2, 59-68, 2004]. The Virtual Reactor allowed us to study the film degradation processes and to find the
featuring trends in system response to the variation of physical and chemical parameters and the operating
actions. We were able to indicate the mechanisms responsible for the irreversible deterioration of film
homogeneity and degradation of the layer composition at the finishing stage of the film production process.
Acknowledgements
The research was conducted with financial support from the Dutch National Science Foundation and the
Russian Foundation for Basic Research within the projects No 047.016.007 and 047.016.018, and with partial
support from the CrossGrid EU project IST-2001-32243 (www.eu-crossgrid.org) and from the Virtual
Laboratory for e-Science project (www.vl-e.nl).

References
1. Krzhizhanovskaya, V.V., Zatevakhin, M.A., Ignatiev, A.A., Gorbachev, Yu.E., Goedheer, W.J. and
Sloot, P.M.A., In Proceedings of the 5th International Bi-Annual ASME/JSME Symposium on
Computational Technologies for Fluid/Thermal/Structural/ Chemical Systems with Industrial
Applications, ASME PVP-vol. 491-2, 59-68, 2004
2. Krzhizhanovskaya, V.V., Sloot, P.M.A. and Gorbachev, Yu.E., Simulation: Transactions of the Society
for Modeling and Simulation International, vol. 81, No. 1, 77-85, 2005
3. Gorbachev, Yu.E., Zatevakhin, M.A., Krzhizhanovskaya, V.V. and Shveigert, V.A., Technical Physics,
vol. 45, No 8, 1032-1041, 2000
4. Krzhizhanovskaya, V.V., Zatevakhin, M.A., Ignatiev, A.A., Gorbachev, Y.E. and Sloot, P.M.A., Lecture
Notes in Computer Science, vol. 2328, 879-888, 2002
5. Gorbachev, Yu.E., Technical Physics, in press, 2005.
6. Nienhuis, G.J. and Goedheer, W.J., Plasma Sources Sci. Technol., v. 8, 295-298, 1999
 42. Scaling and Size Effects 1349

A FRACTAL APPROACH INTERPRETATION FOR THE INDENTATION SIZE

EFFECT

A. Carpinteri and S. Puzzi

Politecnico di Torino, Department of Structural and Geotechnical Engineering
Corso Duca degli Abruzzi 24, 10129 Torino, Italy
alberto.carpinteri@polito.it
simone.puzzi@polito.it

Since the 1950’s (Tabor [1]), the hardness measurements have been recognized to be size
dependent; a hardness increase with decreasing indentation depth (or indenter size) is always
observed. The indentation size effect (ISE) is extensively studied in the literature and research in
this field has been continuously growing in the last decades; this is motivated partly by the
development of nano-composites and the large-scale application of nanometer-thick films in
electronic components, partly by the availability of new methods of probing mechanical properties
in very small volumes. Several different mechanisms have been suggested to be responsible for the
ISE. Proposed mechanisms include: presence of oxides or chemical contamination on the surface,
interfacial friction, increased dominance of edge effects with shallow indents, indenter pile-up or
sink-in and loading rate (for a review, see Xue et al. [2]).
Recently, several Authors proposed a strain gradient theory, in order to explain the ISE.
Among them Stelmashenko [3], Fleck and Hutchinson [4], Nix and Gao [5], Gao and Huang [6],
Al-Rub and Voyiadjis [7]. This theory, based on Taylor’s hardening rule, incorporates in the
classical plasticity theory the material scale length that is needed to characterize and predict the
indentation size effect. This theory, in all its variants, assumes that the material deformation is
ruled by the density of statistically stored and geometrically necessary dislocations. The ISE arises
since the density of the statistically stored dislocations depends on the effective strain, whereas the
density of the geometrically necessary dislocations is a function of the strain gradient.
A different theory, that gained some success in explaining the ISE, is based on the surface to
volume ratio (S/V) and was proposed almost contemporarily by Gerberich et al. [8] and Zhang and
Xu [9]. This theory is based on the remark that the work done by an applied indentation load
contains both bulk and surface terms. The surface work, which is related to the surface stress and
the size and geometry of the indenter tip, prevails if the indentation depth is shallower than a
critical depth, while the bulk deformation prevails when the indentation depth is deeper than the
same depth.
In this paper, we propose an original interpretation of the ISE in single crystal and
polycrystalline metals, which is based on the experimental evidence of the formation of cellular
dislocation patterns during the later stages of plastic deformation, in the strain hardening of face-
centered cubic metals, both under tensile loading and in compression (Hähner et al. [10], Sžekely
et al. [11]). This dislocation patterns have also been explained theoretically by means of a
stochastic model for the evolution of the densities of dislocations (Zaiser [12]). When fractal
patterns are present, the dislocation cell structures are not appropriately characterized by any
regular, periodic or nearly periodic geometry. Thus, the microstructure cannot be characterized at
the meso-scale through an average dislocation density. In fact, the dislocation structure, which is
the damage domain, is a lacunar fractal of dimension D comprised between 2 and 3. In this case,
the same arguments already proposed by Carpinteri [13] to derive the Multifractal Scaling Law
(MFSL) for the tensile strength in brittle and quasi-brittle materials can be applied. As a
consequence, the hardness of metals, evaluated by micro- and nano-indentation, can be interpreted
1350 A. Carpinteri and S. Puzzi

by means of a power-law. Formally, the proposed equation is identical to the one already proposed
by Stelmashenko et al. [3] and Nix and Gao [4], but the underlying physical model is completely
different.
Eventually, some experimental hardness data from micro-indentation on copper have been
fitted with the MFSL, showing a very good agreement.

References
1. Tabor, D., The hardness of metals, Oxford University Press, Oxford, 1951.
2. Xue, Z., Huang, Y., Hwang, K.C. and Li, M., J. Eng. Mater. Technol. (ASME), vol. 124, 371-
379, 2002.
3. Stelmashenko, N.A., Walls, M.G., Brown, L.M. and Milman, Y.V., Acta Metall. Mater., vol.
41, 2855-2865, 1993.
4. Fleck, N.A. and Hutchinson, J.W., Adv. Appl. Mech., vol. 33, 295-361, 1997.
5. Nix, W.D. and Gao, H., J. Mech. Phys. Solids, vol. 46, 411-425, 1998.
6. Gao, H. and Huang, Y., Int. J. Solids Struct., vol. 38, 2615-2637, 2001.
7. Al-Rub, R.K.A. and Voyiadjis, G.Z., Int. J. Plast., vol. 20, 1139-1182, 2004.
8. Gerberich, W.W., Tymiak, N.I. and Grunlan, J.C., J. Appl. Mech., vol. 69, 433-442, 2002.
9. Zhang, T-Y. and Xu, W-H., J. Mater. Res., vol. 17, 1715-1720, 2002.
10. Hahner, P., Bay, K. and Zaiser, M., Phys. Rev. Lett., vol. 81, 2470-2473, 1998.
11. Szekely, F., Groma, I. and Lendvai, J., Mater. Sci. Eng., A, vol. 309–310, 352-355, 2001.
12. Zaiser, M., Mater. Sci. Eng., A, vol. 309–310, 304-315, 2001.
13. Carpinteri, A., Mech. Mater., vol. 18, 89-101, 1994; Internal Report, Laboratory of Fracture
Mechanics, Politecnico di Torino, N. 1/92, 1992.
 42. Scaling and Size Effects 1351

DESCRIPTION OF MULTI-SCALING POWER LAWS IN FRACTURE AND

STRENGTH

A. M. Korsunsky
Department of Engineering Science
University of Oxford, Parks Road, Oxford OX1 3PJ, United Kingdom
alexander.korsunsky@eng.ox.ac.uk

Scaling transitions and size effects in the fracture and strength of materials and structures have
particular significance in modern science and engineering, since the boundaries of mechanical
phenomena studied and devices exploited are being pushed both out, towards global scale
phenomena, and in, towards the nano-scale. These circumstances challenge the conventional
wisdom acquired over many decades of laboratory experiments and modelling at the engineering
scale (sub-mm to a few meters). More specifically, violations of the accepted scaling laws are
observed, requiring new physical deformation mechanisms to be proposed or identified, and new
modelling approaches to be developed and validated.

FIGURE 1. The Irwin et al. [2] data for net failure stress as a function of total slit length 2a, plotted
in bi-logarithmic coordinates, together with the ‘knee’ function describing the transition.

FIGURE 2. Paris fatigue crack growth diagram (Botvina et al., [3]) described by the ’knee’
function. The power law indicated by the straight line represents an intermediate asymptotic.
1352 A. M. Korsunsky

The present study aims to address a fundamental question of the efficient description of size
effects and scale transitions. To this end the functional description of multi-scaling power law
regimes is considered from first principles, and the functional form suitable for the task is
identified (Korsunsky [1]). The newly formulated ‘knee’ function is then applied to a variety of
experimental data containing manifestations of the size effect, including instances of two-criteria
failure strength (stress and toughness), fatigue crack growth threshold (Kitagawa-Takahashi
diagram) and its application in the context of fretting fatigue, Paris fatigue crack growth law,
indentation hardness of coated systems, etc.

References
1. Korsunsky, A.M. Physical Review B, submitted; e-print cond-mat/0508653.
2. Irwin, G.R., Kies J.A. and Smith H.L., ASTM STP, vol.381,1958.
3. Botvina, L.R., Yarema S.Ya., Grechko V.V. and Limar L.V. Physico-Chemical Mechanics of
Materials, vol.6, 41-50, 1981.
 42. Scaling and Size Effects 1353

THE SPALLING FAILURE AROUND DEEP EXCAVATIONS IN ROCK

MASSES

A. P. Fantilli and P. Vallini

Department of Structural and Geotechnical Engineering, Politecnico di Torino
Corso Duca degli Abruzzi 24, Torino – 10129 Italy
alessandro.fantilli@polito.it, paolo.vallini@polito.it

The process of spalling failure around a deep excavation in a rock mass is one of the main
problems in tunnel construction. It consists in a brittle failure, produced by the crushing of
compressed rock, during which strain localization appears as a shear crack. Both the tensile
damage and the confinement reduction, produced by the excavation of the tunnel, play a
fundamental role in the production of the spalling failure [1]. To better understand and to prevent
this failure, a plane strain model able to evaluate the structural response of the rock around the
excavation is here proposed. In particular, under the hypothesis of circular tunnel (of radius R0 ),
within an undefined rock mass having an initial hydrostatic pressure V0 , the stress VT in the
tangential direction is evaluated during the progressive reduction of the radial stress Vr (Fig. 1a).
This reduction represents the loss of confinement produced by the excavation.

FIGURE 1. Spalling failure of a rock mass after the excavation of a tunnel: a) geometrical
properties of the tunnel; b)-c) crushing of compressed rock cylinders.

In the proposed approach, the stress-strain VT-HT relationship of Figs. 1b-1c is considered for
the rock mass subjected to confinement stress Vr . As Fig. 1c shows, in the softening subsequent to
the compressive strength fT , a strain localization and, consequently, a shear crack appear in the
rock cylinder (Fig. 1b). Since this phenomenon is similar to that of compressed concrete observed
by Jansen and Shah [2], the VT-HT relationship already adopted for the crushing of compressed
concrete in RC beams is here considered. By means of this relationship the size effect of ultimate
bending moment, both with [3] and without [4] confinement produced by stirrups, has been
evaluated.
Regarding to a deep tunnel in a rock mass, the proposed model is here applied in two different
situations (case_1 and case_2). In case_1 the tunnel does not have any reinforced concrete RC
1354 A. P. Fantilli and P. Vallini

cover (R0=RC in Fig. 1a), whereas the cover is present in case_2 (RC/R0=1.1). In both cases, the
failure radii Rf , which define the failure zone produced by the tunnel excavation (Fig. 1), are
evaluated by changing the tunnel radius R0. Precisely, for R0<R<Rf the condition of spalling failure
(VT=0) occurs. The results of the two cases, and the related size effect, are reported in Fig. 2. The
ratio Rf/R0 remains equal to 1 (that is, the spalling failure does not exist) till R0 reaches the critical
value Rcrit . When R0=Rcrit , the failure zone rapidly increases (i.e., Rf/R0>1) similarly to an
unstable phenomenon. It must be remarked that the presence of a RC cover, despite its minimum
thickness, can increase the value of Rcrit and, consequently, the possibility of spalling failure is
drastically reduced.

FIGURE 2. The proposed model applied in a tunnel without (Case_1) and with (Case_2) a
reinforced concrete cover.

References
1. Diederichs, M.S., Rock Mech. Rock Engng, vol. 36(5), 339-381, 2003.
2. Jansen, D.C. and Shah, S. P., J. Engrg. Mech., ASCE, vol. 123(1), 25-35, 1997.
3. Fantilli, A.P., Iori, I. and Vallini, P., In Proceedings of the First fib Congress: Concrete
Structures in the 21st century, edited by fib, Osaka, 2002, 43-52.
4. Fantilli, A.P., Ferretti, D., Iori, I. and Vallini, P., J. Struct. Engrg. ASCE, vol. 128(5), 637-
645, 2002.
 42. Scaling and Size Effects 1355

SCALING IN MULTIAXIAL COMPRESSIVE FRACTURE

A. S. Elkadi and J. G. M. Van Mier

Microlab, Delft University of Technology
2628 CN Delft, Netherlands
Institute for Building Materials, Swiss Federal Institute of Technology
CH 8093 Zurich, Switzerland
A.Elkadi@tnodiana.com, Vanmier@ifb.baug.ethz.ch

Size effect studies under multiaxial-compressive loading are experimentally challenging as they
involve three-dimensional scaling, notably if cubical specimens are considered in poly-axial
testing. Laboratory experiments using model openings such as scaled hollow cylinders are suited
for such an investigation. The hollow cylinder geometry lends itself for providing permutations of
various multiaxial states of stress around its inner-hole depending on the stress path applied to its
external boundaries. They are also advantageous in terms of specimen boundary restraint and load
application procedure since radial pressures are applied using a fluid pressure along the
circumference of the cylinder, which is almost unrestrained. This test geometry is frequentely used
as a model test for stability studies of underground tunnels and deep boreholes.
In this study, an experimental investigation concerning size effect in concrete subject to
multiaxial compressive loading was carried out. Laboratory results are presented concerning the
size effect and fracture mechanisms in hollow-cylinders using two concrete model mixtures. The
mixtures were developed with analogous mechanical properties and different maximum aggregate
size; i.e. 4 mm and 2 mm, to accentuate the effect of microstructure and heterogeneity on material
behaviour under complex loading conditions and its size dependence (see Elkadi [1] for material
development and characteristics). Hollow-cylinders in a size range 1:4 were tested under external
hydrostatic pressures while monitoring their inner-hole deformations. A triaxial pressure-cell was
developed for these tests that is capable of accommodating the complete size range. The axial and
radial pressures could be applied independently on the samples and up to 40 MPa each (maximum
safe pressure) under process temperature of ±20° C. The samples are hollow-cylinders with inner-
diameters of 12.5, 25, and 50 mm and with a ratio of inner-diameter/outer-diameter/length equal to
1:4:6. In the remaining of this paper, samples with inner-diameter of 12.5, 25, and 50 mm are
referred to as A, B, and C, respectively.
A mild but significant size effect was observed in the strength of the hollow-cylinders from the
two mixtures. A consistent decease of hollow-cylinder strength with size (at tangential strain of
5‰ strain) was recorded up to 32% and 44% for the 4 mm and 2 mm mixtures, respectively. A
scatter in the size effect results was marked for all sizes from both mixtures. A power relation
could reasonably describe the observed size effect trends with an exponent of –0.19 and –0.26 for
the 4 mm and 2 mm mixtures, respectively. Fig. 1 presents the size effect results for both mixtures
as individual test values on a bi-logarithmic plot. At least six tests were performed per size per
mixture and their results are given in the figure.
1356 A. S. Elkadi and J. G. M. Van Mier

FIGURE 1. Individual values for log outer-stress versus log inner-hole diameter.

Failure around hollow cylinders from the two mixtures was observed to initiate through small
splitting, opening-mode, cracks along grain-boundaries oriented parallel to the maximum principal
stress. These cracks develop firstly close to the hole and under increase of load could develop
further deep in from the wall in a concentric pattern. Depending on the grain size, mineralogy, and
position with respect to orientation of crack growth, cracks could propagate either inter- or intra-
granularly parallel to the principal stress direction. Cracks were observed to preferably grow along
the spring-lines of the hole; however, occasionally they turn to meet the surface of the inner-wall.
Coalescence of these micro-cracks sub-parallel to the inner-wall resulted in the formation of thin
slabs, which then detached and started spalling into the hole. The fracture process seemed to
sequentially repeat itself with new extensional cracks of small width forming behind spall zones,
which are expected to extend later on forming new slabs of spalling potential.

References
1. Elkadi, A.S., van Mier, J.G.M., de Pater, C.J., In Proceedings 4th International Ph.D.
symposium in Civil Engineering, edited by P. Schiel, N. Gebbeken, M. Keuser and K. Zilch
J.C. Smith, Springer-VDI-Verlag GmbH, 2002, 135-142.
 42. Scaling and Size Effects 1357

FRACTURE OF ANTARCTIC FY SEA ICE

J. P. Dempsey1, S. Wang1 and D. M. Cole2

1Clarkson University, Potsdam, New York, USA
2US Army ERDC-CRREL, Hanover, New Hampshire, USA

jdempsey@clarkson.edu

The break-up of sea ice in McMurdo Sound has been investigated during two field trips in the fall
of 2000 and 2001 via in-situ cyclic loading and fracture experiments. Details regarding the test site,
setup and methods used in the in-situ experiments have been reported [1,2]. A viscoelastic
cohesive crack model [3,4,5] is being used to model the cracking behaviour observed and
measured on 5m and 10m square test specimens [6,7]. Bilinear stress-separation curves for the
different tests are being back-calculated for first-year Antarctic sea ice. The evolution of these
curves during crack growth is under investigation.
The in-situ experiments conducted during the October-December 2000 and 2001 field trips
were located on McMurdo Sound approximately 3 km offshore from Cape Barne on Ross Island.
The first-year sheet was predominantly congelation ice with a top layer of granular ice, with
evidence of platelet ice at depth. The experimental setup is shown in Fig.1, which details the crack-
opening-displacement measurement and flatjack locations, the pre-sawn crack, and a schematic of
the evolving crack front. The dimensions are in cm. As the side view shows, the ice thickness for
test A2-SP8 was 142 cm.
Note that the fracture behavior of this first-year sea ice in McMurdo Sound was studied by
measuring the crack-opening-displacements at the surface only (see Fig. 1). The crack front does
not propagate as a vertical straight line, in fact it seems to propagate in a staggered bottom portion,
top portion, bottom portion, and so on. It is an especially difficult matter to determine what the
critical separation is in each test. In other words, to determine what the critical CTOD is for growth
of the traction-free crack to initiate. This type of information is deduced by examining the crack-
opening-displacement at locations on both the pre-sawn crack and on locations ahead of the
cracking itself (locations NCTOD+, F1, F2, and F3).
In the viscoelastic cohesive crack model being used [3,4,5],, the bulk specimen material
behaviour is modelled as linearly viscoelastic. Both the stress-separation curve and the viscoelastic
characteristics of the sea ice are unknown at the outset. This information is back-calculated such
that predicted results match the experimental results. This research has led to certain conclusions:
the viscoelastic cohesive crack model and experimental-predictive back-calculating methodology
can be applied to the fracture of floating first-year sea ice in the Antarctic (as well as the Arctic);
the stress-separation curve for first-year Antarctic sea ice is bilinear; the critical CTOD is larger
for warmer weather whilst the tensile strength is smaller; during stable crack growth, the fracture
energy increases while the length of the fracture process zone decreases.
1358 J. P. Dempsey et al.

FIGURE 1. The test configuration for A2-SP8: (a) plan view, (b) side view.

References
1. Cole, D.M., Dempsey, J.P., Kjestveit, G., Shapiro, S., Shapiro, L.H. and Morley, G.M., In
Proceedings of the 16th IAHR International Symposium on Ice, edited by Vernon Squire and
Pat Langhorne, University of Otago, Dunedin, New Zealand, 2002, vol. 1, 474-481.
2. Cole, D.M. and Dempsey, J.P., J. Cold Regions Eng., vol. 18, 155-174, 2004
3. Mulmule, S.V. and Dempsey, J.P., J. Eng. Mech., vol. 123, 870-877, 1997
4. Mulmule, S.V. and Dempsey, J.P., Mech. Time-Dependent Mat., vol. 1, 331-356, 1998
5. Mulmule, S.V. and Dempsey, J.P., Mech. Cohesive-Frictional Mat., vol. 4, 505-524, 1999
6. Dempsey, J.P., Cole, D.M., Shapiro, S., Kjestveit, G., Shapiro, L.H. and Morley, G.M., In
Proceedings of the 17th International Conference on Port and Ocean Engineering Under
Arctic Conditions, edited by S. Løset, B. Bonnemaire and M. Bjerkås, Norwegian University
of Science and Technology, Trondheim, Norway, 2003, vol. 1, 51-60..
7. Dempsey, J.P., Mu, Z. and Cole, D.M., In Proceedings of the 17th IAHR International
Symposium on Ice, edited by VNIIG, St. Petersburg, Russia, 2004, vol. 2, 299-306.
 42. Scaling and Size Effects 1359

MIXED MODE FRACTURE OF BRICKWORK MASONRY

J. C. Galvez1, E. Reyes2, M. J. Casati2, J. M. Sancho3, J.Planas1 and D. A. Cendon1

1E.T.S. Ingenieros Caminos, Universidad Politecnica Madrid

c/ Profesor Aranguren s/n, 28040 Madrid, Spain

2E.T.S. Ingenieros Caminos, Universidad Castilla La Mancha

Av. Camilo Jose Cela s/n, 13071 Ciudad Real, Spain

3E.T.S. Arquitectura, Universidad Politecnica Madrid

c/ Juan de Herrera 4, 28040 Madrid, Spain

jcgalvez@caminos.upm.es

The brickwork masonry, especially walls, often shows cracking ought to differential settlements
and excessive deflections of the slabs. In these circumstances the cracks are induced by a
combination of tensile and shear stresses, under static loading. Until now the study of the
brickwork masonry failure under static loading has been focussed on the compression and
compression/shear failure mechanisms, and minor effort has been devoted to study the tensile/
shear failure (mixed mode I/II fracture) [1, 2]. This is the reason because there are not enough
experimental data for a good knowledge of the mixed mode fracture of the brickwork masonry.
The experimental results are needed for a better knowledge of the failure mechanisms of the
brickwork masonry under tensile/shear loading and to supply a benchmark to validate the
analytical and numerical models for the mixed mode fracture of the brickwork masonry.
This paper shows a series of brickwork masonry tests under tensile/shear loading. The tests
were performed with two testing geometries: the three point bending test with non-symmetric
notch (Fig.1), and the double-edge notched specimen (Fig 2). Both testing geometries had been
proposed to study the mixed mode fracture of concrete, and satisfactory results had been achieved
by the authors [3].
To take into account the influence of the orientation of the bed joints in the direction of fracture
of the brick masonry, specimens with eight different orientations of the layers, respect to the notch
plane, were tested: 0, 30, 45, 60, 90, -30, -45 and –60 degrees. Besides, to study the size effect
specimens with two similar sizes were tested.
This paper also deals with the numerical simulation of the mixed mode fracture of the
brickwork masonry. Traditionally, the numerical simulation of the mixed mode I/II fracture in
brickwork masonry has been approached with damage models, using macroscopic models, or with
the use of interface elements, in the case of using discrete crack models, from a microscopic
approach. Both procedures had been successfully used by several authors, specially when they
include a cohesive model fracture. The main obstacle when finite elements with cohesive crack
models are used, is the need of remeshing the model at the same time the crack grows. In previous
works [4, 5] the authors has been able to avoid this inconvenience, by using the prediction of the
crack path that provides the Elastic and Lineal Fracture Mechanics, and incorporating interface
elements with a cohesive model along the fissure. This model were implemented in a commercial
finite element code and has been achieved satisfactory results.
Recently has been developed a finite element with a embedded discontinuity based in the
cohesive crack model. The main advantage that this element gives is to avoid the remeshing, or the
previous computation of the crack path, because the crack grows through the embedded
discontinuity elements at the same time the analysis. The model extends the embedded
discontinuity element model proposed by the authors for isotropic cohesive materials to
1360 J. C. Galvez et al.

anisotropic materials. The numerical procedure is implemented in a commercial finite element

code by means of an user material subroutine. This paper presents a summary of the finite element
formulation, and the numerical prediction given by this elements in comparison with the
experimental results done by the authors. The numerical models only includes parameters with
physical meaning and measurable by standard methods.

Figure 1. Testing arrangement, geometry Figure 2. Testing arrangement, geometry

and dimensions of the TPB specimens. and dimensions of the double-edge notched
specimens.

References
1. Lourenco, P.J., Computational Strategies for Masonry Structures, Ph.D. Thesis, 211 pages.,
Delft, The Netherlands, 1996
2. Pande, G.N., Middleton, J. and Kralj, B. (eds.) Computer Methods in Structural Masonry - 4,
E & FN Spon, New York, 343 pages., 1998.
3. Cendon, D.A., Galvez, J.C., Elices, M. and Planas, J., International Journal of Fracture, vol.
103, 293-310, 2000.
4. E. Reyes, M.J. Casati y J.C. Galvez, In Fracture Mechanics of Concrete Structures V, edited
by V.C. Li, C. K. Leung, K. Willam and S. Billington, USA, Vol. 1, 415-422, 2004.
5. Galvez, J.C., Reyes, E. and Casati, M.J., In 6th International Masonry Conference, edited by
Brithish Masonry Society, London, 184-191, 2002.
 42. Scaling and Size Effects 1361

GEOMETRIC SCALING AND INSTABILITY IN FRP-CONCRETE DEBONDING

Kolluru V. Subramaniam1, M. Ali-Ahmad2 and C. Carloni3

1City College of City University of New York

Civil Engineering Department

New York, NY 10031, USA
ksubram@ce.ccny.cuny.edu
2City University of New York
3University of Bologna

Developing rational design guidelines for application of FRP strengthening techniques to concrete
beams requires a fundamental understanding of the shear debonding phenomenon and the
associated failure mechanism. Several finite element analyses have been performed to study the
shear stress transfer across the FRP-concrete interface and several analytical formulations based on
the classical section analysis have been proposed for predicting the ultimate load of a strengthened
beam These previous analyses however did not explicitly consider the material law of the
interfacial bond between concrete and the FRP. Recently, researchers have developed various
constitutive laws that adequately model the FRP-concrete bond behavior during debonding (Ali-
Ahmad et al. 2005, Leung 2004, Savoia et al. 2003). Fracture-based analytical and numerical
formulations have also been proposed to predict the initiation of shear debonding of FRP from
concrete surface (Taljsten 1996, Wu and Niu 2000). However, the full understanding of the
uncontrolled failure in the strengthened beams and influence of size on the observed failure
produced by shear debonding, is still emerging.
The debonding mode of failure, which is observed in concrete beams strengthened using
externally attached FRP sheets, is analyzed. The debonding failure, which results in separation of
the FRP from the concrete substrate, is produced by a crack propagating through the interface
region. Shear debonding is studied as a Mode-II fracture problem. In this paper the scaling of the
load carrying capacity of a given FRP-concrete interface as a function of the width of the FRP
composite laminate is experimentally investigated using a direct shear specimen as shown in
Figure 1. Direct shear tests were performed on FRP composite laminates of varying widths which
were attached to concrete blocks. In the experimental program, the ratio of the width of the FRP to
that of the concrete was varied for a fixed bonded length of the FRP. During each test, spatially
continuous measurements of the surface displacements on the FRP and concrete were obtained
using an optical technique known as digital image correlation. The contribution of the edge effects
to the overall strain distribution and hence the scaling in the load carrying capacity with the width
of the FRP composite laminate is studied. The stress distributions in the FRP during progressive
debonding, for different widths of FRP composite, are presented. Finally, the interfacial material
law and the fracture parameter GF are determined for the geometric scaling produced by varying
the width of the FRP composite.
1362 K. V. Subramaniam et al.

Figure 1: Direct shear test geometry to study debonding of FRP from concrete

A numerical analysis of the direct-shear response of FRP attached to a concrete substrate is

performed to study the influence of bonded length on the debonding failure. The material response
of the bi-material interface, which includes post-peak softening, is considered in the numerical
model. An instability in the load response associated with snap-back is predicted close to failure
and the arc-length method is used to obtain the entire load response. It is shown that there is a
scaling in the load capacity when the bonded length does not allow for the establishment of the full
stress-transfer zone associated with interface crack growth. From the results of the numerical
analysis, a fundamental understanding of interfacial crack propagation and the failure instability of
concrete members strengthened with externally bonded FRP is developed. A simple energy based
formulation demonstrates that the instability at complete debonding would lead to an explosive
failure associated with a sudden release of energy.

References
1. Ali-Ahmad, M., Subramaniam, K., and Ghosn, M. ASCE, Journal of Engineering Mechanics,
accepted for publication
2. Leung, C.K.Y. FraMCoS-5 conference, pp. 12-16, 2005
3. Savoia, M., Ferracuti, B. and Mazzotti, C., FRPRCS-6 Conference Proceedings, Singapore,
pp. 1-10, 2003
4. Taljsten, B., International Journal of Fracture, 82, pp. 253-266, 1996
5. Wu, Z. and Niu, H., J. Materials, Conc. Struct., Pavements, JSCE, 49(662), pp. 231-245,
2000.
 42. Scaling and Size Effects 1363

A SIMPLIFIED MCFT FOR SHEAR CAPACITY SCALING OF R/C BEAMS

M. T. Kazemi and V. Broujerdian

Associate Professor, Ph.D. student
Department of Civil Engineering, Sharif University of Technology, Tehran, Iran
Kazemi@sharif.edu, Broujerdian@mehr.sharif.edu

It is well known that shear strength of beams without stirrups shows a strong size effect. At
present, several formulas of size effect are used by some codes and suggested by various
researchers. Most of those formulas are not in agree with fracture mechanics. Recently, a simple
formula based on modified compression field theory, MCFT, was proposed (Collins and Kuchma
[1], Lubell et al. [2]), which shows a size effect stronger than the size effect formula based on
fracture mechanics, proposed by Bazant (e.g. Bazant and Kazemi [3], Bazant and Yu [4]). In this
paper it is shown that with some refined considerations, formulas are resulted in agree with
Fracture Mechanics concept. In the derived formulas, the effects of other parameters of the R/C
beams are present, which are justified based on Walraven’s research on aggregate interlock [5],
simplified MCFT, and subcommittee 445-F of ACI- ASCE [6] data bank.
Calculating the aggregate interlock capacity across the dominant diagonal crack of the beam,
some similar to the simplified MCFT procedure but considering more refined assumption for crack
opening profile, a new expression for shear capacity of reinforced concrete beams without stirrups
was derived that displays size effect in accordance to fracture mechanics, asymptotically. By more
simplification of the derived formula the following simple formula was obtained:

vc 1
v0 1  d / d0 , v B fcc , d 0 Ad c (1)
0

where vc Vc / bw d is nominal failure shear strength in MPa , Vc is shear capacity of beam in

N , d is beam effective depth in mm , bw is web width in mm , fcc is concrete compressive

strength in MPa , A and B are two constants determined from data fitting, and d c is beam
characteristic length defined as:

UE s ( d a  16)
dc

100 a / d  f cc
(2)

in which U As / bw d is longitudinal reinforcement ratio, As is cross-sectional area of the bars

in mm 2 , E s is modulus of elasticity of steel in MPa, d a is maximum aggregate size in mm

( da 17 mm if it is unknown), and a / d is shear span-to-depth ratio which may be calculated as
Vu d / M u and is not to be taken greater than one. M u and Vu are factored moment and shear
force at the section.
Equation (1) has the same general form of the size effect formula obtained from fracture
mechanics and dimensional analysis arguments and verified by various test results [3, 4], but with
1364 M. T. Kazemi and V. Broujerdian

some different definitions for v0 and d 0 . Fig. 1 represents the best fitting of ACI-445F database
[6] by Equation (1) and the curve obtained by a 5% cut-off to provide additional safety margin for
design. In Fig. 1 the ACI’s simple formula [7], shown with dashed line, is comprised with the
proposed relation.

FIGURE 1. ACI-445F database and shear strength formulas.

As seen from Equations (1) and (2), the proposed formula captures the dependence of size
effect parameters, v0 and d 0 , on beam parameters such as fcc , U , a / d , and d a , in a more
theoretical basis. These contributions were verified by some test series that each of them focused
on one parameter. The higher the beam material-geometrical characteristic length, d c , the lower
the brittleness of shear failure.

References
1. Collins, M.P., and Kuchma, D., ACI Structural Journal, vol. 96, 482-490, 1999.
2. Lubell, A., Sherwood, T., Bentz, E., and Collins, M.P., Concrete International, vol. 26, 67-78,
2004.
3. Bazant, Z.P., and Kazemi, M.T., ACI Structural Journal, vol. 88, 268-276, 1991.
4. Bazant, Z.P., and Yu, Q., Designing Against Size Effect on Shear Strength of Reinforced
Concrete Beams without Stirrups, Northwestern University, Report No. 03-02/A466s, 2004.
5. Walraven, J.C., Proceedings, ASCE, vol. 107, 2245-2270, 1981.
6. Subcommittee 445-F of ACI-ASCE committee 445, Summary of the Evaluations of the
Proposals for the Quick Fix for Reinforced Concrete Members without Transverse
Reinforcement, 2003.
7. ACI Committee 318, Building Code Requirements for Structural Concrete (ACI 318-02) and
Commentary (ACI 318R-02), American Concrete Institute, 2002.
 42. Scaling and Size Effects 1365

INTERPLAY OF SOURCES OF SIZE EFFECTS IN CONCRETE SPECIMENS

M. Vorechovsky and D. Matesova

Institute of Structural Mechanics, Faculty of Civil Engineering,
Brno University of Technology
Veveri 95, 602 00 Brno, Czech Republic
vorechovsky.m@fce.vutbr.cz, matesova.d@fce.vutbr.cz

The paper studies well published results of direct tensile tests on dog-bone specimens with rotating
boundary conditions performed by van Vliet and van Mier [1,2,3,4,5,6,7]. In particular we are
interested in the series of dry concrete specimens A to F (dimension D varying from 50 to 1600
mm, see Fig. 1); a series accompanied by verification tensile splitting tests. The paper attempts at
explaining of the complex size effect on mean and variance of nominal strength by combination of
random field simulation of local material properties, “weak boundary“ effects and a nonlinear
fracture mechanics software based on a cohesive crack model.
There has been spent much effort on different explanations of experimentally obtained size
effect on strength from several different points of view. Firstly, the effect of a non-uniform
distribution of strains in the smallest cross-section was studied with simple linear constitutive law
[2,3] and a separation of structural and material size effects was commented. The results were also
compared to the Weibull theory [8] based on the weakest-link theory which was found to fit the
mean nominal strength of sizes B to F [1,2,3,4,5]. Secondly, the effect of Gaussian stress
fluctuation with non-uniform loading was studied by Dyskin et al. [6] and the developed model
employing a limiting distribution of independent Gaussian variables with linear trend agrees with
the data very well. Van Mier and van Vliet also compared the data to “Delft lattice model” using a
simple local elastic-brittle material with regular and random lattices with good results.

FIGURE 1. Dog-bone specimens tested by van Vliet and van Mier: series A to F.

In this paper, the authors attempt firstly to explain the mean size effect curve by deterministic
effects (not assuming the local material strength random). The major contribution to non-
uniformity of the nominal strength is the “energetic-deterministic” size effect (caused by a constant
FPZ size in all beams), see e.g. Bažant and Planas [9]. It partly explains the decreasing slope of the
mean size effect curve (MSEC) in double-logarithmic plot (nominal strength versus characteristic
size). However, the strong decrease of the mean strength of the smallest specimen A is believed to
be sufficiently captured by modelling a weak surface layer of thickness of about 8 mm. This
thickness roughly corresponds to the maximum aggregate size dmax = 8 mm reported in [1]. A
parametric study of the influence of a “weak layer” thickness and the percentage reduction of the
layer strength compared to the bulk strength will be presented with regard to resulting MSEC. It
1366 M. Vorechovsky and D. Matesova

will be shown that these features of the model are able to explain majority of the experimentally
measured MSEC.
Next, the authors consider the local material strength as autocorrelated random field attempting
at capturing the whole size effect, scatter inclusive. A noticeable fact shown by Voechovský and
Chudoba [10] is that such an approach has the large size asymptotic behaviour coinciding with
Weibull while the small size asymptote deviates strongly from Weibull (the small size mean
nominal strength does not lie on the Weibull straight line in double-log plot, but on the horizontal
asymptote, MSEC is curved between two asymptotes with transition centred at the autocorrelation
length). The last important fact taken into account by the authors is that the beam of the smallest
size A has a considerable out-of-plane thickness leading to plane strain condition rather that plane
stress and most importantly the crack front does not cross the whole section but cracks can initiate
from one face of the beam only, see Fig. 1.
The computational approach represents a marriage of advanced computational nonlinear
fracture mechanics with simulation techniques for random fields representing spatially varying
material properties [10].

References
1. van Vliet, M.R.A. and van Mier, J.G.M., In Proceedings of FRAMCOS-3, edited by H.
Mihashi and K. Rokugo, Aedificatio Publishers, Japan, 1998, 1923-1936.
2. van Vliet, M.R.A. and van Mier, J.G.M., International Journal of Fracture, vol. 95, 195-219,
1999
3. van Vliet, M.R.A. and van Mier, J.G.M., Engineering Fracture Mechanics, vol. 65, 165-188,
2000.
4. van Vliet, M.R.A. and van Mier, J.G.M., Heron, vol. 45, 91-108, 2000.
5. van Mier, J.G.M. and van Vliet, M.R.A., Engineering Fracture Mechanics, vol. 70, 2281-
2306, 2003.
6. Dyskin, A.V., van Vliet, M.R.A. and van Mier, J.G.M., International Journal of Fracture,
vol. 108, 43-61, 2001.
7. van Mier, J.G.M. and van Vliet, M.R.A., Engineering Fracture Mechanics, vol. 70, 2281-
2306, 2003.
8. Weibull, W., Proceedings of the Royal Swedish Institute, Engineering Research No.151,
Stockholm, Sweden, 1939
9. Bazant, Z.P. and Planas, J. Fracture and Size Effect in Concrete and other Quasibrittle
Materials, CRC Press, Boca Raton, Florida, 1998.
10. Vorechovsky, M and Chudoba, R., International journal of Solids and Structures, accepted,
2005.
 42. Scaling and Size Effects 1367

SCALE EFFECT IN ELASTIC AND STRENGTH PROPERTIES OF

NANOSTRUCTURES

Olga S. Loboda1, Anton M. Krivtsov2 and Nikita F. Morozov3

1Institute for Problems in Mechanical Engineering RAS

Work address: Bolshoy pr. V.O., 61, 199178, St. Petersburg, Russia
loboda_o@mail.ru
2Institute for Problems in Mechanical Engineering RAS

Work address: Bolshoy pr. V.O., 61, 199178, St. Petersburg, Russia
krivtsov@AK5744.spb.edu
3Institute for Problems in Mechanical Engineering RAS

Work address: Bolshoy pr. V.O., 61, 199178, St. Petersburg, Russia
morozov@nm1016.spb.edu

Rapid development of nanotechnologies in the last decades demands thorough analysis of

mechanical properties of the objects at the nanosize scale level. Natural discreteness of the
nanoobjects requires new methods involving consideration of their properties from both discrete
and continuum points of view. In this work we apply this approach to analysis of mechanical
behaviour of close-packed nanocrystals having finite number of atomic layers in some directions
and infinite in others (2D nanostripes, 3D nanoplates and nanobars). In the current talk we will
focus on elastic and strength properties of nanocrystals being studied using both static and dynamic
theories. The scale effect (dependence of the properties on the number of atomic layers) is one of
the main topics of interest for this investigation.
Analysis of the elastic properties we start from static consideration of 3D-nanocrystal with a
face-centered cubic lattice. Two variants are considered: nanoplate (finite number of layers in one
direction and an infinite number in other directions) and nanobar (finite number of layers in two
directions and an infinite number in one direction). Uniform deformation of the discrete
nanocrystal is studied, which allows obtaining analytical formulae for the elasticity constants —
see Fig.1. The Poisson coefficient and Young modules are determined for three directions and the
dependence of their values on the size of the nanocrystal is investigated. It is shown that elasticity
tensor of nanocrystal is not symmetric. The symmetry occurs only for those directions in which the
crystal is infinite. The size of the crystal in the direction with the finite number of layers is
determined ambiguously, which leads to the ambiguousness in determination of its mechanical
properties. The dependence of elasticity modules is determined according to the thickness of the
crystal in different directions. These results are compared with the previous study of the elastic
properties for 2D nanostripe.
1368 O. S. Loboda et al.

FIGURE 1. Nanocrystal, [111] direction, gray (black) – upper (lower) layer.

For dynamical analysis of the elastic properties we consider longitudinal and bending
vibrations of a 2D nanocrystal stripe having square and triangular lattice. The vibration spectrum
and dispersion relations are obtained. On the other hand the spectrum is determined by the
continuum theory of rods where longitudinal and bending vibrations are considered. The
eigenfrequencies of the nanostripe are expressed in the terms of elastic moduli. Comparing
eigenfrequencies being found using discrete and continuum approaches the conditions for elastic
moduli determination are obtained. The obtained elastic characteristics for nanostripe are
compared with the same characteristics obtained from the static approach.
Obtained scale relations for the elastic properties of the nanostructures are used for the
further analysis of their strength and stability. Strength analysis requires usage of nonlinear
interatomic potentials for the nanostructures. From the static approach the ultimate stress for the
nanostructures is obtained in the terms of the interatomic bond strength. In the dynamic approach
we investigate wave propagation in the nanocrystal. The vibration spectrum and dispersion
relations are obtained. The speed of wave propagation decreases with the stress increase and
vanish in the fracture. This condition is used for determinations of the ultimate strength from the
dynamical point of view. Scale dependences of the mechanical characteristics obtained by
different approaches are compared.
 42. Scaling and Size Effects 1369

FRACTURE TOUGHNESS ASSESSMENT OF A C-MN STEEL USING

MINIATURE SPECIMENS

P. J. Apps1, W. Geary1, J. W. Hobbs1 and G. Wardle2

1Health and Safety Laboratory, 2Warhelle Consulting Ltd
1Health and Safety Laboratory, Harpur Hill, Buxton, Derbyshire, SK17 9JN, UK.
2Warhelle Consulting Ltd, Hawthorns, 79 Garton Drive, Lowton, Cheshire, WA3 2TR, UK.

pete.apps@hsl.gov.uk, william.geary@hsl.gov.uk, warhelle@blueyonder.co.uk

The fracture toughness of a material is of prime importance as input data to the assessment of
defects within a structure. However, in many instances there are insufficient fracture toughness
data available and little or no archive material available for testing. This has led to a drive towards
testing smaller specimens that do not meet the minimum size requirements as given in current
testing standards, or alternatively the development of re-constitution techniques. Furthermore,
‘boat’ or ‘scoop’ sampling techniques are available for removing small amounts of materials from
operating plant and this again focuses attention on the desirability for testing smaller test pieces.
Previous work by Wardle et al [1] has shown that the Master Curve approach (to ASTM E-
1921 [2]) may be utilised in the determination of fracture toughness data from small fracture
mechanics specimens, manufactured from a nuclear grade pressure vessel steel (A533B-1). The
Master Curve technique, first developed by Wallin [3], [4], essentially models the temperature and
size dependence of fracture toughness data obtained in the ductile-brittle transition regime.
The question arises as to whether or not this statistical approach, originally developed for
nuclear grade ferritic steels, could be used to assess the large number of C-Mn pressure vessels,
designed and built in the 1960s and 70s, which are still in operation in many industries throughout
the UK, but which are now reaching their original design lives. Operators are interested in trying
to extend the operating lives of these vessels based on engineering critical assessments (ECA)
using limited fracture toughness data.
This paper describes the experimental and analysis work undertaken on a BS 1501:213 C-Mn
steel, typical of a general engineering pressure vessel material, supplied by Advantica
Technologies. This material was obtained from a decommissioned high-pressure storage vessel
(HPSV) and contained both longitudinal and circumferential welds. Fracture toughness tests have
been carried out at a range of temperatures in the transition regime on 25mm x 25mm (1T) and
5mm x 5mm single-edge-notch bend specimens extracted from both the parent plate and weld
regions. In addition, 5mm x 5mm specimens were extracted from three different locations through
the weld thickness to assess the variability of fracture properties within the different weld regions.
The data obtained have been analysed according to ASTM E-1921 and the validity of these
data obtained from the 5mm x 5mm specimens have been compared with those obtained from the
25mm x 25mm specimens. The applicability of the Master Curve technique to general engineering
grade steels is discussed, along with practical and economic factors associated with the extraction
and testing of miniature fracture toughness specimens.
1370 P. J. Apps et al.

References
1. Wardle G, Birkett R P, Knowles G, Shuter D M, Dutton J T and Geary W. ‘Final report on
transition toughness testing of an A533B-1 steel using miniature specimens. Serco Assurance
technical report. SA/RJCB/01284001/R07. 2002.
2. ASTM E1921 – 97 ‘Standard test method for determination of reference temperature, T0 for
ferritic steels in the transition range’ American Society for the Testing of Materials. Annual
Book of Standards, Vol 03.01, 2000.
3. Wallin K ‘Validity of small specimen fracture toughness estimates neglecting constraint
corrections’ in Constraint effects in fracture theory and Applications, ASTM STP 1244,
American Society for Testing and Materials, p. 519. 1995.
4. Wallin K. ‘Master curve analysis of ductile to brittle transition region fracture toughness
round robin data. The ‘EURO’ fracture toughness curve.’ Technical Research Centre of
Finland, VVT Publications 367. Espoo 1998.
 42. Scaling and Size Effects 1371

SIZE EFFECT IN THE BONDING OF SMOOTH AND DEFORMED BARS: NSC

VERSUS HPC

Patrick Bamonte1, Dario Coronelli2 and Pietro G. Gambarova3

Milan University of Technology, Milan, Italy
pietro.gambarova@polimi.it

Size effect is unanimously recognized as a peculiar – and critical – aspect of the mechanical
behavior of many structures, provided that either the material or the interfaces among the various
parts have a quasi-brittle behavior, characterized by a softening answer under increasing strains or
displacements.
This is the case of bond in reinforced concrete structures, since the interface between a bar and
the surrounding concrete exhibits a typical softening behavior, both in smooth and in deformed
bars. However, in smooth bars - where bond is mostly ensured by the embedment length - size
effect is mainly related to bar debonding, and to the balance between the energy required by
debonding and that stored in the surrounding concrete. Since the embedment length plays a critical
role, size effect has been studied by testing long anchored bars (pull-out tests [1]).
On the contrary, in deformed bars – where bond is mostly ensured by the interlocking action of
bar ribs – size effect is mainly related to concrete behavior, and eventually to concrete crushing in
front of the ribs and to the microcracks radiating from the tip of the ribs. Since bond efficiency is
definitely greater than in smooth bars and bar yielding may precede bond failure, size effect has
been studied by testing short anchored bars (pull-out and push-off tests [2,3], often with well-
confined bars to avoid cover splitting).
In this paper, the results obtained during two recent experimental campaigns on long smooth
bars and on short deformed bars are recalled [4,5]. In both cases size effect is shown to be very
strong and in good agreement with the well-known size-effect power law [6].
Size effect is also modeled by numerically solving the 2nd-order bond equation in one
dimension. In the case of smooth bars, the simple but effective approach proposed by Shah et al.
[7] is adopted, by assuming an elastic-frictional behavior for the interface and two criteria
(strength-based and energy-based respectively) for the evolution of bar debonding.
In the case of deformed bars, the test results make it possible to formulate a size-dependent
local bond-stress/slip law, that fits fairly well also the scanty results obtained by other authors.
This law is probably the most useful finding of the whole research project, since its formulation is
an extension of the well-known power law introduced by EC-2, and is equally handy, to the
advantage of the designer, who may model any anchorage, should special circumstances require a
very refined analysis.
1 PhD Candidate, Dept. of Structural Engineering, Milan University of Technology, Milan, Italy
2 Assn. Professor, Dept. of Structural Engineering, Milan University of Technology, Milan, Italy
3 Professor, Dept. of Structural Engineering, Milan University of Technology, Milan, Italy
1372 P. Bamonte et al.

References
1. Bazant Z.P., Li Z. and Thoma,M. (1995). Identification of stress-slip law for bar or fiber
pullout by size effect tests. ASCE – J. of Engrg. Mech. 121 (5): 620-625.
2. Soroushian, P. and Choi, K.B. (1989). Local bond of deformed bars with different diameters
in confined concrete. ACI – Structural J. 86 (2): 217-222.
3. de Larrard. F., Schaller I. and Fuchs J. (1993). Effect of bar diameter on the bond strength of
passive reinforcement in HPC. ACI – Mat. J. 90 (4): 333-339.
4. Bamonte P., Coronelli D. And Gambarova P.G. (2003). Smooth Anchored Bars in NSC and
HPC: a Study on Size Effect. J. of Advanced Concrete Technology,V.1, No.1: 42-53.
5. Bamonte P., Coronelli D. And Gambarova P.G. (2004). Local Bond-Stress/Slip Law and Size
Effect in High-Bond Bars. Proc. 5th Int. Conf. on Fracture Mechanics of Concrete and
Concrete Structures-5th FraMCos, V.2, Vail (Co, USA): 869-876.
6. Bazant Z.P. and Chen E.P. (1997). Scaling of structural failure. Appl. Mech. Review 50 (10),
1997: 593-627.
7. Stang H., Li Z. and Shah,S.P. (1990). Pull - out problem : stress versus fracture mechanical
approach. ASCE – J. of Engrg. Mech. 116 (10): 2136-2150.
 42. Scaling and Size Effects 1373

SIZE EFFECT IN THE CRACKING OF DRYING SOIL

Pere C. Prat, Alberto Ledesma, and M. R. Lakshmikantha

School of Civil Engineering, Technical University of Catalonia
Jordi Girona 1-3, Building D2 — 08034 Barcelona, Spain
pere.prat@upc.edu

Cracking in soils due to water loss is a problem not much studied from a mechanical point of view,
despite its environmental implications. For instance, if a clayey soil is used as an impervious
barrier in open waste sites, an intense drought may origin cracks and therefore preferential flow
paths for polluted water. Cracks produced by environmental agents also reduce the bearing
capacity of the soil and increases its propensity to erosion. Previous works have studied the
problem either from a Fracture Mechanics perspective (Vallejo [1], Prat et al. [2], Ávila [3],
Harison et al. [4, 5], Hallet and Newson [6]), analysing the conditions for crack propagation, or
from a classical Soil Mechanics approach (Kodikara et al. [7], Abu-Hejleh and Znidarcic [8],
Konrad and Ayad [9], Morris et al. [10], Lloret et al. [11], using the effective stress principle. In
this case it has been observed that cracks initiate when soil is still close to saturation.
The purpose of the work presented in this paper is to investigate further key aspects of the
process of formation and propagation of desiccation cracks in soil. Several tests have been
conducted in an environment-controlled laboratory, starting with the work of Rodríguez et al. [11,
12] on circular samples. Further tests performed on rectangular samples of varying size, shape and
thickness suggest that there is a size-effect-related component in the process of formation and
propagation of the cracks, see Fig. 1. The conclusions from these preliminary tests are, however,
masked by the fact that the specimens were not of geometrically similar shapes, thus introducing
an additional factor in the analysis that makes difficult the interpretation of the size effect in the
problem under study.

FIGURE 1. Average crack width and average area of cracked cell as a function of specimen size.

Therefore, to investigate the size effect, a new series of tests has been designed in which the
samples are geometrically similar rectangles of sizes (in mm) A0 (841×1189), A1 (595×841), A2
(421×595), A3 (297×421) and A4 (210×297). Two series of tests are being performed, with
specimen thicknesses of 10 and 20 mm respectively. At the time of writing this Summary, tests
with the A0 (Fig. 2) and A1 specimens have been completed, and the remaining tests are still under
way. Fig. 2 shows a sequence of images showing the changing crack pattern on the A0 specimen
(20 mm thickness) as it dries, from the appearance of the first cracks until the end of the test. The
complete series of tests will be finished and analyzed by the time the full paper is to be submitted.
1374 P. C. Prat et al.

FIGURE 2. Evolution of cracking on a 20 mm thick A0 specimen at different stages.

References
1. Vallejo, L.E. In Fracture Mechanics Applied to Geotechnical Engineering, edited by ASCE,
1994, 1–20.
2. Prat, P.C., Ledesma, A. and Cabeza, L. In 8th Int. Conf. on Numerical Models in
Geomechanics, edited by G. Pande and S. Pietruszczak, Swets & Zielinger, 2002, 705–711
3. Avila, G. Estudio de la retraccion y el agrietamiento de arcillas. Aplicación a la arcilla de
Bogota, Ph.D. Thesis, Dept. of Geotech. Engng. and Geosciences, UPC, Barcelona, 2004.
4. Harison, J.A., Hardin, B. and Mahboub, K. J. of Geotech. Engng., vol. 120, 872891, 1994.
5. Harison, J.A. and Hardin, B. Int. J. Num. Anal. Meth. in Geomech., vol. 18, 467–484, 1994.
6. Hallett, P.D. and Newson, T. European Journal of Soil Science, vol. 56, 31–38, 2005.
7. Kodikara, J., Barbour, S.L. and Fredlund, D.G. Canadian Geotechnical Journal, vol. 35,
11121114, 1998.
8. Abu-Hejleh, A.N. and Znidarcic, D. J. of Geotech. Engng., vol. 121, 493502, 1995.
9. Konrad, J.-M. and Ayad, R. Canadian Geotechnical Journal, vol. 34, 477488, 1997.
10. Morris, P.H., Graham, J. and Williams, D.J. Canadian Geotechnical Journal, vol. 29, 263–
277, 1991.
11. Lloret, A., et al. In 2nd Int. Conf. on Unsaturated Soils. International Academic Publishers,
1998, 497–502.
12. Rodriguez, R.L., Estudio experimental de flujo y transporte de cromo, níquel y manganeso en
residuos de la zona minera de Moa (Cuba): influencia del comportamiento hidro-mecánico,
Ph.D. Thesis, Dept. of Geotech. Engng. and Geosciences, UPC, Barcelona, 2002.
 42. Scaling and Size Effects 1375

MODELLING OF THE VOLUME EFFECTS RELATED TO THE UNIXIAL

BEHAVIOUR OF CONCRETE. FROM A DISCONTINUOUS TO A
MACROSCOPIC APPROACH.

P. Rossi, J. L. Tailhan, J. Lombart and A. Deleurence

Laboratoire Central des Ponts et Chaussees (Paris, France)
58 Bvd Lefebvre, 75732 Paris cedex 15
Pierre.rossi@lcpc.fr, jean-louis.tailhan@lcpc.fr

The modelling of the cracking behaviour of cementitious composites can be viewed as one of the
actual major stakes in civil engineering. Although models are more and more sophisticated and
relatively appropriate to represent the macroscopic mechanical behaviour, they are not usually
fully pertinent for describing the complex localised or diffused multi-cracking processes arising in
such materials and eventually their possible couplings with other physical or chemo-physical
phenomena. Difficulties are essentially lying in combining the descriptions of microscopic
mechanisms at the level of the material and their impacts on the macroscopic behaviours at the
structural level, thus describing the whole range of scales (and one can note that complexities are
even larger if the heterogeneous character of the material and the size effects it induces [1] are
taken into account). The goal is then to develop models allowing obtaining the finest information
on the cracking processes but keeping a robust and reliable description of the macroscopic
mechanical behaviour.
This paper deals with that kind of goal through a modelling of the volume effects related to the
uniaxial tensile behaviour of concrete. Arising from results of numerical simulations representing
the – discrete and probabilistic – cracking behaviour of the concrete in tension, the mechanical
uniaxial pre-peak and post-peak responses as well as the opening of the major crack are analysed in
statistical terms. The bases of the implementation of such information in a macroscopic continuous
approach for modelling the material behaviour are proposed.
First, different series of numerical tests are performed in pure tension using the probabilistic
and discrete approach for representing the initiation and propagation of cracks in concrete,
developed by Rossi et al. [2]. This model was especially aimed to take into account volume effects
in the material. It is based on the hypotheses that the mechanical properties, especially the Young
modulus and the tensile strength of the material, are mainly related to the heterogeneity,
represented in the model by the ratio between the volume of the specimen and the volume of the
coarsest grain, and to the quality of the matrix directly related to the compressive strength of the
material. Empirical formulas are used to determine the mean values and the standard deviations of
these mechanical properties. The formulas were experimentally obtained [3] on different
geometries of cylindrical specimens, keeping a constant ratio height/surface equals to two, and
different formulations of concretes. The tested range of ratio ‘volume of the specimen/volume of
the coarsest grain’ extended from 10 to 7000 approximately and the tested range of compressive
strength extended from 30 MPa to 130 MPa.
In this study, numerical tests are replacing experiments. In order to test size effects, the
specimens are represented by regular meshes of different dimensions, but using a constant ratio
height (H) / surface (S): couples (H,S), (2H,2S), (H/2,S/2), (H/4,S/4) and (H/8,S/8) are used with H
equals to 276mm and S equals to 138mm x 138mm. The simulations are performed in plane
stresses because the model is still only developed in two dimensions. The uncracked concrete is
modelled using massive triangular elements while contact elements interfacing the massive
elements are representing cracks. Since the mechanical properties are considered as random
1376 P. Rossi et al.

variables, they are distributed on the meshes according to distributions characteristics following
the empirical formulas. Two sizes of coarsest grain (10mm and 25mm) and four compressive
strength are used (25 MPa, 40 MPa, 60 MPa and 120 MPa). This kind of model correctly
represents the initiation of a crack but does not take into account explicitly its propagation.
Nevertheless, using ‘sufficiently’ fine meshes, the propagation can be represented with a quite
good accuracy by the successive opening of the contact elements on the crack path. The sizes of
the (massive and contact) elements of the regular meshes are also chosen so that the duration of the
computation is kept reasonable. The ratio between the surface of the elements and the surface of
the plane meshes are maintained constant for all the meshes. For different meshes and the same
concrete, this leads to different ratios ‘volume of the specimen/volume of the coarsest grain’. And
considering the sizes of the finite elements, these ratios are leaving the initial range cited above.
The consistency of the model is then tested and criticized, and, using and inverse approach, a
correction of the empirical formulas is proposed for the small ratios ‘volume of the specimen/
volume of the coarsest grain’.
Then, an analyse of the mechanical pre- and post-peak behaviours of the material, as well as its
cracking behaviour, is proposed in terms of random processes, size effects, and influential
parameters. A parallel between a macroscopic description of the concrete’s cracking and the
microscopic statistical information obtained by the previous numerical probabilistic simulations is
drawn.

References
1. Bazant, Z. P., ‘Size effect’, International Journal of Solids and Structures, vol. 37, 69-80,
2000.
2. Rossi, P., Wu, X., ‘Probabilistic model for material behaviour analysis and appraisement of
concrete structures’, Materials and Structures, vol. 44, n°161, 271-280, 1992
3. Rossi, P., Wu, X., Le Maou, F. and Belloc, A., ‘Scale effect on concrete in tension’, Materials
and Stuctures, vol. 27, 437-444, 1994
 42. Scaling and Size Effects 1377

SIZE EFFECT AND R-CURVE IN QUASIBRITTLE FRACTURE

S. Morel1, E. Bouchaud2 and G. Valentin1

1LRBB, UMR 5103 (CNRS/INRA/Univ. Bx1), 69 route d'Arcachon, 33612 Cestas, France
2CEA-Saclay,DSM/DRECAM/SPCSI, 91130 Gif-Sur-Yvette Cedex, France
morel@lrbb.u-bordeaux.fr

In solid mechanics, an essential scaling problem is the effect of the structure size on its nominal
strength. This effect is particularly important in the case of quasibrittle materials which are
characterized by the existence of a large Fracture Process Zone (FPZ) where various toughening
mechanism take place such as microcracking, crack branching or crack bridging. Materials as
different as concretes, mortar and rocks, some composites and wood belong to this category. From
a classical equivalent linear elastic approach (where the increase of the specimen compliance due
to the FPZ development is attributed to the propagation of an elastically equivalent crack [1,2]) the
fracture behaviour of quasibrittle materials is characterised by a more or less pronounced rising
resistance curve, commonly called R-curve [3]. This R-curve behaviour emphasizes stress
redistributions and stored energy release which take place in such large FPZ.
Since 1984, Bažant and co-workers [4,5] have shown that in the case of quasibrittle materials,
contrary to what happens for Weibull's statistics [6], the size effect is linked to the very existence
of the R-curve behaviour. From an energy-based asymptotic analysis founded on a single R-curve,
Bažant [7] has shown that the nominal strength of structure should be transitional between two
asymptotic behaviours as shown in Fig. 1(a). In the case of small structures sizes, it is expected that
there is no size effect. For these small structure sizes, the fracture process zone is expected to
occupy the whole volume of the structure, inducing no stress concentration and, as a consequence,
the failure occurs with no crack propagation: this is the domain of the strength theory. For large
structures sizes, the size effect is expected to be the one of the linear elastic fracture mechanics
(LEFM): VN ~ D-1/2.where D is the characteristic size of the structure A possible justification is
that in large structures, the process zone is expected to lie within only an infinitesimal volume
fraction of the body and hence, the stress and displacements fields surrounding the FPZ are the
asymptotic elastic fields: this is the domain of LEFM. However, despite the success of the
Bažant’s Size Effect Law (SEL) to describe size effect of quasibrittle materials, the crossover
regime between both asymptotic behaviours (estimated from the intermediate asymptotic theory)
does not appear accurately defined. This point merits thinking about, especially since this is
usually the range of the experimental values from which the SEL is entirely defined.
In this study, within the framework of Bažant’s theory [7], we show that the size effect
obtained from a theoretical analytical expression of an R-curve (including size effect) appears
more complicated than the one proposed in the SEL especially in the crossover regime. In order to
estimate accurately the size effect, we propose to use an arbitrary defined analytical R-curve taking
into account (1) a size effect on the plateau value GRC of the R-curve but also (2) a size effect on
the crack length increment for which this plateau value occurs. The scaling of both characteristics
of the R-curve has been chosen to be in agreement with the scaling of the FPZ proposed in the
Bazant’s SEL [7].
1378 S. Morel et al.

FIGURE 1. Bazant’s SEL (a) and size effect estimated from a theoretical R-curve (b).

From the theoretical R-curve and as a function of a given structure geometry, the crack length
au corresponding to the peak load is estimated according to the recent proposal [2]. Thus, knowing
the crack length at peak load au and hence the corresponding resistance to crack growth GRu, the
size effect on the nominal strength is estimated following the framework of Bažant’s theory [7].
We show that, if the size effects obtained for small and large structures are in agreement with the
SEL proposals, the crossover regime between both asymptotic behaviours is more complicated
than a simple crossover and exhibits two new asymptotic regimes as shown in Fig. 1(b). The first
regime, which takes place just after the domain of strength theory, corresponds to the size effect
linked to constant relative crack length Du=au/D at peak load while the second regime is related to
a resistance at peak load which corresponds to the plateau value of the R-curve, GRu=GRC [8].
Finally the obtained size effect is compared to the Bažant’s SEL and discussed.

References
1. Bazant, Z.P. and Kazemi, M.T., J. Am. Ceram. Soc. vol. 73, 1841-1853, 1990.
2. Morel, S., Dourado, N., Valentin, G. and Morais, J., Int. J. Fract. vol. 131, 385-400, 2005.
3. Lawn, B.R., Fracture of Brittle Solids, (2nd ed.) Cambridge University Press, England, 1993.
4. Bazant, Z.P., J. Eng. Mech. Vol. 110, 518, 1984.
5. Bazant, Z.P., Appl. Mech. Rev. vol. 50 (10), 593, 1997(a) and references therein.
6. Weibull, W., Proc. Royal Swedish Inst. of Eng. Res. vol. 153, 1, 1939.
7. Bazant, Z.P., Int. J. Fract. vol. 83, 19-40, 1997(b).
8. Morel, S., Bouchaud, E. and Valentin, G., Phys. Rev. B. vol. 64, 104101, 2002.
 42. Scaling and Size Effects 1379

BIFURCATION AND SIZE EFFECT IN A VISCOELASTIC NON-LOCAL

DAMAGEABLE CONTINUUM

Th. Baxevanis, G. Pijaudier-Cabot and F. Dufour

R&DO, Institut de recherches en Genie Civil
Ecole Centrale de Nantes, 1 rue de la Noe
BP 92101, F-44321 Nantes cedex 3, FRANCE
theocharis@tem.uoc.gr

The growth of microcracks in progressively fracturing rate-independent materials is a fairly

distributed process which is associated with stable material response. However, for certain stress
trajectories such as tertiary creep, a different deformation mode may prevail, consisting of
formation of discrete failure planes (macrocracks). In the latter case, the mechanical response as
observed on the macroscale becomes, in general, unstable. The inception of such a localized mode
may be considered as a bifurcation problem related to the loss of positive definiteness of the
tangent material stiffness operator governing the homogeneous deformation (Rudnicki and Rice
[11]). In mathematical terms, the elliptic character of the set of governing differential equations, in
quasi-static problems, is lost (ill-posedness of the related boundary value problem). This result was
derived for the linearized rate equation problem, considering what is commonly denoted as a linear
comparison solid (Hill [5]).
In common practice, it is usually assumed that for low stress levels linear viscoelasticity takes
place and that the instantaneous mechanical behavior is elastic. Viscoelasticity of the material
causes strong time dependence of fracture growth. Bazant and Gettu [2] conducted size effect tests
on geometrically similar specimens loaded at different rates and observed that the slower the
loading rate, the more brittle the response. Pijaudier-Cabot et al [9] performed also a series of
three point bending beam tests (notched specimens) applying a sustained load for 3 months up to
85 percent of the maximum load prior to the fracture test. They arrived to a similar conclusion,
namely, that upon size effect tests, the specimens subjected to creep exhibited a more brittle
response compared to the ones that were not. This property yields a shift to the right of the data set
on the size effect plot.
For high stress levels the linearity hypothesis fails. Nevertheless, creep cannot be captured
merely by nonlinear generalization of viscoelastic stress-strain relation (Bazant [3]). The
experimental evidence associates creep strains at high stress levels to microcracking nucleation
and growth - cracks grow and interact with visco-elasticity - (Rüsch [12] was the first one to
conduct such experiments). Some experimental and analytical results concerning non-linear creep
can be found in the literature (see among others Bazant [1], Bazant and Gettu [2], Mazzotti and
Savoia [7]). The coupling of the mechanical behaviour and the visco-elastic behaviour of the
material assumed as a continuum, and its nonlinear instantaneous behaviour has been tackled by
many authors using a linear creep model (viscoelastic behaviour) coupled with a plasticity model
(De Borst et al [4]), or a smeared crack model (Rots et al [10]), or a damage model (Mazzotti and
Savoia [7]) (nonlinear instantaneous behaviour).
The aim of this paper is to investigate the localization properties of a rate-independent material
described by a non-local damage-based constitutive relation coupled with a creep relation of the
Kelvin type. It is proved that the coupled model preserves the properties of localization limiters as
they were pointed out in the paper of Pijaudier-Cabot and Benallal [8] and reproduces the
experimentally observed rate-effects of creep. Namely, the minimum wavelength of the
localization modes can not be zero in accordance with energy considerations and the material
1380 Th. Baxevanis et al.

response due to creep becomes more brittle, at least under a certain condition on the evolution of
the yield function of damage. Moreover, the normal n to the localization band depends only upon
the elastic state of strain at the onset of bifurcation and a shift on the size effect plot occurs towards
the strength criterion when the material parameter (0 1), which is the fraction of creep release
energy which contributes to damage, is decreasing. The motivation for introducing this parameter
into the constitutive law is that for low stress levels although creep strain can be large (even larger
than that corresponding to peak stress for short term loading) there is no significant variation of the
elastic modulus; that is no significant damage (Mazzotti and Savoia [6]). The proposed model fails
to predict creep rate-effects when damage is supposed, as it is common practice, to be driven solely
from the elastic release energy (=0).

References
1. Bazant Z.P., Mathematical modeling of creep and shrinkage of concrete, John Wiley & Sons
Ltd, 1988.
2. Bazant, Z.P. and Gettu, R., ACI Materials Journal, vol 89, 456-468, 1992.
3. Bazant, Z.P., In Proceedings of the 5th Int. RILEM Symp. on Creep and Shrinkage of
Concrete, edited by Z.P. Bazant and I. Carol, Barcelona, E&FN Spon, London, 291-307,
1993.
4. De Borst, R., Van Den Boogaard, A. H. And Sluys, L. J., In the Proccedings of the 5th Int.
RILEM Symp. on Creep and Shrinkage of Concrete, edited by Z.P. Bazant and I. Carol,
Barcelona, E&FN Spon, London, 309-325, 1993.
5. Hill, R., J. Mech. Phys. Solids, vol 7, 209-225, 1959.
6. Mazzotti, C. and Savoia, M., In Proceedings of Concreep 6, edited by Z.P. Bazant, F.-J. Ulm
and F.H. Whitmann, Cambridge, England, Elsevier, Amsterdam, 259-264, 2001.
7. Mazzotti, C. and Savoia, M., J. Engrg. Mech. ASCE, vol 129, 1065-1075, 2003.
8. Pijaudier-Cabot, G. and Benallal, A., Int. J. Solids Structures, vol 30, 1761-1775, 1993.
9. Pijaudier-Cabot, G., Omar, M., Loukili, A. and Le Pape, Y., In Proccedings of ICF 11,
Torino, March 2005.
10. Rots, J. G., Dissertation, Delft University of Technology, Delft, the Netherlands.
11. Rudnicki, J. and Rice, J. R., J. Mech. Phys. Solids, vol 23, 371-394, 1975.
12. Rusch, H., ACI j., vol 57 (1), 1-28, 1960.
 42. Scaling and Size Effects 1381

ULTISCALE NECESSARY AND SUFFICIENT STRENGTH CRITERIA

V. M. Kornev
Lavrentyev Institute of Hydrodynamics SB RAS, Novosibirsk, 630090, Russia
kornev@hydro.nsc.ru

A reasonably detailed description of multiscale fracture has reviewed by Cih [1]. Earlier necessary
multiscale strength criteria for brittle solids with hierarchy have been obtained by Kornev [2,3].
Later, Kornev [4,5] has considered some refinements of sufficient criteria at one of structural
levels. It should be noted that in [4,5], the Neuber-Novozhilov approach is used. This approach
allows one to implement a limiting passage from sufficient criteria to necessary ones when the
prefracture zone length tends to zero.
Multiscale discrete-integral strength criteria. Materials with hierarchy of regular structures are
under consideration. Introduce notation for numbers i 1, 2, ..., i 0 , the index i 1 corresponds to
a macrostructure. A set of normal rupture criteria agreed with each i structure is proposed for
normal rupture cracks
n i ri
1 (i)
³V ( x i , 0 ) d V m( i )
k i ri 0
y
, 0 d x i d n i ri , i 1, 2, ..., i 0 . (1)

2 v ( i ) ( x i , 0) d G m( i ) ,  ' i d x i d 0 . (2)

(i )
Here V y ( x i , 0) are normal stresses; O i x i y i are Cartesian coordinate systems oriented
relative the right crack parts (the coordinate system origin for each structure coincides with the tip
of a fictitious crack in the Leonov-Panasyuk-Dugdale model); 2 li 2 l 0  2 ' i is the length of

the model inner crack for the i structure; 2l0 is the length of a real inner crack; ' i is the
prefracture zone length for the i structure; ri is the specific linear size of the i structure of

initial material, ri ri 1 ; ni , ki are integers ( n i t k i ); ni ri is the averaging interval;

( n i  k i ) / n i is the damage coefficient of initial material within the averaging interval; V m(i ) - the

“theoretical” granular material strength; 2 v ( i ) 2 v ( i ) ( x i , 0) is the model crack opening

( i )
displacement; 2 v (  ' i , 0) G m( i ) is the critical model crack opening displacement.

As a rule, quasi-brittle fracture is considered for sufficient criteria (1), (2) when ' i l0 . For
'i o 0 ( i 1, 2, ..., i 0 ), we have the set of necessary strength criteria for brittle structured
solids [2, 3]. In the sufficient criterion (1), (2) the first relation for the i structure governs a model
crack tip advance, and the second relation describes a break of the bond nearest to the real crack
center. Using criteria (1) and (2) provides description of satellite crack initiation in the front of a
real crack.
Dual scale damage model. Consider stress corrosion cracking for steels. Dual scale damage
model is chosen for a cracking model when only two structural scales i 0 2 are considered. At
the macrolevel i 1 , relations from [4, 5] describing prefracture zones are used. At the microlevel
1382 V. M. Kornev

i 2 , the effect of surface-active substances on strength and deformation properties of solids is

accounted for (Rehbinder’s effect [6, 7]). Quantitative description of this effect is taken from [8,
9]. Atomic hydrogen is the most dangerous component of surfactants. Atomic hydrogen arises in
solutions after dissociation of molecules. At the microlevel, atomic hydrogen breaks overstrained
bonds at the crack tips [8, 9] producing either initiation of a crack or its advance. The rate of
microcrack tip advance is defined by the rate of atomic hydrogen delivery to the crack tip. At the
macrolevel, atomic hydrogen changes deformation properties of steel directly at the crack tip. In
this case, embrittlement of steel occurs due to its hydrogenation. Breaking of the overstrained bond
nearest to the center of a real crack occurs. The rate of the macrocrack tip advance is defined by
intensity of material hydrogenation.

References
1. Sih, G.C. Fracture mechanics in retrospect in contrast to multiscaling in prospect. In
Proceedings of the 17-th National Conference of Italian Group of Fracture, edited by
A.Finelli and L.Nobile, Bologna, June 16-18, 2004, 15-37.
2. Kornev, V.M. Hierarchy of strength criteria of structured brittle media. Satellite initiation of
microcracks, J. Appl. Mech. Techn. Phys., vol. 41, No. 2, 367-376, 2000.
3. Kornev, V.M. Multiscale criteria for the shear strength of block brittle media. Satellite
nucleation of micropores, J. Mining Sience, vol. 36, No. 5, 417-425, 2000.
4. Kornev, V.M. Generalized sufficient strength criteria. Description of pre-fracture zone, J.
Appl. Mech. Techn. Phys., vol. 43, No. 5, 763-769, 2002.
5. Kornev, V.M. Stress distribution and crack opening in the prefrature zone (the Neuber-
Novozhilov approach), Physical mesomechanics, vol. 7, No. 3, 53-62, 2004 (in Russian).
6. Rehbinder P. A. About influence of changing surface energy on cleavage, hardness, and other
crystal properties. VI s"ezd russkikh phisikov. State Publishing House, Moscow, p. 29, 1928
(in Russian).
7. Shchukin E. D. and Rebinder P. A. The formation of new surfaces under deformation and
destruction of solid in surfactants. Kolloidn. Zh. , 645-654, 1958 (in Russian).
8. Kornev, V.M. Quantitative description of the Rehbinder effect (brittle and qusi-brittle solids):
from fracture retardation to spontaneous dispertion, Physical mesomechanics, vol. 6, No. 3, 9-
18, 2003 (in Russian).
9. Kornev, V.M. Strength reduction of metals upon hydrogen chemisorption at the tip of a crack.
J. Appl. Mech. Techn. Phys., vol. 39, No. 3, 476-480, 1998.
 42. Scaling and Size Effects 1383

SIZE EFFECTS: MOVING FORWARDS

X. Hu and K. Duan
School of Mechanical Engineering
University of Western Australia
Perth, WA 6009, Australia
xhu@mech.uwa.edu.au

It is known that fracture behaviours of brittle thin films and brittle MEMS structures are controlled
by their surface conditions such as surface roughness and defects [1-4], which can vary from nano-
meters (nm) to microns (Pm). The surface roughness and surface defects are in the same length
order, which implies fracture initiated from a critical surface flaw is not only controlled by the
intrinsic fracture toughness, but also the interactions between the flaw and surface roughness. In
this case, the surface roughness is similar to a fracture process zone (FPZ), commonly found in
front of a crack tip in brittle heterogeneous materials such as concrete and polycrystalline
ceramics. Therefore, the effects of surface or boundary conditions have to be taken into account to
fully understand the size effects of brittle thin films and MEMS structures.
The surface finish or surface roughness is still important to brittle polycrystalline ceramics, but
is not so important to concrete. The material micro-structure, FPZ and their relative sizes to
surface roughness determine the quasi-brittle fracture or size effects. Such a comparison is even
more important when FPZ is close to the surface or specimen boundary. Therefore, FPZ close to
the surface or specimen boundary can have a huge influence on the size effect or quasi-brittle
fracture behavior of brittle heterogeneous materials including concrete and polycrystalline
ceramics. Emphasizing the interactions between FPZ and specimen boundaries, the concept of
“boundary effects” has been proposed recently to explain the size effects on the specific fracture
energy GF and the nominal fracture strength, studied extensively by various size effects models [5-
10]. It is therefore possible that size effects study on the macro- and micro- scales could be related
to size effects study under the nano-scale by emphasizing the common mechanism, i.e. the
boundary effects.
The purpose of the present paper is to show how the interaction of FPZ and specimen boundary
can influence the quasi-brittle fracture behavior of brittle heterogeneous materials such as
concrete, and to show that the common size effects can also be explained by the boundary effects
models. Similarity and difference between the boundary effects models and commonly-used size
effects models will also be studied using experimental results reported in the literature. It is hoped
that a common ground can be found among various size effects models by emphasizing the
fundamental size effects mechanisms.

References
1. Namazu, T., Isono, Y. and Tanaka, T., J. Microelectromechanical Systems, vol. 9, 450-459,
2000
2. Ding, J.N., Meng, Y.G. and Wen, S.Z., J. Mater. Res., vol. 16, 2223-2228, 2001
3. Ding, J.N., Yang, Y.C., Cai, L. and Wen, Inter. J. of Nonlinear Sci. and Numerical
Simulation, vol. 3, 499-502, 2002
4. Hu, X.Z. and Duan, K., J. Mater. Sci. and Tech., vol. 12, Suppl. 1, 2005 (in press)
5. Duan, K., Hu, X.Z. and Wittmann, F.H., Eng. Fract. Mech., vol. 70, 2257-2268, 2003
1384 X. Hu and K. Duan

6. Hu, X.Z. and Duan, K., Cement and Concrete Res., vol. 34, 1321-1330, 2004
7. Duan, K. and Hu, X.Z., Strength, Fracture and Complexity, vol. 2, 47-68, 2004
8. Duan, K., Hu, X.Z. and Wittmann, F.H., Mech. of Mater., 2005 (in press)
9. Duan, K. and Hu, X.Z., In Fracture Mechanics of Concrete Structures (Proceedings of
Framcos-5, April 12-16, 2004, Colorado, USA), ed: by V.C. Li et al., 2004, vol 1, 197-204.
10. Duan, K., Hu, X.Z. and Wittmann, F.H., In Fracture Mechanics of Concrete Structures
(Proceedings of Framcos-5, April 12-16, 2004, Colorado, USA), ed: by V.C. Li et al., 2004,
vol 1, 205-212.
 42. Scaling and Size Effects 1385

AN EXPERIMENTAL STUDY ON RAPID SETTING CONCRETE REPAIR

MATERIALS

J. P. Richards and Y. Xi
Department of Civil & Mechanical Engineering
United States Military Academy at West Point, U.S.A.
Department of Civil, Environmental and Architectural Engineering
University of Colorado at Boulder, U.S.A.
yunping.xi@colorado.edu

As the transportation infrastructure in the United States and around the world deteriorates, the
rapid and economical repair of rigid (concrete) pavement becomes increasingly important. There
are an ever growing number of repair methods and materials, but the method of selecting the
appropriate repair method and repair material is still very subjective, and very often it does not lead
to the most compatible repair.
This study reviews and summarizes the current state of research in the areas of concrete repair
methods and materials for highway pavement. Two commercially available rapid setting concrete
repair materials were selected and tested for their mechanical properties and fracture properties.
The fracture properties were determined based on the size effect method. Comparisons were then
made between the mechanical and fracture properties of the two repair materials, and those of high
strength and normal strength concretes.
The first observation is that quality control and workmanship are major issues for repairs and
repair material properties. Since the samples were made in numerous small batches rather than one
single large batch (the setting time is too fast to make and pour large batches in the laboratory
setting), the water content and other mixing and casting conditions can vary. With the small
amount of water that is needed for mixing (about 2.5-3.5 liters per 50 lb bag), even small variations
in water content may have a great influence on the material properties. This may explain why the
values for the mechanical properties in this study were often times lower than those published by
the material manufacturers. In a field environment, the variations could even be greater, leading to
larger differences between expected material performance and actual field performance.
Setting time also plays an important role in quality control. Since the materials set so rapidly,
careful attention must be paid to how long it takes to mix and place the repair material, or else it
will harden too fast and not bond appropriately.
The results of the mechanical properties tested under ASTM C 928 for the two repair materials
followed the trend noted in previous studies that the mechanical properties of repaired specimens
exceeded those of the original concrete. A more thorough review of these results makes it clear
that the mechanical properties alone are not enough to evaluate repair materials for their suitability
for specific repair situations. There is either not enough difference in the mechanical properties of
the materials to distinguish them, or there is not a strong enough correlation between the laboratory
results and the field performance of the materials.
For the non-standard mechanical properties tested, it was observed that temperature rise may
also play a large role in the material’s mechanical and fracture properties. Since the repair
materials have a higher and more rapid temperature profile than normal strength concrete, this may
lead to a lack of compatibility between old and new concretes, as shown by the early cracking ages
in the restrained shrinkage test.
1386 J. P. Richards and Y. Xi

The tests for the fracture properties of the repair materials may be good tests of material
compatibility, since fracture energy represents the material’s resistance to crack propagation. The
higher the fracture energy, the higher the resistance to crack propagation. The fracture energy for
the repair materials was similar to the fracture energy for normal strength concrete. HD 50 had a
slightly lower fracture energy and SikaQuick 2500 had a slightly higher fracture energy. This may
explain the problems that the Colorado Department of Transportation has had with repairs using
HD 50.
 44. Multiple cracking and delamination 1387

HIERARCHICAL FAILURE MODELING AND RELATED SCALE-INVARIANT

PROBABILITY DISTRIBUTIONS OF STRENGTH

D. A. Onishchenko
Institute for Problems in Mechanics of the Russian Academy of Sciences
Prospekt Vernadskogo, 101/1, Moscow 119526, Russia
onish_da@mail.ru, onish@ipmnet.ru

The paper deals with the strength problem for a discrete system with regular multilevel branching
structure. We will use the term 'hierarchical structure' (HS) for such system. In the variety of
probability strength models (Freudental [1]) the HS model can be attributed to an intermediate
position between the weakest link model and the Daniels stochastic bundle. Such models make it
possible to simulate salient features of propagating failure processes with varying scales
(hierarchical levels) of their actions. Some examples of their applications are: damage
accumulation and cracking in solids, the interaction of faults in earthquakes, breakage of fiber
structures, failure of complex systems, including progressive fire propagation (e.g., Newman and
A. M. Gabrielov [2], Onishchenko [3] and references inside). The characteristic feature of these
apparently different phenomena is that they involve some cascade processes which consecutively
cover one structural level of an underlying system after another and, eventually, envelop the entire
system.

FIGURE 1. Hierarchical structure model for the case k = 2.

Main model statements are as follows: (i) the HS of hierarchy depth n and with coordination
(m )
number k t 2 is a unity of stochastically identical blocks Bi related to different structural
levels, m = 0, …, n; i = 1, …, kn-m (see Fig. 1); (ii) zero-level blocks are the proper elements Ei;
their strength r is a random quantity with a probability distribution F0(x); (iii) the failure of
elements of any block B changes the load within this block alone until it becomes totally failed;
(iv) elements are treated as broken, if the failure criterion is satisfied: r < q, where q is the current
load applied to the given element; the blocks are failed when all their elements are broken. Under
imposed loading t, the progressive failure of the HS can occur with a certain probability that is
needed to be found.
The analysis of the HS model gives, in particular, the following results.
1 It can be shown that the constitutive equation of the HS model has the form
F m T { F m  1 } , where Fm(x) is the strength distribution for the level m block and T is a
corresponding functional operator. For k = 2 (Fig. 1) the governing equation is

Fm ( x ) 2 F m 1 ( x ) F m 1 ( g ( x ))  >F m 1 ( x ) @ 2 , m 1,  n
(1)
1388 D. A. Onishchenko

with a load redistribution function g. For instance, g ( x ) ax , a ! 1 corresponds to the so-

called model of fractal bundle of threads, where the tension load applied to any thread (block of
threads) passes in full to the neighbor one (the case a = 2 has been studied in part in [2]). The
relation g ( x ) x  a , a ! 0 can be employed in fire propagation models, when burning
elements exert additional influence a upon their neighbors. A number of other load redistribution
rules can be used in the problem of hierarchical failure [3].
2 Given the function g(x), there exists the distribution F(s)(x) such that the equality
F (s) T{ F ( s ) } holds, i.e., the function F(x) is a fixed point of transformation (1).
Distributions of such kind can be named scale-invariant (SI) distributions because they
remain the same for all structural levels. Here is some examples ( x  [ x min , x max ] ):

g ( x) x  a , a ! 0 : F ( s ) ( x ) 1  e  J ( x  xmin ) , xmin t 0, x max f, J ln 2 a ;

ax , a ! 1 : F ( s ) ( x ) 1  x min x  , x min ! 0, x max

J
g ( x) f, J ln 2 ln a
.
3 SI distributions are Lyapunov stable. Thus, when the distribution F0(x) belongs to the
attraction zone of the SI distribution, then under n o f the system "forgets" about
strength properties of its elements and generates a universal asymptotic strength
distribution governed only by the load redistribution law g. Such feature is similar in a
sense to the self-organization.
In spite of the fixed skeleton of the HS, such models are potentially of a great variety due to the
arbitrariness of properties of the function g(x). This provides the possibility of the use of SI
distributions for studying the hierarchical failure processes in different physical systems.
Moreover, various complications of HS models can be introduced, including the accounting for
factors diminishing the external action (Goldstein and Onishchenko [4]) and the availability of
several failure criteria (Onishchenko et al. [5]).

References
1. Freudental, A.M., In Fracture (ed. H. Liebowitz), vol. 2, Acad. Press, New York, 1968.
2. Newman, W.I. and Gabrielov, A.M., Intern. J. Fracture, vol. 50, No. 1, 1–14, 1991.
3. Onishchenko, D.A., Mechanics of Solids (Engl. transl.), vol. 34, No. 5, 21-38, 1999.
4. Goldstein, R.V. and Onishchenko, D.A., Phys.–Dokl. (Engl. transl.), vol. 41, no. 1, 37-39,
1996.
5. Onishchenko, D.A., Borodich, F.M., and Gomatam, J., Mathematical background for
modelling of the collapse of probabilistic hierarchical structures (fractal trees) with several
criteria of failure, Techn. Report TR/MAT/DAO-FMB-JG/97-94, Glasgow Caledonian
Univ., Glasgow, 1997.
 44. Multiple cracking and delamination 1389

INTERACTION OF TWO ADHESIVELY BONDED WEAK ZONES

I. V. Simonov and B. L. Karihaloo

Institute for Problems in Mechanics RAS
Vernadskogo av. 101-1, 119526 Moscow, Russia
simonov@ipmnet.ru
School of Engineering, Cardiff University, The Parade,
Cardiff CF24 3AA, UK
karihaloob@cardiff.ac.uk

This work focuses on the analytical-numerical analysis of the interaction of two crack-like defects
situated at the interface between dissimilar elastic materials and subjected to a remote tensile
stress field (the plane problem). These defects are modelled by weak zones (WZs), i.e. cuts
subjected to adhesive forces over their entire length. A small but strong microstructural feature of
the material , say, a small obstacle, separates the WZs. Contrary to a traction-free crack or a crack
with a small or a large process zone at its tips (the so-called cohesive crack), the WZ is normally
closed but opens gradually under sufficiently large external tensile stresses and becomes the
nucleus of a cohesive crack at some critical load. The adhesive forces can be of very different
physical origins – atomic, dislocational, localized porosity, etc. Healed cracks in glaciers and in the
earth’s crust can be regarded as WZs. The length of WZ can thus range from a few nm to
hundreds of km. Because of this, it is expedient to investigate the fundamental feature of the
interaction between WZs located close to each other. The key questions that will be addressed are:
(i) when does one of the WZs become the nucleus of a cohesive crack on its own without linking
with the other, and (ii) when do the WZs force the obstacle to rupture allowing the WZs to link
with each other before event (i) occurs.
Note that the known direct approaches using some adhesive force-opening relationship require
the solution of a cumbersome system of non-linear singular integral equations. The dislocation
model of a symmetric WZ, allowed us to obtain surprisingly simple analytical results [1]. As a
continuation, a generic approach to examining the behaviour of a WZ embedded in a very
asymmetric external tensile stress field was proposed in [2] by prescribing a priori the WZ
asymmetric opening by a two-parameter basic function which meets all physical constraints. This
was exemplified in [2] on the interaction between a short WZ and a semi-infinite collinear crack
(with a small process zone at the tip) subjected to wedge opening forces.
Here we apply this generic asymmetric WZ approach to obtain useful practical results for the
WZ interaction. The solution in the Cauchy integrals is computed for an extensive range of
parameters, in particular, tractions along the interface. The adhesive force distribution along the
WZs is calculated and it reaches finite peak values at all the edges of the WZs and minima inside
them. These minima decrease as the external load increases, and when one of them vanishes at a
critical load, the corresponding WZ becomes the nucleus of a cohesive crack. On the other hand,
the limit load is reached when the bridge between zones breaks down according to the Neuber-
Novozhilov criterion. Both critical and limit loads are completely determined in the framework of
this approach and can be compared in each concrete case. Overall, the asymmetric WZ model
shows considerable promise for solving analogous problems.
This work was partly supported by the Program N13 of the Russian Academy of Sciences
and the Russian Foundation of Basic Researches, N 05-01-00628.
1390 I. V. Simonov and B. L. Karihaloo

References
1. Simonov, I. V. and Karihaloo, B.L., Int. J. Solids Struct., vol. 37, 7055-7069, 2000.
2. Simonov, I. V. and Karihaloo, B.L., Philos. Magazine A, 2005 (in press).
 44. Multiple cracking and delamination 1391

MULTIPLE CRACKING IN SURFACE-HARDENED TENSILE SPECIMENS

AND THEIR FRACTURE MECHANISMS

L. S. Derevyagina, V. E. Panin, R. V. Goldstein1, N. A. Antipina and I. L. Strelkova

Institute of Strength Physics and Materials Science SB RAS
Tomsk, 634020, Russia
1Institute for Problems in Mechanics of the RAS

Moscow, 117526, Russia

strela@ispms.tsc.ru

In tension of metal materials with a surface-hardened layer can be clearly observed during the
entire development of localized deformation and neck formation. Quasiperiodic cracks formed on
the surface of such specimens are stress mesoconcentrators that generate high plasticity zones in
the material bulk. Sometimes their evolution passes through the stage of macroband formation and
ends in material fracture.
The kinetics of the process was studied on two sets of flat 65X13 grade steel specimens having
gage length 15 u 3 u 1.2 mm and a surface layer ~ 70 Pm thick hardened by ion-plasma nitriding.
The type-I specimens had surface-hardened layer only on the lateral surfaces and the type-II
specimens on the upper and bottom surfaces. The present study is devoted to the kinetics of local
stress-strain state in these specimens. Using television-optical complex TOMSC >1@, the plastic
strain fields were measured. On the base of these data we plotted the distribution patterns of the
main characteristics of deformation as a function of macro-deformation. The strain-induced relief
images were compared to the distribution of strain rate intensity given by:
2 2 2 2
Hi (H 1  H 2 )  (H 2  H 3 )  (H 3  H 1 )
3 .

FIGURE. 1. Interaction of high plasticity zones on the upper specimen surface in the region of two
neighboring cracks. The type-I specimen: the fields of longitudinal (a) and transverse (b)
displacement vector components; strain intensity distribution (c).

Fig. 1 (a–c) shows respectively the displacement vector component fields and the strain rate
intensity distribution at the stage of interaction between high plasticity zones. The inhomogeneous
behavior of the deformation is seen to develop near the nitrided surface layer. The plastic flow
exhibits a nonuniform behavior on a macro-scale level within a region constituting about one
eighth of the specimen width. The maximum strain intensity occurs at the crack tips as well as in
the region of overlapping of the high plasticity zones. The strain rate in the points of maximum
strain intensity (near the cracks) the is 8 times as large and in the points of zone interaction 3 times
as large as that in the middle part of the quasi-uniformly deforming specimen. The mechanical
characteristics of the type-I specimens and uncoated specimens are almost the same. Consequently,
the mechanical properties of the material are practically unaffected by the nonuniform plastic flow
1392 L. S. Derevyagina et al.

behavior near the hardened lateral surface layer. The tipe-1 specimen undergoes macrofracture in
the same way as the specimen having no nitrided layer.

FIGURE. 2. Interaction of high plasticity zones on the lateral specimen surface in the region of two
neighboring cracks. The type-II specimen: fields of longitudinal (a, d) and transverse (b, e)
displacement vector components; strain intensity distribution (c, f).

Fig. 2 (a-c) presents the fields of longitudinal and transverse displacement vector components
and the calculated strain rate intensity distribution for the type-II specimen at the prefracture stage
at H = 7.5–8.5 % (fracture takes place at H = 9.6 %). An analysis of the given patterns suggests that
high plasticity zone development is affected by crack propagation and extends over the entire
specimen thickness. The regions in which the isothetics of longitudinal and transverse components
are parallel to each other correspond to the maximum shear and the maximum strain rate intensity.
At the prefracture stage the isothetic configuration changes (Fig. 2, d-f): the maximum shear
region increases and extends over the whole specimen cross-section from one crack to another on
the opposite specimen surface. The region of maximum strain rate intensity coincides with the
maximum shear region. In the case of type-II specimen the trajectory of fracture is defined by the
maximum value of strain rate intensity. The specimen fracture occurs by shear.
Thus, a coated sample in with high plasticity zones associated with coating cracking extend
over the entire cross section would exhibit lower plasticity relative to the substrate material and
would undergo fracture by shear.

Reference
1. Panin S.V., Computer technologies, vol. 8, 10-25, 2003
 44. Multiple cracking and delamination 1393

FRACTURE CRITERION OF CRACKS INITIATION AND GROWTH

M. Perelmuter
Institute for Problems in Mechanics of RAS
Pr. Vernadskogo 101-1, 119526, Moscow, Russia
perelm@ipmnet.ru

The two-parametric fracture criterion of cracks initiation and growth is proposed. The application
of this criterion to the problems of cracks initiation and growth in homogeneous media or at the
interface of two different materials is considered. It is supposed that the mechanical bonds or fibers
constraining the crack opening exist between the crack surfaces. For polymers and nanocomposites
with polymers or ceramics matrix these mechanical bonds are chains of molecules, nanotubes and
nanoparticles. The size of the zone filled with the bonds (the bridged zone) can be comparable to
the size of the crack. The energetic characteristics of the crack such as the energy release rate
(ERR) associated with the crack tip stress intensity factors (SIF) and the rate of the energy
dissipation by the bonds (RED) associated with the energy necessary to create a unit of new
surface of the bridged zone are considered. In the model the ERR and the RED depend on the
bridged zone size and the bonds characteristics. The two-parametric fracture criterion consists of
the following conditions: 1) the necessary energetic condition of the crack tip limit equilibrium
(this condition is based on the equality of the ERR and the RED values); 2) the sufficient condition
of the bond limit stretching at the trailing edge of the bridge zone.
Based on these two fracture conditions the regimes of the bridged zone and the crack tip
equilibrium and growth are considered:

1 the crack tip propagates and the size of the bridged zone d increases without rupture of
the bonds if

­ G tip ( d , " ) t G b on d ( d , " ) ( a )

®
¯ u ( "  d )  G cr (b)
(1)

where G tip ( d , " ) is the energy release rate , G bond ( d , " ) is the rate of the energy dissipation by
the bonds, " is the half of the crack size, u ( "  d ) is the crack opening at the trailing edge of the
bridged zone, G cr is the bond limit stretching;
2 the size of the bridged zone decreases due to rupture of the bond without the crack tip
propagation if

­ G tip ( d , " )  G b o n d ( d , " ) ( a )

®
¯ u ( "  d ) t G cr (b ) (2)
3 the crack tip propagates with simultaneous bond rupture at the trailing edge of the bridged
zone if

­Gtip ( d , ") t Gbond ( d , ") ( a )

®
¯u( "  d ) t G cr ( b)
(3)
1394 M. Perelmuter

The last case corresponds to quasistatic crack growth because the both fracture conditions are
fulfilled. The equilibrium size of the bridged zone and the critical external stress are determined
from these two conditions. The value of the RED calculated in this case represents the energetic
characteristic of the adhesion fracture resistance that generally does not remain constant at the
crack length variation. Analytical application of the proposed criterion was performed for the
problem of the straight crack in homogeneous plane with the rectilinear law of bond stress.
Analytical expressions of the ERR, the RED and the dependencies of the critical external stress
and the bridged zone size versus the crack size at the limit equilibrium state are derived. The
application of the proposed model and the comparison of the proposed criterion and the classical
fracture criterions for different problems from macro to nanoscale of the interface cracks are
presented. Asymptotic expressions corresponding to the classical fracture models are found. The
problem of the bond stress determination is transformed to the system of singular integral-
differential equations in the case of the bond stresses dependent on the crack opening and all steps
of the fracture criterion application are performed numerically. Parametric analysis of proposed
model under the different properties of materials, bonds and bridged zone size is considered. The
distributions of the crack opening, the bond tractions, the SIF and the energetic characteristics are
presented. The estimations of the equilibrium size of the bridged zone, the adhesion fracture
energy and the external critical stress (the interface strength) depend on the crack size are found.
Some preliminary steps of the problem solution also are presented in the papers [1-4].

References
1. Goldstein R.V., Perelmuter M.N. Modeling of bonding at the interface crack, Int. J. of
Fracture, 1999, vol.99, N 1-2, p.53-79
2. Perelmuter M. N. An interface crack with bonds under tension and shear loading, 6th Int.
Conference on Biaxial / Multiaxial Fatigue and Fracture, Portugal, Barselona, June, 2001,
vol.2, p.1015-1022
3. Perelmuter M. Fracture criterions for bridged crack: from macro to nanoscale, In: The
abstracts book of 21st International Congress of Theoretical and Applied Mechanics, August
15-21, 2004, Warsaw, Poland
4. Perelmuter M. Growth of cracks bridged by nanofibers, In: The abstract book of 11th
International Conference on Fracture, March, 20-25, 2005, Turin, Italy

Work was partly supported by RFBR, grant number is 05-01-000191

 44. Multiple cracking and delamination 1395

INTERFACIAL CRACKS EMANATING FROM PARTIALLY DEBONDED

SUBSURFACE CIRCULAR ELASTIC INCLUSIONS

P. B. N. Prasad
Ground Floor, #2197, Sampada, 11th Main, 2nd Stage, Rajajinagar,
A-Block, Bangalore 560010, INDIA
bnprasad1@yahoo.com

It is well known that sub-surface defects like voids and inclusions are potential crack nucleation
sites in many applications. One typical example is sub-surface defects in railway wheels and rails.
It is well known that cracks emanate from these defects and propagate thereby removing chunks of
material from the wheel/rail that may lead to derailment and even loss of human life. It is therefore
important to understand the behaviour of cracks emanating from such defects and their interaction
with the free boundaries and neighbouring defects.
There is another source of stress singularity with sub-surface inclusion problems. The
interaction between the inclusion and the free surface; as the inclusion comes closer to the free
surface, the stresses in the thin ligament between the inclusion and the free boundary become
unbounded (Markenscoff and Dundurs, 1992; Wu and Markenscoff, 1996; Prasad and Simha,
2002; Prasad and Simha, 2004). The presence of cracks along the inclusion interface adds up to the
existing stresses and makes debonding even more easy (it depends on the orientation of the arc
crack and remote loads). These cracks propagate along the interface to a certain extent and then
deviate into the surrounding material (propagate parallel to the free surface before removing
cracked portion).
Another interesting crack interaction arises out of the fact that multiple debonds can occur
along the interface of the inclusion. In contrast to the single debond case, multiple debonds interact
with each other as well as with the free surface.
Unlike straight cracks, SIF for curvilinear cracks do not increase monotonically as the crack
propagates along the interface. Instead, SIF increases upto an arc angle beyond which it drops of as
it propagates. This suggests that under favourable conditions, interface cracks may kink into the
surrounding material.
In this paper, the problem of a circular arc crack at the interface of a sub-surface circular elastic
inclusion beneath the half-plane is examined (see Fig. 1) using complex variable approach
(Muskhelshvili, 1954). The variation of the SIF at the arc crack tips as the inclusion comes closer
to the half-plane is obtained from results for a partially debonded circular elastic inclusion in an
infinite plane by Prasad and Simha (2002, 2003). Solution for remote loads and point forces are
used to cancel desired tractions on the half-plane. The integral equation is then solved for varying
distance of the inclusion from the free boundary as well as varying elastic mismatch between the
half-plane and the elastic inclusion (Prasad et. al., 20051,2,3).
1396 P. B. N. Prasad

FIGURE 1. Half-plane with a sub-surface elastic inclusion with debond at the interface.

It has been shown by Ramakrishna (1994) analytically and experimentally that stress fields
become singular as the distance between the sub-surface void or circular elastic inclusion and the
half-plane decreases (half-plane subjected to remote stress). The current paper deals with a more
general problem of debonded circular elastic inclusion the limiting cases of which are hole and
rigid inclusion (by changing the elastic modulus of the inclusion and keeping the arc length
negligibly small). As the arc length becomes larger the competition between the crack tip
singularity and singularity due to the thin ligament stresses are examined in detail, numerically.

References
1. Markenscoff, X., Dundurs, J., Int. J. Solids. Structures., vol. 29, 1883-1888, 1992.
2. Muskhelishvili, N. I., Some Basic Problems of the Mathematical Theory of Elasticity,
Noordhof, Holland, 1954.
3. Prasad, P.B.N., Simha, K.R.Y., Int. J. Frac., vol. 117(1), 39-62, 2002.
4. Prasad, P.B.N., Simha, K.R.Y., Int. J. Frac., vol. 116, L3-L8, 2002.
5. Prasad, P.B.N., Simha, K.R.Y., Engng. Frac. Mech., vol. 70(2), 285-307, 2003.
6. Prasad, P.B.N., Simha, K.R.Y., Int. J. Frac., vol. 125(1), 51-71, 2004.
7. Prasad, P.B.N., Norio Hasebe, Wang, X.F., Shirai, Y., J. App. Mech., vol. 72, 389-393, 2005.
8. Prasad, P.B.N., Norio Hasebe, Wang, X.F., J. App. Mech., vol. 72, 393-399, 2005.
9. Prasad, P.B.N., Norio Hasebe, Wang, X.F., Shirai, Y., Int. J. Solids. Structures., vol. 42(5-6),
1513-1535, 2005.
10. Ramakrishna, D.S., Stress field distortion due to voids and inclusions in contact problems,
PhD thesis, IISc, 1994.
11. Wu, L., Markenscoff, X., J. Elasticity, vol. 44, 131-144, 1996.
 44. Multiple cracking and delamination 1397

MECHANICS OF BLOCK STRUCTURES AND ITS APPLICATIONS TO

GEODYNAMICS

P. V. Makarov
Institute of Strength Physics and Materials Science SB RAS
Pr. Academicheskii 2/1, Tomsk, 634021, Russia
pvm@ms.tsc.ru

A review of experimental and theoretical investigations of fracture of brittle and plastic solids and
media is presented. The investigations irrefutably testify the hierarchy and self-similarity of their
deformation and fracture. It is shown that the principle of infinite divisibility of geomaterials and
geomedia can be extended to meso and micro scales where self-similarity of their destruction,
determined for macroobjects, is preserved [1]. It is also shown that this principle can be applied to
plastic materials and media.
A media under loading is considered as a hierarchically organized system of blocks of various
sizes Li. Each subsequent block in the hierarchy is n times bigger than the previous (n 3). This
approach allows explaining of many effects of fracture and plasticity, for instance, “size effect”,
phenomenon of energy rearrangement (aftershocks following earthquakes). In plastic media the
hierarchy block formation is connected with generation of corresponding hierarchy of shear bands.
The consideration in the framework of general theory of systems is an alternative to statistical
methods of fracture and deformation analysis and allows to reveal the scales of fracture and its
location.
Nonlinear processes of localization of heat, deformation and damages accumulation in the
Earth’s crust form its image: hierarchies of blocks and cracks of different scales from microscopic
damages and geomaterial grains to tectonic faults and gigantic plates.
The multilevel approach of physical mesomechanics to study of deformation and fracture of
geological objects is a perspective line of investigation [2, 3]. The mechanical behavior of a
geological medium, strain localization and block-like structure formation on different scales were
studied in various loading conditions.
The realistic rheologic relaxation model of a geomaterial, which allows describing its
prolonged creep for geological times, was proposed. The model takes into account both brittle and
viscoplastic temperature-dependent properties of the material as well as damage accumulation.
Numerical modeling of behavior of quasi-brittle geomaterials and formation of block
structures in conditions of tension, compression and shear was carried out. For such conditions
when the geomaterials show plastic properties a numerical model allowing to take into account the
influence of pressure on the elasticity limit, volumetric changes, damaging and growth of cracks
has been developed.
A model of rift genesis was suggested. The model takes into consideration heat transfer and
mechanisms of active rifting as well as dynamics of plates and stress-strain state in Earth’s crust
elements. When developing the rheologic model of the litosphere, significant structure
heterogeneities are considered explicitly.
Model calculations of the formation of various graben types were carried out. It was shown
that no matter whether rifting is passive or active (structural and mechanical characteristics of the
geological medium varied in the model calculations) we can obtain all known types of grabens,
such as classical graben, semigraben, step-like graben, etc. The form and size of the formed
1398 P. V. Makarov

grabens turned to be sensitive to mechanical parameters of the model as well as to the geometry of
structural elements. Undoubtedly, this gives rise to certain difficulties in modeling real geological
models. The observed types of the geological objects and their motion can be simulated for various
and relatively close combinations of physical, mechanical and structural parameters of the
medium. Only complex problem solution (thermal history, structure examination, considering the
development of a fault system and block motion, etc.) makes it possible to determine the most
probable scenario of a particular rift development.
The analysis of influence of different angles of internal friction and dilatation, including ones
varying during deformation (or sliding on rough, serrated boundaries), on the development of
nonlinear process of the strain localization patterns is carried out.
Simulation of quasi-static processes is based on the variational-difference method of solving
the nonlinear connected tasks of the theory of plasticity and thermal condition. The model takes
into account accumulation of damages, effect of a dilatation (due to opening of internal
microcracks), dependence of material properties on temperature. Results of simulation of
localization of both deformation and accumulation of damages which lead to formation of
hierarchy of blocks, linear and ring structures of Earth during thermoforce loading of
heterogeneous geological medium are presented.
On the base of this approach the thermomechanical model of the Baikal rift zone was
developed.
This work was performed under financial support of Russian Academy of Sciences (Projects
Nos. 13.12 and 6.5.2) and Russian Foundation for Basic Research (Grants Nos. 02-05-65346 and
05-05-64659).
The author wishes to thank Dr. I.Yu. Smolin and Dr. Yu.P. Stefanov for calculations carried
out and useful discussions of the work.

References
1. Makarov, P.V., Phys. Mesomechanics, vol. 7, No. 4, 25-34, 2004
2. Makarov, P.V., Phys. Mesomechanics, vol. 1, No.1, 57-78, 1998
3. Makarov, P.V., Phys. Mesomechanics, vol. 6, No. 4, 99-111, 2003
 44. Multiple cracking and delamination 1399

STATIC AND DYNAMIC RESPONSE OF MULTIPLE DELAMINATIONS

M. G. Andrews and R. Massabo

1Department of Civil and Environmental Engineering,
Northwestern University, Evanston, IL, U.S.A.
2 Department of Structural and Geotechnical Engineering,

University of Genova, Genova, 16145, Italy

massabo@diseg.unige.it

Multiple delamination fracture of plane structures, Fig. 1, has been studied using a
particularization of beam theory applied to static and dynamic fracture problems. For static
problems, the model utilizes Timoshenko Beam theory applied to a set of sub-laminates defined by
the coordinates of the delamination tips. The model accounts for the possibility of relative rotations
between each of the two sub-laminates forming a delamination tip and the intact sub-laminate
ahead of the delamination tip, or relative root rotations, using localized rotational springs
(Andrews et al. [1]). Frictional contact is accounted for along each delamination surface.
Application of the model results in analytical and semi-analytical solutions. The model has been
shown to be qualitatively accurate as well as quantitatively accurate when the delaminations tips
have sufficient longitudinal separation.

FIGURE 1. Beam with multiple delaminations

For static loading, important interaction effects arise between the delaminations that strongly
depend on their through-thickness spacing. These effects include phenomena of shielding and
amplification of the energy release rates of the delaminations. ([1] and Andrews et al. [2]) In
addition, discontinuities in energy release rate are predicted when the delaminations tips reach the
same longitudinal coordinate, leading to local shielding and amplification. The discontinuities
agree well with the continuous but sharp transitions found in the 2D solution. These effects
strongly alter the macrostructural behavior of the system, leading to phenomena such as local snap-
back and snap-through instabilities, hyperstrength, crack pull-along and arrest.
For systems of equally spaced delaminations, the equality of length of the delaminations is
always stable (i.e. the system of delaminations will tend to grow self-similarly). For unequally
spaced delaminations the equality of length is often, but not always, unstable resulting in the
localized propagation of one or few delaminations. This leads to a more brittle response with
reduced energy absorption
For dynamic loading, a cohesive interface model has been formulated that represents a multiply
delaminated plate as a stack of sub-laminates connected by interfaces. Each sublaminate is studied
as a Timoshenko beam accounting for inertia. The interfaces are represented by linear springs in
the intact portions of the beam and follow a cohesive interface law. Fully formed delaminations
are represented by traction free conditions when opening between the sublaminates forming a
1400 M. G. Andrews and R. Massabo

delamination surface occurs and a contact law that resist interpenetration. A two level, implicit
finite difference solution scheme has been utilized.
A preliminary application of the model examines the problem of transient growth of a single
delamination in a double cantilever beam. A Dugdale type cohesive law, with an initial stiff branch
(stiffness k) representing the intact beam, shown in the inset of Fig.1, has been used for the
interfaces. The analysis proceeds in two steps. First, the free ends are statically displaced to an
initial displacement. Delamination growth is suppressed during this first step. Second, the
delamination growth constraint is released allowing the delamination to propagate dynamically.
Fig. 1 shows the delamination length as a function of time. The results of the proposed model are
compared in the figure with the 2D finite element results of Camacho and Ortiz [3]. The
comparison highlights how the simplified model is able to closely follow dynamic crack growth
both from qualitative (i.e., changes in speed and crack arrest) and quantitative point of views.
The problem of the dynamic behaviour of systems of multiple delaminations is currently being
studied by the authors and results will be presented at the meeting.

FIGURE 1: Time history of the crack tip position in the DCB specimen. Results of the proposed
model are compared with finite element results of [3]. Geometrical and material properties: h =
400 Pm, a = 400Pm, L = 6 mm to minimize wave reflection; material = alumina with E = 260 GPa,
Q = 0.21, U = 3,690 kg/m3 , Gc=34 J/m2, longitudinal wave speed cl = 8,394 m/s, Gc= 0.1674 Pm.
Loading: w0 = 4Pm.

References
1. Andrews, M. G., Massabò, R. and Cox B. N., In Preparation, (2005a)
2. Andrews, M. G., Massabò, R. and Cox B. N., Int. J. Solids Structures, In press, (2005b)
3. Camacho, G. T. and Ortiz, M., Int. J. Solids Structures, vol. 33, 2899-2938, 1996
 44. Multiple cracking and delamination 1401

MODELING CRACK GROWTH IN STRUCTURE- NONHOMOGENEOUS

MEDIUM UNDER COMPLEX STRESS STATE

Robert V. Goldstein, Yuri V. Zhitnikov and Nikolai M. Osipenko

Institute for Problems in Mechanics of the Russian Academy of Sciences
Prospect Vernadskogo, 101-1, 119526, Moscow, Russia
goldst@ipmnet.ru

The paper is devoted to modeling of an influence of cracking related to the material structure on
the macro fracture toughness. As examples of the structure we consider porous and piece – wise
homogeneous materials (media). It is assumed that the loading of the medium or structural
component provides multiaxial stress state in the region surrounding the macrocrack. The local
stress state is characterized by existence of zones of compression and tension.
The analysis is performed within the following assumptions:
- the size of a zone of nonlinear deformation and fracture preparing near the crack tips is
essentialy smaller as compared to the crack length (2D problem is considered). In that case the
conditions of crack growth are determined by local characteristics of the stress deformation state
near the crack tips;
- the size of a zone of nonlinear deformation and fracture preparing near the crack tips is
essentialy larger as compared to the size of a structural element of the material (medium). For
instance in case porous medium the last size equals the maximum of two values – characteristic
pore size and distance between ajacent pores. In that case the strength characteristics are
determined by average values with averaging scale of order the size of a zone of nonlinear
deformation and fracture preparing near the crack tips.
The crack growth criteria are formulated accounting for a possible influence of various
characteristic scales of changing the stress deformation state and parameters of the material
(medium) structure. The following main possibilities of crack growth in the materials (media)
under consideration are analyzed:
- rupture near the crack tip;
- rupture at some distance from the crack tip caused by a stress concentration at the structural
elements of the material (medium).
It was cases the conditions of local rupture are evaluated accounting for specific mechanisms
of cracking at the structural elements (pores, boundaries of the homogeneity regions).
It is shown that the stress state in the crack tip end region is characterized at certain loading
conditions by presence of two zones: rather small zone of tension adjoining the crack tip and
compression zone. The specific mechanisms of local rupture at the same structural elements (e.g.,
pores) depend on the tip of the stress state. Two types of plastic zones can be realized in zones of
tension and compression, respectively. The limit value of the effective stress intensity factor KIc is
essentially determined by the sizes of a plastic zone within the compression one.
Numerical modeling of microcracks formation in the region surrounding the main crack was
performed. The conditions of coalescence of the microcracks and the main crack were determined;
a possibility of the main crack deviation was analyzed.
For the porous media the mechanism of echelonlike structure of cracks formation in the region
surrounding the main crack was studied. It was shown that one can separate the stages of stable
1402 R. V. Goldstein et al.

increment of microcracks emanating on the pore contours and their unstable growth and
coalescence. In dependence on the stress state (combination of the main stresses in the region
surrounding the main crack) in a medium containing structural elements (pores, etc.) of the same
scale a region surrounding a mode I macrocrack tip has a complex structure which can be
represented schematically as follows.
Near the effective crack tip a fracture zone is located at the outer boundary of which the
conditions of unstable coalescence of microcracks joining adjacent pores are fulfilled. This zone is
surrounded by an annular zone in which joining of the adjacent structural elements by microcracks
is only possible in some directions.
As a result an echelonlike structure of microcracks can be formed within this annular zone. In
turn, the annular zone is surrounded by an external one where the stable growth of microcracks
emanating at the boundary of structural elements (e.g., pores) can occur. The outer boundary of
this zone is determined by the conditions of microcrack initiation.
The effective fracture toughness of the material (medium) KIc is determined by the features of
energy dissipation in all these three zones forming a region of pre – fracture.
Different scenarios of local rupture within these zones including the main crack deviation were
analyzed.
It is essential that a presence of active microstructures of the material (medium) leads to an
influence of a spherical part of the stress tensor caused by the external loads on the fracture
toughness value since this part changes the loading conditions of structural elements at the
microscale. The resistance of a porous body to the macrocrack growth increases under confining
compression. Note, that traditionally it is assumed that in conditions of a uniform loading the KIc
value only depends on a difference of pressure between the inner region of the macrocrack and
surrounding material.
Another essential factor is the pore pressure. Its action is opposite to an influence of confining
pressure. Thus, the effective macrofracture toughness (resistance to the macrocrack growth) of the
medium with typical local nonhomogeneities (e.g. pores) as structural elements of a certain
characteristic scale is influenced by additional factors which characterize the conditions of the
limit equilibrium of the local fracture processes at the scale of the structural element of the medium
(material). For instance, the spherical part of the tensor of the stresses caused by external loads and
pore pressure along with the geometric parameters of the medium belong to these factors for a
porous body. As a result the process (end) zone of a macrocrack has a complex structure. The
energy dissipation on new surfaces formation occurs in a volume embracing essential amount of
microstructural elements. The effects of the pore pressure on the effective fracture resistance in the
processes zone were evaluated.
 44. Multiple cracking and delamination 1403

NONIDEAL INTERFACE OF A BIMATERIAL WITH DEFECTS UNDER

THERMAL LOAD

V. E. Petrova and K. P. Herrmann

Voronezh State University of Architecture and Civil Engineering,
ul. 20-letija Oktjabrja 84, 394006 Voronezh, Russia
vera.petrova@math.vsu.ru
Paderborn University, LTM, Pohlweg 47-49, D-33098 Paderborn, Germany

Thermal cracking is often observed in bimaterial compounds as well as in composite materials due
to thermal mismatch of materials (Herrmann [1]). In addition, interfaces, defects and their
interactions play an important role in understanding the fracture behavior of multiphase solids, and
have received considerable attention (Theotokoglou and Tsamasphyros [2], Han et al. [3]).
Numerous investigations reveal that the interface strength influences especially also the strength of
a bimaterial as a whole.
In a number of previous authors’ works (see, for example, [4, 5]) the thermoelastic fracture
problems have been considered for a bimaterial with an interface crack and internal defects. The
bimaterial was subjected to thermal loading caused by a heat source or by a uniform heat flux,
applied at infinity and normal to the interface. These papers presented a method for the solution of
those problems. This method is based on the theory of complex potentials, the analytical
continuation theorem and the superposition technique and leads to singular integral equations for
the problems of interest. Due to the superposition principle the main temperature and thermoelastic
problems decompose into a series of simple subproblems each of them has simple geometry. The
subproblems are solved separately where the known complex potentials for the crack/singularity
problem can be used or new potentials can be constructed. The superposition scheme for such a
thermal crack problem is presented in Fig.1.

FIGURE 1. Superposition scheme

In previous studies a perfect interface (with the exception of an interface crack) was
considered. Now nonideal interfaces are considered. These nonideal interfaces can be modelled by
special interface boundary conditions for temperature functions and for stresses.
For example, the thermal complex potential for the problem (a) for a bimaterial with a virgin
interface and a crack in the material 2 (y<0) is given by

­° (1  K ) F 0 ( z ), z 1
F (z) ® 0
°̄ K F ( z )  F 0 ( z ), z  2
(1)

where F0(z) is the solution for the same crack in a homogeneous medium.
1404 V. E. Petrova and K. P. Herrmann

The complex potential F(z) is expressed by the solution F0(z) for a crack in an infinite
homogeneous plane by consideration of the different material properties of the given bimaterial.
Within certain limits the relation (1) is universal in the sense that it is independent of the physical
nature of the function F0(z). Therefore, F0(z) can be the complex potential for other singularities,
i.e. a dislocation or a point heat source/sink, in an infinite plane.
The constant K represents the mismatch of the conductive constants kj of the two constituent
materials and includes also the information about the interface. For a perfect interface with a fully
thermal conductive interface K ( k 2  k1 ) /( k 2  k1 ) is valid. If one of the both materials is
thermally insulated, for instance material 1 (y>0), then K=1 holds true. Moreover, the integral
equations for the problem of interaction of an interface crack with an internal crack also contain
this coefficient.
The previously established method has been applied to special cases. Thereby the effects of the
heat conductivity of a bimaterial interface on the stress-strain state along the interface and on the
stress intensity factors have been investigated in the presence of internal defects. The influence of
the thermoelastic parameters of the bimaterial as well as the location and orientation of different
systems of internal cracks on the stress intensity factors at an interface crack tip have also been
studied.
Acknowledgements
One of the authors (V.Petrova) acknowledges the Russian Fund for Basic Research for
financial support via Grant 05-01-00749.

References
1. Herrmann, K.P., ZAMM 77, 163-188, 1997
2. Theotokoglou, E.E., Tsamasphyros, G., Int. J. Fracture 55, 1-16, 1992
3. Han, X., Ellyin, F., Xia, Z, Int. J. Solids Structures 39, 1575-1590, 2002
4. Petrova, V., Herrmann, K., In Proceedings of Int. Conf New Challenges in Mesomechanics,
Aalborg University, 2, 591-597,2002
5. Petrova, V., Herrmann, K., Int. J. Fracture 128, 49–63, 2004
 44. Multiple cracking and delamination 1405

MULTIPLE CRACKING DEVELOPMENT AT THE PREFRUCTURE STAGE OF

ION CRYSTALS

Ye. Ye. Deryugin, V. E. Panin, V. Hadjicontis1, K. Mavromatou1

Institute of Strength Physics and Materials Science SB RAS
Tomsk, 634021, Russia
1Athenian university, Athens, 16121, Greece

dee@ispms.tsc.ru

By means of physical mesomechanics three scale levels of plastic flow localization in LiF single
crystals under compression were revealed: fine slip at micrometer range, kink bands at a millimeter
range and localization of macroshear-bands accompanied by extremely pronounced effects of
shear – bending – torsion. In a zone of macro shear-bands localization a crack of longitudinal
delaminating nucleates whose development is completed by specimen fracture. Scale levels of
shear stability loss of ionic LiF crystals under compression are qualitatively similar to those under
tension of metal materials. It testifies the universality of a principle of the physical mesomechanics
that the plastic deformation and fracture of solid is related to its shear stability loss at micro-,
meso- and macroscale levels.
Another principle of physical mesomechanics stating that a shear with a constrained material
rotation in a continuous medium generates on its path a zone of bending – torsion, which is the new
stress concentrator, has got the convincing confirmation. The scale levels of strain localization in a
deformed solid are related to generation of the appropriate scale hierarchy of stress concentrators.
The association of the generated stress concentrator with the sample bending governs the
dependence of material mechanical performances upon geometrical sizes of a sample. It requires
the criteria on allowance for scale factor during mechanical tests of materials to be designed on a
basis of mesomechanics.
Evolution of the growth of arrested longitudinal crack on the narrow side of the specimen is
shown on Fig. 1.
The relation between material failure under compression and its global shear stability loss was
shown for the first time. Earlier, the appropriate principle of physical mesomechanics was
formulated on the basis of tests by uniaxial tension scheme only. The given result has, in essence,
important value for understanding the nature of brittle and brittle-ductile fracture of materials, as
well as brittle-ductile transition at low temperatures and high loading rates etc.
Finally, within fields of bending-torsion stress gradients the intensive flows of surface defects
should occur. In ionic crystals it will stipulate the effect of electrical polarization as well as origin
of electrical charges on a surface of deformed crystal. Comparing electrical polarization effect
stages and plastic flow localization mechanisms up to fracture makes it possible to arrive to some
important conclusions:
1 On the stage of quasi-uniform deformation of LiF single crystals under compression the
currents of electric charges of different signs appear on the crystal sides {100}. Electrical
polarization current i=i(t) reaches its maximum in the centre of the compressed specimen
where fine sliding appears. At the opposite specimen ends the currents of electric charges
have opposite signs.
1406 Y. Y. Deryugin et al.

FIGURE 1. Evolution of the growth of arrested longitudinal crack on the narrow side of the
specimen (-c) and displacement vector field (d): Hp = 6.8 (, d), 7.4 (b), 8.4% (c).

2 The currents of electric charges have different signs on the opposite specimen faces.
3 Curves i=i(t) become serrated when coarse kink bands and cracks arise against the
background of the thin sliding.
4 Macrolocalization of deformation and multiple cracks appearing on the prefracture stage
qualitatively change the character of the curves i=i(t). Electrical polarization current
sharply going up, electric signals abnormal sign-alternating oscillations appear. Their peak
value may be several times higher as compared to the averaged curve i=i(t). Average
values of i=i(t) have the same sign in different zones of one side (100), thought they may
be different in quantity.
These regularities testify to fundamental assosiation between electric charges flows and
character of crystal shears and crack development. Fine sliding forms small steps on crystal
surface. That steps cause electrical polarization of the same sign. It defines average level of curve
i=i(t). The kink bands form more large steps on crystal surface with electrical polarization of the
opposite sign. It causes serrated character of averaged curve i=i(t).
Sharp increase of electrical polarization currents and abnormal signal oscillation on the
prefracture stage are of great practical interest. Detected result well correlates with development of
macrobands of localized deformation and appearance multiple cracks in specimen, causing
material fragmentation in zone of macrolocalization. This effect can serve as a precursor of
earthquake at electrical polarization measurements in the Earth crust.

References
1. Panin, V. E. (Ed.), Physical Mesomechanics of Heterogeneous Media and Computer-Aided
Design of Materials, Cambridge Interscience Publishing, Cambridge, 1998.
2. Panin, V. E., Physical Mesomechanics, vol. 3, No 6, 5-36, 2000.
 Author Index

A Angelova, D., 175 Bannister, M., 583

Abbadi, M., 289 Anifantis, N.K., 1313 Barbezat, M., 291
Abbasi, K.M., 349 , 399 Anthymidis, K.G., 219 Bardenheier, R., 507
Abdallah, A.A., 77 Antipina, N.A., 1391 Barlas, B., 837
Abdelaziz, M.N., 1271 Antolovich, B.F., 1245 Baron, A., 593
Adams, D.P., 17 Antolovich, S.D., 1245 Barre, T., 81
Adib, H., 1267 Antypas, C., 155 Barsoum, I., 1169
Adziev, G., 1043 Anvari, M., 971 Bartali, A.E., 843
Adziev, T., 1043 Aoi, S., 255 Bate, S.K., 1331
Agiasofitou, E.K., 529 Apps, P.J., 1369 Bathias, C., 761 , 771
Agioutantis, Z., 413 , 459 Araujo, J.A., 1099 Batra, R.C., 805
Agrianidis, P., 219 Aravas, N., 699 , 725 Baxevanis, Th., 1379
Ahmad, A., 893 Aria, F.R., 609 Bayraktar, E., 761
Ahmar, W. El, 1317 Arsi, M., 1057 Bazant, Z.P., 5
Ahmed-benyahia, A., 419 Artamonov, M.A., 1213 Beauchene, J., 1187
Ahrens, G., 579 Asferg, J.L., 397 Beaumont, P.W.R., 1279
Aiello, L., 323 Ashkenazi, R., 237 Bechet, E., 939
Aifantis, E.C., 3 , 37 , 345 Asiz, A., 1193 Becker, W., 277
Aifantis, K. E., 63 Astrup, T., 1209 Beke, D.L., 101
Ainsworth, R.A., 833 Atanasovska, I., 787 Bekku, N., 235
Ait Hocine, N., 767 , 769 Athavale, M., 69 Belayachi, N., 1271
Akiniwa, Y., 653 , 1131 , 1217 Athiniotis, N., 305 Belhouari, M., 405 , 541
Akisanya, A.R., 915 Atkins, T., 1211 Belingardi, G., 1257
Akita, M., 229 Atrash, F., 917 Belotteau, J., 827
Alamr, A., 673 Atrens, A., 979 Belouchrani, M.A., 131
Albaut, G., 435 , 451 Atzori, B., 201 Belytschko, T., 397
Aleksi, R., 1035 Aubin, V., 843 Ben Fredj, N., 1303
Aleksi, V., 1057 Audoly, B., 119 Benguediab, M., 407
Aleksic, B., 1047 Auer, T., 591 Bennani, A., 765
Alexandrov, S., 357 Auersperg, J., 733 Bennati, S., 1175
Alexopoulos, N.D., 509 Auriault, J.L., 529 Benseddiq, N., 769 , 1271
Alfaro, M.P.M., 547 Awaji, H., 91 Benzerga, A.A., 649
Ali, A., 163 Ayaso, F.J., 215 , 441 , 639 Berdin, B., 827
Aliabadi, M.H., 937 Ayatollahi, M.R., 1171 , 1173 Berdin, C., 817
Ali-Ahmad, M., 1361 Azari, Z., 289 Beretta, S., 263
Aliha, M.R.M., 1173 Azouaoui, K., 419 Berger, J., 681
Allix, O., 941 Bernabei, M., 323
Alsem, D.D., 671 Bertacchini, O. W., 1309
B Berthaud, Y., 537
Altenbach, H., 43
Altsadt, V., 93 Babout, L., 897 Berto, F., 1109
Aluko, O., 1097 Babski, K., 1335 Bertolino, G., 1223
Alush, H., 905 Bae, D.-H., 489 Besson, J., 109 , 815 , 831
Alvarez, J.A., 253 , 431 , 611 , Baek, D.-C., 721 Bessri, K., 761
615 Baek, S.-Y., 489 Betzwar-Kotas, A., 645
Amaro A.M. 525 Baer, W., 521 Bezensek, B., 369 , 593
Amaro, A.M., 525 Baets, P. De, 315 Bhat, S., 359
Amel-Farzad, H., 483 Bahr, D.F., 17 , 673 Biallas, G., 1241
Andersen, N.L., 1339 Bajric, A., 123 Bieroegel, C., 507
Andjelic, B., 1081 Baker, J., 613 Bilat, A.S., 109
Andjelkovic, M., 1041 Bakic, G., 1081 , 1083 Bilbie, G., 629
Ando, K., 1051 , 1065 , 1315 Balitskii, A., 981 Birkett, P., 261
Andreev, A.V., 381 , 449 Ballarini, R., 929 Bivin, Y.K., 103
Andrews, M.G., 1399 Balopoulos, V.D., 559 , 561 , Bjerken, C., 927
Andrews, R.M., 151 567 Blackman, B.R.K., 493
Andrianopoulos, N.P., 585 , Balueva, A.V., 485 Blai, I., 1079
587 Bamonte, P., 1371 Bland, D., 911
Anelkovi, Z., 1057 Banks-Sills, L., 923 Blawert, C., 979
1408
Blazquez, A., 1155
Blekhman, I.I., 785
Bocaneala, B., 1163
Bocciarelli, M., 575
Bochkaryova, L.V., 43
Boczkowska, A., 1335
Boellinghaus, Th., 985
Boer, H. de, 597
Boguszewski, T., 1335
Bolle, B., 1319
Bolzon, G., 575
Bombach, C., 731
Bompard, P., 817
Bontozoglou, V., 1293 , 1297
Borboudaki, C., 155
Bordet, S.R., 835
Borger, A., 589
Bosch, M.J. van der, 973
Bouchard, P.J., 1327
Bouchard, R., 901
Bouchaud, E., 1377
Bouchet, C., 815
Bouffioux, P., 1133
Bougault, A., 813
Bouiadjra, A.B., 541
Bouiadjra, B.B., 405 , 541
Boukharouba, T., 419
Boulahia, R., 419
Bourke, P., 331
Boursier, J.M., 1151
Boutabout, B., 405
Bouten, P.C.P., 77
Bouvard, J.L., 959
Bozhevolnaya, E., 557 , 565
Braham, C., 1303
Brajovi, L., 1035
Brameshube, W., 553
Bravo, R.E., 603
Breguet, J.M., 61
Brennan, F.P., 1017 , 1019
Britton, J.R., 1031
Brochet, S., 887
Brocks, W., 857 , 965
Brodin, H., 273
Brodling, N., 845
Broujerdian, V., 1363
Brown, M.W., 163
Bruening, J., 245
Brunet, M., 143
Brunner, A.J., 291
Bruschi, R., 987
Bryscejn, J., 571
Budden, P., 261
Bugat, S., 815 , 817 , 835
Bukhanko, A.A., 125
Bumbieler, F., 1147
Burzic, M., 1085
Burzic, Z., 1047 , 1075 , 1085 ,
1145
Busby, J.T., 677
Busso, E.P., 1161
Author Index
C Chentsov, A.V., 51
Cabrini, M., 1005
Cacchione, F., 647
Cadalen, E., 733
Cailletaud, G., 837
Cain, K.J., 319
Cajner, F., 347
Cako, J., 533
Calabri, L., 691 , 707
Camarao, A.F., 243
Cannillo, V., 379
Cannizzaro, D., 137
Caperaa, S., 943
Carboni, M., 231
Care, S., 537
Carloni, C., 353 , 1361
Carpinteri, Al., 9 , 239 , 601 ,
781 , 1105 , 1107 ,
1349
Carpinteri, An., 1091
Cartie, D.D.R., 493
Casati, M.J., 1359
Casavola, C., 429
Cassenti, B., 527
Castagnet, S., 337
Castillo, E., 183 , 377
Castro, J.T.P., 225 , 227
Castro-Prada, E.M., 439
Catherine, C.S., 153
Caucci, D., 323
Cavaliere, P., 307
Cavatorta, M.P., 1257
Cazacu, O., 549 , 789
Cendon, D.A., 967 , 1167 ,
1359
Cerny, M., 873
Ceyhan, U., 595
Chaboche, J.L., 959
Chai, G., 193 , 195
Chamat, A., 289
Chambon, L., 1291
Chambon, R., 629
Chamis, C.C., 543 , 635
Champaney, L., 941
Champion, Y., 903
Chang, J.H., 1157
Chang, S., 89 , 1137
Chaouadi, R., 149 , 515
Chaplain, M., 1195
Chapuliot, S., 621
Charkaluk, E., 181
Charles, R., 1331
Chatterjee, U.K., 1013
Chau, K.T., 747 , 825
Chau, T.T., 1333
Chazistergos, P., 417
Chen, C.R., 317 , 975
Chen, X., 691 , 707
Chen, Z., 727
Chen, Z.J., 69
Cheng, W., 423
Cherkaska, V., 867
Chernev, G., 1037
Chevrier, P., 81 , 1319
Chiang, C.-R., 371
Chiang, F.-P., 89, 1137
Chiarbonello, M., 433
Chieragatti, R., 609
Ching, H.K., 1139
Chiozzi, S., 265
Choi, D.J., 717
Choi, J.W., 875
Choi, S.-M., 91
Chou, T.-W., 95
Choukri, S., 329
Christ, H.J., 277 , 863 , 1247
Christensen, T.A., 1339
Chu, D.-Y., 469
Chue, C.-H., 363 , 365
Chung, S., 751
Chursin, A.S., 103
Chwalik, P., 511
Cicero, R., 253
Cicero, S., 611 , 615
Cincala, M., 877
Cipiere, M.F., 925
Cizek, J., 867
Clark, G., 305 , 1125
Clavel, M., 1133
Cleri, F., 41
Cocheteux, F., 289
Cole, D.M., 1357
Colombo, L., 41 , 1105
Combescure, A., 933 , 1321
Comi, C., 931
Constantinescu, D.M., 1163
Cooke, F.W., 1023
Copuroglu, O., 637
Cordill, M.J., 13 , 17
Coret, M., 1321
Corigliano, A., 647 , 1251
Cornetti, P., 345 , 1107
Cornuault, C., 941
Coronelli, D., 1371
Corre, V.Le, 621
Correia, J.R., 375
Costopoulos, Th., 1239
Coureau, J.L., 1201
Crocker, A., 821 , 823
Cross, C.E., 985
Cuamatzi, R., 121
Cui, H.Z., 309
Cvijovic, Z., 1069 , 1087
D
Dag, S., 1177
Daguerre, F., 603
Dahl, W., 123
Damkilde, L., 1209
Daniali, H.R.M., 333 , 335

Daniel, I.M., 11
Danzer, R., 317 , 589
Darwish, F.A., 243
Darwish, F.A.I., 547
Dascalu, C., 529 , 629
Dattoma, V., 265
David, K., 219
Davidkov, A., 175
Davids, W.G., 1191
Dear, J.P., 995 , 1263
Debruyne, G 961
Decoopman, X., 203
Degallaix, S., 621 , 843
DeGiorgi, V.G., 1345
Degrieck, J., 315
Delannay, F., 839
Deleurence, A., 1375
Dempsey, J.P., 63 , 1357
Deng, G., 437
Deng, M., 757
Derevyagina, L.S., 1391
Derimay, S., 277
Deryugin, Ye.Ye., 1405
Desjarlais, J., 1191
Deviatkin, E.A., 103
Dey, S., 145
Diaconu, C., 605
Diamantakos, I., 1295
Diamantopoulos, G-S.P., 393
Dietzel, W., 979 , 993
Dineva, P.S., 391
Ding, W., 691 , 707
Ding, Y., 89
Djambaski, P., 1037
Djukic, M., 1081 , 1083
Dlouhy, I., 865 , 1269
Dmytrakh, I.M., 991
Dogan, B., 595
Doglione, R., 137
Doi, S., 85 , 267
Doker, H., 851
Dong, H., 655
Dong, S., 771
Donnadieu, P., 79
Donoso, J.R., 139
Doquet, V., 171 , 1223
Doring, R., 1221
Dourado, N., 1199
Dover, W.D., 1015
Drobnjak, Dj., 1073
Du, J., 667
Duan, K., 1383
Duber, O., 863
Dudek, R., 731 , 733 , 737
Dufour, F., 1379
Dugger, M.T., 671
Duhamel, C., 903
Dulieu-Barton, J.M., 577
Dumont, D., 997
Dumouchel, P.E., 961
Dunger, S., 93
Author Index
Duranton, P., 1325
E Fissolo, A., 621
Ebara, R., 255
Eck, T., 553
Edwards, D.P., 709
Eftaxiopoulos, A., 155
Eichlseder, W., 205 , 241
Eifler, D., 191
Elata, D., 237
Elboujdaini, M., 1007
Elguedj, T., 933
Eliasi, R., 923
Eliezer, D., 985
Elizalde, M.R., 685
Elkadi, A.S., 1355
Elmegueni, M., 407
Emrich, A., 409
Endo, M., 1219
Engelberg, D., 897
Engelke, R., 579
Enzl, R., 889
Eric, O., 1145
Ernst, H.A., 603
Escobedo, C., 851
Evans, S.L., 271
Eve, S., 697
Everett, D., 1331
F Furukawa, T., 1341 , 1343
Fabregue, D., 133 , 839
Fajdiga, G., 197
Faleschini, M., 445
Faleskog, J., 105 , 1169
Fang, H.E., 727
Fang, J., 503
Fantilli, A.P., 1353
Farahmand, B., 383
Farhat, C., 1337
Fasce, L.A., 159
Fathy, A.M., 967
Faust, W., 731
Fedelich, B., 853
Fedina, M.E., 157
Fedorchenko, D.G., 401
Feichter, Ch., 777
Feng, X.-Q., 59
Ferentinos, G., 415 , 417
Fernandez-Canteli, A., 183 ,
377 , 1155
Fernandez-Zuniga, D., 377 ,
1155
Ferreira, J.L.A., 185
Ferreno, D., 599
Ferro, G., 601
Feyel, F., 959
Filipovic, N., 1071
Fillery, B.P., 385
Finnie, I., 423
1409
Firrao, D., 433
Fischer, F.D., 317 , 975
Fisher, G., 385
Flasker, J., 197 , 369 , 1233
Flewitt, P., 821 , 823
Floer, W., 863
Foct, F., 1151
Fok, A., 891
Follstaedt, D., 683
Forest, S., 827
Forth, S.C., 457
Fourman, V., 923
Fournier, B., 813
Fragomeni, J.M., 351
Frangi, A., 647
Freed, Y., 923
Freire, J.L.F., 227
Friedmann, V., 829
Fritzen, C.P., 863
Frontini, P.M., 159
Fujibayashi, S., 617
Fujimoto, H., 187
Fujimoto, T., 1181
Fukui, K., 187
Fukui, Y., 1141
Fukumori, M., 299
Fukumoto, M., 187
Fukuyama, S., 1003
Furlong, C., 65 , 71
Furnemont, Q., 819
Furuya, Y., 233
G
Gabetta, G., 987
Gable, B., 709
Gabor, P., 209
Gaenser, H.P., 205
Galiotis, C., 1277
Galkiewicz, J., 113
Gallo, C., 431
Galvez, J.C., 967 , 1167 , 1359
Gama, B.A., 1255
Gambarova, P.G., 1371
Gan, Y., 727
Ganguly, P., 969
Gao, H., 845
Garic, N., 1047
Gasparini, D.A., 343
Gasqueres, C., 997
Gay, R., 465
Gdoutos, E.E., 559 , 561 , 567
Geary, W., 1369
Geers, M.G.D., 387 , 411 , 973
Geniaut, S., 951
Gentilini, C., 353
George, S.M., 693
Gerberich, W.W., 13 , 17 , 693
Geric, K., 1053 , 1071
1410
Gerosa, R., 433
Getsov, L.B., 401
Ghassemieh, M., 581
Ghazavizadeh, A., 581
Gheshlaghi, R.M., 333 , 335
Ghidini, A., 433
Ghisi, A., 1251
Ghonem, H., 989
Gibala, R., 703
Gilles, P., 925 , 1325
Gillespie Jr., J.W., 1255
Gil-Sevillano, J., 685
Giovanola, J., 61 , 117 , 137
Glez, J.C., 887
Gliha, V., 1085
Glodez, S., 1233
Goedor, I., 205
Goettner, W., 1261
Goldstein, R.V., 51 , 1391 ,
1401
Goldthorpe, M.R., 1299
Gollhardt, A., 735
Gonzalez, B., 215
Gonzalez, D., 685
Gorbachev, Y.E., 1347
Gordon, A.P., 1243
Gorjo, K.L., 771
Goroshko, O.O., 803
Gorrochategui, I., 253 , 599
Goss, C., 173
Goto, M., 247
Gotsis, P.K., 543
Gourgues-Lorenzon,
A.F., 109
Graba, M., 113
Grabulov, V., 1079 , 1085
Graham, S.M., 461
Grandidier, J.-C., 337
Gravouil, A., 933
Gregoriou, V., 1285
Grellmann, W., 507 , 763
Groeger, V., 645
Gross, D., 53
Gruescu, I.-C., 789
Grutzner, G., 579
Guagliano, M., 1231
Gubeljak, N., 1049
Guerin, S., 903
Guerrero-Tovar, A., 1247
Guidault, P.A., 941
Guillen, T., 277
Guo, R., 655
Guo, Y.S.H., 825
Guo, Z., 661
Gutierrez-Solana, F., 431 ,
599 , 611 , 615
H Hoffmeyer, J., 1221
Hachemi, A., 131
Hachez, F., 965
Author Index
Hack, K., 531
Hackney, S.A., 63
Hadavinia, H., 913
Hadidi-Moud, S., 619
Hadjicontis, V., 1405
Haidemenopoulos, G.N., 725 ,
1293 , 1297
Hajjaj, M., 817
Hal, B.A.E., 387
Haller, P., 453
Hallman, D., 13
Ham, S.Il, 717
Hamaguchi, T., 285
Hamam, R., 1147
Hamdi, A., 767
Hamidi, B., 311
Han, J., 683
Han, S.-H., 169
Han, W., 73
Hancock, J.W., 369 , 593
Hansbo, P., 947
Hansson, P., 251
Harmuth, H., 297 , 591
Harrer, W., 589
Hartmaier, A., 845
Hasanifard, S., 269 , 481
Hasegawa, M., 783
Hashemi, S.H., 151 , 465
Hattar, K., 683
Hattori, T., 1115
Hausild, P., 107 , 495
Hawong, J.S., 711
Hayakawa, M., 233
He, J., 685
Headley, T., 695
Hearne, S.J., 683
Heaton-Adegbile, P., 1027
Heckel, T.K., 1247
Hedicke, K., 93
Hedrih, K.S., 793
Hefti, P., 67
Heinrich, G., 763 , 773
Heintz, P., 947
Herrmann, K.P., 1403
Hertel, O., 245 , 1221
Heuillet, P., 767
Heusden, J. van, 597
Hirakata, H., 743
Hirano, S., 1215
Hirotugu, I., 627
Hirth, J.P., 701
Hitzenberger, C.K., 579
Hoagland, R.G., 701
Hobbs, J.W., 1369
Hochstetter, G., 1143
Hodgkins, A., 891
Hoey, D., 1025
Hoffmann, M., 681 , 921
Hoffmeyer, P., 1209
Hohe, J., 277 , 829
Hojjati, M.H., 333 , 335
Holy, T., 573
Holzmann, M., 865 , 1269
Honchar, K., 327
Hora, P., 849
Hornikova, J., 869
Horsfall, I., 331
Horst, T., 773
Hortig, C., 389
Hoshide, T., 1183
Hosseini-Toudeshky, H., 311
Houdkova, S., 889
Hourlier, F., 771
Howard, I.C., 121 , 151 , 465
Hraam, H., 1073
Hsu, W.H., 363
Hu, J., 659 , 689
Hu, X., 385 , 1383
Huang, X., 49 , 321
Huber, N., 697
Huber, R., 189
Hughes, G., 821 , 823
Hutar, P., 147
Hytch, M.J., 903
I
Ienco, M.G., 433
Ikeda, T., 753 , 755
Iliopoulos, A.P., 587
Imad, A., 407
Imade, M., 1003
Inoue, H., 505
Ioannides, S., 211
Ionescu, I.K., 549
Ipperico, M., 601
Ippolito, M., 41 , 1105
Ipser, H., 645
Ishida, T., 653
Ishikawa, K., 477
Isogimi, K., 299 , 523
Ito, T., 1117
Itoga, H., 1121
Ivanova, E., 45
Ivanova, O., 451
Ivers, H., 775
J
Jacob, K.I., 655
Jacobsen, T.K., 957
Jacobsson, L., 223
Jacques, P.J., 165 , 819
Jakobsen, J., 565
Jakubek, J., 571 , 573
Jakubek, M., 573
Jang, S.-C., 489
Jelaska, D.T., 1227
Jesner, G., 199
Jha, M., 359
Jiang, J.C., 675

Jiao, Z., 677
Jivkov, A.P., 897
Joffe, R., 1281
Johannes, M., 565
Johansson, S., 193
Johnson, A., 1253
Johnston, W.M., 457
Joly, D., 79
Jones, R., 257 , 583 , 613 ,
1125
Jovanovic, M.T., 1145
Jullien, J.F., 1317
Jungk, J.M., 693
Jurasek, L., 1269
K Kleber, S., 861
Kaddouri, K., 405
Kadi, M., 419
Kalaitzidis, P.A., 115 , 393 ,
559 , 561 , 567
Kalthoff, J.F., 1155
Kamakura, M., 235
Kamoutsi, E., 1297
Kamoutsi, H., 1293
Kanamori, H., 25
Kandilioti, G., 1285
Kaneko, K., 555
Kanezaki, T., 907
Kang, B.-G., 553
Kao-Walter, S., 927
Karakostas, Th., 87
Karaman, I., 209
Kare, R.F., 1015 , 1021
Karihaloo, B.L., 935 , 1389
Karimi, A., 61
Karlik, M., 107 , 495
Karolczuk, A., 847
Kashchieva, E., 1037
Kashtanov, A., 99
Kassner, M.E., 55
Katerelos, D.G., 1277
Katsareas, D.E., 1311 , 1313
Katsikini, M., 87
Katz, Y., 905
Kavouras, P., 87
Kawada, H., 339 , 473
Kawagoishi, N., 247
Kawashita, L., 913
Kazemi, M.T., 1363
Kefalas, V., 155
Keller, J., 735 , 737
Keller, T., 375
Kelly, J., 695
Kennedy, M.S., 17
Keppas, L., 1313
Kermanidis, T., 1259
Keunecke, D., 1189
Kfouri, A.P., 1089
Kharin, V., 1001
Khatibi, G., 645
Author Index 1411
Khromov, A.I., 125 Kulkarni, A., 665
Kim, D.H., 713 , 715 Kumar, A., 633
Kim, I., 719 Kumar, B., 167 , 1023 , 1301
Kim, S.H., 875 Kumazawa, N., 1141
Kim, S.-Y., 721 Kun, F., 101 , 517
Kimura, H., 653 Kuna, M., 409 , 535
Kimura, S., 339 Kunkler, B., 863
Kinloch, A.J., 493 , 911 , 913 Kunz, L., 877
Kirchheim, R., 705 Kuo, C.C., 217
Kireitseu, M.V., 43 , 47 Kurai, J., 1047
Kiric, M., 1061 , 1063 Kurata, H., 555
Kishimoto, K., 83 , 505 , 627 Kuroda, M., 895
Kitagawa, M., 325 Kurzydlowski, K.J., 909,
Kitamura, T., 743 , 899 1335
Kiyak, Y., 853 Kushima, A., 899
Klapetek, P., 871 Kuzel, R., 867
Kwon, O.S., 711
Klempner, A.R., 67
Klepaczko, J.R., 511
Knabl, W., 445 L
Knight, S., 709 L’Hostis, V., 537
Ko, H.N., 1121 Labeas, G., 1259 , 1295 , 1297
Kobayashi, M., 141 Laborde, P., 953
Kobayashi, T., 141 Lacalle, R., 599
Kocanda, D., 177 Lacidogna, G., 781
Kocanda, S., 177 Lacroix, G., 819
Kocherov, E.P., 125 Lagoda, T., 279
Koguchi, H., 745 Lagoudas, D.C., 649 , 1309
Kohlhaas, K., 691 Lai, J., 211
Kohlhaas, K.M., 707 Laiarinandrasana, L., 765 ,
Kohut, M., 279 1143
Kojovi, A., 1035 Lakshmikantha, M.R., 1373
Kolednik, O., 161 , 317 , 885 , Lamazouade, M., 355
975 , 1049 Lamberti, L., 429
Koller, R., 183 Lambrou, P., 945
Komninou, Ph., 87 Landa, M., 849
Kompis, V., 43 , 47 Landes, J.D., 139
Kondo, D., 181 , 789 Landis, E.N., 1191
Kondo, Y., 1225 Lang, R.W., 777
Konsta, M., 561 Langlade, C., 1133
Konstantinidis, A., 345 Lashari, A.N., 179
Kontchakova, N., 357 Lassance, D., 839
Kontos, G., 459 Lassen, T., 281
Koponen, S., 1203 Lau, D.C.Y., 749
Kornev, V.M., 1381 Lauke, B., 293
Korobeynikov, S.N., 57 Laverne, J., 961
Korsunsky, A.M., 425 , 1351 Lavernia, E.J., 669
Kotoul, M., 879 Lazzarin, P., 1109
Kourkoulis, S.K., 413 , 415 , Lebaillif, D., 1159
417 , 459 , 467 Ledesma, A., 1373
Kovacevi, D., 779 Lee, G.-Y., 489
Koyanagi, J., 339 , 473 Lee, H., 1263
Kozodaev, D., 77 Lee, O.S., 713 , 715
Kraft, O., 697 Lee, S.-B., 719 , 721
Kramberger, J., 1237 Lee, S.W.R., 749
Krishnapillai, M., 583 Leendertz, J.S., 315
Krivtsov, A.M., 1367 Lefranc, P., 171
Krupp, U., 277 , 863 Legarth, B.N., 977
Krzhizhanovskaya, Leguillon, D., 919 , 1093
V.V., 1347 Leitner, H., 205
Kubota, M., 1225 Lempidaki, D.E., 1161
1412 Author Index
Lenain, A., 165 Machova, A., 849
Lenci, S., 811 Madia, M., 231 , 263
Lenkey, G. B., 101 Magnissalis, E.A., 415 , 417
Lennon, A.B., 1031 Mahapatra, D.R., 651
Leroy, R., 79 Maier, G., 575
Lespine, C., 1201 Maier, H.J., 209 , 1241
Lewandowska, M., 1335 Maiti, S.K., 1179
Li, J., 283 , 1119 Maiz, J., 685
Li, L.Y., 825 Major, Z., 777
Li, S.C., 825 Makarov, P.V., 1397
Li, X., 661 Maksimovic, K., 795 , 1075
Li, X.H., 273 Maksimovic, S., 795 , 1075 ,
Liang, C.-C., 519 1077
Liang, W., 663 Man, H-K., 963
Liao, M., 1289 Man, J., 871
Lichtendonk, W., 439 Manea, I., 605
Liechti, K.M., 687 Manfredini, T., 379
Lieske, D., 729 Manhart, C., 297
Lin, C.-K., 469 Manjgo, M., 1085
Lipinski, P., 81 , 545 Manolis, G.D., 391
Liu, C.C., 217 Marchi, C.S., 695, 669
Liu, C.T., 1135 , 1139 Margolin, B.Z., 401
Liu, R., 259 , 1249 Mariani, S., 931 , 1251
Liu, W.-J., 365 Mariano, P.M., 949
Liu, X.Y., 935 Marinakis, E.M., 393
Livieri, P., 1109 , 1111 , 1113 Marinelli, A., 467
Lo, T.Y., 309 Marinis, T.F., 67
Loboda, O.S., 1367 Marki, C., 1189
Locke, J.E., 167 , 1301 Markides, Ch., 413
Loibnegger, F., 189 Marrow, T.J., 891, 893 , 895 ,
Lombart, J., 1375 897
Lopes, L.H.M., 1099 Marsavina, L., 443 , 1163
Lopez-Aenlle, M., 183 Marsden, B.J., 891
López-Crespo, P., 427 Marshall, I.H., 583
Louter, C., 301 , 597 Marsoner, S., 199
Loutridis, S.J., 1235 Martha, L.F., 225
Love, A.J., 1017 Martin, E., 919 , 1093
Love, B.M., 805 Martinez-Esnaola, J.M., 685
Lovreni, M., 347 Martin-Meizoso, A., 111
Lu, C.-C., 519 Martsiniouk, D.Y., 449
Lube, T., 317 , 589 Martynyak, R., 327
Lucht, T., 421 Marumiya, T., 255
Ludwig, A., 513 Marunic, G., 1229
Lukas, P., 877 Maruszewska, W., 1263
Luke, M., 607 Marx, M., 881
Lundmark, P., 1277 Masaki, K., 285
Lupascu, D.C., 53 Masi, B. De, 647
Lusvarghi, L., 379 Mason, N.S., 995
Luu, T.T., 831 Massabo, R., 1399
Luyten, H., 439 Massin, P., 951
Lyckegaard, A., 557 Massinon, D., 837
Lynch, M., 261 Mastorakos, I.N., 679
Lynch, S.P., 641 , 709 Matczynski, M., 327
Matesova, D., 1365
Matj, Z., 867
M Matos, J.C., 215
Ma, S., 283 Matos, P.F.P., 249
Ma, Y.W., 875 Matsumoto, S., 743
Mabru, C., 609 Matsumura, T., 285
Macha, E., 847 , 855 Matsuoka, S., 233 , 907
Machmud, M.N., 505 Matsuyama, Y., 247
Matteis, P., 433
Mattoni, A., 41 , 1105
Mattsson, D., 1281
Matvienko, Y.G., 623 , 1273
Mavromatou, K., 1405
May, T., 853
Mayer, H., 1123
Mayer, T.M., 693
McCartney, L.N., 1149
McDanels, S., 15
McDowell, D.L., 657 , 1243
McEvily, A.J., 1219
McKown, S., 1265
Mecklenburg, T., 507
Mediavilla, J., 411
Medvedeva, N.Y., 449
Meggiolaro, M.A., 225 , 227
Mehiri, K., 545
Meletis, E.I., 675
Melin, S., 251
Melnik, R.V.N., 651
Meneghetti, G., 201
Menin, E.C.G., 185
Mesmacque, G., 203
Metaxa, Z.S., 559
Michel, B., 7 , 729 , 731 , 733 ,
735 , 737
Michelis, P., 463
Michler, J., 61 , 117
Michopoulos, J.G., 1337 ,
1341 , 1343
Mier, G.M. van, 963
Mijalkovic, M., 797
Mikula, P., 1329
Miladinovic, S., 1041
Millard, A., 537
Milovic, L., 1055
Mine, Y., 907
Miner, B., 685
Mines, R.A.W., 1265
Minichmayr, E., 241
Minnebo, H., 939
Minnetyan, L., 635
Miranda, A.C.O., 225 , 227
Mirzaee-Sisan, A., 1305
Misra, A., 701
Misumi, M., 1215
Mitasov, V., 451
Mithieux, J.-D., 887
Mitic, S., 809
Mitrakovi, D., 1035
Miyahara, M., 187
Miyamoto, N., 1131
Miyazaki, N., 753 , 755
Mliha-Touati, M., 329
Moes, N., 939 , 951
Mohammadi, B., 311
Moinereau, D., 835
Molina, J., 685
Monchiet, V., 181 , 789
Montorsi, M., 379

Moody, N.R., 17, 693
Mook, W.M., 13
Moore, D.R., 913
Morais, J., 1199
Morasch, K.R., 673
Morel, S., 1199 , 1201 , 1377
Morestin, F., 143
Moretti, M., 307
Morino, K., 247
Morozov, N., 45
Morozov, N.F., 1367
Mortarino, G., 433
Moskovic, R., 833
Mostafavi, M., 1171
Motta, S.H., 243
Mottot, M., 813
Moulin, D., 81
Moura, M.F.M.S. de, 525
Moura, M.F.S.F. de, 1199
Mourad, A.H.I., 1179
Moutsos, S., 641 , 709
Mouzakis, D.E., 1285
Muddle, B.C., 709
Mueller, R., 53
Mughrabi, H., 1127
Muhlich, U., 409
Muhlstein, C.L., 671
Mukhopadhyay, N., 1013
Mummery, P., 891
Murakami, Y., 907
Myoung, N.H., 713 , 715
N Ocana, I., 111 , 685
Nagai, M., 755
Nagaya, K., 791
Nagy, S., 517
Nait-Abdelaziz, M., 407 ,
767 , 769
Nakajima, M., 229 , 1009 ,
1121
Nakanishi, E., 299 , 523
Nakanishi, T., 437
Nakao, W., 1051 , 1065
Nakatani, M., 1065
Nakayama, E., 187
Nalla, R.K., 23
Nam, J.H., 711
Namdeo, S., 1179
Nara, Y., 555
Narasaiah, N., 221
Narazaki, C., 907
Nasu, K., 437
Naumenko, V.P., 361 , 367
Navi, P., 1197
Negishi, T., 1215
Negru, M., 605
Neimitz, A., 113
Neme, A., 329
Neov, D., 1307 , 1323
Neu, R.W., 1243
Author Index
Neviere, R., 1135
Newman, R.C., 897
Ngala, W.O., 1241
Ngiam, S.S., 1019
Nicoletto, G., 877
Nicouleau, E., 837
Nied, H.F., 395
Nielsen, L.O., 397
Niemz, P., 1189
Nikbin, K.M., 19
Nikiforovskij, V., 451
Nikitin, E., 705
Nikolic, R.R., 807
Nikolic-Stanojevic, V., 795
Nikoli-Stanojevi, V., 787
Nirgude, S.M., 359
Nishi, D., 325
Nishimura, N., 1115
Nishimura, T., 255
Nishioka, T., 1181
Nistor, I., 943
Nobile, L., 353
Nouet, L., 1325
Nowakowski, K.A., 73
Nowell, D., 249
Nykyforchyn, H.M., 1011
O Pardoen, T., 133 , 165 , 819 ,
O’Dowd, N.P., 1161
O’Gara, D., 1331
O’Reilly, P., 1025
Obrtlik, K., 871 , 883
Ochi, Y., 285
Ogawa, F., 473
Ogin, S.L., 1283
Ognjanovic, M., 1059
Oh, S.T., 1263
Ohgaki, T., 141
Ohguma, K., 1225
Ohkawa, I., 1215
Ohms, C., 1307 , 1311 , 1323
Ohrndorf, A., 277
Ojaghnezhad, F., 1165
Okada, H., 1141 , 1315
Okamura, K., 187
Olesen, J.F., 1209
Omiya, M., 83 , 505
Onishchenko, D., 1387
Ono, M., 1065
Osada, T., 1051
Osipenko, N.M., 1401
Ostby, E., 127
Oster, R., 579
Ouali, M.O., 403
Ouali, N., 419
Ovejero, E., 999
Ovesi, H.R., 311
1413
P
Paepegem, W. Van, 315
Paggi, M., 239
Pakiela, Z., 909
Paloura, E.C., 87
Panasyuk, V.V., 991
Pandermarakis, Z.G., 539
Pandey, R.K., 569
Panella, F.W., 265
Pang, S.-D., 5
Pangaev, V., 447
Panin, A.V., 35
Panin, S., 631
Panin, V.E., 35 , 1391 , 1405
Panoskaltsis, V.P., 343
Pantale, O., 943
Pantelakis, S.G., 1289, 1293 ,
1297
Pantuso, D., 685
Paolino, D.S., 1257
Papadopoulos, G.A., 303,
1239
Papadopoulos, P.G., 945
Papakaliatakis, G.E., 393
Papanastasiou, P., 551
Papanikos, P., 39
Papoulia, K.D., 969
Pappalettere, C., 429
839 , 857
Park, S., 751
Park, S.D., 717
Park, T.S., 719
Parodi, A., 433
Parrot, A., 827
Partheepan, G., 569
Pascual, J., 317
Pastore, T., 1005
Patoor, E., 1309
Patterson, E.A., 427 , 443
Pavisic, M., 1067
Pazis, D.N., 413 , 459
Peerlings, R.H.J., 387 , 411
Pehan, S., 1237
Peivandi, M.-T., 483
Pelikan, V., 849
Pell, R., 305
Peng, D.J., 1157
Pentecote, N., 1253
Pereira, M.V., 243
Perelmuter, M., 1393
Peres, F.M., 295
Perrin, G., 831
Persson, C., 223 , 251
Peshkov, I.A., 449
Petit, J., 997
Petraskovi, Z., 1041
Petrenec, M., 883
Petrova, V.E., 1403
Petrovski, B., 595
1414 Author Index

Petroyiannis, P.V., 1293 , Puzzi, S., 9 , 1349 Rozenburg, K., 681 , 921
1297 Pyttel, B., 607 Rozumek, D., 855
Pfennig, A., 853 Ruiz-Sabariego, J.A., 491
Pfuff, M., 993 Ruoff, R.S., 691, 707
Piazza, A., 1231 Q Rusinek, A., 511
Picard, D., 1093 Qian, L., 141 Rymkiewicz, S., 1153
Pignata, V., 601
Pijaudier-Cabot, G., 1379
Pikrakis, A., 585 R S
Pinasco, M.R., 433 Rabeau, V., 813 Saarela, O., 341
Pineau, A., 109 , 813 , 815 , Radi, E., 135 Sabate, N., 729
831 , 835 , 837 Radovi, A., 1079 Sabatier, L., 843
Pippan, R., 199 , 445 , 851 , Radovic, N., 1073 Saif, T., 683
859 , 861 Radu, C., 605 Sakai, T., 1129
Piques, R., 765 Rajic, N., 583 Salapete, R., 837
Pircher, M., 579 Rajicic, B., 1081 , 1083 Salaun, M., 953
Pirondi, A., 955 Rajnovic, D., 1145 Samuneva, B., 1037
Pitt, S., 257 Rakin, M., 1069 , 1087 Samyn, P., 315
Planas, J., 967 , 1167 , 1359 Ramos, A., 183 Sanchez, J.M., 685
Plasatis, D., 945 Ranganathan, N., 79 Sancho, J.M., 967 , 1167 ,
Plumtree, A., 319 Rangelov, T.V., 391 1359
Pluvinage, G., 1267 Ravi-Chandar, K., 497 Sandera, P., 869
Podrug, S., 1227 Ravichandran, G., 1135 Sandler, J.K.W., 93
Poitou, B., 919 Ray, K.K., 221 Sandvik, A., 127
Pokluda, J., 869 , 873 Recho, N., 281 , 283 , 1159 Sante, R. Di, 307
Polak, J., 871 , 883 Redda, T.D., 437 Sanz, L., 1025
Polanco, J.A., 431 Rega, G., 811 Saribay, M., 395
Polatoglou, H.M., 87 Reifsnider, K.L., 49, 321, Sarikaya, D., 1177
Polyzois, D., 179 1287 Sarrazin-Baudoux, C., 171,
Pommier, J., 953 Reimanis, I., 681 , 921 997
Pommier, S., 491 , 1147 Reimerdes, H.G., 1261 Sarris, E., 413
Poshtovaya, O.A., 449 Reincke, K., 763 Sathon, N., 577
Potapenko, A.Y., 841 Reis, P.M., 119 Sato, H., 471
Potapenko, Y.A., 1213 Reis, P.N.B., 525 Sato, S., 1051
Poulon-Quintin, A., 887 Renard, Y., 953 Satraki, M., 415
Poulsen, P.N., 397 Renault, L., 609 Sauzay, M., 147 , 813
Poulton, M., 995 Reusch, F., 389 Savaidis, G., 245
Poussard, C., 153 Reyes, E., 1167 , 1359 Sawaki, Y., 299 , 523
Pouvreau, C., 61 , 117 Richards, J.P., 1385 Scannell, P.T., 1031
Powers, L.M., 343 Riedler, M., 241 Scavino, G., 433
Prasad, P.B.N., 1395 Rishmany, J., 609 Schaf, W., 881
Prat, P.C., 1373 Ritchie, R.O., 23, 671 Scheider, I., 965 , 993
Predan, J., 1049 Rivolta, B., 433 Schemmel, I., 199
Prendergast, P.J., 1031 Roberts, S.G., 633 Schepdael, L. Van, 315
Prioul, C., 107 , 827 Robertson, I.M., 683 Scherban, T., 685
Priymak, O.A., 623 Robin, C., 203 Schick, A., 863
Prochazka, I., 867 Robinson, S.L., 669 Schlangen, E., 637
Profant, T., 879 Rodopoulos, C.A., 163 , 169 , Schodel, M., 263
Pryputniewicz, E.J., 65 1289 Schon, C.G., 295
Pryputniewicz, R.J., 21 , 65 , Rodriguez-Martin, R., 111 Schonwalder, J., 475
67 , 69 , 71 , 73 , 75 Rodriquez-Sanchez, F.S., 493 Schossig, M., 507
Przekwas, A.J., 69 Rokach, I.V., 499 Schrade, D., 53
Puchko, N.P., 801 Roman, B., 119 Schreurs, P.J.G., 973
Pugno, N., 345 , 691 , 1105 , Romein, T., 301 Schuller, R., 1127
1107 Ronde-Oustau, F., 1291 Schutte, H., 349 , 399
Puiggali, M., 355 Rosakis, A.J., 25 Schwob, C., 1291
Pundt, A., 705 Rossi, G.L., 307 Sciammarella, C.A., 429
Pusch, G., 513 Rossi, P., 1375 Scibetta, M., 599
Pustai, D., 347 Rossmanith, H.P., 27 , 189 Sedighi-Gilani, M., 1197
Putot, C., 1093 Rossoll, A., 137 Sedmak, A., 1043 , 1061 ,
Putzger, R., 453 Rots, J.G., 475 1069
 Author Index 1415

Sedmak, S., 1053 , 1055 , Spitas, C., 1239 Tanguy, B., 815 , 831 , 835
1069 , 1079 Spitas, V., 463 , 1239 Tarafder, M., 145
Seeger, T., 1221 Springmann, M., 535 Tarafder, S., 145 , 221
Sehgal, D.K., 569 Sraml, M., 197 Tarfaoui, M., 329
Seiler, B., 733 Srinivasan, K., 161 Tay, T.E., 757
Semenov, B., 45 Stach, E.A., 671 Taylor, D., 1025 , 1095 , 1099 ,
Serier, B., 541 Stagno, E., 433 1107
Seshadri, B.R., 457 Stahle, P., 927 Tche, K., 711
Sevecek, O., 879 Stanzl-Tschegg, S.E., 1205, Teng, T.-L., 519
Shanyavskiy, A.A., 841 , 1123, 1127 Teo, W.S., 493
1213 Staroselsky, A., 527 Tesch, A., 851
Sharp, P.K., 1125 Statnikov, E., 1289 Thakre, P.R., 649
Shen, G., 901 Stazi, F.L., 949 Thanh, N., 537
Shen, L., 727 Stec, M., 105 Thaulow, C., 127 , 971
Shenoy, M.M., 1243 Steglich, D., 857 Theotokoglou, E.E., 563
Sherman, D., 917 Steinberger, R., 777 Thibaut, B., 1187
Sherry, A.H., 895, 1299 Steinmann, P., 37 Thomsen, O.T., 557 , 565
Shi, Z., 1185 Stekovic, S., 275 Thostenson, E.T., 95
Shibata, H., 235 , 287 Stepanov, S.L., 125 Tidu, A., 545 , 1319
Shimizu, T., 1009 Stepanova, L.V., 157 Tijeras, V., 545
Shinozaki, A., 627 Stevens, N., 897 Tilbrook, M., 681 , 921
Shodja, H.M., 313 , 1165 Stifter, D., 579 Timkin, S.V., 449
Shterenlikht, A., 427 Stiles, D.J., 461 Timmerman, E., 439
Shukla, A., 29 Stock, R., 859 Timmerman, R., 671
Shur, V.Ya, 53 Stone, N., 1015 Timofeev, B.T., 487
Siad, L., 403 Strelkova, I.L., 1391 Toda, H., 141
Sideridis, E., 303 Strnadel, B., 495 Tokaji, K., 229 , 235 , 287 ,
Sidhom, H., 1303 Student, O.Z., 1011 1009 , 1121
Sidjanin, L., 1145 Stulikova, I., 867 Tokunaga, Y., 255
Siegele, D., 829 Stumpf, H., 531 Tomlinson, G.R., 43 , 47
Siegmund, T., 161 Subramaniam, K.V., 1361 Tomlinson, N., 1097
Sih, G.C., 31 , 723 Suister, E., 315 Tomlinson, R.A., 443
Sikorski, Z., 69 Sumarac, D., 797 , 1041 Tong, J., 1027
Siligardi, C., 379 Sun, B., 685 Tonkovi, Z., 625
Silva, G., 433 Sun, C.T., 741 Toonder, den J.M.J., 77
Simonov, I.V., 103 , 1389 Sun, Y.Z., 747 Toribio, J., 215 , 441 , 639 ,
Siores, E., 257 Supancic, P., 589 999 , 1001
Sivaprasad, S., 145 Suryavanshi, A., 659 , 689 Torzewski, J., 177
Sjoestroem, S., 273 Susmel, L., 1099 Touzet, M., 355
Skozrit, I., 625 Svendsen, B., 389 Tovo, R., 1111
Skrypnyk, Y.D., 367 Swieszkowsk, W., 1335 Tozaki, Y., 287
Skvortsov, V., 565 Trajkovi, M., 1041
Skytta, V., 341 Trajkovic, M., 797
Sloot, P.M.A., 1347 T Trappe, V., 775
Smith, D.J., 619 , 1305 Tabanyukhova, M., 447 , 451 Trattnig, G., 861
Smith, G., 821 , 823 Tabatabaei, S.M., 313 Trautmann, K.H., 851
Smith, I., 1193 Tai, Y.H., 465 Trisovic, N., 1041
Smith, P.A., 1283 Tailhan, J.L., 1375 Trubitz, P., 513
Smola, B., 867 Takacs, N., 101 Truman, C.E., 619 , 1305
Sniezek, L., 173 Takahashi, K., 1051 , 1065 Tschegg, E.K., 1205
Snow, M., 1193 Takahashi, Y., 743 Tschegg, S., 1207
Sofronis, P., 699 Tal-Gutelmacher, E., 985 Tserpes, K.I., 39
Sola, A., 379 Talia, J.E., 703 Tsipas, D.N., 219
Soltani, N., 581 Tamuzs, V., 1287 Tsirigas, P., 459
Song, G., 979 Tan, K.T., 911 Tsuru, H., 1131
Sorbello, F., 821 Tan, V.B.C., 757 Tukiainen, P., 1203
Sorensen, B.F., 957 Tanaka, K., 653 , 1131 , 1217 Turnbull, A., 983 , 1149
Sori, J., 625 Tanaka, S., 1141 Tushentsov, A.L., 1213
Sotiropoulou, A., 539 Tang, X.S., 723 Tvergaard, V., 977
Soucy, J.W., 67 Tang, Z., 751 Tyson, W.R., 901
Spagnoli, A., 1091 Tange, A., 1315 Tzavalas, S., 1285
1416 Author Index

U Wakita, M., 1217 Xing, L., 321

Ubertalli, G., 433
Uca, O., 1323
Uematsu, Y., 229 , 287 , 1009
Uesugi, K., 141
Ugrcic, M., 799
Ukadgaonker, V.G., 359
Ulbin, M., 1233
Umeno, Y., 899
Unterweger, K., 885
Utschig, T., 53
V Wang, S., 1357
Vaillant, F., 1151
Vaisberg, L.A., 785
Vakili-Tahami, F., 481
Valach, J., 571
Valentin, G., 1195 , 1199 ,
1201 , 1377
Vallee, T., 375
Vallini, P., 1353
Valvo, P.S., 1175
Vambersky, J., 597
Van Mier, J.G.M., 1355
Vantatori, S., 1091
Varfolomeyev, I., 213 , 607
Varna, J., 1275 , 1277
Vasic, S., 1207
Vasques, B., 79
Vavrik, D., 571 , 573
Vayas, I., 467
Veer, F., 301 , 597
Vegt, I., 501
Vehoff, H., 881
Veit, R., 455
Velasquez, J.D.P., 1319
Veljkovic, J.M., 807
VerganiK, L., 1231
Vergara, D., 215
Versteegen, J., 597
Viappiani, D., 1091
Vieville, P., 545
Vimercati, M., 1231
Vitorri, J.F., 771
Vliet, T., 439
Vogel, D., 729 , 735 , 737
Vogt, J.B., 887
Volkert, C., 697
Vorechovsky, M., 1365
Vormwald, M., 245 , 1221
Vrana, M., 1329
Vratnica, M., 1087
Vrellos, N., 1283
Vuherer, T., 1085
Vykydal, Z., 573
W Xia, K., 25
Waas, A., 739
Wainstein, J., 159
Walle, E. van, 599 Xu, D., 687
Walle, E., 1151 Xu, F., 699
Wallin, M., 341 Xu, G., 685
Walter, H., 731 , 737
Walter-Leberre, H., 143
Walther, F., 191 Y
Wang, D., 697 Yabuno, H., 783
Wang, G.S., 373 Yamashita, M., 1115
Wang, K., 1137 Yang, N., 695
Wang, K.Y., 675 Yang, N.Y.C., 669
Wang, M., 1029 Yang, W., 661
Wang, QY., 247 Yang, X.G., 667
Yang, Z., 741
Wang, Y., 929 Yasuoka, M., 85 , 267
Wang, Z., 659 , 689 Yates, J.R., 121, 151 , 427 ,
Wardle, G., 1369 443 , 465
Warren, A., 1331 Yellowlees, S., 1331
Was, G.S., 677 Yen, S.C., 1139
Wasmer, K., 61 , 117 , 479 Yildirim, B., 1177
Watanabe, J., 523 Yokogawa, K., 1003
Watts, J.F., 911 Yoon, K.B., 875
Weck, A., 129 Yosof-Sani, S.M.-R, 483
Weerheijm, J., 501 Youtsos, A.G., 1307 , 1311 ,
Weichert, D., 131 1313 , 1323
Weiss, B., 645 Yu, M.F., 659, 689
Weiss, M.P., 237
Wen, M., 1003
Wendler, E., 87 Z
Wenzel, H., 455 Zacharopoulos, D.A., 115 ,
Wesch, W., 87 559 , 567
Wetscher, F., 859 Zafosnik, B., 1237
Whitworth, H.A., 1097 Zahalka, F., 889
Wiesauer, K., 579 Zaïri, F., 1271
Wilkinson, D.S., 129 Zakrzewski, D., 203
Willatzen, M., 1339 Zant, N.P., 1027
Williams, J.G., 913 Zbib, H.M., 679
Williams, R.A., 47 Zehsaz, M., 269
Wimmer, S.A., 1345 Zeng, P., 667
Wimpory R. C. 1307 Zeravcic, V.S., 1081 , 1083
Wind, R.A., 693 Zerbst, U., 263
Winkler, H.-P., 513 Zhang, G.Q., 387
Winter, G., 241 Zhang, H., 497
Winzer, N., 979 Zhang, L., 1003
With, de G., 77 Zhang, T., 259
Withers, P.J., 427 , 897 , 1327 Zhang, X.B., 283 , 1119 , 1159
Wittler, O., 737 Zhang, Z., 1249
Wong, R.C.K., 747 Zhao, J., 259
Wong, R.H.C., 825 Zhilko, V.N., 449
Wood, P., 897 Zhitnikov, Y.V., 1401
Wu, B., 203 Zhou, M., 663 , 665
Wu, J.H., 217 Zhou, S., 983 , 1149
Wu, T., 1321 Zhu, W.S., 825
Wu, X.J., 259, 1249 Zielinski, W., 909
Ziering, A., 645
Zijl, G.P.A.G van, 475
X Zimprich, P., 645
Xi, Y., 1385 Zine, A., 769
Zivkovi, I., 1035
Xiao, Q.Z., 935 Zou, Y., 49
Xie, De, 739 Zrilic, M., 1047 , 1069