Theory: Z Soil - PC 2009 Manual

THEORY
Z Soil.PC 2009 manual
A. Truty Th. Zimmermann K. Podles

with contribution by A. Urba
nski, A. Wr
oblewski and Jean-Luc Sarf
Zace Services Ltd, Software engineering

P.O.Box 2, 1015 Lausanne, Switzerland
(T) +41 21 802 46 05
(F) +41 21 802 46 06
http://www.zace.com,
hotline: zsoil@zace.com
since 1982
WARNING
Z Soil.PC is regularly updated for minor changes. We recommend that you send us your e-mail, as Z Soil owner, so that we can inform
you of latest changes. Otherwise, consult our site regularly and download free upgrades to your version.
Latest updates to the manual are always included in the online help, so that slight differences with your printed manual will appear with
time; always refer to the online manual for latest version, in case of doubt.
Z Soil.PC 2009 manual:

1. Data preparation
2. Tutorials and benchmarks
3. Theory
ISBN 2-940009-08-2
c
Copyright 19852009
by Zace Services Ltd, Software engineering. All rights reserved.
Published by Elmepress International, Lausanne, Switzerland
END-USER LICENSE AGREEMENT FOR ZACEs Z SOIL.PC SOFTWARE

Read carefully this document, it is a legal agreement between you and Zace Services Ltd for the software product identified above. By
installing, copying, or otherwise using the software product identified above, you agree to be bound by the terms of this agreement. If
you do not agree to the terms of this agreement, promptly return the unused software product to the place from which you obtained it
for full refund (less shipping).
ZACE SERVICES LTD OFFERS A 60 DAYS MONEYBACK GUARANTEE ON Z SOIL.PC
Z SOIL.PC (the Software) SOFTWARE PRODUCT LICENSE

The software Z Soil.PC is protected by copyright laws and international copyright treaties, as well as other intellectual property laws
and treaties. The Z Soil.PC software product is licensed, not sold.
1. GRANT OF LICENSE
A: Zace Services Ltd grants you, the customer, a non-exclusive license to use copies of Z Soil.PC. you may install copies of Z Soil.PC
on an unlimited number of computers, provided that you use only one copy at the time.
B: You may make an unlimited number of copies of documents accompanying Z Soil.PC, provided that such copies shall be used only
for internal purposes and are not republished or distributed to any third party.
2. COPYRIGHT All title and copyrights in and to the Software product (including but not limited to images, photographs, text, applets,
etc), the accompanying materials, and any copies of Z Soil.PC are owned by Zace Services Ltd. Z Soil.PC is protected by copyright
laws and international treaties provisions. Therefore, you must treat Z Soil.PC like any other copyrighted material except that you
may make copies of the software for backup or archival purposes or install the software as stipulated under 1.
3. OTHER RIGHTS AND LIMITATIONS Limitations on Reverse Engineering, Decompilation, Disassembly. You may not reverse engineer,
decompile, or disassemble the Software
A: No separation of components. Z Soil.PC is licensed as a single product and neither the Softwares components, nor any upgrade
may be separated for use by more than one user at the time.
B: Rental. You may not rent or lease the software product.
C: Software transfer. You may permanently transfer all of your rights under this agreement, provided you do not retain any copies,
and the recepient agrees to all the terms of this agreement.
D: Termination. Without prejudice to any other rights, Zace Services Ltd may terminate this agreement if you fail to comply with the
conditions of this agreement. In such event, you must destroy all copies of the Software.
LIMITED WARRANTY
Zace Services Ltd. warrants that Z Soil.PC will a)perform substantially in accordance with the accompanying written material for a
period of 90 days from the date of receipt, and b) any hardware accompanying the product will be free from defects in materials and
workmanship under normal use and service for a period of one year, from the date of receipt.
CUSTOMER REMEDIES
Zace Services Ltd entire liability and your exclusive remedy shall be at Zaces option, either a)return of the price paid, or b) repair or
replacement of the software or hardware component which does not meet Zaces limited warranty, and which is returned to Zace Services
Ltd, with a copy of proof of payment of Z Soil.PC. This limited warranty is void if failure of the Software or hardware component has
resulted from accident, abuse, or misapplication. Any replacement of software or hardware will be warranted for the remainder of the
original warranty period or 30 days, whichever is longer.
NO OTHER WARRANTIES
TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, ZACE SERVICES LTD DISCLAIMS ALL OTHER WARRANTIES, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, WITH REGARD TO THE SOFTWARE PRODUCT, AND ANY ACCOMPANYING HARDWARE.
NO LIABILITY FOR CONSEQUENTIAL DAMAGES

TO THE MAXIMUM EXTENT PERMITTED BY LAW, IN NO EVENT SHALL ZACE SERVICES LTD BE LIABLE FOR ANY SPECIAL
INCIDENTAL, INDIRECT, OR CONSEQUENTIAL DAMAGES WHATSOEVER (INCLUDING, WITHOUT LIMITATION, DAMAGES
FOR LOSS OF BUSINESS, PROFITS, BUSINESS INTERRUPTION, LOSS OF BUSINESS INFORMATION, OR ANY OTHER PECUNIARY LOSS) ARISING OUT OF THE USE OF OR INABILITY TO USE THE SOFTWARE PRODUCT, EVEN IF ZACE SERVICES
LTD HAS BEEN ADVISED OF THE POSSIBITY OF SUCH DAMAGES.
HOTLINE
During the first year following purchase, hotline assistance will be provided by Zace Services Ltd, by fax or e-mail exclusively. This
service excludes all forms of consulting on actual projects. This hotline assistance can be renewed, for following years, at a cost of 10%
of current full package price.
THIS AGREEMENT IS GOVERNED BY THE LAWS OF SWITZERLAND
Lausanne, 1.01.2009
December 3, 2008
Z Soilr -3D-2PHASE v.2009
QuickHelp DataPrep Benchmarks Tutorials

TM4
Contents of Theoretical Manual

1 INTRODUCTION
25
1.1
NOTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
1.2
SOME IMPORTANT FORMULAE IN TENSOR ALGEBRA AND ANALYSIS
28
2 PROBLEM STATEMENT
37
2.1
SINGLE PHASE, SOLID MEDIUM . . . . . . . . . . . . . . . . . . . . . .
38
2.2
TWO-PHASE PARTIALLY SATURATED MEDIUM . . . . . . . . . . . . .
39
2.3
TRANSIENT FLOW . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
2.4
HEAT TRANSFER
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5
HUMIDITY TRANSFER
. . . . . . . . . . . . . . . . . . . . . . . . . . .
3 MATERIAL MODELS
3.1
ELASTICITY
3.2
CONSOLIDATION
3.3
3.4
45
47
49
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.1
GENERALIZED DARCY LAW
3.2.2
FLUID MOTION
50
54
. . . . . . . . . . . . . . . . . . . .
55
. . . . . . . . . . . . . . . . . . . . . . . . . . .
56
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
57
3.3.1
SKETCH OF THE PLASTICITY APPROACH . . . . . . . . . . . .
58
3.3.2
MOHRCOULOMB CRITERION . . . . . . . . . . . . . . . . . . .
60
3.3.3
DRUCKER-PRAGER CRITERION . . . . . . . . . . . . . . . . . . .
61
3.3.4
CAP MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
65
3.3.5
MOHR-COULOMB (M-W) . . . . . . . . . . . . . . . . . . . . . .
72
3.3.6
HOEKBROWN CRITERION (SMOOTH) . . . . . . . . . . . . . .
76
3.3.7
CUT-OFF CONDITION AND TREATMENT OF THE APEX . . . .
77
3.3.8
MULTILAMINATE MODEL
. . . . . . . . . . . . . . . . . . . . .
78
3.3.9
MODIFIED CAM CLAY MODEL . . . . . . . . . . . . . . . . . . .
82
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
86
PLASTICITY
CREEP
3.4.1
CREEP UNDER VARIABLE STRESS
December 3, 2008
. . . . . . . . . . . . . . . .
87

TM5
3.4.2
CREEP PARAMETER IDENTIFICATION FROM EXPERIMENTS
91
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
93
3.5
SWELLING
3.6
AGING CONCRETE
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
3.7
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
98
3.7.1
99
SAFETY FACTORS AND STRESS LEVELS . . . . . . . . . . . . .
4 NUMERICAL IMPLEMENTATION
4.1
4.2
4.3
101
WEAK FORM AND MATRIX FORMS OF THE PROBLEM
. . . . . . . . 102
4.1.1
SINGLE PHASE MEDIUM, TIME INDEPENDENT LOADING . . . . 103
4.1.2
TWO-PHASE MEDIUM, RHEOLOGICAL BEHAVIOUR . . . . . . . 104
4.1.3
HEAT TRANSFER
. . . . . . . . . . . . . . . . . . . . . . . . . . 106
ELEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.2.1
FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS . . . . 110
4.2.2
NUMERICAL INTEGRATION
4.2.3
STRAINS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.2.4
STIFFNESS MATRIX . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.2.5
BODY FORCES AND DISTRIBUTED LOADS
4.2.6
INITIAL STRESSES, STRAINS . . . . . . . . . . . . . . . . . . . . 115
. . . . . . . . . . . . . . . . . . . . 111
. . . . . . . . . . . 114
INCOMPRESSIBLE AND DILATANT MEDIA . . . . . . . . . . . . . . . . 116

4.3.1
INCOMPRESSIBLE MEDIA : B-BAR STRAIN PROJECTION METHOD117
4.3.2
DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD . . 119

4.3.2.1
INTRODUCTION TO ENHANCED ASSSUMED STRAIN

(EAS) APPROACH . . . . . . . . . . . . . . . . . . . . . 120
4.3.2.2
EXTENSION OF THE EAS METHOD TO NONLINEAR

ELASTOPLASTIC ANALYSIS . . . . . . . . . . . . . . . 124
4.3.2.3
REMARKS AND ASSESSMENT OF EAS ELEMENTS . . 125
4.4
FAR FIELD
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.5
OVERLAID MESHES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.6
ALGORITHMS
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.6.1
FULL/MODIFIED NEWTON-RAPHSON ALGORITHM . . . . . . . 135
4.6.2
CONVERGENCE NORMS
4.6.3
INITIAL STATE ANALYSIS
4.6.4
STABILITY ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . 141
4.6.5
ULTIMATE LOAD ANALYSIS
. . . . . . . . . . . . . . . . . . . . 143
4.6.6
CONSOLIDATION ANALYSIS
. . . . . . . . . . . . . . . . . . . . 144
December 3, 2008
. . . . . . . . . . . . . . . . . . . . . . 136
. . . . . . . . . . . . . . . . . . . . . 139

TM6
4.7
4.6.7
CREEP ANALYSIS
. . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.6.8
LOAD FUNCTION AND TIME STEPPING PROCEDURE
4.6.9
SIMULATION OF EXCAVATION AND CONSTRUCTION STAGES . 148
. . . . . 147
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
4.7.1
SHAPE FUNCTION DEFINITION AND REFERENCE ELEMENTS

FOR 2/3 D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.7.2
NUMERICAL INTEGRATION DATA FOR DIFFERENT ELEMENTS

IN 1/2/3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
4.7.3
MULTISURFACE PLASTICITY CLOSEST POINT PROJECTION ALGORITHM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
4.7.4
SINGLE SURFACE PLASTICITY CLOSEST POINT PROJECTION

ALGORITHM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
5 STRUCTURES
5.1
TRUSSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
5.1.1
5.1.2
5.1.3
TRUSS ELEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . 161

5.1.1.1
GENERAL IDEA OF TRUSS ELEMENT
5.1.1.2
GEOMETRY AND DOF OF TRUSS ELEMENTS . . . . . 164
5.1.1.3
INTERPOLATION OF THE DISPLACEMENTS AND STRAINS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
5.1.1.4
WEAK FORMULATION OF THE EQUILIBRIUM . . . . . 167
5.1.1.5
STIFFNESS MATRIX AND ELEMENT FORCE VECTOR
RING ELEMENT
5.1.4
. . . . . . . . . 162
168
. . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.1.2.1
GEOMETRY AND KINEMATICS OF A RING ELEMENT
5.1.2.2
WEAK FORMULATION OF THE EQUILIBRIUM . . . . . 171
5.1.2.3
STIFFNESS MATRIX AND ELEMENT FORCES . . . . . 172
ANCHORING OF TRUSS AND RING ELEMENTS

5.1.3.1
5.2
159
170
. . . . . . . . . 173
NUMERICAL REALIZATION OF ANCHORING . . . . . . 174
PRESTRESSING OF TRUSS AND RING ELEMENTS . . . . . . . 175
BEAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
5.2.1
GEOMETRY OF BEAM ELEMENT . . . . . . . . . . . . . . . . . 177
5.2.2
KINEMATICS OF BEAM THEORY
5.2.3
WEAK FORMULATION OF THE EQUILIBRIUM . . . . . . . . . . 184
5.2.4
INTERPOLATION OF THE DISPLACEMENT FIELD . . . . . . . . 186
5.2.5
STRAIN REPRESENTATION
5.2.6
STIFFNESS MATRIX AND ELEMENT FORCES
December 3, 2008
. . . . . . . . . . . . . . . . . 181
. . . . . . . . . . . . . . . . . . . . 187
. . . . . . . . . . 189

TM7
5.3
5.2.7
MASTER-CENTROID (OFFSET) TRANSFORMATION
. . . . . . 193
5.2.8
RELAXATION OF INTERNAL DOF . . . . . . . . . . . . . . . . . 194
5.2.9
BEAM ELEMENT RESULTS . . . . . . . . . . . . . . . . . . . . . 195
SHELLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
5.3.1
BASIC ASSUMPTIONS OF SHELL ANALYSIS
. . . . . . . . . . . 198
5.3.2
SETTING OF THE ELEMENT GEOMETRY
5.3.3
ELEMENT MAPPING AND COORDINATE SYSTEMS . . . . . . . 202
5.3.4
CROSSSECTION MODELS . . . . . . . . . . . . . . . . . . . . . 204
5.3.5
DISPLACEMENT AND STRAIN FIELD WITHIN THE ELEMENT . 206
5.3.6
TREATMENT OF TRANSVERSAL SHEAR . . . . . . . . . . . . . 209
5.3.7
WEAK FORMULATION OF EQUILIBRIUM . . . . . . . . . . . . . 210
5.3.8
5.3.9
LOADS
. . . . . . . . . . . . 199
. . . . . . . . . . 211
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
5.3.10 SHELL ELEMENT RESULTS . . . . . . . . . . . . . . . . . . . . . 213

5.4
5.5
MEMBRANES
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
5.4.1
ELEMENT GEOMETRY MAPPING AND COORDINATE SYSTEM
5.4.2
DISPLACEMENTS AND STRAINS FIELD . . . . . . . . . . . . . . 217
5.4.3
CONSTITUTIVE MODELS . . . . . . . . . . . . . . . . . . . . . . 219
5.4.4
WEAK FORMULATION OF EQUILIBRIUM . . . . . . . . . . . . . 221
5.4.5
6.2
. . . . . . . . . . 222
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
5.5.1
MASTER-SLAVE (OFFSET) TRANSFORMATION
5.5.2
SETTING THE DIRECTION ON SURFACE ELEMENTS . . . . . . 226
5.5.3
SETTING OF THE LOCAL BASE ON A SURFACE ELEMENT
5.5.4
UNI-AXIAL ELASTO-PLASTIC MATERIAL MODEL
6 INTERFACE
6.1
216
. . . . . . . . . 224
. . 227
. . . . . . . . 228
229
CONTACT OF SOLIDS AND FLUID INTERFACE . . . . . . . . . . . . . . 230

6.1.1
GENERAL OUTLOOK
6.1.2
DISPLACEMENTS AND STRAINS
6.1.3
CONSTITUTIVE MODEL
6.1.4
STIFFNESS MATRIX AND ELEMENT FORCE VECTOR . . . . . . 238
6.1.5
AUGMENTED LAGRANGIAN APPROACH
6.1.6
CONTRIBUTION TO CONTINUITY EQUATION . . . . . . . . . . 240
PILE CONTACT INTERFACE
December 3, 2008
. . . . . . . . . . . . . . . . . . . . . . . . 231
. . . . . . . . . . . . . . . . . 233
. . . . . . . . . . . . . . . . . . . . . . 234
. . . . . . . . . . . . . 239
. . . . . . . . . . . . . . . . . . . . . . . . 241

TM8
6.3
6.2.1
GENERAL OUTLOOK
. . . . . . . . . . . . . . . . . . . . . . . . 242
6.2.2
6.2.3
CONSTITUTIVE MODEL
6.2.4
. . . . . . . . . . . . . . . . . 243
. . . . . . . . . . . . . . . . . . . . . . 244
PILE FOOT CONTACT INTERFACE
. . . . . . . . . . . . . . . . . . . . 246
6.3.1
GENERAL OUTLOOK
. . . . . . . . . . . . . . . . . . . . . . . . 247
6.3.2
6.3.3
CONSTITUTIVE MODEL
6.3.4
. . . . . . . . . . . . . . . . . 248
. . . . . . . . . . . . . . . . . . . . . . 249
7 GEOTECHNICAL ASPECTS
251
7.1
TWO-PHASE MEDIUM
. . . . . . . . . . . . . . . . . . . . . . . . . . . 252
7.2
EFFECTIVE STRESSES
. . . . . . . . . . . . . . . . . . . . . . . . . . . 253
7.3
SOIL PLASTICITY
7.4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
7.3.1
DRUCKER-PRAGER VERSUS MOHR-COULOMB CRITERION . . . 255
7.3.2
CAP MODEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
7.3.3
DILATANCY
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
INITIAL STATE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260

7.4.1
COEFFICIENT OF EARTH PRESSURE AT REST, K0
7.4.2
STATES OF PLASTIC EQUILIBRIUM . . . . . . . . . . . . . . . . 263

7.4.2.1
MOHR-COULOMB MATERIAL
7.4.2.2
DRUCKER-PRAGER MATERIAL
. . . . . . . 261
. . . . . . . . . . . . . . 264
. . . . . . . . . . . . . 269
7.4.3
INFLUENCE OF POISSONS RATIO
. . . . . . . . . . . . . . . . 272
7.4.4
COMPUTATION OF THE INITIAL STATE
7.4.5
INFLUENCE OF WATER . . . . . . . . . . . . . . . . . . . . . . . 277
. . . . . . . . . . . . . 275
7.5
SOIL RHEOLOGY
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
7.6
ALGORITHMIC STRATEGIES . . . . . . . . . . . . . . . . . . . . . . . . 279

7.6.1
SEQUENCES OF ANALYSES
7.6.2
EXCAVATION, CONSTRUCTION ALGORITHM
December 3, 2008
. . . . . . . . . . . . . . . . . . . . 280
. . . . . . . . . . 281

TM9
December 3, 2008

TM10
Contents of Data Preparation

Acknowledgments
iii
System requirements
Getting started
vii
PREFACE
1 INTRODUCTION
1.1
SIGN CONVENTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2
DEFINITIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.1
UNIT WEIGHTS
1.2.2
ELASTICITY CONSTANTS
1.3
UNITS TABLE
. . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 DATA PREPARATION AND POSTPROCESSING DESCRIPTION

2.1
2.2
FILES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
2.1.1
. . . . . . . . . . . . . . . . . . . . . .
13
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
PARAMETRIC ANALYSIS
CONTROL
2.2.1
ANALYSIS & DRIVERS

2.2.1.1
. . . . . . . . . . . . . . . . . . . . . . .
16
PROBLEM TYPE AND DRIVERS . . . . . . . . . . . . .
19
2.2.1.1.1
SINGLE PHASE (DEFORMATION) ANALYSIS .
20
2.2.1.1.1.1
INITIAL STATE . . . . . . . . . . . . . .
21
2.2.1.1.1.2
TIME DEPENDENT
. . . . . . . . . . .
25
2.2.1.1.1.3
STABILITY . . . . . . . . . . . . . . . .
30
2.2.1.1.2
. . .
34
2.2.1.1.2.1
INITIAL STATE . . . . . . . . . . . . . .
35
2.2.1.1.2.2
TIME DEPENDENT
. . . . . . . . . . .
37
2.2.1.1.2.3
STABILITY . . . . . . . . . . . . . . . .
40
FLOW (STEADY STATE AND TRANSIENT) . .
44
2.2.1.1.3
DEFORMATION COUPLED WITH FLOW
2.2.1.1.3.1
December 3, 2008
INITIAL STATE . . . . . . . . . . . . . .
46

TM11
2.2.1.1.3.2
2.2.1.1.4
. . . . . . . . . . .
47
HEAT TRANSFER . . . . . . . . . . . . . . . .
49
2.2.1.1.4.1
INITIAL STATE . . . . . . . . . . . . . .
51
2.2.1.1.4.2
TIME DEPENDENT
. . . . . . . . . . .
52
. . . . . . . . . . . . .
54
2.2.1.1.5.1
INITIAL STATE . . . . . . . . . . . . . .
56
2.2.1.1.5.2
TIME DEPENDENT
. . . . . . . . . . .
57
. . . . . . . . . . . .
59
. . . . . . . . . . . . . . . . . . .
63
2.2.1.2
ANALYSIS TYPES . . . . . . . . . . . . . . . . . . . . .
67
2.2.1.3
ASSOCIATED PREPROCESSED PROJECTS . . . . . . .
68
2.2.1.1.5
2.3
TIME DEPENDENT
HUMIDITY TRANSFER
2.2.1.1.6
TRANSIENT DYNAMICS
2.2.1.1.7
PUSHOVER
2.2.2
CONTROL
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
2.2.3
DYNAMICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
70
2.2.4
PUSHOVER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
71
2.2.5
CONTACT ALGORITHM . . . . . . . . . . . . . . . . . . . . . . .
72
2.2.6
LINEAR EQUATION SOLVERS
. . . . . . . . . . . . . . . . . . .
73
2.2.7
UNITS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
2.2.8
FINITE ELEMENTS
. . . . . . . . . . . . . . . . . . . . . . . . .
78
2.2.9
RESULTS CONTENT
. . . . . . . . . . . . . . . . . . . . . . . .
81
ASSEMBLY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84
2.3.1
85
PREPROCESSING
2.3.1.1
. . . . . . . . . . . . . . . . . . . . . . . . . .
USER INTERFACE, MAIN MENU AND BASIC TOOLS
86
. . . . . . . . . . . .
87
. . . . . . . . . . . . . . . . . . .
88
2.3.1.1.2.1
FILE . . . . . . . . . . . . . . . . . . . .
89
2.3.1.1.2.2
EDIT . . . . . . . . . . . . . . . . . . . .
90
2.3.1.1.2.3
VIEW . . . . . . . . . . . . . . . . . . . .
93
2.3.1.1.2.4
VISUALIZATION . . . . . . . . . . . . . .
95
2.3.1.1.2.5
CURSOR MODE . . . . . . . . . . . . . .
97
2.3.1.1.2.6
SELECTIONS . . . . . . . . . . . . . . .
99
2.3.1.1.2.7
ELEMENT NODES SELECTIONS . . . . 105
2.3.1.1.2.8
TOOLS . . . . . . . . . . . . . . . . . . . 106
2.3.1.1.2.9
SETTINGS . . . . . . . . . . . . . . . . . 107
2.3.1.1.2.10
LAYERS . . . . . . . . . . . . . . . . . . 110
2.3.1.1.1
Main preprocessor window
2.3.1.1.2
Popup menu
December 3, 2008

TM12
2.3.1.1.2.11
2.3.1.1.3
2.3.1.2
HELP . . . . . . . . . . . . . . . . . . . . 112
Toolbars . . . . . . . . . . . . . . . . . . . . . . 113
MACRO-MODEL . . . . . . . . . . . . . . . . . . . . . . 127
2.3.1.2.1
POINT
. . . . . . . . . . . . . . . . . . . . . . 128
2.3.1.2.2
OBJECTS . . . . . . . . . . . . . . . . . . . . . 130
2.3.1.2.3
2D mesh mapping
. . . . . . . . . . . . . . . . 140
2.3.1.2.4
Extrusion direction
. . . . . . . . . . . . . . . . 143
2.3.1.2.5
Subdomain
2.3.1.2.6
SEEPAGE . . . . . . . . . . . . . . . . . . . . . 185
2.3.1.2.7
CONVECTION . . . . . . . . . . . . . . . . . . 188
2.3.1.2.8
INTERFACE
2.3.1.2.9
PRESSURE BC . . . . . . . . . . . . . . . . . . 195
. . . . . . . . . . . . . . . . . . . . 144
. . . . . . . . . . . . . . . . . . . 191
2.3.1.2.10 TEMPERATURE BC . . . . . . . . . . . . . . . 199

2.3.1.2.11 HUMIDITY BC . . . . . . . . . . . . . . . . . . 202
2.3.1.2.12 FLUID FLUX . . . . . . . . . . . . . . . . . . . 205
2.3.1.2.13 HEAT FLUX
. . . . . . . . . . . . . . . . . . . 208
2.3.1.2.14 HUMIDITY FLUX
. . . . . . . . . . . . . . . . 211
2.3.1.2.15 SURFACE LOAD . . . . . . . . . . . . . . . . . 214

2.3.1.2.16 PILE . . . . . . . . . . . . . . . . . . . . . . . . 222
2.3.1.2.17 AUXILIARY PLANES . . . . . . . . . . . . . . . 227
2.3.1.2.18 POINT LOAD . . . . . . . . . . . . . . . . . . . 232
2.3.1.3
FE MODEL . . . . . . . . . . . . . . . . . . . . . . . . . 235
2.3.1.3.1
COMMON METHODS FOR ALL FE MODEL COMPONENTS . . . . . . . . . . . . . . . . . . . . . 236
2.3.1.3.2
NODE . . . . . . . . . . . . . . . . . . . . . . . 244
2.3.1.3.3
BEAM . . . . . . . . . . . . . . . . . . . . . . . 249
2.3.1.3.4
ANCHOR . . . . . . . . . . . . . . . . . . . . . 256
2.3.1.3.5
CONTINUUM 2D . . . . . . . . . . . . . . . . . 264
2.3.1.3.6
CONTINUUM 3D . . . . . . . . . . . . . . . . . 269
2.3.1.3.7
SHELL . . . . . . . . . . . . . . . . . . . . . . . 275
2.3.1.3.8
SHELL-ONE LAYER . . . . . . . . . . . . . . . . 281
2.3.1.3.9
MEMBRANE . . . . . . . . . . . . . . . . . . . 286
2.3.1.3.10 INFINITE ELEMENT . . . . . . . . . . . . . . . 291

2.3.1.3.11 INTERFACE (SMALL DEFORMATION) . . . . . 296
2.3.1.3.12 INTERFACE (LARGE DEFORMATION) . . . . . 305
December 3, 2008

TM13
2.3.1.3.13 SEEPAGE . . . . . . . . . . . . . . . . . . . . . 309

2.3.1.3.14 CONVECTION . . . . . . . . . . . . . . . . . . 313
2.3.1.3.15 SHELL HINGES . . . . . . . . . . . . . . . . . . 317
2.3.1.3.16 BOUNDARY CONDITIONS . . . . . . . . . . . . 320
2.3.1.3.16.1
SOLID BC . . . . . . . . . . . . . . . . . 321
2.3.1.3.16.2
TEMPERATURE BC . . . . . . . . . . . 328
2.3.1.3.16.3
HUMIDITY BC . . . . . . . . . . . . . . 333
2.3.1.3.16.4
PRESSURE BC . . . . . . . . . . . . . . 334
2.3.1.3.17 LOADS . . . . . . . . . . . . . . . . . . . . . . 340

2.3.1.3.17.1
BODY LOAD . . . . . . . . . . . . . . . 341
2.3.1.3.17.2
NODAL LOAD . . . . . . . . . . . . . . . 343
2.3.1.3.17.3
SURFACE LOAD . . . . . . . . . . . . . . 345
2.3.1.3.17.4
BEAM LOAD . . . . . . . . . . . . . . . 352
2.3.1.3.18 ADDED MASSES . . . . . . . . . . . . . . . . . 354

2.3.1.3.18.1
NODAL ADDED MASSES . . . . . . . . 355
2.3.1.3.18.2
DISTRIBUTED ADDED MASSES . . . . 357
2.3.1.3.18.3
FILTERING OF ADDED MASSES . . . . 359
2.3.1.3.19 DISTRIBUTED FLUX . . . . . . . . . . . . . . . 361

2.3.1.3.19.1
FLUID FLUX . . . . . . . . . . . . . . . . 362
2.3.1.3.19.2
HEAT FLUX . . . . . . . . . . . . . . . . 365
2.3.1.3.19.3
HUMIDITY FLUX . . . . . . . . . . . . . 366
2.3.1.3.20 INITIAL CONDITION . . . . . . . . . . . . . . . 367

2.3.1.3.20.1
INITIAL PRESSURE . . . . . . . . . . . . 368
2.3.1.3.20.2
INITIAL TEMPERATURE
2.3.1.3.20.3
INITIAL HUMIDITY . . . . . . . . . . . . 370
2.3.1.3.20.4
INITIAL STRESS . . . . . . . . . . . . . 371
2.3.1.3.20.5
INITIAL STRAIN
2.3.1.3.20.6
INITIAL DISPLACEMENT/VELOCITY . . 375
. . . . . . . . 369
. . . . . . . . . . . . . 373
2.3.1.3.21 KINEMATIC CONSTRAINTS . . . . . . . . . . . 378

2.3.1.3.22 NODAL LINK . . . . . . . . . . . . . . . . . . . 380
2.3.1.3.23 MESH TYING . . . . . . . . . . . . . . . . . . . 383
2.3.2
2.3.1.4
PUSHOVER CONTROL NODE
2.3.1.5
BOREHOLES . . . . . . . . . . . . . . . . . . . . . . . . 386
MESH INFO
December 3, 2008
. . . . . . . . . . . . . . 385
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391

TM14
2.3.3
MATERIALS
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
2.3.3.1
MATERIAL DATA BASE . . . . . . . . . . . . . . . . . . 393
2.3.3.2
CONTINUUM / STRUCTURE TYPE . . . . . . . . . . . 394

2.3.3.2.1
BEAMS AND AXISYMMETRIC SHELLS . . . . . 395
2.3.3.2.1.1
BEAM MODEL . . . . . . . . . . . . . . 396
2.3.3.2.1.2
AXISYMMETRIC SHELL MODEL . . . . 405
2.3.3.2.2
CONTACT
. . . . . . . . . . . . . . . . . . . . 409
2.3.3.2.3
PILE INTERFACE . . . . . . . . . . . . . . . . . 413
2.3.3.2.4
PILE FOOT INTERFACE . . . . . . . . . . . . . 416
2.3.3.2.5
CONTINUUM . . . . . . . . . . . . . . . . . . . 419
2.3.3.2.5.1
ELASTIC
. . . . . . . . . . . . . . . . . 420
2.3.3.2.5.2
UNIT WEIGHT . . . . . . . . . . . . . . 421
2.3.3.2.5.3
FLOW . . . . . . . . . . . . . . . . . . . 423
2.3.3.2.5.4
CREEP
2.3.3.2.5.5
INITIAL K0 STATE . . . . . . . . . . . . 428
2.3.3.2.5.6
HEAT . . . . . . . . . . . . . . . . . . . 430
2.3.3.2.5.7
HUMIDITY
2.3.3.2.5.8
LOCAL STABILITY . . . . . . . . . . . . 433
2.3.3.2.5.9
CONTINUUM: ELASTIC . . . . . . . . . 434
2.3.3.2.5.10
CONTINUUM: AGING CONCRETE
2.3.3.2.5.11
CONTINUUM: CAP
2.3.3.2.5.12
CONTINUUM: MODIFIED CAM-CLAY
2.3.3.2.5.13
CONTINUUM: DRUCKERPRAGER . . . 442
2.3.3.2.5.14
CONTINUUM: DUNCANCHANG . . . . 444
2.3.3.2.5.15
CONTINUUM: HOEKBROWN (M-W)
2.3.3.2.5.16
CONTINUUM: MOHRCOULOMB (M-W)

. . . . . . . . . . . . . . . . . . . . . . . 447
2.3.3.2.5.17
CONTINUUM: MULTILAMINATE . . . . 449
2.3.3.2.5.18
CONTINUUM: RANKINE (M-W)
2.3.3.2.5.19
CONTINUUM: MOHR-COULOMB . . . . 452
2.3.3.2.5.20
CONTINUUM: ECP-HUJEUX
2.3.3.2.5.21
CONTINUUM: HS-small
. . . . . . . . . . . . . . . . . . 426
. . . . . . . . . . . . . . . . 432
. . . 435
. . . . . . . . . . . 437
. 440
. 445
. . . . 451
. . . . . . 454
. . . . . . . . . 457
2.3.3.2.6
CONTINUUM FOR STRUCTURES . . . . . . . . 460
2.3.3.2.7
HEAT CONVECTION
2.3.3.2.8
HUMIDITY CONVECTION
December 3, 2008
. . . . . . . . . . . . . . 461
. . . . . . . . . . . 462

TM15
2.3.3.2.9
INFINITE MEDIA . . . . . . . . . . . . . . . . . 463
2.3.3.2.10 MEMBRANES
. . . . . . . . . . . . . . . . . . 464
2.3.3.2.10.1
FIBER MODEL . . . . . . . . . . . . . . 465
2.3.3.2.10.2
PLANE STRESS MODELS . . . . . . . . 467
2.3.3.2.10.3
MEMBRANE - FABRIC MODELS
. . . . 469
2.3.3.2.11 SEEPAGE . . . . . . . . . . . . . . . . . . . . . 471

2.3.3.2.12 MODEL FOR SHELLS . . . . . . . . . . . . . . 472
2.3.3.2.13 TRUSS (ANCHOR) MODEL . . . . . . . . . . . 479
2.4
2.5
2.3.3.3
NON-STANDARD MATERIAL DATA . . . . . . . . . . . 482
2.3.3.4
MATERIAL DATA VALUES
. . . . . . . . . . . . . . . . 484
2.3.4
EXISTENCE FUNCTION . . . . . . . . . . . . . . . . . . . . . . . 487
2.3.5
LOAD FUNCTION
2.3.6
GRAVITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492
2.3.7
SEISMIC INPUT . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
. . . . . . . . . . . . . . . . . . . . . . . . . . 490
ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
2.4.1
RESTART . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 496
2.4.2
BATCH PROCESSING . . . . . . . . . . . . . . . . . . . . . . . . 497
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 498
2.5.1
POSTPROCESSING . . . . . . . . . . . . . . . . . . . . . . . . . . 499
2.5.1.1
HOW DO I...... . . . . . . . . . . . . . . . . . . . . . . . 500
2.5.1.2
HOW TO USE MACRO
2.5.1.3
WORKING OUT RESULTS FROM PARAMETRIC ANALYSIS504
. . . . . . . . . . . . . . . . . . 502
2.6
EXTRAS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508
2.7
SYSTEM CONFIGURATION . . . . . . . . . . . . . . . . . . . . . . . . . 509
2.8
HELP
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
3 TROUBLESHOOTING
3.1
511
CALCULATION MODULE
. . . . . . . . . . . . . . . . . . . . . . . . . . 512
3.1.1
ERRORS AND WARNINGS HANDLING . . . . . . . . . . . . . . . 513
3.1.2
FREQUENTLY ASKED QUESTIONS
December 3, 2008
. . . . . . . . . . . . . . . . 519

TM16
Contents of Tutorial
PREFACE
23
1 2D PROBLEMS
25
1.1
CUT STABILITY ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . .
26
1.2
HOW TO RUN AN AXISYMMETRIC PROBLEM WITH A DRIVEN EXTERNAL FORCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
1.3
HOW TO RUN A CONSOLIDATION PROBLEM
. . . . . . . . . . . . . .
33
1.4
HOW TO RUN SHEET-PILE WALL PROBLEM . . . . . . . . . . . . . . .
37
1.5
HOW TO RUN A STEADY-STATE FLOW PROBLEM . . . . . . . . . . .
41
1.6
HOW TO SIMULATE A TUNNEL IN AN URBAN ENVIRONMENT . . . .
44
1.7
HEAT TRANSFER FOLLOWED BY MECHANICAL ANALYSIS . . . . . . .
51
1.7.1
THERMAL ANALYSIS
55
1.7.2
MECHANICAL ANALYSIS (TUNNEL LINING MODELED WITH CONTINUUM ELEMENTS) . . . . . . . . . . . . . . . . . . . . . . . . 56
1.7.3
MECHANICAL ANALYSIS (TUNNEL LINING MODELED WITH BEAMS)

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
. . . . . . . . . . . . . . . . . . . . . . . .
2 3D PROBLEMS
61
2.1
CONCRETE BOX CONTAINER
. . . . . . . . . . . . . . . . . . . . . . .
62
2.2
DRAINING CONCRETE DAM
. . . . . . . . . . . . . . . . . . . . . . . .
67
2.3
REINFORCED SOIL ABUTMENT . . . . . . . . . . . . . . . . . . . . . .
71
2.4
FOUNDATION RAFT STRENGTHENED BY PILES
78
December 3, 2008
. . . . . . . . . . . .

TM17
December 3, 2008

TM18
Contents of Benchmarks
PREFACE
1 INTRODUCTION
2 ELEMENTARY BOUNDARY VALUE PROBLEMS
2.1
BOX-SHAPED MEDIUM, PLANE STRAIN AND AXISYMMETRY . . . . .
10
2.2
BOX-SHAPED MEDIUM, WITH WATER BOUNDARY CONDITIONS . . .
12
2.3
PLANE STRAIN BOX-SHAPED MEDIUM WITH OVERPRESSURE . . . .
14
3 SOIL MECHANICS BENCHMARKS

3.1
3.2
3.3
LOAD CARRYING AND SETTLEMENTS OF FOUNDATIONS . . . . . . .
16
3.1.1
SUPERFICIAL FOUNDATION (PLANE STRAIN) . . . . . . . . . .
17
3.1.2
EMBEDDED FOUNDATION . . . . . . . . . . . . . . . . . . . . .
19
3.1.3
AXISYMMETRIC SUPERFICIAL FOUNDATION
. . . . . . . . . .
23
STABILITY ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
3.2.1
SLOPES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
3.2.2
SEISMIC BEARING CAPACITY OF STRIP FOOTING ON SLOPES
29
3.2.3
SLOPE STABILITY IN PRESENCE OF SEEPAGE FLOW . . . . . .
32
PRESTRESS
3.3.1
3.4
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SINGLE ANCHOR
. . . . . . . . . . . . . . . . . . . . . . . . . .
EXCAVATION AND CONSTRUCTION STAGES

3.4.1
3.5
15
34
35
. . . . . . . . . . . . . .
37
EXCAVATION WITH PROGRESSIVE UNLOADING . . . . . . . . .
38
CONSOLIDATION PROBLEMS
. . . . . . . . . . . . . . . . . . . . . . .
39
3.5.1
OEDOMETRIC TEST (TERZAGHI CONSOLIDATION)
. . . . . .
40
3.5.2
TERZAGHI CONSOLIDATION, TWO LAYERS MEDIUM . . . . . .
42
3.5.3
TERZAGHI CONSOLIDATION, TWO LAYERS WITH WATER TABLE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
43
3.5.4
TWODIMENSIONAL FOOTING SETTLEMENT . . . . . . . . . .
44
3.5.5
ELASTOPLASTIC COMPRESSION (COMP.INP) . . . . . . . . . .
45
December 3, 2008

TM19
3.6
CREEP (CREEP1.INP) . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.7
SWELLING
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
OEDOMETER TEST UNDER FORCE CONTROL

(SWELL FCTRL.INP) . . . . . . . . . . . . . . . . . . . . . . . .
48
OEDOMETER UNLOADING-LOADING TEST UNDER FORCE CONTROL (SWELL UNLREL.INP) . . . . . . . . . . . . . . . . . . . .
50
INFINITE MEDIA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
3.8.1
A GAP IN INFINITE MEDIUM . . . . . . . . . . . . . . . . . . . .
52
3.8.2
HALF-SPACE UNDER COMPRESSIVE LOAD (3D)
. . . . . . . .
53
3.8.3
CIRCULAR CAVITY UNDER THE PRESSURE
. . . . . . . . . . .
54
3.7.1
3.7.2
3.8
4 FLOW BENCHMARKS
57
4.1
RECTANGULAR DAM WITH TAILWATER . . . . . . . . . . . . . . . . .
58
4.2
RECTANGULAR DAM WITH TOEDRAIN . . . . . . . . . . . . . . . . .
59
4.3
MODELLING BOUNDARY CONDITIONS FOR TRANSIENT AND STEADY

STATE FLOW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
5 HEAT PROBLEMS
5.1
63
TRANSIENT HEAT PROBLEM . . . . . . . . . . . . . . . . . . . . . . . .
6 STRUCTURAL BENCHMARKS
6.1
6.2
6.3
46
64
65
BEAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
66
6.1.1
ELASTO-PLASTIC FIXED-END BEAM . . . . . . . . . . . . . . .
67
6.1.2
ELASTO-PLASTIC BEAM WITH SUPPORTS VARIABLE IN TIME
69
6.1.3
REINFORCED CONCRETE 2-SPAN BEAM . . . . . . . . . . . . .
71
6.1.4
REINFORCED CONCRETE 2-FLOOR FRAME
. . . . . . . . . . .
73
6.1.5
TWISTED BEAM . . . . . . . . . . . . . . . . . . . . . . . . . . .
75
6.1.6
RING
76
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
AXISYMMETRIC SHELLS
. . . . . . . . . . . . . . . . . . . . . . . . . .
78
6.2.1
TUBE TO SPHERE CONNECTION . . . . . . . . . . . . . . . . .
79
6.2.2
CYLINDER SUBJECTED TO PRESSURE . . . . . . . . . . . . . .
80
6.2.3
CIRCULAR ELASTO-PLASTIC PLATE
. . . . . . . . . . . . . . .
81
SHELLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
82
6.3.1
SCORDELIS-LO ROOF . . . . . . . . . . . . . . . . . . . . . . . .
83
6.3.2
TWISTED BEAM (SHELL MODEL) . . . . . . . . . . . . . . . . .
84
6.3.3
HEMISPHERE
85
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
concerns versions: ACADEMIC, PROFESSIONAL, EXPERT only
December 3, 2008

TM20
6.4
6.3.4
SQUARE ELASTOPLASTIC PLATE . . . . . . . . . . . . . . . . .
86
6.3.5
ELASTOPLASTIC CYLINDRICAL SHELL . . . . . . . . . . . . . .
87
MEMBRANES
6.4.1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
88
SOIL SLOPE REINFORCED BY MEMBRANES . . . . . . . . . . .
89
7 SOIL-STRUCTURE INTERACTION BENCHMARKS
91
7.1
CYLINDRICAL CONTAINER . . . . . . . . . . . . . . . . . . . . . . . . .
92
7.2
DIAPHRAGM WALL
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
7.3
BURRIED PIPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
98
7.4
PILE 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
December 3, 2008

TM21
December 3, 2008

TM22
Preface
Document THEORY MANUAL presents the constitutive model and the finite element implementation. The discussion is limited to features which are actually used in the program.
No attempt is made to give a general overview of numerical methods in soil mechanics.
The proposed models include elasticity, various plasticity models, time dependent behavior
resulting from consolidation, and actual creep.
Sign convention are different in continuum and soil mechanics. Both sign conventions are
used in this text; variables which are positive in compression are underlined, in order to avoid
possible confusion.
INTRODUCTION
PROBLEM STATEMENT
MATERIAL MODELS
NUMERICAL IMPLEMENTATION
STRUCTURES
INTERFACE
GEOTECHNICAL ASPECTS
December 3, 2008

TM23
December 3, 2008

TM24
N Preface
Chapter 1
INTRODUCTION
The following sections present the constitutive model and the finite element implementation.
The discussion is limited to features which are actually used in the program. No attempt is
made to give a general overview of numerical methods in soil mechanics.
The proposed models include elasticity, various plasticity models, time dependent behavior
resulting from consolidation, and creep.
Sign convention are different in continuum and soil mechanics. Both sign conventions are
used in this text; variables which are positive in compression are underlined, in order to avoid
possible confusion.
December 3, 2008

TM25
N Preface N N Introduction
1.1
NOTATION
General rules
underlined variables (like stress , pressure p, etc.) are positive in compression
overbarred symbols mean the parameters with prescribed (known) value (
p, u)
abbreviations in sub/superscripts
cr
e
e
eq
Ext
F
L
m
max
min
n
p
tot
0
creep
elastic part
element
equilibrium
external
fluid
liquid
mean
maximum
minimum
normal direction
plastic part
total
reference state
infinity
Symbols
Latin symbol
SI units
bi , b
N/m3
Cijkl , C
N/m2
Dijkl , D
N/m2
G
N/m2
g
m/s2
E
N/m2
e
eij
kij , k
m/s
K
N/m2
ni , n
ui , u
m
p
N/m2
Q kg/(s m2 )
Vi , v
m/s
MC
R
N/m2
S
Sr
sij
N/m2
December 3, 2008
Meaning
body force vector
compliance constitutive tensor
stiffness
Kirchhoff modulus
earth acceleration
Young modulus
void ratio
strain deviator
permeability tensor
bulk modulus (solid)
normal vector
porosity
displacement vector
pressure
mass source
relative fluid velocity
radius of the MohrCoulomb circle
saturation coefficient
residual saturation coefficient
stress deviator

TM26
Greek symbol
SI units
Meaning
ij ,
=
p
q
s
t
u
ij ,
0ij , 0
ij ,
Ns/m2
N/m3
m
m
m
m
m
m
N/m2
kg/m3
N/m2
N/m2
N/m2
strain tensor
fluid viscosity
specific weight
boundary of the domain
boundary with imposed pressure conditions
boundary with imposed flow conditions
boundary with imposed seepage (i.e. pressure dependent) flow conditions
boundary with imposed traction conditions
boundary with imposed displacement conditions
Lame constant
Poisson coefficient
mass density
stress tensor
effective stress tensor
shear stress tensor
December 3, 2008

TM27
1.2
SOME IMPORTANT FORMULAE IN TENSOR ALGEBRA AND

ANALYSIS
In mechanics several tensorial variables of different rank are used. Examples are:
scalar (zeroth rank tensor) density , temperature T , energy W , . . .
vector (first rank tensor) displacement vector u, velocity vector v, . . .
dyad (second rank tensor) stress tensor , deformation tensor , . . .
fourth, sixth and higher rank tensor material tensor . . .
Some rules of calculations with tensors in the three-dimensional Euclidean space are presented
in this section. The direct (symbolic) and the component notation of tensor quantities are
used. For shorter writing we introduce the Einsteins summation convention (repeated index
in some term in the expression requires summation)
. . . + ai b i + . . . = +
3
X
ai b i + . . . = . . . + a1 b 1 + a2 b 2 + a3 b 3 + . . .
i=1
i is a dummy index.
December 3, 2008

TM28
Window 1-1: Scalars, vectors and tensors
SCALAR
Scalars are variables, which are fully independent on the choice of coordinate system (invariant
variables) because they have no orientation.
VECTOR
Vectors can be written as
a = a1 e1 + a2 e2 + a3 e3 ,
a = (a1 , a2 , a3 ),
a = (ai ), i = 1, 2, 3.
The ai are the coordinates of the vector, which are related to the vector basis ei with respect
to the given coordinate system. This vector basis is assumed to be an orthonormal basis

1 i=j
|ei | = 1, ei ej =
0 i 6= j
The scalar product (inner product, dot product) of two vectors a and b is defined as

1 i=j
a b = ai ei bj ej = ai bj ei ej = ai bj ij = ai bi = ; ij =
0 i 6= j
The dyadic product of two vectors a and b
ab = ai ei bj ej = ai bj ei ej = Tij ei ej = T.
In some textbooks for this product the following designation is used
a b ab.
SECOND RANK TENSOR With the help of the dyadic product the second rank tensor
T can be introduced
T = ab = ai bj ei ej = Tij ei ej .
For the second rank tensors T and S we define the following products:
Scalar product e.g. tensor product with the contraction
T : S = Tij ei ej : Skl ek el = Tij Skl ei jk el = Tij Sjl ei el = Mil ei el ,
which leads to a second rank tensor.
Double scalar product e.g. tensor product with the double contraction
T :: S = Tij ei ej :: Skl ek el = Tij Skl jk il = Tij Sji = ,
resulting in a scalar.
HIGHER RANK TENSOR
In a similar way (by dyadic product) we can define tensors of higher ranks:
the 4th rank tensor
the 6th rank tensor
(4)
A = abcd = ai bj ck dl ei ej ek el = TS =
Tij Skl ei ej ek el = Aijkl ei ej ek el
(6)
B = abcdgh = TSP =
Tij Skl Pmn ei ej ek el em en = Bijklmn ei ej ek el em en .
Also tensor products (from uni- to multi- scalar) can be defined.

Window 1-1
December 3, 2008

TM29
Window 1-2: Special for second rank tensors
Unit tensor (identity tensor) I
I = ij ei ej
Transposed tensor TT
T = ab = TT = ba ,
T = Tij ei ej = TT = Tij ej ei = Tji ei ej ,
Symmetric tensor
If T = TT (Tij = Tji ) then tensor T is symmetric
Antisymmetric tensor
If T = TT (Tij = Tji ) then tensor T is antisymmetric
Trace of the tensor
trT = I :: T = Tii = T11 + T22 + T33 .
Tensor decomposition
The tensor T may be decomposed into two parts: axiator and deviator defined as follow:
Axiator (spherical tensor); denoted AT or TA
1
1
AT = (I :: T)I = Tkk ij ei Ij
3
3
Deviator denoted DT or TD
1
DT = T AT = (Tij Tkk ij )ei ej
3
so
T = AT + DT
Window 1-2
December 3, 2008

TM30
Window 1-3: Transformation rules for tensors
The rules of transformation from one coordinate system to a rotated system (marked with )
for tensors of the rank 2, 4 or 6 are (all indices range from 1 to 3)
a0ij = mi nj amn ,
b0ijkl = mi nj sk tl bmnst ,
c0ijklop = mi nj sk tl uo vp cmnstuv .
The ij are the elements of the transformation matrix (direction cosines):
ij = cos(e0i , ej ).
Window 1-3
Window 1-4: Eigenvalue problem for a second rank tensor
The eigenvalues and the eigenvectors (eigendirections) n for a second rank tensor T can
be obtained from the solution of the following equations
(T I) n = 0,
(Tij ij )nj = 0
The eigenvalues follow from the condition that nontrivial solutions are existing, which leads
to the characteristic equation:
det(T I) = 0;
det(Tij ij ) = 0.
(1)
The roots of this equation () , = 1, 2, 3 sort in the ascending order (e.g. 1 2 3 )

are called the principal values. It can be shown that in the case of symmetric second rank
tensors all principle values are real.
()
For each root we get the eigenvector (eigendirections, principal directions) nj , = 1, 2, 3

from the system
(T11 )n1 + T12 n2 + T13 n3
T21 n1 + (T22 )n2 + T23 n3
T31 n1 + T32 n2 + (T33 )n3
n21 + n22 + n23
=
=
=
=
0,
0,
0,
1.
Window 1-4
December 3, 2008

TM31
Window 1-5: Invariants of a second rank tensor
Invariant terms are independent on the choice of the coordinate system. Such a system of
invariants can be related to the coefficients of the characteristic equation (1) rewrite in the
form:
det(T I) = 3 I1 (T)2 + I2 (T) I3 (T) = 0.
(1)
The Ii are called principal invariants, defined as:
Linear principal invariant
I1 (T) = trT T :: I Tii ,
Quadratic principal invariant
I2 (T) =
1
1 2
J1 (T) J1 (T2 ) = (Tii Tjj Tij Tji ),
2
2
Cubic principal invariant

1
J1 (T3 ) + 3J1 (T)J2 (T) J13 (T)
3
1
1
1
J1 (T3 ) J1 (T2 )J1 (T) + J13 (T)
=
3
2
6
= det(Tij ).
I3 (T) =
In the stress space very often we use invariants of the stress deviator D = sij :
J1 = 0
1
sij sji
J2 =
2
1
J3 =
sij sjk ski
3
Cylindrical invariants of stress tensor (Haigh and Westergaard):
1
= I1
3
3 3
3/2
cos 3 =
J3 J2
p2
=
2J2
Invariants commonly used in geotechnical material models are:
1
p = I1
p3
q = 3J2
Window 1-5
December 3, 2008

TM32
Window 1-6: Cayley-Hamilton theorem
The second rank tensor satisfies characteristic equation
T3 I1 (T)T2 + I2 (T)T I3 (T)I = 0,
which enables the representation of Tn (n 3) as a linear function of T2 , T, T0 = I, e.g.,
T3 = I1 (T)T2 I2 (T)T + I3 (T)I.
Window 1-6
Window 1-7: Derivatives of the invariants of a second rank tensor
A scalar-valued function of a second rank tensor can be represented by
= (T) = (T11 , T22 , . . . , T31 ).
Then we can calculate the derivative by the following equation
,T =
=
ek el .
T
Tkl
On the other hand the derivatives of the invariants are

T
J1 (T),T = I, J1 (T2 ),T = 2TT ,

J2 (T),T = J1 (T)I TT ,
J1 (T3 ),T = 3T2 ,
J3 (T),T = T2 J1 (T)TT + J2 (T)I = J3 (T)(TT )1 .

So, we finally get

[J1 , J2 , J3 ],T =
+ J1
+ J2
J1
J2
J3

I
+ J1
J2
J3
TT +
2 T
T .
J3
These calculations can be helpful for the use of the representation theorem of an isotropic
function
P = F(T) = 0 I + 1 T + 2 T2 .
The coefficients i itself are functions of the invariants
i = i [J1 (T), J2 (T), J3 (T)] .
(1)
Window 1-7
December 3, 2008

TM33
Window 1-8: Transition from tensorial to matrix notation
It is assumed that stress/strain components are ordered in column vectors as follows:

T
= xx yy xy zz xz yz

T
= xx yy xy zz xz yz
The following table is used to extract vector components from appropriate tensorial objects:
I/J
1
2
3
4
5
6
i/k
1
2
1
3
1
2
j/l
1
2
2
3
3
3
With the above table we can set:

- I-th component of stress vector :
I = i(I)j(I)
- I-th component of strain vector :
I = i(I)j(I) (shear terms have to doubled)
- I-th, J-th component of stiffness matrix D :
DIJ = Di(I)j(I)k(J)l(J)
Window 1-8
December 3, 2008

TM34
Window 1-9: Example: 2nd rank tensor in 2D space
Transformation rule
The (u, v) coordinates system is rotated from (x, y) one by angle
txx + tyy txx tyy
+
cos 2 + txy sin 2
2
2
txx + tyy txx tyy
cos 2 txy sin 2

= txx sin2 + tyy cos2 2txy sin cos =
2
2
txx tyy
= (txx tyy ) sin cos + txy cos2 sin2 =
sin 2 txy cos 2
2
tuu = txx cos2 + tyy sin2 + 2txy sin cos =

tvv
tuv
Principal coordinate system ( )
t12 = t21
1
= 0 = arctan
2
def
2txy
txx tyy
Diagonal, principal (max/min) components
t1,2 = tmax,min
txx + tyy
=
s
txx tyy
2
2
+ t2xy
Maximal outof diagonal component

Position of the coordinate system:
t12
1
max = arctan
2
tyy txx
2txy
e.g. such system is rotated from principal one by 45

Maximal value of out-of diagonal component
tmax
uv =
t1 t2
2
Window 1-9
December 3, 2008

TM35
December 3, 2008

TM36
N Preface
Chapter 2
PROBLEM STATEMENT
In following sections formulations of problems available to be solved are given. In particular:
SINGLE PHASE
TWO PHASE
TRANSIENT FLOW
HEAT TRANSFER
HUMIDITY TRANSFER
They contain governing differential equations and boundary conditions (strong formulation).
Despite this, for each problem variational formulations (weak form) are given which are the
basis for numerical solution.
December 3, 2008

TM37
N Preface N N Problem Statement

2.1
SINGLE PHASE, SOLID MEDIUM
Window 2-1: Strong form of the boundary value problem
Equilibrium of the single phase medium

Equilibrium equation:
ij,j + bi = 0,
in
Boundary conditions (BC):

traction BC: ij nj = ti , on t
displacement BC:
ui = ui , on u
= t u
Straindisplacement relation (analysis of the small strain tensor and the linear relation is
assumed):
1
ij = (ui ,j +uj ,i )
2
Constitutive equation (incremental form e.g. means a stress increment )
ij = Cijkl kl
where Cijkl is the 4th rank constitutive tensor
Window 2-1
Related Topics
THEORY MANUAL: TWO-PHASE MEDIA
THEORY MANUAL: WEAK FORM
December 3, 2008

TM38

2.2
TWO-PHASE PARTIALLY SATURATED MEDIUM
The simulation of a twophase medium is necessary in order to account for time-dependent

behaviour resulting from consolidation and/or transient flow. Actual creep will be discussed
later on.
The boundaryvalueproblem to be solved requires the coupled solution of conservation of
mass and momentum in both the solid and the liquid phases, together with boundary and
initial conditions. The general transient case is considered here.
The Windows in section present:
Two phase medium model Window 2-2
Equilibrium of two phase medium Window 2-3
Strong form of BVP for two phase partially saturated medium Window 2-4
Strong form of BVP for two phase fully saturated medium Window 2-5
Window 2-2: Twophase medium model
air
n(1 S)
n
fluid
(1 n)
solid
nS
n porosity
S saturation ratio
The two-phase medium is in fact an approximation of a three-phase medium where it is

assumed that the air bubbles are trapped in the liquid phase so that the mixture (fluid + air)
forms a compressible fluid obeying Darcys law.
Window 2-2
December 3, 2008

TM39

Window 2-3: Equilibrium of twophase medium
bi denote body force, p is the fluid pressure (positive in tension). 0ij is used here for the
effective stresses (positive in tension) see problem statement for:
Partially saturated medium Window 2-4
Fully saturated medium Window 2-5
Equilibrium equation:
ij,j + bi = 0 on T
Straindisplacement relation (analysis of the small strain tensor and the linear relation is
assumed):
1
ij = (ui ,j +uj ,i )
2
Constitutive equation (incremental form e.g. means a stress increment )
ij = Cijkl kl
where Cijkl is the 4th rank constitutive tensor
Window 2-3
December 3, 2008

TM40

Window 2-4: Strong form of BVP for 2phase partially saturated medium problem
Effective stresses 0ij :
0ij = ij Sp ij ,
where S is the saturation coefficient1
S = S(p) =
S +
r "
1 Sr

2 #1/2 if p > 0
p
1+
F
if p 0
F stands for proper weight of fluid, Sr is the residual saturation ratio, a parameter,
Flow equation (the Darcy law Window 3-4)

p
qi = kij + z ,j
F
The permeability tensor kij is obtained by scaling the kij tensor for fully saturated medium
by scalar valued function kr dependent on the saturation ratio.
kij = kr (S)kij
kr
(S Sr )3
=
(1 Sr )3
1
= "

2 #3/2
p
1+
F
Mass balance:
= nSw + (1 n)s
where n is the porosity defined by the void ratio e
n=
e
,
1+e
e=
void volume
solid volume
Van Genuchten (1980), A closed form of the equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sciences Am. Soc., 44, 892898.
December 3, 2008

TM41

Continuity equation:
S kk + qk,k = cp
with c the specific storage coefficient

c = c(p) = n
S
dS
+
KF
dp
KF fluid bulk modulus

Boundary conditions:
on solid phase
ij nj = ti on t
ui = ui on u ,
= u + t
on fluid phase
qj nj = q on q
qj nj = qs on s
p = p on p
= p + q + s
The seepage surface requires a special treatment as the applicable boundary condition is
unknown a priori. Boundary condition on s :
p = 0 on s if S = 1
p = pExt on s if S = 1 and pExt imposed
q = 0 on s if S < 1
This boundary condition is satisfied through penalization imposing
q = 0 if p 0 in the domain and pExt = 0
q = kv p if p < 0 in the domain and pExt = 0
q = kv (p pExt )
p in the domain and pExt 6= 0
with kv , a fictitious permeability (penalty parameter) and pExt the pressure on the external
face of s ; pExt = 0 corresponds to atmospheric pressure.
Boundary conditions for typical flow problem (damping)

Initial conditions
ui (t = t0 ) = ui0 on
p(t = t0 ) = p0 on
Window 2-4
December 3, 2008

TM42

Window 2-5: Strong form of BVP for 2phase fully saturated medium problem
It can be treated as the particular case of partially saturated medium for which Sr 1 e.g.
S 1.
Effective stresses 0ij :
0ij = ij p ij
Continuity equation:
kk + qk,k
n
p = 0 in T
KF
KF fluid bulk modulus

Flow equation (the Darcy law Window 3-4)

p
qi = kij ,j = kij = + z
F
,j
kij is the permeability tensor and F stands for proper weight of fluid.
on solid phase
ij nj = ti on t
ui = ui on u ,
= u + t
on fluid phase
qj nj = q on q
qj nj = qs on s
= +q + s
Initial conditions
ui (t = t0 ) = ui0 on
p(t = t0 ) = p0 on
Window 2-5
Related Topics
THEORY MANUAL: TRANSIENT FLOW
THEORY MANUAL: SINGLE PHASE, SOLID MEDIUM
THEORY MANUAL: TWOPHASE MEDIA. APPROXIMATION AND MATRIX FORM
THEORY MANUAL: TWOPHASE MEDIA. WEAK FORM
December 3, 2008

TM43

2.3
TRANSIENT FLOW
Transient flow problem formulation may be derived from twophase media formulation, see
section 2.2. The only primary state variable is pore pressures p. In continuity equation, term
resulting from skeleton volume changes kk should be neglected. Constitutive relation for the
flow (generalized Darcy law will take identical form). Also boundary contitions for fluid phase
are identical as in the case of twophase media.
December 3, 2008

TM44

2.4
HEAT TRANSFER
Following section gives steady state/transient heat transfer formulation for the isotropic case
of 2/3D continuum, including differential equation, boundary and initial conditions.
Window 2-6: Heat transfer formulation in strong form
Fourier equation:
( T,i ),i +
T
H
= c
t
t
on
F
Temperature BC , with prescribed temperature T :

T = T on T
Heat flow BC , with prescribed heat flux q:
T
= q
on q
n
Convective BC , with prescribed ambient temperature Te :
T
= h(T Te ) on c
n
Note: (q c ) T = ; but (q c ) T =
Initial condition:
Known temperature field T0 at time t = 0:
T (x,0) = T0 (x)
on
where :
T
t
c = c
c
q
Te
h
H
temperature,
time,
heat conductivity,
heat capacity,
specific heat,
mass density,
external heat flux,
ambient temperature,
heat convection coefficient,
heat source
[ K]
[day]
[kN/( K day)]
[kN/(m2 K)]
[kN m/(kg K)]
[kg/m3 ]
[kN/(m day)]
[ K]
[kN/(m K day)]
[kN/m2 ]
Window 2-6
Note: (q c )T = ; but (q c )T = which means that setting temperature B.C.

exclude other boundary conditions at given part T , while heat flow B.C. and convective
B.C. may coexist on a common part of the boundary i.e. mixed condition may be set on
q c as: T
+ h(T Te ) + q = 0. Leaving a part of the boundary with no B.C. specified
n
explicitly, corresponds to setting no heat flow, i.e. adiabatic condition T
= 0 on it.
n
Steady state problem described by Laplace equation
( T,i ),i = 0
December 3, 2008

TM45

can be formulated. Physically, it describes continuum at thermal equilibrium state, while
mathematically it corresponds to a limit of the transient problem at time t with all
state variables independent from time.
The source term H adopted in the formulation is related to the phenomena of heat emission
during concrete hydration process and is described by the following set of equations:
Window 2-7: Concrete hydration heat source
heat source as a function of maturity M :
H(t, T ) = H
aM
1 + aM
maturity M as a function of absolute temperature T and time t:
Zt
M (t, T ) =
Q
exp
R
1
1
Tf
T

dt
td
where:
H
a
Q/R
Tf
td
total value of concrete hydration heat per unit volume [kJ/m3 ],

heat source parameter
[1/day]
activation energy/universal gas constant
[ K]
reference temperature, normally 20 C = 293 K

[ K]
dormant period
[day]
Window 2-7
December 3, 2008

TM46

2.5
HUMIDITY TRANSFER
Following section gives steady state/transient humidity transfer formulation for the isotropic
case of 2/3D continuum, including differential equation, boundary and initial conditions.
Window 2-8: Humidity transfer formulation in strong form
Ficks equation:
(D(W )W,i ),i =
W
t
on
F
Humidity BC, with prescribed relative humidity

W = W on W
Perfect isolation, with humidity flux q W = 0:
(DW )
q W =
=0
on q
n
Note: q W = ; but q W =
Initial condition:
Known relative humidity field W0 at time t = 0:
W (x,0) = W0 (x)
on
with :
W
moisturre potential, i.e. relative humidity
t
time,
D(W ) diffusion coeffcient
as a function
of W ,
D(W ) = D1 a +
D1
W1
a
1a

4
[]
[day]
[m2 /day]
where:
1W
1W1
1+
diffusion coefficient for a moisture potential of W = 1, [m2 /day]
moisture potential at which D(W ) = 21 D1 (1 + a)
[]
factor, to define diffusion at low relative humidity
[]
Window 2-8
Note: q W = ; but q W = which means that setting humidity B.C. exclude

perfect isolation BC at given part W . Leaving a part of the boundary with no B.C. specified
explicitly (default), corresponds to setting no humidity flux, i.e. perfect isolation condition
(DW )
= 0 on it.
n
Steady state problem described by Laplace equation
(D W,i ),i = 0
can be formulated. Physically, it describes continuum at diffusion equilibrium state, while
mathematically it corresponds to a limit of the transient problem at time t with all
state variables independent from time.
December 3, 2008

TM47
December 3, 2008

TM48
N Preface
Chapter 3
MATERIAL MODELS
ELASTICITY
CREEP
CONSOLIDATION
PLASTICITY
SWELLING
AGING CONCRETE
APPENDICES
December 3, 2008

TM49
N Preface N N Material Models

3.1
ELASTICITY
Hookes law is used as the basis of the model. In the subsequent windows a review of formulae
from general to some particular cases is presented:
Hookes law
Plane Strain
Axisymmetric Analysis
To simplify the writing, different sets of material constant are introduced. The relations
between Lames constant and , the coefficient of compressibility K, Youngs modulus E
and Poissons ratio are given in Table 3.1:
Table 3.1: Elastic constants
,
E,
K,
G=
E
(1 + )(1 2)
3K 2
3
December 3, 2008
E
(3 + 2)
+
2( + )
E
2(1 + }
9K
3K +
3K 2
2(3K + )
K
3 + 2
3
E
3(1 2)
K

TM50

Window 3-1: Hookes law
Generalized constitutive equation
ij = Dijkl kl
(1)
where Dijkl is the modulus tensor with 36 independent components

Orthotropic case 9 independent material constants
21
31
1
11
22
33
E1
E2
E3
12
1
32
= 11
22
33
E1
E2
E3
13
23
1
= 11
22
33
E1
E2
E3
1
1
=
12 ,
13 =
13 ,
12
13
11 =
22
33
12
(2)
23 =
1
23
23
Isotropic case 2 independent material constants

Isotropy conditions:
= 12 = 21 = 31 = 13 = 23 = 32
E = E1 = E2 = E3
= 12 = 13 = 23
(3)
one obtains:
1
( 11 22 33 )
E
1
=
( 11 + 22 33 )
E
1
=
( 11 22 + 33 )
E
1
=
ij ,
i 6= j
11 =
22
33
ij
(4)
Isotropic case: tensor notation

Strain vs stress
ij =
1
[(1 + ) ij kk ij ]
E
(5)
Stress vs strain
ij = kk ij + 2ij
(6)
Isotropic case: volumetricdeviatoric split
kk = 3Kkk
sij = 2eij
(7)
(8)
sij , eij denote the components of stress and strain deviators respectively
Window 3-1
December 3, 2008

TM51

Window 3-2: Plane Strain
Plane Strain assumptions:
33 = 13 = 23 0
Under the above assumptions from Eqs (4) (Win.(3-1)), one obtains:
33 = ( 11 + 22 )
and
11
E(1
22 =
(1 + )(1 2)
12
1
0
1
1
0
(1)
11
0
22
1 2 12
2(1 )
(2)
or,
11
1
0
11
22 = + 2 0 22
12
0
0
12
Window 3-2
December 3, 2008

TM52

Window 3-3: Axisymmetric Analysis
In an axisymmetric analysis, the following notation for coordinates and components of displacements, in the cylindrical coordinates system, are used:
x1 = r
x2 = y
x3 =
the radial coordinate

the axial coordinate
the circumferential coordinates
u1 = ur
u2 = uy
u3 = u
the radial displacement

the axial displacement
the circumferential displacement
Analysis of axisymmetric body is assumed as well as that all state variables are independent
of i.e. they are dependent on r and y only. Hence three dimensional problems can be
reduced to 2dimensional ones.
In the axisymmetric torsionless case it is additionally assumed:
u = 0
which results in
r = y = 0
r = 0 and y = 0
Rewriting the constitutive equations in vector form, one gets:
1
0
11
1
0
22
1
12 E 0
0
0
33
0 1
First inverting the above stressstrain relations, one obtains:

11
+ 2
0
22
+ 2 0

12 = 0
0
0
33
0 + 2
11
22
.
12
33
(1)
11
22
12 .
33
Remark:
Below matrix consists of the equivalent planestrain matrix plus a fourth row and column.
Hence planestrain conditions can be obtained from the axisymmetric case by ignoring the
fourth row and column, and setting 33 = ( 11 + 22 ).
Window 3-3
December 3, 2008

TM53

3.2
CONSOLIDATION
Primary consolidation is discussed in this section.

It results from the coupling of loadinduced Darcy flow with the motion of a quasisaturated
medium.
DARCY LAW
FLUID MOTION
Related Topics
THEORY: TWO PHASE MEDIA
NUMERICAL IMPLEMENTATION: CONSOLIDATION
GEOTECHNICAL ASPECTS: TWO PHASE MEDIA
December 3, 2008

TM54
N Preface N N Material Models N N Consolidation

3.2.1
GENERALIZED DARCY LAW
The generalized Darcy law is summarized in Window 3-4

Window 3-4: Darcy flow
Darcys flow velocity1 :
q=
K
ip = k i
(1)
where:
q = nvF
vF
n
K
k=K
F
K
= F g
F
g
ip = i
F = F g
i
relative fluid velocity [m/s]

average velocity through holes [m/s]
porosity
permeability [m2 ] (function of porosity,
independent of fluid properties)
permeability coefficient [m/s]
fluid mass density [kg/m3 ]

earth acceleration [m/s2 ]
fluid viscosity [N s/m2 ]
pressure gradient [N/m3 ]
specific weight [N/m3 ]
hydraulic gradient [nondim.]
Threedimensional extension (with appropriate sign convention),

q = k : grad = k : grad(pF / F + z)
and in indicial notation

qi = kij
(2)

pF
+ z ,j
F
(3)
Window 3-4
See K. Terzaghi (1943) Theoretical Soil Mechanics, Wiley.
December 3, 2008

TM55
N Preface N N Material Models N N Consolidation

3.2.2
FLUID MOTION
Conservation of mass in the liquid phase is expressed by Eq. (1) in the following Window.
The time variation of apparent specific mass splits into two terms as shown in Eq. (2). The
variation of porosity can in turn be related to the volumetric strain, Eq. (3). The fluid density
change is related to the fluids volumetric strain by Eq. (4). Assuming a slightly compressible
fluid, Eq. (5), replacing then in the mass conservation equation (Eq. (1), using Eq. (1)) and
with Q = 0, Eq. (6) is obtained.
The convective contribution can be neglected for small strain and Darcy flow, (v gradF ) is
small, this leads to equation (7) after division by F .
If fluid compressibility is negligible, then the term in t can be ignored. If consolidation effects
are negligible then the square brackets in (6) and (7) is ignored.
Window 3-5: Stress induced fluid motion in porous medium
Conservation of mass

F
+ div F v F = Q
t
(1)
where:
F = nF
F
q = nv F
Q
n
apparent specific mass

specific mass of fluid phase
velocity
mass source term (zero if no source)
porosity
Further,
n
F
=
F + F n
t
t
t
(2)
with
n
kk
=
t
t
F
F
= F kk
t
t
p
F
= KF kk
t
t
results in

F
(3)
(4)
(5)

kk
n p
+ div (F q) = 0.
t
KF t
Finally
kk
(6)
n
p + qk ,k = 0
KF
(7)
with KF the fluid bulk modulus.

Window 3-5
December 3, 2008

TM56

3.3
PLASTICITY
SKETCH OF PLASTICITY APPROACH

MOHRCOULOMB CRITERION
DRUCKERPRAGER CRITERION
CAP MODEL
CAM-CLAY MODEL
December 3, 2008

TM57
N Preface N N Material Models N N N Plasticity

3.3.1
SKETCH OF THE PLASTICITY APPROACH
Plasticity is a nonlinear constitutive theory and leads to a nonlinear system of equations, which
is solved iteratively for d, the displacement increment, using a tangent (local) stiffness.
Once the displacement increment is known the corresponding strain increment results from
the usual straindisplacement relations. From the strain increment a trial stress can be
deduced which, if it lies outside of the yield criterion, must be returned onto the criterion
using a flow rule. This flow rule defines the direction of the stress return. The amplitude of
the return results from the consistency condition, which requires the new stateofstress to
lie on the yield criterion. The objective, in this section, is to define precisely the new keywords
introduced for plasticity theory. The basic ingredients of the elastoplasticity theory are as
follows:
Strain decomposition
Assume that the total strain increment (or rate) is the sum of an elastic e and a plastic p
contribution:
= e + p
and that the following constitutive equation holds:
= De ( p )
with De the elastic constitutive tensor.
Flow rule
The flow rule defines the direction of the plastic flow by:
p = d r(, q)
where d is a positive scalar which defines the amplitude of the plastic flow and r (in general
a function of the stress state and set of a hardening parameters q) defines the direction in
space. The calculation of d will be described later on.
For associative plasticity the direction of the flow r coincides with that of the normal a to
the yield surface:
F
a=r=
while for nonassociative plasticity, we assume the existence of a plastic potential surface Q
such that:
Q
r=
Hardening law
The most general form of hardening law can be expressed in rate form as follows:
q = d h(, q)
where h (in general a function of the stress state and of a set of hardening parameters q)
is called hardening function.
December 3, 2008

TM58

The amplitude of the plastic flow
The amplitude of plastic flow can be derived from the consistency condition, which expresses
that the stress point remains on the yield surface during plastic flow:
F
F
F
F =
: +
: q = a : De : ( p ) +
: q
q
q
F
= a : De : ( d r) +
: d h = 0.
q
From the above equation, in which the flow rule and the hardening rule have been applied,
the amplitude d can be evaluated:
d =
(a : De : )
F
(a : De : r)
:h
q
Derivation of the elastoplastic tangent matrix

Applying the amplitude d to the general constitutive equation the tangent elastoplastic
constitutive matrix is defined in the following way:
= De : ( p ) = De : ( d r) = De :
r
e
=
D
(a : De : )
F
:h
(a : De : r)
q
De : r : a : De
= Dep
F
(a : De : r)
:h
q
In case of perfect elastoplasticity (no hardening is introduced) the definition of the tangent
matrix reduces to:
De : r : a : De
Dep = De
.
(a : De : r)
For a nonassociative plastic flow rule the resulting tangent matrix is nonsymmetric.
December 3, 2008

TM59

3.3.2
MOHRCOULOMB CRITERION
The MohrCoulomb (MC) criterion is more common in soil mechanics. Traditionally, a soil
is described by its cohesion C and its angle of friction . The MC criterion then states
that the shear stress required for yielding depends on the cohesion, the friction angle and the
pressure normal to the slip surface.
Window 3-6: Mohr-Coulomb yield criterion
Mohr-Coulomb yield criterion
| |= c + n tan
where
=
(1)
11 + 22
2
(2)
and,
RMC = c cos + sin
(3)
is the maximum shear stress equal to radius of the Mohr circle at failure, i.e.:
s
( 11 22 )2
+ 212
RMC =
4
(4)
Window 3-6
December 3, 2008

TM60

3.3.3
DRUCKER-PRAGER CRITERION
The DruckerPrager (DP) yield criterion is, mathematically speaking, the most convenient
choice and often numerically the most efficient. The DP criterion is defined in stress space,
by the following equation:
p
F () =a I1 + J2 k = 0
where the invariants I1 and J2 are defined in Section 1.2 and a and k are positive material
properties. For a = 0, HuberMises criterion results.
Flow rule
Associative plasticity: the direction of the flow r coincides with that of the normal a to the
yield surface:
1
F
= a ij + sij
a=r=
2 J2
Nonassociative plasticity: the existence of a plastic potential surface Q of DruckerPrager
type is assumed:
1
Q
= a ij + sij
rij =
ij
2 J2
Notice that the corresponding flow is associative in the deviatoric component and non
associative in the volumetric components, as a = a .
Cut-off condition
The following tensile cuttoff plasticity condition can be activated in conjunction with the
DruckerPrager plasticity criterion:
p
1 0
1
F () = I1 + J2 I1T
= 0.
3
3
It has two basic features, first that maximum first stress invariant I1 is limited to the value
0
for zero deviatoric stress s and the second that the maximum stress ratio defined as:
I1T
3 3J2
q
=
3
p
I1
is limited to the value which can be reached in the uniaxial compression test.
The flow rule has been assumed as fully associated so the plastic flow vector r is:
F
1
1
rij =
= ij + sij .
ij
2 J2
3
Matching of DruckerPrager criterion
DruckerPrager constants could be derived directly from experiments, instead of calculated
from cohesion and friction angle. Assuming that the material is properly identified by a
MohrCoulomb criterion, the matching with the DruckerPrager criterion can be done for
various stress states.
Three dimensional matching
Collapse load matching
Elastic domains matching
December 3, 2008

TM61

Window 3-7: DP criterion: Three dimensional matching
Both criteria are represented in the deviatoric plane along with threedimensional matching
coefficients.
Deviatoric sections of MohrCoulomb (MC) and DruckerPrager (DP) criteria
External apices of the MC criterion yields (axial compression):

a =
2 sin
3(3 sin )
k=
6c cos
3(3 sin )
(1)
2 sin
3(3 + sin )
k=
6c cos
3(3 + sin )
(2)
Internal apices:
a =
Window 3-7
December 3, 2008

TM62

Window 3-8: Matching of collapse load (plane strain conditions)
Matching collapse load of DP and M-C criteria under plane strain conditions is the default
adjustment adopted in the program when plane strain is activated.
Assumptions:
e p e = 0; = p
p33 = p13 = p23 = 0

1
p
ij = d rij = d a ij + sij ,
2 J2
Q
rij =
ij
perfect plasticity:
plane strain:
flow rule :
(1)
(2)
(3)
From (3)
s33 = 2a
p
J2 ;
s13 = s23 = 0
and invariants
[( 11 22 ) /2]2 + 212
J2 =
(1 3a2 )

p
3
I1 = ( 11 + 22 ) 3a J2 ;
2
From DP criterion (F () =a I1 +
(RMC )2
(1 3a2 )
J2 k = 0):
3
RMC (1 3a a )
a ( 11 + 22 ) + q
k =0
2
(1 3a2 )
(4)
q

(1 3a2 ) 3a ( + )
11
22
+k
=
(1 3a a )
2
(5)
one obtains
RMC
Identification with MohrCoulomb criterion, Eq. 4

q
q
(1 3a2 )
(1 3a2 )
sin = 3a
,
c cos = k
(1 3a a )
(1 3a a )
(6)
Associated flow a = a
a = p
tan
2
9 + 12 tan
3c
k=p
9 + 12 tan2
(7)
k = c cos
(8)
Deviatoric flow a = 0
a =
sin
,
3
a specified:
a =
q
1
sin
a sin + 1 3a2
,
3
q

1
k = c cos a sin + 1 3a2
(9)
Window 3-8
December 3, 2008

TM63

Window 3-9: Matching of elastic domain
Plane strain conditions are assumed.
DP criterion (square form of F () =a I1 +
J2 k = 0):
a2 I12 2a I1 k + k 2 = J2
(1)
and invariants:
I1 = ( 11 + 22 )(1 + )

1
J2 =
( 11 22 )2 (1 + 2 ) + 11 22 (1 2)2 + 212
3
(DP) :
(MC) :
11 22
2
2
+ 212 = k 2 2a k(1 + )( 11 + 22 )

1
2
2
2
2
+( 11 + 22 ) a (1 + ) (1 2)
12

2
11 22
11
22
sin c cos
+ 212 = c2 cos2 2
2
2

2
11 + 22
+
sin2
2
Matching the constant, linear and quadratic terms ( 11 + 22 ) yields:

sin
2(1 + )
k = c cos
= 0.5
a =
i.e. stressstateindependent matching is possible for arbitrary c and only when = 0.5.
Alternatively, when c = 0, matching is possible for arbitrary for a specified .
Window 3-9
December 3, 2008

TM64

3.3.4
CAP MODEL
While the constitutive models described previously could be applied to any kind of material,
the following one is more specific to soils.
Model is describe in the subsequent windows:
Yield surface
It combines the DruckerPrager criterion with an ellipsoidal cap closure analogous to the
CAMCLAY ellipse and the tensile cuttoff defined in Section 3.3.3 (if needed). Multisurface plasticity algorithms require the cap definition to be extended to the zone which is
covered by the DP criterion (that is for p < pcs ) where it takes the form of a cylinder. pc
denotes the preconsolidation pressure defines the current cap size.
Flow rule
Associative flow is assumed on the cap; the corresponding flow vector is derived in Win.(310).
Hardening law
The hardening law defines the evolution of the size of the cap yield surface. This requires
the evolution law for pc as a function of plastic strain. The corresponding derivation
is given in Window 3-11, where Eqs (2) and (3) define respectively the total and the
elastic contributions in (4); Eq.( 5) results which relates the hardening parameter pc to the
volumetric plastic strain
Remark:
Underlined variables are positive in compression.
December 3, 2008

TM65

Window 3-10: Cap model: Yield surface and plastic flow vectors
Yield function criterion

DruckerPrager criterion:
FDP = a I1 +
p
J2 k = 0
Cap:
M2
(p pc )(p + pc 2pcs ) = 0 if p pcs
(R 1)2
M2
(p pc )(pc pcs ) = 0 if p < pcs
= q2 +
(R 1)2 cs
FC1 = q 2 +
FC2
Tensile cutoff:
p
1
1 0
=0
FCT = I1 + J2 I1T
3
3
with
q =
3J2 ,
I1
p = ,
3
M = 3 3a ,
December 3, 2008
J2 =
pcs =
pT
1
sij sij
2
pc + (1 R)pT
k
=
3a

TM66

Flow vectors
Defined in reduced stress space
= {q, p} are as follows:
(DP): generally nonassociated flow defined by a

1/ 3
rDP =
3a
CAP: associated flow rule is used for both segments of CAP surface
2q

2
if p p
rC1 =
M
cs
(2p 2pcs )
R1

2q
rC2 =
if p < pcs
0
CutOff: associated flow rule is used

rCT =

1/ 3
3/ 3
Window 3-10
December 3, 2008

TM67

Window 3-11: Cap hardening
Recall:
relation between volumetric strain rate and void ratio rate
dkk =
de
1 + e0
(1)
void ratio evolution : derived from the logarithmic (e ln p) approximation of the virgin
consolidation path
de = dpc
(2)
pc
void ratio variation for elastic unloading path
des = (1 + e0 )dkk = (1 + e0 )
dpc
(3)
Evolution law for pc

elasto-plastic volumetric strain increment
tot
ekk + pkk
kk =
(4)
From (13) the hardening law for pc is derived

dpkk =
1
1
1 + e0 pc K
!
dpc
Evolution law for cap shape R parameter

R = RIN (RIN R0 )
pc pc0
(5)
a + pc pc0
parameters RIN and a are set automatically by the numerical procedure to preserve approxQ/p
q
M
imately the same dilatancy d =
for stress paths with stress ratio =
. These
Q/q
p
2
depend on initial preconsolidation pressure pc0 and on pT . R0 is given by the user. This
parameter enables proper modelling of the K0 coefficient for normally consolidated state
(R0 > 1).
Window 3-11
December 3, 2008

TM68

Window 3-12: Evaluation of pco , Ro from oedometer test
The initial preconsolidation stress pco and cap shape parameter Ro can be set based on
oedometer test once VM (vertical stress at which transition from secondary to primary
consolidation path occurs) and KoNC (Ko coefficient at state of normal consolidation) values
are given (see Window 3-13).
In the oedometer test the following relation holds (assuming that plastic strains are large
compared to elastic ones):
3
dpV = dpD
2
r
2 p p
where: dpD =
de de , dp = dpii .
3 ij ij V
This equation can be rewritten in the form (using plastic flow rule and effect of hardening):

1
3 1
1
1 2
n dp + np nq dq =
nq nq dq + nq np dp
H p
H
2 H
H
where:
np =
nq =
QC1
(QC1 = FC1 elliptic cap surface)
p
QC1
q
dq/dp = Ko (along Ko path)

H=
Ko =
FC1 pc
np (plastic modulus for constant shape ratio parameter R)
pc pV
3 (1 KoNC )
2 KoNC + 1
Window 3-12
Window 3-13: Procedure of evaluation of pco , Ro from oedometer test

Given material properties: eo , E, , , , DP-size adjustment (ak , a )
VM (vertical stress at the transition point
NC
Ko (Ko at state of normal consolidation)
from secondary to primary consolidation line),
Find: pco , Ro
initialize:
ak
2 KoNC + 1
(i=0)
, M = 3 3a pco =
VM
3a
3
next iteration: i = i + 1
find Ro (see Window 3-14) (for pT = 0)
find modified shape ratio parameter RIN for real value of pT and p(i)
(see Window 3-15)
co
(i+1)
find corrected pco value (see Window 3-16)
iterate until | p(i+1)
p(i)
| > 108
co
co
i = 0: pT =
Window 3-13
December 3, 2008

TM69

Window 3-14: Ro evaluation
M, pco , Ko = q/p (at Ko path) =
Given:
Find:
3 (1 KoNC )
2 KoNC + 1
Ro (using bisection method)
Initialize i = 0:
(i=0)
Ro
= 1.01; R = 103
Step i = i + 1:
(i)
(i1)
Ro = Ro
+ R
(i)
for given: M , Ro , pco , Ko compute mean stress p at the intersection point of elliptic
cap surface FC1 and Ko line
FC1
compute corresponding deviatoric stress: q = Ko p and np , nq ,
pc
3
compute residuum of the governing equation for oedometer test: fKo = np /nq ;
2
if i > 1 then
fKlast
fKo 0 then
o
if
(i+1)
set: Ro
(i)
= (Ro Ro /2) and EXIT
else
= fKo and go to next iteration
save: fKlast
o
end if
end if
Window 3-14
December 3, 2008

TM70

Window 3-15: RIN evaluation
Given:
pco
Find:
modified shape ratio parameter RIN such that dilatancy parameter d = np /nq is
the same along trial stress path M = M/2 (some arbitrary path) both for elliptic cap with
pT = 0 and cap surface with real pT value.
Initialize i = 0:
Compute dilatancy parameter do = np /nq for given: pco , pT = 0, M and shape ratio
parameter Ro
set: RIN = 0.01
set: Rlast = Ro
Step i = i + 1
(i)
(i1)
RIN = RIN
+ R
(i)
for given M , pco , pT , M and shape ratio parameter R = RIN compute compute mean
stress p at the intersection point of elliptic cap surface FC1 and stress path line q/p = M
compute corresponding deviatoric stress: q = Ko p and np , nq
compute dilatancy parameter d = np /nq
if
i > 1 then
if
do > dlast
(i+1)
set: RIN
AND
(i)
= RIN
do < d OR do > d AND

dlast do
RIN and EXIT
+ last
d d
do < dlast
then
else
set: dlast = d and go to next iteration
end if
end if
Window 3-15
Window 3-16: Evaluation of corrected p(i+1)
value
co
Given: M , RIN , pT , , VM
(i+1)
Find: pco
Set:
Koel
3(1 Koel )
(1 + 2Koel ) VM
Ko
o
, el =
,p=
, q = K
=
el p
el
1
1 + 2Ko
3
For p, q, pT , M solve quadratic equation (elliptic cap equation) FC1 = 0 for unknown
(i+1)
pco value
Window 3-16
December 3, 2008

TM71

3.3.5
MOHR-COULOMB (M-W)
Yield surface The original MC criterion (see Window 3-6) which leads to a nonsmooth
multisurface plasticity problem, is substituted by its smooth, singlesurface approximation,
being a particular case of a general 3parameter criterion developed recently by Mentrey
(see Menetrey, Willam: A triaxial failure criterion for concrete and its generalization. ACI
Structural Journal 92(3) p.311318). This criterion takes the form described in Window 3-17.
Window 3-17: Menetrey criterion
.
F (, , ) = (Af )2 + mf [Bf rf (, e) + Cf ] Df = 0
(1)
where , , are Haigh-Westergaard stress coordinates equal to:

1
= I1
3
3
3 3
J3 J2 2
cos 3 =
2
p
=
2J2
(2)
(3)
(4)
with I1 , J2 , J3 being the usual stress invariants (????)

Function rf = rf (, e), 0.5 < e 1, describes the shape of the surface in deviatoric section
rf (, e) =
4 (1 e2 ) cos2 + (2e 1)2

1/2
2 (1 e2 ) cos + (2e 1) 4 (1 e2 ) cos2 + (2e 1)2 (1 e2 )
(5)
Window 3-17
The eccentricity parameter e can be calibrated to fit exactly the MohrCoulomb surface
on both extension and compression meridians which leads to the smooth MohrCoulomb
surface. All other parameters of the generalized criterion are also expressed in terms of the
MohrCoulomb friction angle and cohesion c as shown in Window 3-18.
December 3, 2008

TM72

Window 3-18: MohrCoulomb yield criterion by Menetrey criterion
Given:

c
friction angle, 0 < < 90

cohesion c 0
e =
Af =
Bf =
Cf =
mf =
Df =
3 sin
3 + sin
0
3 sin
24c cos
1
tan
3c
1
1
(1)
(2)
(3)
(4)
(5)
(6)
Deviatoric sections of the smooth Mohr-Coulomb yield surface: a) =10 , e=0.89, b)

=50 , e=0.59
Window 3-18
Flow rule
The flow rule which defines the direction of the plastic flow is given in a standard form:
Q
p =
The flow potential adopted here takes a form similar to the yield surface, but with assumption
that r is independent of Lodes angle
Q(, ) = (Aq )2 + mq (Bq rq + Cq )
radius rq is taken as the radius rf for the extension meridian.
December 3, 2008

TM73

Other parameters of flow potential Aq , Bq , Cq , mq are evaluated in a manner analogous to
yield parameters Af , Bf , Cf , mf by formulae of Window 3-18, but using the dilatancy angle
instead of the friction angle .
The recommended form of the flow potential is identical with the DruckerPrager flow potential. It leads to non associated plasticity even in case when = due to the different
forms of rq and rf . Alternative forms proposed here are tabulated below.
Plastic flow parameters

Flow type
Aq
DruckerPrager
0 a a
Bq
Cq
1
2k
3a
k
k and a are computed from , c using tensile meridian adjustment

(putting instead of into formula for k one gets k value)
Axisymmetric HoekBrown
(HoekBrown or
concrete models only)

2 2 tan c fc ft

1
3 tan c
2
c - dilatancy angle for uniaxial compression
1
2
Bq + ft
2
3
fc
= ,
3
r
=
2
fc ,
3
4 < c < 35.3
=
3
Matching the smooth MohrCoulomb criterion with MohrCoulomb criterion

The proposed smooth MohrCoulomb criterion reduces to a vonMis`es criterion through
external vertices when the excentricity factor e tends towards 1, i.e when tends to zero. A
size adjustment must therefore be activated i.o. to achieve the proper stability results. This
consists in replacing the user defined cohesion c and friction angle by adjusted values which
can be derived following the same reasoning as for the DruckerPrager criterion; this is done
in Window 3.3.3-3.
Window 3-19: Matching of collapse load (planestrain conditions)
pij
1
= d rij = d a ij + sij
2 J2

(1)
(plane strain, e <<)

p33 = p13 = p23 = 0
December 3, 2008
(2)

TM74

from (1)
p
s33 = 2a J2 ;
s13 = s23 = 0
p
3
I1 =
( 11 + 22 ) 3a J2 = 3
2

[( 11 22 ) /2]2 + 212
J2 =
(1 3a2 )
J2 =
(3)
(4)
(5)
2
R2
=
(1 3a2 )
2
(6)
replacing in Eq. (1) one gets, after some transformations
6
6
R=
cf cos f
sin f ( 11 + 22 );
A
2A
(7)
index f is added on c and sin , cos to avoid confusion in Eq. (10) below:

2 3 sin f r(e, ) 2 6a sin f
q
A=
.
2 1 3a2
Identifying with the MohrCoulomb criterion of Window 3-17
6
cf cos f
c cos =
A
6
sin f
sin =
A
and alternatively
c
tan
.
=
cf
tan f
(8)
(9)
(10)
(11)
Special cases:
arbitrary flow
r(e, )(3 sin f ) 2 3a sin f

sin f =
sin
q
2 3 1 3a2
(12)
3 3
J3
cos 3 =
a 3/2
2
J2
(13)
results from
which for plane strain failure and e neglected yields:

deviatoric flow yields
a = 0
yields
= 30
r(e, )(3 sin f )
sin f =
sin
2 3
(14)
(15)
if in addition = 0 , then cf = 0.866c

Window 3-19
December 3, 2008

TM75

3.3.6
HOEKBROWN CRITERION (SMOOTH)
The empirical strength criterion proposed by Hoek and Brown for rock masses is:

2
1 3
1
f ( 1 , 2 ) =
+ hb chb = 0
fc
fc
where 1 , 3 the material parameters chb and hb are measures of cohesive and frictional
strength, and fc designates for uniaxial compressive strength. This yield surface is generated
with the general yield surface presented in Window 3-20 by identification of the adjustment
parameters as presented next.
Window 3-20: Smooth Hoek-Brown criterion
Given fc , ft and e , the uniaxial compressive, tensile strength, and the surface eccentricity.
0.5 < e 1 and the following eccentricity value is recommended 0.5 < e 0.6.
1.5ft
1
1
3 (fc2 ft2 ) e
,
Bf =
,
c=1
Af =
,
Cf =
,
mf =
fc
fc ft
1+e
6 fc
3 fc
It has to be emphasized that for given fc , ft , fb (fb -biaxial compressive strength) one may
compute the eccentricity ratio using the following formula:
e=
fb (fc2 ft2 ) + ft (fb2 fc2 )

2 fb (ft2 fc2 ) ft (fb2 fc2 )
Calibration of the generalized criterion to: 1) HuberMises, 2) parabolic Leon, 3) smooth

HoekBrown
Window 3-20
Remark:: The same flow rules as for the Mohr-Coulomb criterion applies.
December 3, 2008

TM76

3.3.7
CUT-OFF CONDITION AND TREATMENT OF THE APEX
As an additional feature of most criteria a tensile Cutoff condition of the following form may
be attached if required
I1+max I1t
The presence of the Cutoff condition places the problem into the class of multisurface
plasticity problems as the cutoff may be formally treated as additional stress constraint.
Treatment of the apex

The flow potential accompanying the Cutoff surface has an associative form:
Q() = I1 .
If the Cutoff is disregarded the surface possesses an apex located at the stress point
3 A
3 A
3 A
A = 3 , 3 , 0, 3 |, 0, 0
|, 3D
with
A =
Df
mf Cf
If the trial stress state is located inside of the apex cone which means it fulfills condition:
tr < ( tr A )
GBf rq
KCq
it would return to the apex.

In the case of an active Cutoff condition, limiting value I1t should be compatible with the
position of the surface apex, i.e.
t < (1 ) A
with
I1
t = t
3
and
= 102
If the above condition is not met, then the maximum possible cutoff position is set automatically as:
I1t = 3 (1 ) A .
December 3, 2008

TM77

3.3.8
MULTILAMINATE MODEL
In this model the existence of up to M Lmax = 3 weakness planes characterizing anisotropy

of the material behaviour is assumed. On each plane separately, MohrCoulomb plasticity
condition and a tension cutoff condition must be fullfilled.
Window 3-21: Weakness plane plasticity conditions
Yield function and flow potential isolines
F (1i) = + n tan i ci
Q(1i) = + n tan i
F (2i) = + n tan i ci
Q(2i) = + n tan i

T
b F (1i) = tan i , 1

T
b Q(1i) = tan i , 1

T
b F (2i) = tan i , 1

T
b Q(2i) = tan i , 1
(1)
(2)
(3)
(4)
F (3i) = n ft
Q(3i) = n
b F (3i) = {1, 0}
b Q(3i) = {1, 0}T
(5)
(6)
Note:
b2b = 0.
Window 3-21
This leads to a multisurface plasticity problem which require that a set of up to 3*MLmax
plasticity conditions must be simultaneously fullfiled by any stress state in the multilaminate
material, i.e

F
b(i) 0; J; J : {1, . . . , 3M L}
T (i) , T are linear transformation matrices describing transition between ( xx , yy , xy , zz ),
December 3, 2008

TM78

global coordinate system stress components and the i-th weakness plane components { n , }T

(i)
2

x
n
s
c2 2sc
(i)
(i)
(i)
y
b =
=T
,
T =
sc sc c2 s2
xy
where :
s = sin (i)
c = cos (i) .
Window 3-22: Multilaminate model (3D, Generalized Plane Strain analysis type case)
For i = 1, M L (M L < 3)
Weakness plane setting:
ith weakness plane plasticity conditions:
Weakness plane unit normal n: if 6= 0:
Yield function expressed in terms of n , :
sin cos
cos cos
sin sin
n=
,
,
a
a
a
T
F (1i) ( n , ) = k k + tan n c
F (1i) ( n , ) = n ft .
where:
Flow potential expressed in terns of n , :
q
a = cos2 + sin2 sin2 ;
Q(1i) ( n , ) = k k + tan n c
if = 0:
Q(1i) ( n , ) = n ft .
n = [0, cos , sin ]T .

T
Expressions of F (ki) , q (ki) by full stress vector = xx , yy , xy , zz , xz , yz
normal stress:

T
n = nT n = vT , where v = n2x , n2y , 2nx ny , n2z , 2nx nz , 2ny nz
,
tangent stress vector:
=
x, y , z
T
= n n n = A
where
nz 1 n2x
A = ny n2x
nz n2x

nx n2y ny 1 2n2x
nx n2z
nz 1 2n2x
2nx ny nz
ny 1 n2y nz 1 2n2y
ny n2z
2nx n2y
nz 1 2n2y
nz n2y
2nx ny nz nz 1 n2z nx (1 2ny nz ) ny 1 2n2z
December 3, 2008

TM79

thus:
p
F (1i) () =
T (AT A) + tan vT c
F (2i) () = vT ft
p
Q(1i) () =
T (AT A) + tan vT c
Q(2i) () = vT ft .
Gradients required by multisurface plasticity closest point projection algorithm:

1
AT A + tan v
k k
= v

1
=
AT A + tan v
k k
= v
"

T #
1
=
AT A A T
AT
.
k k
k k
k k
F (1i) =
F (2i)
Q(1i)
Q(2i)
Q(2i)
Window 3-22
This leads to a multisurface plasticity problem which require that a set of up to 3M L M ax
plasticity conditions must be simultaneously fulfilled by any stress state in the multilaminate
material, i.e.:

F () () 0; J; J : 1, . . . , 3M L .
Plastic strains emerge due to violation of any of those conditions by the trial elastic stress.
The total plastic strain is the sum of each planes contribution.
p = b Q
X
p =
p
NB : A perfectly elastoplastic behaviour (no hardening) is assumed.

The flow rule is governed by a flow potential Q, the form of which is analogous to the form
of corresponding yield function F . As the dilatancy angle i specified for each weakness
plane may in general differ from the corresponding friction angle, the flow rule adopted is
nonassociative (Q 6= F ) .
The model requires the data of M L, the number of assumed weakness planes (M L 3).
For each i-th plane , (i = 1, .., 3M L) the following data should be specified:
inclination angle of ith weakness plane
(90 degrees, positive counterclockwise)
friction angle
dilatancy angle
cohesion.
(i)
(i)
(i)
c(i)
December 3, 2008

TM80

The following constitutive model is derived.
Window 3-23: Physical origin of flow potential Q
During plastic slip ( = tan n + c):
Frictional slide model to explain physical origin of flow potential

dilatancy angle, 6=
if = associativity is preserved.
Window 3-23
Window 3-24: Constitutive equations of multilaminate model
= D : ( p )
X
b Q
p =
(1)
(2)
with yield and loading/unloading conditions for {1, 2, ...3M L}

F ()
F ()
F ()
=
=
0
0
0
0.
(3)
(4)
(5)
(6)
In expanded form
if F () 0 or F () = 0 and F () < 0 = 0
if F () = 0 and F () = 0 = 0
( constraint is active).
Window 3-24
December 3, 2008

TM81

3.3.9
MODIFIED CAM CLAY MODEL
Window 3-25: Yield and flow potential surface

The elliptical yield surface (and flow potential as well) is described by the equation (see Fig.
below)
F (,pc ) = q 2 + Mc2 r2 () p (p pc ) = 0
Modified-Cam Clay yield surface
The Mc parameter is the slope of the critical state line along compression meridian, pc is a
preconsolidation pressure adjusted along p axis and r() is a function of Lode parameter
describing shape of the yield surface in the deviatoric plane. The r() function is taken after
van Ekelen.

r () =
1 sin(3)
1
n
3 3 J3
sin(3) =
2 J 23
n = 0.229
0.7925
The relation between k = ME /MC (ME is the slope of critical state line for the tension
meridian) and parameter is as follows
1
kn 1
1
kn + 1
and the default setting for the parameter k can be set as

k=
3
3 + Mc
Remark:
The applied r() function is applicable for friction angles up to 46.55o .
Window 3-25
December 3, 2008

TM82

Window 3-26: Nonlinear elastic behavior within yield surface
The reversible part of the deformation is governed via nonlinear elasticity assuming that
K = K(p), G/K =const. and the well-known formula for bulk modulus is applied
K=
1 + eo
p
where eo is an initial void ratio, is the slope of secondary compression line in e ln(p) axes.
Remarks
Shear modulus G is a linear function of p
Poissons coefficient is a constant
Mean stress may reach a zero value for infinitely large tensile volumetric strains
Any calculation carried out with Cam Clay model requires explicit setting of the initial
stresses
Window 3-26
Window 3-27: Hardening/softening law
The evolution of the hardening parameter is defined through the following equation
dpc =
1 + eo
pc (dpkk )
where is a slope of a primary compression line in eln(p) system (see Fig. given below).
Isotropic compression test
Window 3-27
December 3, 2008

TM83

Window 3-28: Modeling normally consolidated soils with modified C-C model
The Modified Cam-Clay model can describe basic macroscopic phenomena observed for normally consolidated cohesive soils. The model behavior is well represented if we consider a
standard triaxial compression test. In that case (OCR = 1), as shown in Fig. below, the
yield surface follows the current stress state.
350
300
CSL
OCR=1
250
200
Stress path
MC
150
Limit state
100
Intermediate state
50
Initial state
0
0
50
100
150
200
250
300
350
Drained triaxial compression test (OCR = 1)
During the application of an axial strain to the specimen the compressive volumetric strain
is produced while deviatoric stress grows monotonically up to the value q = Mc p (see Fig.
below).
160
140
120
100
OCR=1
80
60
40
20
0
0
0.02
0.04
0.06
0.08
e ps -de v
0
0.02
0.04
0.06
0.08
e ps -vo l
-0.001
-0.002
-0.003
-0.004
-0.005
OCR=1
densification
-0.006
e ps -de v
Stress/strain characteristics (OCR = 1)
Window 3-28
December 3, 2008

TM84

Window 3-29: Modeling overconsolidated soils with modified Cam-Clay model
The Modified Cam-Clay model describes also some macroscopic phenomena observed for
overconsolidated cohesive soils. A typical model behavior, analyzed for a drained triaxial
compression test (for OCR = 5) is shown in figures given below. In that case the effective
stress path passes the critical state line until the current yield surface is met and then goes
down until critical state is achieved. This effect corresponds to strain softening.
600
500
CSL
OCR=5
400
MC
Intermediate state
300
200
Limit state
100
Initial state
0
0
100
200
300
400
500
600
Drained triaxial compression test (OCR = 5)

During the application of the axial strain to the specimen the compressive elastic volumetric
strain is produced first. Once the yield surface is met a dilatant volumetric strain is growing up tending to an asymptote at the critical state. The deviatoric stress does not grow
monotonically exhibiting peak and residual values.
250
Strain softening
200
150
100
OCR=5
50
0
0
0.01
0.02
0.03
0.04
e ps -de v
0
0.01
0.02
0.03
0.04
0.0025
e ps -vo l
0.002
0.0015
dilation
0.001
0.0005
OCR=5
0
-0.0005
e ps -de v
Stress/strain characteristics (OCR = 5)

Window 3-29
December 3, 2008

TM85

3.4
CREEP
Creep is a timedependent deformation under maintained stress. It is assumed that the stress
can be split into volumetric and deviatoric components and the corresponding timedependent
strain components are the volumetric and the deviatoric creep. The following formulation is
adopted for onedimensional creep :
cr = einst f (t) = C (t)
Creep is considered to be proportional to the instantaneous elastic deformation. C(t) is the
creep law corresponding to a unit stress.
The three-dimensional creep law is then:
cr = ED1 C(t) = D1
o C(t)
and C(t) is assumed to be e.g. of the form
C(t) = A(t t0 )m
where A and m are material parameters.
In most situations the same creep law will be adopted for both the volumetric and the
deviatoric components. Different parameters can however be chosen for the volumetric and
deviatoric creep components. Great care has to be taken in that case because this may
generate a Poisson coefficient which varies in time.

CREEP PARAMETER IDENTIFICATION FROM EXPERIMENT
December 3, 2008

TM86
N Preface N N Material Models N N N Creep

3.4.1
Creep under variable stress requires a principle of superposition. The adopted power law does
not lend itself easily to such in principle. It is therefore replaced, in the implementation, by
a series of Kelvin elements. This is described next in Windows 3-31 and 3-30.
The automatic adjustment of the Kelvin element parameters to the prescribed power law is
derived in Window .3-32
Window 3-30: Kelvin element under constant unit stress
Constitutive equation
q = G +
Creep strain increment, due to
q = H(t t0 )
where: H Heaviside unit step function

G
1
1
cr
1 exp (t t0 )
= A 1 exp (t t0 )
=
=
G
B
= C(t t0 ) with = 1
N.B.:
cr = 0 at t = t0
q
cr =
at t =
G
Kelvin mechanical model

Window 3-30
December 3, 2008

TM87

Window 3-31: Creep under variable stress
Let the creep strain under variable stress be
Zt
C(t )
cr
1
d
(t) = D0
[ (t) ( 0 )]
with D0 = E 1 D.
For unit stress and a single Kelvin element

1
C(t, ) = A 1 exp (t )
B

A
t
C
= exp
B
B
For a chain of N Kelvin elements in series, and introducing the volumetric-deviatoric split

N
X
1
v
Cv (t, ) =
Ai 1 exp v (t )
Bi
i=1

N
X
1
d
Cd (t, ) =
Ai 1 exp d (t )
Bi
i=1
Zt

Cv
cr
v 1
m (t) = D0v
[ m (t) m ( 0 )]
d
Zt

Cd
cr
d 1
d
e (t) = D0
[s (t) s ( 0 )]
The creep strain increment, required by the general nonlinear incremental scheme, as described in section FULL/MODIFIED NEWTON-RAPHSON ALGORITHM (Section 4.6.1)
can be derived by recurrence
cr
cr
n+1
cr
n
D1
0
N
X
e
cr
i
i=1
and
(i )cr
n+1
(e
i )cr
n+1

tn+1
tn+1
+ exp
1 (e
i )cr
= [ n + n+1 ( 0 )] Ai 1 exp
n
Bi
Bi
cr
= (e
i )cr
n + i
In the case of nonlinear creep the above recurrence formula is modifed by an additional term
scaling the amplitude of the creep strain Ai in the following manner
(i )cr
n+1 = [ n + n+1 ( 0 )] (C1 n + C1 n+1 )

tn+1
tn+1
Ai 1 exp
+ exp
1 (e
i )cr
n
Bi
Bi
December 3, 2008

TM88

The nonlinear term C1 () is expressed by the equation
C1 () =1+a SLb
where a (0 a 10) and b (1 b 10) are material parameters and SL is a stress level
expressing the relative distance of the stress state from the yield surface.
Eqs (????) also apply for volumetric and deviatoric creep components when , , D0 , Ai ,
Bi are specialized appropriately.
(D0v )1 =
and Dd0
1
E
= 1 2
3K
is a (4 4) matrix with the following diagonal

1

diag Dd0
= (1 + ) , (1 + ) , 2 (1 + ) , (1 + )
which is valid for axisymmetry and for plane strain.

Window 3-31
Window 3-32: Approximation of the creep law
Creep function identification

Parameter identification procedure
Initial approximation
1. Select
B1 , B2 , . . . , Bj
104 , 103 , . . .
N.B.: Six components [104 , . . . , 10] seem to be appropriate, less may be sufficient

2. Define t1 , t2 , . . . tj such that

tj
Aj = Aj 1 exp
Bj
Select : 0.5 e.g., then
tj = Bj ln(0.5)
December 3, 2008

TM89

3. Define such that for j = 1 to J
C(tj ) =
j1
X

Ak + Aj
k=1
hence

tj
Aj = 1 exp
Bj

tj
1 exp
Bj
1 "
C (tj )
j1
X
#
Ak
k=1
2nd approximation
A(2)
tj
= 1 exp
Bj
(1)
1 "
C (tj )
#
Fk (tj )
k=1
Fk (tj ) = Ak
j1
X

1 exp
tj
Bj

.
Iterative scheme
(0)
Aj
#

1 "
j1
X
tj
(0)
= 1 exp
C (tj )
Ak
Bj
k=1

i-th approximation
(i)
Aj
#

1 "
X
tj
= 1 exp
C (tj )
Fk (tj )
Bj
k6=j
Fk (tj ) =
(i1)
Ak
tj
1 exp
Bk

.
tj = Bj ln (0.5)
Bj = 10j4
[days] 1 < j < 6.
Convergence test
"
(i1)
(i)
max Aj
Aj
(i)
Aj
#
< 102
Window 3-32
December 3, 2008

TM90

3.4.2
CREEP PARAMETER IDENTIFICATION FROM EXPERIMENTS
The easiest way to adjust creep parameters A and m for C(t, t0 ) = A(t t0 )m from an
experimental curve f (t, t0 ) is to pick two points situated at (t t0 ) = 1 for the adjustment
of A and then at a large value of (t t0 ) for the adjustment of m.
Depending on the available test the procedure differs only slightly. Four typical experiments
are analyzed next; additional situations can easily be extrapoled using D1
0 with
1
0
1
0

.
D1
0 =
0
0 2(1 + ) 0

0
1
If volumetric and deviatoric creep are different i.e.
Cv (t, t0 ) = Av (t t0 )mv 6= Cd (t, t0 ) = Ad (t t0 )md
then
(D0v )1 =
and
E
= 1 2
3K
Dd0
1
1+
0
0
0
0
1+
0
0
.
=
0
0
2 (1 + )
0
0
0
0
1+
Window 3-33: Uniaxial test
c = D1
0 C (t t0 )
c1
1
0

c2
1
0
=
0
0
0
2(1
+
)
0
c
3

0
1
1
0
0
0
C(t t0 )
hence with
C(t t0 ) = A(t t0 )m
c1 = 1 A(t t0 )m .
Window 3-33
December 3, 2008

TM91

Window 3-34: Triaxial test
Assuming cv is measured in the experiment:
c
1
0

1
1
1
0
2
2
=
c
0
0
2(1
+
)
0
0
12
c

0
1
3
3
C(t t0 )
hence, with
C(t t0 ) = A(t t0 )m
cv = 3 (1 2) m A(t t0 )m .
Window 3-34
Window 3-35: Triaxial deviatoric test
Assuming c2 c2 = c is measured for a unit q = ( 1 2 )
2
c
1
d = D 0
and 3 = 1
0
3
then
c2 = ( 2 2 1 ) C (t, t0 )
c1 = [ 1 ( 2 + 1 )] C (t, t0 )
c = (1 + ) qC (t t0 ) .
Window 3-35
Window 3-36: Oedometer test
K0 2
2
c = D01
0
K0 2
C (t, t0 )
then
c2 = (1 2K0 ) 2 C (t, t0 ) .
Window 3-36
December 3, 2008

TM92

3.5
SWELLING
A nonlinear creep approach has been adopted here to model swelling phenomena. Creep seems
to be the simplest phenomenological approach which can reproduce macroscopic behaviour of
swelling soils/rocks both quantitatively and qualitatively without considering all the microscale
effects.
Oedometric swelling test
Memorizing of in situ stress 0
Correction of os with respect to o state
Three-dimensional generalization
December 3, 2008

TM93

Window 3-37: Oedometric swelling test
An oedometric swelling test is taken as the basis for further three-dimensional generalization
according to the suggestion of Wittke and Kiehl 2 .
A typical relation between swelling strain and vertical stress (after Huder and Amberg) obtained from oedometer test is shown in figure below.
Five basic parameters are needed to calibrate swelling: os , cs , , Bo , and s .

Parameter os defines a minimum vertical stress value which stops swelling evolution while
cs bounds the excessive increase of the swelling for small or even tensile stress state.
Parameter defines the slope of the line sy y for stress range cs y os in semilogarithmic system. A parameter Bo , the time to reach the steady state for a sudden total
unloading, is introduced, which defines maximum swelling evolution rate while s reduces or
increases swelling evolution rate according to the formula:
Bo
1 exp(s )
k 0 k
= MIN (1, REF
)
k os REF
cs k
B() =
where the current effective stress is denoted by , in situ stress is denoted by 0 and
REF
appropriate reference stress states REF
os , cs are defined in the following windows.
Window 3-37
W.Wittke.
Stability Analysis of Tunnels.Fundamentals.Verlag Glueckauf Essen, 2000.
J.R.Kiehl.Interpretation der Ergebnisse von Grossquellversuchen in situ durch dreidimensionale numerische
Berechnungen, Proc. 7th.ISRM Congress, Aachen 1991,pp.1534-1538)
December 3, 2008

TM94

Window 3-38: Memorizing of in situ stress 0
The reference stress 0 is required for evaluation of swelling rate function B() and has to
be memorized during computation. As the computation scenario can be very complex the
following recipe is used to set up 0 reference state:
If the initial state analysis is run then for all active (at time t = 0) materials 0 is taken as
the result of the last step of the initial state.
If the initial state is not specified in list of drivers but swelling is activated for some materials
then 0 is taken as the stress state at first converged step since material activation. NB. In
such case swelling strain increment will not be generated in the first time step (since material
activation).
Window 3-38
Window 3-39: Correction of os parameter with respect to 0 state
If Ko axes do not coincide with global x-y-z system axes
transform 0 =TGL 0
else
0 = 0
IF 0y < os modify parameter os : os = 0y .
Window 3-39
Window 3-40: Three-dimensional generalization
We assume that the generalized reference stress state REF
and REF
os
cs , in xyz system, can
be defined as follows:

T
K
0
K
0
0
REF
=
ox
o
o
oz
o
os,xyz

T
Kox c c 0 Koz c 0 0
REF
cs,xyz =
v
(if not explicitely defined)
1v
It is possible to set up the in situ Ko coefficients Kox , Koz in local coordinate system x0y0z0
rotated with respect to xyz system. In such a case the reference stress states are defined as
where Kox = Koz =
REF
os, x0 y 0 z 0 =
Kox o o 0 Koz o 0 0
REF
cs, x0 y 0 z 0 =
Kox c c 0 Koz c 0 0
and the following transformation takes place to set them up in xyz system:
LG REF
REF
os, x0 y0 z0
os, xyz = T
LG REF
REF
cs, x0 y0 z0
cs, xyz = T
where TLG is a transformation matrix from local to global system.
December 3, 2008

TM95

The increment of creep strain (swelling strain) is computed using following recurrence formula:
CR =
t
CR(*)
B()
The following approach is used to compute CR(*) for given effective stress state :
find eigenvalues i and eigendirections di of

REF(*)
REF(*)
REF
transform REF
os,xyz , cs,xyz to principal directions of os,xyz , cs,xyz
transform accumulated creep strain ac-CR to principal directions of ac-CR(*)
compute predicted swelling strain components in principal directions of using following
formula:
i
REF(*)
REF(*)
ln
if
c
o
i
ii
ii
REF(*)
oii
REF(*)
CR
0
if
o
i
ii
i =
REF(*)
cii
REF(*)
if
i cii
ln REF(*)
oii
ac-CR(*)
compute CR(*) based on CR
i , di and

ac-CR(*)
ac-CR(*)
CR(*) = MAX(CR
,
di dTi
i
ii
ii
Remarks:
An explicit integration scheme has been adopted here and thus the maximum time step value
is limited and should satisfy the condition:
0 < t <
2B MIN
Eoed
where MIN is the principal normal stress which yields the largest swelling strain.
Window 3-40
December 3, 2008

TM96

3.6
AGING CONCRETE
The aging concrete model represent time dependent mechanical properties as well as rheological behavior of concrete in early age.
It consists of a set of parallel Maxwell units, as shown in the Window 3-41. Each unit is
described by a maturity dependent Young modulus Ek (M ) = Wk (M )E, and retardation
Ek
. The contribution of each unit depends on maturity measure M (expressed
time k =
k
P
in time units), by the set of weighting factors Wk , ( k Wk = 1), given for specified time
instants ti .
Window 3-41: Aging concrete model
Maxwell chain model for aging concrete
Window 3-41
Remarks:
For details of the numerical implementation see:
AGING CONCRETE MODEL - IMPLEMENTATION SCHEME
December 3, 2008

TM97

3.7
APPENDICES
SAFETY FACTORS AND STRESS LEVELS
December 3, 2008

TM98
N Preface N N Material Models N N N Appendix

3.7.1
SAFETY FACTORS AND STRESS LEVELS
The following definitions are used in Z Soilr ; of course these definitions are not unique.
Window 3-42: Stress level (SL)
I. for axisymmetric criteria
SL =
q
qfailure
with q =
p
3J2
II. for Lodes angle dependent criteria

SL =
with failure = m + s
Remarks:
1. I is a subset of II
2. obviously will be different for each material point.
Window 3-42
Window 3-43: Safety factor (SF)
I. Let SF1 be the multiplier which can be applied uniformly to each deviatoric stress in the
structure, such that failure occurs when
0 = m + SF1 s
II. For twoparameter criteria (C, ) an alternative definition is available which is
R

failure d

SF2 = R

where is the failure surface and can, as shown elsewhere, be obtained numerically by a
progressive reduction of C and tan using:
c
c0 =
SF2
(tan )0 =
tan
.
SF2
In the general MentreyWillam definition this amounts to replace the parameters of the
criterion by
A0f = SF1 Af
Bf0 = SF1 Bf
Window 3-43
December 3, 2008

TM99
N Preface N N Material Models N N N Appendix
December 3, 2008

TM100
N Preface
Chapter 4
NUMERICAL IMPLEMENTATION
WEAK FORMULATION
ELEMENTS
INCOMPRESSIBLE AND DILATANT MEDIA
FAR FIELD
OVERLAID MESHES
ALGORITHMS
December 3, 2008

TM101
N Preface N N Numerical Implementation

4.1
WEAK FORM AND MATRIX FORMS OF THE PROBLEM
SINGLE PHASE MEDIUM, TIME INDEPENDENT LOADING

TWOPHASE MEDIUM, RHEOLOGICAL BEHAVIOUR
HEAT TRANSFER
In the subsequent sections the following Hughes notations is used:
w(i,j) =
December 3, 2008
1
(wi,j + wj,i )
2

TM102
N Preface N N Numerical Implementation N N N Weak Form

4.1.1
SINGLE PHASE MEDIUM, TIME INDEPENDENT LOADING
In the finite element solution procedure one does not solve the strong form of the boundary
value problem but an equivalent weak (variational or virtual work) form which is usually
discretized following a Galerkin technique. The equivalence of the strong or differential form
of the problem statement and the weak form is discussed in texts concerning finite element
implementation (see [Hughes, 1987]).
Window 4-1: Weak and matrix form of single phase medium
Weak form:
wi fi d +
w(i,j) ij d =
(1)
wi ti d
ui = ui
on u
where t and u are boundaries with prescribed tractions and displacements, respectively,
and
= = u + t .
Approximation functions for w and u
wi = Na wia
ui = Na uia
(2)
Index a identifies the element node number.

Matrix form
The matrix form results from the introduction of the appropriate derivatives of these approximations into the weak form and invoking the arbitrariness of the weighting functions w
Ku = F
uia = uia
(3)
on u
where:
K=
e=1,N
(Ke ) ,
A the assembly of elemental stiffness contributions
and Ke is the elemental stiffness contribution

Z
e
K = BT D B de
(4)
will be established when elements are discussed.

Internal force: term Ku
It can alternatively be written as
Ku=
e=1,N
BT de
(5)
which will be needed later.

Window 4-1
December 3, 2008

TM103

4.1.2
TWO-PHASE MEDIUM, RHEOLOGICAL BEHAVIOUR
Twophase medium. Weak form

Integration in time domain
Choice of shape functions
The strong statement of the twophase boundary value problem was given in Section 2.2.
The corresponding weak statement and matrix form are shown in Window 4-2.
Window 4-2 describes the weak form of the overall equilibrium equation (expressed in terms
of total stresses according to extended Bishops effective stress principle) and weak form
of the fluid flow continuity equation with associated boundary conditions. The problem is
formulated in terms of nodal displacements u and pore pressures pF .
The actual implementation of the seepage boundary conditions is done using a twosided
surface element (see Window 2-4) in which boundary conditions are imposed on the external
face by user.
Window 4-2: Twophase medium. Weak form
Equilibrium equations
Z
w(i,j) tot
ij
Z
d =
(total internal force)
wF kk d
(skeleton def. rate)
F F
w,k
v,k d
(internal flux)
Z
wi bi d +
(body load)
wi ti d
t
(surface traction)
Z
Z

wF c pF pF d + wF q d wF kv (pF pF ext ) d = 0
(storage)
q
(surface flux)
s
(seepage penalty term)
Solid boundary conditions

ui
tot
ij nj
= ui
= ti
on u T
on t T
Fluid boundary conditions

pF = pF
viF ni = q
on
on

p + 2s T

q + 1s T
Initial conditions
ui (t0 ) = ui0
F
p (t0 ) =
pF0
on
on
Window 4-2
December 3, 2008

TM104

The matrix form of the problem results from the specific choice of shape functions for interpolation of displacements and pore pressure fields within the element. The matrix equations
are shown in Window ??. The Fint (u) is a nonlinear internal force term due to plasticity and
thus a linearized problem statement is required. The application of the implicit integration
scheme in time permits an elimination of time derivatives from set of linearized equations
which is finally solved for u the increment of displacement u and pF the increment of pore
pressure pF . Iterations are needed if the constitutive law of the solid skeleton is nonlinear.
Window 4-3: Integration in time domain
The following predictor-corrector scheme has been adopted for integration of the displacements and pore pressures in time.
uiN +1 = uiN + (1 ) t u iN + t u iN +1 = uiN + t u iN + t u iN +1
{z
}
|
| {z }
upredictor
u i
u iN
+1 p
pN +1 = pN + (1 ) t pN + t pN +1 = pN +iNt
+ t Np+1
|
{z N}
| {zN +1}
ppredictor
N +1
p N +1 p N
The integration coefficient has been assumed to satisfy the sufficient condition for stability
1
.
2
Window 4-3
Window 4-4: Choice of shape functions
Let us introduce the following finite element interpolation functions for approximated field(s),
i.e. displacements uh and pore pressure ph , within a finite element
uh = Nu ue
w h = Nu w e
ph = Np pe
q h = Np q e
With these definitions the strain-displacement relation is expressed in the standard form

uh = B ue
Window 4-4
December 3, 2008

TM105

4.1.3
HEAT TRANSFER
Following section describe the solution algorithm for transient and steady state heat transfer
problem formulated in section 2.4. In particular hints on
Weak and matrix form of heat transfer analysis

Time marching algorithm for transient analysis
Heat source term
are given.
Notation
T -temperature field
-time
TE
H
qT
c
-external (ambient) temperature X

-heat source
-heat flux
O
-heat capacity
X
-heat conductivity
Xe
-maturity
-time derivative of X
-time increment
-gradient
-matrix/vector representation of X
element contribution of X
Weak form and matrix formulation

Window 4-5: Weak and matrix form of heat transfer analysis
Approximation for temperature field T (x,t)
T (x,t) = N(x)T
(1)
where N are shape functions and T are nodal temperatures.

Matrix form of governing differential equation for transient heat transfer analysis
CT + (K + Kb )T =H + Kb T
(2)
As Z Soilr system is generally based on incremental approach, equations have to be linearized

which finally leads to following linearized form:
CT + (K + Kb + H )T =Q
(3)
where :
Q = Q+H + Kb TE CT (K + Kb )T
December 3, 2008
(4)

TM106

For each matrix X =
e=1,N
(Xe ), where symbol
e=1,N
represents the assembly of element
contributions Xe , as listed below.

Z
e
K = ONT ON de
e
e
C =c
NT N de
Keb
Z
= hT
Qe =
NT N de
NT qT de
He =
NT He
H M
H
H M
H
H =
+
+
=
M T
T
M T
T
e
NT N de
All those matrices are evaluated using Gauss quadratures.

Window 4-5
The general system can be reduced to steady state conditions which is expressed by the set
of equations:
(K + Kb )T = Q + Kb TE (K + Kb )T
(4.1)
Window 4-6: Time marching algorithm for transient analysis

Applying finite differences in time domain , with (0, 1):

Tn+1 = Tn + (1 )tTn + tT n+1 = Tn + tTn +t T n+1
(1)
predictor
thus:
T n+1 =
Tn+1
t
(2)
Introducing above into the linearized form 3, the following set of incremental equations is
obtained:
[C + t (K + Kb + H )] T =t Q
(3)
Window 4-6
December 3, 2008

TM107

Handling of concrete hydration heat source term.
In the case of heat source term related to concrete hydration (1), which appear in definition
of matrices H,H, integration of both heat source term H and maturity factor derivative
M
with respect to temperature
via Eqn. (2), in each integration point in an element are
T
required. Even with assumption that in given time increment the temperature varies linearly,
the integrals (2) can not be evaluated using elementary functions, so it has to be integrated
numerically as follows:
Window 4-7: Heat source term
heat source:
H = H
aM
1 + aM
(1)
maturity factor:
Zt

exp
M=
Q 1
1
(
)
R Tref
T

dt
(2)
td
numerical integration shema:

Q
1
1
Mn+1 = Mn + exp
) tn+1
Tref T
R

M
M
=
+
T n+1
T n

2

Q
1
1
1
Q

exp
tn+1
R Tn+1
R Tref
Tn+1
=0
Mn+1
M
=0
T n+1
if
tn+1 > tBEG + td
if
tn+1 > tBEG + td
if tn+1 tBEG + td
if tn+1 tBEG + td
where:
Tn+1
= Tn (1 ) + Tn+1
Window 4-7
The first existence time of an element to which the integration point belongs is denoted by
tBEG . In the above scheme in case when tn+1 < tn+1 tBEG td the value tn+1 should
be replaced by appropriate value.
December 3, 2008

TM108

4.2
ELEMENTS
FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS

STRAINS
STIFFNESS
BODY FORCES
INITIAL STRESSES
December 3, 2008

TM109
N Preface N N Numerical Implementation N N N Elements

4.2.1
FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS
A family of 2/3D isoparametric elements with 1st order interpolation function is implemented
in the program, see Appendix 4.7.1.
In case of single phase analysis the nodal displacements (active for given analysis type) are
the basic unknowns. Derivatives of shape function are necessary for strains analysis and then
for stresses calculation. Basic features are summarized in the following table:
Finite elements for 2/3D continuum problems
Analysis type
Ndim
Elementary
volume dV
Active
displacement
DOF / node
Kinematic
constraints
Strain vector
= Bd
Straindisplacement
operator
Ba
B = [Ba ]
a = 1, N EN
Plane Strain
2
Axisymmetry, r = x
2
dx dy 1
2r dr dy
u(x, y) = [u, v]T
u(r, y) = [u, v]T
=0
z

T
xx , yy , xy
=0
T

xx , yy , xy ,
Na
0
r
N
a
0
Na Na
Na
0
r
w = 0,
Na
x
0
Na
y
0
Na
y
Na
x
w = 0,
3D
3
dx dy dz
u(x, y) = [u, v, w]T
3

T
xx , yy , xy , zz , xz , yz
Na
0
0
x
Na
0
0
Na Na
0
y
Na
z
Na
Na
0
z
x
Na Na
0
z
x
December 3, 2008

TM110

4.2.2
Numerical integration is required for the evaluation all integrals resulting from the problem
definition. Gaussian quadrature is used and for a onedimensional integral can be represented
as follows:
Z
X
f (x) dx =
f ( i ) Jx , ( i ) wi
where:
f ( i ) function being integrated, value at i
Jx , jacobian determinant
wi Gaussian weighting factor
The extension of the above for multidimensional case is straightforward : the double (triple)
integration is replaced by a double (triple) summation.
In the case of axisymmetry, all integrands include a factor equal to 2r where r = radius of
the integration point under consideration, and 2 can be devided out.
Standard quadrature types used for stiffness/forces evaluation for different elements are given
in the table below:
Standard quadrature for typical elements
Element:
N gauss
T3
1
Q4
22
TH4
1
W6
21
B8
222
Details on integration point positions, weighting factors, for different quadrature types are
given in Appendix 4.7.2
December 3, 2008

TM111

4.2.3
STRAINS
Strains at any integration point of a continuum element are evaluated by an unified procedure given in Window 4-8. Beside the standard approach presented here, in the case of
incompressible or dilatant media another approach is recommended, see Section 4.3.
Window 4-8: Evaluation of strains
Quadrilateral element (Q4) outlook

An unified procedure for all 2/3D standard continuum elements include:
isoparametric mapping:
x () =
Nen
X
Na () xa
a=1
displacement interpolation:
u () =
Nen
X
Na () ua
a=1
strain:
() = B () u =
Nen
X
Ba () ua
a=1
where the standard form of matrix B is given in section 4.2.1. In the case of strain
projection technique (necessary to simulate incompressible media) resulting form of the
Bmatrix, called BBar, will be described in Window 4-9.
shape function derivatives, present in all concerned Ba are evaluated as:
T Na
Na
= J1
x
where Jacobian J = x/ for 2D and 3D is calculated as follows
x x
x x

y y

2D: J = y y ;
3D: J =

z z

Window 4-8
December 3, 2008

TM112

4.2.4
STIFFNESS MATRIX
The element stiffness matrix can now be evaluated numerically; recalling that:
Z
e
K = BT D B d
Fe =
BT d
Numerical integration yields:

Ngauss
e
K =
BT ( i ) Di B ( i ) det J ( i ) Wi
i
Ngauss
Fe =
BT ( i ) ( i ) det J ( i ) Wi
where the integrand is evaluated at the integration point i.

The standard integration rules are used for each element, see Section 4.2.2.
December 3, 2008

TM113

4.2.5
BODY FORCES AND DISTRIBUTED LOADS
The introduction of the interpolation functions into the weak form yields an elemental body
force term as follows:
Z
e
Fb = NT b de
e
For distributed forces, the equivalent nodal force is seen to be:

Z
e
NT t de .
Ft =
e
December 3, 2008

TM114

4.2.6
INITIAL STRESSES, STRAINS
If the initial stress field is in equilibrium the initial stresses can simply be added to the stresses
resulting from deformations. If not, equivalent nodal forces must be calculated. Nodal forces
equivalent to initial stresses and/or strains can be introduced as follows; this calls for a
generalized stress-strain relation.
Let
= D ( 0 ) + 0
where 0 are the initial stresses and 0 are the initial strains. By the definition of the internal
force term:
Z
Z
Z
T
T
A B d = A B D ( 0) d + A BT0 d
e=1,N
e=1,N
e=1,N
from which the following expression of the nonlinear equilibrium equation results
Fint (u) = F F + F0
with
F0 =
F =
e=1,N
e=1,N
BT 0 d
BT 0 d.
Note that if initial stresses are in equilibrium F0 = 0 no deformations will result from the
initial stresses. If not, deformations will result. Similarly if initial strains are compatible with
boundary conditions, no stresses result.
December 3, 2008

TM115

4.3
INCOMPRESSIBLE AND DILATANT MEDIA
A strain projection technique is necessary in order to simulate incompressible media. The

resulting form of the Bmatrix, called BBar, will be described later on.
Another finite element technique to simulate incompressible and highly dilatant plastic media
is the Enhanced Assumed Strain method (EAS). The general idea and detailed algorithmic
scheme is described in Section 4.3.2. This option is activated in ZSOIL when dilatant plastic
flow is active.
INCOMPRESSIBLE MEDIA: B-BAR STRAIN PROJECTION METHOD
DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD
December 3, 2008

TM116
N Preface N N Numer. Impl. N N N Incompressible Media

4.3.1
INCOMPRESSIBLE MEDIA : B-BAR STRAIN PROJECTION

METHOD
Material data include the elastic constants i.e.Youngs modulus and Poisson ratio. In fully
incompressible cases, Poissons ratio is 0.5 which within the program is taken as 0.499999.
This leads to mesh locking when a full (2 2) integration scheme is used. Underintegration
(i.e. the use of a single integration point) is therefore required in the volumetric part of the
stiffness, while full integration is used for the shear term.
It can be seen from the iterative scheme that the use of seective underintegration would
result in different integration schemes for the left and righthand side of the equilibrium
equations. This can, and will, disturb convergence, so strain projection is used to circumvent
this problem.
First, the BBAR matrix is established for the case of torsionless axisymmetry and then the
plane strain case is derived.
Beginning with the standard Bmatrix for node a defined previously in this section, we can
split the matrix into dilatational and deviatoric parts as shown in Window 4-9.
a is used where the dilatational term has been underintegrated.
In place of using Ba the B
entries are calculated using a reduced 1 1 integration rule whereas the
The dilatational B
entries are calculated using the normal (2 2) integration rule.
deviatoric B
Window 4-9: Strain projection B-BAR method
dil
Ba = Bdev
a + Ba
where

(B0 + B1 ) B2
1 (B0 + B1 ) B2
=
0
0
3
0 + B
1 B
2
B
Bdil
a
and
a:
In place of use Ba use B
Bdev
a

1
2
1
B1 B0
B2
3
3
3

1
1
2
B2
B0 B1
3
3
3
B2
B1
2
1
1
B0 B1
B2
3
3
3
a = Bdev + B
dil
B
a
a
i.e.:
B12
B10
a =
B
B2
11
B
B6
B7
B1
6
B
where
B0 = Na /x1
1 B1 /3
B4 = B

0 B0 /3
B10 = B4 + B
B1 = Na,x1

2 B2 /3
B6 = B
B11 = B0 + B10
B2 = Na,x2
B7 = B2 + B6
B12 = B1 + B10
Window 4-9
December 3, 2008

TM117

Remarks:
1. The plane strain case many be obtained from the above formulation by simply setting
0 = B0 = 0 and using Cartesian rather than cylindrical coordinates. The 2r (or r)
B
factor included in the integrands in the case of axisymmetry must also be ignored.
the formulation is used.
2. Notice that the D matrix remains 4 4 for plane strain, when B
December 3, 2008

TM118

4.3.2
DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD
INTRODUCTION TO ENHANCED ASSSUMED STRAIN (EAS) APPROACH

EXTENSION OF EAS TO NONLINEAR ELASTOPLASTIC ANALYSIS
REMARKS AND ASSESSMENT OF EAS ELEMENTS
December 3, 2008

TM119
N Numer. Impl. N N Incompressible... N N N Dilatant Media

4.3.2.1
INTRODUCTION TO ENHANCED ASSSUMED STRAIN

(EAS) APPROACH
In EAS methods1 the strain is approximated by two fields, i.e. the compatible and the
enhanced one via equation :
= Bu + M
where B is the standard strain operator, M is the interpolation operator for the additional
strain field which may be discontinuous across element edges.
The variational basis of the method is the HuWashizu functional in the form:

Z
1 T
T
T
T T
D + Bu d N F d.
U=
2
Substitution of the modified strain field into functional yields:

Z
1
T
T
T
T T
U=
(Bu + M) D (Bu + M) (Bu + M) + Bu u N F d.
2
In order to eliminate the statically admissible stress field from functional the following condition has to be satisfied:
Z
T M d = 0.
The system of equations obtained by taking the variation of functional with respect to u and
takes the form:

ext
Kuu Ku
u
F
=
0
Ku K
where:
Z
Kuu =
B DB d ;
Ku =
BT DM d
M DB d ;
Ku =
K =
MT DM d
Assuming discontinuous approximation for these extra degrees of freedom can be eliminated
at the element level through standard condensation procedure. The interpolation functions
assumed for M matrix, defined in isoparametric space, for plane strain case and axisymmetry
are given in Windows (4-10) and (4-11).
The forms of M matrix for quasi-plane strain and 3D are given in Windows (??) and (4-12).
These interpolation functions assumed for M (here index is used to distinguish between
isoparametric and physical spaces) have to satisfy the orthogonality condition introduced
already in 4-th equation in this section. The transformation of M matrix from isoparametric
space to physical one, summarized in Window (4-13), is performed with aid of Jacobian of
the isoparametric mapping.
1
Simo J.C., Rifai M.S., A class of mixed assumed strain methods and method of incompatible modes.
International Journal for Numerical Methods in Engineering. Vol.29, pp.1595-1638, 1990
Groen A.E., Improvement of low order elements using assumed strain concepts. TU-Delft report Nr
25.2.94.203, 1994.
December 3, 2008

TM120

Window 4-10: M matrices for Q4 element in plane strain
EASelement
Number of
extra modes
EAS2
EAS4
EAS7
M matrix for plane

= x , y , xy ,
0
0
M =
0 0
0 0
0 0 0
0 0 0
M =
0 0
0 0 0 0
0 0 0
0 0 0 0
M =
0 0 0
0 0 0 0 0
strain
T
z
0 0
0
0
0 0
Window 4-10
Window 4-11: M matrices for Q4 element in axisymmetric case

EASelement
Number of
extra modes
EAS3
M =
EAS5
M =
where:
=
rT a0 =
M matrix for axisymmetry

T
= r , y , ry ,
0
0
0
0
0
0
0
J (, )
0
0
r (, ) J0
0
0
0
0
0
0
0
0
0
0

0
J (, )
0
0
0
0
r (, ) J0
1 rT a1
3 rT a0
1
(r1 + r2 + r3 + r4 )
4
1 rT a2
3 rT a0
1
rT a1 = (r1 + r2 + r3 r4 )
4
=
1
(r1 r2 + r3 + r4 )
4
r1 , r2 , r3 , r4 radial coordinates of element nodes.
rT a2 =
Window 4-11
December 3, 2008

TM121

Window 4-12: M matrices for B8 element in 3D case
EAS-element
Nr of extra modes
EAS6
EAS9
M matrix for 3D case

( =
{ x y xy z
0 0
0 0
0 0 0 0 0
M =
0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 0
M =
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0 0
xz yz
0
0
0
0
0 0 0
0 0 0
0 0 0
0 0 0
0 0
0
}T )
Window 4-12
December 3, 2008

TM122

Window 4-13: Mapping from isoparametric to physical space
The operator M is mapped onto physical space through the Jacobian of the isoparametric
transformation at the element centroid (which is the origin of the isoparametric space). For
second order tensors the following transformation from the isoparametric to physical space
holds (J is evaluated in current integration point while J0 is evaluated at the center):
isoparametric
=
ij
det J
Jki0 physical
Jjl0 .
kl
det J0
det J0 1 isoparametric
T
.
det J 0
det J0 1
From the above equation one gets the definition of M matrix: M =
T M
det J 0
For plane strain and axisymmetry the transformation matrix T0 takes the form:
2
2
J11
J21
J11 J21
0
2
2
J12
J22
J12 J22
0
T0 =
at ( = 0, = 0)
2J11 J12 2J21 J22 J11 J22 + J12 J21 0
0
0
0
1
With matrix notation this mapping can be rewritten in form : physical =
For 3Dcase the transformation matrix T0 takes the form:

2
2
2
J11 J31
J21 J31
J11 J21
J31
J21
J11
2
2
2
J12
J12 J32
J22 J32
J12 J22
J32
J22
2J11 J12 2J21 J22 J11 J22 + J21 J12 2J31 J32 J11 J32 + J31 J12 J21 J32 + J31 J22
T0 =
2
2
2
J13
J23
J13 J23
J33
J13 J33
J23 J33
Window 4-13
December 3, 2008

TM123

4.3.2.2
EXTENSION OF THE EAS METHOD TO NONLINEAR

ELASTOPLASTIC ANALYSIS
With the EAS method the resulting nonlinear system of equations to be solved is as follows:
ext

F Fint (ue , e ) = 0
A
e=1,N
Z
he (ue , e ) = MT d = 0 (e = 1, 2, . . . , Nele)
Linearization of the above yields:

ext R

T
Kuu Ku
u
F
B
R
=
Ku K
0 MT d
where:
ep
ep
B D B d
Kuu =
Z
M D B d
Ku =
December 3, 2008
Z
Ku =
Z
K =
BT Dep M d
MT Dep M d

TM124

4.3.2.3
REMARKS AND ASSESSMENT OF EAS ELEMENTS
1. the EAS2 element in plane strain, EAS3 in axisymmetry and EAS6 in 3D are strongly
recommended for stability and ultimate state analysis when using isotropic plastic models
with dilatant plastic (for ex. Drucker-Prager if > 0)
2. the EAS4 element in plane strain, EAS5 in axisymmetry and EAS9 in 3D are strongly
recommended for stability and ultimate state analysis when using anisotropic plastic models
with any type of plastic flow (only multilaminate model)
3. if one wants to get highly accurate results for beams/shells modeled with aid of continuum
elements (retaining walls, etc... ) the Q4-EAS7 (for plane strain case exclusively) and
B8-EAS15 (3D) elements give superior results even for coarse distorted meshes, but their
application is limited to elastic type of materials only (these elements are activated once
Continuum for structures instead of Continuum is used for Material formulation
to be defined at the material level).
December 3, 2008

TM125

4.4
FAR FIELD
Standard finite elements can be used only to discretize a bounded domain. In the case of
a problem given in an unbounded domain, infinite elements, attached to the boundary of
a bounded domain, can be used to describe interaction between bounded domain and its
infinite surroundings. The formulation is based on the concept of infinite mapped elements
(see 2 ). The infinite mapped elements concept is summarized in the following windows.
Formulation of the mapped infinite element
Mapping for 2D infinite element
Mapping for 3D brick type of infinite element
Mapping for 3D wedge type of infinite element
Evaluation of element matrices for infinite mapped elements
O.C. Zienkiewicz, C. Emson, P. Bettes, A Novel Boundary Infinite Element, International Journal for
Numerical Methods in Engineering, vol. 19, p.393-404, 1983
December 3, 2008

TM126

Window 4-14: Formulation of the mapped infinite element
To summarize the concept of infinite mapped elements let us cosider the following onedimensional situation given in the Fig. The element which extends from node 1 through node
2 up to node 3, placed at infinity, is mapped onto parent element in the local coordinate
system. Once the arbitrary position of the pole of expansion x0 is choosen one can define the
position of node 2 using the expression
x2 = 2 x1 x0
The mapping from local coordinate system to the global one can be done in standard way
using the following rule
2
X
x() =
Ni () xi
i=1
where node 3 is disregarded from the summation.

The infinite mapping functions are defined as follows
N1 () = 2/(1 )
N2 () = 1 N1 ()
It can be easily seen that = 1, 0, +1 corresponds respectively to positions of node 1, node
2 and node 3 in global coordinate system.
The basic field variable is interpolated using standard shape functions and written in the
polynomial form takes the form
u() = a0 + a1 + a2 2 + ............
in which coefficients ai depend on element shape functions. Solving the mapping expression
for we get
2a
=1
r
where the distance from the pole O to any point within the element is denoted by r and
a = x2 x1 . Substituting the expression for into expression for basic field variable we can
get it in the form expressed in terms of global coordinate r
b1 b2
+ 2 + ..........
r
r
where b0 = 0 is implied since the variable u vanishes at infinity. From the above one can
show that suitable decay functions 1/rn for basic variable one can get by choosing proper
shape functions for it.
Window 4-14
u() = b0 +
December 3, 2008

TM127

Window 4-15: Mapping for 2D infinite element
The mapping functions for the 2D element can be easily obtained as a product of onedimensional Lagrangian shape functions along local direction and infinite shape functions
along direction . If we assume that the poles positions are defined then we can define
positions of nodes 2 and 3 as
x2 = 2 x1 xO1
x3 = 2 x4 xO4
altough the definition of the infinite element geometry is usually done by introduction of
nodes 1, 2, 3 and 4 rather then by poles positions O1 and O2.
The set of the infinite shape functions for quadrilateral infinite 4-node element INFQ4 is as
follows
M1 (, ) = N1 () N1 ()
M2 (, ) = N2 () N1 ()
M3 (, ) = N2 () N2 ()
M4 (, ) = N1 () N2 ()
in which the Lagrangian shape functions Ni ()
1
(1 )
2
1
N2 () = (1 + )
2
N1 () =
The coordinate element mapping is defined as

x(, ) =
4
X
Mi (, ) xi
i=1
The interpolation of the displacement field within the element is assumed in the form
u(, ) = N1 (, ) u1 + N4 (, ) u4
Window 4-15
December 3, 2008

TM128

Window 4-16: Mapping for 3D brick type of infinite element
The scheme of mapping for 3D brick type of infinite element is shown in the following figure
The appropriate set of the shape functions is as follows

M1 (, ) = N1 () N1 (, )
M2 (, ) = N1 () N2 (, )
M3 (, ) = N1 () N3 (, )
M4 (, ) = N1 () N4 (, )
M6 (, ) = N2 () N1 (, )
M7 (, ) = N2 () N2 (, )
M8 (, ) = N2 () N3 (, )
M8 (, ) = N2 () N4 (, )
in which the shape functions Ni (, ) are the standard shape functions as for standard 4-node
quadrilateral element.
x(, , ) =
8
X
Mi (, , ) xi
i=1
u(, ) =
4
X
Ni (, ) ui
i=1
Window 4-16
December 3, 2008

TM129

Window 4-17: Mapping for 3D wedge type of infinite element
The scheme of mapping for 3D wedge type of infinite element is shown in the following figure
The appropriate set of the shape functions is as follows

M1 (, ) = N1 () N1 (, )
M2 (, ) = N1 () N2 (, )
M3 (, ) = N1 () N3 (, )
M4 (, ) = N2 () N1 (, )
M5 (, ) = N2 () N2 (, )
M6 (, ) = N2 () N3 (, )
in which the shape functions Ni (, ) are the standard shape functions as for standard 3-node
triangular element.
x(, , ) =
6
X
Mi (, , ) xi
i=1
u(, ) =
3
X
Ni (, ) ui
i=1
Window 4-17
December 3, 2008

TM130

Window 4-18: Evaluation of element matrices for infinite mapped elements
The numerical integration of required infinite element matrices is performed with aid of
standard Gauss integration scheme. The Jacobian, Cartesian derivatives of standard shape
functions Ni and domain differential d, for general 3D case, are computed as follows
Mi
Mi
Mi
xi
yi
zi
NX
EN
M
M
M
i
i
i
x
y
z
J=
i
i
i
i=1
Mi
Mi
Mi
xi
yi
zi
Ni
N
x
N
N
i
i
1
=J
Ni
Ni
d = dx dy dz = det(J) d d d
Window 4-18
December 3, 2008

TM131

4.5
OVERLAID MESHES
To explain the idea of overlaid meshes let us consider a 2D domain 30m wide and 30m high,
with a square excavation zone of size 6m by 6m placed at the center of the domain. During
the excavation a tunnel lining is installed (see Fig. below).
q = 1 kN/m/m
Tunnel
lining
Kinematic
constraints
In the context of the overlaid meshes approach we can create a relatively coarse mesh, labeled
as layer number 0, which will represent soil behavior far from the excavation zone. In the
neighborhood of the excavation zone, to enhance the quality of the results, we need a denser
mesh and thus we superpose a second mesh labeled as layer number 1. In the 2D case
we could easily generate a mesh in layer 1 in such a way that its external boundary nodes
positions coincide with the nodes of the coarse mesh. However, in 3D applications this is
usually not that easy and it is simpler, from the users point of view, to superpose layer 1
with layer 0 without additional restrictions. This corresponds to overlaid meshes for which
we have to handle the problem of overlapping of few meshes plus the continuity of primary
variables (displacements, pressures etc..) handled via Lagrange multipliers.
In the considered example a mesh consisting of 10 x 10 elements is generated in layer 0 and 20
x 20 elements in layer 1 (layer 1 is 14m wide and 14m high) (see Fig. above). As it is shown in
Fig. below some of the elements in layer 0 are fully overlapped (identified with circles) while
some others are overlapped only partially (identified with triangles). All these elements which
are fully overlapped, and nodes which exclusively belong to these elements, are neglected
during computations. Problems arise in the case of partially overlapped elements as part
of their domain must be deleted from the resulting stiffness and right-hand-side vector. To
handle this case an exact geometrical intersection of a given element in layer 0 and elements
in layer 1 must be computed to estimate the overlapped volume. For further application, this
intersection is computed for each integration point subdomain of a given element in layer 0
and set of elements in layer 1.
December 3, 2008

TM132

Element fully overlapped
Element partially overlapped
If we analyse the situation shown in Fig. below we can notice that for the integration point
number 2 the ratio between nonoverlapped and total volume becomes zero (as the integration
area associated with point 2 is entirely overlapped) while for other integration points it is
smaller than 1 but greater than zero. Hence a simple computation scheme, in which each
integration point volume is weighted by the aforementioned ratio, can be used.
Element in layer 0
4
Elements in layer 1
Non overlapped
area associated
with point 1
December 3, 2008

TM133

4.6
ALGORITHMS
Several solution algorithms can be activated either individually or in a sequence: initial state,
stability, time dependent (driven load/consolidation/creep).
All refer in some way to a NewtonRaphson iterative scheme which is therefore described in
Window 4-19. The specific algorithms are described next.
The simulation of excavation/construction stages is managed through existence functions
associated with elements at the level of the data preparation.
MODIFIED NEWTON-RAPHSON ALGORITHM
CONVERGENCE NORMS
STABILITY ANALYSIS
CONSOLIDATION
CREEP
LOAD TIME FUNCTIONS AND TIME STEPPING
EXCAVATION AND CONSTRUCTION STAGES
December 3, 2008

TM134
N Preface N N Numerical Impl. N N N Algorithms

4.6.1
FULL/MODIFIED NEWTON-RAPHSON ALGORITHM
Window 4-19: Full/Modified Newton-Raphson algorithm
1. Step initialization : n = n + 1
set iteration counter i = 0
(i=0)
set un+1 = 0
Accumulate total external force vector Fext
n+1
2. At the element level, at each integration point
if creep is active then compute the creep strain increment cr
n+1 according to the
Section 3.4 (MATERIAL MODELS/CREEP)
set the initial stress/strain increments 0n+1 , 0n+1
3. i = i + 1
4. Assemble internal force vector and stiffness matrix using computed at the element level
(i+1)tot
and at each integration point current total stress n+1
and current tangent stiffness
(i+1)e
ep
matrix Dn+1 ( an elastic stress state is computed via formula n+1 = n + 0n+1 +
(i+1)
De (n+1 cr
n+1 0n+1 ) )
Z
(i+1)tot
int (i+1)
BT n+1
d
Fn+1
= A
e=1,N
e
(i+1)
Kn+1
e=1,N
ep(i+1)
BT Dn+1
B d
5. Solve system of equations

int (i+1)
(i+1)
Kn+1 u(i+1) =Fext

n+1 Fn+1
(i+1)
(for modified NR technique assume Kn+1 = Kn

(i+1)
(for full NR use curent Kn+1 matrix)

(i+1)
(i)
6. Update accumulated increment of displacements: un+1 = un+1 + u(i+1)

7. If convergence of the iteration process has not been achieved go to step 1
Window 4-19
December 3, 2008

TM135

4.6.2
CONVERGENCE NORMS
RHS OUT OF BALANCE EUCLIDEAN NORM

ENERGY INCREMENT NORM
TOTAL ENERGY NORM
CONVERGENCE/DIVERGENCE/CONTINUATION STRATEGY
The right-hand side (RHS) out of balance Euclidean norm, total energy and incremental
energy norms are used in the code to detect convergence or divergence of the NewtonRaphson iterations. These norms are evaluated separtely for solid and for fluid phases in
case of twophase computations. The RHS out of balance norm, incremental energy norm,
and total energy norm are defined in Windows 4-20, 4-21, 4-22 respectively. Detection of
convergence/divergence state and possible continuation strategy is setup inWindow 4-23.
Window 4-20: RHS-out of balance Euclidean norm
The RHS-out of balance Euclidean norm is defined as follows:
int (i+1)
k
kFext
n+1 Fn+1
TOL
int
kFext
n+1 Fn k
This norm is evaluated selectively for different degree of freedom ( translations u, rotations
, pore pressures pF , temperatures T and humidities W ). The following table shows active
RHS norms for different analysis types and different degrees of freedom (+ means active and
- nonactive).
Analysis Type/DOF type
Deformation
Deformation+Flow
Flow
Heat transfer
Humidity transfer
u
+
+
-
pF
+
+
+
-
+()
+()
-
T
+
-
W
+
The two following tollerances for RHS norm are used i.e. FTOL (set to 0.01 by default)
and QTOL (set to 0.001 by default) related to solid phase DOFs and fluid phase DOFs
respectively. The following table shows corresponding tollerances for selected DOFs.
DOF
u
pF
T
W
Active tollerance TOL

FTOL
QTOL
FTOL
FTOL
FTOL
()
int
ext
The reference value, kFext
n+1 Fn k, related to rotational DOFs is set to 0.1 kFn+1
u
Fint
n k L where L is some averaged finite element size, factor 0.1 has been assumed by rule
int u
of thumb and kFext
n+1 Fn k is an Euclidean norm of the RHS corresponding to translational
DOFs.
Window 4-20
December 3, 2008

TM136

Window 4-21: Energy increment norm
Th increment of energy and then the energy increment norm (valid for iteration i 2) are
evaluated as follows (for solid and fluid phase separately):
S(i+1)
En+1
F(i+1)
En+1
S(i+1)
En+1
1 int u(i+1) (i+1)

= Fn+1
u
2
1 int (i+1)
= Fn+1 pF (i+1)
2
S(i)
En+1
S(i=2)
ESTOL
by default ESTOL = 103
En+1
F(i+1)
En+1
F(i)
En+1
F(i=2)
EFTOL
by default EFTOL = 103
En+1
Window 4-21
Window 4-22: Total energy norm
The total energy and then the total energy norm (valid for step n > 1) are evaluated as
follows (separately for solid and fluid phase):
S(i+1)
= EnS + En+1
F(i+1)
= EnF + En+1
En+1
En+1
S(i+1)
En+1
F(i+1)
S(i)
En+1
ESTOL
EnS
F(i+1)
En+1
S(i+1)
by default ESTOL = 103
by default EFTOL = 103
F(i)
En+1
EFTOL
EnF
Window 4-22
December 3, 2008

TM137

Window 4-23: Convergence / divergence / continuation
Convergence strategy
The step is converged if all active RHS out of unbalance norms are satsfied simutaneously
(see Window 4-20)
int (i+1)
k
kFext
n+1 Fn+1
TOL (FTOL/QTOL)
ext
int
kFn+1 Fn k
or if the following set of conditions is satisfied:
int (i+1)
k
kFext
n+1 Fn+1
10 TOL (for all active RHS norms)
ext
int
kFn+1 Fn k
S(i+1)
S(i)
F(i+1)
F(i)
En+1
En+1
ESTOL/10
EnS
En+1
EFTOL/10 (if active)
EnF
En+1
Detection of divergence
The divergence of the iteration process is detected if the following set of conditions is satisfied:
int (i+1)
kFext
k
n+1 Fn+1
(for all active RHS norms)
ext
int
kFn+1 Fn k
S(i+1)
En+1
S(i+1)
En+1
S(i)
En+1
EnS
and
S(i)
En+1
(i=2)S
En
and
F(i+1)
F(i)
F(i+1)
En+1
En+1
En+1
En+1
(if active)
EnF
(i=2)F
F(i)
(if active)
En
For stability driver = 5.0 and for all others = 10.0.

Continuation strategy
In case when the maximum number of iterations is reached the following action is performed
depending on the status of the check box Auto in calculation window called Increase max.
nr of iterations. If it is ON (default setting) and all RHS out of balance norms are below value
1.0 (100%) then the maximum number of iterations is updated automaticaly (increased by 5)
and the process continues unless the new updated maximum number of iterations overreaches
the absolute maximum number of iterations (set in the menu under item Control). If this
check box is OFF, when reaching current maximum number of iterations a dialog box will be
displayed on a screen waiting for new maximum number of iterations to be defined by the
user.
Window 4-23
December 3, 2008

TM138

4.6.3
The initial state in soil mechanics results essentially (but not exclusively) from gravity. The
correct implementation of gravity requires simultaneous application (or superposition) of the
gravity itself and the corresponding initial stresses, which may undergo additional constraints
set by the user (initial state K0 effect ). The corresponding multistep algorithm is described
in Window 4-24:
Window 4-24: Initial state multistep algorithm
Initialize:
set time t = 0
assume first gravity increment and increment size for further steps
set = 0
set step counter n = 0
for each element and for each integration point set 0 = 0
1. Continue
2. Set counter of initial stress reevaluations (for superposition) l = 0
3. Set iteration counter i = 0
4. Continue
5. Evaluate equivalent partial gravity nodal load: Fg =
A
e=1,N
R
e
NT b d, where b is the
gravity load
6. Accumulate total external force vector Fext
n+1 adding to Fg additional external forces if
needed (also weighted by the factor )
(i)
ep(i)
7. Evaluate corresponding stress state n+1 and constitutive tangent matrix Dn+1 for given
(i)
(i)
(i)
un+1 , n+1 = B un+1 and given 0 , using the following expression:
(i)
(i)
p(i)
n+1 = n + 0 + De (n+1 n+1 )

p(i)
ep(i)
where the increment of plastic strain n+1 and Dn+1 are obtained from return mapping
algorithm
8. Assemble internal forces (adding the effect of initial pore pressures if needed, here pore
pressures are also weighted by the current gravity multiplier )
Z

int (i)
(i)
F (i)
T
T
Fn+1 = A
B n+1 + pn+1 d
e=1,N
9. Assemble stiffness matrix: K =
December 3, 2008
A
e=1,N
R
e
ep(i)
BT Dn+1 B d

TM139

int (i)
10. Solve system of equations: K u(i+1) = Fext

n+1 Fn+1
(i+1)
(i)
11. Update accumulated increment of displacements and strains: un+1 = un+1 +u(i+1) ,
(i+1)
(i+1)
n+1 = B un+1
12. If convergence of the iteration process has not been achieved set i = i + 1 and go to step
3
13. Increase l = l + 1
14. IF l <MAX REPEAT (by default MAX REPEAT=2) then
Update the initial stress increment 0 (via procedure given in Window 4-25 Initial
stress increment correction
Repeat last step introducing corrected initial stresses, GO TO step 2
(i+1)
(i+1)
15. Set : un+1 = un+1 = 0, n+1 = n+1 = 0

16. If n < nmax set n = n + 1, = + and GO TO step 1
Window 4-24
Window 4-25: Initial stress increment correction
Correction of the assumed initial stresses is done based on the previous stress state n ,
current stress state n+1 and additional set of constraints which include the effect of K0
(the ratio between selected normal stress components can be defined in a local coordinate
system defined by the user). We assume that the y component of the stress state is computed
()
()
()
()
accurately and we want to fulfil the conditions : x / y = Kox and z / y = Koz . The
following steps lead to the evaluation of corrected initial stress increment 0 .
()
()
1. Transform n+1 n+1 , n n (upper index () means local system, one in which
K0 has been set up)
()
()
()
2. Compute stress increment between n and n + 1 steps: n+1 = n+1 n

()
()
3. Assume : 0 = n+1
()
()
()
()
()
()
4. IF K0x is specified by the user then correct 0x 3 : 0x = K0x 0y

5. IF K0z is specified by the user then correct 0z : 0z = K0z 0y
()
()
6. Transform 0 from local K0 system to global one: 0 0

Window 4-25
Specification of K0 can be done exclusively in confined directions
December 3, 2008

TM140

4.6.4
STABILITY ANALYSIS
The available definitions of safety factors were introduced in Appendix 3.7.1, i.e.
1. , such that global instability corresponds to
global failure = m + s
where is a uniform deviatoric stress multiplier in the domain. From the implementation
point of view the approach amounts to a progressive modification of the yield criteria
parameters until failure occurs.
2. Alternatively for two-parameter (C ) criteria, can be used

R

d
Rs y s
SF2 =

s ds
where y = C + 0n tan 0 is the yield stress according to the MohrCoulomb criterion, C
the cohesion, 0n the effective normal stress, the friction angle, SF2 the safety factor and
s defines the sliding surface. An algorithm which fits the plasticity based approach can
be deduced from the above equation; rewriting the equation as
R
R
Z
ds
(C + 0n tan 0 ) ds
s y
s
ds =
=
SF2
SF2
s
it is observed that SF2 can be viewed as the dividing factor of C and tan0 for which
instability is reached.
The stability algorithm is summarized in Window 4-26
Window 4-26: Stability algorithm for SF=SF2
1. Initialization
Set SFn = SF0
(start with the prescribed lower bound of the safety factor SF0 )
2. For each step

SFn+1 = SFn + SF (increment the safety factor)
C
Cn+1 =
(scale the cohesion)
SFn+1
tan 0
0
(scale the tangens of friction angle)
tan n+1 =
SFn+1
3. Solve the boundary-value problem, iterate as needed
4. Go to 2 until divergence occurs
5. At divergence SFn SFSFn+1 ; estimate of the safety factor
Window 4-26
December 3, 2008

TM141

Remarks:
1. Notice that the material characteristics are modified by the algorithm, which is, strictly
speaking, only valid close to SF= 1.
2. Divergence is normally accompanied by a localized strain field on a slip surface, in which
SF corresponds to the given definition.
3. Let SF2 be the default definition of the safety factor for two-parameter (C ) criteria.
December 3, 2008

TM142

4.6.5
An ultimate load analysis follows the classical NewtonRaphson scheme. Computation are
pursued until divergence is reached. Refer to Section 4.6.1. The simulation of drained
behaviour should be done using drivers as for 1phase or 2phase medium but using large
time step. The undrained behavior can be done only with 2phase algorithm (Section 4.6.6).
December 3, 2008

TM143

4.6.6
CONSOLIDATION ANALYSIS
The consolidation algorithm simulates the transient behavior of the twophase medium. The
time step can vary starting from small values at the beginning up to large steps at the end
of the process. The only limitation is that minimum time step should be greater than the
critical one estimated via following relation:

F h2
1 1
t tcrit =
+ Eoed c
(4.2)
Eoed k 4 6
where:
c
k
F
Eoed
h
compressibility of fluid = n/ F ; n porosity, F fluid bulk modulus

Darcy coefficient
fluid specific weight
integration coefficient (in Z SOIL = 1)
E (1 )
=
oedometric stiffness modulus
(1 + ) (1 2)
element size.
The one-dimensional test is taken as the basis for the estimation. tot
The condition for nonoscil
F
(with a positive sign
latory pore pressure distribution can be formulated as p
1 + Eoed c
for tensile stresses). In the incompressibility limit (c = 0) this condition means that the
stress transferred by the fluid cannot be higher then the total stress value. The estimate
will be derived for linear two-node element with an equal interpolation for displacements and
pressures. Let us take a mesh which consists of a single finite element, as shown in following
figure, and assume that the initial pressure is zero. Solving the resulting system of equations
One-dimensional test
with the set of the boundary conditions (u1 = 0 and p2 = 0) we obtain the following value
for the pressure increment in the first time step:
p1 =
h
4Eoed
qh
2Eoed
h kt 1
+c +
3
F h
(4.3)
From the consistency condition, which in a general case can be expressed in the form,(see 4 ):
p1
4
Vermeer P., Verruijt A. An accuracy

Int.J.Numer.Anal.Meth.Geomech.5, p.1-14, 1981
December 3, 2008
q
1 + Eoed c
condition
for
(4.4)
consolidation
by
finite
elements.

TM144

the condition given in eq. 4.2 for the time step t is derived.
Remark:To avoid this restrictive condition the two-phase stabilization method is available in
Z Soil program.
December 3, 2008

TM145

4.6.7
CREEP ANALYSIS
Creep introduces a time dependent deformation at constant or variable stress. The implementation scheme is summarized in Sect. 4.6.1. An important aspect is that the current
formulation does not produce creeping effect due to initial stresses. The phenomenon starts
due to stress variation starting at time at which the creep is activated. This means that for
example the effect of secondary consolidation understood as a result of creeping should be
modelled running the consolidation and creep simultaneously. Since creep is a time dependent
process it requires therefore a time stepping procedure, which is shown in Window 4-27.
December 3, 2008

TM146

4.6.8
LOAD FUNCTION AND TIME STEPPING PROCEDURE
For time dependent processes like creep, consolidation, transient flow, each item consisting
load or prescribed boundary condition may have attached a function describing its variability
in time (physical or pseudo) during the process. This is called a Load Time Function and is
defined as a list of pairs (tk , vk ), as shown in Window 4-27 assuming linear interpolation for intermediate points. Different loads acting on the model may have attached different load time
functions, but its argument must be common for all time function in a job. It must have the
meaning of a physical time (consistently with constitutive data ) for such a Time Dependent
problems as: Transient Fluid, Heat, Humidity Flow, Creep, Consolidation, while for
the remaining (i.e. Driven Load, Deformation, all Steady State) ones, may be treated
as the non-physical one.
Window 4-27: Time stepping procedure
Load time history
Applied loads are characterized by a prescribed load amplitude and a load multiplier (load
factor). At time tn+1 the applied load is then
Fn+1 = F0 LFn+1 .
The load factor LF is defined as a function of time. Each load can be associated with a
different load factor.
Load timehistory
Time step
The time dependent processes start always with the initial time step increment defined by the
user t1 = tBEG . Each next time step is predicted through relation tn+1 = tn tMULT
with time multiplier tMULT defined by the user. This next time step increment can be
corrected automatically by the system if in between tn tn + tpredicted
at least one of the
n+1
existence functions changes its value (from OFF to ON or vice versa) or some characteristic
point on one of the load time functions is detected. Once such situation is detected the time
step increment is reset to the initial value tBEG .
Window 4-27
December 3, 2008

TM147

4.6.9
SIMULATION OF EXCAVATION AND CONSTRUCTION STAGES
The simulation of excavation/construction stages is normally embedded in every time dependent driver (driven load / consolidation etc..). This type of simulation requires a stiffness
update at each excavation stage. The corresponding appropriate algorithmic choice is done
automatically by the system.
On the other hand, the input mesh must correspond to the maximal mesh. This may correspond to the final stage in a construction simulation or to the initial state for an excavation
simulation.
Management of stiffness
The simulation of construction stages leads to a variable size of the stiffness matrix and
therefore to a variable number of nodes and elements. This requires additional data management. The input mesh must correspond to the maximal mesh. The corresponding node and
element numbering will be kept throughout the analysis.
At each excavation stage an optimization of the nodal numbering will be performed, and
a correspondence table with the input-output node numbering will be established. The
optimized mesh node numbering is used for analysis purposes.
Progressive unloading after excavation
If no LT F (unloading function) is specified for excavated elements, interaction forces from
excavated media will vanish immediately at the moment of excavation. In the case of elastoplastic media this may cause difficulties in obtaining converged solution. To prevent this,
progressive unloading after excavation can be used which helps the system to redistribute
stresses in the surroundings of the excavated domain, and in consequence to obtain convergent
solution concerning new equilibrium state.
Unloading after excavation can be controlled. The interaction force between the excavated
and the remaining medium will be computed by performing

Z
LT F (t)
intEXC
T tot
F
=
B d
0 if no unloading function is given
exc
over the excavated medium. The association of a load timehistory with this set of forces
provides the means for the control of unloading. For more details see section 7.6.2.
Related Topics
GEOTECHNICAL ASPECT. EXCAVATION, CONSTRUCTION ALGORITHM
December 3, 2008

TM148

4.7
APPENDICES
SHAPE FUNCTIONS AND REFERENCE ELEMENTS FOR 2D/3D

NUMERICAL INTEGRATION DATA
MULTISURFACE PLASTICITY CLOSEST POINT PROJECT ALGORITHM
SINGLE SURFACE PLASTICITY CLOSEST POINT PROJECTION ALGORITHM
December 3, 2008

TM149
N Preface N N Numerical Implementation N N N Appendices

4.7.1
SHAPE FUNCTION DEFINITION AND REFERENCE ELEMENTS FOR 2/3 D
Reference element
Shape functions
N1 =
N2 =
1
Ni = (1 + i ) (1 + i )
N3 =
N4 =
1
(1 ) (1 )
4
1
(1 + ) (1 )
4
1
(1 + ) (1 + )
4
1
(1 ) (1 + )
4
N1 = 1
N2 =
N3 = .
1
Ni = (1 + i ) (1 + i ) (1 + i )
8
1
N1 = (1 ) (1 ) (1 ) , N5 =
N2 = 1 (1 + ) (1 ) (1 ) , N6 =
8
1
N3 = (1 + ) (1 + ) (1 ) , N7 =
N4 = 1 (1 ) (1 + ) (1 ) , N8 =
8
1
T3
(, )
Ni = (1 + i ) Nk(i)
2
where: k (i) = (i 1) mod 3 + 1
N1
N2
N3
N4
December 3, 2008
1
(1 ) (1 ) (1 + )
8
1
(1 ) (1 ) (1 + )
8
1
(1 ) (1 ) (1 + )
8
1
(1 ) (1 ) (1 + )
8
=1
=
=
=

TM150

4.7.2
NUMERICAL INTEGRATION DATA FOR DIFFERENT ELEMENTS IN 1/2/3D
3
=
= 0.5773502691896
3
Element shape
Integration points
Positions
i
i
number
Ngauss
i:
0.0
2.0
1:
2:
1.0
1.0
1:
0.33333
0.33333
0.5
1:
2:
3:
0.16666
0.16666
0.66666
0.16666
0.66666
0.16666
0.16666
0.16666
0.16666
1:
0.0
0.0
4.0
22=4
1:
2:
3:
4:
1.0
1.0
1.0
1.0
in
December 3, 2008
Weigthing
factors Wi

TM151

Element shape
number
Ngauss
1
0.13819
0.58541
0.13819
0.13819
0.13819
0.13819
0.58541
0.13819
0.13819
0.13819
0.13819
0.58541
0.04166
0.04166
0.04166
0.04166
1:
0.33333
0.33333
0.0
1.0
1:
2:
3:
4:
5:
6:
1:
0.166666
0.666666
0.166666
0.166666
0.666666
0.166666
0.0
0.166666
0.166666
0.666666
0.166666
0.166666
0.666666
0.0
0.0
0.166666
0.166666
0.166666
0.166666
0.166666
0.166666
1.0
1:
2:
3:
4:
5:
6:
7:
8:
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
222
=
8
i
0.25
Weigthing
factors Wi
0.16666
1:
2:
3:
4:
23=6
December 3, 2008
i:
1:
Integration points
Positions
i
i
0.25
0.25

TM152

4.7.3
MULTISURFACE PLASTICITY CLOSEST POINT PROJECTION ALGORITHM
A generalized, unconditionally stable, algorithm designed for the integration of constitutive

equations for class of multisurface elastoplastic models is presented here. This subject has
been worked out first by Simo & Hughes but procedures described therein were not as general
as needed. Three crucial problems e.a.: selection of the reduced set of active mechanisms
solving the plastic corrector problem in case when the number of active mechanisms is greater
than the stress space dimension , activation of initially nonactive plastic mechanisms due to
effects of hardening/softening and finally definition of the general form of a consistent tangent
matrix for multimechanism models with generally nonassociated flow rule, and including hardening/softening, were introduced later by Szarlinski and Truty. Slightly modified algorithm
including an additional substeping scheme is described here. The complete set of information
on Multisurface Closest Point Projection Stress Return Algorithm ( MCPPSS) can be found
in the following Windows:
Notation
Basic set of equations
Consistent tangent operator Dep
Window 4-28: Notation
M
Jact
a
= t
r (, q)
h (, q)
D
C
k
k+1
n
n+1
number of all plastic mechanisms,

actual set of active plastic mechanisms,
mechanism index,
plastic multiplier value,
flow vector,
column matrix of hardening/softening functions,
column matrix of plastic (hardening/softening)
parameters
column matrix of total plastic strains,
the value of plasticity condition for given stress
state and plastic parameters
elastic stiffness matrix
elastic compliance matrix
index of previous closest point projection algorithm iteration
index of actual iteration
index of last configuration of equilibrium
index of actual configuration of equilibrium
Window 4-28
Window 4-29: Set of basic equations

The integration of constitutive equations for any elastoplastic model consists of two stages
e.g. elastic predictor and plastic corrector one. Assuming the fully implicit integration scheme
December 3, 2008

TM153

for plastic strains and hardening/softening parameters this second stage is expressed by the
following set of equations which is the basis of the closest point projection algorithm (in the
following windows this will be called the basic set of equations):
X
p(k+1)
(k+1) (k+1)
n+1
= pn +
n+1 rn+1
=1,M
(k+1)
(k+1)
(k+1)
qn+1 = qn +
n+1 hn+1
=1,M
(k)
(k+1)
(k)
Fn+1
= Fn+1 + Fn+1 = 0

Jact
(k+1)
p(k)
n+1 = Dn+1
In most cases this system is nonlinear and NewtonRaphson procedure is needed for its
solution. To apply this procedure the consistent linerization of the above set of equations
has to be done. Denoting by R and Rq the residuals of the first two equations which are
defined by the following expressions:
X
p(k)
(k) (k)
R = pn n+1 +
n+1 rn+1
=1,M
Rq = q n
(k)
qn+1
(k) (k)
n+1 hn+1
=1,M
the linearized system of first two equations of the basic set using expressions for R and Rq
and 4th equation from basic set is as follows (summation for ):
Jact
(k)
An+1
1

=
D 0
0 I
(
p(k)
n+1
(k)
qn+1
X
(k) (k)
)
R +
n+1 rn+1
=1,M
X
=
(k) (k)
R
+
n+1 hn+1
q
=1,M
where:
(k)
An+1
r(k)
D +
=1,M
=
(k)
X
h
=1,M
=1,M
=1,M
r(k)
q
h(k)
In order to find increments of plastic multipliers the third equation from basic set written for
each mechanism F : Jact has to be solved. After linearization it takes the form:
(k)
(k+1)
(k+1)
(k)
(k+1)
Fn+1 = Fn+1 + Fn+1 = Fn+1 +
(k)
Fn+1
Fn+1
(k)
(k)
n+1 +
qn+1
(1)
It can be written in matrix form using 4-th equation from basic set which leads finally to the
following:
#
"
(k)
(k)
F
(k)
n+1
n+1
Fn+1 =
(2)
D
q
In the above equation vector of unknowns which consists ofplasticstrain and plastic variables
(k)
increments can be eliminated via expression derived for An+1

matrix manipulations this expression can be rewritten in the form:
(k)
(k)
(k)
(k)
(k)
(k)
and finally after some
(k)
Un+1 An+1 Pn+1 (k) = fn+1 Un+1 An+1 Rn+1
December 3, 2008

TM154

where:
(k)
Rn+1
(k)
Pn+1
(k)
Un+1

=
R
Rq

r 1 . . . r N
=
h1 . . . hn
T T
f 1
f 1
.
..
.
=
. T . T
f N
f N

Window 4-29
Window 4-30: Consistent tangent operator Dep
The general definition of the tangent matrix is as follows:
Dep =
d
.
d
This definition after introducing of the applied integration scheme leads to the so called
consistent tangent matrix. The starting point for its derivation is the basic set of equations
given in Window 4-29 which expresses the fully implicit scheme applied for the integration
of plastic strains and plastic variables. All next steps will lead to derivation of the relation
between d d and encapsulating from it the algorithmic constitutive operator. To do that
lets differentiate this basic set of equations which yields:
X
X
r
r
d+
dq+
d r
q
=1,M
=1,M
=1,M
X
X
X
dq =
d+
dq+
d h
q
=1,M
=1,M
=1,M
F
F
d F =
d+
dq = 0;
Jact
q
d = D ( d dp )
dp =
Considering only first two equations from the above set and eliminating the increment of
plastic strains using fourth equation we can write the following matrix equation:
X
r
r
C+
q
=1,M
=1,M
X
X
h
h
q
=1,M
=1,M
{z
A1
December 3, 2008
d r

=1,M
d
d
=
X
0
dq
d h
=1,M

TM155

Let us evaluate the stress increment considering once again the fourth equation from linearized
basic set
!
X
X
X
r

p
r
d
dq
d r
d = D ( d d ) = D
q
=1,M
=1,M
=1,M
" X
#

X
r
r
d
= D d
d r .
=1,M
q
dq
=1,M
=1,M
{z
}
|
a
By elimination of the plastic strain differential from the above equation we can rewrite as:
r

=1,M
d
.
X
d = D d aA
+ aA
0
d h
=1,M
Let us introduce some auxiliary matrices to simplify further derivations:

matrix e
r which consists of column stored flow vectors for
[r1 , . . . , rN ]
d 1
..
vector of plastic multiplier increments:
d = .
each active mechanism
e
r=
d N

matrix K
[(NSTRE+NHARD)NSTRE]
:K=
I
0

.
With these defnitions one can reform the equation for d:

d
d = D d aA
+ (aAPe
r) d
0
but still it is necessary to define vector d. It can be done writing appropriate consistency
condition for each active mechanism. Thus for all (NJACT) active mechanisms one gets the
following system
F F

0

q
d
.
.
.
..
..
dq = .. .
NF NF
0
q
Eliminating vector of unknowns the expression for increments of plastic multipliers will take
finally the form:

d
1
d = (UAP) UA
.
0
Introducing the above formula into the equation for d, after some matrix manipulations,
we get the most general form of the multimechanism consistent elasto-plastic matrix:

d = D I aAK + (aAPe
r) (UAP)1 UAK d.
|
{z
}
Depconsistent
December 3, 2008

TM156

4.7.4
SINGLE SURFACE PLASTICITY CLOSEST POINT PROJECTION ALGORITHM
The case of a single surface plasticity with/without hardening is handled via multisurface
plasticity theory and algorithms assuming M=1 (nr of plastic surfaces), see Appendix 4.7.3
December 3, 2008

TM157
December 3, 2008

TM158
N Preface
Chapter 5
STRUCTURES
TRUSSES
BEAMS
SHELLS
MEMBRANES
APPENDICES
December 3, 2008

TM159
N Preface N N Structures
5.1
TRUSSES
TRUSS ELEMENT
RING ELEMENT
PRE-STRESSING OF TRUSS AND RING ELEMENTS
UNI-AXIAL ELASTO-PLASTIC MODEL
December 3, 2008

TM160
N Preface N N Structures N N N Trusses

5.1.1
TRUSS ELEMENT

GEOMETRY AND DOF OF TRUSS ELEMENTS
WEAK FORMULATION OF THE EQUILIBRIUM
December 3, 2008

TM161

5.1.1.1
Elements are designed to model the presence of different kind of anchors, reinforcements ,
or separate bar (3D case). Note the difference between truss/anchor and 2D membrane
elements , see Section 5.4 .
Window 5-1: Truss elements in Plane Strain
Plane Strain case

Window 5-1
Window 5-2: Truss elements in axisymmetry
Axisymmetric case
Window 5-2
December 3, 2008

TM162

Window 5-3: Truss elements in 3D analysis type
3D case: single bar
Window 5-3
December 3, 2008

TM163

5.1.1.2
GEOMETRY AND DOF OF TRUSS ELEMENTS
Window 5-4: Truss element geometry
The reference element, shape functions, construction of isoparametric mapping, DOF

setting for 2/3 node truss element
Window 5-4
Geometry of the truss element is identified by the geometry of the centroid line:
X
o
x() =
Ni ()xi
i=1,2
Local cross-sectional coordinate system {xL , yL , zL } directions are set as:

o
e xL =
x,
;
ko x, k
e zL =
0, 0, 1
eyL = ezL exL ,
which for 2D cases correspond to the following rules:

1. xL axis is tangent to element with the direction pointing from 1st to 2nd node of the
element,
2. zL is perpendicular to global XY plane (meridian plane for Axisymmetric) and equal to
global Z axis,
3. yL is set from xL , zL by righthand screwdriver rule .
For an arbitrary point of an element identified by its reference coordinate and local cross
sectional positions r{yL , (zL )}, the mapping to the global coordinate may be put as:
x(, yL , zL ) =o x() + yL eyL + zL ezL
December 3, 2008

TM164

The Jacobi matrix of the above mapping may be put as:

J = o x, ; eyL ; ezL .
The localglobal {xL } {xG } transformation matrix may be put as:

T = exL ; eyL ; ezL ,
hexL i
TT = heyL i
hezL i
For for any vector:

vG = TvL , vL = TT vG
Evaluation of strains requires the derivatives of displacement components versus local xL axis
:
v,xL = v,
= v, D with D = hJ1 i1 exL
xL
For 2D cases a relation between local and global DOF as well as forces may be established
in a form :

c s
T
vG = TvL , vL = T vG , with T =
,
s c
where:
exL =
December 3, 2008
c, s
e yL =
s, c

TM165

5.1.1.3
For the point with given reference coordinate translation displacement components uG
(referred to global coordinate system {xG }) are interpolated from its values at nodal points
X
uG () =
Ni ()uGi
i=1,2
The only strain component taken into account while evaluating element stress is strain along
the element, and, as no bending is taken into account, this strain is assumed to remain
constant in the whole cross-section of the element:
xxL (xL , yL ) =
uG
uL
= e xL T
,
xL
xL
The formulae relating the only strains component xxL any point within the element, with its
DOF vector u may put in general for:
N en
X
xxL () = BxL ()u = BixL ()ui ,
i=1
For each analysis type in takes form:

Plane Strain, Axisymmetry:
BixL

T
ui = ui vi

= exL x DNi , exL y DNi ,
3D:
ui =
BixL =
December 3, 2008
ui , vi , wi
T
exL x DNi , exL y DNi , exL z DNi ,

TM166

5.1.1.4
The virtual work principle expressing equilibrium of a system may be put in the general form:
find such that:
for any , u
Z
dV
V
uT pdV = 0
The contribution of the truss elements in the above may be easily derived, with integrals taken
along the element length. A is an area attributed to the assumed computational domain i.e.
to the unit slice for the Plane strain, Generalized Plane Strain , to the whole circumference
(independently from current radius) for the Axisymmetric case or to one distinct bar for the
3D case.
for any , u
Z
Z
xxL xxL AdL
uT pdL = 0
For the description of 1D uniaxial elastoplastic material model used for truss elements see
the Window 5-25.
December 3, 2008

TM167

5.1.1.5
Stiffness matrix Ke and force vector f e of the truss element are derived in a standard way
from the weak formulation of a problem. It would require constitutive module Dxxxx as well
as xxL stress evaluation for given strain increment. This will be performed by 1D uniaxial
elastoplastic model.
Numerical integration technique is used to evaluate integrals over the length of the element.
In the case of a 2node linear element, 1 integration point is used (N gaus = 1, W1 = 2.0,
1 = 0.0). Integration over the cross section of the element is hidden in given values of
integral characteristics of the cross section area A.
Ke =
Z1
BxL T ()Dxxxx BxL ()ko x()kd =
1
N gaus
BxL T ( igaus )Dxxxx BxL ( igaus )ko x()kWigaus
igaus=1
fe =
Z1
BxL T () xx ()ko x()kd =
1
N gaus
BxL T ( igaus ) xx ( igaus )ko x( igaus )kWigaus
igaus=1
The element p load is assumed to act along the centroid line leading to load induced forces
evaluated as:
Z1
fp =
NT ()pko x()kd =
1
N gaus
NT ( igaus )pko x( igaus )kWigaus
igaus=1
where:

N() = Ni ()I(dim) ,
December 3, 2008
i = 1, N en

TM168

5.1.2
RING ELEMENT

December 3, 2008

TM169
N Structures N N Trusses N N N Ring Element

5.1.2.1
The geometry of the ring element (available exclusively for Axisymmetric analysis type) is
represented by the coordinates of its 1st node and attached area.
Window 5-5: Geometry and DOF of a ring elements
Ring element. Geometry and DOF
Window 5-5
Although only u contribute to element strain both {u, v} displacement components are kept
as element DOF, to allow force resulting from vertical load to be transmitted to the system
via ring element.
The only strain component taken into account while evaluating element stress is strain in
circumferential direction, constant in the whole crosssection of the element, related only to
the radial component of the displacement and current radius:
zz =
u
r
The formulae relating zz , with elements DOF vector u:

zz = Bzz u,
u=

Bzz =
December 3, 2008
1
, 0
r
u, v

T

TM170

5.1.2.2
From the general form of the V.W.P it may be stated as:

for any , u :
zz zz (2rA) uT p(2r) = 0
For the description of 1D uniaxial elastoplastic material model used for ring element see
the Window 5-25
December 3, 2008

TM171

5.1.2.3
Stiffness matrix Ke and force vector f e of the truss element is derived in a standard way
from the weak formulation of a problem. It would require constitutive module Dzzzz as well
as zz stress evaluation for given strain increment. This will be performed by 1D uniaxial
elastoplastic model, see the Window 5-25
Ke = BT
zz Dzzzz Bzz 2rA
f e = BT
zz zz 2rA
December 3, 2008

TM172

5.1.3
For the sake of generality the nodal points of truss elements may be placed in arbitrary
position within the domain occupied by continuum or structural element. Despite modeling
convenience, the option also possesses some physical meaning as then the truss element
force are distributed over surrounding nodes of continuum element and the effect of force
concentration is diminished.
Window 5-6: Anchoring of truss and ring elements
Conditions to be fulfilled:
displacement compatibility:
uT russ = uCont
force equivalence (weak form):
for any u :
uT
T russ fT russ
N en
X
(i)T
(i)
uCont fCont
i=1
Anchoring of a truss/ring node within continuum element
Window 5-6
December 3, 2008

TM173
N Structures N N Trusses N N N Anchoring ...

5.1.3.1
NUMERICAL REALIZATION OF ANCHORING
Let Ni are shape functions of the element in which kth node of a truss is anchored, and
are local coordinates of the anchorage point within its reference element. Displacement of a
truss node uT russ are to be evaluated from displacement of surrounding element uCont via
compatibility condition :
uT russ = uCont ( ) =
N enCont
X
Ni ( )uiCont = NT uCont
i=1
where:
N = [Ni ( )IN dm ] ,
i = 1, N enCont
In the case of anchoring within structural (i.e. beam or shell element), shape function matrix
N must be understood in more general sense i.e. as a matrix relating translation displacement
at arbitrary point within the element with all its DOF and takes the form given in point 1.2,
1.3.
In turn, forces evaluated at the node of a truss fT russ are transferred to the nodes of the
surrounding element as fCont basing on (weakly formulated) equivalency
for any u :
uT
T russ fT russ
N en
X
(i)T
(i)
uCont fCont = uCont fCont
i=1
T
uT
T russ fT russ = uCont Nf Cont = uCont fCont
fCont = Nf T russ
Finally, the stiffness of the truss element is assembled on the DOF of surrounding element,
and takes a form as bellow, where Kjk are jth and kth nodal submatrix of the truss
stiffness:
.
Kjj .. Kjk NT
b =
s
s
K
s
..
T
NKkj . NKkk N
In the case of the ring element, its stiffness assembled on the DOF of surrounding elements
takes form:
b = NKkk NT .
K
December 3, 2008

TM174

5.1.4
PRESTRESSING OF TRUSS AND RING ELEMENTS
Pre-stress can be applied only to anchor/truss or ring elements. Prestress differs from a
standard initial stress situation by the fact, that the internal force is assumed to be known a
priori. According to the definition of current tangent stiffness matrix it follows that:
K=
f (u)
=0
u
since internal force f (u) is constant. As a consequence, if pre-stress is being applied then
the stiffness of the anchor is always set to zero while it is evaluated via standard procedures
if prestress is nonactive.
Strains in the anchor / ring element are monitored during the pre-stress as:
i = i1 + i
prestres (ti ) i1
i
=
E
December 3, 2008

TM175
5.2
BEAMS
GEOMETRY OF BEAM ELEMENT

INTERPOLATION OF THE DISPLACEMENT FIELD
RELAXATION OF INTERNAL DOF
BEAM ELEMENT RESULTS
December 3, 2008

TM176
N Preface N N Structures N N N Beams

5.2.1
GEOMETRY OF BEAM ELEMENT
Beams in 2D analysis types (Plane Strain , Generalized Plane Strain, Axisymmetry, Plane Stress)
Window 5-7: Beams in 2D analysis type modeling situations
In the case of Plane Strain
analysis, beam elements can be used for two - modeling situations:
Modeling situations for beam elements in Plane Strain and 2.5D analysis
In the case of Axisymmetric analysis, only continuum shell option can be used:
Beam elements in axisymmetric analysis

Window 5-7
System {xG , yG , zG } is the global one. Local element system {xL , yL , zL } is created in the
element in a following way:
1. xL - axis is tangent to the elements centroid axis at a given point, with the direction
pointing from 1-st to 2-nd centroid node of the element
December 3, 2008

TM177

2. zL - is perpendicular to global XY plane (meridian plane for Axisymmetric case) and
corresponds to global Z axis
3. yL -is set from xL , zL by right-hand screw driver rule
2D beam element: nodal DOF and coordinates systems
exL =
x,
ko x
, k
ezL =
0, 0, 1
eyL = ezL exL
Relation between local and global DOF (degrees of freedom) as well as forces may be established in a form :
c s 0
vG = TvL , vL = TT vG ,
T= s c 0
0 0 1
are local coordinate unit vectors.
Beams in 3D analysis type
The geometry of a 2 node beam element allowing for moderate (when 3 nodes are used) and
data, shown on the plot for the most general situation.
December 3, 2008

TM178

Window 5-8: Beams in 3D analysis type
Beam element in 3D analysis
Window 5-8
The versors of local cross sectional axis are evaluated from the above data at any point of
the element with reference coordinates as:
o
exL =
x,
ko x
, k
ezL =
exL d()
,
kexL d()k
eyL = ezL exL
Note, that iso-parametric mapping is used to interpolate both centroid position vector o x()
and directional vector d() from their nodal representatives.
X
X
o
x() =
Ni ()xi , d() =
Ni ()di
i=1,2
i=1,2
Centroids may coincide with corresponding masters. Directional nodes may be different for
each node enabling for moderate twist of cross-sectional axis or only 1 common point for all
nodes in the element may be given. The directional planes may be different then the element
curvature plane. The only obligation for the positions of the directional nodes is to omit the
situation when cross product is indeterminable, i.e. when tangent vector exL is parallel to the
directional vector d.
Local cross sectional axis are these to which the geometry of the crosssection is referred.
In the case of elastic beam model they must strictly correspond to centroidal and principal
crosssectional inertia axis . In the case of layered, nonlinear beam model this demand may
be satisfied only approximately, as most meaningful effects, related to axial strain and stresses
resulting from axial force and bidirectional bending action are taken into account by cross
sectional numerical integration. Then cross sectional axis setting will be used to input layer
centers positions. Moreover, cross sectional axis setting will be used to refer bidirectional
shear and torsion elastic characteristics.
December 3, 2008

TM179

For local cross sectional positions r{yL , (zL )}, the mapping to the global coordinate may be
put as:
x(, yL , zL ) = o x() + yL eyL + zL ezL
The Jacobi matrix of the above mapping may be put as:

J(, yL , zL ) = o x, + yL eyL , +zL ezL , ; eyL ; ezL
The local-global {xL } {xG } transformation matrix may be put as:
G
hexL i

v = TvL ,
T
T = exL ; eyL ; ezL , T = heyL i

thus for any vector :
v L = TT v G .
hezL i
Evaluation of strains requires the derivatives of any displacement components versus local xL
axis :
v,xL = v,
= v, D with D = hJ1 i1 exL
xL
In 3D analysis case derivatives of base vectors eL appear, evaluated as:

o
1
x, o x, o
exL , = o
I o
x,
k x, k
x, o x,

1
(exL d) (exL d)
e zL =
I
(exL , d + exL d, )
kexL dk
(exL d) (exL d)
eyL = ezL , exL + ezL exL ,
December 3, 2008

TM180

5.2.2
The adopted kinematics of a beam is based on Timoshenkos hypothesis, i.e. that originally
plane crosssection perpendicular to beam centroid line remains plane but not necessarily
perpendicular to deformed beam axis . Moreover small displacements & strains are assumed
Window 5-9: Kinematic assumption of beam theory
Kinematic assumption of beam theory
Window 5-9
Displacements at any point r with given local coordinate {xL , yL , zL } are then given by
displacement o u of centroid line and independent rotation of a cross section plane (fiber)
as :
u(x) = o u(xL ) + (xL ) r
For 2D case this leads to:
uL = o uL (xL ) (xL )yL
vL = o vL (xL )
Two formulations of beam element are used:
layered approach (Nonlinear Beam option, analysis type Plane Strain, Axisymmetry,
3D analysis )
composite sections allowed, nonlinear or elastic material models, setting of the centroid of
the crosssection may be done in approximate manner as M N coupling is taken into
account in the model);
integral approach (Elastic Beam option, Plane strain, Axisymmetry,
3D analysis),
uniform section described by its integral characteristics, elastic material model only, setting
of the centroid of the crosssection must be done in rigorously exact way, as M N coupling
is disregarded by the model)
December 3, 2008

TM181

For the layered approach strains are evaluated at each layer. These (related to local coordinate
system) are:
(2D case)
normal strain
xxL (xL , yL ) =
shear strain
xyL =
zL (xL )
o uG (xL )
uL
= eT
yL
,
xL
xL
xL
xL
vL uL
o uG (xL )
+
= eT
zL (xL )
yL
xL yL
xL
additionally, for Axisymmetry , circumferential strain is set as:

zzL (xL , yL ) =
uG
1
=
r
r

uG (xL ) cyL zL (xL )
In 3D analysis case and layered beam model strains related to the cross sectional axis are:
normal strain:
uL
xxL (xL , yL ) =
= eT
xL
xL
o uG (xL ) ((xL ) r)
+
xL
xL
average strains due to shear:

xyL (xL ) =
vL uL
vL
o uG (xL )
+
=
zL = eT
eT
yL
zL (xL )
xL yL
xL
xL
xzL (xL ) =
wL uL
wL
o uG (xL )
+
=
+ yL = eT
eT
zL
yL (xL )
xL
zL
xL
xL
torsion angle:
(xL ) =
xL
(xL )
= eT
xL
xL
xL
For the integral approach strains (in generalized sense) are (for Plane Strain ):
extension of {xL , 0, 0} line:
xxL (xL ) =
o uL
o uG (xL )
= eT
,
xL
xL
xL
curvature:
z (xL ) =
December 3, 2008
zL (xL )
,
xL

TM182

average shear strain:
xyL (xL ) =
vL uL
o uG (xL )
+
= eT
zL (xL ).
yL
xL yL
xL
In 3D case (integral approach) generalized strains are:

extension of {xL , 0, 0} line:
o
xxL (xL ) =
o uG (xL )
o uL
= eT
,
xL
xL
xL
curvatures:
y (xL ) =
z (xL ) =
yL (xL )
(xL )
= eT
,
yL
xL
xL
zL (xL )
(xL )
= eT
,
zL
xL
xL
Also average strains due to shear and torsion angle taking the form identical as for layered
beam are taken into account.
December 3, 2008

TM183

5.2.3
In both layered and integral approach, the treatment of a transversal shear (and torsion
for the 3D case) is decoupled from bending and extension state analysis and limited to
linear stress- strain relation. This kind of procedure is exact in linear cases, while for the
nonlinear one, consists a commonly accepted simplification. Coupled treatment of both,
shear & bending / extension state would require additional significant numerical effort and
much more complex formulation. In both formulations, the shear behavior is treated in
integral way, with usage of averaged shear angle, stress-resultant tangent force and a shear
correction factor.
The virtual work principle expressing the equilibrium of a beam may be put in the general
form:
find such that:
for any , u
Z
dV
V
uT p dV = 0
From the above, taking into account kinematic assumptions both layered and integral
approach, detailed formulation for each analysis type may be derived, i.e.:
layered formulation:
(2D, Plane Strain case)
for any , , u :
XZ
i
(i)
xxL xxL
dV
(i)
Z
xyL (ky GA) xyL dL
+
L
V (i)
uT p dV = 0
(2D, Axisymmetric case, dV is conical volume element)

for any , , u :
XZ
i
(i)
xxL xxL
(i)
(i)
zzL zzL
dV
(i)
Z
xyL (ky GA) xyL r dL
+ 2
L
V (i)
uT p dV = 0
(2D analysis case antiplane shear included optionally (hi),in each layer separately
for any , , u :
XZ
i
(i)
xxL xxL
(i)
h (i)
xzL xzL i
V (i)
dV
(i)
Z
xyL (ky GA) xyL dL
+
L
uT p dV = 0
(3D case)
December 3, 2008

TM184

for any , , , u :
XZ
i
(i)
xxL xxL
dV
(i)
X Z
Z
Z
xhiL (khi GA) xhiL dL+ GIx dL uT p dV = 0
hi=y,z L
V (i)
integral formulation for elastic beam model

(2D, Plane Strain case )
for any , , z , u :
[ xxL N + z M + (ky GA)] dL

L
uT p dV = 0
(3D case)
for any , hi , hi , u :
Z
L
Z
X

hi Mhi + xhi (khi GA) xhi ) + GIx dL uT p dV = 0
( xxL N +
o
hi=y,z
In the aboveshear correction factor k is introduced to account for nonuniform shear stress
strain distribution over the cross section. On the basis of energetic equivalency this may
be put as:
Z 2
A
1
S
= 2
dA.
k
I
b2
A
Moreover, shear module G must be specified by the user.

see the Appendix, Window 5-25.
December 3, 2008
For material model description

TM185

5.2.4
INTERPOLATION OF THE DISPLACEMENT FIELD
For the point lying on the element centroid line with given reference coordinate , both
translations u and rotational , displacement components (referred to global coordinate
system ) are interpolated from its values at the centroid nodes Ci. Note, that these may
be obtained from element DOF by master to centroid offset transformation ,see the
Appendix, Window 5-22.
X
o G
u () =
Ni ()uGCi
i=1,2
o
G () =
Ni ()GCi
i=1,2
December 3, 2008

TM186

5.2.5
The formulae relating strains any point within the element with its DOF vector q, based on
kinematic assumptions is given in general form:
(, yL , zL ) = B(, yL , zL )q =
N en
X

Bi (, yL , zL )qi ,
qi =
i=1
uCi
Ci
The displacement vector q consist of:

2DPlane Strain, Axisymmetry:
qi =
u, v, z
T
3D analysis:
qi =
u, v, w, x , y , z
T
i-th node submatrix of B-matrix relating shear strain with nodal DOF has a form in the case
of:
2DPlane Strain, Axisymmetry:
B i =
eyL x DNi , ; eyL y DNi , ; Ni
3D analysis:
B xyi =
T
DNi , eT
yL ; Ni eyL
B xzi =
T
DNi , eT
zL ; Ni ezL
In that case also torsion angle matrix B should be set as :

Bi = 013 ; DNi , eT
xL
Remaining strain component onenode submatrix Bi takes form (for straight geometry
element):
layered approach:
2DPlane Strain:
axial strain
BxLi =
exL x DNi , ; exL y DNi , ; DNi , yL
2DAxisymmetry:
axial strain:
December 3, 2008

TM187

BxLi =
exL x DNi , ; exL y DNi , ; DNi , yL
circumferential strain:

BxLi
Ni
=
;
r
0;
Ni
yL eyL y
r
3D analysis:
BxLi =
T
DNi , eT
xL ; D(Ni r exL + Ni , r, exL )
where:
r = yL eyL + zL ezL
integral approach:
2D-Plane Strain:
axial strain at centroid line:
Bi =
exL x DNi , ; exL y DNi , ; 0
curvature:
Bi =
0; 0; DNi ,
DNi , eT
xL ; 013
3D analysis:
axial strain at centroid line:
Bi =
curvature in xL zL plane:
Byi =

013 ; D(Ni , eTyL + Ni eT
,
)
,
yL
curvature in xL yL plane:
Bzi =
December 3, 2008

013 ; D(Ni , eTzL + Ni eT
zL , ) .

TM188

5.2.6
Stiffness matrix Ke and force vector f e of a beam element are derived in a standard way from
the weak formulation of a problem. It would require constitutive module matrix D for each
stressstrain component. Numerical integration technique is used to evaluate integrals over
the length of the element. In the case of a 2node linear element selected reduced integration
technique (SRI) with 1 integration point is used ( N gaus = 1, W1 = 2.0, 1 = 0.0).
Integration over the cross section of the element is performed numerically in the case of
layered approach, or it is hidden in given values of integral characteristics of the cross section
( shear area for uncoupled treatment of shear, area and rotational inertia for elastic beam).
Integrals in layered approach case use longitudinal jacobians l J = det(J(, yLl , zLl )) evaluated at each layer l separately, in order to enhance accuracy in case of curved elements.,
while other integrals over the element length use jacobians referred to centroid line o J =
det(J(, 0, 0)). For all 2D cases, contribution of shear part is evaluated in a common way
as:
Z1
K =
o
BT
()(kGA)B () J d
1
N gaus
o
BT
( igaus )(kGA)B ( igaus ) Jigaus Wigaus
igaus=1
Z1
f =
BT ()(kGA()) o J d
1
N gaus
o
BT
( igaus )(kGA( igaus )) Jigaus Wigaus
igaus=1
2DPlane Strain, layered approach (Nonlinear beam):

e
K =
Z1 X
1
l
l
BT
xL (, yLl )Dxxxx BxL (, yLl ) J d + K
N gaus
X X
l
l
BT
xL ( igaus , yLl )Dxxxx BxL ( igaus , yLl ) Jigaus Wigaus + K
igaus=1 l
Z1 X
l
fe =
BT
xL (, yLl ) xx (, yLl ) J d + f
1
N gaus
X X
l
BT
xL ( igaus , yLl ) xxL ( igaus , yLl ) Jigaus Wigaus + f
igaus=1 l
December 3, 2008

TM189
Plane Strain , integral approach (Elastic beam model):

Z1
K =
BT
()
BT
();

EA
0
0 ELz
B ()
B ()
J d + K
1
N gaus
X
BT
( igaus );
BT
( igaus )
igaus=1
o
B ( igaus )
B ( igaus )
Jigaus Wigaus + K
Z1

EA
0
0 ELz
f =
BT
();
BT
()
EA()
EIz ()
J d + f =
1
N gaus
X
BT
( igaus );
BT
( igaus )
igaus=1
EA( igaus )
EIz ( igaus )
Jigaus Wigaus + f
2DAxisymmetry, layered approach (Nonlinear beam):

1

Z X
T
l BxL (, yLl )
BxL (, yLl ); BT
Ke = 2
D
r() l Jd + rK
zL (, yLl )
BzL (, yLl )
l
N gaus
X X
= 2
l
T
BT
D
xL ( igaus , yLl ); BzL ( igaus , yLl )
igaus=1 l
BxL ( igaus , yLl )

BzL ( igaus , yLl )
!
r( igaus ) l Jigaus Wigaus + 2rK
f e = 2
Z1 X
T
BT
xL (, yLl ); BzL (, yLl )
xx (, yLl )
xz (, yLl )
N gaus
X X
T
BT
xL ( igaus , yLl ); BzL ( igaus , yLl )
igaus=1 l
r() l J d + rf =
xx ( igaus , yLl )
xz ( igaus , yLl )
!
r( igaus ) l Jigaus Wigaus + 2rf

where constitutive module matrix:
Dl(22)

=
Dxxxx Dxxzz
Dzzxx Dzzzz

.
3D analysis:
For both layered and integral approaches, treatment of transversal shear and torsion is de
coupled from treatment of bending and extension states. Thus element stiffness and forces
are split according to:
X
Ke = KeN M +
Kei + Ke
i=x,y
December 3, 2008

TM190

f e = fNe M +
e
fi
+ fe
i=x,y
The contributions common for layered and integral approaches are:

shear :
Z1
Kei =
o
BT
xi ()ki GABxi () J d
1
N gaus
o
BT
xi ( igaus )ki GABxi ( igaus ) Jigaus Wigaus
igaus=1
e
fi
=
Z1
o
BT
xi ()ki GA xi () J d
1
N gaus
o
BT
xi ( igaus )ki GA xi ( igaus ) Jigaus Wigaus
igaus=1
torsion:
Ke =
Z1
o
BT
()GIx B () J d
1
N gaus
o
BT
( igaus )GIx B ( igaus ) Jigaus Wigaus
igaus=1
fe =
Z1
o
BT
()GIx () J d
1
N gaus
o
BT
( igaus )GIx ( igaus ) Jigaus Wigaus
igaus=1
Bending and extension states are treated in different ways:

layered approach (Nonlinear beam model):
KeN M
Z1 X
l
l
=
BT
xL (, yLl , zLl )Dxxxx BxL (, yLl , zLl ) J d
1
N gaus
X X
l
l
BT
xL ( igaus , yLl , zLl )Dxxxx BxL ( igaus , yLl , zLl ) Jigaus Wigaus
igaus=1 l
December 3, 2008

TM191

fNe M
Z1 X
l
=
BT
xL (, yLl , zLl ) xx (, yLl , zLl ) J d
l
N gaus
X X
l
BT
xL ( igaus , yLl , zLl ) xx ( igaus , yLl , zLl ) Jigaus Wigaus
igaus=1 l
integral approach (Elastic beam model):
EA 0
0
T
T
0 EIy 0
BT
KeN M =
(); By (); Bz ()
0
0 EIz
1
N gaus
EA
X

T
T
T
B ( igaus ); By ( igaus ); Bz ( igaus )

0
=
igaus=1
0
B ( igaus )
By ( igaus ) o Jigaus Wigaus
Bz ( igaus )
Z1
Z1
fNe M =
B ()
By () o J d
Bz ()
0
0
EIy 0
0 EIz
EA()
T
T
EIy y () o J d
BT
() By () Bz ()
EIz z ()
N gaus
X
T
T
BT
( igaus ); By ( igaus ); Bz ( igaus )
igaus=1
EA( igaus )
EIy y ( igaus ) o Jigaus Wigaus
EIz z ( igaus )
The element load p is assumed to act along the centroid line leading to load induced forces
evaluated as:
Z1
fp =
N gaus
T
N ()p J d =
NT ( igaus )p o Jigaus Wigaus
igaus=1

N() = Ni ()I(dim) ,
i = 1, N en
All element forces and stiffness, as given above, are referred to centroidal nodes of the
element. Prior to the agregation to global ones, they are submitted to Mastercentroid
(offset) transformation, see Appendix 5.5.1 .
December 3, 2008

TM192

5.2.7
In order to deal with the frequently encountered situation where beam elements are connected
to other elements of the model by nodes, which are not the centroids of a beam cross section,
offset transformation of element displacement, forces and stiffness is introduced. The DOF
of the element are placed on its master nodes defining connectivity. Based on rigid body
movement of the master, translation and rotation displacements of the centroid are
evaluated.
For the details of master-slave offset transformation see the Appendix 5.5.1 .
December 3, 2008

TM193

5.2.8
RELAXATION OF INTERNAL DOF
Stiffness matrix and force vector derived above concerns element with all DOF active. Situation when user demanded group of DOF is relaxed (what means that no forces are transmitted
by the element in selected directions).The directions of relaxed DOF might be related to local
element directions at both ends of the element.
If given directions at beam node are relaxed, the node is implicitly duplicated and DSC
(Node-to-node interface ) elements are introduced to the model. The stiffness of that interface remains 0 at relaxed direction, while it is assumed to take penalty values, estimated
automaticaly from beam stiffness on fixed DOF.
December 3, 2008

TM194

5.2.9
BEAM ELEMENT RESULTS
Element stress resultants are referred to cross section local coordinate system at the integration point. The sign convention is as follows:
N normal force
positive in tension
Mi bending moment positive, are such which leads to positive (tensile) stresses in the
points with positive local coordinate on the complementary axis
Qi shear forces
positive is such that produce positive shear stresses acting in the
ith local direction
Mx torsion moment positive moment is represented by the vector directed towards
outer normal
Note, that in case of all analysis types, signs of some stress resultants are related to node
order C1 C2
Window 5-10: 2D beam element result setting
2D beam element results sign convention
Window 5-10
Note, that for Plane Strain and case of discrete beam system in Z direction stress resultants
are referred to single beam but not a unit slice of a model.
December 3, 2008

TM195

Window 5-11: Axisymetric shell element result setting
Axisymmetric shell element results sign convention
Window 5-11
Note, that in the case of Axisymmetry, evaluated stress resultants are referred to the unit
length of the axisymmetric shell both in circumference and meridian direction.
Window 5-12: 3D beam element result setting
3D beam element results sign convention
Window 5-12
December 3, 2008

TM196
5.3
SHELLS
Following paragraphs present most important aspects of applied shell elements

ELEMENT MAPPING AND COORDINATE SYSTEMS
CROSSSECTION MODELS
DISPLACEMENT AND STRAIN FIELD WITHIN THE ELEMENT
TREATMENT OF TRANSVERSAL SHEAR
WEAK FORMULATION OF EQUILIBRIUM
LOADS
SHELL ELEMENT RESULTS
December 3, 2008

TM197
N Preface N N Structures N N N Shells

5.3.1
Shell elements are 2D elements in 3D space, which are of course compatible with the continuum, beam, truss and, contact elements. The degenerated continuum approach is used, in
which deformation field within the element is described by translations and independent fiber
rotations field (Mindlin-Reissner hypothesis) of the reference surface. The fiber is assumed
to be in-extensive. The principles of the shell kinematics are shown in the following Figure.
Shell kinematics principle

The displacement field u within the element is split into two parts. First is related to reference
surface translation 0 u, and second U is related to fiber rotation:
u(, ) =
u() + U(, )
U(, ) = z()(2 () e1 1 () e2 )
where:
T
n= [, ]
o point coordinates at the reference element,
ei (), i = 1, 2, 3 fiber coordinate system,
fiber rotation vector components referred to fiber
i (), i = 1, 2
coordinate system,
z()
lamina coordiante.
In each lamina plane stress hypothesis is adopted. Moreover transversal shear strains and
stresses are accounted for. Each node of the reference surface posses 6 DOF, i.e.:
q = {u1 , u2 , 1 , 2 , 3 }T
Nodal rotations are transformed to nodal fiber coordinate systems
n
o
eai (), i = 1, 2, 3; a = 1, N en .
Bending rotation components i (), i = 1, 2 are used to evaluate element strains and then
stresses, while only stabilization stiffness is attached to so called drilling rotation 3 .
December 3, 2008

TM198

5.3.2
Four (8) and eight (16) nodes shell elements are available. Two modeling options and related
classes of shell elements are:
A. Shell elements with 2 layers of nodes (Window 5-13)
B. Shell elements with 1 layer of nodes (Window 5-14)
Window 5-13: Shell elements 2 layers
Shell elements with 2 layers of nodes

Window 5-13
Elements are designed to be fully compatible in all kind of connections with surrounding
volume elements, without overlapping. They have DOF attached only to one (master) layer
of nodes creating reference surface, slave nodes are used to define the geometry. Master
nodes 1 to 4 characterize the SHQ4 element with bilinear interpolation are slave nodes which
help constructing meshes and manage variable thickness. Shell elements are connected (or
not) to the other elements through either master or slave elements. As the only degreesof-freedom retained for analysis correspond to master nodes all loads applied to slave nodes
will be transferred to master nodes assuming a rigid link, with possibly resulting moments.
Similarly results (displacements) obtained at master nodes are transferred to slave nodes
assuming a rigid link, so that a master node rotation may result in an additional slave node
translation. Moreover, offset transformation, see Appendix 5.5.1 , is performed on force
vectors and stiffness matrices of all other elements adjacent to the slave node layer. Described
modeling option posses however following limitations:
1. mechanical boundary conditions can not be applied to slave nodes,
2. while connecting beam to shell elements Master-to-Master and slave-to-slave condition
must be respected.
December 3, 2008

TM199

Window 5-14: Shell elements 1 layer
Shell elements with 1 layer of nodes

Window 5-14
Elements are designed to be used in thin shell-structural systems. Element thickness (variable
allowed) must be set as an additional data. It is of course taken into account during element
stiffness/forces evaluation, but in the sense of geometrical modeling 0 thickness is in fact
assumed. This means that the volume occupied by a shell may be over-lapped in the case
when a volume element is adjacent to the shell element (i.e. when shell nodes consist a face
of a volume element). Fiber is assumed to be strictly orthogonal to the reference surface.
Window 5-15: Setting of the variable thickness data
Variable thickness of the shell elements is set using the following algorithm:
A family of fictitious (and temporary) thickness interpolation macro-elements is introduced.
It consist of linear L1, surface triangular T3,T6 and surface quadrilateral Q4, elements.
User has to define the thickness values {hi , i = 1, N en} at the nodes of these macroelements.
Current point x at the shell element is projected ortogonally onto the surface of fictitious
element x x0 , and then local coordinates 0 of a point x0 at the macro-element are set.
The thickness at the current point h(x) is equal to the value at the projection point h(x0 )
and, as such, is interpolated from the nodal values of the thickness using macro-element
0
shape functions Ni .
h(x) = h(x0 ) =
hi Ni ( (x0 ))
i=1,N en
Window 5-15
Despite above difference in element geometry description all others mechanical features of
the element (kinematics, constitutive modeling) are common for both classes A. and B.
December 3, 2008

TM200
Different possibilities of variable thickness setting
Practical examples of the variable thickness setting
December 3, 2008

TM201

5.3.3
ELEMENT MAPPING AND COORDINATE SYSTEMS
Given construction of the shell element is common for each type of shell element SXQ4,
SHQ4.
Window 5-16: Shell elements coordinate systems and mapping
Nodal basis, after
for each node a = 1, N en
{b
eai ,
i = 1, 2, 3} =
[b
ea1 , b
ea2 , b
ea3 ] =
1
1
n2
nx ny nx
nz +
1 + nz y
1 + nz
1
1
2
nx ny nz +
nx ny
, if nz 6= 1
1 + nz
1 + nz
nx
ny
nz
0
1 0
0 1 0
0 0 1
, if nz = 1
where: na = [nx , ny , nz ]T is element reference surface normal at node a.

Mapping from the reference element is constructed as follows, after2 : x(, , ) =
X(, , )
with:
x(, )+
reference surface:
o
x(, ) =
Na (, ) o xa ;
a=1,N en
position on the director:

X(, , ) =
Na (, ) za ()b
ea3;
a=1,N en
where:
1
2
1
1
za () = (1 + )za+ + (1 )za
2
2
J.L. Batoz, G. Dhatt, Modelisation des structures par element finis, Ed.Hermes, Paris 1992.
T.J.R. Hughes, The Finite Element method, Ed.PrenticeHall Inc.1988.
December 3, 2008

TM202

za+ = ha ;
za = 0;
za+
za
= ha /2;
for SHQ4
za , = ha /2.
= ha /2; for SXQ4
Jacobi matrix of the mapping:

J = x, ; x, ; x, = o x, ;o x, ; 0 +
X
X
+
za () [b
ea3 Na , ; b
ea3 Na , ; 0] +
[0; 0; b
ea3 Na , ]
a=1,N en
Shape function global derivatives:
Na ,
Na
= JT Na , ;
x
0
a=1,N en
Na ,
0
zNa
= JT z Na , + Na 0
x
0
z,
Derivatives versus local directions {xiL }, i = 1, 2, 3 are evaluated as:
= eT
.
i
xiL
x
Gauss point base, at layer after3 , see Appendix: 5.5.3
Shell element coordinate systems and mapping
Window 5-16
T.J.R. Hughes, The Finite Element method, Ed.PrenticeHall Inc.1988.
December 3, 2008

TM203

5.3.4
CROSSSECTION MODELS
Independently from the element geometry description, two approaches to modeling of the
cross sectional behavior of a shell are available. They are:
(i) homogeneous section, elastic model.
The cross section is fully described by elastic material data, i.e. Young module E and Poisson
ratio .Note, that element thickness is set as geometric data, independent from cross section
model. Numerical integration along the depth of an element is performed using 2-point Gauss
rule.
(ii) layered model (nonhomogenous section or/and material nonlinearity).
The cross section is understood as a set of layers with possible different material models in
each. In the case of homogenous section but non-linear material model introduction of layers
may be understood as the specification of integration rule, different than 2-point Gauss rule
used in the case of homogenous elastic section.
The following data must be specified for the whole section:
number of layers N layer N LAY ER M AX = 20
elastic properties E,
averaged volume unit weight ( for evaluation of gravity load )
while for each layer required are:
zi , the position of the layer center (relative or the distance from top/bottom)
wi =
hi
,
h
the relative depth of the layer or, for the fiber model, the area of fiber
per unit length.

indication of a material model attached to the layer.
The models to be used in that context are : elastic , elasto-plastic Menetrey
Willam, 1D elasto-plastic fiber model, In the latter case, for each elasto
plastic fiber model
E, the elastic modulus,
fty , the tensile yield stress,
fcy , the compressive yield stress,
nx , ny , nz direction cosines of a unit vector; the projection of n on the element
surface will define the fiber orientation, see Appendix 5.5.2
figure.
December 3, 2008
and example in the following

TM204
Shell crosssection and fiber orientation models
December 3, 2008

TM205

5.3.5
DISPLACEMENT AND STRAIN FIELD WITHIN THE ELEMENT
Assumed kinematic hypothesis together with interpolating (shape) functions built upon reference element with node number N en = 4 (SHQ4, SXQ4) lead to the following expressions
for the displacement field within the element. The displacement field may be expressed :
in terms of nodal translations ua and rotations a1 , a2 in the directions of nodal base
vectors b
e1 , b
e2 tangent to the element reference surface, as:
X
X
u(, ) =
Na ()ua +
Na ()z()(b
ea1 a2 b
ea2 a1 )
a=1,N en
a=1,N en
Above will be used to evaluate element strains.

in terms of element DOF vector q, including global components of rotations using
generalized shape function matrix N, as:
u(, ) = N(, )q
q = [ua1 , ua2 , ua3 , a1 , a2 , a3 ]T , a = 1, N en

T
b
b
N(, ) = Na ()I33 ; z()Na ()(b
ea1b
eT
e
e
)
, a = 1, N en
a2
a2
a1
Above, will be used for evaluation of element equivalent forces, and for numerical realization
of anchoring, see p. 5.1.3.1 Strain approximation is built upon nodal translations ua
and rotations a1 , a2 in the directions of nodal base vectors b
e1 , b
e2 tangent to the element
reference surface, consisting 5 components of DOF vector q(5)
q(5) = [ua1 , ua2 , ua3 , a1 , a2 ]T ,
u1L
,
m =
x1L

u2L
,
x2L
u1L u3L
s =
+
,
x3L x1L
a = 1, N en

u1L u2L
+
= Bm q(5)
x2L x1L

u2L u3L
+
= Bs q(5)
x3L x2L
Membrane strains are evaluated in unified way for all options considered, with Bm matrix
taking form:
..
.
.
, = 1, 2 ..
.
u ..
zNa
Na
a1
eT
1
x
x
1L
1L
zNa
..
T Na
Bm =
, a = 1, N en
.
e
a2
1
x
x
1L
2L
zNa
zNa
T Na
T Na
e1
+ e2
a1
+ a2
x2L
x1L
x2L
x1L
December 3, 2008

TM206

where:
ak1 = eT
ea2 ;
kb
ak2 = eT
ea1 .
kb
For transversal shear strains (i.e. mean shear angles), evaluated at the mid-surface standard
and enhanced options are provided. While standard option is used, shear strains are evaluated
with use of:
..
.
.
.
u ..
, = 1, 2 ..
zNa
zNa
eT Na + eT Na
a1
+ a3
3
1 x
x1L
x3L
x1L
3L
o
..
Bs =
.
Na
Na
zNa
zNa
eT
+ eT
a2
+ a3
2
3
x3L
x2L
x3L
x2L
a = 1, N en.
For enhanced option, in order to omit shear-locking problem in low order Q4 elements Mixed
Interpolation of Tensorial Components (MITC) approach is optionally introduced, after Batoz4 . In that case appropriate form of the o Bs matrix may be derived from following considT

eration. Covariant components of mean shear angle = , are interpolated from its
midside representatives as:
Mean shear angle components
1 A1 1 + A2
+
;
2
2
1 B1 1 + B2
+
;
2
2
where
T o
T
u, )|=0,=1 ;
A1
= (a1 + n
T
T o
B1
u, )|=1,=0 ;
= (a2 + n
a1 = o x, ;
a1 = o x, ;
T
T o
A2
u, )|=0,=1
= (a1 + n
T
T o
B2
u, )|=1,=0
= (a2 + n
X
=
Na ()(b
ea1 a2 b
ea2 a1 ).
a=1,N en
J.L.Batoz, G.Dhatt, Modelisation des structures par element finis, Ed.Hermes, Paris 1992.
December 3, 2008

TM207

After transformation components of mean shear angle S = [ 1L , 2L ]T in local Gauss point
base {Q} are obtained as:
1
S = C T ;
with: C =
December 3, 2008
aT
1 a1
aT
1 a2
symm aT
2 a2
aT
1 e1
aT
1 e2
T
aT
1 e2 a2 e2

TM208

5.3.6
TREATMENT OF TRANSVERSAL SHEAR
Standard and enhanced options of treatment of the transversal shear are available. This
concern interpolation of the shear strain, constitutive modeling and shear strain decomposition
along the cross-section depth as summarized in widow below.
Window 5-17: Options of transversal shear treatment
Treatment:
Context:
Constitutive treatment
of shear in case of layered model
Shear interpolation and

numerical integration rule
for elements of type:
Standard
Q4
(SHQ4,
SXQ4)
SRI
Enhanced
MITC/URI
Linear, shear behavior decoupled from in

plane nonlinear modeling
Shear included to nonlinear model,
Assumed shear strain

decomposition along
the
crosssection
depth in case of
layered model
Uniform shear correction factor = 5/6 is

used
Nonuniform, resulting from elastic stiffness ( )
URI Uniform Reduced Integration (2x2).

SRI Selective Reduced Integration 1x1 (for shear), 2x2 (for membrane & bending part).
MITC Mixed Interpolation of Tensorial Components approach.
()
S(z)
G(z)
(z) =
1
h
h/2
R
h/2
S(z)
G(z)
dz
= (z) ,
Zz
S(z) =
E(z)z dz.
h/2
Window 5-17
December 3, 2008

TM209

5.3.7
Formulation of the equilibrium equations are derived from virtual work principle (VWP), which in takes
general form:
find such that:
for any , u
Z
T dV
uT b dV = 0
In each detailed case, the particular form of VWP is as follows, using the notation:

T
m = xxL yyL xyL
membrane strains

T
membrane stresses
m = xxL yyL xyL

T
s = xzL yzL
shear strains

T
s = xzL yzL
shear stresses
T
= [m , s ]
full strains and stresses
V
b
Del
ss =
volume of the element

reference surface area
body forces

G 0
0 G

shear part of linear constituive matrix
with
shear correction factor (=5/6 for homogeneous section)
G Kirchhoff moduli
h element thickness.
homogenous section, elastic model
find such that:
for any , , u
Z
Z
Z
T
el
T
dV
+
hD
uT b dV = 0
m m
s
ss s
V
layered section, possible nonlinear model, standard treatment of shear (shear de-coupled from membrane
part in constitutive model)
find such that:
for any , , u
Z X
Z
Z
l T l
T
el
(m ) m hl d + s hDss s d uT b dV = 0
layered section, possible nonlinear model, enhanced treatment of shear (shear included in constitutive
model)
find such that:
for any , , u
Z X
Z
l T l
(m ) m hl d uT b dV = 0
Moreover, in the case of enhanced shear treatment for SHQ4,SXQ4 elements MITC ( Mixed Interpolation of
Tensorial Components ) approach is applied.
December 3, 2008

TM210

5.3.8
The formulas for element stiffness matrix K(55) and force vectors concern 5DOF component vectors
q(5) , f(5) with rotations /moments related to nodal tangent directions.
layered model with standard treatment of shear or homogenous section , elastic model:
NX
layer
Z
K(55) =
BT
m Dmm Bm jhilayer d d +
ilayer=1
NX
layer
Z
f(5) =
el o
BT
s hDss Bs j d d
BT
m m jhilayer d d +
ilayer=1
el o
BT
s hDss s j d d
where () concerns mean shear strains.

layered model with enhanced treatment of shear:
Z
K(55) =
NX
layer
T T
Bm , Bs
ilayer=1
Z
f(5) =
NX
layer
Dmm
Dsm
T T
Bm , Bs
ilayer=1
Dms
Dss
m
s

Bm
Bs

jhilayer d d

jhilayer d d
Bs = (zilayer ) o Bs
In all above integrals Gauss type integration is performed over the surface of the element as specified in
Window 5-17. Integration in the direction of element depth is based on user defined layer setting (layered
model) or on 2 point Gauss rule in case of homogenous section, elastic model.
5 DOF vectors and matrix are put into full 6 DOF ones and stabilizing terms are added to stiffness
matrix on the positions of 6th DOF (drilling rotations), in order to avoid singularity in case of co-planarity
of elements adjacent to a node.

Kab(55) 0
K=
,
a, b = 1, N en
drill
0
kab

f = fa(5) , 0 ,
a = 1, N en
Z
drill
kab = Na Nb Gh d,
= 106
Subsequently, rotational part is transformed to global coordinate system :

f G = Tf
KG = TKTT
where
T1 s
0
s
..
.
.
.
.
0
Ta
T=
..
..
.
.
0
s
0
s

I33 0
Ta =
,
0
[b
ea1 , b
ea2 , b
ea3 ]
December 3, 2008
0
..
.
0
..
.
TN en
a = 1, N en.

TM211

5.3.9
LOADS
Gravity, nodal and surface loads can be applied. Gravity corresponds to body forces. In a case of shell 2
node layer elements (option A) nodal forces can be applied to either master or slave nodes. Also surface
loads are specified on fictitious surface elements S Q4 which may be set on slave or master layer of nodes.
Equivalent nodal force vector consist of forces as well as moments, evaluated with use of generalized shape
function matrix N as:
Z
fp = NT pdV
Ve
Thermal loads can also be accounted for; thermal analysis is always performed on a continuum element and
then thermal is projected into shell material data at each layer.
December 3, 2008

TM212

5.3.10
SHELL ELEMENT RESULTS
Nodal results include 3 translations and 3 rotations. Element results include (at each numerical integration
point) membrane forces , moments and shear forces .
Window 5-18: Shell element stress resultants and sign convention
Shell element stress resultants and sign convention
Window 5-18
Window 5-18 indicates the sign convention. These results are evaluated, stored and printed into text file at
integration point and are referred to integration point fiber coordinate system and to the midsurface (not
to the reference surface). During the visualization phase users reference system must be defined. This is
documented in the reference manual, under postprocessing.
December 3, 2008

TM213
5.4
MEMBRANES

DISPLACEMENTS AND STRAINS FIELD
CONSTITUTIVE MODELS
Elements are designed to model different kind of soil reinforcement (solid phase) such as geotextiles, geogrids. In modern geo-technical practice there is a variety of different types of such a means. Membrane
elements are available in all 2D and 3D analysis types. Although membrane and truss elements use the same
DOF (at least for statics), note the difference between membrane and truss elements for all 2D analysis.
The difference concerns the possibility of using a wider list of constitutive models in the case of membrane
elements.
Window 5-19: Membrane elements
The family of membrane elements
Window 5-19
Note the significant difference between membrane and truss elements for the Axisymmetry analysis. For
the membrane elementary volume dV changes with the current point radius r, while for the truss elements
December 3, 2008

TM214
it remains constant, see the Window 5-1. The geometry of the element is set as iso-parametric surface
element (all T3/Q4) in 3D space or as 2 node linear element in 2D space. It posses one layer of nodes as it is required in statics. As element has only 1 layer of nodes it can not be used to model a flow in the
normal direction - (2 node layer necessary) -thus membrane elements can not be used to model impermeable
surface. When necessary to simulate the drainage or impermeable surface in the normal direction, one node
layer membrane element should be placed together with 2-node layer flow interface:
Window 5-20: Flow trough membrane elements
Modelling of flow trough membrane elements
Window 5-20
December 3, 2008

TM215
N Preface N N Structures N N N Membranes

5.4.1
3D case
Window 5-21: 3D membrane elements co-ordinate systems and mapping
Mapping from the reference element is constructed as:
x(, ) =
Na (, )xa ;
a=1,N en
Gauss point local base, by unified procedure see Appendix 5.5.3

Derivatives of any function f versus local directions xL ,yL are evaluated as:
f

x
f0
L = (J1 )T
,
f
f,
yL

with: J =
eT
1 x,
eT
2 x,
eT
1 x,
eT
1 x,
Window 5-21
In all 2D cases geometry setting for membrane element is identical with these for truss element, see Window
5-4.
December 3, 2008

TM216

5.4.2
DISPLACEMENTS AND STRAINS FIELD
The displacements within the element are interpolated from nodal values as:
u() =
Na ()ua
a=1,N en
The strains within the element (only in-plane are taken into account)
3D case:
xxL
= yyL =
xyL
uxL
xL
uyL
yL
uxL
uyL
+
yL
xL
u
xL
u
eT
yL
yL
u
u
eT
+ eT
yL
xL
xL
yL
eT
xL
2D analysis types:
F
Plane Strain:
xxL
= yyL
xyL

uxL
u
T
x exL x
L
L

= 0 =
0

0
0
Axisymmetry:
xxL
= yyL
=
xyL
uxL
xL
uxG
r
0
xL
uxG
eT
xL
The formulae relating strains any point within the element, with its DOF vector u may be put in the general
form:
= Bu =
N
en
X
Ba ua ,
a=1
Plane Strain:
ua = [ua , va ]
Ba () = [exLx DNa0 ,
exLy DNa0 ]
Axisymmetry:
exLx DNa0
Na
Ba () =
December 3, 2008
exLx DNa0
,
0

TM217

3D:
ua = [ua , va , wa ]T
Ba (, ) =
e1x Na0 xL
e1y Na0 xL
e1z Na0 xL
e2x Na0 yL
e2y Na0 yL
e2z Na0 yL
e1x Na0 yL + e2x Na0 xL
e1y Na0 yL + e2y Na0 xL
e1z Na0 yL + e2z Na0 xL
December 3, 2008

TM218

5.4.3
CONSTITUTIVE MODELS
Constitutive models used in membrane elements (solid phase) for all analysis types relate in- Plane Strains
= [xxL , yyL , xyL , ]T with membrane stress components = [ xxL , yyL , xyL , ]T in local directions of
membrane element. For some models, constitutive data concerning stiffness and strength must be given in
reference to the whole thickness of the element, in units [force/length]. The models are:
Elasto-plastic membrane (isotropic)
Data:
stiffness K [force/length], Poisson ratio , tensile

and compressive strength ft , fc [force/length]
Applications:
geotextile
Stress criterion:
1 ft , 2
where: 1 , 2
K
K
De =
0
Elasticity matrix:
> fc
principal stresses
K
K
0
K
2(1 + )
Elasto-plastic membrane (anisotropic)

Data:
stifness K11 , K22 , K12 [force/length] tensile

strength ft1 , ft2
[force/length] compressive
strength fc1 , fc2 [force/length] angle between
local element axis xL and an-isotropy 1-st axis x1 ,
evaluated from projection of a direction vector onto
element surface, see Appendix 5.5.2 .
Applications:
geo-grids

11
= T
22
K
K
11
12
T
De = TT
K12 K22
2

c s2 sc
where: T =
;
s2 c2 sc
s = sin , c = cos
Stress criterion:
Elasticity matrix:
Elasto-plastic fibre
Data:
elasticity moduleE, area per unit lengthA, tensile and compressive strengthft , fc , angle between local element axis xL and fibre direction x1 ,
evaluated from projection of a direction vector onto
element surface, see Appendix 5.5.2 .
Applications:
reinforcement layer
Note: In case of Axisymmetry, the model is suitable for modelling circumferential reinforcement ( = 90 ),
while for longitudinal one, usage of truss elements is recommended.
Stress criterion
ft , > fc
uniaxial stress in the fibre direction
Elasticity matrix:
De = Et tT

where: t = c2
December 3, 2008
.
s2
sc
T

TM219

Elasto plastic-plane stress member
Data:
elasticity moduleE, Poisson ratio , area per unit

lengthA, MenetreyWillam criterion data
Applications:
thin lining (steel, concrete, etc.)
Stress criterion
MenetreyWillam (plane stress)
Elasticity matrix:
standard plane stress
December 3, 2008

TM220

5.4.4
The contribution of the membrane elements to the virtual work principle expressing equilibrium of a system
may be put as:
find such that:
for any , u
Z
T AdS
with
uT p dS = 0

T
= xxL , yyL , xyL , ;
T
= [ xxL , yyL , xyL , ] ;

(
u=
[u
[u
v]
for Plane Strain, Axisymmetry
T
v w]
for Gen. plane strain, 3D
A thickness of the element (= 1 for models using membrane forces instead of stresses).
Integrals are taken over the surface of the element. S is an area attributed to the assumed computational
domain i.e. to the unit slice for the Plane Strain or to the whole circumference for the Axisymmetric case.
December 3, 2008

TM221

5.4.5
Stiffness matrix Ke and force vector f e of the membrane element are derived in a standard way from weak
formulation of a problem. It would require constitutive module matrix D as well as stress evaluation for
given strain increment. Numerical integration technique is used to evaluate integrals over the element with
inegration point number as shown in the table.
Element
Ngaus
M L2
1
M T3
1
M Q4
22
3D case:
Z1 Z1
K =
Z1 Z1
NX
gaus
B |J| Add =
f =
BT DB|J|A Wigaus
igaus=1
1 1
NX
gaus
BT DB|J| Add =
BT |J|A Wigaus
igaus=1
1 1
2D cases:
F
Plane Strain:
Z1 Z1
B DB k x, k Ad =
K =
Z1 Z1
B k x, k Add =
f =
NX
gaus
BT k x, k A Wigaus
igaus=1
1 1
F
BT DB k x, k A Wigaus
igaus=1
1 1
NX
gaus
Axisymmetry
Ke =
Z1 Z1
BT DB k x, k 2r Ad =
Z1 Z1
B k x, k 2rAdd =
f =
BT DB k x, k 2rA Wigaus
igaus=1
1 1
NX
gaus
1 1
December 3, 2008
NX
gaus
BT k x, k 2rA Wigaus .
igaus=1

TM222
N Preface N N Structures N N N Appendices

5.5
APPENDICES

SETTING THE DIRECTION ON SURFACE ELEMENTS
UNI-AXIAL ELASTO-PLASTIC MODEL
December 3, 2008

TM223

5.5.1
In order to deal with the frequently encountered situation where elements are connected to other elements
of the model by nodes, which are not the centroids of a cross section, offset transformation of element
displacement, forces and stiffness is introduced. The DOF of the element are placed on its master nodes
defining connectivity. Based on rigid body movement of the master, translation and rotation displacements
of the centroid being the slave node are evaluated. The above concerns beams as well as shell elements.
Window 5-22: Master-slave (offset) transformation
Displacements at the slave:
uS
= uM + M o
= M
where offset vector is used:

o = xSlave xMaster
o = {xi, yi, (zi)}T .

In turn, forces and moments evaluated initially at slave are moved to masters in a way preserving static
equivalency:
tM
mM
= tS
= mS + o tS
Expressing the above in a matrix form one can get:

displacement transformation :
qS = OqM
force transformation:
f M = OT f S
stiffness transformation:
as
KM
f M = OT f S = OT KS qS = OT KS OqM = KM qM
= OT KS O
Offset transformation matrix takes the form:

(all 2D cases)
1
O= 0
0
0
1
0
yi
xi
1
(3D case)
December 3, 2008

TM224
O=
1
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
zi
yi
1
0
0
zi
0
xi
0
1
0
yi
xi
0
0
0
1
Window 5-22
December 3, 2008

TM225

5.5.2
SETTING THE DIRECTION ON SURFACE ELEMENTS
A unified procedure of setting the direction on the surface element (shell, membrane, surface load, interface)
is proposed. The direction setting is independent from node order, orientation of the element, what is not
the case of local element base {e1, e2, e3} Moreover the same method of setting user defined coordinate
system to present the stress resultants in shell and membrane element is used in post-processor
Window 5-23: Direction on the surface elements
Distinction of the direction on the surface elements

Evaluation of the angle :
v0
cos
v (eT
3 v)e3
T 0
e1 v
eT v0
;
sin = 2 0
0
kv k
kv k
Note:
1. An error will be reported in the case when v is orthogonal to element surface, leading to k v0 k =0 .
2. xi is the direction closest to given v among all directions tangent to the element surface.
3. In a case when v is tangent to element surface, xi coincide with v, i.e. xi v.
Window 5-23
December 3, 2008

TM226

5.5.3
The following construction of the local base in the integration point, after T.J.R.Hughes5 , is used in all kind
of surface elements (i.e. shell, membrane, contact, fictitious for surface load elements). Moreover for these
elements, stress-type results which are stored in *.str ASCII file, are referred to defined bellow coordinate
system.
Note, that element local base depends on node numbering order and its orientation.
Window 5-24: Local base on a surface element
Setting of the local base on a surface element
{Q}(, ) = [e1 , e2 , e3 ]
2
2
e e
; e1 =
(a b); e2 =
(a + b);
e3 =
k e e k
2
2
where
e =
x,
;
k x, k
e =
x,
;
k x, k
a=
e + e
;
k e + e k
b=
e3 a
;
k e3 a k
Window 5-24
T.J.R.Hughes, Finite Element Method, Ed.PrenticeHall 1988
December 3, 2008

TM227

5.5.4
UNI-AXIAL ELASTO-PLASTIC MATERIAL MODEL
The uni-axial stress-strain relationship to be used commonly for truss, ring, beam, fibers is given as follows:
Window 5-25: Uniaxial elastoplastic material model
Uniaxial elasto-plastic material model
Window 5-25
December 3, 2008

TM228
N Preface
Chapter 6
INTERFACE
CONTACT
PILE INTERFACE
PILE FOOT INTERFACE
December 3, 2008

TM229
N Preface N N Interface
6.1
CONTACT OF SOLIDS AND FLUID INTERFACE
GENERAL OUTLOOK
DISPLACEMENT & STRAINS
CONSTITUTIVE MODEL
STIFFNESS MATRIX AND FORCE VECTOR
CONTRIBUTION TO CONTINUITY EQUATION
December 3, 2008

TM230
N Preface N N Interface N N Contact

6.1.1
GENERAL OUTLOOK
Window 6-1: Interface elements: General remarks

Mechanical contact as well as flux (of fluid, heat, humidity) through the surface between two bodies is
modelled by finite element discretization of the interface between them. The interface elements use nodes
belonging to the FEmesh of both adjacent solids with assumed invariable topology (small displacement
theory). Moreover the compatibility of the initial positions of nodes is required (this is assured by pre
processing tools, see Interface option).
Interface (contact) elements between 2 adjacent bodies
Interface element geometry is based on the iso-parametric mapping from the reference element. As nodes of
both layers of element are assumed to occupy the same position :
x() =
Ni ()xi .
i=1,..,N en
Family of interface elements
Window 6-1
December 3, 2008

TM231

Window 6-2: Interface: Mechanical contact
As an additional feature, for mechanical contact, the displacements continuity option is introduced. This is
related to interface element status.
Status CONTACT: The interface elements reproduce the force action between the two bodies based on the
relative displacements of the interface nodes. The elastoplastic friction model is used, allowing for sliding
and separation, while the elastic properties of the interface impose penalty constraints multipliers excluding
penetration, see Window 6-4 for the details.
Status CONTINUITY(u,p,T) or CONTINUITY(u,T): Displacements continuity across the interface is
enforced. Nodes on both sides of the interface share the same kinematical DOFs.
Window 6-2
Window 6-3: Fluid Phase
In case of Flow or Deformation+Flow analysis mode, interface elements may posses pressure DOF at nodes
of both layers. Interface elements can be used to model following situations:
Fully permeable interface
Pressures continuity across the interface is enforced. Nodes on both sides of the interface share the same
pressure DOFs.
Impermeable interface
No flow takes place in the direction normal to the interface (no flux qn = 0 condition on both faces is
imposed). Resulting pressures on both interface faces will be (in general) discontinuos. In this case the
interface element does not contribute to the equation system.
Permeable interface
Both isotropic and anisotropic flow conditions can be handled.
Window 6-3
December 3, 2008

TM232

6.1.2
Displacements at each layer of the interface L1 and L2 are interpolated using standard approach, from given
nodal displacements u:
X
u L1 () =
Ni ()ui
L2
i=1,..,N en
i=N en+1,..,2N en
. The generalized strains at the integration point of the interface element are mutual displacements of both
layers transformed to the local basis {t, n} of the element. In 3D case local base on the element surface is
created according to unified procedure given in Appendix, Window 5-24.
Generalized strains in contact element

The relation between the above generalized strain and nodal displacements may be put in unified form,
() = B()u =
2N
en
X
Bi ()ui
i=1
with B matrix given as:

T
B()= T() [Ni ()IN DOF , Ni ()IN DOF ] ,
i = 1, N en
where:
T()transormation matrix such that UT N = T()u
IN DOF unit matrix, N DOF is displacement component number per node
2D cases
Plane Strain, Axisymmetry:
ui = [ui , vi ]
() = [ut , un ] = [utL2 utL1 , unL2 unL1 ]

x,
y,
c, s
T() =
, c= q
, s= q
s, c
2
2
x, +y,
x,2 +y,2
3D case:
T
ui = [ui , wi , vi ]
T
(, ) = [ut1 , ut2 , un ] = [ut1L1 ut1L1 , ut2L2 ut2L1 , unL2 unL1 ]

T
.. ..
T(, ) = e1 .e2 .e3
Moreover, initial gap may be accounted for while evaluating element strains.
December 3, 2008

TM233

6.1.3
CONSTITUTIVE MODEL
Constitutive behavior of the interface is described in the terms of:

generalized strains , evaluated from nodal displacements of the interface
effective stresses 0 , submitted to appropriate stress criterions resulting from
1. cohesive Mohr - Coulomb condition
2. no-tension condition.
For Plane Strain and Axisymmetry number of stress components N stre = 2, while for 3D analysis N stre =
3. In the case of analysis mode Deformation+Flow, the concept of effective stress is used taking into account
pressures p and saturation ratio S:
N stre = 2

0
0
= +S
p
N stre =3
0
= 0 + S 0
p
flow rule with a flow potential in the form analogous to Mohr-Coulomb yield function allowing for nonassociative flow rule in the case when 6= ,
constitutive matrix D.
Both 0 and D are evaluated within the frame of perfect multisurface elastoplasticity theory, with components related to the plane of the interface and its normal.
Formulation of both cases of contact constitutive law is given in the Window 6-4.
The trial stresses are evaluated as:
= n + Del
using the previous stress n , strain increment , the elastic (penalty) interface stiffness Del .
The elastic stiffness Kn should be large enough to prevent significant penetration in the case of compression,
but can not undertake arbitrarily large values as it might spoil conditioning of the resulting FE equation
system and lead to difficulties in obtaining convergence of the solution. Estimation of penalty stiffness is
done as follows in the Window 6-5.
December 3, 2008

TM234

Window 6-4: Constitutive law of frictional contact
Analysis type:
Nstre
Data:
Plane Strain, Axisymmetry

2
c cohesion
friction angle
dilatancy angle (0 )
= [ , n ]T
with:
the shear interface stress

n the stress normal to the
interface
Stresses:
Stress conditions
Flow potential
a=
Gradients
b=
Slip activated when:

F1 ( 0 ) = + tan() 0n c > 0 if 0
F2 ( 0 ) = + tan() 0n c > 0 if < 0
No tension (cutoff) activated when:
F3 ( 0 ) = 0n > 0
Graphic presentation:
Q1 ( 0 ) = + tan() 0n c > 0 if 0
Q2 ( 0 ) = + tan() 0n c > 0 if < 0
F
,
Q
:

T
a1/2 = 1, tan
,

T
b1/2 = 1, tan
.

Elasticity matrix:
Del =
Kt
0
0
Kn
3D
3
= [ 1 , 2 , n ]T
with:
1 , 2 the shear interface stresses
in local directions e1 , e2
n
the stress normal to the
interface
p
F1 ( 0 ) = 21 + 22 +
tan() 0n c > 0
F2 ( 0 ) = 0n > 0
p
Q1 ( 0 ) = 21 + 22 +
tan() 0n c > 0
h
iT
2
1
,
, tan
a1 =
h
iT
2
1
,
, tan
b1 =
p
= 21+ 22
Kt 0
0
Del = 0 Kt 0
0
0 Kn
Window 6-4
December 3, 2008

TM235

Window 6-5: Estimation of the penalty stiffness and permeability
This is done automatically (default) based on the following algorithm:
1. Find neighbouring elements ( active at the current time tn + 1)
2. For neighbouring elements find the maximum size in the direction normal to the interface
Penalty stiffness estimation
1. Estimate normal stiffness as:

Kn = min
E1 E2
,
h1 h2
A
N eq
2. Set tangent stiffness Kt of the interface as:

Kt = 0.01Kn
3. In the case of Flow or Deformation+Flow analysis mode estimate penalty permeability:

s X
B
k1 k2
2 k (S(p))
kf =
min
kii
,
where km =
r
h1 h2
w N eq
i=1,N dim
taking into account permeability multiplier kr dependent on current saturation S = S(p)

In the above:
A, B
N eq
arbitrary factors(default A = 104 , B = 103 )set by numerical experience

total equation number in the system
precision (machine dependent small number)
Parameters adopted under points 3, 4, 5 may be multiplied by user-defined factors.

Window 6-5
December 3, 2008

TM236

Window 6-6: Stress point algorithm
Depending on the trial stress , 3 different types of the behavior will be modelled:
sticking,
sliding,
separation.
The algorithm of stress and constitutive matrix evaluation is as follows:
Stress-point algorithm
if n > 0 then
separation:
n+1 = 0
D=0
else
if
N stre=2
}|
{
z
F1 ( ) 0F2 ( ) 0then
if
sticking
n+1 =
D = Del
else
sliding (F1 ( ) > 0) :
n+1 = n + Del ( b)
Fi ( )
= T
a Del b
(Del b) : (Del a)T
D = Del
aT Del b
Fi
Qi
where : a =
, b=
end if
end if
Note that stress return for the sliding case is performed with a one step cutting-plane procedure. This is
due to the linear form of the yield function and flow potential in 2D case (N stre = 2) as well as possible
radial return for the 3D case (N stre = 3).
Window 6-6
December 3, 2008

TM237

6.1.4
These are evaluated as: 3D case:

Z1 Z1
B DB | J |dd =
K =
B DB | J |Wigaus
igaus=1
1 1
Z1 Z1
NX
gaus
B | J |dd =
f =
NX
gaus
B | J |Wigaus
igaus=1
1 1
2D cases:
Plane Strain
Z1
B DB k x, k d =
K =
Z1
B k x, k d =
f =
B DB k x, k Wigaus
igaus=1
NX
gaus
NX
gaus
B k x, k Wigaus
igaus=1
Axisymmetry
Z1
B DB k x, k 2rd =
K =
Z1
B k x, k 2rd =
f =
B DB k x, k 2rWigaus
igaus=1
NX
gaus
NX
gaus
B k x, k 2rWigaus
igaus=1
with constitutive matrix D and stresses being the result of the point level algorithm, see Window 6.5.4 .
Note that in the case of Analysis mode Deformation +Flow, the total stresses including pressure is used
to evaluate element forces. In order to avoid oscillatory patterns of normal stress, integration is performed at
nodes and not standard Gauss points. In addition normal vectors at each node are averaged from all adjacent
elements. The B matrix is evaluated as follows:
Bt ( igaus )
s
B( igaus ) =
Bn ( igaus )
December 3, 2008

TM238

6.1.5
A common problem of the standard penalty approach used for problems of contacting elasto-plastic media
is that the resulting overpenetration is too large and contact stresses may be underestimated. Usage of high
values of penalty stifnesses usually results in loss of convergence and oscillatory contact stress distribution.
In the case of soil-structure contact interaction the stress resultants may be underestimated as well. To
handle this deficiency an Augmented Lagrangian Approach can be used. In the current contact formulation
(segment to segment approach) each time the state of the global static equlibrium is achieved the contact
resulting normal stresses are memorized, penalty stiffness is increased (by default through factor of 2). The
Augmented Lagrangian Approach is summarized in window given below.
Window 6-7: Augmented Lagrangian Approach
1. initialize: = 0,
(=0)
f =1
2. at each integration point at contact element set:
(=0)
nN +1 = nN
3. solve: FextN +1 Fint (uN +1 ) = 0 assuming that that the trial normal stress at each integration point of
()
nN +1 +() f kn nN +1
contact element is computed as: trial
nN +1 =
4. at each integration point of contact element check overpenetration: |nN +1 | >TOL (?)
5. if overpenetration is too large (at any integration point) perform augmentation procedure:
6. set: = + 1
7. increase penalty parameter:
8. if
()
f > fmax set:
()
()
f =(1) f g
(g = 2 by default)
f = fmax
9. if <MAX-AUGMENTATIONS go to step (2)

Window 6-7
December 3, 2008

TM239

6.1.6
CONTRIBUTION TO CONTINUITY EQUATION
In the case of active fluid phase interface elements contribute to matrix H and flux vector Q (see Section
4.1.2) with the following:
Z
HInterface =
N
N

kf
NT , NT
QInterface = HInterface tw
where:

()
NT = N1 () . . . NN en
shape function vector

t
,
.
.
.
t
tT
=
nodal pressure vector.
w1
wN
en
w
The integration technique analogous to the one used for Stiffness matrix and element force vector evaluation
is used.
December 3, 2008

TM240
6.2
PILE CONTACT INTERFACE
GENERAL OUTLOOK
CONSTITUTIVE MODEL
December 3, 2008

TM241

6.2.1
GENERAL OUTLOOK
Window 6-8: Pile interface element: General remarks

Pile interface element is used to model frictional contact between pile (beam elements) and continuum in
which pile is embedded. This element consists of the two linear segments called master and slave respectively.
Nodal points of the master segment are linked to the continuum via Nodal link option while slave segment
coincides with beam element. This element allows to model relative movement of the pile and continuum
while the interface behavior is controlled by the standard Coulombs friction law. This element assumes full
displacement continuity, enforced by the penalty method, in the plane perpendicular to the pile axis.
x
master
slave
z
Pile as set of beam elements
Window 6-8
December 3, 2008

TM242

6.2.2
Window 6-9: Pile interface: Generalized strains

Displacements at slave and master segments are interpolated by using standard approach, from given nodal
displacementsPat master segment umaster
and slave segment uslave
:
i
i
master
master
u
() =P i=1..N en Ni ()ui
uslave () = i=N en+1..2N en Ni ()uslave
i
The generalized strains at any integration point of the interface element are understood as relative displacements of both segments transformed to the local element basis {t, n1 , n2 }. The local n1 and n2 axes are
created in a random way due to axial symmetry. If t axis is parallel to the global y axis (standard situation)
then local n1 and n2 axes are parallel to the global x and z axes.
u 2
4
master
slave
n2
n1
u1
The relation between generalized strains and nodal global displacements:

() = B()u =
2N
Xen
Bi ()ui
i=1
where B is defined as:

T
B() = [Ni ()T(), Ni ()T()]
for i = 1..N en
local
T()transformation matrix such that u
= T()uglobal

T
master
master T
() = [ut , un1 , un2 ] = uslave
umaster
, uslave
, uslave
t
t
n1 un1
n2 un2
Transformation matrix is composed of unit vectors et , en1 , en2 expressed in global coordinate system
T
T() = [et , en1 , en2 ]
Window 6-9
December 3, 2008

TM243

6.2.3
CONSTITUTIVE MODEL
Window 6-10: Pile interface: Constitutive aspects and stress return algorithm
Constitutive behavior of the pile interface is described in terms of:
Generalized strain t , evaluated from nodal tangential displacements of the interface
Effective stress 0n estimated in an explicit manner from the continuum in which pile is embedded; note
that displacement continuity along n1 and n2 directions is always preserved
Frictional Colulombs law: F ( 0n , ) = | | + 0n tan() c
Constitutive elastic matrix
Kt
De = 0
0
0
Kn
0
0
0
Kn
Stress evaluation consists of the following steps:

Compute trial averaged shear stress trial
N +1 = N + Kt ut
If 0n > 0 set N +1 = 0 and
Dep
0
= 0
0
0
Kn
0
0
0
Kn
If 0n 0 check plasticity condition: F ( 0n , )

F
F
ep
If F ( 0n , ) < 0 (sticking) then set N +1 = trial
= De
N +1 and D
If F ( 0n , ) > 0 (sliding) then set N +1 = ( 0n tan() + c) sign( trial

N +1 ) and
0
0
0
0
Dep = 0 Kn
0
0
Kn
Compute the two remaining stress vector components n1 , n2 which are concerned with the enforced
displacement continuity in n1 and n2 directions
n1 N +1 = n1 N + Kn un1
n2 N +1 = n2 N + Kn un2
T
Compose stress vector = { N +1 , n1 , n2 }

Remarks:
1. The elastic stiffness Kt should be large enough to prevent significant loss of displacement continuity in
the tangential direction for case of full sticking
2. The normal elastic stiffness Kn should also be large enough to prevent loss of continuity of the displacement
fields in the plane perpendicular to the pile axis; however, too large values for Kn and Kt may lead to the
lack of convergence, oscillations.
3. Dilatancy angle is not meaningful for this type of the interface
4. Estimation of Kn and Kt factors follows the procedure described in Win.(6-5)
Window 6-10
December 3, 2008

TM244

6.2.4

e
K = 2 r
pile
Z1
B DB | J |d = 2 r
pile
f = 2 r
pile
Z1
BT DB | J |Wigaus
igaus=1
NX
gaus
BT | J |d = 2 rpile
NX
gaus
BT | J |Wigaus
igaus=1
Remarks:
1. rpile is a radius of pile (reinforcement)
December 3, 2008

TM245
6.3
PILE FOOT CONTACT INTERFACE
GENERAL OUTLOOK
CONSTITUTIVE MODEL
December 3, 2008

TM246

6.3.1
GENERAL OUTLOOK
Window 6-11: Pile foot interface element: General remarks

Pile foot interface element is a simple node to node interface element which is used to put limits on the tensile
and compressive stresses in the contact zone of foot of the pile and underlying subsoil. This element consists
of the two nodes called master (it is linked to the continuum via Nodal link option) and slave ((pile) beam
element endpoint). This element allows to model separation of the foot of the pile and soil during pull out
and limited compressive strength. The latter effect is important as due to coarseness of the continuum finite
element mesh in the zone of the foot of the pile bearing capacity of the pile could be overestimated. This
interface controls relative movement of the pile and continuum only in the axial pile direction (xL ); in the
two remaining directions (yL , zL ) full continuity is enforced.
xL
x
Slave node
Master node zL
yL
Window 6-11
December 3, 2008

TM247

6.3.2
Window 6-12: Pile foot interface: Generalized strains

The generalized strains in the interface are understood as relative displacements of both nodes transformed
to the local element basis {xL , yL , zL }. The local yL and zL axes are created in a random way due to axial
symmetry. If xL axis is parallel to the global y axis (standard situation) then local yL and zL axes are parallel
to the global x and z axes.
u 2
xL
master
slave
zL
u
n1
1
3
yL
The relation between generalized strains and nodal global displacements:
= Bu =
2
X
Bi ui
i=1
where B is defined as:

T
B = [1 T, 1 T]
Ttransformation matrix such that ulocal = Tuglobal

T

T
master
master
master T
= [uxL , uyL , uzL ] = uslave
, uslave
, uslave
xL uxL
yL uyL
zL uzL
Transformation matrix is composed of unit vectors exL , eyL , ezL expressed in global coordinate system
T
T = [et , en1 , en2 ]
Window 6-12
December 3, 2008

TM248

6.3.3
CONSTITUTIVE MODEL
Window 6-13: Pile foot interface: Constitutive aspects and stress return algorithm
Constitutive behavior of the pile foot interface is described in terms of:
Generalized strain x , evaluated from nodal displacements of the interface
Stick-separation law: qc x qt ; qt is a tensile bearing capacity (default qt = 0 kPa) and qc is the
compressive bearing capacity (can be found in standard codes for pile design) (default qc = 1038 kPa)
Constitutive elastic matrix
Kn
De = 0
0
0
Kn
0
0
0
Kn
Stress evaluation consists of the following steps:

Compute trial normal stress x trial
N +1 = xN + Kn ux
If x trial
N +1 > qt set xN +1 = qt and
Dep
0
= 0
0
0
Kn
0
0
0
Kn
0
Kt
0
0
0
Kt
If x trial
N +1 < qc set xN +1 = qc and
Dep
0
= 0
0
ep
trial
= De
If qc < x trial
N +1 < qt set xN +1 = x N +1 and D
Compute the two remaining stress vector components xz , xy which are concerned with the enforced
displacement continuity in yL and zL directions
xy N +1 = xy N + Kn uy
xz N +1 = xz N + Kn uz
Compose stress vector = { x , xy , xz }
Remarks:
1. The elastic stiffness Kn should be large enough to prevent significant over-penetration in case of full
sticking
2. Too large values for Kn may lead to the lack of convergence (oscillations)
3. Estimation of Kn follows the procedure described in Win.(6-5)
Window 6-13
December 3, 2008

TM249

6.3.4

2
Ke = rpile BT DB
2
f e = rpile BT
Remarks:
1. rpile is a radius of pile
December 3, 2008

TM250
N Preface
Chapter 7
GEOTECHNICAL ASPECTS
TWOPHASE MEDIUM
EFFECTIVE STRESSES
SOIL PLASTICITY
INITIAL STATE
SOIL RHEOLOGY
ALGORITHMIC STRATEGIES
December 3, 2008

TM251
N Preface N N Geotechnical aspects

7.1
TWO-PHASE MEDIUM
In this section, an attempt is made to relate modelling parameters to geotechnical aspects. The soil is modeled
as a two-phase medium, this means that equilibrium of the medium requires the solution of a coupled system
of differential equations where one set of equations represents the equilibrium of the solid and the second
set of equations represents the continuity of the fluid flow. Both sets include coupling terms. Drained and
undrained conditions are limiting cases of particular interest. The corresponding boundary conditions are
shown in Window 7-1
Drained conditions
Boundary conditions are such that, in the long term, the local stress is carried by the skeleton ( = 0 ).
Undrained conditions
When boundary conditions and material properties are such that no fluid motion relative to the solid is
possible, the condition is undrained and the medium behaves in an essentially incompressible manner.
Window 7-1: Drained and undrained condition
Drained condition: pF = pF on p (left); Undrained condition q = 0 on v (right)
Window 7-1
December 3, 2008

TM252

7.2
EFFECTIVE STRESSES
Effective stresses allow a unified approach to the analysis of the drained and undrained conditions, this
concept is extended here to account for partially saturated media. Let:
= 0 + pF S
where 0 is the effective (grain to grain stress) , pF the interstitial pressure, and S the saturation ratio.
December 3, 2008

TM253

7.3
SOIL PLASTICITY
DRUCKER-PRAGER VERSUS MOHR-COULOMB CRITERION

CAP MODEL
DILATANCY
December 3, 2008

TM254
N Preface N N Geotechnical aspects N N N Soil plasticity

7.3.1
DRUCKER-PRAGER VERSUS MOHR-COULOMB CRITERION
It is common to describe soils as elasticperfectly plastic MohrCoulomb materials. A smooth MohrCoulomb

criterion described earlier is available in this program. If a DruckerPrager criterion with cap closure is
preferred, the size of the DruckerPrager criterion can be adjusted to match the MohrCoulomb criterion.
This is illustrated in Window 7-2. The various matching options are derived in the theoretical section; for
plane strain, the most meaningful matching for an ultimate load analysis is the matching of collapse loads.
The most important sizeadjustments are summarized in Window 7-2.
Window 7-2: Matching Mohr-Coulomb and Drucker-Prager criteria
Deviatoric sections of MohrCoulomb (MC) and DruckerPrager
Matching
External apices
Internal apices
sin
;
3D
Planestrain collapse
Elastic domain
(plane strain t = 0.5)
a , k
:
:
:
:
a
2 sin
3(3 sin )
2 sin
3(3 + sin )

q
D = a sin + 1 3a2
sin
3
k
6c cos
3(3 sin )
6c sin
3(3 + sin )
c cos D1
c cos
parameters of the DruckerPrager criterion

friction angle
cohesion
angle of nonassociativity (in doubt, use default value)
Window 7-2
Related Topics
THEORY: M-C VERSUS D-P
December 3, 2008

TM255

7.3.2
CAP MODEL
The cap model accounts for nonlinear behavior of soil under dominant volumetric (pressure) stress. The
initial size of the cap is derived from the oedometric test, it is determined by the preconsolidation pressure.
Once the yield point (i.e. the cap) is reached by the stress, hardening takes place.
Hardening results from the reduction of the void ratio under increasing pressure, it is again controlled by the
oedometric test.
Related Topics
CAP MODEL
CAP MODEL - STRESS POINT ALGORITHM
OEDOMETRIC TEST
December 3, 2008

TM256

7.3.3
DILATANCY
Two alternative dilatancy parameters can be accommodated by the material models proposed earlier. The
first parameter, d, relies on the availability of experimental results. The second one, , assumes the empirical
knowledge of , by analogy with the friction angle .
Dilatancy parameter d
Volumes changes in soils, in the plastic regime, are conveniently described as follows; let d be the dilatancy
d=
dpV
Q/p
rp
=
p =
dD
Q/q
rq
where dpV is the volumetric plastic strain increment and dpD is the plastic deviatoric strain increment.
Q is the plastic potential, rp is the norm of the volumetric plastic flow component and rq , the deviatoric
one. dpV , dpD will usually be retrieved from a tri-axial test and rp /rq will be derived from the plastic
model.
Dilatancy extraction from triaxial test
Experimental results from a triaxial test can be plotted as follows:
V = 1 + 23
D = (2/3)(1 3 )
P hence: dV = d
In the plastic regime dV
= dP
V and d = d
dD
Dilatancy with Drucker-Prager plasticity
The plastic potential is in this case:
p
Q = a I1 + J2
then,
Q I1
Q
=
= 3a
p
I1 p
Q
Q J2
rq =
=
= 1/ 3
q
J2 p
rp =
and finally,
d = 3 3a
Given d from the experiment, a can be found for the material model.
Dilatancy with Mohr-Coulomb plasticity
Plastic flow is again governed by a Drucker-Prager type surface in the program, even when yield is governed
by a Mohr-Coulomb criterion. The rate of non-associativity generated by a given (experimental) d is
therefore dependent on the size adjustment.
In 2D, for size adjustment option matching plane strain collapse load , given , c and d, one gets:
December 3, 2008

TM257
a = d/3 3
q
1
a = sin [a sin + 1 3a2 ]1
3
q
k = C cos [a sin + 1 3a2 ]1
For a 3D situation in which defines the size adjustment with respect to external and internal matching,
we get
a
=
d/3
3
a = (1 )[2 sin /( 3(3 sin ))] + [2 sin /(

3(3 + sin ))]
k = (1 )[6C cos /( 3(3 sin ))] + [6C cos /( 3(3 + sin ))]
NB: a should be such that (0 < a < a ).
Alternative experimental results can also be used in order to define d, as illustrated next.
Alternative experimental representations
d=
3
2
1 + 2d0
1 d0

d=
d00
d00
1
3
Dilatancy angle
Assuming a plastic potential given by Q = | | n tan . The value of can result from empirical
knowledge or be retrieved from experiments as before, using the following formula,
tan =
pn
pt
where pn and pt are normal and tangential plastic strain increments or alternatively:
sin =
pV
p1 + p3
=
pmax
p1 p3
Application with Drucker-Prager plasticity

If is specified, then the same size adjustment as for will be assumed in order to retrieve a . For
example, assuming a 2D situation for which , C, and size adjustment plane strain collapse are
specified, then introduced into the associated flow option yields:
q
a = tan / 9 + 12 tan2
can then in turn be introduced into formulas in order to retrieve k and a .
In a 3D situation with f , k, and specified, a can be retrieved from formula, replacing ; a and
k are, in this case independent of . These operations are, of course, done automatically and hidden to
the user.
Dilatancy with smooth Hoek-Brown criterion
Flow options available on Hoek-Brown criterion include:
December 3, 2008

TM258

F
Deviatoric, corresponding to incompressible flow.
Tensile meridian. The flow will be radial in the deviatoric plane and follow the normal to the tensile
meridian in the meridian plane.
c prescribed (Hoek-Brown flow).

c is defined as the angle of dilatancy at failure under uniaxial compression c = arctan(d/d) and

ft
1
arctan(
) < c < arctan( ) .
2fc
2
1 dI1
Alternatively c = arctan[ ; the resulting dilatancy will vary with loading paths: from coinci6 d J2
dence with the loading path under uniaxial tension to c under uniaxial compression.
Following table contains indicative values of dilatancy characteristics
Clay
Sand
Gravel
Rock
Concrete
December 3, 2008
/
0
0.67 1
-
d
0
-
c , []o
o
4 35o

TM259

7.4
INITIAL STATE

STATES OF PLASTIC EQUILIBRIUM
INFLUENCE OF WATER
December 3, 2008

TM260
N Preface N N Geotechnical aspects N N N Initial state

7.4.1
K0 is by definition the ratio of horizontal effective stresses to vertical effective stresses

K0 =
01
.
02
The coefficient can be determined from triaxial experiments, measured with a pressuremeter or evaluated
using approximate formula. Commonly used formula are given in Window 7-3.
Window 7-3: Coefficient of earth pressure at rest, K0
Normally consolidated soil (NC)

Overconsolidated soil (OC)
Confined elastic medium
(K0 )NC = 1 sin

(K0 )OC = (K0 )NC OCR
= sin
K0 = 1
[JAK48]
Window 7-3
Overconsolidation ratio (OCR)
The overconsolidation ratio OCR is obtained from an oedometric test. It is maximum close to soil surface
and tends to 1 a depth. The identification of OCR is illustrated in
Window 7-4.
Given the void ratio, the water content, a preconsolidation pressure can be associated with each vertical
stress and the corresponding overconsolidation ratio can be computed as illustrated in Figs 7-4, 7-4.
December 3, 2008

TM261

Window 7-4: Overconsolidation ratio OCR
Effective stress at depth z :
0z = 0 z,
initial void ratio:

e0 =
1+w

1
where:
s : unit weight of solid particles
w : water content ratio
Oedometric test
Vertical stresse and preconsolidation pressure as a function of depth

(right).Overconsolidation ratio as a function of stress (left)
Window 7-4
December 3, 2008

TM262

7.4.2
STATES OF PLASTIC EQUILIBRIUM
The section deals with the case of semi-infinite soil mass with horizontal or inclined surface subjected to
gravity load. Coefficients of horizontal pressure K fulfilling different stress criterion are investigated.
MOHR-COULOMB CRITERION
DRUCKER-PRAGER CRITERION
December 3, 2008

TM263
N Geotech. asp. N N Initial ... N N N States of plastic ...

7.4.2.1
MOHR-COULOMB MATERIAL
The plastic equilibrium at depth h of a semiinfinite soil mass with horizontal surface, considered as a
MohrCoulomb material, subjected to gravity loading is characterized by two circles in a Mohr diagram
(Window 7-5).
The two Mohr circles correspond to Rankine states. The small circle corresponds to the active Rankine state,
the large circle to the passive state.
A cohesionless material is considered first. Note that the principal stress orientation coincides with axes 1
(horizontal) and 2 (vertical). The plastic stress state at depth h is
2 = h
(vertical)
1 = 3
(assumption)
1 = A = KA h
1 = p = KA h
(horizontal, active state)

(horizontal, passive state)
Stress states such that,

1 = K0 h
with KA < K0 < KP
are elastic. K0 is the earth pressure coefficient at rest.

From geometrical considerations equations 1 and 2 (Window 7-5) can be derived for KA and KP are plotted
and provide a useful way to define possible horizontal stress states, given the friction angle. For an elastic
perfectly plastic material, no stress state outside of these limits is tolerable.
December 3, 2008

TM264

Window 7-5: Rankine state
Mohr diagram of the plastic state of a semi-infinite medium: Cohesionless material (left),
Cohesive material (right)
Earth pressure coefficient at rest K0 , for a cohesionless semi-infinite soil mass with
horizontal surface
Active state:

1 sin
1
2
KA =
= tan 45
=
1 + sin
2
N
(1)

1 + sin
= tan2 45 +
= N
1 sin
2
(2)
Passive state:
KP =
Window 7-5
Similar expressions for the horizontal stress are derived in Window 7-5 for a cohesive material, from
Window 7-6 and geometrical considerations.
December 3, 2008

TM265

Window 7-6: Rankine states for a Cohesive MohrCoulomb material
Active state

1 = KA
C
2 +
tan
or
1 = KA 2 (1 KA )
Passive state

1 = KP
2 +
C
tan
C
.
tan
or
1 = KP 2 (1 KP )
C
tan
C
tan
C
.
tan
Window 7-6
Given the friction angle, KA and KP can be read from Window 7-5. Introducing the cohesion C, f and KA
or KP into the expressions for 1 yields limiting values of the horizontal stress 1 . These values define the
elastic range. The same discussion holds when the semifinite soil mass is loaded on its surface by a uniform
load q. In this case 2 = h is replaced by 2 = (h + q).
The presence of a water table can be accounted for similarly. Plastic states are defined in terms of effective
stresses. The limiting values KA and KP of K0 are therefore the same for a saturated medium as for a dry
medium.
If the water table is located at a depth d, the upper layer can be viewed as a surface load on a saturated
medium, and KA and KP are again the same.
Particular situations
If the semiinfinite soil mass is limited by an upper surface inclined at an angle the Mohr diagram is
given in Window 7-7.
The stress state at depth h on a plane inclined at angle can be calculated as:
= h cos2
= h sin cos .
The corresponding point in the Mohr diagram is Z and the Mohr circles corresponding to active and passive
states can be constructed. The values obtained for KA

and KP (K = 11 /h) are reported in Window 7-7
as functions of the friction angle and the angle of the slope .
The case of a semiinfinite soil mass limited by an upper surface inclined at an angle , submerged by
, KP apply, associated with 0 .

water can be solved similarly and the same values of , KA
The case of an infinite slope under conditions of seepage flow leads to the following stress state on a plane
inclined at angle :
= h cos2
= sat h sin cos .
A Mohr circle can be drawn as in the previous case and the corresponding limiting states can be calculated.
The values of KA
and KP can again be read from Window 7-5 , with tan replaced by
tan = sat tan .

b
Similar derivations can be performed for the case of a cohesive material leading to the same expressions for
KA
and KP if is replaced by 0 such that:

tan 0 = tan
.
+ C/ tan
For seepage flow replaces .
December 3, 2008

TM266

Window 7-7: Rankine states, inclined surface
Mohr diagram for a semiinfinite soil mass with inclined surface. Cohesionless soil
Earth pressure coefficient at rest K0 , for a cohesionless semiinfinite soil mass with surface
inclined at angle

p
1
KA
= cos2 cos cos2 cos2 cos2
(1 sin cos )

p
1
KP = cos2 cos cos2 cos2 cos2
(1 sin cos )
with
(1)
(2)

sin
= arcsin
;
sin

sin
= arcsin
+
sin

The horizontal stresses corresponding to active and passive states are expressed as:
active state:
( 11 )A = KA
h
December 3, 2008
C
tan

TM267

passive state:
( 11 )P = KP h
C
tan
where KA
and KP are read from Window 7-7.
Window 7-7
December 3, 2008

TM268

7.4.2.2
DRUCKER-PRAGER MATERIAL
Rankine states can be derived similarly for a DruckerPrager cohesionless material, with the stress state
defined previously. With the assumption 3 = 1 , equations (DP1) Window 7-8 are derived. With the
assumption 3 = 0.5 ( 1 + 2 ), relations (DP2) result.
The results obtained for KA and KP as function of are reported in Window 7-8. It is observed that
the elastic ranges of MohrCoulomb and DruckerPrager materials coincide, for the assumption that 3 =
0.5 ( 1 + 2 ).
As before, the presence of a surface load or of a water table can be accounted for, as for the MohrCoulomb
case.
Window 7-8: Rankine states for Drucker-Prager material
Earth pressure coefficient at rest K0 for cohesionless material, MohrCoulomb (MC),

DruckerPrager 3 = 1 (DP1) and Drucker-Prager with 3 = 0.5 ( 1 + 2 ) (DP2)
3 sin
,
KA =
2 sin + 3
KA =
1 sin
,
1 + sin
3 sin
KP =
2 sin 3
KP =

(1)
1 + sin
1 sin
(2)
Window 7-8
December 3, 2008

TM269

If the semiinfinite soil mass, always considered as a cohesionless material, is limited by an upper surface
inclined at an angle , values of KA

and KP can again be derived for both case, i.e.:
KA
= cos2
with
1 D tan
,
1 + D tan
KP = cos2
1 + D tan
1 D tan
!#
6 3 sin
D = tan arccos sin
if 3 = 1
3 3 sin

sin
D = tan arccos
if 3 = 0.5 ( 1 + 2 ) .
sin
"
The obtained values of KA

and KP (K = 11 /h) are plotted in Window 7-9 as functions of the friction
angle and the angle of slope , the same assumptions for 3 as before are made.
It is noted again that the elastic ranges of MohrCoulomb and DruckerPrager materials, with the assumption
of 3 = 0.5 ( 1 + 2 ), coincide (see Window 7-8 and Window 7-9).
Similar derivation can be performed for the case of a cohesive material. The case of a horizontal surface can
be treated using KA and KP from Window 7-7. The horizontal stresses corresponding to active and passive
states are obtained from:
Active state:
3C cos
( 11 )A = KA h
3 + 2 sin
( 11 )A = KA h
Passive state:
2C cos
1 + sin
if 3 = 0.5 ( 1 + 2 )
3C cos
( 11 )P = KP h
3 + 2 sin
( 11 )P = KP h
2C cos
1 sin
if 3 = 1
if 3 = 1
if 3 = 0.5 ( 1 + 2 ) .
and KP cannot be read directly from Window 7-9, they

Under condition of seepage flow the value of KA
need to be derived explicitly. The same remark holds for the case of a cohesive material with surface inclined
at angle .
December 3, 2008

TM270

Window 7-9: Rankine states. Inclined surface
Earth pressure coefficient at rest K for a cohesionless semiinfinite soil mass with surface
inclined at angle . DruckerPrager with 1 = 3 .
Earth pressure coefficient at rest K for a cohesionless semiinfinite soil mass with surface
inclined at angle . DruckerPrager with 3 =0.5( 1 + 2 ).
Window 7-9
December 3, 2008

TM271

7.4.3
Let 12 be the plane containing axes 1 and 2 in which the plane strain problem is defined. If failure is to
occur in the 12 plane, 3 must be the intermediate stress; hence the following condition must apply:
1 3 2 .
Simultaneously, plane strain holds, i.e. for the elastic case:
and 0.5.
3 = ( 1 + 2 )
The following limiting conditions results:
1.

3 = 1 =

2
2.
3 = 0.5 ( 1 + 2 ) .
The second one result from the limits of Poissons ratio. If no tectonic stresses are present, all elastic states
lies within these limits.
These conditions are sometimes met a priori by the boundaryvalue problem (e.g. for the box shaped
medium) or by the adopted matching of DruckerPrager/MohrCoulomb criteria, as e.g. for the elastic
matching proposed earlier. Violating these conditions can have a significant effect on the solution.
In addition, some choices of material data lead to plastic behavior already under gravity loading. Since this is
the most common loading in soil mechanics, it is interesting to investigate the corresponding limits of elastic
behavior. Some situations of special interest are analyzed next for a cohesionless material.
Case 1: Box-shaped medium under gravity load, dry, matching collapse load.
The box-shaped medium is the default configuration adopted in the program. Combining plane strain and
the boundary conditions associated with the boxshaped medium leads for the isotropic medium to:

1 = 3 =
2 ; 2 = h
1
Notice that this coincides with a limiting condition established previously for failure to occur in the plane 12
and to the active Rankine state. This condition is therefore satisfied a priori. The invariants corresponding
to this stress state are:

2
1+
1 1 2
I1 =
J2 =
2 = 3
2 .
1
3 1
In stress space, the corresponding stress point is located on a cone with its vertex at the origin, characterized
by:
p
a I1 J2 = 0
from which,
3a =
J2
= 3
1 2
1+

.
For a cohesionless soil, a characterizes the position of the stress point with respect to the yield surface.
Elastic and plastic stress states can be identified as follows:
a < a
a = a
a > a
: elastic state
: plastic limit
: outofbalance state
Using the matching rules discussed earlier with a replacing a , relations are established which define the
elastic limit as a function of and .
Matching the collapse loads corresponding to MohrCoulomb and DruckerPrager criteria, under plane strain
conditions and deviatoric flow, yields the following result:
December 3, 2008

TM272

a =
sin
3
or
sin y =
3 12
1+ .
The lower index y denotes yielding. This curve is shown in Window 7-10.
Note that a similar curve can be derived using directly the MohrCoulomb criterion without any consideration
of matching with the DruckerPrager criterion, leading to:
sin y = 1 2.
Material data corresponding to a point located below the curve will automatically generate a plastic state,
and a point located above will generate elastic behavior. Different situations can be analyzed in a similar
way.
Window 7-10: Influence of Poissons ratio
Influence of Poissons ratio (box-shaped medium, cohesionless soil) (left); Influence of

Poissons ratio (infinite slope, cohesionless soil, dry medium) (right)
Window 7-10
Case 2: Box-shaped medium under gravity load, dry, matching elastic domains.
Matching of orthotropic elastic domains of DruckerPrager and MohrCoulomb criteria yields:

4 + 1
2 ;
2 = h
1 =
3
3 = 0.5 ( 1 + 2 ) .
The corresponding invariants were computed earlier and (see matching of elastic domains) ratio a is such
that yielding occurs if:
1 2
J2
=
sin
3a =
2 (1 + )
This curve is plotted in Window 7-10. Material data corresponding to a point located below the curve will
generate a plastic state.
Case 3: Saturated medium.
The same results as before apply to effective stresses.
December 3, 2008

TM273

Case 4: Infinite slope at angle , dry, matching collapse loads.
Principal stresses are:
1 = h(1 sin )
2 = h(1 + sin )
3 = ( 1 + 2 ) = 2h
Corresponding invariants are:
I1 = 2 (1 + ) h
J2 =
then

1
2
(h) 3 sin2 + 1 4 + 4 2
3
3a =
J2
=

3 3 sin2 + (1 2 2 )
2 (1 + )
Note that, for failure to occur in plane 12 , 3 must be the intermediate stress. This yields:
1 sin
2
(see Window 7-10 ) .
Similarly, the adjustment of yield criteria for place strain collapse load yields:
q

3 3 sin2 + (1 2 2 )
3a =
> sin .
2 (1 + )
For each given slope in a cohesionless soil, a curve = f () can be drawn, which characterizes the limit
of elastic behavior and, for the given boundary-value-problem, the limit of stability (Window 7-5).
Case 5: Infinite slope at angle , dry medium, matching elastic domains (Window 7-10).
Adjustment of yield criteria for coincidence of elastic domains, with t = 0.5. The state of principal stresses
is:
1 = h(1 sin )
2 = h(1 + sin )
3 = 0.5( 1 + 2 ) = h
The corresponding invariants are:
I1 = 3h
J2 = 2 h2 sin2
with elastic matching of failure criteria, this corresponds to yield if:
sin > sin .
The stress state in a cohesionless soil, for the adopted adjustment with the MohrCoulomb criterion, will be
elastic if < and plastic if > . Poissons ratio has no influence in this particular case.
Case 6: Infinite slope at angle , saturated.
The same results as before apply, corresponding to effective stresses.
December 3, 2008

TM274

7.4.4
Box-shaped medium in plane strain

For most static plane strain problems the soil halfplane can be conveniently approximated by a boxshaped
medium with smooth lateral boundaries (Window 5.4.9). The particular stress-strain state which results
can easily be derived.
From plane strain and lateral boundary conditions :
3 = 0 3 = ( 1 + 2 )
1 = 0 1 = ( 2 + 3 )
therefore:
1 2
The stress-strain fields which result for some typical loading cases are summarized in Window 7-11.
1 =
Gravity field
As can be seen in Window 7-11 the correct implementation of gravity requires simultaneous application of
and corresponding initial stresses. This combination is characterized by the capital in this text.
An initial state corresponding to an urban environment can be established using the same procedure.
Axisymmetric medium
The default boundary conditions for the axisymmetric case are the same as for planestrain.
December 3, 2008

TM275

Window 7-11: Box-shaped medium with smooth lateral boundaries
Box-shaped medium (1 = 3 = 0)
No
APPLICATION OF:
Deadweight
downwards
Initial stress
02
Initial stress
01
gravity field
02 = h
01 = K0 02
03 = K0 02
NB: K0 = 1
by default
YIELDS:
2 = h
1 = 3 = 1
h
h
2 2
2 = E (1 1 )
2 = 0
1 = 3 = 1
02
02
2 2
2 = E (1 1 )
2 = 0, 1 = 01
3 = 0, 2 = 0
2 = h
1 = K0 02
3 = K0 02
2 = 0
WHERE:
hdepth
unit weight
(BOXD1)
(BOXD2)
(BOXD3)
(BOXD4)
Window 7-11
December 3, 2008

TM276

7.4.5
INFLUENCE OF WATER
A steady state Darcy flow model only is included, although the pressure b.c. can vary in time and for each
step steady state solution can be obtained.
December 3, 2008

TM277

7.5
SOIL RHEOLOGY
Soil is subjected to long term deformations which cannot be avoided. This phenomenon is called consolidation.
Modern consolidation theories split the deformations into several mechanisms and two time periods associated
with primary and secondary consolidation.
Primary consolidation is dominated by a mechanism of stressinduced seepage flow which transfers progressively the part of load carried by the interstitial water to the soil skeleton. During secondary consolidation,
after stabilization of primary consolidation, the deformation is dominated by creep mechanisms.
Creep can be split into volumetric and deviatoric components.
A carreful choice of boundary to avoid meaningless results.
December 3, 2008

TM278

7.6
ALGORITHMIC STRATEGIES
December 3, 2008

TM279
N Preface N N Geotech. aspects N N N Alg. strategies

7.6.1
Most combinations of drivers are possible provided they are meaningful: initial state, stability, ultimate load,
prestress, consolidation, creep, flow. Obviously, an initial state analysis should come first; a stability analysis
should not be followed by any other type of analysis unless provisions are taken to restart before the stability
analysis; recall that the stability analysis goes through the change of the material properties.
December 3, 2008

TM280

7.6.2
The excavationconstruction process shows an analogy with loads and associated loadtime histories. Each
element is associated with an existence timehistory which takes values (0 or 1) depending if the element
exists at a given time.
It is possible to simulate excavation and construction processes; corresponding restrictions are specified in
the following remarks.
Remarks
1. The initial mesh numbering will be referred to throughout the analysis, for display of results. It must
therefore include all elements appearing during analysis; some may however, be inactive at the beginning
of the analysis.
2. When performing an excavationconstruction analysis, stiffness update must obviously be required at the
beginning of each step. The corresponding algorithmic choice must be done in the input definition.
3. Excavation stages can not be associated with some types of stability e.g. algorithms; this options would
not be meaningful.
4. If an excavation is followed by a time dependent analysis, progressive unloading will occur.
5. Unloading can be controlled using load time functions attached to the elements, named as unloading
functions. The interaction forces of the excavated medium on the surrounding medium can be computed
as follows:

Z
LT F (t)
intEXC
T tot
F
=
B d
0, if no unloading function is given
exc
where the integration is carried out over the excavated domain. Each excavated element is associated with
a load function which can be used to control progressive unloading.
Simulation of excavation. Events sequence

If no LT F (unloading function) is specified for excavated elements, interaction forces from excavated
media will vanish immediately at the moment of excavation (situation 2a in the above figure). Forces
FintREM will act as a load in a first step after the excavation. If compressive stresses dominated in the
December 3, 2008

TM281

area before the excavation then FintREM will generate tensile load around newly created boundary. In
the case of elasto-plastic media this may cause difficulties in obtaining converged solution. In that case
progressive unloading after excavation (situation 2b in the above figure) will deminish destabilizing
effect of FintREM and helps to redistribute stresses in the surroundings of the excavated domain, and in
consequence to obtain convergent solution of the new equilibrium state, (situation 3 in the above figure).
The excavation data should include existence function and, if needed, unloading function. Both are shown
in the below figure
Functions controlling excavation process

Above numerical procedure, despite being useful in obtaining solution, corresponds to technical measures
normally undertaken at the construction site in order to prevent failure of a soil mass during the excavation,
like temporary supports or spacers.
December 3, 2008

TM282
Index
3D
analysis and drivers, DP: 16, DP: 67
3D analysis
beams, TM: 179, TM: 184, TM: 187, TM:
189, TM: 196
continuum finite elements, TM: 109, 110
EAS, TM: 119
elastic model, TM: 51
membranes, TM: 214, TM: 217, TM: 222
numerical integration, TM: 111, TM: 151
shells, TM: 204
trusses, TM: 163, TM: 166
Analysis
batch processing, DP: 496
restart computation, DP: 495
run computation, DP: 494
run computation without writing *.dat, DP:
507
Auxiliary planes, DP: 226
Axisymmetry
analysis and drivers, DP: 16, TU: 29, DP: 67
beams (shells), TM: 177, TM: 184, TM: 187,
TM: 189, TM: 196
EAS, TM: 119121, TM: 123
foot benchmark, BM: 21, BM: 23
trusses and rings, TM: 162, TM: 166, TM:
170
Beams, TU: 37, TM: 176
analytical solution benchmarks, BM: 66, 67,
BM: 69, BM: 75, 76
axisymmetric shell benchmarks, BM: 7981
create/outline/update/delete elements, DP:
248
create/outline/update/delete subdomain 2D,
TU: 44, DP: 146
create/outline/update/delete subdomain 3D,
DP: 146
hinges, TM: 194
orientation 2D, TU: 44, TM: 177
orientation 3D, TM: 177
reinforced concrete benchmarks, BM: 71, BM:
73
subdomain 2D parameters, DP: 165
subdomain 3D parameters, DP: 165
December 3, 2008
Bearing capacity, DP: 2729

foot benchmark, BM: 16
Boundary conditions
for humidity, DP: 201, DP: 332
for humidity on macro-elements, DP: 201
for pore pressure, DP: 194, DP: 333
for pore pressure on macro-elements, DP: 194
for solid phase, DP: 320
for temperature, DP: 198, DP: 327
for temperature on macro-elements, DP: 198
Consolidation, TU: 33
algorithm, TM: 144
analytical solution benchmarks, BM: 38
geotechnical aspects, TM: 252
material model, TM: 54
numerical implementation, TM: 104
overconsolidation ratio, TM: 261
problem statement, TM: 39
Construction algorithm
analysis and drivers, DP: 16, TM: 281
Contact
macromodel, TU: 37, TU: 61, TU: 71, TU:
77
Continuum 2D elements, DP: 263
Continuum 2D macromodel
Automatic mesh generation, DP: 180, 181
Mesh morphing, DP: 182
Semi-automatic mesh generation, DP: 176,
177
Subdomain generation, DP: 145, 146
Virtual mesh, DP: 175178
Continuum 3D elements, DP: 268
Continuum 3D macromodel
Mesh morphing, DP: 183
Semi-automatic mesh generation, DP: 178,
179
Subdomain generation, DP: 147150, DP: 152,
DP: 154, DP: 156
Virtual mesh, DP: 178, 179
Convection 2D macromodel
Subdomain generation, DP: 187
Convection 3D macromodel
Convection elements, DP: 312
Convergence, TM: 136
Creep
analytical solution benchmark, BM: 45
standard properties, DP: 425
swelling properties, TM: 93, DP: 425

TM283

Dynamics
added masses, DP: 353
consistent mass matrix, DP: 70
control parameters, DP: 70
HHT scheme, DP: 70
lumped mass matrix, DP: 70
mass filtering, DP: 71
Newmark scheme, DP: 70
Reylaigh damping, DP: 70
Seismic input, DP: 493
humidity, DP: 365

Fluxes on macromodel
fluid, DP: 204
heat, DP: 207
humidity, DP: 210
Gravity, DP: 491
body load components, DP: 491
direction, DP: 491
Heat
Elasticity
constants, DP: 7
material properties, DP: 419
Elasto-plastic 1D
model, TM: 228, DP: 478
Excavation/Stage construction
algorithm, TM: 281
benchmark, BM: 95
existence function, TM: 281
show steps in preprocessor, TU: 44, TU: 61,
TU: 71, TU: 77, DP: 94
unloading function, TM: 281
unloading function benchmark, BM: 36
Existence functions, TM: 281, DP: 486
FE model preprocessing
common methods to copy elements/nodes,
DP: 235
common methods to delete elements/nodes,
DP: 235
common methods to move elements/nodes,
DP: 235
common methods to outline elements/nodes,
DP: 235
common methods to rotate elements/nodes,
DP: 235
Finite elements
selection strategy, DP: 78, DP: 80
stabilization of pressure oscillations, DP: 80
volumetric locking, DP: 79
Flow
analysis, DP: 16, TU: 41, DP: 44, TU: 67
benchmarks, BM: 57
fluid head boundary condition, TU: 41, TU:
67
flux boundary condition, DP: 45, DP: 204
initial conditions, DP: 45
initial state driver, DP: 44, DP: 46, TU: 67
material data, TU: 41, DP: 44, TM: 5456,
TU: 67, DP: 422
pressure boundary condition, TU: 41, DP: 45,
DP: 194
steady state driver, TU: 41, DP: 44, DP: 47
time dependent drivers, DP: 47
transient driver, DP: 44, DP: 48
Fluxes on elements
fluid, DP: 361
heat, DP: 364
December 3, 2008
analysis, DP: 16, DP: 49, TU: 51

analytical solution benchmark, BM: 61
flux boundary conditions, TM: 45, DP: 50,
DP: 207
initial conditions, TM: 45, DP: 50, 51, TU:
51
initial state driver, DP: 49, DP: 51, TU: 51
material properties, TM: 45, TU: 51, DP: 429
steady state driver, DP: 49, DP: 52
temperature boundary conditions, TM: 45,
DP: 50, DP: 198
thermal strains, TU: 51, DP: 68
time dependent drivers, DP: 49, DP: 52
transient driver, DP: 49, TU: 51, DP: 53
Hinges
in beam elements, DP: 254
in shell elements, DP: 316
Humidity
analysis, DP: 16, DP: 54
flux boundary conditions, TM: 47, DP: 55,
DP: 210
humidity boundary conditions, TM: 47, DP:
55, DP: 201
hygral strains, DP: 54, DP: 68
initial conditions, TM: 47, DP: 55, 56
initial state driver, DP: 54, DP: 56
material properties, TM: 47, DP: 54, DP: 431
steady state driver, DP: 57
time dependent drivers, DP: 54, DP: 57
transient driver, DP: 58
Infinite elements, TM: 126, DP: 290
Initial condition
for displacements, DP: 374
for humidity, DP: 369, 370
for pore pressure, DP: 367
for solid velocities, DP: 374
for temperature, DP: 368
Initial state
algorithm, TM: 139
earth pressure at rest (Ko), TM: 261, TM:
263
geotechnical aspects, TM: 260
Initial state Ko

TM284

Interface 2D
Material data groups, TM: 230, TM: 241,
TM: 246
Material models, TM: 230, TM: 241, TM:
246
Interface 2D macromodel
Interface 3D
Material data groups, TM: 230, TM: 241,
TM: 246
Material models, TM: 230, TM: 241, TM:
246
Interface 3D macromodel
Interface elements, TM: 230, TM: 241, TM: 246,
DP: 295
Interface elements for large deformations, DP: 304
Kinematic constraints, DP: 377
Large deformations
analysis and drivers, DP: 16
Linear equation solvers, DP: 73
skyline, DP: 73
sparse, DP: 73
Load time functions, DP: 489
Loads
body, DP: 340
on beam elements, DP: 351
on element surfaces, DP: 344
on nodes, DP: 342
on subdomain boundaries, DP: 213
Macromodeling
2D mesh mapping, DP: 139
Bore holes, DP: 385
Extrusion direction, DP: 142
Fluid flux, DP: 204
Heat flux, DP: 207
Humidity BC, DP: 201
Humidity flux, DP: 210
Objects, DP: 129
Piles, TU: 77, DP: 221
Point, DP: 127
Point loads, DP: 231
Pressure BC, DP: 194
Subdomain, DP: 143
Subdomain parameters, DP: 164
Surface load, DP: 213
Temperature BC, DP: 198
Virtual to real mesh conversion, DP: 183
Macromodeling objects
Arc, DP: 132
Circle, DP: 133
common methods, DP: 130
Delete methods, DP: 138
DXF import, DP: 137
Line, DP: 131
December 3, 2008
Line(s) on edge(s), DP: 133

Outline methods, DP: 138
Point, DP: 131
Split by plane, DP: 137
Surface intersection, DP: 134, DP: 184
Surface on Q4 skeleton, DP: 134
Surface on T3 skeleton, DP: 134
Update methods, DP: 139
Main Z Soil menu
Analysis options, DP: 494
Assembly options, DP: 83
Control options, DP: 14, 15
Control:analysis and drivers, DP: 16
Extras, DP: 507
File options, DP: 11
Help, DP: 509
Postpro, DP: 497
System configuration, DP: v, DP: 508
toolbars, DP: 10
Materials
creep properties group, TM: 86, DP: 425
data groups, DP: 419
databases, DP: 393
elastic properties group, DP: 419
flow properties group, TM: 54, DP: 422
handling, DP: 392
heat properties group, DP: 429
humidity properties group, DP: 431
initial Ko properties group, DP: 427
local stability properties group, DP: 432
models, TM: 49, DP: 393
properties varying in space, DP: 481
properties varying in time, DP: 481
unit weight properties group, DP: 420
Materials for axisymmetric shell elements
linear elastic model, DP: 404, DP: 406
nonlinear (layered) model, DP: 404, DP: 406
Materials for beam elements, DP: 394
fiber model, TM: 228
linear elastic model, TU: 37, DP: 395, DP:
399
nonlinear (layered) model, DP: 395, DP: 400,
DP: 402
Materials for contact elements, DP: 408
frictional contact model, DP: 409
Materials for continuum elements, TM: 49, DP:
418
Aging concrete model, TM: 97, DP: 434
Cap model, TM: 65, DP: 436
Drucker-Prager model, TU: 29, TM: 61, DP:
442
Duncan Chang model, DP: 443
ECP Hujeux model, DP: 453
Hoek-Brown(M-W) model, TM: 76, DP: 444
HS-small model, DP: 456
Linear elastic model, TM: 50, DP: 433
Modified Cam Clay model, TM: 82, DP: 439
Mohr-Coulomb (M-W) model, TM: 72, DP:
446

TM285

Mohr-Coulomb model, TU: 25, TU: 33, TM:
60, DP: 451
Multilaminate model, TM: 78, DP: 448
Rankine(M-W) model, DP: 450
Materials for continuum elements for structures,
DP: 460
Materials for heat convection elements
convection model, DP: 460
Materials for humidity convection elements
convection model, DP: 461
Materials for infinite elements
Linear elastic model, DP: 462
Materials for membrane elements, TM: 219, DP:
463
Anisotropic elasto-plastic model, TM: 219,
DP: 468
Fiber elasto-plastic model, TM: 219, DP: 464
Isotropic elasto-plastic model, TM: 219, DP:
468
Plane stress elastic model, TM: 219, DP: 467
Plane stress Hoek-Brown model, TM: 219,
DP: 467
Plane stress Huber-Mises model, TM: 219,
DP: 467
Plane stress Rankine model, TM: 219, DP:
467
Materials for pile foot interface elements, TU: 77,
DP: 415
Materials for pile interface elements, TU: 77, DP:
412
frictional contact model, DP: 413, DP: 416
Materials for seepage elements, DP: 470
Materials for shell elements, TM: 204, DP: 471
Aging concrete model, TM: 204
Fiber model, TM: 204
Linear elastic model, DP: 473
Nonlinear (layered) model, TM: 204, DP: 473
475
Orthotropic elastic model, DP: 477, 478
Membrane 2D
Thickness, TM: 221
Membrane 2D macromodel
Membrane 3D macromodel
Subdomain generation, TU: 71, DP: 162
Membranes, TU: 71, TM: 214, DP: 285
reinforced soil benchmark, BM: 89
Mesh tying, DP: 382
Nodal link, DP: 379
Node, DP: 243
Nonlinear solvers, DP: 69
Overlaid meshes
generation, DP: 110
Plane strain
analysis and drivers, DP: 16, DP: 67
December 3, 2008
beams, TM: 177, TM: 184, TM: 187, TM:

189, TM: 195
box-shaped medium benchmarks, BM: 9
EAS, TM: 119, TM: 123
trusses, TM: 162, TM: 166
Postprocessing
how do I...., DP: 499
using macros, DP: 501
Preferences, DP: 107
Preprocessing
FE model, DP: 85, DP: 234
Macromodelling, DP: 85, DP: 126
Main menu, DP: 86
User interface, DP: 85
Preprocessor
construction lines, DP: 108, 109
copy by rotation selected objects, DP: 92
copy by symmetry, DP: 92
copy by translation selected objects, DP: 92
grid, DP: 109
import geometrical model, DP: 89
move selected objects, DP: 91
rotate selected objects, DP: 91
selection by plate, DP: 104
selection inside box, DP: 103
selection lists, DP: 100
selection of finite elements, DP: 101
selection of nodes, DP: 102
selection with oriented plane, DP: 103
show distance, DP: 106
show node coordinates, DP: 106
show vector, DP: 106
show volume of continuum element, DP: 106
snap options, DP: 109
visibility setup, DP: 96
Preprocessor menu
Cursor mode, DP: 96
Edit, DP: 89
Element nodes selection, DP: 104
Files, DP: 88
Help, DP: 111
Layers, DP: 109
Selections, DP: 98
Settings, DP: 106
Tools, DP: 106
View, DP: 92
Visualization, DP: 94
Preprocessor toolbars
2D mesh refinement, DP: 123
3D mesh refinement, DP: 124
3D view, DP: 116
camera position, DP: 116
construction lines, DP: 114
edges selection, DP: 118

TM286

edit component, DP: 122
edit selected components, DP: 115
faces selection, DP: 117
grid, DP: 114
hide selected components, DP: 114
mesh refinement, DP: 122
selection lists, DP: 119
selection modes, DP: 126
snap to options, DP: 114
special selections methods, DP: 125
split elements, DP: 124
standard selection cursor, DP: 121
toolbars, DP: 113
tools, DP: 120
view 3D/2D scene, DP: 121
visibility, DP: 115
visualizalization of materials, existence and
unloading functions, DP: 126
working plane in 3D, DP: 116
zoom in/out, DP: 115
Problem statement, TM: 37
heat, TM: 45
humidity, TM: 47
single phase, TM: 38
two phase, TM: 39
Pushover
control parameters, DP: 71
results, DP: 498
Pushover control node, DP: 384
Restart
control, DP: 69
Results
content, DP: 81
for beam elements, DP: 82, TM: 195
for continuum elements, DP: 82
for shell/membrane elements, DP: 82, TM:
213
nodal solid accelerations, DP: 83
nodal solid velocities, DP: 83
residuals at nodes, DP: 83
standard nodal results, DP: 83
storage frequency, DP: 69
Seepage 2D macromodel
Seepage 3D macromodel
Subdomain generation, TU: 67, DP: 184
Seepage elements, DP: 308
Shell 1L
Thickness, TU: 61, TU: 77, TM: 200
Shell 1L macromodel
Extrusion, TU: 61, TU: 77
Virtual mesh, TU: 61, TU: 77, DP: 179
Shell elements, TM: 197, DP: 274
benchmarks, BM: 8387
Shell elements with one layer of nodes, TU: 61,
TU: 77, DP: 280
December 3, 2008
Single phase
driven load driver, DP: 25, TU: 29
effective stress analysis, DP: 20
initial state driver, DP: 21
stability driver, TU: 25, TU: 71
total stress analysis, DP: 20, DP: 24, DP: 33
Stability, TU: 25, DP: 30
algorithm, TM: 141
analysis, DP: 32, 33
driver, TU: 25, DP: 3033
local material setting, DP: 31, DP: 432
slope benchmark, BM: 24
Strains
imposed, DP: 372
thermal, DP: 49
Stress
sign convention, DP: 3
Structures
beams, TM: 177
direction on surface, TM: 226
local base, TM: 227
membranes, TM: 222
offset, TM: 224
shells, TM: 197
trusses, TM: 161
Subdomain
Excavation front, DP: 167
Project subdomain on surface, DP: 184
Swelling
material properties, TM: 93, DP: 425
Truss 2D macromodel
Truss 3D macromodel
Truss elements, TM: 162, DP: 255
prestress benchmark, BM: 34
Two phase
box-shaped medium benchmarks, BM: 11, BM:
13
consolidation driver, DP: 38
driven load driver, DP: 37
initial state driver, DP: 35
slope stability benchmark, BM: 32
stability driver, DP: 40, 41
Unit weight
definitions, DP: 6
in single phase problems, DP: 20, DP: 24

TM287

in two phase problems, DP: 41
Units, DP: 74
basic, DP: 7, DP: 76
compound, DP: 77
conversion, DP: 75
setting basic units, DP: 74
December 3, 2008

TM288

Theory: Z Soil - PC 2009 Manual

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Theory: Z Soil - PC 2009 Manual

Загружено:

Авторское право:

Доступные форматы

THEORY

Z Soil.PC 2009 manual

A. Truty Th. Zimmermann K. Podles

Zace Services Ltd, Software engineering

Z Soil.PC 2009 manual:

END-USER LICENSE AGREEMENT FOR ZACEs Z SOIL.PC SOFTWARE

Z SOIL.PC (the Software) SOFTWARE PRODUCT LICENSE

NO LIABILITY FOR CONSEQUENTIAL DAMAGES

QuickHelp DataPrep Benchmarks Tutorials

Contents of Theoretical Manual

SOME IMPORTANT FORMULAE IN TENSOR ALGEBRA AND ANALYSIS

SINGLE PHASE, SOLID MEDIUM . . . . . . . . . . . . . . . . . . . . . .

TWO-PHASE PARTIALLY SATURATED MEDIUM . . . . . . . . . . . . .

GENERALIZED DARCY LAW

SKETCH OF THE PLASTICITY APPROACH . . . . . . . . . . . .

HOEKBROWN CRITERION (SMOOTH) . . . . . . . . . . . . . .

CUT-OFF CONDITION AND TREATMENT OF THE APEX . . . .

MODIFIED CAM CLAY MODEL . . . . . . . . . . . . . . . . . . .

CREEP UNDER VARIABLE STRESS

QuickHelp DataPrep Benchmarks Tutorials

CREEP PARAMETER IDENTIFICATION FROM EXPERIMENTS

SAFETY FACTORS AND STRESS LEVELS . . . . . . . . . . . . .

WEAK FORM AND MATRIX FORMS OF THE PROBLEM

SINGLE PHASE MEDIUM, TIME INDEPENDENT LOADING . . . . 103

TWO-PHASE MEDIUM, RHEOLOGICAL BEHAVIOUR . . . . . . . 104

FINITE ELEMENTS FOR 2D/3D CONTINUUM PROBLEMS . . . . 110

STIFFNESS MATRIX . . . . . . . . . . . . . . . . . . . . . . . . . 113

BODY FORCES AND DISTRIBUTED LOADS

INITIAL STRESSES, STRAINS . . . . . . . . . . . . . . . . . . . . 115

INCOMPRESSIBLE AND DILATANT MEDIA . . . . . . . . . . . . . . . . 116

INCOMPRESSIBLE MEDIA : B-BAR STRAIN PROJECTION METHOD117

DILATANT MEDIA: ENHANCED ASSUMED STRAIN METHOD . . 119

INTRODUCTION TO ENHANCED ASSSUMED STRAIN

EXTENSION OF THE EAS METHOD TO NONLINEAR

REMARKS AND ASSESSMENT OF EAS ELEMENTS . . 125

OVERLAID MESHES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132

FULL/MODIFIED NEWTON-RAPHSON ALGORITHM . . . . . . . 135

INITIAL STATE ANALYSIS

STABILITY ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . 141

ULTIMATE LOAD ANALYSIS

QuickHelp DataPrep Benchmarks Tutorials

LOAD FUNCTION AND TIME STEPPING PROCEDURE

SIMULATION OF EXCAVATION AND CONSTRUCTION STAGES . 148

SHAPE FUNCTION DEFINITION AND REFERENCE ELEMENTS

NUMERICAL INTEGRATION DATA FOR DIFFERENT ELEMENTS

MULTISURFACE PLASTICITY CLOSEST POINT PROJECTION ALGORITHM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

SINGLE SURFACE PLASTICITY CLOSEST POINT PROJECTION

TRUSS ELEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . 161

GENERAL IDEA OF TRUSS ELEMENT

GEOMETRY AND DOF OF TRUSS ELEMENTS . . . . . 164

INTERPOLATION OF THE DISPLACEMENTS AND STRAINS

WEAK FORMULATION OF THE EQUILIBRIUM . . . . . 167

STIFFNESS MATRIX AND ELEMENT FORCE VECTOR

GEOMETRY AND KINEMATICS OF A RING ELEMENT

WEAK FORMULATION OF THE EQUILIBRIUM . . . . . 171

STIFFNESS MATRIX AND ELEMENT FORCES . . . . . 172

ANCHORING OF TRUSS AND RING ELEMENTS

NUMERICAL REALIZATION OF ANCHORING . . . . . . 174

PRESTRESSING OF TRUSS AND RING ELEMENTS . . . . . . . 175

GEOMETRY OF BEAM ELEMENT . . . . . . . . . . . . . . . . . 177

KINEMATICS OF BEAM THEORY

WEAK FORMULATION OF THE EQUILIBRIUM . . . . . . . . . . 184

INTERPOLATION OF THE DISPLACEMENT FIELD . . . . . . . . 186