Full 10088

AGHUniversityofScienceandTechnology
FacultyofMechanicalEngineeringandRobotics
DoctoralThesis
WeldingsequenceAnalysis
By
IsaacHernndezArriaga
CoAdvisor:Dr.Hab.PiotrRusek,Prof.AGH
Dr.EduardoAguileraGmez
September2009

Welding Sequence Analysis

i
Isaac Hernndez Arriaga

ABSTRACT

This thesis has been divided in nine chapters.

Chapter 1 provides brief background and general and specific objectives of this work.

Chapter 2 presents the methods and advantages of reducing or controlling residual stresses and distortion
induced by the welding process as well as a definition and classification of the welding sequence. It also
presents a study of the welding sequence analysis, a survey of previous research in the field of welding
sequences, and a discussion of its advantages, disadvantages, scope, and limitations.

The subject of Chapter 3 is the finite element modeling of the welding processes, defining the boundary
and initial conditions of the welding process and studying the effects of the welding sequence on the residual
stress distribution and distortion in symmetrical structures. It should be noted that the proposed numerical
model has general applicability and is not limited to symmetrical structures. The proposed sequentially-
coupled thermo-mechanical analysis involves two steps. A transient heat transfer analysis is performed
followed by a thermal elastic plastic analysis. This numerical simulation is performed in an I-type specimen
subject to tension and validated with experimental data [25]. Finally, the chapter presents a numerical
simulation of the welding sequence in an L-type structure to demonstrate that the proposed numerical model
accurately simulates the effects of the welding sequence on residual stresses and distortion.

Chapter 4 presents a study of the effects of the welding sequence on residual stresses and distortion in a
stiffened symmetrical flat frame. Selected welding sequences reduce residual stresses, distortion, or the
relation between both parameters. These proper welding sequences are obtained from empirical welding
rules, axis of symmetry, center of gravity of the frame, and concentric circles. The origin of the circles coincides
with the center of gravity of the frame, and the radius of the circles is formed by the center of gravity of the
frame with the center of gravity of each of the weld beads. The different welding sequences are analyzed with
the numerical model developed in chapter 3. Finally, the chapter presents a procedure to determine the
proper welding sequences to reduce residual stress, distortion or the relation between both parameters for 2-
dimensional symmetrical structures.

The main goal chapter 5 is to demonstrate that the procedure to determine the proper welding sequences
to reduce residual stress, distortion or the relation between both parameters in 2-dimensional symmetrical
structures can be applied to 3-dimensional symmetrical structures. This is done by studying the effects of the


ii
welding sequence on the residual stresses and distortion in a 3-dimensional unitary cell-type symmetrical
structure. Now the weld bead circles become spheres. To demonstrate the procedure to determine the proper
welding sequence for 3-dimensional symmetrical structures, four numerical simulations are performed in the
proposed symmetrical structure. Two of these numerical simulations deal with the proper welding sequence
to reduce residual stress and the other two deals with the proper welding sequence to reduce distortion. Also,
the numerical simulation of a special welding sequence is performed for comparison with the proper welding
sequence to reduce distortion. This special welding sequence applies the external weld beads first and the
internal weld beads later. All the numerical simulations are based on the proposed numerical model of the
welding process developed in Chapter 3.

Chapter 6 presents a methodology for the development of the experimental tests. This methodology helps
to plan, execute, and control each of the stages of the experimental tests. The methodology starts with the
material selection of the specimen, configuration selection, welding process selection, metal transfer mode
selection, welding parameter selection, design and fabrication of the equipment needed to run the test,
design and fabrication of the mounting locks, residual stresses relief caused by the manufacturing process,
transportation, handling, storage and cutting of the plates, measurement of the initial distortion of the plates,
design and fabrication of a holder-mounting device to hold the plates, design and fabrication of a square-
mounting device to square the holder-mounting device, application of welding tacks, measurement of the
distortion after applying the welding tacks, installation of the run-off tabs, application of the welding, removal
of the run-off tabs of the welded structure, measurement of the distortion after welding, and measurement of
the final distortion induced by the welding process.

Chapter 7 covers the results of the experimental tests performed in 3-dimensional unitary cell-type
symmetrical specimens. In these experimental tests, the effects of the welding sequence on distortion are
studied. Eight symmetrical specimens are prepared. Four welding sequences are considered: two of them are
adequate to reduce distortion and the other two reduce residual stresses. The chapter also studies the effects
that occur when a welding bead is divided into 3 sub-weld beads, as well as the effects of relieving the residual
stresses caused by manufacturing process, transportation, storage and cutting of plates. Welding tacks are
applied to all specimens before the actual weld process begins. The measurement of the distortion is
periodically performed to observe if rheological effects occur in the specimens after welding. The experimental
tests were performed in the Department of Machine Strength and Manufacturing in the Faculty of Mechanical
Engineering at the University of Science and Technology in Krakow, Poland.

Chapter 8 presents a comparison between the numerical results obtained in Chapter 5 and the
experimental results obtained in Chapter 7 for the 3-dimensional unitary cell-type symmetrical structure. The
comparison discusses the distortion modes and the distortion in the 24 points of interest. The chapter
presents the procedures to determine the proper welding sequences to reduce the residual stresses,
distortion, or a relation between both parameters in symmetrical and asymmetrical structures in 2 and 3


iii
dimensions. These procedures were developed in chapters 4 and 5 to determine the proper welding
sequences to reduce residual stresses, distortion, or relation between both parameters in symmetrical
structures in 2 and 3 dimensions.

Chapter 9 presents the conclusions, contributions and suggestions for future work.


iv

ACKNOWLEDGMENTS

I am extremely grateful for the support of the University of Guanajuato and AGH University of Science and
Technology. They provide employment and resources which made it possible for me to pursue this degree.

I wish to express my sincere appreciation to Dr. Eduardo Aguilera for his guidance, encouragement and
insight throughout the duration of this research.

I would also express my gratitude to Professor Piotr Rusek for his encouragement and support, his
influence extends far beyond my academic work.

I wish to tanks to Dr. Arturo Lara, Dr. Elias Ledesma, Professor Stanisaw Wolny, and Professor Andrzej
Skorupa for serving as dissertation committee members and providing positive suggestion and comments.

I acknowledge the Consejo Nacional de Ciencia y Tecnologa (CONACYT), Direccin de Relaciones
Academicas Internacionales e Interinstitucionales (DRAII), and Direccin de Investigacin y Posgrado (DINPO)
of the University of Guanajuato for the funding of doctoral studies, doctoral research and stay at AGH.
I would also like to express my gratitude to Director of the engineering division of the Irapuato-Salamanca
Campus; Dr. Oscar Ibarra, for his invaluable support in the completion of my doctoral studies.
I would especially like to thank the Authorities of the AGH University of Science and Technology; Rector,
Professor Antoni Tajdu, Vice-Rector for Cooperation and Development, Professor Jerzy Lis, and Dean of the
faculty of Mechanical Engineering and Robotics, Professor Janusz Kowal for thier invaluable collaboration with
the University of Guanajuato.
I am also grateful to Dr. Hector Plascencia for his interest and help to my research. Also, I would like to
thank to Dr. Pedro de Jess Garca and to Dr. Rogelio Navarro for their initial help, interest and advice.

I am very grateful to Dr. Tomasz Gral for his help on experimental research. Thanks are also extended to
Drs. Jerzy Haduch and Andrzej Tyka for their assistance.


v
During my stay at AGH, I was fortunate to have had a number of talented technical workers. Kazimierz
Nawrot, Wodzimierz Rusek, and Artur Konopczak all helped with the fabrication of the equipment needed to
run the test.

My thanks to Salvador Martnez, for his friendly help in the experimental measurements at AGH. The same
quality of special thanks goes to Mr. Guadalupe Negrete for his expertise and generous help in welding .

My special thanks go to Renato Snchez for giving me their generous and solidary support.

To all my friends and classmates, especilly Hijinio Jurez, Alejandro Len, Sergio Pacheco and Mr.
Baldomero Lucero, thank you for your warm friendships.

To Ms. Ma. Eugenia Gallardo, secretary of our Mechanical Department, for helping me during my studies.

My academic achievements would have been impossible without the spiritual support of my family. Special
thanks are due to my parents, Beny Arriaga and Daniel Hernndez. Their sacrifice for my education made me
who I am. Thanks are also extended to my sister and brother; Ruth Hernndez and Daniel Hernndez, for their
understanding and support for my studies. Special love goes to my wife, Maria Victoria Cabrera whose
boundless love and encouragement made my time at University of Guanajuato and AGH easy and pleasant.
Also, my cute son, Samuel Isaac Hernndez, enabled me to periodically escape the academic pressure with his
smiles.

Finally, I thank an anonymous editor for assisting with the English version of my thesis.


vi
TABLE OF CONTENTS

Abstract i
Acknowledgments iv
Table of contents vi
List of figures xii
List of tables xvii
Nomenclature xix

Chapter I INTRODUCTION

1.1 Background 1
1.2 General objective 2
1.3 Specific objectives 2

Chapter II WELDING SEQUENCE BACKGROUND AND METHODS FOR CONTROLLING RESIDUAL
STRESSES AND DISTORTION INDUCED BY WELDING

2.1 Introduction 3
2.2 Advantages of residual stress and distortion control 3
2.3 Methods to control welding-induced residual stress and distortion 4
2.3.1 Welding sequence 4
2.3.2 Definition of weld parameter 4
2.3.3 Weld procedure 5
2.3.4 Fixture design 5
2.3.5 Precambering 5
2.3.6 Prebending 6
2.3.7 Thermal tensioning 6
2.3.8 Heat sink welding 7
2.3.9 Preheating 7
2.10 Post-weld heat treatment 7
2.11 Post-weld corrective methods 7
2.4 Welding sequence definition 8
2.5 Welding sequence classification 8
2.5.1 Welding sequence for single pass welds 8
2.5.2 Welding sequence for multiple pass welds 9


vii
2.6 Welding sequence selection based on empiric rules 10
2.7 Welding sequence background 11
2.8 Summary of the welding sequence analysis background 43
2.9 Matrix of the welding sequence analysis background 44

Chapter III PROPOSAL OF A NUMERICAL SIMULATION OF THE WELDING PROCESS AND A NUMERICAL
SIMULATION OF THE WELDING SEQUENCE IN AN L-TYPE STRUCTURE

3.1 Introduction 45
3.2 Heat transfer in welding 45
3.2.1 Analytical solution for the temperature field 46
3.2.2 Thermal Initial and boundary Conditions 48
3.3 Thermal elastic plastic stress analysis in welding 50
3.3.1 Mechanical equations 50
3.3.2 Mechanical initial and boundary conditions 51
3.4 Finite element solution of the welding 51
3.4.1 Finite element solution of heat transfer in welding 51
3.4.2 Finite element solution of the thermal elastic plastic stress analysis in welding 52
3.5 Geometric configuration of I-type specimen subject to tension 55
3.6 Material selection for the I-type specimen subject to tension 56
3.7 Temperature-dependent thermal and mechanical properties of ASTM A36 56
3.8 Finite element model of the I-type specimen subject to tension 57
3.8.1 Definition and justification of the applied finite elements 58
3.8.2 Thermal initial and boundary conditions 61
3.8.3 Mechanical boundary condition 61
3.8.4 Body load 61
3.8.5. Solution of the finite element model 62
3.9 Points of interest in the finite element model of the I-type specimen subject to tension 62
3.10 Residual stresses in the I-type specimen subject to tension obtained in the numerical simulation 62
3.11 Comparison between the numerical and experimental results of an I-type specimen subject to
tension
63
3.12 Conclusions of the numerical simulation of the welding process in an I-type specimen subject to
tension
64
3.13 Numerical simulation of the welding sequence in an L-type structure 65
3.14 Geometric configuration of the L-type structure 65
3.15 Finite element model of the L-type structure 65
3.15.2 Mechanical boundary conditions 67


viii
3.15.3 Solution of the finite element model of the L-type structure 67
3.16 Configuration of welding sequences for the L-type structure 67
3.17 Localization of the point of interest in the L-type structure 68
3.18 Numerical results in the L-type structure 68
3.18.1 Distortion profile in the L-type structure 70
3.18.2 Residual stress distribution in the L-type structure 70
3.19 Experimental tests for the L-type structure 71
3.19.1 Selection of points of interest in the L-type structure 72
3.19.2 Configuration of the welding sequences in the L-type specimens 72
3.19.3 Measurement of distortion on L-type specimens 73
3.20 Conclusions of the welding sequence analysis of the L-type structure 75

Chapter IV WELDING SEQUENCE ANALYSIS IN A STIFFENED SYMMETRICAL 2-DIMENSIONAL FRAME

4.1 Introduction 76
4.2 Geometric configuration of a stiffened symmetrical flat frame 76
4.3 Welding configuration in the stiffened symmetrical flat frame 77
4.4 Finite element model of the stiffened symmetrical flat frame 80
4.4.3 Solution of the finite element model of the stiffened symmetrical flat frame 82
4.5 Numerical results in the stiffened symmetrical flat frame 82
4.5.1 Distribution of the residual stresses in the stiffened symmetrical flat frame 82
4.5.2 Distortion profile in the stiffened symmetrical flat frame 83
4.6 Analysis of residual stress-distortion relations analysis 83
4.7 Order of importance of the welding sequences to reduce residual stress, distortion, or the
relation between them in the stiffened symmetrical flat frame
84
4.8 Proper welding sequences to reduce the residual stress, distortion, or a relation between them
in the stiffened symmetrical flat frame
84
4.8.1 Proper welding sequence to reduce the residual stress in the stiffened symmetrical flat
frame
86
4.8.2 Proper welding sequence to reduce distortion in the stiffened symmetrical flat frame 87
4.8.3 Proper welding sequence to improve the relation between both critical parameters in the
stiffened symmetrical flat frame
87
4.9 Hypothesis to determine the proper welding sequence to reduce the residual stress, distortion,
or a relation between them in symmetrical flat structures
88
4.10 Experimental tests in a stiffened symmetrical flat frame specimen 89
4.10.1 Selection of points of interest in the stiffened symmetrical flat specimen 90


ix
4.10.2 Configuration of the welding sequence in the stiffened symmetrical flat specimen 90
4.10.3 Measurement of distortion on the stiffened symmetrical flat specimen 91
4.11 Conclusions of the welding sequence analysis of stiffened symmetrical flat frame 93

Chapter V WELDING SEQUENCE ANALYSIS IN A 3-DIMENSIONAL UNITARY CELL-TYPE
SYMMETRICAL STRUCTURE

5.1 Introduction 95
5.2 Hypothesis to determine the proper welding sequence to reduce the residual stress, distortion,
or a relation between them in 3-dimensional symmetrical structures
96
5.3 Geometric configuration of the 3-dimensional unitary cell 97
5.4 Selection of the number of weld beads in the 3-dimensional unitary cell 98
5.5 Symmetry axis selection and formation of the concentric spheres in the 3-dimensional unitary
cell
98
5.6 Proper welding sequence to reduce the residual stress and proper welding sequence to reduce
the distortion in the 3-dimensional unitary cell
100
5.7 Material selection of the 3-dimensional unitary cell 101
5.8 Fillet weld shape used in the 3-dimensional unitary cell 101
5.9 Finite element model of the 3-dimensional unitary cell 101
5.9.3 Solution of the finite element model of the 3-dimensional unitary cell 103
5.10 Localization of the points of interest in the 3-dimensional unitary cell 103
5.11 Configuration of the numerical simulation for the 3-dimensional unitary cell 104
5.12 Numerical results of the different welding sequences analyzed in the 3-dimensional unitary cell 105
5.12.1 Maximum von Mises residual stress in the 3-dimensional unitary cell 105
5.12.2 Distortion modes in the 3-dimensional unitary cell 106
5.12.3 Maximum distortion and distortion in 24 points of interest in the 3-dimensional unitary
cell
106
5.13 Numerical comparison between the proper welding sequence to reduce distortion and a special
welding sequence in the 3-dimensional unitary cell
107
5.14 Conclusions of the welding sequence analysis of the 3-dimensional unitary cell 109

Chapter VI METHODOLOGY OF EXPERIMENTAL TESTS

6.1 Introduction 110
6.2 Specimen material selection 110
6.3 Selection of specimen configuration 111
6.4 Selection of the welding process 111


x
6.5 Selection of metal transfer mode 111
6.6 Selection of welding parameters (Operating variables) 112
6.6.1 Arc voltage and welding current 112
6.6.2 Welding speed 113
6.6.3 Wire feed rate 113
6.6.4 Selecting of contact tip to work distance 114
6.6.5 Electrode orientation 114
6.6.6 Electrode diameter 115
6.6.7 Shielding gas composition 115
6.6.8 Gas flow rate 116
6.7 Design and fabrication of the equipment needed to run the test 116
6.8 Design and fabrication of the mounting locks 117
6.9 Relief of residual stresses caused by the manufacturing process, transportation, handling,
storage and cutting of the plates
118
6.10 Measurement of the initial plate distortion 119
6.11 Design and fabrication of a holder-mounting device to hold the plates 120
6.12 Design and fabrication of a square-mounting device 120
6.13 Application of the welding tacks 121
6.14 Distortion measurement after welding tack application 123
6.15 Installation of the run-off tabs 123
6.16 Application of the welding 124
6.17 Removing the run-off tabs from the welded structure 124
6.18 Measurement of the distortion after applying the welding 125
6.19 Measurement of the final distortion 126

Chapter VII RESULTS OF THE EXPERIMENTAL TESTS IN 3-DIMENSIONAL UNITARY CELL-TYPE SPECIMENS

7.2 Configuration of the 3-dimensional unitary cell specimens 127
7.3 Localization of the points of interest in the 3-dimensional unitary cell specimens 128
7.4 Configuration of the experiment 128
7.5 Distortion after applying welding tacks in the 3-dimensional unitary cell specimens 128
7.6 Distortion after welding in the 3-dimensional unitary cell specimens 129
7.7 Distortion modes of the 3-dimensional unitary specimens 132
7.8 Final distortion of the 3-dimensional unitary cell specimens 133
7.9 Final Remarks for distortion of the 3-dimensional unitary cell specimens 135
7.10 Conclusions of the results of the experimental test in 3-dimensional unitary cell specimens 136


xi
Chapter VIII COMPARISON BETWEEN THE 3- DIMENSIONAL UNITARY CELL-TYPE
STRUCTURES/SPECIMENS

8.2 Comparison of distortion modes 137
8.3 Comparison of distortion 137
8.4 Conclusions of the comparison between numerical and experimental results 140
8.5 Procedures to determine the proper welding sequences to reduce residual stress, distortion, or
a relation between them in symmetrical and asymmetrical structures in 2 and 3 dimensions
140
8.5.1 Symmetrical structures in 2 and 3 dimensions 140
8.5.2 Asymmetrical structures in 2 and 3 dimensions 142

Chapter IX CONCLUSIONS, CONTRIBUTIONS, AND SUGGESTIONS FOR FUTURE RESEARCH

9.1 Conclusions 146
9.2 Contributions 147
9.3 Suggestion for future research 147

REFERENCES 149

APPENDIX

1 Plasticity theory applied to welding process and its formulation by finite element method 151
2 Definition and justification of the applied finite elements 171
3 Response to critical comments 179
4 Proper welding sequence to reduce residual stress and distortion in common symmetrical
structures in 2 and 3 dimensions based on the hypothesis developed in the sections 4.9 and 5.2.
188
5 Listing of commands of the numerical simulation of the welding process (I-type specimen
subject to tension)
191
6 Listing of commands of the numerical simulation of the welding sequence in an L-type structure
(Welding sequence No.1)
194
7 Listing of the commands of the numerical simulation of stiffened symmetrical flat frame
(welding sequence No.5 with welding tacks)
197
8 Listing of the commands of the numerical simulation of the 3-dimensional unitary cell-type
symmetrical structure (welding sequence most appropriate to reduce distortion with 24 weld
beads and welding tacks)
201
9 Construction drawings 205


xii

LIST OF FIGURES

CHAPTER II

Figure 2.1 Welded frame distortion [4]: (a) Without considering a proper welding sequence,
(b) considering a proper welding sequence
4
Figure 2.2 Rigid supports [4] 5
Figure 2.3 Precamber with a curved surface [2] 6
Figure 2.4 Pre-bending [2] 6
Figure 2.5 Welding with the thermal tensioning process [11] 6
Figure 2.6 Heat sink welding [1] 7
Figure 2.7 Sequences for thin-wall butt-welds [3]: (a) Progressive, (b) backstep, (c) symmetric, and
(d) jump
8
Figure 2.8 Built-Up welding sequence on thick-wall butt-weld [3] 9
Figure 2.9 Block welding sequence [3] 9
Figure 2.10 Cascade welding sequence [3] 9
Figure 2.11 Welding sequences for thin-wall butt-welds [16]: (a) Progressive, (b) backstep, and (c)
symmetric
11
Figure 2.12 Longitudinal residual stress distribution [16]: (a) Along the X-direction and (b) along the
Y-direction
12
Figure 2.13 Different welding sequence for thick-wall butt-welds [16] 13
Figure 2.14 Residual stresses distribution along the X-direction in various welding sequence for
thick-wall butt-welds [16]: (a) Longitudinal and (b) transverse
14
Figure 2.15 Geometry and various welding sequence for circular patch [16]: (a) Geometry of circular
patch welding, (b) Progressive sequence, (c) backstep sequence, and (d) jump sequence
15
Figure 2.16 Residual stresses distribution for various welding sequence [16]:(a) Circumferential and
(b) radial
16
Figure 2.17 Dimensions of the specimen, mm. [17] 17
Figure 2.18 Schematic of Weld beads delamination [17] 17
Figure 2.19 Comparison of transverse residual stress between in the same direction and welding in
the inverse direction [17]
18
Figure 2.20 Configuration of welded blocks in a multi-block welding sequence [18] 19
Figure 2.21 Structural boundary conditions of welded plates (clamp fixture at both sides) [18] 19
Figure 2.22 Resulted distortion profile for two different block sequences [18]: (a) Welding sequence
No. 1 and (b) welding sequence No. 2
20


xiii
Figure 2.23 Resulted distribution of the Von Mises stress using different block sequences. [18]:
(a) Welding sequence No. 1 and (b) welding sequence No. 2
21
Figure 2.24 Configuration of a large-diameter multi-pass pass butt-welded pipe joints and its cross
section [19]
22
Figure 2.25 Welding sequences for a multi-pass Welded pipe joint [19] 22
Figure 2.26 Comparison of circumferential residual stress in multi-pass welded pipe joints [19]: (a)
Inner surface and (b) outer surface
23
Figure 2.27 Comparison of axial residual stress in multi-pass welded pipe joints [19]: (a) Inner
surface and (b) outer surface
24
Figure 2.28 Comparison of residual stress across through-thickness along the heat-affected zone in
multi-pass welded pipe joints [19]: (a) Circumferential and (b) axial
25
Figure 2.29 Configuration of a multi-pass fillet Weld joint, mm. [20] 27
Figure 2.30 Different welding sequences in multi-pass fillet weld joint [20] 27
Figure 2.31 Comparison of residual stress in a multi-pass fillet weld joint [20]: (a) Case 1 and
(b) case 2
28
Figure 2.32 Relation between nominal stress range and fatigue life in multi-pass fillet weld joints
[20]
29
Figure 2.33 Aluminum panel for study on welding sequence effect on angular distortion [21] 30
Figure 2.34 Welding sequences for angular distortion analysis of aluminum panel structure[21] 31
Figure 2.35 Distortion displacements at four cross sections of the panel from four welding
sequences [21]: (a) At x= 16 inch; (b) at x=32 inch; (c) at y= 9.3 inch; y (d) at y=30 inch.
32
Figure 2.36 Optimum welding sequence determined by JRM [21] 33
Figure 2.37 Comparison of distortion resulting from welding sequence and the optimum welding
sequence [21]: (a) at x= 16 inch, (b) at x=32 inch, (c) at y= 9.3 inch, and (d) at y=30 inch.
33
Figure 2.38 T-joint configuration [22]: (a) Finite element model and (b) dimensions of the cross
section of the hollow extrusions
34
Figure 2.39 Different welding sequences [22]: (a) Case 1 and (b) case 2 34
Figure 2.40 Comparison of measured (M) and calculated (S) distortion evolution for two welding
sequence [22]: (a) Case 1 and (b) case 2
35
Figure 2.41 Geometric configuration of a sub-assembly [15] 36
Figure 2.42 Simplified model with torsion springs [15] 36
Figure 2.43 Final exaggerated deformed shapes for different sequences [15] 37
Figure 2.44 Geometric configuration and dimensions of weldment [23] 38
Figure 2.45 Code to designate the welding sequence and welding direction [23] 39
Figure 2.46 Radial displacement of the edge of plate with respect to for continuous welding and
the optimum sequence[23]
39
Figure 2.47 Tail bearing housing [24] 41
Figure 2.48 Finite element model of welding process and designation of welding paths showing 41


xiv
positive orientation of weld and reference number in sequence [24]
Figure 2.49 Run 28 is the optimized displacement with a clamped structure [24] 42
Figure 2.50 The absolute values of the X displacement for all 28 sequences [24] 42

CHAPTER III

Figure 3.1 Eulerian frame and thermal initial and boundary conditions 49
Figure 3.2 Geometric configuration of I-type specimen subject to tension (mm) [25] 55
Figure 3.3 Thermal and mechanical properties of ASTM A36 as a function of temperature [34] 56
Figure 3.4 Stress-strain behavior of ASTM A36 carbon steel for different temperatures [Based in
figure 3.3]
57
Figure 3.5 Finite element mesh of the I-type specimen subject to tension 58
Figure 3.6 Weld thermal cycle of ASTM A36 carbon steel [34-35] 60
Figure 3.7 Localization of the points of interest on finite element model 62
Figure 3.8 Distribution of residual stresses in the X-direction in the I-type specimen subject to
tension
63
Figure 3.9 Experimental tests: a) Localization of the points of interest in the I-type specimen
subject to tension, mm. b) welding parameters employed [25]
64
Figure 3.10 Geometric configuration of the L-type structure (mm) 65
Figure 3.11 Finite element model of the L-type structure 66
Figure 3.12 Localization of the node 1533 in the L-type structure 68
Figure 3.13 Maximum distortions in L-type structure for three different welding sequences 69
Figure 3.14 Maximum Von Mises residual stress in the L-type structure for three different welding
sequences
69
Figure 3.15 Distortion profile in the L-type structure corresponding to welding sequence 3 70
Figure 3.16 Distribution of Von Mises residual stress corresponding to the welding sequence 2 71
Figure 3.17 Localization of the interest points in the L-type specimen 72
Figure 3.18 L-type specimen 2 after applying the welding sequence 3 73
Figure 3.19 L-type specimen 2 (welding sequence 3) mounted on the coordinate measuring
machine
73
Figure 3.20 Comparison of the welding distortion (exaggerated) between two L-type specimens 75

CHAPTER IV

Figure 4.1 Geometric configuration of the stiffened symmetrical flat frame (mm) 77
Figure 4.2 Finite element model of the stiffened symmetrical flat frame 80
Figure 4.3 Distribution of Von Mises residual stresses corresponding to the welding sequence
(5 I-O)
82


xv
Figure 4.4 Distortion profile corresponding to the welding sequence 5 O-I WT 83
Figure 4.5 Centers of gravity of the structure, centers of gravity of the weld beads, axis of
symmetry and concentric circles in the stiffened symmetrical flat frame
86
Figure 4.6 Ranges of the values of P and W for welding sequence 1 I-O and combined welding
sequence
87
Figure 4.7 Localization of the point of interest in the stiffened symmetrical flat specimen 90
Figure 4.8 Stiffened symmetrical flat specimen after applying the welding sequence 5 O-I WT 91
Figure 4.9 Stiffened symmetrical flat specimen mounted on the coordinate measuring machine 91
Figure 4.10 Distortion profile in the stiffened symmetrical flat specimen corresponding to welding
sequence 5 O-I WT (exaggerated)
93

CHAPTER V

Figure 5.1 Panel used in the welding industry 97
Figure 5.2 Configurations and dimensions of the 3-dimensional unitary cell (mm) 98
Figure 5.3 Axis of symmetry and concentric spheres of the 3-dimensional unitary cell formed by 24
fillet welds
99
Figure 5.4 Different welding sequences for the 3-dimensional unitary cell 100
Figure 5.5 Fillet weld shape used in the 3-dimensional unitary cell 101
Figure 5.6 Finite element model of the 3-dimensional unitary cell 102
Figure 5.7 Localization of the points of interest in the 3-dimensional unitary cell 104
Figure 5.8 Distribution of residual stresses corresponding to numerical simulation 2 105
Figure 5.9 Isometric view of the distortion profile corresponding to the numerical simulation 3 106
Figure 5.10 Configuration of the special welding sequence 108

CHAPTER VI

Figure 6.1 Semiautomatic welding machine OPTYMAG 501 112
Figure 6.2 Typical zone of good short circuit welding conditions [37] 113
Figure 6.3 Contact tip to work distance [40] 114
Figure 6.4 Positioning of electrode gun with respect to the base metal plate [37] 114
Figure 6.5 Normal work angle for fillet welds [37] 115
Figure 6.6 Arm-holder device 116
Figure 6.7 Welding torch holder device 117
Figure 6.8 Localization of the locks mounted on the traveler carriage 118
Figure 6.9 Electric oven used to initial residual stresses relief 119
Figure 6.10 Measurement of the initial distortion with standard gages 119
Figure 6.11 Holder-mounting device 120


xvi
Figure 6.12 Square-mounting device 121
Figure 6.13 C-type clamps mounted on the specimen 122
Figure 6.14 Welding tacks application in the specimen 122
Figure 6.15 Installation of the run-off tabs on the specimen 123
Figure 6.16 Application of the welding to the specimen 124
Figure 6.17 Removing of the run-off tabs of the specimen 125
Figure 6.18 Measurement of the distortion after applying welding using standard gages 125

CHAPTER VII

Figure 7.1 Distortion (exaggerated) after applying welding tacks and after welding for specimen T9 131
Figure 7.2 Distorted shape of the 3-dimensional unitary cell welded specimens 132
Figure 7.3 Final distortion (exaggerated) for welded specimen T9 134

CHAPTER VIII

Figure 8.1 Comparison between numerical (simulation 3) and experimental distortion (specimen
T9). Exaggerated distortion.
139
Figure 8.2 Flow Diagram of the analysis of the welding sequence for symmetrical and asymmetrical
structures in 2 and 3 dimensions
144


xvii

LIST OF TABLES
CHAPTER II

Table 2.1 Different welding sequences used by Ji and Fang [17] 17
Table 2.2 Residual stresss peak value for several cases [17] 18
Table 2.3 Welding sequences and orders applied into the multi-block model [18] 19
Table 2.4 Different welding sequence used by Bart, Deepak and Kyoung [15] 36
Table 2.5 for various welding conditions [23] 40

CHAPTER III

Table 3.1 Chemical composition of ASTM A36 carbon steel [33] 56
Table 3.2 Residual stresses obtained in the numerical simulation of I-type specimen subject to
tension
63
Table 3.3 Comparison between numerical data and experimental data of two I-type specimen
subject to tension
64
Table 3.4 Welding sequence configuration in the L-type specimen 68
Table 3.5 Distortion at the points of interest in two L-type specimens 74

CHAPTER IV

Table 4.1 Welding sequences from the inside to the outside used in the stiffened symmetrical flat
frame
78
Table 4.2 Welding sequences from the outside to the inside used in the stiffened symmetrical flat
frame
79
Table 4.3 Order of importance of the analyzed welding sequences in the stiffened symmetrical flat
frame
85
Table 4.4 Configuration of welding sequence 5 O-I WT used in the stiffened symmetrical flat
specimen
90
Table 4.5 Distortion at the points of interest (mm) in the stiffened symmetrical flat specimen
corresponding to welding sequence 5 O-I WT
92

CHAPTER V

Table 5.1 Configuration of the numerical simulations for the 3-dimensional unitary cell 104
Table 5.2 Distortion at 24 points of interest (mm) corresponding to different welding sequences
In the 3-dimensional unitary cell
107


xviii
Table 5.3 Comparison of numerical results between both welding sequences In the
3-dimensional unitary cell
108

CHAPTER VI

Table 6.1 Electric settings used in the short circuit transfer mode [39] 113
Table 6.2 Common blends of shielding gas composition for short transfer mode [40] 115

CHAPTER VII

Table 7.1 3-dimensional unitary cell specimen configuration 128
Table 7.2 Distortion after application of the welding tacks for the different 3-dimensional unitary
cell specimens, mm.
129
Table 7.3 Distortion after welding for the different 3-dimensional unitary cell specimens 24 hrs
after welding, mm.
130
Table 7.4 Final distortion in the 24 points of interest of the 3-dimensional unitary cell specimens,
mm.
133

CHAPTER VIII

Table 8.1 Difference (%) between numerical and experimental results 138


xix
NOMENCLATURE

Mathematical symbols Units

Rectangular matrix -
Column vector, row vector -
Matrix transpose -
Matrix inverse -
,
Partial differentiation if the following subscript is a letter -

Latin symbols

[ ]
B
Strain-displacement matrix for each element -
C Specific heat J/KgC
[ ]
C

Damping matrix Kg/s

Power of viscoplastic straining 1/s

Increment of; for example ,
Tot
ij
d
-
ep
D

Elastic-plastic stiffness matrix -
e
ij
e
Deviation component of elastic strain tensor
E Youngs modulus N/m
ijkl
E
2

Elastic tensor N/m
ep
ijkl
E
2

Elastic-plastic tensor N/m
t
ijkl
E
2

Tangent elastic tensor N/m
i
f
2

Sum of the body force N

Elastic forces N

Inertia forces N
Damping forces N
( ) 0 f =
Yield surface -
{ }
e
F
Element load vector N
{ } F
Global load vector N


xx
{ }
t
F
Global load vector corresponding at time t N
{ }
i
t t
F
+
Nodal load vector corresponding to the state of stress for the time t -
( ) 2 1
E
G
v
=
+

Shear modulus N/m
( ) 0 G =
2

Plastic potential -
h Convection heat transfer coefficient
c

W/m
2
H
C

Hardening modulus -
i Iteration number -
k
x
, k
y
, k Thermal conductivity in the x, y and z directions
z
W/mC
( ) 3 1 2
E
K
v
=

Bulk modulus N/m
[ ]
K
2

Global stiffness matrix N/m
[ ]
e
K
Element stiffness matrices N/m
t
K

Stiffness matrix in the time N/m
[ ]
M

Mass matrix Kg
n

Number of nodal points for each element -
( , , )
i
N x y z Shape functions -
N
x
, N
y
, N Cosine directors
z
-
q Boundary heat flux
s
W/m
Q(x,y,z,t)

2

Source of heat generation W/m
{ }
t
R
3

Nodal point force corresponding at time t -
{ } s
Stress deviation increment vector -
s Stress deviation tensor
ij
N/m
T(x,y,z,t)
2

Current temperature C
Prescribed surface temperature C
T Surrounding temperature C
m
T
Melting temperature C
{ } dT
Thermal dilatation vector -
t Time interval sec
cr
t

Critical time step sec

Displacement vector m

Velocity vector m/s

Acceleration vector m/s
2


xxi
, i j
u Displacement gradient -
( )
r
U
Radial distortion with respect to mm
Maximun radial distortion mm
{ } U
Element nodal increments -
{ }
i
U
Nodal increment vector in the iteration -
{ } U
Admissible virtual nodal increase -
V Volume of the body m

3

Plastic work

Greek symbols

Thermal expansion coefficient m/m C
i

Parameters that control the yield surface size -
Kronecker symbol
ij
-
Emissivity
Cauchy strain tensor
Power of elastic straining -
e
ij
d
Elastic strain increment
e
kk

Spherical component of elastic strain tensor
p
ij
d
Plastic strain increment
tr
ij
d
Phase transformation strain increment
ij
d

Thermal strain increment
p
d
Equivalent strain increment
ij
d Variation in the strain increment

Parameter related with the strain hardening effects -
d Plastic multiplier -
v Poissons ratio -
Density Kg/m

3

Stefan-Boltzmann constant
Cauchy stress tensor
ij
N/m
2

Hydrostatic stress tensor
kk
N/m

2

Spherical invariant N/m

2

Radial residual stress N/m
2


xxii

Von Mises stress N/m
or
2

Transverse residual stresses N/m
2

Yield stress
y
N/m

2

Longitudinal residual stress N/m

2

Axial residual stress N/m
or
2

Circumferential residual stress N/m
2


1

CHAPTER I

INTRODUCTION

1.1 Background

During the heating and cooling cycle in the welding process, thermal strain occurs in the filler material and
in the base metal regions close to the weld. The strain produced during heating is accompanied by plastic
deformation. The non-uniform plastic deformation that occurs in the weld structure is what leads to residual
stresses. These residual stresses react to produce internal forces which must be equilibrated and cause
distortion [1].

The residual stress and distortion in weldments depend on several interrelated factors such as thermal
cycle, material properties, structural restraints, welding conditions and geometry [2]. Of these parameters, the
thermal cycle has the greatest influence on the thermal loads in the welded structures. At the same time, the
temperature distribution is a function of parameters such as welding sequence, welding speed, energy of the
source, and environmental conditions.

A high level of tensile residual stresses near the seam can induce brittle fracture, cracking due to corrosion
stress, and reduced fatigue strength. Compressive residual stresses in the base metal located some distance
away from the weld line can substantially decrease the critical buckling stress [3]. The main effects of
distortion are the loss of tolerance in the welded components and deformation of structural elements that
results in inadequate support to transfer applied loads [4]. Therefore, residual stresses and distortion should
be reduced to meet all geometry and strength requirements.

Some of the most popular methods for reducing residual stresses and distortion in weld fabrication are:
welding sequence, definition of weld parameters, definition of weld procedure, use of precambering fixtures,
prebending, thermal tensioning, heat sink welding, post-weld treatment, control of weld consumables, and
post-weld corrective methods [1].


2
1.2 General objective

Develop a welding sequence-based methodology to reduce residual stresses, distortion, or a relation
between both parameters in symmetrical structures.

1.3 Specific Objectives

Obtain temperature-dependent material properties of ASTM A36 steel used in this investigation.
Conduct a literature survey on welding sequences.
Investigate the theory of plasticity applied to the welding process and its finite element formulation.
Perform a finite element simulation of the welding process through a thermo-mechanical analysis
based on the von Mises criterion and flow rule, assuming a lineal isotropic hardening and temperature
dependent materials while neglecting micro-structural evolution.
Validate the proposed numerical model of the welding process with experimental data to determine
model accuracy.
Apply the proposed numerical model of the welding process to determine whether the welding
sequence in an L-type structure affects the residual stresses and distortion.
Apply the proposed numerical model of the welding process in a welding sequence analysis in 2 and 3
Analyze the effects of the welding sequence on residual stresses and distortion in 2 and 3 dimensional
symmetrical structures.
Analyze the relationship between residual stress and distortion due to the welding sequence in 2-
Determine the proper welding sequence to reduce residual stresses, distortion, or a relation between
both parameters in 2 and 3 dimensional symmetrical structures.
Verify if the proper welding sequence for 2-dimensional symmetrical structures is applicable to 3-
dimensional symmetrical structures, considering appropriate modifications.
Develop a methodology for experimental tests.
Perform experimental tests of the proper welding sequence to reduce distortion in a 2-dimensional
symmetrical structure.
Compare the numerical and experimental results on the proper welding sequence to reduce residual
stresses or distortion in a 3-dimensional symmetrical structure.


3

CHAPTER II

WELDING SEQUENCE BACKGROUND AND
METHODS FOR CONTROLLING RESIDUAL STRESSES AND
DISTORTION INDUCED BY WELDING

2.1 Introduction

This chapter presents the methods and advantages of reducing or controlling residual stresses and
distortion induced by the welding process as well as a definition and classification of the welding sequence. It
also includes a study of the welding sequence analysis, as well as a summary of previous welding sequence
research, and a discussion of its advantages, disadvantages, scope, and limitations.

2.2 Advantages of residual stress and distortion control

Controlling the residual stress and distortion in weldments provides two main advantages: (1) reduced
fabrication costs by minimizing or controlling distortion, and (2) increased service life of the welded structure
by controlling the induced residual stress.

The benefits of distortion control are [1]:

1. Eliminate the need of expensive distortion correction and loss of accuracy.
2. Reduce machining requirements
3. Improve quality

And residual stress control benefits are no less important [2]:

1. Maximize fatigue performance.
2. Minimize costly service problems.
3. Improve resistance to environmental damage.


4
2.3 Methods to control welding-induced residual stress and distortion

Some of the most popular methods used in the industry to control welding-induced residual stress and
distortion include: welding sequence, definition of weld parameters, weld procedure, fixture design,
precambering, prebending, thermal tensioning, heat sink welding, post-weld heat treatment, control of weld
consumables and post-weld corrective methods [2].

2.3.1 Welding sequence

The proper welding sequence can minimize distortion and affect the distribution of the residual stress [4].
Figure 2.1 shows two welded frames. In the first frame (a) a proper welding sequence was not performed and
a large distortion was produced. The second frame (a) shows a proper welding sequence which leads to less
distortion.

Figure 2.1 Welded frame distortion [4]: (a) without considering a proper welding sequence,
(b) considering a proper welding sequence

2.3.2 Definition of weld parameter

This method is based on control of weld parameters such as heat input, weld groove geometry, single-pass
versus multiple pass welds, and type of joint [5]. The input heat is the most influential parameter in weld-
induced distortion. Reducing welding heat input decreases all kinds of weld-induced distortions [2]. The heat
input can be controlled through the welding speed and weld size. Faster welding not only reduces the amount
of adjacent material affected by the heat of the arc, but also progressively decreases the residual stress. The
important difference lies in the fact that faster welding produces a slightly narrower isotherm. The width of
the isotherm influences the transverse shrinkage of butt welds, explaining why faster welding generally result
in less residual stress [6]. When the specimen thickness decreases, the weld size also decreases, and a
reduction in the volume of weld metal usually results in less residual stresses and distortion. However, when


5
the specimen thickness decreases, the tensile residual stress in the areas near the fusion zone and distortion
increase [6]. This is because thin weldments absorb more energy per unit volume. The use of small groove
angles and root openings decrease the volume of weld metal, resulting in lower transverse shrinkage. For
example, the use of a U-groove instead of a V-groove should reduce the amount of weld metal [3]. Welding is
frequently performed in one pass, especially for thin plates. However, when welding is performed in multiple
passes, particularly when welding thick plates, shrinkage accumulates [3].

2.3.3 Weld procedure

Welding procedures have considerable effect on distortion. Fusion welding often leads to the largest
distortion, while laser (LBW), electron beam (EBW), and stir welding (FRW) result in lower distortion. However,
friction stir welding can impart large plastic strains to the structure. These large strains, which locally strain-
harden the material, can influence the fracture response of the structure [8]. In fusion welding, residual stress
distributions are similar; this is true when the design and relative size of the weldments are also similar. The
most important fusion welding processes are Shield Metal Arc Welding (SMAW), Gas Metal Arc Welding
(GMAW), Submerged Arc Welding (SAW), and Gas Tungsten Arc Welding (GTAW). However, the automatic or
semiautomatic welding processes present more advantages in the residual stresses and distortion control due
to their repeatability.

2.3.4 Fixture design

This method is based on the design of clamps, jigs and rigid supports that restrain displacements and
rotations of some portions of the welded components or the complete structure. However, the use of these
devices increases residual stresses [7,9-10]. Figure 2.2 shows a rigid support formed by a back plate and two
clamps. The clamps restrain the angular distortion of the welded joint.

Figure 2.2 Rigid supports [4]

2.3.5 Precambering

This method consists on elastically bending some of the components (usually in a specially designed fixture)
in a predefined manner before welding. After welding, the precamber is released and the fabricated structure


6
springs back to minimally distorted shape [2]. Figure 2.3 shows a precamber with a curved surface. The
structure is fixed to the device by clamps.

Figure 2.3 Precamber with a curved surface [2]

2.3.6 Prebending

This method consists of plastically bending some of the components before welding, and possibly before
placing them in a fixture. After welding, the desired non-distorted shape result. The welding is performed
with or without a fixture [2]. Figure 2.4 shows a prebending in a fillet weld.

Figure 2.4 Pre-bending [2]
2.3.7 Thermal tensioning

This method consist on strategically moving a heat source ahead of, beside, behind, (or combinations of
these) the moving weld torch. This method can control distortions and residual stresses during welding by
controlling the heating and cooling rates [11,12]. Figure 2.5 shows thermal tensioning welding.

Figure 2.5 Welding with the thermal tensioning process [11]


7
2.3.8 Heat sink welding

This method is similar to thermal tensioning, except that a cooling source is strategically moved (or kept
stationary) during welding (Figure 2.6).

Figure 2.6 Heat sink welding [1]

2.3.9 Preheating

Preheating the components being welded reduces residual stresses and distortion by reducing thermal
gradients around the weld bead. Preheating has beneficial effects when welding steels by reducing cracks in
the heat affected zones and weld metal [3].

2.3.10 Post-weld heat treatment

Heating all or parts of the welded fabrication to high temperatures (depending on the material) for a period
on time may relieve welding stresses. Often the stresses cannot be fully relieved, i.e., some level of residual
stress remains. This method is expensive and is often used to prevent service fracture problems such as
corrosion, fatigue, creep, or combinations of them.

2.3.11 Post-weld corrective methods

Corrective methods may reduce distortion or residual stresses in a welded component. Corrections made
after weld are often expensive and time consuming. The most important post-weld corrective methods are
press straightening, shot peening, laser shock peening, vibratory stresses relief, and hammer peening [3].

Remark: The control methods previously described can increase the production costs due to energy
consumption, time, and/or expensive equipment. Other methods slow down production by requiring fixture
devices.

Welding sequence is inexpensive because it directly affects the temperature field of the welded
structure, and consequently the residual stresses and distortion. Therefore, sequence analysis is
fundamental for controlling residual stresses and distortion in welded structures.


8
2.4 Welding sequence definition

The American Welding Society (AWS) defines welding sequence as the order of making welds in a
weldment [13].

2.5 Welding sequence classification

Welding sequences are classified by the number of passes: single pass and multiple pass weld sequences.
However, single pass sequences can be applied to multiple pass welds between beads [3].

2.5.1 Welding sequence for single pass welds

For thin components (up to inch) welding is performed in a single pass [3]. The weld bead is divided in
short sections and welded considering the order and direction of the welds [3,14]. The more common welding
sequences for single pass welds are: progressive, backstep, symmetric, and jump (Figure 2.7).

Figure 2.7 Sequences for thin-wall butt-welds [3]: (a) Progressive, (b) backstep, (c) symmetric, and (d) jump

Figure 2.7 (a) corresponds to progressive welding, where the weld beads are set down continually from one
end of the joint to the other. In the backstep sequence (Figure 2.7 b), the weld beads are deposited in the
opposite direction to the welding progress. Figure 2.7 (c) is the symmetric welding sequence, where the weld
beads are deposited from the axis of symmetry of the joint. In the jump welding sequence (Figure 2.7 d) the
weld beads are deposited in intermittent form.


9
2.5.2 Welding sequence for multiple pass welds

For thick components (over 1/4 inch), welding sequences are classified as [3]:

Built-Up: The first layer is completed along the entire weld length through the previously described
single pass sequences (progressive, backstep, symmetric, jump sequences, etc.), followed by the
second, third, etc. (Figure 2.8). This sequence applies to large-diameter butt-welded pipe joints
frequently used in boiling water reactors, oil pipe transport systems, and steam piping systems.

Figure 2.8 Built-Up welding sequence on thick-wall butt-weld [3]

Block welding: A given block of the joint is welded completely and then the next block is welded, and
so on. This kind of welding sequence is applied mainly to very long joints (e.g., ship hulls). Figure 2.9
shows the block welding sequence on thick-wall butt-weld, where first the end blocks of the joint are
welded in, and later the central block is added to the joint.

Figure 2.9 Block welding sequence [3]

Cascade welding: It is similar to the block welding sequence; the main difference is that the ends of the
blocks overlap. An application of this sequence is welding of long thick plates (Figure 2.10).

Figure 2.10 Cascade welding sequence [3]


10

2.6 Welding sequence selection based on empirical rules

For complex geometries, several empirical rules useful to decide the welding sequence have been
introduced [15]:

Rule 1. The weld bead closest to the previous can be selected next.

Rule 2. The weld bead farthest from the current can be next.

Rule 3. Weld beads with greater restraint should be chosen next.

Rule 4. Weld beads symmetric to the neutral axis are selected next.

Rule 5. Weld beads originate from the center points of a structure progressing outwards.

Rule 6. Weld beads that are not adjacent to the current can be next.

Optimizing welding productivity demands minimization of the torch moving distance between weld beads,
as in rule 1. However, rule 1 is not appropriate for welding quality because successively welding close beads
can generate a very high heat flux that results in serious thermal distortion. Rules 2, 3 and 4 can improve
welding quality at the expense of welding time. To tackle both issues simultaneously, rules 5 and 6 are
introduced.

Remark: The authors in [15] do not clearly define "weld quality, nor do they mention what specific weld
parameters improve or worsen with the previously mentioned algorithms.


11

2.7 Welding sequence background

Teng and Peng [16] investigated the reduction in residual stresses caused by welding by analyzing the
effects of welding sequence on residual stress distribution in single and multi-pass butt-welded plates and
circular patch welds. The research was conducted through finite element-based thermo elastic plastic analysis
and simulated weld thermal cycles. The test specimen dimensions and the different welding sequences used
in the thin-wall butt-weld analysis are shown in figure 2.11. The authors worked with three different welding
sequences: (1) progressive, (2) backstep, and (3) symmetric.

Figure 2.11 Welding sequences for thin-wall butt-welds [16]:
(a) Progressive, (b) backstep, and (c) symmetric

Figure 2.12 (a) and (b) show the distribution of the longitudinal residual stresses along the X-direction (at
Y=150 mm) and Y-direction obtained with progressive, backstep, and symmetric sequences. These figures
reveal that the symmetric sequence produced the lowest residual stresses.


12

Figure 2.12. Longitudinal residual stress distribution [16]:
(a) Along the X-direction and (b) along the Y-direction


13

Reference [16] also considered butt-welded thick plate joints (figure 2.13). Three different cases are
considered. Case (A): welding half of the upper groove, the whole lower groove and then the remaining upper
groove. Case (B): welding half of the upper groove, half of the lower groove, the remaining of the upper
groove and then the remaining lower groove. Case (C): welding the whole lower groove before the whole
upper groove.

Figure 2.13 Different welding sequence for thick-wall butt-welds [16]

Figure 2.14 (a) and (b) depict the distribution of the longitudinal and transverse residual stresses obtained
with various types of welding sequences. Longitudinal residual stresses between various welding sequences
did not appear to differ significantly. However, the transverse residual stresses of case (A) were smaller than
those of the other welding sequences. This difference might be attributed to two reasons: (1) the symmetric
welding sequence can reduce the residual shrinkage or (2) the symmetric welding sequence has pre-heating
and post-heating effects.


14

Figure 2.14 Residual stresses distribution along the X-direction in various welding sequences
for thick-wall butt-welds [16]: (a) Longitudinal and (b) transverse


15
Finally, the effect of sequences on residual stresses for circular plates is reported [16]. Figure 2.15 shows
the various welding sequences for circular patch welds.

Figure 2.15 Geometry and various welding sequence for circular patch [16]:
(a) Geometry of circular patch welding, (b) Progressive sequence,
(c) backstep sequence, and (d) jump sequence

Figure 2.16 (a) depicts the distribution of circumferential residual stresses and reveals that the various
welding sequences do not appear to differ significantly. Figure 2.16 (b) depicts the distribution of radial
residual stress and reveals that the backstep sequence has smaller radial residual stresses than the other
welding sequences. This is because the post-weld treatment and the pre-heating effect of backstep sequence
are better than in the other welding sequences.

Remark: Reference [16] is applicable only to simple structures, and the numerical simulations do not
consider the welding direction. No experiments were conducted for validating the numerical results for the
different welding sequences.


16

Figure 2.16 Residual stress distribution for various welding sequences [16]:
(a) Circumferential and (b) radial


17
Ji and Fang [17] investigated the influence of welding sequence on the residual stresses of a thick plate.
Authors worked with double V-groove multiple-pass butt-welds and adopted the converse welding method
between adjacent layers, or between adjacent weld beads in every layer. They analyzed a coupled thermo-
mechanical model using finite element and an ellipsoidal heat source. The numerical results were validated
against experimental results (the x-ray method). Figure 2.17 depicts the dimensions of the analyzed specimen
and the double V-groove configuration. In the numerical simulation the weld was divided into nine layers
(figure 2.18).

Figure 2.17 Dimensions of the specimen, mm. [17] Figure 2.18 Schematic of Weld beads
delamination [17]

Table 2.1 displays several welding sequences analyzed to study the effect on residual stresses. The
converse welding method, consisting of applying the opposite direction between adjacent layers in multi-layer
welds, or between beads in every layer [17], was adopted between adjacent layers.

Welding sequence
Case (A) 2 3 1 4 8 5 9 6 7
Case (B) 2 3 4 1 5 8 6 9 7
Case (C) 2 3 1 4 5 8 6 7 9
Case (D) 1 2 8 3 9 4 5 6 7
Case (E) 1 2 3 8 4 5 9 6 7
Case (F) 1 2 3 8 9 4 5 6 7
Case (G) 1 2 3 4 5 8 6 9 7
Case (H) 2 3 4 5 1 6 8 7 9

Table 2.1 Different welding sequences used by Ji and Fang [17]

Table 2.2 shows the resulting peak value of the residual stresses. It can be seen from the table that the
peak values of transverse residual stress or longitudinal residual stress obtained from case (C) are lowest. This


18
is because the filler metal is being more uniformly applied, the angular deformation is minimized, and the
residual stress is low.

Parameter Case
(A) (B) (C) (D) (E) (F) (G) (H)
Equivalent residual stress (MPa) 701 712 571 885 653 658 921 912
Transverse residual stress (MPa) 623 544 405 718 505 511 829 857
Longitudinal residual stress (MPa) 634 636 507 823 592 607 879 865

Table 2.2 Residual stresss peak value for several cases [17]

To validate if the residual stress obtained by the converse welding method between adjacent layers is the
minimum, the authors studied the welding stress of the double V-groove plate under classic and converse
welding. The simulation results show that the peak value of the transverse residual stress by the method of
converse welding is 18.2 % less than the stress obtained by the method of classic welding (Figure 2.19).
Longitudinal residual stresses behave similar to the results shown in figure 2.19, but the peak value decreased
by 16.9%. When converse welding is adopted, the residual stresses have an opposite distribution to the
previous weld bead, reducing the residual stress value. Therefore, the beads produce a relatively uniform
residual stress distribution and a lower residual stress value.

Figure 2.19 Comparison of transverse residual stress between welding in the
same direction and welding in the inverse direction [17]

Remark: An important disadvantage of [17] is the computational cost to simulate the welding process
because the model considered the effects of phase transformation and the type of heat source.


19

Nami, Kadivar and Jafarpur [18] studied the welding sequence in multiple blocks for the effect on the
thermal and mechanical response of thick plate weldments by the use of a 3-D thermo-viscoplastic model.
Anands viscoplastic model was used to simulate the rate dependent plastic deformation of welded materials.
Also, they considered the temperature dependence of thermal and mechanical properties of material, welding
speed, welding lag, and the effect of the filling material added to the weld. The model was compared with the
results of two analytical and experimental works.

Figure 2.20 depicts the configuration and dimensions of the welded blocks. The length of the welded strip
was divided into seven parts and welded by different sequences. The arc was allowed to move in a forward
(+X3) or in a backward direction (-X3). Figure 2.21 depicts the structural boundary conditions of the welded
plates.

Figure 2.20 Configuration of welded blocks in Figure 2.21 Structural boundary conditions of welded
a multi-block welding sequence [18] plates (clamp fixture at both sides) [18]

The selected welding sequences in table 2.3 are commonly used in practice. In the first sequence the
joining was done inwardly (toward the center of the plates) and in second sequence the joining happened
toward the edge of the plates (outwardly).

Table 2.3 Welding sequences and orders applied into
the multi-block model [18]

Welding sequence Configuration
1 +1, +7, +2, +6, +3, +5, -4
2 +2, +4, +6, +1, +3, +5,+7


20
Figure 2.22 shows the final distortion profile in weldments, indicating the magnitude and location of the
maximum plate distortion. As can be seen, the maximum distortion in sequence 1 is greater than in sequence
2.

Figure 2.22 Resulted distortion profile for two different block sequences [18]:


21
Figure 2.23 shows two very different residual stress contours. The magnitude of the maximum von Mises
stress that has been generated by sequence No.2 is greater than produced by sequence No.1. A large stress
variation occurs in the region close to the welding pattern. In the region between the welding pattern and the
edges, a small variation in the stress value is observed. In the region close to the edges, the boundary
conditions introduce large residual stresses with sharp variations.

Figure 2.23 Distribution of the Von Mises stress using different block sequences. [18]:

Remark: Reference [18] does not research the effects of the welding sequence on the relationship
between distortion and residual stresses. The proposed methodology is computationally expensive because
it considers visco-plastic effects.


22
Mochizuki and Hayashi [19] investigated the residual stress in large-diameter, multi-pass, butt-welded pipe
joints for various welding sequences. The pipe joints had an x-shaped groove. The mechanism that produces
residual stress in the welded pipe joints was studied in detail using a simple prediction model. The authors
worked with a thermo-elastic-plastic analysis using finite element method with an axisymmetric model. Also,
they determined an optimum welding sequence for preventing stress-corrosion cracking from the residual
stress distribution.

The configuration of a large-diameter, multi-pass, butt-welded pipe joint and its cross section is shown in
figure 2.24.

Figure 2.24 Configuration of a large-diameter multi-pass Figure 2.25 Welding sequences for a multi-pass
pass butt-welded pipe joints and its cross section [19] Welded pipe joint [19]

The authors [19] proposed six welding sequences (figure 2.25) to study the dependence of the residual
stress on the welding sequence. In case 1 the inner side of the groove is welded before the outer surface of
the groove. In case 2, the outer side of the groove is welded before the inner surface of the groove. In case 3,
half of the inner side of the groove is welded, then the whole outer side, and later the remaining inner side
groove. In case 4, half of the outer side of the groove is welded, then the whole inner side of the groove, and
later the remaining outer groove. In case 5, half of the inner side of the groove is welded, then half of the
outer groove, later the remaining inner groove, and lastly the remaining outer groove. In case 6, half of the
outside groove is welded, then half of the inside groove, later the remaining outside groove and at the end the
remaining inside groove.

Figure 2.26 (a) and (b) show a comparison between the circumferential residual stresses on the inner and
outer surfaces of the groove; Figure 2.27 (a) and (b) show a comparison of the axial residual stresses on the
inner and outer surfaces of the groove; and Figure 2.28 (a) and (b) show a comparison between the
circumferential and axial residual stresses through the plate thickness along the heat-affected zone. All these
figures consider multi-pass, butt-welded, pipe joints.


23

Figure 2.26 Comparison of circumferential residual stress in multi-pass welded
pipe joints [19]: (a) Inner surface and (b) outer surface


24

Figure 2.27 Comparison of axial residual stress in multi-pass welded
pipe joints [19]: (a) Inner surface and (b) outer surface


25

Figure 2.28 Comparison of residual stress across through-thickness along the heat-affected zone
in multi-pass welded pipe joints [19]: (a) Circumferential and (b) axial


26

In figures 2.26 (a), 2.26 (b) and 2.28 (a), the circumferential residual stresses behaved similarly: tensile
circumferential stresses are distributed near the welding deposit on the inner and outer surfaces. The
maximum stress occurs near the welded metal. Tensile stress then decreases and finally becomes compressive
about 40 mm from the center of welded metal.

In figures 2.27 (a), 2.27 (b) and 2.28 (b), the distribution of the axial residual stress near the welding deposit
differs depending on the welding sequence. Stresses in the heat-affected zone vary with the welding
sequence, both on the surface and through the thickness, but the axial residual stress distribution away from
the welded metal is not affected by the welding sequence.

Through-thickness axial residual stresses along the heat-affected zone have a big influence in the
generation and propagation of stress-corrosion cracking in multi-pass, welded pipe joints. The inner surface is
exposed to a more severe environment than the outer surface because the pipe may contain corrosive
substances. There are two steps in selecting an optimum welding sequence for preventing stress-corrosion
cracking: (i) lowering the axial residual stress on the inner surface along the heat-affected zone, because crack
generation should be prevented first; and (ii) lowering the through-thickness axial stress near the inner surface
to reduce or eliminate crack propagation rate, even if a crack begins to propagate.

According to figure 2.28 (b), cases 2, 3, and 6 are good candidates since they produced lower stresses on
the inner surface of the heat affected zone. Among these, case 6 was the best because the axial through-
thickness stress near the inner surface is almost zero up to a depth of 6 mm. This welding sequence should
have the lowest probability of generating and propagating stress-corrosion cracking.

Remark: In the work performed by Mochizuki and Hayashi [19], the analytical method proposed to
determine the residual stresses through-thickness is only applicable to multi-pass, welded pipe joints.
Therefore, the method is not valid for pipe joints of small diameter (single pass joints). The method
presented has a good qualitative correlation with experimental and numerical data. However, quantitative
correlation was not good.


27
Mochizuki, Hattori and Nakakado [20] studied the effect of residual stress on fatigue strength at a weld toe
in a multi-pass fillet weld joint. The residual stress in the specimen was varied by controlling the welding
sequence. They calculated the residual stresses by thermo-elastic-plastic analysis and compared them to strain
gage and X-ray diffraction measurements.

A weld joint was fabricated to evaluate the residual stress and fatigue strength. Two attachments were
fillet-welded on both sides of a main plate, as shown in figure 2.29.

Figure 2.29 Configuration of a multi-pass fillet Weld joint, mm. [20]

Two joints were fabricated by changing the welding sequence, as shown in figure 2.30. In case 1, the final
welding pass was set down in the attachment side, and in case 2 the final welding pass was set down in the
main plate side.

Figure 2.30 Different welding sequences in multi-pass fillet weld joint [20]

Figure 2.31 depicts the experimental and numerical results for transverse residual stresses for the two
welding sequences. The measured and analytical distributions of residual stress agree well. Therefore, the
results from the thermo-elastic-plastic analysis were used to define the residual stress needed to evaluate
fatigue strength. The transverse residual stress in the weld toe of the main plate was 170 MPa for the weld


28
joint whose final welding pass was deposited on the attachment side (case 1), and 80 MPa for the joint with
the final pass on the main plate (case 2).

Figure 2.31 Comparison of residual stress in a multi-pass fillet
weld joint [20]: (a) case 1 and (b) case 2


29
Figure 2.32 compares the fatigue strength curves for the two multi-pass fillet weld joints. The vertical axis
shows the nominal stress range along the loading direction, y, and the horizontal axis shows the number of
cycles to failure, Nf. It was confirmed from observation during the fatigue test that the initial surface crack
nucleated at the center of the weld toe and propagated as a semi-elliptical crack. The fatigue strength
resulting from the two welding sequences was nearly the same in the low cycle range. A clear difference
appears around 10
5

Figure 2.32 Relation between nominal stress range and fatigue
life in multi-pass fillet weld joints [20]

Remark: In the work presented by Mochizuki, Hattori and Nakakado [20], the methodology is valid for
simple fillet joints only. Two similar welding sequences were used, differing only in the order of application
of two weld beads.
cycles, indicating that high cycle fatigue strength can be improved by varying the welding
sequence. Therefore, the welding sequence corresponding to case 2 is better for multi-pass, fillet weld joints.


30

Tsai and Park [21] studied the distortion mechanisms and the effect of the welding sequence on panel
distortion. In this study, distortion behaviors, including local plate bending and buckling, as well as global
girder bending, were investigated using finite element analysis. It was found that buckling does not occur in
structures with a skin-plate thickness of more than 1.6 mm, unless the stiffening girder bends excessively. They
applied the joint rigidity method (JRM) to determine the optimum welding sequence for minimum panel
warping. The JRM consists of starting with more rigid joints and progressively moving toward less rigid joints
[21].

Figure 2.33 shows the geometrical configuration of the panel structure. This panel is formed by one skin
plate, three longitudinal, and three transverse T-stiffeners. The welding sequence simulation includes: i) laying
tack welds along the joints and, ii) laying structural welds at various joints with different sequences.

Figure 2.33 Aluminum panel for study on welding sequence effect on angular distortion [21]

Four welding sequences were investigated in this study (figure 2.34). Sequence No. 1 deposits the weld
from the inner panels moving outward. Sequence No. 2 lays the weld from the outer panels moving inward.
Sequences No. 3 and 4 are respectively similar to sequences 1 and 2. Sequence No. 3 searches for the joint


31
with the highest constraint to deposit the next weld as the welding process progresses. Sequence No. 4 lays
the next weld at the least constrained joint.

Figure 2.34 Welding sequences for angular distortion analysis of aluminum panel structure [21]

Figure 2.35 shows the vertical displacement along the cross section of the four panels for the four welding
sequences being investigated. The origin of the coordinates is at the lower left corner (point A in figure 2.34).
The global distortion of the panel in all cases shows a downward movement and tilting toward the
unsupported corner due to the structural weight. The high peak values in the displacement curves indicate
the location of the stiffeners.


32

Figure 2.35 Distortion displacements at four cross sections of the panel from four welding sequences [21]:
(a) At x= 16 inch; (b) at x=32 inch; (c) at y= 9.3 inch; y (d) at y=30 inch.

Sequences No. 2 and No. 4 result in greater angular bending curvatures than the other sequences. The
angular distortions of the skin plate from sequences No. 1 and No. 3 are similar in magnitude and shape. It
appears that the welding sequence that follows the most restrained joint for depositing the next weld during
the welding process results in smaller angular distortion of the skin plate. Therefore, welding sequences No. 1
and No. 3, which deposit the weld from the inner panels to the outward direction, are better for the
production of structural panels.

Joint rigidity can be defined as the resistance to angular bending of a T-joint under a unit moment applied
to the joint. The welding sequence that starts with more rigid joints and moves progressively towards less rigid
joints would result in less bending of the skin plate. Using the concept of JRM, the optimum welding sequence
(figure 2.36) begins welding both sides of the middle stiffener in any order, then welding the inside joint of the
edge stiffeners in any order, and finally, welding the outer joints of the edge stiffeners in any order.


33

Figure 2.36 Optimum welding sequence determined by JRM [21]

Figure 2.37 depicts a comparison of the distortion values between sequence No. 1 and the optimum
sequence. The optimum welding sequence produces a significant reduction in angular distortion in the skin
plate.

Figure 2.37 Comparison of distortion resulting from welding sequence No. 1 and the optimum welding sequence [21]:
(a) at x= 16 inch, (b) at x=32 inch, (c) at y= 9.3 inch, and (d) at y=30 inch.

Remark: The work performed by Tsai and Park [21] did not consider the welding in the perpendicular
direction and between the T-stiffeners. Therefore, the proposed method to optimize the welding sequence
is limited to rigid, co-lineal joints.


34

Hackmair and Werner [22] investigated the welding sequence and its effect on distortion in a T-joint
structure. The T-joint structure is formed by two hollow extrusions of a 6060 T6 aluminum alloy. The
extrusions are joined by four weld operations, whose order of application defines the two different welding
sequences. Figure 2.38 shows the finite element mesh of the T-joint and the dimensions of the hollow
extrusions. The T-joint was clamped on both ends of the horizontal tube. The evolution of the deformations in
the x-direction was measured on the upper end of the vertical tube with an inductive measuring device. The
different welding sequences are shown in figure 2.39.

Figure 2.38 T-joint configuration [22]: (a) Finite Figure 2.39 Different welding sequences [22]:
element model and (b) dimensions of the (a) Case 1 and (b) case 2
cross section of the hollow extrusions

Figure 2.40 (a) shows the good qualitative agreement obtained when comparing measured (M) and
numerical (S) distortion values for the welding sequence of case 1. Quantitatively, however, the numerical
results show less pronounced peaks. The difference may be due to the FE-mesh of the T-joint not considering
two separate extrusions. The T-joint in the simulation was considered as one continuous part, whereas there
are two extrusions being joined in the welding experiment, separated by an unavoidable gap. Figure 2.40 (b)
shows the comparison of measured (M) and calculated (S) distortion values for the welding sequence of case
2. Similar to case 1, there was a difference in the magnitude of the distortions.

Case 2 resulted in higher distortion. The average experimental distortion increased from 0,4 mm for case 1,
to 2.8 mm for case 2 7 times more. In the simulation, however, distortion increased from 0.2 mm for case 1,
to 0,4 mm for case 2, a factor of 2. Therefore, further research is necessary to improve the quantitative model
predictions through detailed modeling of the gap between the two extrusions.


35

Figure 2.40 Comparison of measured (M) and calculated (S) distortion evolution
for two welding sequences [22]: (a) Case 1 and (b) case 2

Remark: the modeling performed by Hackmair and Werner [22] does not consider the gap between two
separate extrusions, therefore producing a quantitative difference with experimental results. The authors
performed an analysis of the optimum welding sequence and applied it to a component of an automotive
front axle carrier. However, the optimum sequence was not shown. The proposed methodology is only
applied to simple joints. Therefore, a local/global method is required for more complex structures, which
implies a high computational cost.


36

Bart, Deepak and Kyoung [15] investigated the effect of the welding sequence on a sub-assembly
composed of thin-walled aluminum alloy extruded beams. The main factor considered was the quality of the
assembly after welding, which was measured by the deformation at pre-defined critical locations. The
aluminum alloy extruded beam structure was modeled with a 2-D beam element model. Their methodology
consisted of applying pre-estimated angular shrinkages for each welding step, thus eliminating the use of a
complex nonlinear transient analysis, which would require consideration of thermo-mechanical interactions
and plasticity. Two distortion modes (angular shrinkage and tilting shrinkage) were investigated and applied to
model the welding distortion.

Figure 2.41 depicts the geometric configuration of a sub-assembly composed of thin-walled aluminum alloy
extruded beams. These beams have a rectangular cross section and were joined by 2 weld seams (top and
bottom) per joint. Figure 2.42 shows a simplified model with torsion springs (stiffness k) to model the pre-
calculated angular shrinkage.

Figure 2.41 Geometric configuration of a sub-assembly [15] Figure 2.42 Simplified model with torsion springs [15]

Four representative welding sequences based on empirical algorithms (table 2.4) were selected to simulate
the effect of welding sequence on welding deformation,

Case 1 1, 6, 4, 3, 2, 5
Case 2 6, 1, 3, 4, 2, 5
Case 3 1, 6, 2, 5, 3, 4
Case 4 1, 2, 4, 3, 5, 6

Table 2.4 Different welding sequences used
by Bart, Deepak and Kyoung [15]


37

Figure 2.43 depicts the final (exaggerated) deformed shapes for the different sequences. Case 3 resulted in
minimum final distortion. This is because weld seams on joints with greater restraint were applied first.

Figure 2.43 Final exaggerated deformed shapes for different sequences [15]

Remark: the work performed by Bart, Deepak and Kyoung [15] does not consider the contact between
the beams and does not analyze the effect of the welding sequence on the distribution of the residual
stresses. This study is of great importance because it shows that the residual stress has a critical effect in
constrained structures.


38
Kadivar, Jafarpur and Baradaran [23] utilized a genetic algorithm method with a thermo-mechanical model
to determine an optimum welding sequence. The thermo-mechanical model developed for this purpose
predicts the residual stress and distortion in thin plates. The thermal history of the plate was computed with a
transient, two dimensional finite element model which serves as an input to the mechanical analysis. The
mechanical response of the plate was estimated through a thermo-elastic, viscoplastic finite element model.
The proposed model was verified by comparison with experimental data where available. The authors
observed that the welding sequence changes the distribution of residual stress, but has little influence on the
maximum residual stress levels. However, the welding sequence does have an effect on the weldment
distortion. Therefore, the authors chose the minimization of distortion as the target function.

Figure 2.44 depicts the geometric configuration and dimensions of a weldment chosen to investigate the
best welding sequence. In the optimization procedure, the circular weld line was divided into eight parts. The
order of welding for these parts, and the direction of welding for each part, were considered as the problem
variables. It was assumed that the plate was welded by one welder; i.e., eight parts of weld line were welded
one after another, not simultaneously. Two different types of restraints were considered. In one, several tack
welds with equal distances among them were employed to hold the inner circular plate in place. In the other
type (full restrain), all of the weld line elements were considered active at the beginning of the computation.
In other words, it was assumed that the workpiece is continuous and the region associated with the weld line
is heated to the melting temperature. To eliminate the rigid body motion, the central node of the plate was
considered fixed and the vertical freedom of the edge node at =0 was suppressed.

Figure 2.44 Geometric configuration and dimensions of weldment [23]

For the problem under consideration, a typical welding sequence was coded in the form of a string with 16
digits as shown in figure 2.45. The first eight digits in the string (1, 4, 3, 2, 5, 7, 6, 8) represent the sequence of


39
welding. The second eight digits (1, 2, 2, 2, 1, 2, 2, 1) represent the direction of welding, where 1 and 2
respectively mean clockwise or counterclockwise.

Figure 2.45 Code to designate the welding sequence
and welding direction [23]
As stated earlier, the effects of the welding sequence on the maximum residual stress values are negligible.
Therefore, the distortion of the weldment was considered as the objective function. Because the plate was
thin, and its deformation was modeled with an in-plane model, the maximum radial displacement at the edge
of the plate
( )
max
r
U

represented the maximum radial weldment distortion.
Figure 2.46 shows the variation of ( )
r
U with respect to when using tack welds. The displacement
produced by the optimum sequence is nearly uniform. In other words, when the optimum sequence is applied,
the deformed shape of the plate remains circular with a small contraction; whereas using a continuous
sequence the deformed shape of the plate is irregular and with a distorted shape. The maximum displacement
at the edge of the plate
( )
max
r
U

for the optimum sequence was about 70% less than that of the continuous
sequence. Similar results were obtained for the case of welding with full restrain (without tack welds).

Figure 2.46 Radial displacement of the edge of plate with respect to
for continuous welding and the optimum sequence[23]


40
In table 2.5, the values of for the two cases of restraint are compared with the condition of the
perfect heat sink. This table shows that the distortion obtained by the optimum sequence is smaller than
that of the perfect heat sink condition. It should be noted that the generation of a perfect heat sink
condition is almost impossible, and a real heat sink would produce a considerably larger deformation.

Welding
condition
Full restraint With tack welds Welding with
perfect heat sink
(Continuous seq)
1 2 3 4 5 6 7 8
1 1 1 1 1 1 1 1
Continuous
sequence
1 2 3 4 5 6 7 8
1 1 1 1 1 1 1 1
Optimum
sequence
1 4 7 6 2 5 8 3
1 2 1 2 2 2 2 2
Continuous
sequence
1 2 3 4 5 6 7 8
1 1 1 1 1 1 1 1
Optimum
sequence
1 4 7 2 6 5 8 3
1 1 1 1 2 2 2 2
0.409 0.137 0.371 0.110 0.145

Table 2.5 for various welding conditions [23]

Remark: in the work performed by Kadivar, Jafarpur and Baradaran [23], the applied genetic algorithm
method is valid for simple joints only. Therefore, the use of other genetic algorithms is required for complex
joints. In addition, experimental tests to check the optimum sequence obtained by genetic algorithm were
not performed.


41
Voutchkov, Keane, Bhaskar and Olsen [24] proposed a surrogate model that substantially reduces the
computational expense in sequential combinatorial finite element problems. The model was applied to a weld
path planning problem in a tail bearing housing (TBH). The TBH is a crucial component of most gas turbines
and is used to help mount the engine to the body of the aircraft. The welding sequence used to attach the
vanes to the inner ring was chosen to minimize distortion during the welding process.

Figure 2.47 depicts the tail bearing housing. Its major structural components are the outer ring, the inner
ring, and the vanes. The vanes are welded to the rings using the process of Gas Tungsten Arc Welding. The
welds are broken down into six sub-weld segments, considering the order and direction of the welds. Figure
2.48 shows the finite element model of the welding process and designation of the welding paths showing the
number in the sequence. The arrows indicate the positive direction of each path. The point of interest to
measure the distortion was located at the top of the vane (node 10). Only the component in the X-direction of
node 10 was used for the optimization process.

Figure 2.47 Tail bearing housing [24] Figure 2.48 Finite element model of welding process and designation
of welding paths showing positive orientation of weld and
reference number in sequence [24]

To study the effects of the welding sequence, the authors applied a design of experiments (DOE) and
proposed a combinatorial model to obtain an optimum sequence. The final optimization result is illustrated in
figure 2.49, which shows the FE simulation for Run 28 (displacement versus time). The curve reaches zero
displacement just before 300 s, when the clamps were released. Run 28 is shown together with some previous
runs (numbers 1, 5, 6, 16, 24, 26) for comparison purposes. The figure illustrates the improvement of the X-
displacement in comparison to run 1, which is the standard sequence adopted in most TBH production.


42

Figure 2.49 Run 28 is the optimized displacement with a clamped structure [24]

The values for the displacement for all 28 runs are shown in figure 2.50, showing that the optimized run
produces minimal displacement.

Figure 2.50 The absolute values of the X displacement for all 28 sequences [24]

Remark: Although Voutchkov, Keane, Bhaskar and Olsen [24] applied a surrogate model to reduce the
computational expense in determining the optimum welding sequence, the numerical models that
calculated the response for the design of experiments were too computationally expensive. For example, a
single run required 32 hrs and 28 runs were analyzed. If the authors had worked with the sequential
combinatorial model and with a simplified model of the welding process, the required computational
processing time for the analysis of the welding sequence could have been considerably shorter.


43

2.8 Summary of the welding sequence analysis background

Previous research [15-24] has demonstrated that the welding sequence is an efficient method for
controlling residual stresses and distortion.
When the welding sequence is combined with the converse welding method, the residual stress
distribution and distortion are considerably reduced.
To date, there is no study showing the residual stress-distortion relation due to welding sequence.
The values of residual stresses and distortion obtained by numerical simulation (without considering
micro-structural evolution) of a welding process applied to the analysis of the welding sequence show
good agreement with experimental measurements.
Except for the work of Tsai, Park and Cheng [21], analyses of the welding sequences presented
correspond to simple structures with no more than two components.
Previous research did not consider the contact effects between the welded components.
The process of welding sequence optimization depends on the specifics of the restraint and joint type,
and therefore needs to be carried out for each structure.
The welding sequence analysis background is summarized in the next table. This matrix summarizes the
work and investigations performed for the welding sequence analysis, as well as the main topics that
are to be performed in this investigation.

Weldingsequenceanalysis

44
2.9Matrixoftheweldingsequenceanalysisbackground
Article
Numerical
model
Weldingsequence Jointtype Material Analized
parameters
Dimension
ofanalysis
Welding
process
Remarks
Effectofweldingsequenceonresidual
stresses.
Teng,Peng,ChangandTseng,2003.
Thermoelastic
plastic(weldbead
thermalcycleis
simulated)
Progressivewelding,
backstepwelding,
symmetricwelding,and
jumpwelding
Singlepass andmulti
passbuttweldjoints
andcircularpatchweld
joints
SAE1020
steel
Residualstresses Intheplane GMAW The methodology is applicable only to simple structures; in addition, the welding
directionisnotconsideredinthenumericalsimulation.Noexperimentaltestswere
performedtocomparethenumericalresultsofthedifferentweldingsequences.
Influenceofaweldingsequence
ontheweldingresidualstressofathickplate.
Ji,FangandLiu,2005.
Thermoelastic
plastic
8weldingsequences
combinedwith
converseweldingmethod
DoubleVgroovewith
multilayerweldjoints
Martensite
stainlesssteel
1Cr12NiWMoNb
Residualstresses I Inthe
plane
GMAW Animportantdisadvantageisthecomputationalcostresultingfromtheprocessing
timerequiredtosimulatetheweldingprocess.Thisisbecausetheeffectsofphase
transformationandthetypeofsourceofheatusedwereconsidered.
Threedimensionalthermalresponseofthick
plateweldments:effectoflayerwiseand
piecewisewelding.
Nami,KadivarandJafarpur,2004.
Thermovisco
plastic
Piecewisewelding
(blocksequences)
Vgroovewithsingle
passbuttweldjoints
Mildsteel Residualstresses
anddistortion
Inthespace Thereisno
information
Theauthorsdidnotpresentastudyoftheeffectsoftheweldingsequenceonthe
relationshipbetweendistortionandresidualstresses.Furthermore,theproposed
methodologyisexpensivewithrespecttothetimeofprocessingtosimulatethe
weldingprocessduetotheconsiderationofviscoplasticeffects.
Residualstressdistributiondependingon
weldingsequenceinmultipassweldedjoints
withXshapedgroove.
MochizukiandHayashi,2000.
Thermalelastic
plastic
6weldingsequencesfora
multipassweldedpipe
joints
Multipasswelded
jointswith
Xshapedgrove
Austenitic
stainlesssteel
Residualstresses Intheplane GTAW Theanalyticalmethodproposedtodeterminetheresidualstressesthrough

thicknessisonlyapplicabletomultipass,weldedpipejoints.Therefore,themethod
isnotvalidforpipejointsofsmalldiameter.Themethodpresentedhasagood
qualitativecorrelationwiththeexperimentalandnumericaldata.However,there
wasnotagoodcorrelationquantitatively.
Residualstressreductionandfatiguestrength
improvementbycontrollingweldingpass
sequence.
Mochizuki,HattoriandNakakado,2000.
Thermoelastic
plastic
2weldingsequencefor
multipassfilletweldjoints
Multipassfilletweld
joints
Carbonsteel
JISSS400
Residualstresses Intheplane CO
2
GasArc
Themethodologyisvalidforsimplefilletjointsonly.Twoverysimilarwelding
sequenceswereusedwhereonlytwoweldbeadswerechanged.
Weldingdistortionofathinplatepanel
structure.
Tsai,ParkandCheng,1999.
Inherent
shrinkagemethod
4weldingsequencesfor
singlepassfilletweld
joints
Singlepassfilletweld
joints
Aluminum
5456H116
Distortion Inthespace GMAW
TheweldingintheperpendiculardirectionandbetweentheTstiffenerswasnot
considered.Therefore,theproposedmethodtooptimizetheweldingsequenceis
limitedtorigid,colinealjoints.
Applicationofweldingsimulation
forchassiscomponentswithinthe
developmentofmanufacturingmethods.
Hackmair,WernerandPnisch,2003.
Thermoelastic
plastic
2weldingsequencefor
joints
joints
Aluminum alloy
6060T6
Temperature
distributionand
distortion
Inthespace MIG
Thegapeffectsbetweentwoseparateextrusionsarenotconsidered.Forthis
reason,thequantitativeresultsshowaconsiderabledifference.Theproposed
methodologyisonlyappliedtosimplejoints.Therefore,alocal/globalmethodis
requiredformorecomplexstructures,whichimpliesahighcomputationalcost
(timeofprocessing).
Weldingdistortionminimizationforan
aluminumalloyextrudedbeamstructureusing
a2Dmodel.
Bart,Nnaji,DeepakandKim,2004.
Preestimated
angular
shrinkages
4weldingsequencesfor
joints
joints
Aluminum alloy
Distortion Intheplane Thereisno

information
Thecontacteffectbetweenthebeamswasnotconsidered.Also,astudyofthe
effectsoftheweldingsequenceonthedistributionoftheresidualstresseswasnot
performed.Thisstudyisofgreatimportancebecauseitshowsthatinconstrained
structurestheresidualstresshasacriticaleffect.
Optimizingweldingsequencewithgenetic
algorithm.
Kadivar,JafarpurandBaradaran.2000.
ThermoelastIc
viscoplastic
Continuoussequence,
optimumsequencesand4
randomsequences
Circularpatchweld
joints
Stainless
304L
Distortion Intheplane Thereisno
information
Theappliedgeneticalgorithmmethodisvalidforsimplejointsonly.Therefore,the
usesofothergeneticalgorithmsarerequiredforcomplexjoints.Inaddition,
experimentalteststochecktheoptimumsequenceobtainedbygeneticalgorithm
werenotperformed.
Weldsequenceoptimization:theuseof
surrogatemodelsforsolvingsequential
combinatorialproblems.
Voutchkov,Keane,BhaskarandOlsen,2005.
Thermoelastic
plastic
27Randomwelding
sequencedeterminedby
DOEandtheoptimum
weldingsequence
joints
Inconel718
Distortion Inthespace TIG Althoughasurrogatemodelthatcanbeusedtoreducethecomputationalexpense

todeterminetheoptimumweldingsequencewaspresented,thenumericmodels
usedtomeasuretheresponseforthedesignofexperimentsweretooexpensive
withrespecttotherequiredcomputationalprocessingtime
HernndezA.I.
Researchwork
2009
Thermoelastic
plastic
(weldbead
thermalcycleis
simulated)
36weldingsequences
baseonempiricsrules
Singlepassbuttweld
jointsand
joints
ASTMA36 Steel
(PolandSt3s)
Residualstresses,
distortion
Andarelationship
betweenboth
parameters
Intheplane
andinthe
space
SMAW
and
GMAW
MAG
Formulationoftheweldingsequencesmostappropriatestoreduceresidualstress,
distortionorarelationshipbetweenbothparametersforsymmetricalstructuresin
theplaneandinthespace.Possiblegeneralizationoftheexample


45

CHAPTER III

PROPOSAL OF A NUMERICAL SIMULATION
OF THE WELDING PROCESS
AND
A NUMERICAL SIMULATION OF THE WELDING
SEQUENCE IN AN L-TYPE STRUCTURE

3.1 Introduction

This chapter presents a finite element simulation of the welding processes with the objective of defining
the initial and boundary conditions and studying the effects of the welding sequence on the residual stress
distribution and distortion in symmetrical structures in 2 and 3 dimensions. Although the analysis focuses on
symmetrical structures, the proposed numerical model is applicable to asymmetrical structures. The proposed
sequentially-coupled thermo-mechanical analysis involves two steps: a transient heat transfer analysis is
performed followed by a thermal elastic plastic analysis. This numerical simulation is performed in an I-type
specimen subject to tension and validated with experimental data [25] to check the accuracy of the proposed
numerical model. The end of this chapter presents a numerical simulation of the welding sequence in an L-
type structure to test whether the proposed numerical model can simulate the effects of the welding
sequence on residual stresses and distortion, with respective adaptations.

3.2 Heat transfer in welding

The temperature distribution in a weldment is mainly determined by the total heat input, the preheat
temperature, the welding process, and the type and geometry of joint. The latent heat released during phase
transformation only marginally affects the temperature distribution. The temperature field in the molten pool
is in general assumed to be governed by the same equation as is applied to the solid metal. In most
metallurgical and structural studies of welding, the energy transfer from the electrode to the weld metal is
represented either by a surface heat source or by momentarily deposited energy that gives the weld metal an
initial temperature distribution.


46
In general the temperature field should be taken as thermodynamically coupled to the mechanical field.
The governing equation of the heat flow follows the first law of thermodynamics (conservation of energy). This
law states that the rate of change of internal energy and conduction must be in equilibrium with
the heat production and the power of elastic and viscoplastic straining and , respectively [26]:

(1)

The parameter takes on the value of 1 if all inelastic dissipation is converted into heat. The
mechanical coupling terms in (1) are in most cases not considered because their influence on the temperature
field is very small [27]. It is therefore possible to divide the thermo-mechanical analysis of a welding process
into two main parts: the analysis of the temperature field and subsequent analysis of the mechanical fields.

3.2.1 Analytical solution for the temperature field

Neglecting the mechanical coupling terms in Eq. (1), the energy equation becomes [27]:

(2)

Equations (1) and (2) are statements of the First Law of Thermodynamics [28]. Fouriers law of heat transfer
[28] defines a relationship between the heat flux and the gradient of the temperature field . For an
isotropic material this relationship is the Fourier heat conduction Law:

(3)

Where is the thermal conductivity for the material. Equation (2) and (3) give

(4)

In Cartesian coordinates x, y, z, equation (4) becomes

(5)

Where: is the thermal conductivity in the , and directions, respectively, is the
temperature, is the density, is the specific heat and is time. Equation (5) becomes nonlinear if material
properties and C are a function of temperature. The first term is conduction of heat through the
material. The second term is the source of heat generation. The last term is the rate of change of
internal energy.


47
There are only three terms in the heat equation (with zero velocity) that can be manipulated in addition
to the associated boundary conditions of either prescribed flux or prescribed temperature type. Therefore any
model of a weld heat source based on the heat equation must specify something about one or more terms in
this heat equation or its BCs.

Suppose that a heat source is moving at a constant speed in the positive direction. We can define an
Eulerian (moving) frame with origin at the center of the source (i.e., a welding arc), and coordinates .
The transformation from to is given by [29]:

(6)

Where: is time. Noting that the material time derivate in equation (4) becomes:

(7)

We obtain the equation of conservation of energy in the Eulerian frame:

(8)

Here, represents the time derivate of temperature at a point fixed with respect to the heat source. In
steady state, this derivate is zero, and so equation (8) becomes:

(9)

When and are constant, equation (9) can be solved more easily by applying the transformation:

(10)

Substituting equation (10) into (9), we have:

(11)

(12)

Equation (12) is symmetric positive definite. After properly transforming the boundary conditions from
to function, it is easy to solve this equation with standard Lagrangian FEM code that


48
contains a solver for positive definite symmetric matrix. Equation (12) applies only to linear systems (i.e., is
independent of ) because of equation (11).

3.2.2 Thermal initial and boundary Conditions

The solution of the heat conduction equation (4), involves a number of arbitrary constants to be
determined by specified initial and boundary conditions. These conditions are necessary to translate the real
physical conditions into mathematical expressions [28].

Initial conditions are required only when dealing with transient heat transfer problems (in which the
temperature field changes with time). The common initial condition in a material can be expressed
mathematically as [28]:

(13)

On the boundary of the domain either the essential (prescribed temperature) or natural (prescribed flux)
boundary conditions must be satisfied. Five common types of boundary conditions are [28]:

Essential boundary condition:

a) Prescribed surface temperature, it is often necessary to prescribe a surface temperature for a structure.
The mathematically expression takes the form:

(14)

Natural boundary condition

b) Prescribed heat input, many structures, particularly in the case of welding, have the boundary surface
exposed to a heat source or sink. Let denote the outward normal to the surface at the point , then
the mathematical formulation for the heat input across a solid boundary is:

(15)

(c) Perfectly insulated surface, by definition, a perfectly insulated surface is one across which there is no
heat flux. Equation (15) becomes:

(16)


49
(d) Convection boundary condition, most structures have surfaces which are in contact with fluids, either
gases or liquids. The heat flux across a bounding surface may be taken as proportional to the difference
between the surface temperature and the known temperature of the surrounding medium.
Equation (15) then takes the form:

(17)

Where: is the convection heat transfer coefficient.

(e) Radiation heat exchange, when the rate of the heat flow across a boundary is specified in terms of the
emitted energy from the surface and the incident radiant thermal energy emitted and reflected from other
solids and /or fluids, the boundary condition is:

(18)

Where: is the Stefan-Boltzmann constant and is the emissivity.

If the partial differential equation (4), the initial conditions (13) and the boundary conditions (14), (15),
(16), (17), and (18) are consistent, the problem is well posed and a unique solution exists.

The most convenient domains in which to apply equations (11) and (12) are prisms with constant cross-
sections, as found in the plate shown in figure 3.1. The temperature of the boundary between the weld pool
and the solid is prescribed to the solidus temperature. The flux due to complex radiation and convection on
the surface of the weld pool is not taken into account.

Figure 3.1 Eulerian frame and thermal initial and boundary conditions


50
3.3 Thermal elastic plastic stress analysis in welding

In the mechanical analysis, the temperature history obtained from the thermal analysis is introduced as a
thermal loading. Thermal strains and stresses can be calculated at each time increment, and the final state of
residual stresses will be accumulated by thermal strains and stresses. Residual stresses in each temperature
increment are added to those at nodal points to update the behavior of the model before the next
temperature increment.

3.3.1 Mechanical equations

Three basic sets of equation relating to the mechanical model, the equilibrium and compatibility equations
and the constitutive equations for thermal elastic plastic material, are considered as follows [1]:

,
0
ij j i
f + = (19)

Where:
i
f

is the sum of the body force and
ij
is the Cauchy stress tensor. To ensure that the body
remains continuous during the deformation, the compatibility equations must be satisfied [2].

, , , ,
0
ij kl kl ij ik jl jl ik
+ = (20)

The thermal elastic plastic material model will be derived assuming isotropic material, von Mises yield
criterion and the associated flow rule and linear isotropic hardening. Thermal and mechanical properties of the
selected material are a function of temperature. The micro-structural evolution is not considered. The initial
assumption is that the total strain increment can be decomposed as (valid for small strains and rotations) [1]:

e p
ij ij ij ij
d d d d

= + + (21)

Stress-strain relations can be written as

2
2
3
1 1
2
3 1 2 3
1
3
ij kl kl
ij ij kk ij ij kk
VM
Gs s d
E
d d Gd K dT dK
H v K
G
| | (
= + +
|
(
| |
\ .
+
|
\ .

1
1
1 1
3 3
ij
ij
VM
f
s dT
s
T
dG
H H G
G G
| |
|

\ .
+ +
`
| | | |

+ +
| |

\ . \ .
)
(22)


51

The variation of the material properties dGand dK in equation (22) is due to the temperature only [2].

These equations are treated in more detail in appendix No.1

3.3.2 Mechanical initial and boundary conditions

Initial conditions are necessary only when the analysis is transient [30] and therefore will not be specified
in the static mechanical analysis described here.

In a mechanical problem, the boundary conditions can be specified in three different ways:

Type I: Specification of the primary variable (degree of freedom).
Type II: Specification of variables related to the derivative of the primary variable.
Type III: Specification of a linear combination of the primary variable and its derivate.

The primary variables are the displacement components. When Type I boundary conditions are used, the
displacement constraints are specified along a segment of the boundary. Rotations at the boundary are Type II
because rotations are related to the derivatives of the displacement components.

3.4 Finite element solution of the welding

We have selected finite element modeling has over existing alternatives for detailed analysis (finite
difference and boundary element) for its capability for nonlinear analysis in complex geometries. Finite
element is also most compatible with modern CAD/CAM software systems. For thermal analysis alone, a
strong argument can be made in favor of finite difference methods. However, for thermal elastic plastic
analysis finite element analysis holds an advantage. The boundary element method is not well developed for
nonlinear analysis. Briefly, these considerations led to the choice of FEA as the most effective numerical
method for developing a complete analysis capability for computer modeling or simulation of welds. Finite
element is a generalized Rayleigh-Ritz method, which uses interpolation to express the variables in terms of its
values in a finite number of nodes.

3.4.1 Finite element solution of heat transfer in welding

The finite element method FEM usually imposes a piece-wise polynomial approximation of the temperature
field within each element [29]:

(23)


52
Where: are basis functions dependent only on the type element and its size and shape. Physically
is the value of the temperature at node at time .

(24)

Where: is abbreviated to .

The next question is how to evaluate ? Garlekins FEM is among the most convenient and general of the
methods available for this purpose. If equation (23) is substituted into equation (4), a residual or error term
must be added. Otherwise equation (23) would be the exact solution. Indeed when equation (23) is the exact
solution, the error in the FEM solution is zero.

Garlekins FEM method requires [29]:

(25)

Mathematically is the residual from equation (4); in equation (25) is a test function and the terms in
equation (23) are the trial functions. Since there are nodes, equation (25) creates a set of ordinary
differential equations which are integrated to form a set of algebraic equations for each time step:

(26)

This set of equations is solved for the nodal temperatures at the end of the time step. Usually some form
of Newton-Raphson method together with a Gaussian elimination and back substitution would be employed.

3.4.2 Finite element solution of the thermal elastic plastic stress analysis in welding

In welding, no external force is applied to the welded structure. Thermal loads are the only source of
residual stress and distortion [1-3]. Therefore, the principle of virtual work in this case, without mechanical
loading, is written as [26]:

0
ij ij
V
d d dV =
(27)

Where:
ij
d is the variation in the strain increment
ij
d . The stress increment
ij
d is given in equation
(22).


53
In matrix notation, the thermal elastic plastic constitutive equation (eqn. 22) becomes [26]:

{ } { } { } { }
ep
d D d C s M dT ( =

(28)

Where:
ep
D (

is the elastic plastic stiffness matrix, { } C s and { } M dT are terms related with the
distortion and volume change due to the temperature.

The virtual work equation (27) is now written as:

{ } { } 0
T
V
d d dV =
(29)

Where: is the volume of the body to be analyzed and is the transpose of . The strain-
displacement for three-dimensional problem can be written in matrix form as [26]:

{ } { }
0 0
0 0
0 0
0
0
0
x
y
z
d du
y x
z y
z x
(
(
(
(
(
(
=
(

(
(
(

(
(
(

(
(

(30)

In the finite element solution, the increments in the displacements for each element are approximated as
[30]:

{ } ( ){ }
1
, ,
n
i
i
du N x y z U
=
=
(31)

Where: n

is the number of nodal points for each element, ( , , )
i
N x y z are the shape functions and { } U
are element nodal increments.


54
Substituting equation (31) into equation (32) produces the following expresion:

{ } | |{ } d B U = (32)

Where:
| | | | ( )
1
, ,
n
i
i
B N x y z
=
=

is the strain-displacement matrix for each element.

Using equation (31) and equation (32) yields:

{ } { } | |
T T T
d U B = (33)
Substituting equations (27), (28) and (32) into equation (29) yields

{ } | | | || | { } | | { } | | { }
0
T T T T
V V V
U B D B dV U C B s dV M B d dV

(
=
(

(34)

The vectors { } U and { } U are not included in the integrals because they are not a function of the
coordinates. Equation (34) must be valid for any admissible virtual nodal increase { } U of equilibrium
configuration. Therefore, equation (34) becomes:

{ } | |{ } { } { }
0
T
U K U F = (35)

Now, the expressions inside the braces must be zero, therefore:

| |{ } { } K U F = (36)

Where: | |
K is the global stiffness matrix for the assembly of elements and is formed as a sum of the
element stiffness matrices and is the global load vector, formed as the sum of the element load
vectors .

The stiffness matrix for each element is:

| | | | | |
e
T
ep
e
V
K B D B dV ( =

(37)


55

And the element load vector is:

{ } | | { } | | { }
e e
T T
e
V V
F M B dT dV C B s dV = +

(38)

Equation (36) is solved for each load (time) increment. The stiffness matrix is in most cases updated at
every load increment. It is often advantageous to use the elastic stiffness matrix only, computed at the
beginning of each time step in a modified Newton-Raphson approach. The non-linearities are compensated for
by equilibrium iterations and the stress increment during each load (time) increment are often calculated
using the subincrement technique. The matrix element in is typically integrated with Gauss quadrature.
The reader is referred to Appendix 1 for detailed treatment of this set of equations.

3.5 Geometric configuration of I-type specimen subject to tension

The finite element welding simulation is performed with a commercial code on an I-type specimen subject
to tension. This geometry is selected due to the availability of experimental data [25]. A static linear model was
used with a shrinkage force (transverse to weld bead) applied in the boundary between the weld metal and
the base metal to simulate the effects of the welding process. In the present work, a thermo-mechanical
analysis that includes plastic behavior and temperature distribution will be developed. The I-type specimen
subject to tension consists of a central bar and two lateral bars (restraining members). When the weld metal
solidifies and shrinks, tensile residual stresses are induced in the central bar, while in the lateral bars
compression residual stresses are induced. This is due to the balance conditions and geometry [3]. Figure 3.2
depicts the geometric configuration of an I-type specimen subject to tension.

Figure 3.2 Geometric configuration of I-type specimen subject to tension (mm) [25]


56
3.6 Material selection for the I-type specimen subject to tension

The material selected is ASTM A36 steel because of its common use in the welding industry. The chemical
composition of ASTM A36 is shown in table 3.1.

Chemical composition of ASTM A36 (%)
C Mn P S Si Fe
0.21-0.29 1.0 0.04 0.05 0.28 98.0 min

Table 3.1 Chemical composition of ASTM A36 carbon steel [33]

3.7 Temperature-dependent thermal and mechanical properties of ASTM A36

Temperatures vary widely within the welded part. Therefore, thermal and mechanical properties of welded
materials vary significantly [34]. The thermal and mechanical properties of ASTM A36 due to temperature are
shown in figure 3.3.

Figure 3.3 Thermal and mechanical properties of ASTM A36 as a function of temperature [34]


57

As this previous figure indicates, when temperature increases, the modulus of elasticity, modulus of
tangent, yield stress and thermal conductivity decrease while the thermal expansion and specific heat
increase. Autogeneous weldment is assumed. This means that weld metal, heat affected zone (HAZ), and base
metal share the same mechanical and thermal properties. Figure 3.4 depicts the stress-strain behavior of
ASTM A36 carbon steel for different temperatures.

Figure 3.4 Stress-strain behavior of ASTM A36 carbon steel
for different temperatures [Based on figure 3.3]

3.8 Finite element model of the I-type specimen subject to tension

We simulate welding with a sequentially coupled thermo-mechanical analysis. The thermo-mechanical
analysis involves two steps. First, a transient heat transfer analysis is performed. The calculated transient
temperatures are applied as a step-by-step thermal load. In the second step, a thermal elastic plastic analysis
(quasi-static) computes the cumulative thermal stresses for each time step. The end result is residual stress
and distortion. All numerical simulations in this investigation were performed in a PC with an Intel Core 2
Duo processor T5300 running at 1.73 GHz, configured with 3072MB of SDRAM memory.

Due to the symmetry of the I-type specimen subject to tension, only a quarter of the model is analyzed.
The finite element mesh is shown in figure 3.5. The weld region is finely meshed, and the zones away from the


58
weld bead have a relatively coarse mesh. This guarantees accuracy while reducing the calculation time. The
numerical model is divided into 1536 hexahedral elements and 2420 nodes.

Figure 3.5 Finite element mesh of the I-type specimen subject to tension

3.8.1 Definition and justification of the applied finite elements

To perform the thermal analysis, it is necessary to select an element type that is able to simulate:

A transient thermal analysis.
The weld thermal cycle [figure 3.6, page 60], which determines the temperature as a function of the
time in the weld bead (i.e., all elements belonging to the weld bead).
The conduction phenomenon that occurs in the base metal as primary mode of heat transfer.
The phenomenon of heat loss by convection to the surrounding in the base metal, moreover it is
required to calculate the rate of change of internal energy.
Activation and de-activation of the weld bead elements.

To perform the structural analysis, it is necessary to select an element type that is able to simulate:

Homogeneous and isotropic material.
Elastic-plastic materials behavior following the von Mises yield criterion and the associated flow rule
with linear isotropic hardening by strain.
Small strains.


59
Furthermore, it is necessary to select an element type that with the temperature in each node and the
thermal expansion coefficient, it can couple these data with the structural unitary strain.

ANSYS offers a broad variety of elements for the analysis of two and three-dimensional thermal and
structural problems. In table 1 and 2 some common element used by ANSYS are presented. In general, all
elements have the same characteristics (degree of freedom, input data, and output data), only there are two
main differences: 1) the order of elements: higher order elements achieve better results and greater accuracy,
however, these elements require more computational time. This time requirement is because numerical
integration of elemental matrices is more involved, also, these elements are well suited to model problems
with curved boundaries. 2) The shape of element: quadrilateral and brick element present better meshing
control than triangular and tetrahedral elements.

For these two reasons, PLANE55 and SOLID70 elements are selected for two and three dimensional thermal
analysis, respectively. Both elements have a single degree of freedom, temperature and isotropic material
properties. These elements alloy simulate these thermal conditions mentioned early without complications,
which is necessary to introduce the thermal conductivity and specific heat as a function of temperature [Fig
3.3, page 56], the convective heat transfer coefficient (natural convection), this coefficient is considered
constant and is equal to 10 W/m
2
C. Heat loss by convection is a boundary condition and is applied only to
surfaces in contact with the surrounding. It is also required to introduce the temperature of the surrounding,
which is a boundary condition (20C) and the initial condition of base metal, which is considered as the
ambient temperature (20C).

The elements PLANE42 and SOLID45 are selected for two and three dimensional structural analysis,
respectively. PLANE 42 and SOLID45 elements have two and three degrees of freedom at each node,
respectively: translation in the nodal directions and isotropic material properties. Both elements couple the
temperature with strain, for this reason it is necessary to introduce the thermal expansion coefficient,
elasticity modulus, tangent modulus, yield stress and Poissons ratio dependent on temperature as shown in
figure 3.3 [page 56]. Finally, to transfer the thermal loads obtained in the transient thermal analysis to the
structural analysis, the concept of body load is used through load step. PLANE 42 and SOLID45 elements have
this capability because the temperatures may be input as body loads at the nodes. The elements PLANE42 and
SOLID55, and SOLID70 and SOLID45 are equivalent to perform coupled thermo-mechanical analysis. Definition
and justification of the applied finite elements are treated in more detail in appendix No.2.

3.8.2 Thermal initial and boundary conditions

As mentioned early, these conditions are necessary to express the real physical conditions into
mathematical expressions. For all numerical simulation developed in the present research, the initial and
boundary conditions are equal, unless otherwise noted.


60
a) Initial condition:

Because the thermal model consists of a transient heat transfer analysis, it is required an initial condition of
the base metal, which is considered as the surrounding temperature ( =20C).

b) Boundary conditions (see figure 3.5, page 58):

Symmetrical plane:

Plane y-z in x=0; and plane x-z in y=0;

Surfaces in contact with the surrounding except surfaces belonging to weld metal:

Nodes belonging to weld metal:

Prescribed heat input: heat loading is simulated via weld thermal cycle curves shown in figure 3.6. The
input heat is applied to all surfaces between the weld metal and the base metal.

Figure 3.6 Weld thermal cycle of ASTM A36 carbon steel [34-35]


61
3.8.3 Mechanical boundary condition

In the structural analysis, the parameters and are the displacements in the X, Y, and Z-direction,
respectively. , , and are the strain in the X, Y, and Z-direction, respectively. , , and are the
rotation with respect to X, Y, and Z axis, respectively.

a) Symmetrical plane (see figure 3.5, page 58):

Plane y-z in x=0;

and

; due to Poissons ratio and pure tension

Plane x-z in y=0;

and

due to Poissons ratio and pure tension

b) Surfaces in contact with the surrounding (see figure 3.5, page 58):

The normal stresses to the surfaces in contact with the surrounding and the shear stresses are zero (i.e., in
the plane x-z in y=50.8 mm, ).

c) For all surfaces including the surfaces belonging to symmetrical planes, the shearing strain is zero due to
pure tension.

3.8.4 Body load

Body loads can be generated internally or externally as the result of physical fields acting volumetrically on
the body. Gravity, inertial loads, and temperature changes represent body loads in a mechanical problem [26].
In the present research, we use the concept of body load to transfer the transient thermal loads to the static


62
structural analysis. The use of load steps is necessary to calculate the response of the structure to
temperatures obtained from transient thermal analysis at specific points in time.

3.8.5. Solution to the finite element model

After completing the finite element mesh and specifying the loading conditions (boundary, initial and body
loads), the solution can be initiated.

In the transient thermal analysis, the time to complete the weld bead is 6 seconds, and the time to return
to its initial temperature is 1200 seconds [25]. The thermal elastic plastic analysis requires 15 load steps to
complete the weld thermal cycles (1, 3, 10, 20, 35, 50, 75, 100, 200, 300, 400, 600, 750, and 1200 seconds). All
the computational analysis was done using ANSYS.

3.9 Points of interest in the finite element model of the I-type specimen subject to tension

Experimental data was taken from an I-type specimen subject to tension [25]. Residual stresses in the X-
direction were obtained in two points of interest. The same points (Figure 3.7) are selected on the finite
element model to carry out a quantitative comparison

Figure 3.7 Localization of the points of interest on finite element model

3.10 Residual stresses in the I-type specimen subject to tension obtained in the numerical simulation

The results of interest are the residual stresses in the two selected points. Table 3.2 depicts the values of
the residual stresses in the X-direction obtained in the numerical simulation.


63
Node Residual stresses in the X direction (MPa)
126 204.53 (Tensile)
2001 12.78 (Compressive)

Table 3.2 Residual stresses obtained in the numerical simulation
of I-type specimen subject to tension

Figure 3.8 depicts the distribution of residual stresses in the X-direction at the end of the thermal cycle of
the weld bead. High tensile residual stresses occur in the region between the weld bead and the metal base
due to resistance to contraction of the material as the cooling begins. This contraction produces tensile
residual stresses in the central bar. Therefore, for self-equilibrium purposes, compressive residual stresses
occur in the lateral bars.

Figure 3.8 Distribution of residual stresses in the X-direction
in the I-type specimen subject to tension

3.11 Comparison between numerical and experimental results of an I-type specimen subject to tension

Experimental and numerical results are compared for four locations [figure 3.9(a)] in I-type specimens
subject to tension. Points 1 and 3 are located in the central zone of the lateral bars, while points 2 and 4 are
located in the central zone of the central bar.


64
Figure 3.9 (b) shows welding parameters employed.

Figure 3.9 Experimental tests: a) Localization of the points of interest in the I-type specimen

Table 3.3 depicts the comparison between numerical and experimental data for two I-type specimens
subject to tension. Due to the symmetry of the specimens, points 1 and 3, and 2 and 4 have equal numerical
residual sresses. The table shows that of the 6 comparison points, 2 of them present a difference of 1%, 2
points present a difference of 5% and the 2 remaining points present a difference of 10%.

Point Numerical
Residual
stress
(MPa)
Specimen 1 Specimen 2
Experimental
residual stress
(MPa)
Difference
(%)
Experimental
residual stress
(MPa)
Difference
(%)
1 -12,78 -11,72 8,29 -11,79 7,74
2 204,53 194,43 4,93 195,66 0,628
3 -12,78 -12,18 4,69 -------- --------
4 204,53 205,38 0,41 -------- --------
Table 3.3 Comparison between numerical data and experimental data
of two I-type specimen subject to tension

3.12 Conclusions of the numerical simulation of the welding process in an I-type specimen subject to tension

The maximum residual stress occurs in the region between the weld bead and the metal base.
Tensile residual stresses occur at the central bar, while compressive residual stresses occur at the lateral
bars.
The tensile residual stresses at the central bar remain constant through all of its length.
The compressive residual stresses at the lateral bars remain constant in all their length.
The numerical values show a good agreement with the experimental measurements.


65
3.13 Numerical simulation of the welding sequence in an L-type structure

We are modeling the effects of the welding sequence on the residual stress distribution and distortion in a
L-type structure to check if we can predict residual stresses and distortion with respective adaptations.

3.14 Geometric configuration of the L-type structure

The L-type structure consists of two bars and a rigid base. The bars are joined by a weld bead between
them and two weld beads to a rigid base (figure 3.10). This configuration is selected because it contains few
weld beads and the optimum welding sequence is well known through experience: minimum distortion is
produced by first applying the weld bead that joins the two bars and then applying the two weld beads that
join the two bars to the rigid base.

Figure 3.10 Geometric configuration of the L-type structure (mm)

3.15 Finite element model of the L-type structure

The numerical simulation of the welding sequence in an L-type structure is based on the proposed model
of the welding process developed in the previous section. The model considers the time between the
application of a weld bead and the next. To simulate the structure with welding tacks, the elements of the
welded region are activated during the analysis. The welding region has a fine mesh and the zones away from


66
the weld bead are meshed coarser, reducing the processing time while still guaranteeing computational
accuracy. Figure 3.11 shows the finite element model of the L-type structure.

Figure 3.11 Finite element model of the L-type structure




b) Boundary conditions(see figure 3.11):

Surfaces in contact with the surrounding except surfaces belonging to the weld beads (Plane x-y in z=0):

Nodes belonging to weld beads (according to the welding sequence):

Prescribed heat input: heat loading is simulated via weld thermal cycle curves (figure 3.6, page 60). Heat is
applied on all surfaces between the weld metal and the base metal according to welding sequence.


67
3.15.2 Mechanical boundary conditions

Three boundary conditions are applied to structural model: a) those just sufficient to prevent rigid body
motion and rotations of the model and b) stress-free on Z-direction (plane-stress problem), which imply that
all out-of-plane stresses vanish, and c) the normal stresses to the surfaces in contact with the surrounding and
the shear stresses are zero.

a) The nodes belonging to the element attached to coordinate systems (X=0 and Y=0) [see figure 3.11,
page 66) are selected to prevent rigid body motion and rigid body rotations:

and

b) In the plane-stress problem, the normal stresses to the plane X-Y ( ) and shearing stress acting in this
plane ( and ) are zero on face of the structure (see figure 3.11). Therefore, .

c) Surfaces in contact with the surrounding (see figure 3.11, page 66)

x=0, ).

3.15.3. Solution of the finite element model of the L-type structure

loads), the solution can be initiated.

In the transient thermal analysis, the time to complete each weld bead is 6 seconds, and the time to return
to its initial temperature is 1200 seconds [25]. The thermal elastic plastic analysis for each simulation requires
15 load steps to complete the weld thermal cycles (2, 5, 15, 32, 35, 45, 62, 65, 75, 100, 200, 300, 400, 600,
1200 seconds).

3.16 Configuration of welding sequences for the L-type structure

Due to the number of weld beads and the diagonal symmetry of the L-type structure, there are only three
different welding sequences (Table 3.4). The first welding sequence applies the weld bead counterclockwise.


68
The second welding sequence first applies the weld beads that join the bars with the rigid base, and then the
weld bead that joins the two bars. The third welding sequence first applies the weld bead that joins the two
bars, and then the weld beads that join the two bars with the rigid base.

1 1-2-3
2 1-3-2
3 2-1-3

Table 3.4 Welding sequence configuration
in the L-type specimen

3.17 Localization of the point of interest in the L-type structure

It has been demonstrated in the previous section that the maximum residual stress in the metal base
occurs in the weld metal boundary. For the L-type structure, this maximum occurs in node 1533 (figure 3.12).

Figure 3.12 Localization of the node 1533 in the L-type structure

3.18 Numerical results in the L-type structure

The maximum von Mises residual stress in the base metal and the maximum distortion in the L-type
structure are the parameters of interest in this analysis. Figure 3.13 depicts the maximum distortion for three


69
different welding sequences. The figure shows that sequence 2 produces highest distortion and sequence 3
reduces distortion by 32%

Figure 3.13 Maximum distortions in L-type structure for
three different welding sequences

Figure 3.14 depicts the maximum von Mises residual stress in the base metal for three different welding
sequences. Welding sequence 3 produces highest residual stress and sequence 2 reduces stress by 5%.

Figure 3.14 Maximum Von Mises residual stress in the L-type structure
for three different welding sequences


70

3.18.1 Distortion profile in the L-type structure

The maximum distortion occurs mainly in the Z-direction in the region near weld bead 2, and then
decreases near weld beads 1 and 3 (Figure 3.15, welding sequence 3).

Figure 3.15 Distortion profile in the L-type structure corresponding
to welding sequence 3

3.18.2 Residual stress distribution in the L-type structure

The von Mises residual stress in the base metal is maximum at the boundary between the weld bead 2 and
the vertical bar. High residual stresses also occur near the rigid base, close to weld beads 1 and 3. These
residual stresses occur because weld beads 1 and 3 are highly constrained. In addition, these residual stresses
change little with the welding sequence. Therefore, they are not considered for the welding sequence analysis.
Figure 3.16 depicts the distribution of Von Mises residual stresses corresponding to welding sequence 2.


71
Based on the variation of the maximum distortion values and on the Von Mises residual stresses in the base
metal, it is demonstrated that the proposed numerical model of the welding process can accurately simulate
the effects of the welding sequence on residual stress and distortion in welded structures.

Figure 3.16 Distribution of Von Mises residual stress corresponding to the welding sequence 2

3.19 Experimental tests for the L-type structure

To qualitatively confirm the results obtained in the numerical simulation of the different welding
sequences, two experimental tests are performed. The L-type specimen 1 follows the sequence that produces
the most distortion, and specimen 2 follows the sequence that produces the least amount of distortion. All
welding parameters used in the numerical simulations are reproduced in the experimental tests. Welding tacks
are applied to all the L-type specimens before the welding process starts. Measurements of the distortion in
both specimens are performed in a BH-V504 MITUTOYO coordinate measuring machine (resolution:
0.0001mm) located in the Manufacturing Processes Laboratory at the Salamanca-Irapuato campus of the
University of Guanajuato.


72
3.19.1 Selection of points of interest in the L-type structure

To measure structure distortion, 26 points of interest are selected. These points are located on the edges
of both bars (figure 3.17). A reference frame at the rigid base is marked to locate the specimen.

Figure 3.17 Localization of the points of interest in the L-type specimen

3.19.2 Configuration of the welding sequences in the L-type specimens

As mentioned earlier, specimen 1 follows welding sequence 2, first joining the bars with the rigid base and
then joining the two bars. Specimen 2 (Figure 3.18) is welded with sequence 3, .first applying the weld bead
that joins the two bars and then applying the weld beads that join the two bars to the rigid base.


73

Figure 3.18 L-type specimen 2 after applying the welding sequence 3

3.19.3 Measurement of distortion on L-type specimens

The final distortion induced by the welding process is determined by subtracting the distortion after
applying the weld tacks from the distortion after applying the weld beads. Figure 3.19 shows specimen 2
(welding sequence 3) mounted on the coordinate measuring machine.

Figure 3.19 L-type specimen 2 (welding sequence 3) mounted on
the coordinate measuring machine


74
The maximum distortion in specimen 1 occurs in point 7 and in specimen 2 occurs in point 8, both near
weld bead 2 (table 3.5). However, specimen 1 has higher distortion than specimen 2. This behavior agrees with
the numerical simulation. The experimental data (Table 3.5) indicate that distortion decreases near weld beads
1 and 3. This is also consistent with the numerical simulations.

Point

L-type specimen No.1

L-type specimen No.2
1 0,25 0,14
2 0,51 0,26
3 0,63 0,59
4 0,92 0,81
5 1,05 1,06
6 1,34 1,20
7 1,77 1,34
8 1,58 1,42
9 1,38 1,22
10 1,24 1,06
11 1,02 0,98
12 0,82 0,59
13 0,51 0,36
14 0,32 0,33
15 0,25 0,17
16 0,42 0,30
17 0,52 0,53
18 0,79 0,81
19 1,05 0,98
20 1,32 1,10
21 1,41 1,10
22 1,25 0,99
23 0,96 0,89
24 0,79 0,59
25 0,65 0,38
26 0,41 0,22

Table 3.5 Distortion at the points of interest in two L-type specimens

The experimental difference in the maximum distortion between the two specimens is about 20%, while
the numerical difference is about 32%. This difference is attributed to the lack of an automatic welder leading
to geometric irregularities in the weld beads.


75

Figure 3.20 depicts the comparison of the welding distortion between the two L-type specimens. This figure
shows that specimen 1 (welding sequence 2), results in the highest distortion.

Figure 3.20 Comparison of the welding distortion (exaggerated) between two L-type specimens

3.20 Conclusions of the welding sequence analysis of the L-type structure

The maximum residual stresses occur in the boundary between the weld metal and the base metal.
The welding sequence that produces the lowest distortion also produces the highest residual stress in
the base metal.
The welding sequence that produces the lowest residual stress in the base metal also produces the
highest distortion.
Numerical predictions for distortion are consistent with experimental results.
The welding sequence that produces the lowest distortion in the numerical simulations also produces
the lowest distortion in the experimental tests.


76

CHAPTER IV

WELDING SEQUENCE ANALYSIS
IN A STIFFENED SYMMETRICAL
2-DIMENSIONAL FRAME

4.1 Introduction

This chapter presents a study of the effects of the welding sequence on residual stresses and distortion in a
stiffened symmetrical flat frame. The proper welding sequences obtained in this analysis are chosen to reduce
residual stresses, distortion, or the relation between them. These proper welding sequences are obtained with
the help of empirical welding rules: the axis of symmetry, center of gravity of the frame, and concentric circle.
The origin of the circles coincides with the center of gravity of the flat frame, and the radius of the circles is
formed by the center of gravity of the flat frame with the center of gravity of each of the weld beads. To carry
out the different welding sequences, the numerical model developed in chapter 3 is used with the help of a
commercial code. At the end of this chapter we propose and demonstrate a hypothesis to determine the
proper welding sequences to reduce residual stress, distortion, or the relation between them for 2-

4.2 Geometric configuration of a stiffened symmetrical flat frame

The proposed symmetrical flat structure has the shape of a stiffened frame consisting of four external bars
and three internal bars joined by ten weld beads (Figure 4.1). This configuration is selected due to its multiple
symmetries. The numbering of the weld beads is random.


77

Figure 4.1 Geometric configuration of the stiffened symmetrical flat frame (mm)

4.3 Welding configuration in the stiffened symmetrical flat frame

The number of possible welding sequences depends on the number of weld beads [24]:

(39)

Applying equation (39), the number of possible welding sequences for the frame in figure 4.1 is 10! =
3,628,800. A possible approach is to run all possible combinations and select the sequence that produces the
lowest distortion, the lowest residual stress or an appropriate relation between them. This is not feasible work
due to the large number of possible combinations. Symmetry reduces the number of welding sequences to
1,088,640 still impossibly large for analysis of all options. We therefore consider empirical rules: applying the
weld beads from the inside to the outside of the structure produces 9 sequences; and applying weld beads
from the outside to the inside of the structure produces 9 more sequences. Therefore, only eighteen welding
sequences need to be considered.


78
Welding
sequence
Weld beads
1 and 2
Weld beads
3, 4, 5 and 6
Weld beads
7, 8, 9 and 10
Configuration
1 I-O

1-2-3-4-5-6-7-8-9-10
2 I-O

1-2-5-3-4-6-8-7-10-9
3 I-O

1-2-3-6-4-5-7-8-9-10
4 I-O

1-2-3-4-5-6-7-9-8-10
5 I-O

1-2-5-3-4-6-8-10-7-9
6 I-O

1-2-3-4-6-5-9-7-8-10
7 I-O

1-2-3-4-5-6-7-10-8-9
8 I-O

1-2-5-3-4-6-8-7-9-10
9 I-O

1-2-3-6-4-5-7-10-8-9

Table 4.1 Welding sequences from the inside to the outside used in the stiffened symmetrical flat frame


79
Welding
sequence
Weld beads
7, 8, 9 Y 10
Weld beads
3, 4, 5 Y 6
Weld beads
1 Y 2
Configuration
1 O-I

10-9-8-7-6-5-4-3-2-1
2 O-I

9-10-7-8-6-4-3-5-2-1
3 O-I

10-9-8-7-5-4-6-3-2-1
4 O-I

10-8-9-7-6-5-4-3-2-1
5 O-I

9-7-10-8-6-4-3-5-2-1
6 O-I

10-8-7-9-5-6-4-3-2-1
7 O-I

9-8-10-7-6-5-4-3-2-1
8 O-I

10-9-7-8-6-4-3-5-2-1
9 O-I

9-8-10-7-5-4-6-3-2-1

Table 4.2 Welding sequences from the outside to the inside used in the stiffened symmetrical flat frame


80
In these welding sequences, the weld beads are split in 3 groups according to their relative position to the
center of gravity. A sequence is specified by the bead order 1,2,.. 9 and the letters I-O (starting from the inside)
or O-I (starting from the outside). The term WT means that welding tacks are used to keep the structure fixed
while performing the welding process. Experience shows that the welding tacks (WT), together with the
welding sequence, have important effects on residual stresses and distortion in the welded structures.
Therefore, each of the 18 welding sequences will be considered with and without welding tacks, for a total of
36 numerical simulations a tractable problem. The configurations of these welding sequences are shown in
tables 4.1 and 4.2.

4.4 Finite element model of the stiffened symmetrical flat frame

The numerical simulations of the different welding sequences are based on the numerical model of the
welding process developed in chapter 3, considering the time between the application of a weld bead and the
next. To simulate the structure with welding tacks, the elements of the welded region are activated during all
the analysis. To simulate the structure without welding tacks, the elements of the welded region are first
deactivated and then activated in accordance with the welding sequence. Thermal and mechanical properties
are deactivated by means of a reduction factor (1e-6). The welding region has a fine mesh and the zones away
from the weld bead are meshed coarser. This consideration reduces the processing time and still guarantees
computational accuracy. Figure 4.2 shows the finite element model of the stiffened symmetrical flat frame.

Figure 4.2 Finite element model of the stiffened symmetrical flat frame


81



b) Boundary conditions (see figure 4.2, page 80):

Surfaces in contact with the surrounding except surfaces belonging to the weld beads (Plane x-y in z=0):




Three boundary conditions are applied to structural model: a) those just sufficient to prevent rigid body
motion and rotations of the model and b) stress-free on Z-direction (plane-stress problem), which imply that
all out-of-plane stresses vanish, and c) the normal stresses to the surfaces in contact with the surrounding and
the shear stresses are zero.

a) The nodes belonging to the element attached to its center of gravity of the structure are selected to
prevent rigid body motion and rigid body rotations (see figure 4.2, page 80):

; ; ; and

b) In the plane-stress problem, the normal stresses to the plane X-Y ( ) and shearing stress acting in this
plane ( and ) are zero on face of the structure (see figure 4.2, page 80). Therefore, .

c) Surfaces in contact with the surrounding (see figure 4.2, page 80):

x=0, ).


82
4.4.3 Solution of the finite element model of the stiffened symmetrical flat frame

loads), the solution can be initiated. In the transient thermal analysis, the time to complete each weld bead is
6 seconds. Thermal elastic plastic analysis for each simulation requires 33 load steps to complete the weld
thermal cycles.

4.5 Numerical results in the stiffened symmetrical flat frame

The parameters of interest are maximum von Mises residual stress in the base metal and maximum
distortion in the stiffened symmetrical flat frame.

4.5.1 Distribution of the residual stresses in the stiffened symmetrical flat frame

The maximum von Mises residual stress in the metal base occurs at the boundary of weld bead 6 and an
internal bar (node 681). This is expected because the internal bars have a high degree of constraint due to the
connection to the external bars. In the central part of all the bars the magnitude of the von Mises residual
stress is almost zero. Welding sequence 5I-O shows the lowest von Mises residual stress (199.84 MPa) of all
the welding sequences analyzed (Figure 4.3).

Figure 4.3 Distribution of Von Mises residual stresses corresponding to welding sequence 5 I-O


83

4.5.2 Distortion profile in the stiffened symmetrical flat frame

The maximum distortion occurs in different areas of the external bars depending on the welding sequence.
The internal bars have low distortion because they are restricted by the external bars. Welding sequence 5 O-I
WT results in the lowest distortion (Figure 4.4).

Figure 4.4 Distortion profile corresponding to welding sequence 5 O-I WT

Analyzing the results, it is observed that the maximum Von Mises residual stress can be reduced by 15%
while the maximum distortion can be reduced by 35% by applying a proper welding sequence.

4.6 Analysis of residual stress-distortion relations analysis

The following equation is proposed for quantitative comparison of the effects of the welding sequence on
residual stress, distortion, or the relation between them:

(40)


84

Where: and are the residual stress and distortion for each welding sequence. The residual
stress corresponds to one selected point on the structure and the distortion corresponds to the maximum
distortion that occurs in each welding sequence. The highest residual stress that occurs in all welding
sequences is accompanied by lowest distortion. The highest distortion that occurs in all welding
sequences , is accompanied by lowest residual stress. Weight factors W and P can be assigned to
residual stress and distortion, respectively. For the case when the welding sequence only reduces the residual
stress, and . When the welding sequence only reduces distortion, and . When the
welding sequence reduces the relation between both parameters, .

4.7 Order of importance of the welding sequences to reduce residual stress, distortion, or the relation
between them in the stiffened symmetrical flat frame

Applying equation (40) [page 83] to the results of the residual stress and distortion obtained in the 36
numerical simulations, the order of importance of the different welding sequences to reduce residual stress,
distortion, or a relation between them is obtained (table 4.3).

Analyzing these results, the proper welding sequence to reduce residual stress is 5 I-O (from the inside to
the outside of the structure), and the proper welding sequence to reduce the distortion is 5 O-I WT (from the
outside to the inside using welding tacks). This implies that the proper welding sequence to reduce the
residual stresses is inappropriate to reduce the distortion and vice versa. Finally, welding sequence 1 I-O
(welding sequence in growing spiral) shows the best overall performance: it is ranked number 4 in reducing
stress and 19 in reducing distortion. Therefore, welding sequence 1 I-O is the proper welding sequence to
improve the relation between the selected critical parameters.

4.8 Proper welding sequences to reduce the residual stress, distortion, or a relation between them in the
stiffened symmetrical flat frame

To formulate the proper welding sequences obtained in table 4.3, the axis of symmetry and the center of
gravity of the flat frame are used. We draw concentric circles centered on the center of gravity of the frame
and extending to the center of gravity of the weld beads. The weld beads located at the same distance from
the center of gravity of the frame fall on the same circle. The circles are numbered from the smallest to the
largest. Weld beads 1 and 2 belong to circle 1, weld beads 3, 4, 5, and 6 belong to circle 2, and weld beads 7, 8,
9, and 10 belong to circle 3 (Figure 4.5).


85
Welding
sequence
Parameters Order of importance of welding sequence to reduce:
Residual stress Distortion
Umax
(mm)
vm
W=1, P=0 Node
681 (MPa)
proper = 1
inappropriate= 36
W=0, P=1 proper = 1
inappropriate= 36
1 I-O 0.286 205.48 0.872 4 0.905 19
2 I-O 0.314 201.29 0.854 2 0.994 35
3 I-O 0.293 208.27 0.883 12 0.927 25
4 I-O 0.305 205.73 0.873 6 0.965 32
5 I-O 0.316 199.84 0.848 1 1 36
6 I-O 0.297 207.6 0.881 10 0.940 28
7 I-O 0.309 206.46 0.876 8 0.978 33
8 I-O 0.312 201.72 0.856 3 0.987 34
9 I-O 0.290 208.54 0.885 14 0.918 23
1 O-I 0.277 206.3 0.875 7 0.877 17
2 O-I 0.220 223.64 0.949 31 0.696 8
3 O-I 0.228 216.30 0.918 30 0.722 10
4 O-I 0.278 205.52 0.872 5 0.880 18
5 O-I 0.214 224.22 0.951 33 0.677 3
6 O-I 0.230 215.32 0.913 27 0.728 11
7 O-I 0.277 210.90 0.895 18 0.877 16
8 O-I 0.223 223.70 0.949 32 0.706 9
9 O-I 0.235 214.29 0.909 24 0.744 12
1 I-O WT 0.296 210.33 0.892 15 0.937 26
2 I-O WT 0.299 208.1 0.883 13 0.946 30
3 I-O WT 0.290 212.3 0.901 23 0.918 22
4 I-O WT 0.289 210.53 0.893 16 0.915 21
5 I-O WT 0.300 206.79 0.877 9 0.949 31
6 I-O WT 0.291 211.95 0.899 22 0.921 24
7 I-O WT 0.297 211.43 0.897 20 0.940 27
8 I-O WT 0.298 207.8 0.881 11 0.943 29
9 I-O WT 0.287 214.67 0.911 26 0.908 20
1 O-I WT 0.260 211.05 0.895 19 0.823 14
2 O-I WT 0.212 231.75 0.983 35 0.671 2
3 O-I WT 0.219 214.58 0.910 25 0.693 7
4 O-I WT 0.266 210.45 0.893 17 0.842 15
5 O-I WT 0.206 235.74 1 36 0.652 1
6 O-I WT 0.216 215.20 0.913 28 0.684 5
7 O-I WT 0.260 211.52 0.897 21 0.823 13
8 O-I WT 0.215 231.39 0.982 34 0.680 4
9 O-I WT 0.217 215.75 0.915 29 0.687 6
Table 4.3 Order of importance of the analyzed welding sequences in the stiffened symmetrical flat frame


86

Figure 4.5 Centers of gravity of the structure, centers of gravity of the weld beads, axis of symmetry,
and concentric circles in the stiffened symmetrical flat frame

4.8.1 Proper welding sequence to reduce the residual stress in the stiffened symmetrical flat frame

This welding sequence begins by selecting the weld beads belonging to the smallest circle and then
continuing with the weld beads belonging to the larger circle until the largest circle is reached. Weld beads
within the same circle should be selected symmetrically. First, the weld beads with diagonal symmetry are
selected. If the weld bead with diagonal symmetry has already been selected, then the farthest symmetrical
weld bead is selected next. When a current weld bead has more than one farthest symmetrical weld bead
(weld beads with the same radius); the adjacent weld bead in counter-clockwise is selected next. Finally, to
move from one circle to the other, the farthest weld bead to the current weld bead is selected next. If a
current weld bead has more than one farthest weld bead, the adjacent weld bead in counter-clockwise is
selected.


87
4.8.2 Proper welding sequence to reduce distortion in the stiffened symmetrical flat frame

This welding sequence requires applying weld tacks to the structure. The sequence begins by selecting the
weld beads belonging to the largest circle, and then continuing with the weld beads belonging to the adjacent
smaller circle until the smallest circle is reached. The application of the weld beads belonging to the same
circle should be selected in a symmetrical form. First, the weld beads with diagonal symmetry are selected. If
the weld bead with diagonal symmetry has already been selected previous to the current weld bead, then the
farthest symmetrical weld bead is selected next. When a current weld bead has more than one farthest
symmetrical weld bead, the adjacent weld bead in counter-clockwise is selected next. Finally, to move from
one circle to the other, the farthest weld bead to the current is selected next. If a current weld bead has more
than one farthest weld bead, the adjacent bead in a clockwise manner is selected.

4.8.3 Proper welding sequence to improve the relation between both critical parameters in the stiffened
symmetrical flat frame

It is observed in table 4.3 that the welding sequence 1 I-O shows the best stress-distortion relation, ranking
number 4 in stress reduction and 19 in distortion reduction. Therefore, welding sequence 1 I-O is selected to
improve the relation between the two parameters. However, to understand the mechanism leading to the
residual stresses and distortion observed in the numerical simulations, a combined welding sequence is
proposed. Therefore, a comparison between welding sequence 1 I-O and the combined welding sequence will
be performed. Figure 4.6 depicts the ranges of the values of P and W for the welding sequence 1 I-O, and the
combined welding sequence. This figure shows that a welding sequence is better than the other depending on
the values of P and W.

Figure 4.6 Ranges of the values of P and W for welding sequence 1 I-O
and combined welding sequence


88

For: and

This welding sequence starts with the beads in the smallest circle and then continuing with the beads in the
larger adjacent circle until the largest circle is reached. Weld beads within the same circle should be applied in
counter-clockwise direction. To move from one circle to the next, the closest weld bead is selected.

For: and

In this welding sequence, welding tacks are required only on the circle next to the larger adjacent circle.
The sequence begins by selecting the weld beads on the smallest circle, and then continuing with the weld
beads on the circle next to the larger adjacent circle and continuing with the beads on the smaller adjacent
circle. The application of the beads on the same circle should be symmetrical. First, the weld beads with
diagonal symmetry are applied. If the weld bead with diagonal symmetry has already been applied, then the
farthest symmetrical weld bead is selected next. When a current weld bead has more than one farthest
symmetrical bead, the adjacent bead in counter-clockwise direction is selected next. Finally, to move from one
circle to the other, the farthest weld bead to the current bead is selected next. If a current weld bead has more
than one farthest bead, the adjacent bead in counter-clockwise direction is selected.

4.9 Hypothesis to determine the proper welding sequence to reduce the residual stress, distortion, or a
relation between them in symmetrical flat structures

To determine the proper welding sequences to reduce the residual stress, distortion, or a relation between
them in flat symmetrical structures, it is necessary to determine the axis of symmetry and the center of gravity
of the structure. We next draw concentric circles centered on the center of gravity of the structure and
extending to the center of gravity of the weld beads. Weld beads that are located at the same distance from
the center of gravity of the structure fall on the same circle. Circles are numbered from the smallest to the
largest diameter.

a) To reduce residual stress: Start with the weld beads on the smallest circle and then continue with the
beads on the larger adjacent circle until the largest circle is reached. The weld beads on the same circle should
be selected in a symmetrical form. First, the weld beads with diagonal symmetry are selected. If the bead with
diagonal symmetry has already been selected, then the farthest symmetrical bead is selected. When a weld
bead has more than one farthest symmetrical weld bead, the adjacent weld bead in counter-clockwise
direction is selected. Finally, to move from one circle to the other, the farthest weld bead to the current weld
bead is selected. If a current weld bead has more than one farthest weld bead, the adjacent weld bead in
counter-clockwise direction is selected.


89

b) To reduce distortion: Apply welding tacks to the structure before welding. First, the weld beads in the
largest circle are selected and then the beads on the adjacent smaller circle are selected until the smallest
circle is reached. Beads on the same circle should be applied symmetrically. Beads with diagonal symmetry are
selected first. If the weld bead with diagonal symmetry has already been previously selected, then the farthest
symmetrical bead is selected next. When a current weld bead has more than one farthest symmetrical weld
bead, the adjacent weld bead in counter-clockwise direction is selected next. Finally, to move from one circle
to the other, the farthest weld bead is selected. If a current bead has more than one farthest bead, the
adjacent bead in a clockwise manner is selected.

c) To improve the relation between residual stress and distortion:

i) Favoring the residual stress: Start with the weld beads belonging to the smallest circle and then
continuing with the beads on the larger adjacent circle until the largest circle is reached. The application of
weld beads belonging to the same circle should counterclockwise into adjacent beads. To move from one
circle to the other, the closest weld bead to the current weld bead is selected.

ii) Favoring the distortion: This welding sequence requires application of welding tacks only on the circle
next to the larger adjacent circle. The welding sequence starts by selecting the beads on the smallest circle,
and then continuing with the beads on the circle next to the larger adjacent circle, and later going back to the
beads on the smaller adjacent circle. The application of the weld beads on the same circle should be in a
symmetrical form. First, the beads with diagonal symmetry are applied. If the weld bead with diagonal
symmetry has already been applied, then the farthest symmetrical bead is selected next. When a current bead
has more than one farthest symmetrical bead, the adjacent bead in counter-clockwise direction is selected
next. Finally, to move from one circle to the other, the farthest bead to the current is selected next. If a current
weld bead has more than one farthest bead, the adjacent bead in counter-clockwise direction is selected.
For the stiffened symmetrical flat frame, the hypothesis to determine the proper welding sequence to
reduce residual stress, distortion, or the relation between them in symmetrical structures in the plane is
verified.

4.10 Experimental tests in a stiffened symmetrical flat frame specimen

We conducted an experiment to validate the numerical results. All welding parameters used in the
numerical simulations are reproduced in the experiment. Welding tacks are applied to a stiffened symmetrical
flat specimen before the welding process starts. Measurement of the distortion in the stiffened symmetrical
plane specimen is performed in a BH-V504 MITUTOYO coordinate measuring machine (resolution: 0.0001
mm). This machine is located in the manufacture laboratory at the University of Guanajuato campus
Salamanca-Irapuato.


90
4.10.1 Selection of points of interest in the stiffened symmetrical flat specimen

To measure distortion, 88 points of interest are selected (Figure 4.7). These points are located on the edges
of all bars. A reference frame at the internal horizontal bar is marked to position the stiffened symmetrical flat
specimen.

Figure 4.7 Localization of the point of interest in the stiffened symmetrical flat specimen

4.10.2 Configuration of the welding sequence in the stiffened symmetrical flat specimen

As previously mentioned, the stiffened symmetrical flat specimen follows the welding sequence that
minimizes distortion (5 O-I WT, Table 4.4 and figure 4.8). This sequence first applies beads at locations farthest
from the center and then moves progressively inward. Weld beads at equal distance from the center of the
specimen are selected symmetrically.

5 O-I WT 9-7-10-8-6-4-3-5-2-1
Table 4.4 Configuration of welding sequence 5 O-I WT
used in the stiffened symmetrical flat specimen


91

Figure 4.8 Stiffened symmetrical flat specimen after applying
the welding sequence 5 O-I WT

4.10.3 Measurement of distortion on the stiffened symmetrical flat specimen

The distortion induced by the welding process is determined by subtracting the distortion after applying
the weld tacks from the distortion after applying the weld beads. Figure 4.9 shows the stiffened symmetrical
flat specimen mounted on the coordinate measuring machine.

Figure 4.9 Stiffened symmetrical flat specimen mounted
on the coordinate measuring machine


92
The maximum distortion in the specimen occurs in the external bars and drops in value as the location
moves progressively inwards to the structure (Table 4.5), in agreement with the numerical results.

Point Distortion Point Distortion
1 0,82 45 0,44
2 0,84 46 0,10
3 0,79 47 0,07
4 0,94 48 0,10
5 0,90 49 0,25
6 0,94 50 0,19
7 0,84 51 0,24
8 0,84 52 0,37
9 0,74 53 0,68
10 0,64 54 0,62
11 0,76 55 0,53
12 0,64 56 0,55
13 0,61 57 0,44
14 0,58 58 0,51
15 0,54 59 0,30
16 0,58 60 0,14
17 0,54 61 0,09
18 0,48 62 0,32
19 0,47 63 0,41
20 0,51 64 0,53
21 0,74 65 0,12
22 0,59 66 0,19
23 0,41 67 0,34
24 0,29 68 0,60
25 0,37 69 0,47
26 0,43 70 0,32
27 0,32 71 0,37
28 0,24 72 0,40
29 0,25 73 0,23
30 0,40 74 0,50
31 0,45 75 0,21
32 0,28 76 0,14
33 0,67 77 0,20
34 0,17 78 0,10
35 0,22 79 0,06
36 0,20 80 0,15
37 0,29 81 0,18
38 0,39 82 0,15
39 0,62 83 0,30
40 0,64 84 0,30
41 0,59 85 0,15
42 0,38 86 0,23
43 0,64 87 0,15
44 0,51 88 0,07
Table 4.5 Distortion at the points of interest (mm) in the stiffened symmetrical flat
specimen corresponding to welding sequence 5 O-I WT


93

The maximum measured distortion is four times higher in the Z-direction. This is probably because the
welding process was not automated, producing geometric irregularities in the weld bead. Figure 4.10 depicts
the measured distortion profile, indicating that the maximum distortion occurs on the external bars and
progressively decreases as it approaches to center of the specimen.

Figure 4.10 Distortion profile in the stiffened symmetrical flat specimen
corresponding to welding sequence 5 O-I WT (exaggerated)

4.11 Conclusions of the welding sequence analysis of stiffened symmetrical flat frame

The maximum residual stress in the metal base occurs in the boundary with the welded metal in the
internal bars.
For the stiffened symmetrical flat frame, the distortion is reduced by 35 %, while the maximum von
Mises residual stress is reduced by 15% when applying the proper welding sequence to respectively
reduce distortion or residual stress. The welding sequence is twice as effective for reducing distortion
than for reducing residual stresses.
Welding tacks are not recommended when trying to reduce residual stress.
Welding tacks are necessary to reduce distortion.


94
Welding tacks followed by weld bead application from the outside to the inside of the structure reduces
distortion but increases residual stresses.
Applying the weld beads from the inside to the outside of the structure without welding tacks reduces
the residual stress but increases the distortion.
The welding sequence that produces the lowest distortion also produces the highest residual stress.
The welding sequence that produces the lowest residual stress also produces the highest distortion.
The relation between P=0,5 and W=0,5 does not mean that both parameters are reduced by the same
percentage, because the proper welding sequence reduces the distortion twice as much as residual
stresses.
There is good qualitative agreement between the measured and numerical results. Quantitatively,
however, the numerical results show lower maximum peak values. The difference is probably due to the
lack of an automated welder.


95

CHAPTER V

WELDING SEQUENCE ANALYSIS IN A
3-DIMENSIONAL UNITARY CELL-TYPE
SYMMETRICAL STRUCTURE

5.1 Introduction

This chapter describes a study of the effects of the welding sequence on the residual stresses and distortion
in a 3-dimensional unitary cell, with the purpose of demonstrating that the methodology for determining the
proper welding sequences to reduce residual stress, distortion, or the relation between them in 2-dimensional
symmetrical structures can be applied to 3-dimensional symmetrical structures, where weld bead circles
become spheres. The methodology is demonstrated by conducting four numerical simulations in the proposed
symmetrical structure. Two of the numerical simulations deal with the proper welding sequence to reduce
residual stress and the other two deals with the proper welding sequence to reduce distortion. A special
welding sequence is also analyzed for comparison with the proper sequence to reduce distortion. This special
welding sequence applies the external beads first and the internal beads later. All the numerical simulations
are based on the proposed numerical model of the welding process developed in Chapter 3 with a commercial
software code.


96

5.2 Hypothesis to determine the proper welding sequence to reduce the residual stress, distortion, or a
relation between them in 3-dimensional symmetrical structures

To determine the proper welding sequences to reduce the residual stress, distortion, or a relation between
them in 3-dimensional symmetrical structures, we start by identifying the axis of symmetry and the center of
gravity of the structure. Next, we draw concentric spheres centered on the center of gravity of the structure,
and extending to the center of gravity of each of the weld beads. Weld beads located at the same distance
from the center of gravity of the structure will fall on the same sphere. The spheres are numbered from the
smallest to the largest.

a) To reduce residual stress: Start with the weld beads belonging to the smallest sphere and then continue
with the weld beads belonging to the larger adjacent sphere until the largest sphere is reached. The weld
beads belonging to the same sphere should be symmetrical. First, the beads with diagonal symmetry are
selected. If the weld bead with diagonal symmetry has already been selected, then the farthest symmetrical
weld bead is selected next. When a current weld bead has more than one farthest symmetrical bead, the
adjacent weld bead in counter-clockwise is selected. Finally, to move from one sphere to the other, the
farthest weld bead to the current bead is selected. If a current bead has more than one farthest weld bead,
the adjacent bead in counter-clockwise is selected. The converse welding method between weld beads is
adopted.

b) To reduce distortion: Apply welding tacks to the structure before welding. First, the weld beads in the
largest sphere are selected and then the weld beads belonging to the adjacent smaller sphere are selected
until the smallest sphere is reached. The application of the weld beads on the same sphere should be in
symmetrical form. The weld beads with diagonal symmetry are selected first. If the weld bead with diagonal
symmetry has already been selected, then the farthest symmetrical bead is selected next. When a weld bead
has more than one farthest symmetrical bead, the adjacent weld bead in counter-clockwise direction is
selected next. Finally, to move from one sphere to the other, the farthest weld bead to the current is selected.
If a weld bead has more than one farthest bead, the adjacent bead in a clockwise direction is selected. The
converse welding method between beads is adopted.

c) To improve the relation between residual stress and distortion:

i) Favoring residual stress: Start with the weld beads on the smallest sphere and then continue with the
weld beads on the larger adjacent sphere until the largest sphere is reached. The application of weld beads on
the same sphere should be in an adjacent form in counter-clockwise direction. To move from one sphere to
the other, the closest weld bead to the current bead is selected. The converse welding method between weld
beads is adopted.


97

ii) Favoring distortion: In this welding sequence, the application of welding tacks is required only on the
sphere next to the larger adjacent sphere. The welding sequence begins by selecting the weld beads on the
smallest sphere, and then continuing with the weld beads on the sphere next to the larger adjacent sphere.
Then, going back to the weld beads on the smaller adjacent sphere. The application of the weld beads on the
same sphere should be symmetrical. First, the weld beads with diagonal symmetry are applied. If the weld
bead with diagonal symmetrical has already been applied, then the farthest symmetrical weld bead is selected
next. When a current weld bead has more than one farthest symmetrical bead, the adjacent bead in counter-
clockwise direction is selected next. Finally, to move from one sphere to the other, the farthest weld bead to
the current weld bead is selected next. If a current weld bead has more than one farthest weld bead, the
adjacent weld bead in counter-clockwise direction is selected. The converse welding method between weld
beads is adopted.

5.3 Geometric configuration of the 3-dimensional unitary cell

A 3-dimensional unitary cell configuration is selected because it is a common structural arrangement in
industry, used in the construction of panels, ships, bridges, etc. This configuration keeps the requirement of
symmetry in the space. Figure 5.1 depicts a structural panel used in the welding industry, where a segment
takes the form of a 3-dimensional unitary cell.

Figure 5.1 Panel used in the welding industry


98

The 3-dimensional unitary cell consists of two horizontal and two vertical plates joined by fillet welds as
shown in figure 5.2. The thickness of the plates is 3 mm.

Figure 5.2 Configurations and dimensions of the 3-dimensional unitary cell (mm)

5.4 Selection of the number of weld beads in the 3-dimensional unitary cell

The 3-dimensional unitary cell is joined by eight fillet welds. The effect of dividing a fillet weld into sub-
welds is studied. For the present investigation, each of the fillet welds is divided in 3 sub-welds.

5.5 Symmetry axis selection and formation of the concentric spheres in the 3-dimensional unitary cell

Due to the multiple symmetry (symmetry and diagonal symmetry axis) of the 3-dimensional unitary cell,
the center of mass coincides with the centroid of the structure, which is also the center of the spheres.


99
Two concentric spheres are formed with 8 weld beads and four concentric spheres with 24 weld beads
(Figure 5.3).

Figure 5.3 Axis of symmetry and concentric spheres of the 3-dimensional
unitary cell formed by 24 fillet welds


100

5.6 Proper welding sequence to reduce the residual stress and proper welding sequence to reduce the
distortion in the 3-dimensional unitary cell

Figures 5.4 (a) and (c) depict the proper welding sequence to reduce distortion with 8 and 24 fillet welds,
respectively. Figures 5.4 (b) and (d) depict the proper welding sequence to reduce the residual stresses with 8
and 24 fillet welds, respectively.

Figure 5.4 Different welding sequences for the 3-dimensional unitary cell


101

5.7 Material selection of the 3-dimensional unitary cell

Structural steel S235JR is selected to define the material properties of the 3-dimensional unitary cell. This
steel is widely used in Europe (Poland: steel ST3S) for construction of ships, buildings, bridges and machinery
[36]. This steel is easy to weld because of its carbon content (0.22% C), and its chemical composition,
mechanical and thermal properties are similar to ASTM A36 structural steel. Steel S235JR is therefore modeled
using properties of ASTM A36 structural steel.

5.8 Fillet weld shape used in the 3-dimensional unitary cell

Because the 3-dimensional unitary cell is joined with fillet welds, it is important to model the shape of the
fillet weld as close as possible to reality. For this reason, a common shape found in the welding literature is
proposed [1]. Figure 5.5 depicts the shape of the fillet weld used in the 3-dimensional unitary cell.

Figure 5.5 Fillet weld shape used in the 3-dimensional unitary cell

5.9 Finite element model of the 3-dimensional unitary cell

The numerical simulations of the welding sequences for the 3-dimensional unitary cell are based on the
numerical model of welding process developed in Chapter 3. To simulate the structure with welding tacks, the
elements of the welded region are activated during all of the analysis. To simulate the structure without
welding tacks, the elements of the welded region are originally turned off and then activated in accordance to
the welding sequence. This is accomplished by reducing thermal and mechanical properties of the elements by
multiplying them by 1e-6. The analysis considers the time between the application of weld beads and the
welding direction. The region being welded has a fine mesh and the zones away from the weld beads are
meshed coarser. This consideration reduces the processing time and still guarantees computationally accuracy.


102
The numerical model of the 3-dimensional unitary cell is divided into 29,362 hexahedral elements and 41,202
nodes. Figure 5.6 shows the finite element model of the 3-dimensional unitary cell.

Figure 5.6 Finite element model of the 3-dimensional unitary cell




b) Boundary conditions (see figure 5.6):

Surfaces in contact with the surrounding except surfaces belonging to the weld beads:



103


a) The specified mechanical boundary conditions are those just sufficient to prevent rigid body motion and
rotations of the model, the nodes belonging to the element localized in the center of one lateral edge at the
bottom horizontal plate are selected to prevent rigid body motion and rigid body rotations (see figure 5.6,
page 102):

; ;

; ;

b) Surfaces in contact with the surrounding (see figure 5.6):

the plane y-z in x=0, ).

5.9.3 Solution of the finite element model of the 3-dimensional unitary cell

loads), the solution can be initiated. In the transient thermal analysis, the time to complete each weld bead is
6 seconds. Thermal elastic plastic analysis for each simulation requires 55 load steps to complete the weld
thermal cycles.

5.10 Localization of the points of interest in the 3-dimensional unitary cell

The selected points of interest belong to important areas in the 3-dimensional unitary cell (Figure 5.7).
Points 1 and 3 are located in the center of the horizontal plates, while points 2 and 4 are located in the center
of the vertical plates. Points 5 and 8 and points 6 and 7 are located in the central part of the lower and upper
flanges, respectively. Points 9, 12, 15 and 18, and points 11, 14, 17 and 20 are respectively located in the edges
of the lower and upper flanges. Points 10, 13, 16 and 19 are located in the central part of the lateral edges of
the vertical plates, while points 21, 22, 23 and 24 are located in the central part of the lateral edges of the


104
horizontal plates. The magnitude of the distortion is positive when the displacement direction is inwards to the
structure.

Figure 5.7 Localization of the points of interest in the 3-dimensional unitary cell

5.11 Configuration of the numerical simulation for the 3-dimensional unitary cell

To demonstrate the hypothesis proposed in Section 5.2, four numerical simulations will be performed. Two
of them deal with the proper welding sequence to reduce the distortion with 8 and 24 weld beads, and the
other two deals with the proper welding sequence to reduce the residual stress with 8 and 24 weld beads. The
numerical simulations for reduced distortion include the effects of the welding tacks. Table 5.1 depicts the
configuration of the numerical simulations for the 3-dimensional unitary cell.

Numerical
simulation
Proper welding sequence
to reduce:
Number of
welds
Welding
tacks
figure
1 Distortion 8 Yes 5.4 (a)
2 Residual stress 8 No 5.4 (b)
3 Distortion 24 Yes 5.4 (c)
4 Residual stress 24 No 5.4 (d)

Table 5.1 Configuration of the numerical simulations for the 3-dimensional unitary cell


105

5.12 Numerical results of the different welding sequences analyzed in the 3-dimensional unitary cell

The numerical results of interest are the maximum von Mises residual stresses in the base metal, the
distortion modes, the maximum distortion, and the distortion in the 24 points selected in the unitary 3-
dimensional unitary cell.

5.12.1 Maximum von Mises residual stress in the 3-dimensional unitary cell

The maximum von Mises residual stress in the base metal occurs in the central internal weld beads, at the
boundary with the weld metal. Compressive residual stress occurs in the central region of the vertical and
horizontal plates, while tensile residual stress occurs in the weld metal and nearby areas. Numerical simulation
2 (Figure 5.8) results in the lowest residual stress, therefore confirming the hypothesis proposed in Section 5.2.

Figure 5.8 Distribution of residual stresses in numerical simulation 2


106
5.12.2 Distortion modes in the 3-dimensional unitary cell

All the numerical simulations produce the same distortion mode. However, the magnitude of the distortion
varies. The lower and upper flanges rotate inward to the structure, while the center plates bend towards the
inside of the structure. The central regions of the vertical and horizontal flanges also bend inward towards the
structure. Numerical simulation 3 (Figure 5.9) results in the lowest distortion. This simulation deals with the
proper welding sequence to reduce distortion, therefore confirming the hypothesis proposed in section 5.2.

Figure 5.9 Isometric view of the distortion profile corresponding to numerical simulation 3

5.12.3 Maximum distortion and distortion in 24 points of interest in the 3-dimensional unitary cell

In all the simulations, the maximum distortion occurs in the central region of the upper and lower flanges
(points 5, 6, 7 and 8). Table 5.2 depicts the distortion at the 24 points of interest for the different welding
sequences.


107
The table shows that simulation 3 produces the lowest distortion. However, this welding sequence results in
the highest von Mises residual stress. Once again, the proper welding sequence to reduce the residual stress
results in highest distortion.

Point Node Simulation
1 2 3 4
1 15756 0,08 0,05 0,06 0,04
2 17536 0,11 0,01 0,21 0,05
3 19015 0,08 0,07 0,06 0,04
4 17238 0,11 0,02 0,21 0,05
5 14571 1,85 1,87 1,68 1,65
6 20554 2,0 2,0 1,75 1,85
7 20508 1,82 1,83 1,67 1,64
8 14498 1,99 1,98 1,73 1,80
9 11 1,36 1,70 1,52 1,12
10 386 0,31 0,18 0,35 0,19
11 758 1,54 1,92 1,53 1,13
12 35172 1,49 1,86 1,55 1,11
13 38137 0,08 0,07 0,2 0,19
14 41155 1,57 1,95 1,74 1,24
15 35099 1,51 1,88 1,50 1,08
16 37839 0,39 0,10 0,3 0,08
17 41109 1,33 1,65 1,54 1,13
18 1 1,56 1,94 1,74 1,21
19 361 0,19 0,10 0,16 0,08
20 754 1,47 1,83 1,56 1,01
21 167 0,57 0,25 0,66 0,29
22 36357 0,59 0,34 0,59 0,34
23 591 0,57 0,31 0,6 0,33
24 39616 0,52 0,22 0,68 0,29
Maximum distortion 2,97 3,05 2,04 2,10

Table 5.2 Distortion at 24 points of interest (mm) corresponding to different welding sequences
In the 3-dimensional unitary cell.

5.13 Numerical comparison between the proper welding sequence to reduce distortion and a special welding
sequence in the 3-dimensional unitary cell

The proper welding sequence to reduce distortion (Figure 5.4 (c), page 100) recommends starting with the
weld beads on sphere 4 (Figure 5.3, page 99), located in the lateral edges of the outside of the structure. Next,


108
the weld beads on sphere 3 are applied. These weld beads are located at the lateral edges, on the inside of the
structure. Next, the weld beads located at the center box, on the outside of the structure are applied (sphere
2). Finally, the weld beads on sphere 1, located in the center box, on the inside of the structure, are applied.
The sequence involves applying alternating external and internal weld beads. For this reason, a special welding
sequence (figure 5.10) that involves applying first the external and then internal weld beads is proposed. This
welding sequence is commonly used to reduce the distortion in this type of joints. Weld beads belonging to
the same sphere are applied in the same way in both welding sequences.

Figure 5.10 Configuration of the special welding sequence

Table 5.3 shows the comparison of the numerical results between both welding sequences. It can be seen
that the proper welding sequence results in the lowest distortion, verifying the original hypothesis.

Welding sequence Maximum distortion (mm) Difference (%)
Proper welding sequence
to reduce distortion
2,04
6.4
Special welding sequence 2.18

Table 5.3 Comparison of numerical results between both welding sequences in the
3-dimensional unitary cell


109

5.14 Conclusions of the welding sequence analysis of the 3-dimensional unitary cell

Numerical simulations using 24 weld beads reduce distortion by up to 31% and residual stress by up to
18% with respect to numerical simulations using 8 weld beads.
Numerical simulations of the proper welding sequence to reduce residual stress confirm the hypothesis
in Section 5.2 by delivering minimum stress.
Numerical simulations of the welding sequence to reduce distortion also confirm the hypothesis
proposed in Section 5.2 by producing minimum distortion
It has been demonstrated that the hypothesis to determine the proper welding sequences for flat
symmetrical structures is also valid for symmetrical 3-dimensional structures.


110

CHAPTER VI

METHODOLOGY OF
EXPERIMENTAL TESTS

6.1 Introduction

This chapter presents a methodology for helping plan, execute, and control each of the stages of the
experimental tests. The methodology starts with the material selection of the specimen, configuration
selection, welding process selection, metal transfer mode selection, welding parameters selection, design and
fabrication of the equipment needed to run the test, design and fabrication of the mounting locks, residual
stresses relief caused by the manufacturing process, transportation, handling, storage and cutting of the
plates, measurement of the initial distortion of the plates, design and fabrication of a holder-mounting device
to hold the plates, design and fabrication of a square-mounting device to square the holder-mounting device,
application of welding tacks, measurement of the distortion after applying the welding tacks, installation of
the run-off tabs, welding, removal of the run-off tabs of the welded structure, measurement of the distortion
after welding, and measurement of the final distortion induced by the welding process.

6.2 Specimen material selection

The specimen material is of great importance because it determines the selection of the welding process,
parameters and material [35]. For this experiment, structural steel S235JR is selected. This steel is widely used
in Europe (Poland: steel ST3S) for construction of ships, buildings, bridges and machinery [36]. This steel is easy
to weld because of its carbon content (0.22% C), and its chemical composition, mechanical and thermal


111
properties are similar to ASTM A36 structural steel. Steel S235JR is therefore modeled using properties of
ASTM A36 structural steel.

6.3 Selection of specimen configuration

The configuration of the specimen determines the joint geometries, mounting configuration and the
devices needed to adapt the components of the welding equipment. For this experiment a "three-dimensional
unitary cell" configuration is selected because of its common applicability in industry for construction of
panels, ships and bridges. The three-dimensional unitary cell consists of two horizontal and two vertical plates
joined by fillet welds (figure 5.2, page 98). The thickness of the plates is 3 mm.

6.4 Selection of the welding process

The appropriate selection of the welding process will produce an efficient application of the welding in the
specimen. To produce the correct application, it is necessary to control the welding parameters, direction, and
position of the welding torch. This implies that the selected welding process has to be semiautomatic or
automatic. For this experiment, the Gas Metal Arc Welding (GMAW) process is selected because it can be
automated. Furthermore, the welding laboratory of the Machine Strength and Manufacturing Department at
the University of Sciences and Technology AGH, where the experiment took place, has a semiautomatic,
OPTYMAG 501, welding machine (Figure 6.1, page 112). This machine is equipped with a microprocessor-
controlled electrode wire feeder PDE-5, to fully adjust all of the welding process parameters (continuously
indicated in a multifunctional LCD display). It also has an MG-2 gas mixer and a RTU-10-300 rotameter with a
computerized system for process monitoring. The OPTYMAG 501 welding machine enables welding processes
with the following methods: MIG/MAG standard, MIG/MAG pulse/synergic and manual mode and a MMA
method for welding processes with coated electrodes.

6.5 Selection of metal transfer mode

The operating features of the welding process are largely determined by the form in which the weld metal
transfer takes place from the electrode to the weld pool [37]. The main metal transfer modes are: short
circuiting transfer, globular transfer, spray transfer, and pulsed transfer. For this experiment we select short
circuiting transfer for low current operation with small electrode diameters leading to a small, fast-cooling
weld pool that is generally well suited for joining thin sections. In short circuiting transfer mode, also called
dip transfer, the molten metal formed on the tip of the electrode wire is transferred by the wire dipping into
the molten weld pool, thus causing a momentary short circuit. Metal is therefore transferred only when the
electrode tip is in contact with the weld pool, and no metal is transferred across the arc gap.


112

Figure 6.1 Semiautomatic welding machine OPTYMAG 501

6.6 Selection of welding parameters (Operating variables)

The weld geometry, depth of penetration, and overall quality of the weld process depends on the welding
parameters [38]. These parameters are determined from the welding process, metal transfer mode, base
metal, metal thickness and joint geometries. These parameters are: arc voltage, welding current, welding
speed, wire feed rate, contact tip to work distance, electrode orientation, electrode diameter, shielding gas
composition, and gas flow rate.

6.6.1 Arc voltage and welding current

The arc voltage and welding current depend mainly on the metal transfer mode and the diameter of the
electrode. According to Rao [39], the ranges of voltage and current recommended for different electrode


113
diameters with short circuit transfer mode for carbon and alloy steels are shown in table 6.1. In this
experiment, a voltage of 22V and a current of 160A are selected.

Electrode
diameter
(mm)
Carbon steel and low alloy steels
Short circuit transfer mode
Voltage (V) Current (A)
0.8 15-21 70-130
0.9 16-22 80-190
1.2 17-22 100-225
1.6 ---- ----
2.4 ---- ----
Table 6.1 Electric settings for short circuit transfer mode [39]

6.6.2 Welding speed

The welding speed determines the quantity of deposition of filler metal [40]. The welding speed depends
mainly on the cross section of the weld bead (6.3 mm
2
), and the wire feed volume (31.80 mm
3

Figure 6.2 Typical regions of good short circuit welding conditions [37]
/s). For this
experiment, the calculated welding speed is 5 mm/s.

6.6.3 Wire feed rate

Several combinations of open circuit voltages and wire feed rates can produce short circuiting transfer of
metal. However, an optimum combination of arc voltage and wire feed rate will produce the best result. Figure
6.2 shows arc voltage vs. wire feed rate [37]. The region of good short circuiting or dip transfer arc is about 2V
wide. This graph shows that for a voltage of 22 V, there is a corresponding wire feed rate of 50 mm/s.


114
6.6.4 Selecting of contact tip to work distance

The contact tip to work distance (CTWD, Figure 6.3) determines the arc efficiency [40]. According to the
Lincoln Electric Company, for short circuiting metal transfer mode, the CTWD should be held between 10 and
12 mm. For this experimental investigation, the selected CTWD is 11 mm.

Figure 6.3 Contact tip to work distance [40]

6.6.5 Electrode orientation

The electrode orientation with respect to the weld joint affects the weld bead shape as well as the weld
penetration. According to Mandal [37], the electrode orientation is defined in two ways:

By the electrode axis orientation with respect to the direction of travel. This is known as the travel
angle. The three possible ways to positioning the welding gun (figure 6.4), are called forehand,
perpendicular and backhand. In this experimental investigation, the perpendicular positioning of the
electrode gun with respect to the base metal plate is selected.

Figure 6.4. Positioning of electrode gun with respect to the base metal plate [37]


115

The angle between the electrode axis and the adjacent work surface. This is called work angle. For fillet
welds in the horizontal position, the electrode should be positioned at about 45 to the vertical plane
(figure 6.5).

Figure 6.5 Normal work angle for fillet welds [37]

6.6.6 Electrode diameter

The appropriate diameter of the electrode depends on the base metal composition, metal thickness, and
the metal transfer mode [38]. In this experiment, the base material is carbon steel with a thickness of 3mm,
and the metal transfer mode is by short circuit. Therefore, according to the AWS, the electrode diameter
should be 0,9 mm.

6.6.7 Shielding gas composition

The primary function of the shielding gas is to protect the arc and the molten weld pool from the
atmospheric oxygen and nitrogen. The selection of the shielding gas composition depends on the base
material and the metal transfer mode [39]. According to the Lincoln Electric Company, there are two common
blends of shielding gases for short circuit transfer mode for carbon steels (Table 6.2). The blend 80% Argon +
20%CO
2
Common short circuiting transfer shielding gas blends
is selected for this investigation.

75% Argon + 25%CO
2
- reduces spatter and improves weld bead appearance in
carbon steel applications.
80% Argon + 20%CO
2
- Another popular blend, which further reduces spatter
and enhances weld bead appearance in carbon steel applications.

Table 6.2 Common blends of shielding gas composition for short transfer mode [40]


116
6.6.8 Gas flow rate

Gas flow rate eliminates and prevents defects in the weld bead [40]. According to the Lincoln Electric
Company, the rate flow for short circuiting transfer mode with either CO
2
Must be a rigid device.
or a mixed shielding gas is usually
12-17 L/min. For this experiment, 12 L/min is selected.

6.7 Design and fabrication of the equipment needed to run the test

Due to the configuration of the three-dimensional unitary cell with fillet welds, it is necessary to design and
fabricate an arm-holder and a welding-torch-holder to efficiently apply the welding fillet. Figure 6.6 depicts the
configuration of the arm-holder mounted in the main-support. See Appendix 9 for the construction drawings
of the arm-holder.

The requirements for the arm-holder device are:

Should have a positioning system (two positions) for the welding torch holder.
Should be easy to mount on the main-support.

Figure 6.6 Arm-holder device


117

The requirements for the welding-torch-holder device are:

Must have a rigid arm.
Must have a system to rotate the welding torch from one position to another 90 apart.
Must have a system to perform a zig-zag rotary motion for the welding torch.
Should be easy to mount and remove on the arm-holder.

Figure 6.7 depicts the configuration of the welding torch holder device. See Appendix 9 for the construction
drawings of the welding torch holder device.

Figure 6.7 Welding torch holder device

6.8 Design and fabrication of the mounting locks

The mounting locks position the specimen with respect to a reference point. The locks constrain the
specimen in the X, Y and Z-directions. Figure 6.8 depicts the localization of the mounting locks on the traveler
carriage.


118

Figure 6.8 Localization of the locks mounted on the traveler carriage

6.9 Relief of residual stresses caused by the manufacturing process, transportation, handling, storage and
cutting of the plates.

The manufacturing process results in residual stresses and distortion. For example, in the rolled plates,
tensile residual stresses are developed near the edges, while compressive residual stresses are developed in
the interior of the plates. In addition, transportation, handling and storage of the plates can alter the state of
initial residual stresses and distortion. The cutting processes also induce residual stresses near the edges of the
plates.

To study the effects of the initial residual stresses and distortion due to the manufacturing process,
transportation, handling, storage and cutting of the plates, four plate specimens are subject to a heat
treatment for stress relief. The main objective is to compare the effect of applying or not applying a heat
treatment for stresses relief. The distortion is the parameter of comparison. The duration of the treatment and
the temperature of the electric oven depend on the geometric characteristics and material properties of the


119
plates. For the present study, the heating time is 5 Hrs at 250 C. Figure 6.9 depicts the electric oven used to
relief the initial residual stresses.

Figure 6.9 Electric oven used to initial residual stresses relief

6.10 Measurement of the initial plate distortion

It is important to measure the initial distortion of the plates to avoid errors in the measurements of
distortion after welding, because, in many cases, the distortion due to the manufacturing process,
transportation, handling, storage and cutting of the plates is larger than the distortion induced by the welding
process. A series of standard gages are used to measure the initial distortion on a polished table. These
standard gages have a minimum precision of 0.05 mm. Figure 6.10 depicts the measurement of the initial
distortion on a polished table with standard gages.

Figure 6.10 Measurement of the initial distortion with standard gages


120
6.11 Design and fabrication of a holder-mounting device to hold the plates

To assemble and hold together the plates with the specimen configuration before applying the welding
tacks to the specimen, a holder-mounting device is required (figure 6.11 and Appendix 9).

The requirements for the holder-mounting device are:

Must be rigid.
Must be a precision device.
Must have a system that reduces the work-volume to remove the specimen after the application of the
welding tacks.

Figure 6.11 Holder-mounting device

6.12 Design and fabrication of a square-mounting device

Because the holder-mounting device will be removed of each specimen after applying the welding tacks, a
device to square the holder-mounting device is required (Figure 6.12). The construction drawings of the
square-mounting device are in Appendix 9.


121

The requirements for the square-mounting device are:

Must be rigid.
Must be a precision device.
Must have a system that increases the work-volume to remove the holder-mounting device.

Figure 6.12 Square-mounting device

6.13 Application of the welding tacks

The welding tacks are used to hold the specimen together before welding. Before applying the welding
tacks, C-type clamps are used to temporarily hold the plates together to the holder-mounting device. Once all
the welding tacks are applied, it is necessary to wait 15 minutes to allow solidification before removing the C-
type clamps. Figure 6.13 depicts the C-type clamps mounted on the specimen and figure 6.14 depicts the
application of the welding tacks on the specimen.


122

Figure 6.13 C-type clamps mounted on the specimen

Figure 6.14 Welding tack application on the specimen


123

6.14 Distortion measurement after welding tack application

Initial distortion can considerably affect the value of the final distortion induced by the welding process. It
is therefore necessary to measure distortion after welding tack application to consider the initial distortion of
the specimen. In this experiment, 24 points of interest have been selected to measure the initial distortion
(figure 5.7, page 104). For some of the points, the measurement is performed with standard gages and a
polished rule. For the remaining points, distortion is measured by tracing the contour of the four faces of the
welded structure (front, back, top and bottom) using a pencil with a sharp tip on a special paper sheet fixed on
a polished table. The sheets with the contours are scanned and save in a BMP format. The BMP files are
converted to a DWG format to be read by a CAD program. Once the file is opened, measurement of the
distortion can be performed by using CAD program tools. The measurements have a dimensional tolerance of
0,2 mm.

6.15 Installation of the run-off tabs

In some applications, it is necessary to completely fill out the groove right to every end of the joint. In such
cases, run-off tabs are used. They, in effect, extend the groove beyond the ends of the members to be welded.
The weld is carried over in tabs. This ensures that the entire length of the joint is filled to the necessary depth
with weld metal. Run-off tabs are excellent appendages on which to start and stop welding. Figure 6.15 depicts
the run-off tabs installed on the specimen.

Figure 6.15 Installation of the run-off tabs on the specimen


124

6.16 Application of the welding

The application of the welding is made by following a sequence. The selection of the welding sequence
depends on the critical parameter or parameters to reduce residual stress, distortion or a relation between
both. Chapters 4 and 5 described the proper welding sequences to reduce residual stress, distortion or a
relation between both for symmetrical structures in 2 and 3 dimensions. For this experiment, we selected the
welding sequence to reduce residual stress and distortion for symmetrical structures in the space. Figure 6.16
depicts the specimen welding.

Figure 6.16 Application of the welding to the specimen

6.17 Removing the run-off tabs from the welded structure

The run-off tabs should be removed to be able to perform the distortion measurements. These run-off tabs
are removed once the welding process is completed. Care must be taken when removing them to not induce
residual stresses, or distortion, or both. First the run-off tab material close to the specimen is removed. This
can be performed with an abrasive disc. Then, each of the run-off tabs is removed, and at the end, the
irregular edges are taken out with a grinder. Figure 6.17 depicts the removal of the run-off tabs from the
specimen.


125

Figure 6.17 Removing the run-off tabs from the specimen

6.18 Measurement of the distortion after applying the welding

The same procedure used in the measurement of the distortion after application of the welding tacks is
used to measure the distortion after application of the welding. The measurement of the distortion should be
performed 24 hrs, 48 hrs, 96 hrs and 384 hrs after the welding process to detect possible rheological effects.
Figure 6.18 depicts the distortion measurement after the welding process using standard gages.

Figure 6.18 Measurement of the distortion after welding using standard gages


126

6.19 Measurement of the final distortion

The final distortion, or the distortion induced by the welding process, is obtained by adding or subtracting
the distortion after applying the welding tacks, and the distortion after welding.

Final distortion = Distortion after applying Distortion after applying (41)
the welding tacks welding


127

CHAPTER VII

RESULTS OF THE EXPERIMENTAL TESTS IN
3-DIMENSIONAL UNITARY CELL-TYPE
SPECIMENS

7.1 Introduction

This chapter covers the results of the experimental tests performed in 3-dimensional unitary cell
specimens. In these experimental tests, the effects of the welding sequence on distortion are studied. Eight
symmetrical specimens are prepared. Four welding sequences are considered: two of them are appropriate to
reduce distortion and the others are appropriate to reduce residual stresses. Also, the effects that occur when
a welding bead is divided into 3 sub-weld beads are studied, as well as the effects of relieving the residual
stresses caused by manufacturing process, transportation, storage and cutting of plates. Welding tacks are
applied to all specimens before the actual weld process begins. The measurement of the distortion is
periodically performed to observe if rheological effects occur in the specimens after welding. The experimental
tests were performed in the Department of Machine Strength and Manufacturing in the Faculty of Mechanical
Engineering at the University of Science and Technology in Krakow, Poland.

7.2 Configuration of the 3-dimensional unitary cell specimens

The 3-dimensional unitary cell specimens have the same configuration analyzed in Chapter 5. The main
objective of using the same configuration is performing a quantitative and qualitative comparison between


128
numerical and experimental results. The 3-dimensional unitary cell specimen consists of two horizontal and
two vertical plates joined together by fillet welds (figure 5.2, page 98). The thickness of the plates is 3 mm.

7.3 Localization of the points of interest in the 3-dimensional unitary cell specimens

The points of interest in the 3-dimensional unitary cell specimens are the same points selected in Chapter
5 (figure 5.7, page 104) to perform a quantitative and qualitative comparison between the numerical and
experimental results.

7.4 Configuration of the experiment

Eight specimens and four welding sequences are studied: two of them are appropriate to reduce distortion,
and the other two are appropriate to reduce residual stress (Table 7.1). Other parameters considered in these
experiments are: the number of weld beads (8 or 24) and whether a relief of the residual stresses induced by
manufacture, transportation, handling, storage and cutting is performed. Welding tacks are applied to all
specimens before welding. The specimen number represents the order in which they are fabricated.
Specimens T1, T2 and T3 were discarded due to measurement errors.

Specimen Number of
the Weld
beads
Welding sequence most
appropriate to reduce:
Residual stress relief Welding
tacks
T7 8 Distortion No Yes
T11 8 Distortion Yes Yes
T8 8 Residual stress No Yes
T4 8 Residual stress Yes Yes
T9 24 Distortion No Yes
T5 24 Distortion Yes Yes
T10 24 Residual stress No Yes
T6 24 Residual stress Yes Yes

Table 7.1 3-dimensional unitary cell specimen configuration

7.5 Distortion after applying welding tacks in the 3-dimensional unitary cell specimens

Distortion is measured after application of the welding tacks and it represents the initial specimen distorted
shape. Table 7.2 lists the values of the distortion after application of the welding tacks for the different
specimens.


129
Point
Specimen

T7 T11 T8 T4 T9 T5 T10 T6
1 0,15 0,55 0,25 0,10 0,40 0,20 0,20 0,30
2 0,15 0,20 0,25 0,30 0,10 0,05 0,20 0,20
3 0,25 0,05 0,2 0,15 0,15 0,20 0,10 0,10
4 0,20 0,20 0,05 0,35 0,15 0,10 0 0,05
5 0,95 0,60 0,25 0,93 0,5 0,55 0,55 0,75
6 0,65 0,38 0,60 1,15 0,43 0,95 0,35 0,93
7 0,93 0,60 0,63 1,43 0,80 0,60 0,80 0,45
8 0,92 0,63 0,28 1,38 0,15 0,58 0,45 0,80
9 (-) 1,10 (-) 1,00 (-) 0,40 (-) 1,70 (-) 1,15 (-) 0,45 (-) 0,30 (-) 0,70
10 (-) 0,15 (-) 0,25 0,10 0,10 0 (-) 0,20 0,25 0,25
11 (-) 0,10 (-) 0,20 (-) 0,10 (-) 1,25 (-) 0,70 (-) 0,50 (-) 0,15 (-) 0,20
12 (-) 0,20 (-) 0,35 0 (-) 0,95 (-) 0,30 (-) 0,55 (-) 0,60 (-) 0,30
13 (-) 0,05 (-) 0,15 (-) 0,05 0,05 0 0,20 0,05 0,20
14 (-) 0,40 (-) 0,45 (-) 1,10 (-) 1,10 (-) 0,55 (-) 1,20 0,30 (-) 1,25
15 (-) 0,85 (-) 0,65 (-) 0,45 (-) 1,45 (-) 0,10 (-) 0,25 0 (-) 0,55
16 (-) 0,10 0,35 0,45 0 (-) 0,15 (-) 0,20 0,35 0,15
17 (-) 0,50 0,30 (-) 0,35 (-) 0,95 (-) 0,50 (-) 0,25 (-) 0,60 (-) 0,20
18 (-) 0,20 (-) 0,10 (-) 0,20 (-) 1,50 (-) 0,60 (-) 0,90 (-) 0,30 (-) 0,65
19 0,25 0,05 0,15 0,10 0,25 (-) 0,10 0,50 0,35
20 (-) 0,65 (-) 1,00 (-) 0,70 (-) 1,80 (-) 0,80 (-) 1,05 (-) 1,10 (-) 0,80
21 (-) 0,50 (-) 0,25 (-) 0,05 (-) 0,05 (-) 0,15 (-) 0,40 0,15 0
22 (-) 0,45 (-) 0,30 (-) 0,05 (-) 0,10 (-) 0,30 (-) 0,35 (-) 0,10 (-) 0,10
23 (-) 0,20 (-) 0,35 0,05 (-) 0,10 (-) 0,20 (-) 0,25 0 0
24 (-) 0,40 (-) 0,35 0,15 0,05 (-) 0,20 (-) 0,25 (-) 0,15 0

Table 7.2 Distortion after application of the welding tacks for the
different 3-dimensional unitary cell specimens, mm.

7.6 Distortion after welding in the 3-dimensional unitary cell specimens

Distortion is measured after welding and it represents the final distorted specimen shape. It is necessary to
note that these measurements do not represent the distortion induced by the welding process because the
specimens were originally distorted by the application of the welding tacks.


130

The measurements were performed periodically (24, 48, 96 and 384 hours) after welding to observe
possible rheological effects. However, the distortion did not show any variation with time. Therefore, the
rheological effects are discarded in the experimental tests. Table 7.3 lists the values of the distortion after
welding for the different three dimensional unitary cell specimens 24 hours after welding.

Point
Specimen

T7

T11

T8

T4

T9

T5

T10

T6
1 0,05 0,45 0,30 0,05 0,35 0,25 0,15 0,25
2 0,30 0,35 0,25 0,30 0,10 0,15 0,25 0,15
3 0,15 0,15 0,25 0,10 0,20 0,15 0,05 0,05
4 0,10 0,35 0,05 0,35 0,05 0,30 0,05 0,10
5 0,80 1,30 1,65 1,0 1,05 1,0 0,50 0,93
6 1,28 1,5 1,25 0,70 1,15 0,75 1,30 0,90
7 0,87 1,45 1,20 0,50 0,83 0,93 0,85 1,30
8 0,93 1,35 1,55 0,55 1,50 0,95 1,15 1,05
9 0,30 0,45 1,30 0,10 0,35 1,10 0,85 0,50
10 0,15 0,10 0,25 0,30 0,40 0,10 0,40 0,40
11 1,40 1,30 1,60 0,55 0,75 1,0 0,95 1,20
12 1,10 1,30 1,90 0,90 1,10 0,80 0,85 0,95
13 0,05 0,25 0,05 0,15 0,25 0,45 0,20 0,35
14 0,95 1,0 0,70 0,70 0,95 0,40 0,80 0
15 0,45 1,0 1,30 0,45 1,40 1,0 1,10 0,65
16 0,25 0,05 0,55 0,10 0,10 0,20 0,45 0,25
17 0,95 2,0 1,30 0,85 1,0 1,20 0,45 1,10
18 1,30 1,40 1,50 0,35 1,0 0,70 0,70 0,65
19 0,45 0,30 0,25 0,25 0,45 0,10 0,60 0,45
20 0,70 0,60 1,10 0,05 0,65 0,35 0,05 0,40
21 0 0,20 0,20 0,20 0,35 0,30 0,40 0,30
22 0,10 0,20 0,25 0,20 0,25 0,25 0,20 0,20
23 0,30 0,10 0,30 0,20 0,40 0,35 0,30 0,30
24 0,10 0,10 0,35 0,35 0,35 0,35 0,20 0,30

Table 7.3 Distortion after welding for the different 3-dimensional
unitary cell specimens 24 hrs after welding, mm.


131
Figure 7.1 shows a comparison between the distortion after applying welding tacks and the distortion after
welding for specimen T9, illustrating the distorted shape before and after welding.

Figure 7.1 Distortion (exaggerated) after applying welding tacks and after
welding for specimen T9


132
7.7 Distortion modes of the 3-dimensional unitary specimens

All welded specimens present the same distortion mode. However, the magnitude of the distortion varies.
The lower and upper flanges rotate inward to the specimen, and the plates of the central box bend toward the
center of the specimen (Figure 7.2).

Figure 7.2 Distorted shape of the 3-dimensional unitary cell welded specimens


133

7.8 Final distortion of the 3-dimensional unitary cell specimens

The final distortion is calculated with equation (41) [page 126] and tables 7.2 (page 129) and 7.3 (page
130). Table 7.4 lists the final distortion in the 24 points of interest of the 3-dimensional unitary cell specimens.

Point
Specimen

T7

T11

T8

T4

T9

T5

T10

T6
1 0,10 0,10 0,05 0,05 0,05 0,05 0,05 0,05
2 0,15 0,15 0 0 0,2 0,2 0,05 0,05
3 0,10 0,10 0.05 0,05 0,05 0,05 0,05 0,05
4 0,10 0,15 0 0 0,2 0,2 0,05 0,05
5 1,75 1,90 1,90 1,93 1,55 1,55 1,90 1,68
6 1,93 1,88 1,85 1,85 1,58 1,70 1,65 1,83
7 1,80 2,05 1,83 1,93 1,63 1,53 1,65 1,75
8 1,85 1,98 1,83 1,93 1,65 1,53 1,60 1,85
9 1,40 1,45 1,70 1,80 1,50 1,55 1,15 1,20
10 0,30 0,35 0,15 0,20 0,40 0,30 0,15 0,15
11 1,50 1,50 1,70 1,80 1,45 1,50 1,10 1,40
12 1,30 1,65 1,90 1,85 1,40 1,35 1,45 1,25
13 0,10 0,10 0,10 0,10 0,25 0,25 0,15 0,15
14 1,35 1,45 1,80 1,80 1,50 1,60 1,10 1,25
15 1,30 1,65 1,75 1,90 1,50 1,25 1,10 1,20
16 0,35 0,40 0,10 0,10 0,25 0,40 0,10 0,10
17 1,45 1,70 1,65 1,80 1,50 1,45 1,05 1,30
18 1,50 1,50 1,70 1,85 1,60 1,60 1,00 1,30
19 0,20 0,25 0,10 0,15 0,20 0,20 0,10 0,10
20 1,35 1,60 1,80 1,85 1,45 1,40 1,15 1,20
21 0,50 0,45 0,25 0,25 0,50 0,70 0,25 0,30
22 0,55 0,50 0,30 0,30 0,55 0,60 0,30 0,30
23 0,50 0,45 0,25 0,30 0,60 0,60 0,30 0,30
24 0,50 0,45 0,20 0,30 0,55 0,60 0,35 0,30

Table 7.4 Final distortion in the 24 points of interest of the 3-dimensional
unitary cell specimens, mm.


134
Figure 7.3 depicts the comparison of the non-distorted and distorted shape of specimen T9, showing a
general agreement with distortions in Section 7.7.

Figure 7.3 Final distortion (exaggerated) for welded specimen T9


135

7.9 Final Remarks for distortion of the 3-dimensional unitary cell specimens

For the points located at the center of the plates (points 1, 2, 3 and 4):

There is symmetry with respect to the horizontal (points 1 and 3) and vertical (points 2 and 4) axes.
Welding with 24 beads results in higher distortion than welding with 8 beads.
When following a welding sequence from the outside to the inside of the structure, specimens with 8
weld beads develop higher distortion.
Little variation is observed between welded specimens with or without stress-relief.

For points 5, 6, 7 and 8 located at the central part of the lower and upper flanges:

Specimens with 8 weld beads result in higher distortion than specimens with 24 weld beads.
Welded specimens with stress-relieved plates develop a slight increase in distortion.
Specimens welded with 24 weld beads that use a sequence from the outside to the inside of the
structure developed less distortion than specimens using the opposite sequence. In specimens welded
with 8 weld beads no significant trend was identified.

For points 9, 11, 12, 14, 15, 17, 18 and 20 located at the edges of the lower and upper flanges:

There is no distortion symmetry in any axis.
Specimens with 8 weld beads show higher distortion than specimens with 24 beads.
Welded specimens with stress-relieved plates develop higher distortion than specimens without stress-
relief.
In 8 bead specimens, following a welding sequence from the outside to the inside of the structure
produces less distortion than using the converse sequence. However, 24 bead specimens using an
inside-to-outside sequence developed less distortion than specimens using the converse sequence.

For points 10, 13, 16 and 19 located at the central part of the lateral edges of the vertical plates:

Welding sequences starting from the inside to the outside of the structure developed less distortion
than specimens using the opposite welding sequence.
Distortion is slightly higher in the specimens welded with 24 beads.
Little difference is observed between specimens with and without stress-relieved plates.


136
For points 21, 22, 23 and 24 located at the central part of the lateral edges of the horizontal plates:

Specimens welded with 24 beads develop higher distortion than specimens with 8 weld beads.
Specimens without stress-relief develop higher distortion than specimens with stress-relievef.

7.10 Conclusions of the results of the experimental test in 3-dimensional unitary cell specimens

In all welded specimens, the maximum distortion occurs at the central part of the lower and upper
flanges (points 5, 6, 7 and 8).
The specimens with 8 weld beads show higher distortion than specimens with 24 weld beads.
Little difference is observed between specimens with or without stress-relieved plates.
Specimens with 24 beads following a welding sequence starting from the outside to the inside of the
structure, show the lowest distortion without taking into account the stress-relief in the plates.


137

CHAPTER VIII

COMPARISON BETWEEN THE
3- DIMENSIONAL UNITARY CELL-TYPE
STRUCTURES/SPECIMENS

8.1 Introduction

This chapter presents a comparison between numerical (Chapter 5) and experimental (Chapter 7) results
for the 3-dimensional unitary cell structure. The comparison discusses the distortion modes, and distortion in
the 24 points of interest. The end of this chapter presents the procedures to determine the proper welding
sequences to reduce the residual stresses, distortion, or a relation between them in symmetrical and
asymmetrical structures in 2 and 3 dimensions. These procedures are based on the demonstrated hypotheses
developed in Chapters 4 and 5.

8.2 Comparison of distortion modes

The distortion modes of all the numerical simulations (figure 5.9, page 106) agree with the experimental
results (figure 7.2, page 132). The lower and upper flanges rotate inward towards the specimen, and the
central area of the flange bends towards the specimen. The central areas of the vertical and horizontal plates
bend inward to the structure.

8.3 Comparison of distortion

Numerical distortion at the 24 points of interest (table 5.2, page 107) can be directly compared with the
experimental results (table 7.4, page 133). It is important to mention that the numerical simulations do not
consider relief of the initial residual stresses. For this reason, only four numerical simulations were developed
vs. eight experiments. Therefore, each of the simulations will be compared against two experiments: with or


138
without stress relief. Equation 42 is used to determine the percentage difference between numerical and
experimental results for each of the 24 points of interest:

(42)

Table 8.1 depicts the difference percentage between numerical and experimental results.

Point Specimen

T7

T11

T8

T4

T9

T5

T10

T6
1 20 20 0 0 -20 -20 20 20
2 26,6 26,6 --- --- -5 -5 0 0
3 20 20 -40 -40 -20 -20 20 20
4 -10 26,6 --- ---- -5 -5 0 0
5 -5,7 2,6 1,6 3,1 -8,4 -8,4 13,1 1,8
6 -3,6 -6.4 -8,1 -8,1 -10,7 -2,9 -12,2 -1,1
7 -1,1 11,2 0 5,2 -2,4 -9,1 0,6 6,3
8 -7.5 -0,5 -8.2 -2.6 -4,8 -13 -12,5 2,7
9 2,8 6,2 0 5,5 -1,3 1,9 2,6 6,6
10 -3,3 11,4 -20 10 12,5 -16,6 -26,6 -26,6
11 -2,6 -2,6 -12,9 -6,6 -5,5 -2 -2,7 19,3
12 -14,6 9,7 2,1 -0,5 -10,7 -13,3 23,4 11,2
13 20 20 30 30 20 20 -26,6 -26,6
14 14 -8,2 -8,3 -8,3 -16 -8,7 -12,7 0,8
15 -16,2 8,5 -7.4 1 0 -20 1,8 10
16 -11,4 2,5 0 0 -20 25 20 20
17 8,2 21,7 0 8,3 -2,6 -6,2 -7,6 13
18 -4 -4 -14,1 -4,8 -8,7 -8,7 -21 6,9
19 5 24 0 33 20 20 20 20
20 -8,8 8,1 -1,6 1,1 -7,6 -11,4 12,1 15,8
21 -14 -26.6 0 0 -32 5,7 -16 3,3
22 -7,2 -18 -13,3 -13,3 -7,2 1,6 -13,3 -13,3
23 -14 -26.6 -24 -3,3 0 0 -10 -10
24 -4 -15,5 -10 26,6 -23.6 -13,3 17,1 3,3

Table 8.1 Difference (%) between numerical and experimental results


139
Figure 8.1 shows a comparison between numerical (table 5.2-simulation 3, page 107) and experimental
distortion (table 7.4 -specimen T9, page 133). The figure shows good correlation between both results.

Figure 8.1 Comparison between numerical (simulation 3) and experimental
distortion (specimen T9). Exaggerated distortion.


140

8.4 Conclusions of the comparison between numerical and experimental results

Of the 192 points of interest, 17 points show a good correlation in distortion between numerical and
experimental results, 87 points present a difference between 0,1-10%, 61 points differ by 10,1-20%, 17
points are different by 20,1-30%, 2 points present a difference between 30,1-40%, and 4 points cannot
be compared. The resulting comparison is considered acceptable.
Welded specimens without stress-relief present smaller differences than specimens fabricated with
stress-relief, with the exception of specimen T10.
The maximum differences in distortion occur in the points located at the center of the horizontal and
vertical plates because the magnitude of the distortion in these points is very small. As an example,
variations as small as 0,01 mm result in large percentage differences.

8.5 Procedures to determine the proper welding sequences to reduce residual stress, distortion, or a relation
between them in symmetrical and asymmetrical structures in 2 and 3 dimensions

Based on the demonstrated hypotheses to determine the proper welding sequence to reduce the residual
stress, distortion, or a relationship between them in symmetrical structures in 2 (Chapter 4) and 3 dimensions
(Chapter 5), the following procedures are obtained to determine proper welding sequences. Hypotheses for
symmetrical structures are also adapted to asymmetrical structures. However, this hypothesis was not
demonstrated.

8.5.1 Symmetrical structures in 2 and 3 dimensions

To determine the proper welding sequence to reduce the residual stress, distortion, or a relation between
them in symmetrical structures in the plane/space, the procedure is as follows: Determine the axis of
symmetry and the center of gravity of the structure. Draw concentric circles/spheres centered on the center of
gravity of the structure and extending to the center of gravity of each of the weld beads. The weld beads
located at the same distance from the center of gravity of the structure fall on the same circle/sphere. The
circles/spheres are numbered from the smallest to the largest.

Procedure No.1- To reduce residual stress:

Start with the weld beads on the smallest circle/sphere and then continue with the weld beads on the
larger adjacent circle/sphere until the largest circle/sphere is reached. The weld beads on the same
circle/sphere should be symmetrical. First, the weld beads with diagonal symmetry are selected. If the weld
bead with diagonal symmetry has already been selected, then the farthest symmetrical weld bead is selected
next. When a current weld bead has more than one farthest symmetrical weld bead, the adjacent weld bead in


141
counter-clockwise direction is selected. Finally, to move from one circle/sphere to the other, the farthest weld
bead to the current weld bead is selected. If a current weld bead has more than one farthest weld bead, the
adjacent weld bead in counter-clockwise direction is selected. The converse welding method between weld
beads is adopted.

Procedure No.2- To reduce distortion:

Apply welding tacks to the structure before welding. First, the weld beads in the largest circle/sphere are
selected, followed by the weld beads on the adjacent smaller circle/sphere, until the smallest circle/sphere is
reached. The application of the weld beads on the same circle/sphere should be in symmetrical form. The weld
beads with diagonal symmetry are selected first. If the weld bead with diagonal symmetry has already been
selected, then the farthest symmetrical weld bead is selected next. When a current weld bead has more than
one farthest symmetrical weld bead, the adjacent weld bead in counter-clockwise direction is selected next.
Finally, to move from one circle/sphere to the other, the farthest weld bead to the current is selected. If a
current weld bead has more than one farthest weld bead, the adjacent weld bead in a clockwise direction is
selected. The converse welding method between weld beads is adopted.

Procedure No.3- To improve the residual stress-distortion relation while prioritizing the reduction in
residual stresses:

First, the weld beads on the smallest circle/sphere are selected and then continuing with the weld beads on
the larger adjacent circle/sphere until the largest circle/sphere is reached. The application of weld beads on
the same circle/sphere should be in an adjacent form in counter-clockwise direction. To move from one
circle/sphere to the other, the closest weld bead to the current is selected. The converse welding method
between weld beads is adopted.

Procedure No.4- To improve the residual stress-distortion relation while prioritizing the reduction of
distortion:

Welding tacks are required only on the circle/sphere next to the larger adjacent circle/sphere. The welding
sequence begins by selecting the weld beads on the smallest circle/sphere, and then continuing with the weld
beads on the circle/sphere next to the larger adjacent circle/sphere. Then, going back to the weld beads on the
smaller adjacent circle/sphere. The application of the weld beads on the same circle/sphere should be in a
symmetrical form. First, the weld beads with diagonal symmetry are applied. If the weld bead with diagonal
symmetrical has already been applied, then the farthest symmetrical weld bead is selected. When a current
weld bead has more than one farthest symmetrical weld bead, the adjacent weld bead in counter-clockwise
direction is selected. Finally, to move from one circle/sphere to the other, the farthest weld bead is selected


142
next. If a current weld bead has more than one farthest weld bead, the adjacent weld bead in counter-
clockwise direction is selected. The converse welding method between weld beads is adopted.

8.5.2 Asymmetrical structures in 2 and 3 dimensions

The procedure to determine the proper welding sequence to reduce the residual stress, distortion, or a
relation between them in asymmetrical structures is as follows: Find the center of gravity of the structure and
draw concentric circles/spheres centered on the center of gravity of the structure and extending to the center
of gravity of each of the weld beads. The weld beads that are located at the same distance from the center of
gravity of the structure will fall on the same circle/sphere. The circles/spheres are numbered from the smallest
to the largest.

Procedure No.5- To reduce the residual stress:

First, the weld beads on the smallest circle/sphere are selected and then continuing with the weld beads
on the adjacent larger circle/sphere until the largest circle/sphere. Next, the farthest weld bead to the current
weld bead on the same circle/sphere is selected. When a current weld bead has more than one farthest weld
bead in the same circle/sphere, the adjacent weld bead in counter-clockwise direction is selected. Finally, to
move from one circle/sphere to the other, the farthest weld bead to the current weld bead is selected. If a
current weld bead has more than one farthest weld bead, the adjacent weld bead in counter-clockwise
direction is selected. The converse welding method is adopted between weld beads.

Procedure No.6- To reduce the distortion:

Apply welding tacks to the structure. First, the weld beads belonging to the largest circle/sphere are
selected and then the weld beads belonging to the adjacent smaller circle/sphere, until the smallest
circle/sphere is reached. Next, the farthest weld bead to the current weld bead on the same circle/sphere is
selected. When a current weld bead has more than one farthest weld bead in the same circle/sphere, the
adjacent weld bead in counter-clockwise direction is selected. Finally, to move from one circle/sphere to the
other, the farthest weld bead to the current is selected. If a current weld bead has more than one farthest
weld bead, the adjacent weld bead in a clockwise direction is selected. The converse welding method is
adopted between weld beads.

residual stresses.

First, the weld beads on the smallest circle/sphere are selected, and then continuing with the weld beads
on the larger adjacent circle/sphere until the largest circle/sphere is reached. The application of the weld


143
beads on the same circle/sphere should be in an adjacent form in counter-clockwise direction. Finally, to move
from one circle/sphere to the other, the closest weld bead to the current weld bead is selected. The converse
welding method is adopted between weld beads.

distortion:

The application of welding tacks is required only on the circle/sphere next to the larger adjacent
circle/sphere. The welding sequence begins by selecting the weld beads on the smallest circle/sphere, and
then continuing with the weld beads on the circle/sphere next to the larger adjacent circle/sphere. Then, going
back to the weld beads on the smaller adjacent circle/sphere. Next, the farthest weld bead to the current weld
bead on the same circle/sphere is applied. When a current weld bead has more than one farthest weld bead,
the adjacent weld bead in counter-clockwise direction is selected next. Finally, to move from one circle/sphere
to the other, the farthest weld bead to the current weld bead is selected next. If a current weld bead has more
than one farthest weld bead, the adjacent weld bead in counter-clockwise direction is selected. The converse
welding method is adopted between weld beads.

Figure 8.2 depicts the flow diagram of the analysis of the welding sequence for symmetrical and
asymmetrical structures in 2 and 3 dimensions. The main goal of this flow diagram is to facilitate reading by a
welder, for determining the proper welding sequences to reduce the required critical parameter, and
additionally, to be used by the weld-cognizant engineer in the design stage to check the welding sequences
with a finite element code.


144
Figure8.2FlowDiagramoftheanalysisoftheweldingsequenceforsymmetricalandasymmetricalstructuresin2and3dimensions


145

CHAPTER IX

CONCLUSIONS, CONTRIBUTIONS,
AND SUGGESTIONS FOR
FUTURE RESEARCH

9.1 Conclusions

The following conclusions are limited to analyzed cases in this thesis:

The maximum residual stress occurs in the base metal, at the boundary with the welded metal.
Validation of the numerical simulation of the welding process vs. experimental results shows
satisfactory results, with less than 8.3% difference.
The welding sequence considerably affects the residual stresses and the distortion in welded structures.
The welding sequence that produces the lowest distortion also produces the highest residual stress and
vice versa.
The welding sequence has a bigger effect on distortion than on residual stresses.
The comparison between numerical results of the 3-dimensional unitary cell with experimental tests
shows satisfactory results, with less than 40% difference.
Dividing a weld bead in several sub-welds reduces distortion but increases residual stresses.
Combining the welding sequence and converse welding method considerably reduces distortion and
residual stresses.
To reduce distortion it is necessary to apply welding tacks to the structure before welding begins.
To reduce the residual stresses, welding tacks are not recommended.
Application of the weld beads from the inside to the outside of the structure, without the use of
welding tacks results in the lowest residual stress, but the highest distortion.
Application of the weld beads from the outside to the inside of the structure, with the use of welding
tacks, results in the lowest distortion, but in the highest residual stresses.
Application of the weld beads from the inside to the outside of the structure with the use of welding
tacks reduces distortion, but increases residual stress.


146
Application of the weld beads from the outside to the inside of the structure without the use of welding
tacks, reduces residual stresses, but increases distortion.
The hypothesis to determine the proper welding sequences for 2-dimensional symmetrical structures is
valid for 3-dimensional symmetrical structures.

9.2 Contributions

A finite element-based simulation of the welding process. A sequentially coupled thermo-mechanical
analysis is proposed. A transient heat transfer analysis is first performed, followed by a thermo elasto-
plastic analysis. The material is assumed to follow the von Mises yield criterion and the associated flow
rule, lineal isotropic hardening, and the temperature-dependent thermal and mechanical properties of
the selected material. The micro-structural evolution is not considered
A comparison of the numerical simulation of the welding process to experimental tests.
A study of the effects of the welding sequence on the residual stress and distortion on particular
symmetrical structures in 2 and 3 dimensions.
A residual stress-distortion relation analysis due to the welding sequence for 2-dimensional
symmetrical structures.
An equation to determine the residual stress-distortion relation:

A hypothesis to determine the proper welding sequence to reduce the residual stress, distortion or a
relation between them for 2-dimensional symmetrical structures (Chapter 4, Section 4.9).
A hypothesis to determine the proper welding sequence to reduce the residual stress, distortion or a
relation between them, for 3-dimensional symmetrical structures (Chapter 5, Section 5.2).
Validation of the hypotheses to determine the proper welding sequence to reduce the residual stress,
distortion or a relation between them, for symmetrical structures in 2 and 3 dimensions.
Procedures to select the proper welding sequence to reduce the residual stress, distortion, or a relation
between them for symmetrical structures in 2 and 3 dimensions.
Development of a methodology to perform experimental tests.
Experimental comparison of the proper welding sequences to reduce the distortion and/or the residual
stress in a 3-dimensional symmetrical structure.

9.3 Suggestion for future research

Include the effects of micro-structural evolution in the numerical simulation of the welding process.
Measurement of the residual stress in the 3-dimensional unitary cell specimens.


147
Application of this methodology to determine the proper welding sequence to reduce the residual
stress, distortion or a relation between them, for other common structures in the welding industry.
Appendix 4 proposes some welding sequences to reduce the residual stresses and distortion in
common symmetrical structures. These welding sequences have not been numerically verified.
Develop a family-group classification of the symmetrical structures in 2 and 3 dimensions to determine
the proper welding sequence to reduce the residual stress, distortion or a relation between them.
Develop a computer program to automatically determine the proper welding sequences to reduce
residual stress, distortion or a relation between them, for symmetrical structures in 2 and 3 dimensions.
Develop a methodology to determine the proper welding sequence to reduce the residual stress,
distortion or a relation between them for asymmetrical structures.


148

REFERENCES

[1] Zhili, F., 2005, Processes and Mechanisms of Welding Residual Stress and Distortion, CRC Press.
[2] Masubuchi, K., 1980, Analysis of Welded Structures: Residual Stresses, Distortion, and their
Consequences, Pergamon Press.
[3] Masubuchi, K., 2001, Welding Handbook: Residual Stress and Distortion, American Welding Society,
9
th
[4]
edition.
Messler, R., 1999, Principles of Welding: Processes, Physics, Chemistry, and Metallurgy, John Wiley
and Sons, Inc.
[5] Murugan, V., Gunaraj, V., 2005, Effects of Process parameters on angular distortion of gas metal arc
welded structural steel plates, Weld. J. 84 (11): 165s-171s.
[6] Tso-Liang Teng., Ching-Cheng Lin., 1998, Effect of welding condition on residual stress due to butt
welds, Int. J. Pressure Vessels and Piping, 75: 857-864.
[7] Jang, B., Kim, H. K., Kang S., 2001, Effects of root opening on mechanical properties, deformation and
residual stress of weldments, Weld. J. 80 (7): 80s-88s.
[8] Bhide, S. R., Michaleris, P., Posada, M., Deloach, J., 2006, Comparison of buckling distortion
propensity for SAW, GMAW, y FSW, Weld. J. 85 (9): 189s-195s.
[9] Wahab, M. A., Alam, M. S., Painter, M. J., Stafford, P. E., 2006, Experimental and numerical simulation
of restraining forces in gas metal arc welded joints, Weld. J. 85 (2): 35s-43s.
[10] Muhammad, Abid., Muhammad, Siddique., 2005, Numerical simulation of the effect of constraints on
welding deformations and residual stresses in a pipe-flange joint, Modeling Simul. Mater. Sci. Eng. 13:
919-933.
[11] Michaleris, P., Sun, X., 1997, Finite Element Analysis of Thermal Tensioning Techniques Mitigating
Weld Buckling Distortion, Weld. J. 76(11): 451s-457s.
[12] Michaleris, P., Dantzig, J., Tortorelli, D., 1999, Minimization of welding residual stress and distortion in
large structures Weld. J. 78 (11): 361-366.
[13] Holdren, R. L., Greer, J. E., 2001, Welding Handbook: Terms and definition, Appendix A., American
Welding Society, 9
th
[14]
edition,
The Lincoln Electric Company, 1933, Procedure handbook of arc welding design and practice,
Cleveland, Ohio, U.S.A.
[15] Bart, O. N., Deepak, G., Kyoung, Y. K., 2004, Welding distortion minimization for an aluminum alloy
extruded beam structure using a 2D model, ASME J. Manufacturing Science and Engineering, 126: 52-
63.
[16] Tso, L. T., Peng, H. C., Wen, C. T., 2003, Effect of welding sequence on residual stresses, Computer
and Structures, 81: 273-286.
[17] Ji, S. D., Fang, H. Y., Liu, X. S., Meng, Q. G., 2005, Influence of a welding sequence on the welding
residual stress of a thick plate, Modeling and Simulation in Materials Science and Engineering, 13:
553-565.


149

[18] Nami, M. R., Kadivar, M. H., Jafarpur, K., 2004, 3D thermo-viscoplastic modeling of welds: Effect of
piece-wise welding on thermo-mechanical response of thick plate weldments, Iranian Journal of
Science and Technology, Transaction B, Vol. 28, No. B4, 467-478.
[19] Mochizuki, M., Hayashi, M., 2000, Residual stress distribution depending on welding sequence in multi-
pass welded joints with X-shaped groove, ASME J. Pressure Vessel Technology. 122: 27-32.
[20] Mochizuki, M., Hattori, T., Nakakado, K., 2000 , Residual stress reduction and fatigue strength
improvement by controlling welding pass sequence, ASME J. Engineering Materials and Technology,
122: 108-112.
[21] Tsai, C. L., Park, S. C., Cheng, W. T., 1999, Welding distortion of a thin-plate panel structure, Weld. J.
78 (11): 156s-165s.
[22] Hackmair, C., Werner, E., Pnisch, M., 2003, Application of welding simulation for chassis components
within the development of manufacturing methods, Computational materials Science, 28: 540-547.
[23] Kadivar, M. H., Jafarpur, K., Baradaran, G. H., 2000, Optimizing welding sequence with genetic
algorithm, Computational Mechanics. 26: 514-519.
[24] Voutchkov, I., Keane, A. J., Bhaskar, A., Olsen, T. M., 2005, Weld sequence optimization: the use of
surrogate models for solving sequential combinatorial problems, Computer methods in applied
mechanics and engineering, 194: 3535-3551.
[25] Hernndez, A. I., 2005, Anlisis de distorsin y esfuerzos residuales en estructuras soldadas, Tesis de
maestra, Universidad de Guanajuato.
[26] Karlsson, L., 1986, Thermal Stresses I-Chapter 5: Thermal Stresses in welding, Edited by Hetnarski R.
B., Elsevier Science Publishers B.V.
[27] Argyris, J. H., Szimmat, J., William, K. J., 1982, Computational aspect of welding stress analysis,
Computer methods in applied mechanics and engineering, 33: 635-666.
[28] Boley, B., Weiner, J., 1962, Theory of thermal stresses, John Wiley and Sons, Second Edition.
[29] Goldak, J.A., Akhlaghi Mehdi., 2005, Computational Welding Mechanics, Springer.
[30] Cook, R., Malkus, D., Plesha, M., Witt, R., 2002, Concepts and Application of Finite Element Analysis,
John Wiley and Sons, Fourth Edition.
[31] Doghri, I., 2000, Mechanics of Deformable Solid: Linear and Nonlinear, Analytical and Computational
Aspects, Springer.
[32] Malvern, L., 1969, Introduction to the Mechanics of a Continuous Medium, Prentice Hall, Inc.
[33] William, D., Callister, Jr., 2000, Materials Science and Engineering an Introduction, John Wiley and
Sons, inc. Fifth Edition.
[34] Chang, P. H., 2003, Numerical and experimental investigation on residual stresses of the butt-welded
joints, Computation Material Science, 29: 511-522.
[35] Tso-Liang Teng., Ching-Cheng Lin., Peng-Hsiang Chang., 2002, Effect of weld geometry and residual
stresses on fatigue in butt-welded joints, Int. J. Pressure Vessels and Piping, 79: 467-482.
[36] Pilarczyk, J., 2003, Poradnik inyniera Spawalnictwo cz 1, Wydawnictwa Naukowo-Techniczne,
Polska. (In polish)
[37]

Mandal, N. R., 2004, Welding and distortion control, Alpha Science International Ltd.
[38] Holdren, R. L., Greer, J. E., 2001, Welding Handbook: Volume 2., American Welding Society, 9
th
[39]

edition,
Rao, P. N., 2001, Manufacturing Technology: Foundry, Forming and Welding, McGraw-Hill
International, Second Edition.


150
[40] Nadzam, J., 2006, Gas Metal Arc Welding Guidelines, The Lincoln Electric Company, Cleveland, Ohio,
U.S.A.
[41] Suzalec, A., 2005, Theory of Thermomechanical Processes in Welding, Springer.
[42] Prat, P., Gens, A., 2000, Leyes de comportamiento de materiales Mtodos Numricos para Clculo y
Diseo en Ingeniera, Barcelona.
[43] Simo, J., Hugnes T., 1998, Computational Inelasticity, Interdisciplinary Applied Mathematics, Vol. 7,
Springer-Verlag, New York, Inc.
[44] Makhnenko, V. I., Veliokoivanenko, E. A., 1999, Numerical Methods for the Predictions of welding
stresses and distortions, E. O. Paton Electric Welding Institute, Volume 13, Part 1.
[45] Nicholson, D., 2003, Finite Element Analysis: Thermomechanics of Solid, CRC Press.
[46] Klaus-Jurgen Bathe, 1996, Finite Element Procedures, Prentice Hall.
[47] Swanson Analysis Systems, Inc. ANSYS users manual: Elements, Vol. III.


151

APPENDIX 1


152
Plasticity theory applied to welding process and its formulation by finite element method

1. Plasticity theory applied to welding process

Plasticity-based analysis provides an engineering basis for study of residual stress and distortion [1]. The
temperature of the metal in welds varies from melting temperature ( )
m
T to room temperature ( )
amb
T . The
domain includes the liquid weld pool and the far field of solid near room temperature. The materials behavior
depends on the temperature range encountered in the different regions in the weld. At temperatures above
0.8
m
T , the solid can be considered to be linear viscous. Between 0.5 and 0.8
m
T , the material is rate
dependent. For 0.5
m
T , the solid can be considered to be viscoplastic and characterized by an elasticity tensor,
viscosity and deformation resistance. Below 0.5
m
T , the solid can be considered to be rate independent plastic
material characterized by an elasticity tensor, yield strength and isotropic hardening modulus [29].

Most thermal stress analyses have used thermal elastic plastic constitutive models with rate independent
plasticity. Rate independent plasticity implies zero viscosity and therefore zero relaxation time. This means the
stress relaxes instantly to the yield stress. A rate independent model is certainly not valid in the liquid region
and is suspected near the melting point where the viscous effects are expected to be important [29].
Furthermore, the mechanical and thermal properties of weld and base metal are dependent on temperature
[1,29,41].

2. Mathematical considerations in the modeling of the welding process

Welding process simulation considers small displacements [2,41]. Therefore, the infinitesimal strain is
defined as:

( )
, ,
1
2
ij i j j i
u u = + (43)

The strain in equation (43) is known as the Cauchy strain tensor. Changes in stress caused by deformation
are also assumed to travel slowly compared to the speed of sound. So, at any instant, an observed group of
material particles is approximately in static equilibrium [1]:

,
0
ij j i
f + = (19)

Where:
i
f

is the sum of the body force and
ij
is the Cauchy stress tensor.

To ensure that the body remains continuous during the deformation, the compatibility equations must be
satisfied [2].

, , , ,
0
ij kl kl ij ik jl jl ik
+ = (20)


153
3. Rate independent isotropic plasticity in the modeling of the welding process

Rate independent plasticity occurs at low temperatures, roughly below 0.5
m
T . The deformation is due to
dislocation glide and strain rate due to fluctuations plays no significant role. The relaxation time is zero. In the
simulation of the welding process it is important to assume the appropriate material behavior of the weld and
base metal. The additive decomposition of strain, yield surface, flow rule and the hardening hypothesis should
be assumed in such a way to describe the welding process as realistic as possible [41].

3.1 Additive decomposition of total strain

This section shows a derivation of the constitutive equations in differential (increment) form for thermal
elastic plastic isotropic material, assuming that total strain increment can be decomposed (valid for small
strains and small rotations) as [1]:

Tot e p tr
ij ij ij ij ij
d d d d d
= + + + (44)

Where:
e
ij
d is the elastic strain increment,
p
ij
d is the plastic strain increment,
ij
d

is the thermal strain
increment, and
tr
ij
d is the phase transformation strain increment. The metallurgical evolution may develop
stresses due to strain incompatibility during solid-state phase transformation. This strain incompatibility
results in stresses between the grains of different phases, which may have a significant influence on the state
of residual stresses in the weld metal and heat affected zone in joining certain types of materials such as
martensitic weldments. However, this phase transformation stress is usually ineffective in influencing the final
state of weld distortion. Equation (44) is therefore reduced to:

e p
ij ij ij ij
d d d d

= + + (21)

Equation (21) represents the additive decomposition of total strain increment considered in this
investigation. Figure 1 depicts the decomposition of total strain into elastic, plastic and thermal.

Figure 1. Stress-strain curve of thermal elastic plastic material [42,43]


154
From figure 1, two fundamental relations are obtained:

( )
e p
ij ijkl kl ijkl kl kl kl
d E d E d d d

= = (45)

t ep
ij ijkl lk ijkl lk
d E d E d = = (46)

where:
ijkl
E is the elastic tensor,
t
ijkl
E is the tangent elastic tensor and
ep
ijkl
E is the elastic-plastic tensor.

If the elastic strain increment is zero
( )
0
ij
d

= , the material is known as rigid- perfectly plastic (figure
2(a)). If the strain increases with no limit when the stress reaches the yield point, the material is known as
elastic-perfectly plastic (figure 2 (b)). If the stress does not remain constant beyond the yield point, increasing
or decreasing with strain, the material is known elastic-plastic with hardening (figure 2(c)) or elastic-plastic
with softening (figure 2(d)). This investigation considers that materials behave as elastic plastic with hardening.

Figure 2. Stress-strain idealized curves in plasticity [42]. (a) Rigid-perfectly plastic, (b) elastic-perfectly plastic,
(c) elastic-plastic with hardening and (d) elastic-plastic with softening.

3.2 Yield surface

A yield criterion is a basic assumption for determining the onset of plastic strain. If a stress state at a point
satisfies the yield criterion, then this point deforms plastically; otherwise, it undergoes elastic strain [30].
Figure 3 depicts the yield surface for isotropic materials.


155
The general expression of the yield surface is written as [42]:

( )
, 0
ij i
f = (47)

where:
i
are parameters that control the yield surface size

Figure 3. Yield surface for isotropic materials [42]

Generalizing the different material behavior showed in figure 2, the yield surface becomes [42]:

Perfect plasticity: the yield surface only depends on the magnitudes of the stress, not changed in size
during the load process. In this case, equation (47) reduces to:

( )
0
ij
f = (48)

Rigid Plasticity: the yield surface expands during the load process.
Soften Plasticity: the yield surface contracts during the load process.

When the body is under plastic domain, the stress state should be on the yield surface. Thus, given a
defined stress state
ij
and parameters,
i
, yields [42]:

( )
, 0
ij i
f < for elastic deformation domain

( )
, 0
ij i
f = for plastic deformation domain

( )
, 0
ij i
f > impossible

For the general case of modeling of the welding process, the yield surface has the form [2]:

( ) ( )
, , , 0
p p
ij ij
f T = (49)

Where: is the parameter related with the strain hardening effects and T is the temperature.


156

3.3 Flow rule

The flow rule relates the stress state with components of the plastic strain increment [8]. In general there is
a function of the stress and other parameters that [42]:

( )
, 0
ij i
G = (50)

The term is known as plastic potential, so that plastic strain increment is obtained as [42]:

p
ij
ij
G
d d
(51)

Where d is a scalar that determines the magnitude of the plastic strain increment and is called plastic
multiplier. The plastic strain increment is parallel to the gradient of the plastic potential. Therefore, the
direction of the plastic strain is tangent to the surface of G (figure 4).

Figure 4. Plastic potential and direction of plastic strain increment [42]

When the yield surface and the plastic potential coincide
( ) ( )
ij ij
f G = , the plasticity is called associate
and no-associate plasticity for the oppositecase [42]. The associate plasticity is used in ductile metals while the
no-associate plasticity is more appropriate for granular materials [31]. For associate plasticity equation (51)
becomes:

p
ij
ij
f
d d
(52)

Applying the consistency condition to equation (49) yields:

0 df = (53)


157
0
p
ij ij p
ij ij
f f f f
d d d dT
T

+ + + =

(54)

The consistency condition means that if yield occurs during a time interval, the solution should always
remain on the yield surface [32].

Since
p
ij p
ij
d d
and the flow rule is

p
ij
ij
f
d d
, substituting these expressions into equation (54),

produces:

kl
kl
f f
d Q d dT
T

| |
= +
|

\ .
(55)

The plastic multiplier takes specific values to remain on the yield surface at all times:

0 d = if 0 f < (Elastic strain)
0 d > if 0 f = and 0 df = (Yielding)
0 d = if 0 f = and 0 df < (Unloading)

If 0 d > , the plastic strain increments are determined from the flow rule. If equation (55) is introduced
into equation (51), the plastic strain increment becomes:

p
ij kl
ij kl ij
f f f f
d Q d dT
T

| |

= +
|
|

\ .
(56)

Where;
1
p p
ij ij ij
Q
f f f

=
| |

+
|
|

\ .
(57)

3.4 Hardening hypothesis

The hardening hypothesis describes how the yield surface is modified by the plastic strain beyond the initial
yield [8]. Under this hypothesis, the yield surface is not constant but may change in size, shape and/or position
during the process of plastic strain [6].

There are three fundamental hardening hypotheses [30-33,42-43]:

a) Null hardening. The hardening energy is absent in an ideally plastic material (perfect plasticity). The yield
surface for this material is not changed by the presence of plastic strain as illustrated in figure 5 (a). If the
stress state remains on the yield surface in the path 1-2, then the evolution of the plastic strain occurs, the
path 2-3 corresponds to elastic unload.


158
b) Isotropic hardening. Assumes that the subsequent yield surface is a uniform expansion of the initial yield
surface, as shown in figure 5(b), and that the materials isotropic response to yielding remains unchanged
during plastic strain. Thus the center of initial and subsequent yield surface is the same. In isotropic hardening,
the variable depends on the plastic strain or plastic work.

For the case of strain hardening,
( )
p
F d =

where
p
d

is the equivalent strain increment defined as:

2
3
p p p
ij ij
d d d = (58)
For the case of work hardening,
( )
p
F dW =

where
p
dW is the work plastic increment defined as:

p p
ij ij
dW d = (59)

c) Kinematic hardening. Assumes that the yield surface moves as a rigid body in the stress space during
plastic strain. As a result the shape of the subsequent yield surface during plastic strain remains unchanged.
This hardening is more appropriate for cyclic plasticity, because isotropic hardening cannot predict the
Bauschingers effect which is observed experimentally [31].

For this investigation, isotropic strain hardening (Figure 5 (b)) is selected.

(a) (b) (c)

Figure 5. Three fundamental hypothesis of hardening [42]. (a) Null hardening,
(b) isotropic hardening, and (c) kinematic hardening.

4. Von Mises model applied to the welding process

Welding process simulation is based on two important considerations [1, 2, 26,27]: 1) Welding is for ductile
metals (associate plasticity) and 2) Deformations occur without volume change (independent of the spherical


159
invariant,
1
3
ij kk ij
p = ). Due to these considerations, the Von Mises model is adopted. Figure 6 depicts the
Von Mises yield surfaces for one, two and three dimensions.

(a) (b) (c)
Figure 6. Von Mises yield surface [42]. (a) One-dimensional,
(b) Two-dimensional and (c) Three-dimensional

For this investigation, the three-dimensional Von Mises yield surface is selected.

The yield surface of Von Mises has the following equivalent forms [1, 2, 42, 44]:

( ) ( )
2
1 1
, 0
2 3
p
ij ij Y
f s s T = = (60)

( ) ( )
3
, 0
2
p
ij ij Y
f s s T = = (61)

Where
ij ij ij
s p = is the stress deviation tensor.

From equation (61), the gradient of the yielding surface is:

3
2
ij
ij VM
s
f

(62)

Introducing equation (62) in equation (52) produces the plastic strain increment for the Von Mises model.

3
2
p
ij ij
VM
d d s
=
3
2
2
2
2
p
x x
p
y y
p
z z
p
xy xy VM
p
xz xz
p
yz yz
s d
s d
s d
d
d
d
d

( (
( (
( (
( (
= ( (
( (
( (
( (
( (

(63)


160
and then, using equation (61), equation (57) becomes:

1
p
ij ij
Q
f f

=
| |

|
|

\ .
(64)

In the Von Mises yield criterion, the plastic multiplier is equal to the equivalent strain increment [30-32,
42]:

p
d d = (65)

Equation (65) into equation (52) becomes:

1
p
p
ij ij
f

=

(66)
Equation (66) into equation (64) becomes:

1 1 1
p
ij
p
p p
ij
Q
f
H
f

= = =

(67)

Where: H

is the hardening modulus.

If 0 H = Perfect plasticity.
0 H > Rigid plasticity.
0 H < Soft plasticity.

Finally, using equations (62), (67), and (56) the plastic strain increment becomes:

1 3 3 3
2 2 2
ij ij p kl
ij kl
VM VM VM
s s
s f
d d dT
H T

| |
= +
|
\ .
(68)

or

1 3 3
2 2
ij ij p
ij VM
VM VM
s s
f
d d dT
H T

| |
= +
|
\ .
(69)

Equations (68) and (69) indicates that if the temperature dependency of the yield surface is expressed as in
(60) or (61), the effect of temperature is indicated by an additional term in the well-known relationship
between strain and stress.


161
The elastic strain tensor can be divided by deviation and spherical components [1]:

1
3
e e e
ij ij ij kk
e = + (70)

The relationship between stress and strain into their deviation and spherical components are, respectively
[1]:

2
ij e
ij
s
e
G
= (71)

1
3
e
kk kk
K
= (72)

Where:
( ) 2 1
E
G
v
=
+
is the shear modulus and
( ) 3 1 2
E
K
v
=
is the bulk modulus.

The incremental forms of equations (71) and (72) are:

2
1 1
2 2
e
ij ij ij
de ds s dG
G G
= (73)

2
1 2 1
3
e
kk kk kk
v
d d dK
E K

| |
=
|
\ .
(74)

Introducing equations (73) and (74) into equation (70), the elastic strain increment becomes:

2
2
1 1 1 1 2 1 2
2 2 3
e
ij ij ij ij kk ij kk
v v
d ds s dG d dK
G G E E

| | | |
= +
| |
\ . \ .
(75)

For the case of elastic condition, the thermal strain is defined as the thermal expansion of the material [45]:

ij
d dT
= (76)

Where: is the thermal expansion coefficient of material.

Substituting equations (69), (75) and (76) into equation (21), the total strain increment becomes:

2
2
1 1 1 1 2 1 2
2 2 3
ij ij ij ij kk ij kk
v v
d ds s dG d dK
G G E E

| | | |
= +
| |
\ . \ .

3 3
2 2
ij ij
ij VM
VM VM
s s
f
dT d dT
H H T

+ + +
(77)


162

The inverse relationship of equation (77) becomes:

2
2
3
1 1
2
3 1 2 3
1
3
ij kl kl
ij ij kk ij ij kk
VM
Gs s d
E
d d Gd K dT dK
H v K
G
| | (
= + +
|
(
| |
\ .
+
|
\ .

1
1
1 1
3 3
ij
ij
VM
f
s dT
s
T
dG
H H G
G G
| |
|

\ .
+ +
`
| | | |

+ +
| |

\ . \ .
)
(22)

The variation of the material properties dGand dK in equation (22) is due to temperature only [2]. In the
case of elastic conditions, the change in the stress can be written as:

2
1 1
2
3 1 2 3
ij
ij ij kk ij ij kk
s
E
d d Gde dG K dT dK
v G K

| | (
= + +
|
(
\ .
(78)

5. Formulation by finite element method of the welding process

The finite element method is a generalized Rayleigh-Ritz method, which uses interpolation to express the
variables in terms of its values in a finite number of nodes [30]. In the welding process, no external force is
applied to the welded structure. Thermal loads are the only source of residual stress and distortion [1-3].
Therefore, The principle of virtual work without mechanical loading, is written as [26]:

0
ij ij
V
d d dV =
(27)

where
ij
d is the variation in the strain increment
ij
d . The stress increment
ij
d is given in equation
(22).

In matrix notation, the thermal elastic plastic constitutive equation (eqn. 22) becomes [26]:

{ } { } { } { }
ep
d D d C s M dT ( =

(28)

where:
ep
D (

is the elastic plastic stiffness matrix, { } C s and { } M dT are terms related with the
distortion and volume change due to the temperature.


163
In extended notation, equation (28) can be expressed as:

2
2
2
1
1 2 1 2 1 2
1
1 2 1 2 1 2
1
1 2 1 2 1 2
2
x y x xy x yz
x x z x zx
x y y y z y xy y yz y zx
x
x
y z z xy z
x z z
x
xy
yz
zx
s s s s s s
s s s s s v v v
v L v L v L L L L
s s s s s s s s s s s
v v v
d
v L v L v L L L L
d
s s s s s
s s s v v v
d
v L v L v L L
G
d
d
d

=
`

)
2
2
2
2
1
2
2
1
2
1
2
x
y
yz
z zx
z
xy
x xy y xy z xy xy xy yz xy zx
y
x yz y yz z yz xy yz yz yz zx
y zx xy zx yz zx
x zx z zx zx
d
d
s
s s
d
L L
d
d
L L L L L L
L L L L L L
s s s s s s
s s s s s
L L L L L L
(
(
(
(
(
(
(
(
(
(

(
(
(

(
(
(

(

2
z
zx
d

`

)

1
1
1
0
0
0
x
y
z
xy
yz
zx
s
s
s
C M
s
s
s

` `

) )
(79)

Where:

2
1 2 2
1
3 3
VM VM
f
C dG dT
G L L T
| | | |
= +
| |
\ .
\ .
(80)

2
3
kk
M K dT dK
K
(
= +
(

(81)

2
2
1
3 3
VM
H
L
G
| |
= +
|
\ .
(82)

The virtual work equation (27) is now written as:

{ } { } 0
T
V
d d dV =
(29)


164
where: is the volume of the body to be analyzed and is the transpose of . The strain-
displacement for three-dimensional problem can be written in matrix form as [26]:

{ } { }
0 0
0 0
0 0
0
0
0
x
y
z
d du
y x
z y
z x
(
(
(
(
(
(
=
(

(
(
(

(
(
(

(
(

(30)

In the finite element solution, the increments in the displacements for each element are approximated as
[30]:

{ } ( ){ }
1
, ,
n
i
i
du N x y z U
=
=
(31)

where: n

is the number of nodal points for each element, ( , , )
i
N x y z are the shape functions and { } U
are element nodal increments.

Substituting equation (31) into equation (30) becomes:

{ } | |{ } d B U = (32)
Where
| | | | ( )
1
, ,
n
i
i
B N x y z
=
=

is the strain-displacement matrix for each element.

Using equation (31) and equation (32) becomes:

{ } { } | |
T T T
d U B = (33)

Substituting equation (27), (28) and (32) into equation (29) becomes:

{ } | | | || | { } | | { } | | { }
0
T T T T
V V V
U B D B dV U C B s dV M B d dV

(
=
(

(34)


165
The vectors { } U and { } U are not included in the integrals because they are not function of the
coordinates. Equation (34) must be valid for any admissible virtual nodal increase { } U of equilibrium
configuration. Therefore, equation (34) becomes:

{ } | |{ } { } { }
0
T
U K U F = (35)

Now, the expressions inside the braces must be zero, therefore:

| |{ } { } K U F = (36)

Where | |
K is the global stiffness matrix for the assembly of elements and is the sum of the element
stiffness matrices , and is the global load vector equal to the sum of the element load vectors .

The stiffness matrix for each element is:

| | | | | |
e
T
ep
e
V
K B D B dV ( =

(37)

And the element load vector is:

{ } | | { } | | { }
e e
T T
e
V V
F M B dT dV C B s dV = +

(38)
Equation (36) is non-linear and can be rewritten as [30]:

{ } { } { }
1 t i i i
t t t t
K U F F

+ +
( =

(83)

{ } { } { }
1 i i i
t t t t
U U U
+ +
= + (84)

where is the iteration number,
t
K (

is the stiffness matrix in the time ,
{ }
i
U is the nodal increment
vector in the iteration , and
{ }
i
t t
F
+
is the nodal load vector corresponding to the state of stress for the time

t .

6. Solution of the equilibrium equations by finite element method

The equations of equilibrium governing the linear dynamic response of a system of finite element are [46]:

| |{ } | |{ } | |{ } { } M U C U K U F + + =

(85)


166

Where: | |
M , | |
C and are the mass, damping and stiffness matrices; is the external load vector;
and and are the displacement, velocity, and acceleration vectors of the finite element assemblage.
Equation (68) is a linear differential equation of second order and is derived from Newtons second law of
motion. Therefore, equation (85) can be rewritten as:

(86)

where: are the inertia forces, are the damping forces and
are the elastic forces , all of them time-dependent.

In finite element analysis, the system of equations (eqn. 85) is solved with two methods of solution [46]:
direct integration and mode superposition. For this investigation we consider the direct integration methods.

6.1 Direct integration methods

In direct integration the equations in (85) are integrated using a numerical step-by-step procedure, the
term direct meaning that prior to the numerical integration the equations are not transformed into a
different form [46]. Direct numerical integration is based on two ideas. First, instead of trying to satisfy (85) at
any time t, it attempts to satisfy (85) only at discrete time intervals t

apart. Second, it assumes a variation of
displacements, velocities and acceleration within each time interval t .

Direct integration assumes that the displacement, velocity and acceleration vectors at time 0 are known
and the solution to (85) is required from time 0 to time . In the solution the time interval , is divided into
equal time intervals t , and the integration scheme establishes an approximate solution at times 0, t , 2 t
,.., t , t t + ,, . Since an algorithm calculates the solution at the next required time from the solution at the
previous time, the algorithms assume that the solution at times 0, t , 2 t ,.., t , is known and the solution at
time t t + is required next.

The central difference method and the Newmark method are the most common method of direct
integration [30, 31, 46].

The central difference method

To approximate the velocities and accelerations in terms of displacement, appropriate finite difference
expressions can be used. This method assumes that the velocity and acceleration are given by:

{ } { } { } { } ( )
2
1
2
t t t t t t
U U U U
t
+
= +
(87)

{ } { } { } ( )
1
2
t t t t t
U U U
t
+
=
(88)


167
Expressing equation (85) in terms of the time ,

| |{ } | |{ } | |{ } { }
t t t t
M U C U K U F + + =

(89)

Introducing equations (87) and (88) into equation (89) and grouping the terms { } { } ,
t t t
U U
+
, and { }
t t
U

,

| | | | { } { } | | | | { } | | | | { }
2 2 2
1 1 2 1 1
2 2
t t t t t t
M C U F K M U M C U
t t t t t
+
( ( (
+ =
( ( (

(90)

The term { }
t t
U
+
is determined from equation (90), the solution of { }
t t
U
+
is thus based on using the
equilibrium conditions at time t . The integration procedure is therefore called explicit, and it is noted that
such integration scheme does not require factorization of the stiffness matrix | |
K in the step-by-step solution.

Newmark Method

This method considers that the displacement, velocity and acceleration are now in time t t + . The
following assumptions are used:

{ } { } { } ( ) { } { }
2
1 1 1
1
2
t t t t t t t
U U U U U
t t
+ +
| |
=
|

\ .

(91)

{ } { } ( ){ } { }
1
t t t t t t
U U U U t
+ +
(
= + +

(92)

{ } { } { } { } { }
2
1
2
t t t t t t t
U U U t U U t
+ +
(
| |
= + + +
| (
\ .

(93)

Now, expressing equation (85) for the time t t + ,

| |{ } | |{ } | |{ } { }
t t t t t t t t
M U C U K U F
+ + + +
+ + =

(94)

Introducing equations (91), (92) and (93) into equation (94),

| | { } { } ( ) { } { }
2
1 1 1
1
2
t t t t t
M U U U U
t t
+
(
| |
+
| (

\ .

| | { } ( ){ } { }
1
t t t t
C U U U t
+
(
(
+ + +

| | { } { } { } { } { }
2
1
2
t t t t t t t
K U U t U U t F
+ +
( ( | |
+ + + =
( | (
\ .

(95)


168

Introducing equation (91) into equation (95),

| | { } { } ( ) { } { }
2
1 1 1
1
2
t t t t t
M U U U U
t t
+
(
| |
+
| (

\ .

| | { } ( ){ } { } { } ( ) { } { }
2
1 1 1
1 1
2
t t t t t t t
C U U U U U U t
t t

+
( ( ( | |
+ + +
( ( | (

\ . (

| | { } { } { } { } { } ( ) { } { } { }
2
2
1 1 1 1
1
2 2
t t t t t t t t t t
K U U t U U U U U t F
t t

+ +
( ( ( | | | |
+ + + =
( ( | | (

\ . \ . (

(96)

Factorizing the term
t t
U
+
in equation (96) and grouping terms,

| | | | | | { } | | { } { } { }
2 2
1 1 1 1
1
2
t t t t t
M C K U M U U U
t t t t

+
(
( | |
+ + =
| (
(

\ .

| | { } ( ) { } { }
1 1
1 1 1
2 2
t t t
C U t U U
t

( ( | | | | (
+ +
( | | ( (
\ . \ .

| | { } ( ){ } ( ){ } { }
2
2
2
1 1
1 1 1
2 2
t t t t t
t
K U t U t U F
t

+
( | | ( | | | | (
+ + +
( | | | ( (
\ . \ . \ .
(97)

Parameters and can be determined to obtain integration accuracy and stability. If these parameters
satisfy the following conditions, the solution of the equations is inconditionally stable [46].

2
1 1
4 2

| |
+
|
\ .
,
1
2
, y
1
0
2
+ + > (98)

With
1
4
= and
1
2
= , the Newmark method becomes a constant-average-acceleration method.
Introducing these values into equation (97),

| | | | | | { } { } | | { } { } { } | | { } { }
2 2
4 4 4 4 2
t t t t t t t t t
M C K U F M U U U C U U
t t t t t
+ +
( ( (
+ + = + + + + +
( ( (

(99)


169
The term { }
t t
U
+
is determined from equation (82). The solution of { }
t t
U
+
is thus based on the equilibrium
conditions at time t t + . The integration procedure is therefore called implicit. It is noted that such
integration scheme requires factorization of the stiffness matrix | |
K in the step-by-step solution.

6.2 Solution of nonlinear equations in dynamic analysis

Solving the nonlinear dynamic response of a finite element system requires a similar procedure to a
nonlinear static analysis. The main difference is adding the integration algorithms at time .

Explicit integration

As in linear analysis, the equilibrium of the finite element system is considered at time in order to
calculate the displacements at time t t + . Neglecting the effect of a damping matrix, the equations of
equilibrium governing the system for each discrete time step are [46]:

| |{ } { } { }
t t t
M U F R =
(100)

where: { }
t
R is the nodal point force at time t .

The solution for the nodal point displacements at time t t +

is obtained using the central differences
approximation for the accelerations at the timet . The main restriction in the use of the central difference
method is the time step size. For stability, the time step size t must be smaller than a critical time step
cr
t .
This critical time step can be calculated from the stiffness and mass properties of the finite element model and
element size.

Implicit integration

As in linear analysis, the implicit integration considers the equilibrium of the system at time t t + . This
requires an iterative procedure for nonlinear analysis. Using the modified Newton-Raphson and neglecting the
effects of a damping, the governing equilibrium equations are [46]:

| |{ } | |{ } { } { }
1 i i i i
t t t t t t t
M U K U F R

+ + +
+ =
(101)

6.3 Comparison between explicit and implicit integration

There are two main differences between explicit an implicit integration: (1) in explicit integration (eqn. 73),
the stiffness matrix does not require factorization in the step-by-step solution, (2) The time step size t in the
explicit integration is restricted by critical time step
cr
t .

In finite element welding process simulation, the effects of inertia and damping are ignored [1-3].
Furthermore, the welding process simulation is performed by nonlinear transient analysis. Therefore, a


170
commercial finite element code with explicit integration is not appropriate because the time step size
limitation may considerably slow down calculations. Also, equation (83), which is obtained by finite element
method for a thermal elastic plastic model has the form of the implicit integration equation (eqn. 101) neglect
the effects of inertia and damping. Therefore, a commercial package that uses the implicit integration is more
appropriate for the simulation of the welding process.


171

APPENDIX 2


172
Definition and justification of the applied finite elements

The proposed sequentially-coupled thermo-mechanical analysis involves two steps. In the first step a
transient heat transfer analysis is performed followed by a thermal elastic plastic analysis.

To perform the thermal analysis, it is necessary to select an element type that can simulate:

Transient thermal analysis.
The weld thermal cycle [figure 3.6, page 58], which determines the temperature as a function of time
in the weld bead (i.e., all elements belonging to the weld bead).
Conduction heat transfer in the base metal.
Heat loss by convection and radiation to the surroundings in the base metal.

To perform the structural analysis, it is necessary to select an element type that can simulate:

Homogeneous and isotropic material.
Elastic-plastic material behavior following the von Mises yield criterion and the associated flow rule
with linear isotropic hardening by strain.
Small strains.

An important factor for the selection of the structural element is the stress state for each analyzed
structure, as determined below (figures 7 to 10). The analysis reveals that all structures develop bending
stiffness effects, except for the I-type structure subject to tension.

ANSYS offers a broad variety of elements for the analysis of two and three-dimensional thermal and
structural problems such as BEAM, SHELL, PLANE and SOLID elements. All these elements can simulate
mechanical and thermal effects in addition to bending stiffness. The plane and solid elements solved this
bending condition with the link between several elements.

Beam elements are appropriate for structural members whose cross-sectional dimensions are small
compared to its length. They are also commonly subjected to transverse loading. Shell elements are designed
to efficiently model thin structures and bending capability. Otherwise, these elements present initial instability
in perfectly flat plates. Avoiding this effect requires an initial displacement. Beam and shell elements are
therefore discarded while the plane and solid elements are selected for the numerical simulation.

1.- I-type structure subject to tension.

This structure consists of a central bar and two lateral bars (three parallel bars joined together). Due to the
symmetry of the structure, only a quarter of the model is analyzed as is shown in figure 7(a). The stress state is


173
obtained in two points of interest. Both points are located on the upper surface. Points 1 and 2 are
respectively located on the geometric center of the central bar and the lateral bar. When the weld thermal
cycle is completed, element 1 is subject to pure tension while element 2 is subject to pure compression, both
in the X-direction (figure 7b). Due to Poisson ratio effects, element 1 undergoes contraction while element 2
undergoes expansion, both in the Y and Z-direction (figure 7c).

(a)

(b)

(c)

Figure 7. Stress state in the I-type structure subject to tension


174
2.- L-type structure

The L-type structure consists of an L-shaped rigid base and two perpendicularly attached thin bars (figure
8a). Due to the symmetry of the structure, the thin bars develop the same stress state, and only the vertical
bar is analyzed. Figure 8(b) shows the load and moment applied to the vertical bar as a result of the weld
thermal cycles. The stress state is obtained in two points of interest located on the upper surface. The first
point is located in the geometric center of the bar and the second is located on the lateral border in the middle
of the bar. Element 1 undergoes pure shearing due to bending and tension in the Y-direction due to weld
shrinkage (figure 8c). Element 2 undergoes tension in the Y-direction due to weld shrinkage and bending
(figure 8d).

Figure 8. Stress state in the L-type structure

The next equation is used to determine the shearing stress

(102)
And the Y-direction stress is.

(103)


175
3.- Stiffened symmetrical flat frame (flat stress condition)

Stiffened symmetrical flat frame consists of four external bars and three internal bars joined by ten welds
(figure 9a). Figure 9(b) shows the load and moment applied to one half of the external vertical bar as a result
of the weld thermal cycles. The stress state is obtained in two points of interest. Both points are located on the
upper surface. The first is located in the geometric center of the extracted segment from the external vertical
bar and the second is located on the lateral border in the middle of the extracted segment. Element 1
undergoes pure shearing due to bending and tension in the Y-direction due to the weld shrinkage (figure 9c).
Element 2 undergoes tension in the Y-direction due to weld shrinkage and bending (figure 9d). The shearing
stress ( ) and stress ( ) are respectively determined by equations (120) and (103).

Figure 9. Stress state in the stiffened symmetrical flat frame


176
4.- 3-Dimensional unitary cell - type symmetrical structure

The 3-dimensional unitary cell consists of two horizontal and two vertical plates joined by fillet welds
(figure 10a). Figure 10(b) shows the moments applied to a segment of the vertical plate as a result of the weld
thermal cycles. This segment is delimited in the Y-direction by the weld beads and in the Z-direction by the
plates lateral borders. The stress state is obtained in one point of interest located on the upper surface in the
geometric center of the segment from the vertical plate. Element 1 undergoes shearing and tension in the Y
and Z-direction due to bending and to the weld shrinkage (figure 10c). The shearing stress ( ) and stress
( ) are respectively determined by equations (120) and (103).

Figure 10. Stress state in the 3-Dimensional unitary cell

Tables 1 and 2 show some plane and solid elements commonly used by ANSYS. In general, all elements
have the same characteristics (degree of freedom, input data, and output data), except for two main
differences: 1) the order of elements: higher order elements achieve better results and greater accuracy,
however, these elements require more computational time because numerical integration of elemental
matrices is more involved. Higher order elements are well suited to problems with curved boundaries. 2) The


177
shape of element: quadrilateral and brick element present better meshing control than triangular and
tetrahedral elements.
For these two reasons, PLANE55 and SOLID70 elements are respectively selected for two and three
dimensional thermal analysis. Both elements have isotropic material properties and a single degree of
freedom: temperature. These elements allow straightforward thermal modeling of the welding process,
including thermal conductivity and specific heat as a function of temperature (Fig 3.3, page 56). The convective
heat transfer coefficient is considered constant at 10 W/m
2
THERMAL ELEMENTS
C (natural convection). Heat loss by convection is
a boundary condition and is applied only to surfaces in contact with the surrounding air assumed at 20C,
same as the initial temperature of the base metal.

The elements PLANE42 and SOLID45 are respectively selected for two and three dimensional structural
analysis. PLANE 42 and SOLID45 elements have isotropic material properties and respectively two and three
degrees of freedom at each node: translation in the nodal directions. Both elements couple temperatures with
strain, therefore requiring the thermal expansion coefficient, elasticity modulus, tangent modulus, yield stress
and Poissons ratio dependent on temperature (figure 3.3 page 56). Finally, thermal loads are transferred to
the structural analysis through the concept of body load used through load step. PLANE 42 and SOLID45
elements have this capability because the temperatures may be specified as body loads at the nodes. The
elements PLANE42 and SOLID55, and SOLID70 and SOLID45 are equivalent to performing coupled thermo-
mechanical analysis.

Element Dimension Number
of nodes
Main
application
Degree of
freedom
Input data Output data Comments
PLANE35 2-D
triangular
element
6 Steady-state
and transient
conduction
Heat transfer
Temperature - Convection or heat fluxes
may be applied to the
elements surfaces.
- Nodal temperatures.
- Thermal gradients.
- Thermal fluxes components.
-This element is
compatible with
the PLANE77
element.
PLANE55 2-D
quadrilater
al element
4 Steady-state
and transient
conduction
Heat transfer
elements surfaces.
- Better meshing
control
PLANE77 2-D
quadrilater
al element
8 Steady-state
and transient
conduction
Heat transfer
elements surfaces.
- This element is
well suited to
model problems
with curved
boundaries.
SOLID70 3-D brick
element
8 Steady-state
and transient
conduction
Heat transfer
elements surfaces.
-Heat generation rates may
be applied at the nodes.
- Average face temperature.
- Temperature-gradient
components.
- Vector sum at the centroid of
the element.
- Heat flux components.
- Better meshing
control
SOLID90 3-D brick
element
20 Steady-state
and transient
conduction
Heat transfer
elements surfaces.
-Heat generation rates may
be applied at the nodes.
- Average face temperature.
- Temperature-gradient
components.
- Vector sum at the centroid of
the element.
- Heat flux components.
- This element is
well suited to
model problems
with curved
boundaries.

Table 1. Thermal (conduction) elements used by ANSYS


178

STRUCTURAL ELEMENTS
Element Dimension Number
of nodes
Main
application
Degree of
freedom
Input data Output data Comments
PLANE42 2-D
quadrilateral
element
4 Isotropic solid
problems
Translation in
the nodal x
and y-
directions
-Distributed surface load may
be applied to elements
surfaces.
-Temperatures may be input
as element body loads at the
nodes.
- Nodal displacements.
- Directional stress and
principal stresses.
- Better meshing
control.
-This element
may be used to
analyze large-
deflection, large-
strain, plasticity,
and creep
problems
PLANE82 2-D
quadrilateral
element
8 Isotropic solid
problems
Translation in
the nodal x
and y-
directions
-Distributed surface load may
surfaces.
nodes.
- Directional stress and
principal stresses.
- This element is
well suited to
model problems
with curved
boundaries.
-This element
may be used to
analyze large-
deflection, large-
strain, plasticity,
and creep
problems
SOLID45 3-D brick
element
8 Isotropic solid
problems
Translation in
the nodal x, y,
and z-
directions
- Distributed surface load may
surfaces.
nodes.
- Normal components of the
stress in x, y, and z-
directions; shear stress and
principal stresses.
-This element
may be used to
analyze large-
deflection, large-
strain, plasticity,
and creep
problems
SOLID92 3-D
tetrahedral
element
10 Isotropic solid
problems
Translation in
the nodal x, y,
and z-
directions
- Distributed surface load may
surfaces.
nodes.
- Normal components of the
stress in x, y, and z-
directions; shear stress and
principal stresses.
-This element
may be used to
analyze large-
deflection, large-
strain, plasticity,
and creep
problems

Table 2. Structural elements used by ANSYS


179

APPENDIX 3


180
Response to critical comments

1) Even though the I and L- structures are considered only in a qualitative sense in order to validate the
numerical model, their absolute dimensions should be given. The same applies to the flat frame analyzed in
chapter IV, which plays a crucial role in the thesis.

The absolute dimensions for the structures analyzed are:

a) I-type structure subject to tension consists of a central bar and two lateral bars (three parallel bars joined
together).

Figure 11. Dimensions of the I-type structure [mm]

b) L-type structure consists of an L-shaped rigid base and two perpendicularly attached thin bars. The thickness
of the rigid base is 12.7 mm and the thickness of the bars is 3.175 mm.

Figure 12. Dimensions of the L-type structure [mm]


181
c) Stiffened symmetrical plane frame consists of four external bars and three internal bars joined by ten welds.

Figure 13. Dimensions of stiffened symmetrical plane frame [mm]

2) No information is given on how the welding parameters were converted into the heat conditions applied as
the input to the FE analyses.

The heat condition applied was obtained from [Y. Higashida, J. D. Burk, and F.V. Lawrence, JR.]. While these
authors did not determine heat transfer as a function of time, they did measure the thermal cycle at HAZ
adjacent to the fusion line. Chromel-alumel thermocouples (0.51 mm diameter) were spot welded onto the
surface of an ASTM A36 steel plate near the fusion line. The localization of the thermocouple was determined
by preliminary measurements to define the fusion line position. The thermocouples were electrically and
thermally shielded and were connected to an oscilloscope. The thermal cycle was photographically recorded.

The amount of energy available in a heat source is called its energy capacity. For an electrical arc, the
available heat at the source is:

(104)

Where E is the arc voltage (volts), and I is the arc current (Amperes). There are similar relationships for
electrical resistance sources, laser beams, etc. Net heat input in welding is the quantity of energy introduced
per unit length of weld from a traveling heat source, whether a flame, electric arc, plasma, or beam. The net
energy input is computed as the ratio of the power of the heat source (in Watts) to its travel speed:

(105)


182
Where H is the net energy input (in watt-seconds or joules per mm or inch), P is the total input power of
the heat source (in watts), and is the welding velocity (in mm per second or inch per minute). For an
electrical arc (considering the heat transfer efficiency):

(106)

This heat could be introduced in FEM to obtain temperature in function of time; neglecting phase change
effects.

3) Based on the data in table 4.3, in page 85 the Author claims that procedure 1 I-O (welding sequence from
the inside to the outside) should be used to improve the stress-distortion relation. However, the data suggest
that procedure 1 O-I yields the same stress-distortion behavior.

Welding sequences 1 I-O and 1O-I present similar results. However, welding sequence 1 I-O was selected as
the best because the order of importance to reduce the residual stress is slightly better than 1O-I as shown in
the table below. Taking in account that the welding sequence for the case of the stiffened symmetrical flat
frame reduces the distortion twice as much as the residual stress, the change from one position to another in
the improvement of the residual stress has greater importance than a change from one position to another in
the improvement of the distortion.

Welding
sequence
Parameters Order of importance of welding sequence to reduce:
Residual stress Distortion
Umax
(mm)
vm
W=1, P=0 Node 681
(MPa)
proper = 1
inappropriate= 36
W=0, P=1 proper = 1
inappropriate= 36
1 I-O 0.286 205.48 0.872 4 0.905 19
1 O-I 0.277 206.30 0.875 7 0.877 17

Table 3. Comparison between welding sequences 1 I-O and 1O-I

4) From distortion measurement for the three dimensional cell the Author concludes in p. 136 that laying 8
weld beads produces more distortion than 24 beads. However, as revealed in table 7.3, the sum of the
measured distortion is greater for 24 welds (14.98 mm) than for 8 welds (12.93 mm). Also the maximum
distortion value is smaller for 8 welds than for 24 welds (1.40 and 1.50 mm, respectively). The same is
indicated by the computed distortion data in table 5.2. In this context it cannot be overlooked that electrode
craters unfavorably affect the fatigue properties of a structure. Therefore, laying 8 welds seems to be more
appropriate than 24 sub-welds.

Table 7.3 [page 130] depicts the distortion measured after application of the weld and it represents the
final distorted shape of the specimens. It is necessary to note that this distortion does not represent


183
the distortion induced by the welding process because the specimens were originally distorted by the
application of the welding tacks. The final distortion (distortion induced by welding) is determined by:

Final distortion = Distortion after applying Distortion after applying (41)
the welding tacks welding

Table 4 depicts the values of the final distortion in the 24 points of interest of the 3-dimensional unitary
cell specimens (the final distortion is calculated with equation (41) [page 126] and tables 7.2 [page 129]
and 7.3 [page 130]). The data in the table show that the sum of the measured distortion is lower in the
specimens with 24 weld beads than specimens with 8 weld beads with the exception of specimen T7.

Point
Specimen
8 Weld beads 24 Weld beads
T7 T11 T8 T4 T9 T5 T10 T6
1 0,10 0,10 0,05 0,05 0,05 0,05 0,05 0,05
2 0,15 0,15 0 0 0,2 0,2 0,05 0,05
3 0,10 0,10 0.05 0,05 0,05 0,05 0,05 0,05
4 0,10 0,15 0 0 0,2 0,2 0,05 0,05
5 1,75 1,90 1,90 1,93 1,55 1,55 1,90 1,68
6 1,93 1,88 1,85 1,85 1,58 1,70 1,65 1,83
7 1,80 2,05 1,83 1,93 1,63 1,53 1,65 1,75
8 1,85 1,98 1,83 1,93 1,65 1,53 1,60 1,85
9 1,40 1,45 1,70 1,80 1,50 1,55 1,15 1,20
10 0,30 0,35 0,15 0,20 0,40 0,30 0,15 0,15
11 1,50 1,50 1,70 1,80 1,45 1,50 1,10 1,40
12 1,30 1,65 1,90 1,85 1,40 1,35 1,45 1,25
13 0,10 0,10 0,10 0,10 0,25 0,25 0,15 0,15
14 1,35 1,45 1,80 1,80 1,50 1,60 1,10 1,25
15 1,30 1,65 1,75 1,90 1,50 1,25 1,10 1,20
16 0,35 0,40 0,10 0,10 0,25 0,40 0,10 0,10
17 1,45 1,70 1,65 1,80 1,50 1,45 1,05 1,30
18 1,50 1,50 1,70 1,85 1,60 1,60 1,00 1,30
19 0,20 0,25 0,10 0,15 0,20 0,20 0,10 0,10
20 1,35 1,60 1,80 1,85 1,45 1,40 1,15 1,20
21 0,50 0,45 0,25 0,25 0,50 0,70 0,25 0,30
22 0,55 0,50 0,30 0,30 0,55 0,60 0,30 0,30
23 0,50 0,45 0,25 0,30 0,60 0,60 0,30 0,30
24 0,50 0,45 0,20 0,30 0,55 0,60 0,35 0,30
Sum of the measured distortion 21.93 23.76 22.96 24.39 22.11 22.16 17.8 19.11

Table 4. (Table 7.4) Final distortion in the 24 points of interest
of the 3-dimensional unitary cell specimens, mm.


184

Table 5 depicts the computed distortion at the 24 points of interest for the different welding
sequences. The table shows that simulations 3 and 4 (with 24 weld beads) produce lower distortion
than the others (with 8 weld beads).

Point Node Simulation
8 weld beads 24 weld beads
1 2 3 4
1 15756 0,08 0,05 0,06 0,04
2 17536 0,11 0,01 0,21 0,05
3 19015 0,08 0,07 0,06 0,04
4 17238 0,11 0,02 0,21 0,05
5 14571 1,85 1,87 1,68 1,65
6 20554 2,0 2,0 1,75 1,85
7 20508 1,82 1,83 1,67 1,64
8 14498 1,99 1,98 1,73 1,80
9 11 1,36 1,70 1,52 1,12
10 386 0,31 0,18 0,35 0,19
11 758 1,54 1,92 1,53 1,13
12 35172 1,49 1,86 1,55 1,11
13 38137 0,08 0,07 0,2 0,19
14 41155 1,57 1,95 1,74 1,24
15 35099 1,51 1,88 1,50 1,08
16 37839 0,39 0,10 0,3 0,08
17 41109 1,33 1,65 1,54 1,13
18 1 1,56 1,94 1,74 1,21
19 361 0,19 0,10 0,16 0,08
20 754 1,47 1,83 1,56 1,01
21 167 0,57 0,25 0,66 0,29
22 36357 0,59 0,34 0,59 0,34
23 591 0,57 0,31 0,6 0,33
24 39616 0,52 0,22 0,68 0,29
Maximum
distortion
2,97 3,05 2,04 2,10

Table 5. (Table 5.2) Distortion at the 24 points of interest corresponding to different welding sequences
In the 3-dimensional unitary cell structure, mm.

Structures with a greater number of sub-welds are more appropriate to reduce distortion, but also
reduce fatigue performance due to higher residual stresses which may reach the yield point in normal
operating conditions. Therefore, to avoid the loss of stability, laying 8 weld beads seems to be more
appropriate than 24 sub-welds.


185
5) The term converse welding method used in pages 140-143 in the context of the welding procedure
description is not explained.

The converse welding method consists of applying the opposite direction between adjacent layers in multi-
layer weld, or between beads in every layer [17].

6) What is the relationship between the residual stress state in a structure made of several welded elements of
various types and strain in the structural elements?

For symmetrical structures with rigid elements, distortion tends to decrease more than residual stress. For
example, for the stiffened symmetrical flat frame, the welding sequence reduces the distortion twice as much
as residual stress. This relationship would tend to be equaled and with elements even more rigid it would tend
to reverse the relationship among residual stress-distortion. On the other hand, the curves of the ranges of the
values of P and W for the welding sequence 1 I-O, and the combined welding sequence, the intersection point
would tend to move toward the left as is shown in figure 14.

Figure 14. (Figure 4.7) Ranges of the values of P and W for welding sequence 1 I-O
and combined welding sequence

For symmetrical structures made from several welded element of various types (e.g., welding of I-type
beams with beams of rectangular cross section), it is not possible to make a preliminary conclusion of the
relationship between residual stress-distortion, however, the proposed numerical model of the welding
process in the present investigation can analyze this type of configuration of structures.

7) How is the residual stress state measured in the I-type structure after welding?

In the I-type specimen, the stress relaxation method was used. These techniques are based on the known
fact that the strains that occur during unloading are elastic even when the material has undergone plastic


186
deformation. The determination of residual stress takes places when the stress is relaxed by cutting the
specimen. Four electrical strain gauges were mounted to measure the strain release as shown in figure 15 (a).
Strain gauges 1 and 3 are located in the central zone of the lateral bars, while strain gauges 2 and 4 are located
in the central zone of the central bar. Figure 15 (b) shows welding parameters employed.

Figure 15. (Figure 3.11) Experimental tests: a) Localization of the points of interest in the I-type specimen

Procedure followed to obtain residual stress on the I-type specimen:

1.- Preparation of the specimen (e.g., machined operations).
2.- Welding application.
3.- Mounting of the electrical strain gauges.
4.- Cutting of the specimen, the cut was performed between the strain gauge 4 and weld bead.
5.- Measurement of the strain relaxation with the Wheatstone bridge.
6.- Application of the generalized Hookes law to obtain the residual stress state.

Tensile residual stresses were obtained in the central bar, while in the lateral bars compression residual
stresses were obtained. This is due to the balance conditions and geometry [3].

8) What about stability of the welded structure, particularly made from plate elements?

Buckling results from the loss of stability of the elements under compression stress induced in the
peripheral areas away from welds. The mechanism of buckling is hidden in the action of the inherent
(incompatible residual plastic) strains formed during welding. Welding-induced buckling differs from bending
distortion by its much greater out-of-plane deflections and several stable patterns. Buckling patterns depend
much more on the geometry of the elements, and types of weld joint especially depend on the thickness of
sheet materials under certain conditions.


187
Buckling distortion caused by longitudinal welds in plates (see figure above), in panels or in shells is mainly
dominated by longitudinal compressive stress produced in areas away from the weld, but the buckling
distortions caused by circular welds either in plates or in shells are mainly determined by transverse shrinkage
of welds in the radial direction whereby compressive stresses are produced in the tangential direction. The
buckling pattern depend much more on rigidity (thickness, dimensions and shape) of the elements to be
welded, as well as on the type of weld joint and welding heat inputs.

Figure 16. Buckling distortion

The most important stage in eliminating buckling distortion is rational design of welded structural
elements: plates, panels and shells. Buckling can be controlled by a variety of methods and technological
measures for removal, mitigation or prevention. In the design state, rational selection of geometry, thickness
of materials and type of weld joint is essential. In the fabrication stage, technological measures and techniques
to eliminating buckling can be classified as:

1.- Methods applied before welding.
2.- Methods adopted during welding operation.
3.- Methods applied after welding.

In the methods applied before welding, e.g. predeformation, post-weld buckling is compensated by
counterdeformation formed in the elements prior to welding by a specially designed fixture die. In the
methods applied after welding, once buckling is in existence, warpage is removed by special flattening
processes, using manual hammer, applying a mechanized weld-rolling technique or utilizing an electric-
magnetic pulse shock. These methods both before welding and after welding are arranged as special
operations on a production line and specific installations or fixtures are needed, resulting in increasing cost
and variable quality of welded elements.

Pretensioning can be classified in the category of methods applied either before or during welding. For
each particular structural design of panels, a device for mechanical tensile loading is required. Owing to their
complexity and reduced efficiency in practical execution, application of these methods is limited. In this sense,
the thermal tensioning is more flexible in stiffened panel fabrication.


188

APPENDIX 4


189
Proper welding sequence to reduce residual stress and distortion in common symmetrical structures in 2
and 3 dimensions based on the hypothesis developed in the sections 4.9 and 5.2.

Welding of the collinear tubes with 8 weld beads

Collinear tubes Welding Proper welding sequence to reduce
residual stress and/or distortion


190
Welding of tubes to 90 with 8 weld beads

Welding of tubes to 90 (Isometric view)

(top view) (top view)
Proper welding sequence to reduce Proper welding sequence to reduce
residual stress distortion


191

APPENDIX 5


192
Listing of commands of the numerical simulation of the welding process (I-type specimen subject to tension)

Transient thermal analysis
/title,transient thermal analysis
/PREP7
k,1,0,0
k,2,0.0015,0
k,3,0.003,0
k,4,0.00635,0
k,5,0.0127,0
k,6,0.0254,0
k,7,0.0381,0
k,8,0.0508,0
k,9,0.06985,0
k,10,0.101575,0
k,11,0.1100840,0
k,12,0.1365,0
k,13,0,0.00635
k,14,0.0015,0.00635
k,15,0.003,0.00635
k,16,0.00635,0.00635
k,17,0.0127,0.00635
k,18,0.0254,0.00635
k,19,0.0381,0.00635
k,20,0.0508,0.00635
k,21,0.06985,0.00635
k,22,0.101575,0.00635
k,23,0.1100840,0.0115944
k,24,0.1365,0.01159443
k,25,0.1096109,0.0209887
k,26,0.1365,0.0381
k,27,0.101575,0.0254
k,28,0.101575,0.0381
k,29,0.101575,0.0508
k,30,0.1365,0.0508
k,31,0,0.0254
k,32,0,0.0381
k,33,0,0.0508
k,34,0.101575,0.015875
k,35,0.0889,0
k,36,0.0889,0.00635
l,1,2
l,2,3
l,3,4
l,4,5
l,5,6
l,6,7
l,7,8
l,8,9
l,9,35
l,35,10
l,10,11
l,11,12
l,13,14
l,14,15
l,15,16
l,16,17
l,17,18
l,18,19
l,19,20
l,20,21
l,21,36
l,36,22
larc,22,23,34,0.009525
l,1,13
l,2,14
l,3,15
l,4,16
l,5,17
l,6,18
l,7,19
l,8,20
l,9,21
l,35,36
l,10,22
l,11,23
l,12,24
l,23,24
larc,23,25,34,0.009525
l,24,26
l,25,26
larc,25,27,34,0.009525
l,27,28
l,28,26
l,26,30
l,29,30
l,28,29
l,31,27
l,32,28
l,33,29
l,31,32
l,32,33
al,1,25,13,24
al,2,26,14,25
al,3,27,15,26
al,4,28,16,27
al,5,29,17,28
al,6,30,18,29
al,7,31,19,30
al,8,32,20,31
al,9,33,21,32
al,10,34,22,33
al,11,35,23,34
al,12,36,37,35
al,37,39,40,38
al,40,43,42,41
al,43,44,45,46
al,47,42,48,50
al,48,46,49,51
aglue,all
ET,1,PLANE55
ET,2,SOLID70
lesize,1,,,4
lesize,2,,,4
lesize,3,,,4
lesize,4,,,4
lesize,5,,,4
lesize,6,,,4
lesize,7,,,4
lesize,8,,,4
lesize,9,,,4
lesize,10,,,4
lesize,38,,,4
lesize,41,,,4
lesize,47,,,10
lesize,12,,,4
lesize,51,,,4
lesize,24,,,4
MSHAPE,0,2D
MSHKEY,1
amesh,all
TYPE,2
EXTOPT,ESIZE,3,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all,, ,0,0,-0.003175,,,,
TYPE,2
EXTOPT,ESIZE,2,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,83,,,0,0,0.0047625,,,,
VEXT,79,,,0,0,0.0047625,,,,
VEXT,75,,,0,0,0.0047625,,,,
VEXT,71,,,0,0,0.0047625,,,,
VEXT,67,,,0,0,0.0047625,,,,
VEXT,63,,,0,0,0.0047625,,,,
VEXT,17,,,0,0,0.0047625,,,,
VEXT,16,,,0,0,0.0047625,,,,
VEXT,15,,,0,0,0.0047625,,,,
VEXT,14,,,0,0,0.0047625,,,,
VEXT,13,,,0,0,0.0047625,,,,
VEXT,12,,,0,0,0.0047625,,,,
nummrg,all
eplot
/VIEW,1,1,1,1
/ANG,1
/REP,FAST
TOFFST,273
MPTEMP,,,,,,,,
MPTEMP,1,20
MPTEMP,2,90
MPTEMP,3,200
MPTEMP,4,314
MPTEMP,5,423.5
MPTEMP,6,533
MPTEMP,7,643
MPTEMP,8,753
MPTEMP,9,863
MPTEMP,10,972.5
MPTEMP,11,1090
MPTEMP,12,1216
MPTEMP,13,1321.5
MPTEMP,14,1423.5
MPTEMP,15,1500
MPDATA,KXX,1,,51.82
MPDATA,KXX,1,,50.73
MPDATA,KXX,1,,48.78
MPDATA,KXX,1,,45.73
MPDATA,KXX,1,,42.07
MPDATA,KXX,1,,37.80
MPDATA,KXX,1,,33.53
MPDATA,KXX,1,,28.04
MPDATA,KXX,1,,25
MPDATA,KXX,1,,26.82
MPDATA,KXX,1,,28.04
MPDATA,KXX,1,,29.87
MPDATA,KXX,1,,31.09
MPDATA,KXX,1,,40.24
MPDATA,KXX,1,,40.85
MPTEMP,,,,,,,,
MPTEMP,1,20
MPTEMP,2,90
MPTEMP,3,200
MPTEMP,4,314
MPTEMP,5,423.5
MPTEMP,6,533
MPTEMP,7,643
MPTEMP,8,753
MPTEMP,9,863
MPTEMP,10,972.5
MPTEMP,11,1090
MPTEMP,12,1216
MPTEMP,13,1321.5
MPTEMP,14,1423.5
MPTEMP,15,1500
MPDATA,C,1,,451.21
MPDATA,C,1,,487.8
MPDATA,C,1,,518.29
MPDATA,C,1,,560.97
MPDATA,C,1,,628.04
MPDATA,C,1,,713.41
MPDATA,C,1,,829.26
MPDATA,C,1,,920.73
MPDATA,C,1,,1051.21
MPDATA,C,1,,1158.53
MPDATA,C,1,,1262.19
MPDATA,C,1,,1384.14
MPDATA,C,1,,1500
MPDATA,C,1,,1603.6
MPDATA,C,1,,1670
MP,DENS,1,7850
SFA,116,1,CONV,10,20
*DIM,timevstemp,TABLE,9,1,1,time,temp
*SET,TIMEVSTEMP(0,1,1) , 20
*SET,TIMEVSTEMP(3,0,1) ,3
SAVE
FINISH
/SOL
ANTYPE,4
TRNOPT,FULL
LUMPM,0
NSUBST,990,990,990
OUTRES,ERASE
OUTRES,ALL,1
LNSRCH,1
TIME,990
DA,22,TEMP, %TIMEVSTEMP%


193
IC,all,TEMP,20,
/STATUS,SOLU
SOLVE

Thermal elastic plastic analysis
/title,thermal elastic plastic analysis
/PREP7
%Commands of the geometric model and
mesh equal to transient thermal
analysis%
TOFFST,273
MPTEMP,,,,,,,,
MPTEMP,1,20
MPTEMP,2,90
MPTEMP,3,200
MPTEMP,4,314
MPTEMP,5,423.5
MPTEMP,6,533
MPTEMP,7,643
MPTEMP,8,753
MPTEMP,9,863
MPTEMP,10,972.5
MPTEMP,11,1090
MPTEMP,12,1216
MPTEMP,13,1321.5
MPTEMP,14,1423.5
MPTEMP,15,1500
MPDATA,EX,1,,207e9
MPDATA,EX,1,,198e9
MPDATA,EX,1,,198e9
MPDATA,EX,1,,186e9
MPDATA,EX,1,,168e9
MPDATA,EX,1,,118e9
MPDATA,EX,1,,54e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,EX,1,,6e9
MPDATA,PRXY,1,,0.3
MPDATA,PRXY,1,,0.308
TB,BISO,1,6,2,
TBTEMP,20
TBDATA,,250e6,11.032e9,,,,
TBTEMP,90
TBDATA,,238.8e6,10.894e9,,,,
TBTEMP,423.5
TBDATA,,172.56e6,9.584e9,,,,
TBTEMP,643
TBDATA,,74.39e6,5.171e9,,,,
TBTEMP,753
TBDATA,,18.29e6,1.862e9,,,,
TBTEMP,1500
TBDATA,,9.75e6,0.069e9,,,,
MPTEMP,,,,,,,,
MPTEMP,1,20
MPTEMP,2,90
MPTEMP,3,200
MPTEMP,4,314
MPTEMP,5,423.5
MPTEMP,6,533
MPTEMP,7,643
MPTEMP,8,753
MPTEMP,9,863
MPTEMP,10,972.5
MPTEMP,11,1090
MPTEMP,12,1216
MPTEMP,13,1321.5
MPTEMP,14,1423.5
MPTEMP,15,1500
UIMP,1,REFT,,,20
MPDATA,ALPX,1,,11.7e-6
MPDATA,ALPX,1,,13e-6
MP,DENS,1,7850
/VIEW,1,1,1,1
/ANG,1
/REP,FAST
DA,82,SYMM
DA,85,SYMM
DA,90,SYMM
DA,95,SYMM
DA,120,SYMM
DA,125,SYMM
DA,22,SYMM
DA,19,SYMM
DA,24,SYMM
DA,28,SYMM
DA,32,SYMM
DA,36,SYMM
DA,40,SYMM
DA,44,SYMM
DA,48,SYMM
DA,52,SYMM
DA,56,SYMM
DA,60,SYMM
DA,142,SYMM
DA,64,SYMM
DA,112,SYMM
TREF,20,
FINISH
/SOL
LDREAD,TEMP,,,1, , 'transient thermal
analysis ','rth',' '
LSWRITE,1,
BFDELE,all,TEMP
LDREAD,TEMP,,,3, , 'transient thermal
LSWRITE,2,
BFDELE,all,TEMP
LDREAD,TEMP,,,10,, 'transient thermal
LSWRITE,3,
BFDELE,all,TEMP
LSWRITE,4,
BFDELE,all,TEMP
LSWRITE,5,
BFDELE,all,TEMP
LSWRITE,6,
BFDELE,all,TEMP
LSWRITE,7,
BFDELE,all,TEMP
LDREAD,TEMP,,,100,,'transient thermal
LSWRITE,8,
BFDELE,all,TEMP
LSWRITE,9,
BFDELE,all,TEMP
LSWRITE,10,
BFDELE,all,TEMP
LSWRITE,11,
BFDELE,all,TEMP
LSWRITE,12,
BFDELE,all,TEMP
LSWRITE,13,
BFDELE,all,TEMP
LSWRITE,14,
LSSOLVE,1,14,1,
FINISH
/SOLU


194

APPENDIX 6


195
Listing of commands of the numerical simulation of the welding sequence in an L-type structure (welding
sequence 1)

/title,transient thermal analysis WS1
/PREP7
k,1,0,0
k,2,0.1,0
k,3,0.227,0
k,4,0.23,0
k,5,0.25,0
k,6,0.2627,0
k,7,0.2827,0
k,8,0.2857,0
k,9,0.32,0
k,10,0.23,0.097
k,11,0.25,0.097
k,12,0.2627,0.097
k,13,0.2827,0.097
k,14,0,0.1
k,15,0.1,0.1
k,16,0.227,0.1
k,17,0.25,0.1
k,18,0.2627,0.1
k,19,0.2857,0.1
k,20,0.32,0.1
k,21,0.25,0.103
k,22,0.2627,0.103
k,23,0,0.227
k,24,0.1,0.227
k,25,0,0.23
k,26,0.097,0.23
k,27,0.25,0.24575736
k,28,0,0.25
k,29,0.097,0.25
k,30,0.1,0.25
k,31,0.103,0.25
k,32,0.24575736,0.25
k,33,0.25,0.25
k,34,0,0.2627
k,35,0.097,0.2627
k,36,0.1,0.2627
k,37,0.103,0.2627
k,38,0.25845736,0.2627
k,39,0.2627,0.2627
k,40,0,0.2827
k,41,0.097,0.2827
k,42,0,0.2857
k,43,0.1,0.2857
k,44,0,0.32
k,45,0.1,0.32
k,46,0.2627,0.25845736
l,1,2
l,2,3
l,3,4
l,4,5
l,5,6
l,6,7
l,7,8
l,8,9
l,1,14
l,2,15
l,3,16
l,4,10
l,5,11
l,6,12
l,7,13
l,8,19
l,9,20
l,16,10
l,10,11
l,11,12
l,12,13
l,13,19
l,14,15
l,15,16
l,16,17
l,17,18
l,18,19
l,19,20
l,14,23
l,15,24
l,17,21
l,18,22
l,21,22
l,21,27
l,22,46
l,23,24
l,23,25
l,24,26
l,25,26
l,25,28
l,26,29
l,24,30
l,27,33
l,27,46
l,28,29
l,29,30
l,30,31
l,31,32
l,32,33
l,33,39
l,46,39
l,28,34
l,29,35
l,30,36
l,31,37
l,32,38
l,34,35
l,35,36
l,36,37
l,37,38
l,38,39
l,34,40
l,35,41
l,36,43
l,40,41
l,40,42
l,41,43
l,42,43
l,42,44
l,43,45
l,44,45
l,11,17
l,12,18
lplot
al,1,10,23,9
al,2,11,24,10
al,3,12,18,11
al,4,13,19,12
al,5,14,20,13
al,6,15,21,14
al,7,16,22,15
al,8,17,28,16
al,19,72,25,18
al,20,73,26,72
al,21,22,27,73
al,23,30,36,29
al,26,32,33,31
al,33,35,44,34
al,36,38,39,37
al,39,41,45,40
al,44,51,50,43
al,45,53,57,52
al,46,54,58,53
al,47,55,59,54
al,48,56,60,55
al,49,50,61,56
al,57,63,65,62
al,58,64,67,63
al,65,67,68,66
al,68,70,71,69
al,38,42,46,41
aglue,all
ET,1,PLANE55
ET,2,SOLID70
lesize,5,,,6
lesize,50,,,6
lesize,52,,,6
lesize,72,,,3
lesize,73,,,3
lesize,31,,,3
lesize,32,,,3
lesize,43,,,3
lesize,51,,,3
lesize,49,,,3
lesize,61,,,3
lesize,46,,,3
lesize,58,,,3
lesize,47,,,3
lesize,59,,,3
lesize,25,,,6
lesize,4,,,6
lesize,27,,,6
lesize,6,,,6
lesize,42,,,6
lesize,40,,,6
lesize,62,,,6
lesize,64,,,6
lesize,66,,,3
lesize,37,,,3
lesize,3,,,3
lesize,7,,,3
lesize,9,,,20
lesize,17,,,20
lesize,1,,,20
lesize,71,,,20
lesize,28,,,3
lesize,8,,,3
lesize,70,,,3
lesize,69,,,3
lesize,2,,,10
lesize,24,,,10
lesize,29,,,10
lesize,30,,,10
lesize,2,,,10
lesize,24,,,10
lesize,29,,,10
lesize,30,,,10
lesize,34,,,30
lesize,35,,,30
lesize,48,,,30
lesize,60,,,30
MSHAPE,0,2D
MSHKEY,1
amesh,all
nummrg,all
%Thermal properties dependent on
temperature used in numerical welding
process of the I-type specimen to
tension%
TYPE,2
EXTOPT,ESIZE,2,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
MAT,1
VEXT,1, , ,0,0,-0.003175,,,,
VEXT,2, , ,0,0,-0.003175,,,,
VEXT,3, , ,0,0,-0.003175,,,,
VEXT,4, , ,0,0,-0.003175,,,,
VEXT,5, , ,0,0,-0.003175,,,,
VEXT,6, , ,0,0,-0.003175,,,,
VEXT,7, , ,0,0,-0.003175,,,,
VEXT,8, , ,0,0,-0.003175,,,,
VEXT,9, , ,0,0,-0.003175,,,,
VEXT,11, , ,0,0,-0.003175,,,,
VEXT,12, , ,0,0,-0.003175,,,,
VEXT,14, , ,0,0,-0.003175,,,,
VEXT,15, , ,0,0,-0.003175,,,,
VEXT,16, , ,0,0,-0.003175,,,,
VEXT,27, , ,0,0,-0.003175,,,,
VEXT,18, , ,0,0,-0.003175,,,,
VEXT,21, , ,0,0,-0.003175,,,,
VEXT,23, , ,0,0,-0.003175,,,,
VEXT,24, , ,0,0,-0.003175,,,,
VEXT,25, , ,0,0,-0.003175,,,,
VEXT,26, , ,0,0,-0.003175,,,,
VEXT,10, , ,0,0,-0.003175,,,,
VEXT,13, , ,0,0,-0.003175,,,,
VEXT,17, , ,0,0,-0.003175,,,,
VEXT,22, , ,0,0,-0.003175,,,,
VEXT,19, , ,0,0,-0.003175,,,,
VEXT,20, , ,0,0,-0.003175,,,,


196
eplot
SFA,1,1,CONV,10,20
SFA,2,1,CONV,10,20
SFA,3,1,CONV,10,20
SFA,4,1,CONV,10,20
SFA,5,1,CONV,10,20
SFA,6,1,CONV,10,20
SFA,7,1,CONV,10,20
SFA,8,1,CONV,10,20
SFA,9,1,CONV,10,20
SFA,11,1,CONV,10,20
SFA,12,1,CONV,10,20
SFA,14,1,CONV,10,20
SFA,15,1,CONV,10,20
SFA,16,1,CONV,10,20
SFA,27,1,CONV,10,20
SFA,18,1,CONV,10,20
SFA,21,1,CONV,10,20
SFA,23,1,CONV,10,20
SFA,25,1,CONV,10,20
SFA,24,1,CONV,10,20
SFA,26,1,CONV,10,20
SFA,83,1,CONV,10,20
SFA,108,1,CONV,10,20
SFA,128,1,CONV,10,20
SFA,123,1,CONV,10,20
SFA,118,1,CONV,10,20
SFA,113,1,CONV,10,20
SFA,103,1,CONV,10,20
SFA,98,1,CONV,10,20
SFA,93,1,CONV,10,20
SFA,88,1,CONV,10,20
SFA,78,1,CONV,10,20
SFA,28,1,CONV,10,20
SFA,33,1,CONV,10,20
SFA,38,1,CONV,10,20
SFA,43,1,CONV,10,20
SFA,48,1,CONV,10,20
SFA,53,1,CONV,10,20
SFA,58,1,CONV,10,20
SFA,63,1,CONV,10,20
SFA,68,1,CONV,10,20
SFA,73,1,CONV,10,20
*DIM,TIMEVSTEMP01,TABLE,9,1,1,TIM
E,TEMP,
*SET,TIMEVSTEMP01(0,1,1) , 20
E,TEMP,
E,TEMP,
FINISH
/SOL
ANTYPE,4
TRNOPT,FULL
LUMPM,0
NSUBST,990,990,990
OUTRES,ERASE
OUTRES,ALL,1
LNSRCH,1
TIME,990
DA,51,TEMP, %TIMEVSTEMP01%
IC,all,TEMP,20,
/STATUS,SOLU
SOLVE
/title,thermal elastic plastic WS1
/PREP7
%Commands of the geometric model
and mesh equal to transient thermal
analysis%
%Mechanical properties dependent on
tension%
FINISH
/SOL
D,1,all,0
D,3,all,0
D,62,all,0
D,2595,all,0
D,2656,all,0
D,2597,all,0
D,3036,all,0
D,3040,all,0
D,3059,all,0
TREF,20,
%Application of the load thermal from
transient thermal analysis, 15 load step
are required%
LDREAD,TEMP,,,2, ,'analisis transiente
termico S1','rth',' '
LSWRITE,1,
BFDELE,all,TEMP
LSSOLVE,1,15,1,
FINISH


197

APPENDIX 7


198
Listing of the commands of the numerical simulation of the stiffened symmetrical flat frame (welding
sequence 5 O-I WT)

S5 O-I WT
/PREP7
k,1,0,0
k,2,0.004243,0
k,3,0.053175,0
k,4,0.08365,0
k,5,0.08865,0
k,6,0.09365,0
k,7,0.10635,0
k,8,0.11135,0
k,9,0.11635,0
k,10,0.146825,0
k,11,0.195757,0
k,12,0.2,0
k,13,0.08865,0.0097
k,14,0.09365,0.0097
k,15,0.10635,0.0097
k,16,0.11135,0.0097
k,17,0.0127,0.0127
k,18,0.016943,0.0127
k,19,0.053175,0.0127
k,20,0.08365,0.0127
k,21,0.09365,0.0127
k,22,0.10635,0.0127
k,23,0.11635,0.0127
k,24,0.146825,0.0127
k,25,0.183057,0.0127
k,26,0.1873,0.0127
k,27,0.0127,0.016943
k,28,0.09365,0.0157
k,29,0.10635,0.0157
k,30,0.1873,0.016943
k,31,0,0.053175
k,32,0.0127,0.053175
k,33,0.09365,0.053175
k,34,0.10635,0.053175
k,35,0.1873,0.053175
k,36,0.2,0.053175
k,37,0,0.08365
k,38,0.0127,0.08365
k,39,0.1873,0.08365
k,40,0.20,0.08365
k,41,0,0.08865
k,42,0.0097,0.08865
k,43,0.1903,0.08865
k,44,0.2,0.08865
k,45,0.09365,0.09065
k,46,0.10635,0.09065
k,47,0,0.09365
k,48,0.0097,0.09365
k,49,0.0127,0.09365
k,50,0.0157,0.09365
k,51,0.0227,0.09365
k,52,0.0537175,0.09365
k,53,0.08365,0.09365
k,54,0.09365,0.09365
k,55,0.10635,0.09365
k,56,0.11635,0.09365
k,57,0.146825,0.09365
k,58,0.1773,0.09365
k,59,0.1843,0.09365
k,60,0.1873,0.09365
k,61,0.1903,0.09365
k,62,0.2,0.09365
k,63,0.08865,0.09665
k,64,0.09365,0.09665
k,65,0.10635,0.09665
k,66,0.11135,0.09665
k,67,0.08865,0.10335
k,68,0.09365,0.10335
k,69,0.10635,0.10335
k,70,0.11135,0.10335
k,71,0,0.10635
k,72,0.0097,0.10635
k,73,0.0127,0.10635
k,74,0.0157,0.10635
k,75,0.0227,0.10635
k,76,0.0537175,0.10635
k,77,0.08365,0.10635
k,78,0.09365,0.10635
k,79,0.10635,0.10635
k,80,0.11635,0.10635
k,81,0.146825,0.10635
k,82,0.1773,0.10635
k,83,0.1843,0.10635
k,84,0.1873,0.10635
k,85,0.1903,0.10635
k,86,0.2,0.10635
k,87,0.09365,0.10935
k,88,0.10635,0.10935
k,89,0,0.11135
k,90,0.0097,0.11135
k,91,0.1903,0.11135
k,92,0.2,0.11135
k,93,0,0.11635
k,94,0.0127,0.11635
k,95,0.1873,0.11635
k,96,0.2,0.11635
k,97,0,0.146825
k,98,0.0127,0.146825
k,99,0.09365,0.146825
k,100,0.10635,0.146825
k,101,0.1873,0.146825
k,102,0.2,0.146825
k,103,0.0127,0.183057
k,104,0.1873,0.183057
k,105,0.09365,0.1843
k,106,0.10635,0.1843
k,107,0.0127,0.1873
k,108,0.016943,0.1873
k,109,0.053175,0.1873
k,110,0.08365,0.1873
k,111,0.09365,0.1873
k,112,0.10635,0.1873
k,113,0.11635,0.1873
k,114,0.146825,0.1873
k,115,0.183057,0.1873
k,116,0.1873,0.1873
k,117,0.08865,0.1903
k,118,0.09365,0.1903
k,119,0.10635,0.1903
k,120,0.11135,0.1903
k,121,0,0.2
k,122,0.004243,0.2
k,123,0.053175,0.2
k,124,0.08365,0.2
k,125,0.08865,0.2
k,126,0.09365,0.2
k,127,0.10635,0.2
k,128,0.11135,0.2
k,129,0.11635,0.2
k,130,0.146825,0.2
k,131,0.195757,0.2
k,132,0.2,0.2
k,133,0,0.004243
k,134,0.2,0.004243
k,135,0,0.195757
k,136,0.2,0.195757
l,1,2
l,2,3
l,3,4
l,4,5
l,5,6
l,6,7
l,7,8
l,8,9
l,9,10
l,10,11
l,11,12
l,1,133
l,1,17
l,2,18
l,3,19
l,4,20
l,5,13
l,6,14
l,7,15
l,8,16
l,9,23
l,10,24
l,11,25
l,12,26
l,12,134
l,133,27
l,134,30
l,13,20
l,13,14
l,14,15
l,15,16
l,16,23
l,17,18
l,18,19
l,19,20
l,20,21
l,21,22
l,22,23
l,23,24
l,24,25
l,25,26
l,17,27
l,21,28
l,22,29
l,26,30
l,133,31
l,27,32
l,28,33
l,29,34
l,30,35
l,134,36
l,31,32
l,33,34
l,35,36
l,31,37
l,32,38
l,33,45
l,34,46
l,35,39
l,36,40
l,37,38
l,45,46
l,39,40
l,37,41
l,38,42
l,38,49
l,39,60
l,39,43
l,40,44
l,41,47
l,41,42
l,42,48
l,45,54
l,46,55
l,43,61
l,43,44
l,44,62
l,47,48
l,48,49
l,49,50
l,50,51
l,51,52
l,52,53
l,53,54
l,54,55
l,55,56
l,56,57
l,57,58
l,58,59
l,59,60
l,60,61
l,61,62
l,47,71
l,48,72
l,49,73
l,50,74
l,51,75
l,52,76
l,53,77
l,63,67
l,64,68
l,65,69
l,66,70
l,56,80


199
l,57,81
l,58,82
l,59,83
l,60,84
l,61,85
l,62,86
l,63,64
l,64,65
l,65,66
l,53,63
l,56,66
l,67,77
l,67,68
l,68,78
l,68,69
l,69,79
l,69,70
l,70,80
l,54,64
l,55,65
l,71,72
l,72,73
l,73,74
l,74,75
l,75,76
l,76,77
l,77,78
l,78,79
l,79,80
l,80,81
l,81,82
l,82,83
l,83,84
l,84,85
l,85,86
l,71,89
l,72,90
l,73,94
l,78,87
l,79,88
l,84,95
l,85,91
l,86,92
l,89,93
l,89,90
l,90,94
l,91,95
l,91,92
l,92,96
l,93,94
l,87,88
l,95,96
l,93,97
l,94,98
l,87,99
l,88,100
l,95,101
l,96,102
l,97,98
l,99,100
l,101,102
l,97,135
l,98,103
l,99,105
l,100,106
l,101,104
l,102,136
l,135,103
l,103,107
l,107,108
l,108,109
l,109,110
l,110,111
l,105,111
l,105,106
l,106,112
l,112,113
l,113,114
l,114,115
l,115,116
l,116,104
l,104,136
l,135,121
l,121,107
l,122,108
l,109,123
l,110,124
l,117,125
l,110,117
l,117,118
l,118,111
l,111,112
l,112,119
l,118,119
l,118,126
l,119,127
l,120,128
l,119,120
l,120,113
l,113,129
l,114,130
l,115,131
l,116,132
l,132,136
l,121,122
l,122,123
l,123,124
l,124,125
l,125,126
l,126,127
l,127,128
l,128,129
l,129,130
l,130,131
l,131,132
l,14,21
l,15,22
l,28,29
lplot
al,1,14,33,13
al,2,15,34,14
al,3,16,35,15
al,4,17,28,16
al,5,18,29,17
al,29,220,36,28
al,6,19,30,18
al,30,221,37,220
al,7,20,31,19
al,8,21,32,20
al,31,32,38,221
al,9,22,39,21
al,10,23,40,22
al,11,24,41,23
al,24,25,27,45
al,27,51,54,50
al,54,60,63,59
al,63,69,76,68
al,76,77,92,75
al,68,75,91,67
al,92,110,139,109
al,91,109,138,108
al,139,147,152,146
al,152,153,156,151
al,138,146,151,145
al,156,162,165,161
al,165,171,186,170
al,186,208,207,185
al,207,219,206,184
al,183,206,218,205
al,182,205,217,204
al,181,203,202,197
al,203,204,216,201
al,202,201,215,200
al,198,200,214,199
al,196,197,198,195
al,177,195,194,193
al,194,199,213,192
al,193,192,212,191
al,176,191,211,190
al,175,190,210,189
al,174,189,209,188
al,172,173,188,187
al,163,167,172,166
al,154,158,163,157
al,149,150,154,148
al,125,141,149,140
al,126,142,150,141
al,78,94,125,93
al,79,95,126,94
al,71,72,78,70
al,61,65,71,64
al,65,66,79,72
al,52,56,61,55
al,26,47,52,46
al,13,42,26,12
al,80,96,127,95
al,81,97,128,96
al,82,98,129,97
al,83,99,130,98
al,84,123,111,114
al,114,100,116,99
al,111,101,117,100
al,117,118,131,116
al,85,124,112,123
al,112,102,119,101
al,119,120,132,118
al,86,115,113,124
al,113,103,121,102
al,121,122,133,120
al,115,104,122,103
al,87,105,134,104
al,88,106,135,105
al,89,107,136,106
al,90,108,137,107
al,132,144,155,143
al,155,160,164,159
al,164,169,179,168
al,179,180,196,178
al,62,74,85,73
al,53,58,62,57
al,222,49,53,48
al,37,44,222,43
aglue,all
%Thermal properties %
ET,1,PLANE55
/REPLOT,RESIZE
TYPE, 1
MAT, 1
REAL,
ESYS, 0
SECNUM,
FLST,5,116,4,ORDE,59
FITEM,5,2
FITEM,5,-3
FITEM,5,6
FITEM,5,9
FITEM,5,-10
FITEM,5,14
FITEM,5,-23
FITEM,5,26
FITEM,5,-27
FITEM,5,30
FITEM,5,34
FITEM,5,-35
FITEM,5,39
FITEM,5,-40
FITEM,5,46
FITEM,5,-63
FITEM,5,71
FITEM,5,76
FITEM,5,78
FITEM,5,82
FITEM,5,-83
FITEM,5,87
FITEM,5,-88
FITEM,5,92
FITEM,5,-94
FITEM,5,96
FITEM,5,-107
FITEM,5,109
FITEM,5,-110
FITEM,5,112
FITEM,5,119
FITEM,5,125
FITEM,5,129
FITEM,5,-130
FITEM,5,134
FITEM,5,-135
FITEM,5,139
FITEM,5,149
FITEM,5,152
FITEM,5,154
FITEM,5,-172
FITEM,5,175
FITEM,5,-176
FITEM,5,179
FITEM,5,182
FITEM,5,-183
FITEM,5,186
FITEM,5,189


200
FITEM,5,-192
FITEM,5,198
FITEM,5,-201
FITEM,5,204
FITEM,5,-206
FITEM,5,210
FITEM,5,-211
FITEM,5,214
FITEM,5,217
FITEM,5,-218
FITEM,5,222
CM,_Y,LINE
LSEL, , , ,P51X
CM,_Y1,LINE
CMSEL,,_Y
LESIZE,_Y1, , ,4, , , , ,1
CLRMSHLN
FLST,5,32,4,ORDE,30
FITEM,5,4
FITEM,5,8
FITEM,5,28
FITEM,5,32
FITEM,5,64
FITEM,5,-65
FITEM,5,68
FITEM,5,-69
FITEM,5,79
FITEM,5,91
FITEM,5,114
FITEM,5,-116
FITEM,5,118
FITEM,5,120
FITEM,5,122
FITEM,5,-124
FITEM,5,126
FITEM,5,138
FITEM,5,148
FITEM,5,150
FITEM,5,-151
FITEM,5,153
FITEM,5,193
FITEM,5,195
FITEM,5,197
FITEM,5,203
FITEM,5,212
FITEM,5,216
FITEM,5,220
FITEM,5,-221
CM,_Y,LINE
LSEL, , , ,P51X
CM,_Y1,LINE
CMSEL,,_Y
LESIZE,_Y1, , ,2, , , , ,1
CLRMSHLN
TYPE, 1
MAT, 1
REAL,
ESYS, 0
SECNUM,
MSHAPE,0,2D
MSHKEY,1
FLST,5,63,5,ORDE,25
FITEM,5,2
FITEM,5,-7
FITEM,5,9
FITEM,5,-13
FITEM,5,16
FITEM,5,-21
FITEM,5,23
FITEM,5,-27
FITEM,5,30
FITEM,5,-35
FITEM,5,37
FITEM,5,-41
FITEM,5,44
FITEM,5,-49
FITEM,5,51
FITEM,5,-55
FITEM,5,58
FITEM,5,-64
FITEM,5,66
FITEM,5,68
FITEM,5,-74
FITEM,5,77
FITEM,5,-78
FITEM,5,81
FITEM,5,-82
CM,_Y,AREA
ASEL, , , ,P51X
CM,_Y1,AREA
CHKMSH,'AREA'
CMSEL,S,_Y
AMESH,_Y1
CMDELE,_Y
CMDELE,_Y1
CMDELE,_Y2
TYPE, 1
MAT, 2
REAL,
ESYS, 0
SECNUM,
FLST,5,4,4,ORDE,4
FITEM,5,13
FITEM,5,24
FITEM,5,188
FITEM,5,207
CM,_Y,LINE
LSEL, , , ,P51X
CM,_Y1,LINE
CMSEL,,_Y
LESIZE,_Y1, , ,4, , , , ,1
CLRMSHLN
FLST,5,28,4,ORDE,25
FITEM,5,1
FITEM,5,11
FITEM,5,-12
FITEM,5,25
FITEM,5,33
FITEM,5,41
FITEM,5,-45
FITEM,5,73
FITEM,5,-74
FITEM,5,80
FITEM,5,90
FITEM,5,127
FITEM,5,137
FITEM,5,143
FITEM,5,-144
FITEM,5,173
FITEM,5,-174
FITEM,5,178
FITEM,5,180
FITEM,5,184
FITEM,5,-185
FITEM,5,187
FITEM,5,208
FITEM,5,-209
FITEM,5,219
CM,_Y,LINE
LSEL, , , ,P51X
CM,_Y1,LINE
CMSEL,,_Y
LESIZE,_Y1, , ,2, , , , ,1
CLRMSHLN
TYPE, 1
MAT, 2
REAL,
ESYS, 0
SECNUM,
FLST,5,20,5,ORDE,20
FITEM,5,1
FITEM,5,8
FITEM,5,14
FITEM,5,-15
FITEM,5,22
FITEM,5,28
FITEM,5,-29
FITEM,5,36
FITEM,5,42
FITEM,5,-43
FITEM,5,50
FITEM,5,56
FITEM,5,-57
FITEM,5,65
FITEM,5,67
FITEM,5,75
FITEM,5,-76
FITEM,5,79
FITEM,5,-80
FITEM,5,83
CM,_Y,AREA
ASEL, , , ,P51X
CM,_Y1,AREA
CHKMSH,'AREA'
CMSEL,S,_Y
AMESH,_Y1
CMDELE,_Y
CMDELE,_Y1
CMDELE,_Y2
/DIST,1,0.924021086472,1
/REP,FAST
eplot
SAVE
/PNUM,KP,0
/PNUM,LINE,0
/PNUM,AREA,0
/PNUM,VOLU,0
/PNUM,NODE,0
/PNUM,TABN,0
/PNUM,SVAL,0
/NUMBER,1
/PNUM,MAT,1
/REPLOT
/UI,MESH,OFF
SFA,2,1,CONV,10,20
% Weld bead thermal cycle, from weld
bead No. 1 to weld bead No.10 is
written%
E,TEMP,
FINISH
/SOL
ANTYPE,4
TRNOPT,FULL
LUMPM,0
NSUBST,990,990,990
OUTRES,ERASE
OUTRES,ALL,1
LNSRCH,1
TIME,990
%Application of the weld bead thermal
cycle to all areas corresponding to
each weld bead, from weld bead No. 1
to weld bead No.10 %
DA,29,TEMP,%TIMEVSTEMP01%
IC,all,TEMP,20,
/STATUS,SOLU
SOLVE
/title,thermal elastic plastic S5 OI WT
/PREP7
%Commands of the geometric model
and mesh equal to transient thermal
analysis%
%Mechanical properties dependent on
tension%
D,778,all,0
D,784,all,0
TREF,20,
%Application of the load thermal from
transient thermal analysis, 33 load step
are required%
LDREAD,TEMP,,,2, ,'analisis transiente
termico S5','rth',' '
LSWRITE,1,
LSSOLVE,1,33,1,
FINISH


201

APPENDIX 8


202
Listing of the commands of the numerical simulation of the 3-dimensional unitary cell (proper welding
sequence to reduce distortion with 24 weld beads and welding tacks)

/PREP7
k,1,0,0
k,2,0.003,0
k,3,0.203,0
k,4,0.206,0
k,5,0,0.014
k,6,0.003,0.014
k,7,0.203,0.014
k,8,0.206,0.014
k,9,0,0.034
k,10,0.003,0.034
k,11,0.203,0.034
k,12,0.206,0.034
k,13,0,0.044
k,14,0.003,0.044
k,15,0.203,0.044
k,16,0.206,0.044
k,17,0.0045,0.0455
k,18,0.2015,0.0455
k,19,0,0.046
k,20,0.003,0.046
k,21,0.004,0.046
k,22,0.202,0.046
k,23,0.203,0.046
k,24,0.206,0.046
k,25,0,0.047
k,26,0.003,0.047
k,27,0.004,0.047
k,28,0.006,0.047
k,29,0.016,0.047
k,30,0.036,0.047
k,31,0.17,0.047
k,32,0.19,0.047
k,33,0.2,0.047
k,34,0.202,0.047
k,35,0.203,0.047
k,36,0.206,0.047
k,37,0,0.05
k,38,0.003,0.05
k,39,0.004,0.05
k,40,0.006,0.05
k,41,0.016,0.05
k,42,0.036,0.05
k,43,0.17,0.05
k,44,0.19,0.05
k,45,0.2,0.05
k,46,0.202,0.05
k,47,0.203,0.05
k,48,0.206,0.05
k,49,0,0.051
k,50,0.003,0.051
k,51,0.004,0.051
k,52,0.202,0.051
k,53,0.203,0.051
k,54,0.206,0.051
k,55,0.0045,0.0515
k,56,0.2015,0.0515
k,57,0,0.053
k,58,0.003,0.053
k,59,0.203,0.053
k,60,0.206,0.053
k,61,0,0.063
k,62,0.003,0.063
k,63,0.203,0.063
k,64,0.206,0.063
k,65,0,0.083
k,66,0.003,0.083
k,67,0.203,0.083
k,68,0.206,0.083
k,69,0,0.167
k,70,0.003,0.167
k,71,0.203,0.167
k,72,0.206,0.167
k,73,0,0.187
k,74,0.003,0.187
k,75,0.203,0.187
k,76,0.206,0.187
k,77,0,0.197
k,78,0.003,0.197
k,79,0.203,0.197
k,80,0.206,0.197
k,81,0.0045,0.1985
k,82,0.2015,0.1985
k,83,0,0.199
k,84,0.003,0.199
k,85,0.004,0.199
k,86,0.202,0.199
k,87,0.203,0.199
k,88,0.206,0.199
k,89,0,0.2
k,90,0.003,0.2
k,91,0.004,0.2
k,92,0.006,0.2
k,93,0.016,0.2
k,94,0.036,0.2
k,95,0.17,0.2
k,96,0.19,0.2
k,97,0.2,0.2
k,98,0.202,0.2
k,99,0.203,0.2
k,100,0.206,0.2
k,101,0,0.203
k,102,0.003,0.203
k,103,0.004,0.203
k,104,0.006,0.203
k,105,0.016,0.203
k,106,0.036,0.203
k,107,0.17,0.203
k,108,0.19,0.203
k,109,0.2,0.203
k,110,0.202,0.203
k,111,0.203,0.203
k,112,0.206,0.203
k,113,0,0.204
k,114,0.003,0.204
k,115,0.004,0.204
k,116,0.202,0.204
k,117,0.203,0.204
k,118,0.206,0.204
k,119,0.0045,0.2045
k,120,0.2015,0.2045
k,121,0,0.206
k,122,0.003,0.206
k,123,0.203,0.206
k,124,0.206,0.206
k,125,0,0.216
k,126,0.003,0.216
k,127,0.203,0.216
k,128,0.206,0.216
k,129,0,0.236
k,130,0.003,0.236
k,131,0.203,0.236
k,132,0.206,0.236
k,133,0,0.25
k,134,0.003,0.25
k,135,0.203,0.25
k,136,0.206,0.25
l,1,2
l,3,4
l,1,5
l,2,6
l,3,7
l,4,8
l,5,6
l,7,8
l,5,9
l,6,10
l,7,11
l,8,12
l,9,10
l,11,12
l,9,13
l,10,14
l,11,15
l,12,16
l,13,14
l,15,16
l,13,19
l,14,20
l,14,17
l,15,18
l,15,23
l,16,24
l,19,20
l,20,21
l,21,17
l,17,28
l,18,33
l,18,22
l,22,23
l,23,24
l,19,25
l,20,26
l,21,27
l,22,34
l,23,35
l,24,36
l,25,26
l,26,27
l,27,28
l,28,29
l,29,30
l,30,31
l,31,32
l,32,33
l,33,34
l,34,35
l,35,36
l,25,37
l,26,38
l,27,39
l,28,40
l,29,41
l,30,42
l,31,43
l,32,44
l,33,45
l,34,46
l,35,47
l,36,48
l,37,38
l,38,39
l,39,40
l,40,41
l,41,42
l,42,43
l,43,44
l,44,45
l,45,46
l,46,47
l,47,48
l,37,49
l,38,50
l,39,51
l,40,55
l,45,56
l,46,52
l,47,53
l,48,54
l,49,50
l,50,51
l,51,55
l,52,56
l,52,53
l,53,54
l,49,57
l,50,58
l,55,58
l,56,59
l,53,59
l,54,60
l,57,58
l,59,60
l,57,61
l,58,62
l,59,63
l,60,64
l,61,62


203
l,63,64
l,61,65
l,62,66
l,63,67
l,64,68
l,65,66
l,67,68
l,65,69
l,66,70
l,67,71
l,68,72
l,69,70
l,71,72
l,69,73
l,70,74
l,71,75
l,72,76
l,73,74
l,75,76
l,73,77
l,74,78
l,75,79
l,76,80
l,77,78
l,79,80
l,77,83
l,78,84
l,78,81
l,79,82
l,79,87
l,80,88
l,83,84
l,84,85
l,81,85
l,82,86
l,86,87
l,87,88
l,83,89
l,84,90
l,85,91
l,81,92
l,82,97
l,86,98
l,87,99
l,88,100
l,89,90
l,90,91
l,91,92
l,92,93
l,93,94
l,94,95
l,95,96
l,96,97
l,97,98
l,98,99
l,99,100
l,89,101
l,90,102
l,91,103
l,92,104
l,93,105
l,94,106
l,95,107
l,96,108
l,97,109
l,98,110
l,99,111
l,100,112
l,101,102
l,102,103
l,103,104
l,104,105
l,105,106
l,106,107
l,107,108
l,108,109
l,109,110
l,110,111
l,111,112
l,101,113
l,102,114
l,103,115
l,104,119
l,109,120
l,110,116
l,111,117
l,112,118
l,113,114
l,114,115
l,115,119
l,116,120
l,116,117
l,117,118
l,113,121
l,114,122
l,119,122
l,120,123
l,117,123
l,118,124
l,121,122
l,123,124
l,121,125
l,122,126
l,123,127
l,124,128
l,125,126
l,127,128
l,125,129
l,126,130
l,127,131
l,128,132
l,129,130
l,131,132
l,129,133
l,130,134
l,131,135
l,132,136
l,133,134
l,135,136
lplot
al,1,4,7,3
al,2,6,8,5
al,7,10,13,9
al,8,12,14,11
al,13,16,19,15
al,14,18,20,17
al,19,22,27,21
al,23,29,28,22
al,24,25,33,32
al,20,26,34,25
al,27,36,41,35
al,28,37,42,36
al,29,30,43,37
al,31,32,38,49
al,33,39,50,38
al,34,40,51,39
al,41,53,64,52
al,42,54,65,53
al,43,55,66,54
al,44,56,67,55
al,45,57,68,56
al,46,58,69,57
al,47,59,70,58
al,48,60,71,59
al,49,61,72,60
al,50,62,73,61
al,51,63,74,62
al,64,76,83,75
al,65,77,84,76
al,66,78,85,77
al,72,80,86,79
al,73,81,87,80
al,74,82,88,81
al,83,90,95,89
al,84,85,91,90
al,87,93,92,86
al,88,94,96,93
al,95,98,101,97
al,96,100,102,99
al,101,104,107,103
al,102,106,108,105
al,107,110,113,109
al,108,112,114,111
al,113,116,119,115
al,114,118,120,117
al,119,122,125,121
al,120,124,126,123
al,125,128,133,127
al,129,135,134,128
al,130,131,137,136
al,126,132,138,131
al,133,140,147,139
al,134,141,148,140
al,142,149,141,135
al,143,136,144,155
al,137,145,156,144
al,138,146,157,145
al,147,159,170,158
al,148,160,171,159
al,149,161,172,160
al,150,162,173,161
al,151,163,174,162
al,152,164,175,163
al,153,165,176,164
al,154,166,177,165
al,155,167,178,166
al,156,168,179,167
al,157,169,180,168
al,170,182,189,181
al,171,183,190,182
al,172,184,191,183
al,178,186,192,185
al,179,187,193,186
al,180,188,194,187
al,189,196,201,195
al,190,191,197,196
al,193,199,198,192
al,194,200,202,199
al,201,204,207,203
al,202,206,208,205
al,207,210,213,209
al,208,212,214,211
al,213,216,219,215
al,214,218,220,217
aplot
aglue,all
ET,1,PLANE55
ET,2,SOLID70
lesize,1,,,2
lesize,2,,,2
lesize,219,,,2
lesize,220,,,2
lesize,52,,,2
lesize,63,,,2
lesize,158,,,2
lesize,169,,,2
lesize,21,,,2
lesize,35,,,2
lesize,75,,,2
lesize,89,,,2
lesize,42,,,2
lesize,65,,,2
lesize,26,,,2
lesize,40,,,2
lesize,82,,,2
lesize,94,,,2
lesize,50,,,2
lesize,73,,,2
lesize,127,,,2
lesize,139,,,2
lesize,181,,,2
lesize,195,,,2
lesize,148,,,2
lesize,171,,,2
lesize,132,,,2
lesize,146,,,2
lesize,188,,,2
lesize,200,,,2
lesize,155,,,2
lesize,178,,,2
lesize,15,,,4
lesize,18,,,4
lesize,97,,,4
lesize,100,,,4
lesize,121,,,4
lesize,124,,,4
lesize,203,,,4
lesize,206,,,4
lesize,44,,,4
lesize,173,,,4
lesize,48,,,4
lesize,177,,,4
lesize,9,,,4
lesize,12,,,4
lesize,103,,,4
lesize,106,,,4
lesize,115,,,4
lesize,118,,,4
lesize,209,,,4
lesize,212,,,4


204
lesize,45,,,4
lesize,174,,,4
lesize,47,,,4
lesize,176,,,4
lesize,3,,,2
lesize,6,,,2
lesize,215,,,2
lesize,218,,,2
lesize,109,,,13
lesize,112,,,13
lesize,46,,,19
lesize,175,,,19
MSHAPE,0,2D
MSHKEY,1
amesh,all
TYPE,2
EXTOPT,ESIZE,8,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.023,,,,
nummrg,all
/VIEW,1,1,1,1
/ANG,1
/REP,FAST
eplot
ASEL,S,LOC,Z,-0.023
aplot
TYPE,2
EXTOPT,ESIZE,1,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.0024,,,,
ALLSEL,ALL
aplot
ASEL,S,LOC,Z,-0.0254
aplot
EXTOPT,ESIZE,8,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.023,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,1,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.0024,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,8,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.023,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,1,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.0024,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,8,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.023,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,1,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.0024,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,8,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.023,,,,
ALLSEL,ALL
aplot
aplot
EXTOPT,ESIZE,1,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.0024,,,,
ALLSEL,ALL
aplot
ASEL,S,LOC,Z,-0.127
aplot
EXTOPT,ESIZE,8,0,
EXTOPT,ACLEAR,1
EXTOPT,ATTR,0,0,0
VEXT,all, , ,0,0,-0.023,,,,
ALLSEL,ALL
aplot
eplot
nummrg,all
/VIEW,1,1,1,1
/ANG,1
/REP,FAST
eplot
%Thermall properties dependent
on temperature used in
numerical welding process of the
I-type specimen to tension%
SFA,386 AREA,1,CONV,10,20
% Weld bead thermal cycle, from
weld bead No. 1 to weld bead
No.24 is written%
*DIM,TIMEVSTEMP01,TABLE,9,1,
1,TIME,TEMP,
*SET,TIMEVSTEMP01(3,1,1) ,
1245
1245
1000
FINISH
/SOL
ANTYPE,4
NLGEOM,1
DELTIM,2,2,2
OUTRES,ERASE
OUTRES,ALL,1
TIME,1980
%Application of the weld bead
thermal cycle to all areas
corresponding to each weld

bead, from weld bead No. 1 to
weld bead No.24 %
DA,1483,TEMP,
%TIMEVSTEMP01%
DA,1501,TEMP,
%TIMEVSTEMP01%
DA,1502,TEMP,
%TIMEVSTEMP01%
DA,1506,TEMP,
%TIMEVSTEMP01%
DA,1505,TEMP,
%TIMEVSTEMP01%
DA,1486,TEMP,%TIMEVSTEMP01
%
IC,all,TEMP,20,
/STATUS,SOLU
SOLVE

/title,thermal elastic plastic
/PREP7
%Commands of the geometric
model and mesh equal to
transient thermal analysis%
%Mechanical properties
dependent on temperature used
in numerical welding process of
the I-type specimen to tension%
FINISH
/SOL
ANTYPE,0
ANTYPE,0
NLGEOM,1
NSUBST,10,20,5
PSTRES,1
D,166,all,0
D,167,all,0
D,2139,all,0
D,2146,all,0
TREF,20,
%Application of the load thermal
from transient thermal analysis,
53 load step are required%
LDREAD,TEMP,,,8, ,'analisis
transiente termico 1','rth',' '
LSWRITE,1,
BFDELE,all,TEMP
LSSOLVE,1,53,1,
FINISH


205

APPENDIX 9

Full 10088

Загружено:

Сведения о документе

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

Full 10088

Загружено:

Авторское право:

Доступные форматы

AGHUniversityofScienceandTechnology

Residualstresses Intheplane GTAW Theanalyticalmethodproposedtodeterminetheresidualstressesthrough

Distortion Intheplane Thereisno

Distortion Inthespace TIG Althoughasurrogatemodelthatcanbeusedtoreducethecomputationalexpense

and the flow rule is

, substituting these expressions into equation (54),

is the bulk modulus.

Вам также может понравиться