You are on page 1of 36


An Overview of Historical Research Issues in Computational Weld Mechanics



In recent years, with the development of powerful computing facilities, Finite Element (FE)
analysis methods have been applied to the simulation of structural behavior using commercial FE
software packages. However, for general usage, especially in routine structural integrity assessments,
the simplified state-of-the-art methods are much more popular than full step-by-step elastic-plastic
analysis methods. The earliest welding technology has been traced as back as 1000 BC when forge
welding had been utilized into weapons. In Germany, first time the use of electric fusion process has
been reported in 1782 by G. Lichtenberg [1]. However, most of the references show commencement
of electric arc welding process in late nineteenth century. A brief history of the development in
welding processes is depicted in Figure 2.1 [2].

Figure 2.1 Brief histories of development phases in welding processes

Analytical determination of welding and its consequences (residual stresses and deformations)
find its roots in mid 1940's and significant contribution was made in 1950s especially for
deformations of welded structures, which ultimately lead to numerical modeling of heat distributions
and structural aspects of welding processes. Inherently, thermo-mechanical analysis of welding is
Discovery of electric arc by Sir Humphrey Devy
Wilde first intentionally joined metal by electric
welding in early 1860 and was granted a patent in
1865 for his work
De Meritens obtained patent of electric arc welding
process in England using carbon electrode
E. Thomson obtained a patent on resistance welding
Benardos, a Russian scientist, got first patent of
electric arc welding for slightly different equipment
then by De Meritens
Another Russian N. Slavianoff replaced
carbon electrode with a metal electrode and
obtained a patent on metal arc welding
Kjellberg, a Swedish, got a patent for coated
welding electrode
Development in joining process continued and
major welding process including
oxyacetylene, MMAW (manual metal arc
welding), GTAW (gas tungsten arc welding),
and GMAW (gas metal arc welding) processes
were successfully implemented
Advanced welding types such as Electron
Beam, Laser and Ultrasonic welding were
developed between 1950-1960
Most recent development is magnetic pulse
welding which was introduced
An Overview of Historical Research Issues in Computational Weld Mechanics

extremely non-linear due to multi-field interactions and non-linear thermal and structural material
response. In the past three decades, finite element analysis of complex welding phenomena is gaining
a considerable rise and a lot of research work has been reported in the published literature.

Published literature reveals significant computational work pertaining to computer simulation of
plate welding having single and/or multiple-pass welds. The main focus was on different
computational issues arising as a consequence of non linear transient heat flow during welding.
Most of the undesired consequences from welding are supposed to be due to the non-linear heat
flow introduced through a moving heat source (welding torch). Therefore, in this section, the
historical development in the heat source modeling is presented and efforts are made to encompass the
significant contributions in this regard. In 1960-1970, most of the research work on thermal-structural
behavior was based on experiments. Before 1970 there were only few series classical finite difference
solutions for the non-linear analysis of transient welding heat transfer. In late 1970's, serious efforts
were for the development of computer codes to analyze the complex mechanism of heat flow through
the weldments. A number of literature reviews on welding simulation research are reported in
literature. In the recent past, Lindgren compiled a detailed review in three parts [3-5]. Another recent
and comprehensive review was given by Dr. Anas Yaghi and Professor Adib Becker [6]. Practically,
several chapters are required to cover the entire work done in different fields of welding simulation.
However, main objective in this chapter would be description of major contributions and current
status of the research in this field specially in circumferential welding of axis-symmetric structures.
The first step towards the simulation of welding phenomenon was the model of moving heat
source presented by Rosenthal [7] for the analytical solution of transient temperature distribution in
arc welding. In his work, the author presented linear two-dimensional and three-dimensional heat flow
in solid of infinite size or bounded by planes. The authors also validated the model through
experimentally measured temperature distributions during plate welding of different geometries
(length, width and thickness). Subsequent researchers revealed that Rosenthal's model gave good
approximation of temperature only in the far field; however, in close proximity to the heat source the
predicted temperature field was a bit higher. Later on some other heat source models were developed
e.g. multiple point heat sources by Rybicki et al. [8] and Debicarri [9], were presented which gave
better approximation of transient temperature distribution. Further, Seo, Yang and Jang [10] used line
heat source is to model heat input from the welding torch. Goldak et al. [11] used a predefined
temperature at some specified locations. A most dominating heat source model with Gaussian heat
source distribution to overcome the issues in previously presented heat source model was given by
Goldak et al. [12, 13]. This model, commonly referred to as Double Ellipsoidal Heat Source model is
the most widely utilized heat source models now a days. Also, in this dissertation Goldak heat
distribution modeling scheme is employed, with some necessary adjustments as and when required.
An Overview of Historical Research Issues in Computational Weld Mechanics

Keeping in view the requirements, certain other heat source models such as presented by Sabapathy et
al. [14] and Ravichandran et al. [15] are also in practice. Ueda and Yamakawa [16] and Hibbitt and
Marcal [17] pioneered the application of finite element techniques in simulation of welding. The
complications and intricacies presented by these researchers, restrained the following researchers to
explore this complex phenomenon. Later, Friedman [18], Rybicki et al. [8] Andersson [19] showed
some interest in this domain and presented some research work which clarifies the methodologies
involved in welding simulations. They presented basic methodology for sequentially coupled analysis
technique. Temperature dependant material properties were used and latent heat associated with
liquid-solid phase transformation was also accounted for. Rybicki [8] simulated girth welding of pipes
by using planer and rotational symmetry. Further, he used an analytical solution of heat flow and
superimposed the effect of 28 point sources to match the temperature field. Both Friedman [18] and
Andersson [19] analyzed butt-welding of plates and used plane strain formulation with half sectional
model for thermo-mechanical analysis of welding. After the initial work, described above, there was a
deluge of simulation work in which finite element technique was implemented. Towards the
numerical aspect of simulation, numbers of experiments were made to evaluate the effect of modeling
techniques, mesh intensity, element types, modeling of filler materials, numerical Integration
procedure and type of solvers etc. Similarly lots of studies were also dedicated to material modeling in
which the effects of different material properties on thermal and structural response were studied.
Regarding the efficiency and integrity of the computational technique most significant contribution is
made by McDill et al. [20-24], who developed graded element for dynamic adaptive meshing.
Automatic mesh refinement was achieved only at the places of high thermal or stress gradients. This
Gradient dependant adaptive meshing scheme substantially reduced the computational expense by
decreasing the total number of elements in the model. Their scheme for adaptive re-meshing was
further improved by Runnemalm et al. [25, 26] and Lindgren et al. [27] and Hyun and Lindgren [28].
Mostly, significant contributions in the field of welding simulations before 2001 are critically
reviewed and discussed by Lindgren L. E. in three parts as referenced before in [3-5]. Further, the
contributions before 2004 were also discussed in [6]. The authors discussed the effects of different
parameters in detail and proposed recommendations for future work. Now making some efforts to
discuss the same will be a duplication of efforts. Here, in this dissertation the author will only discuss
the previous contributions in circumferential welding of pipes and cylinders. Few very useful
recommendations by the welding simulation legend, professor Lindgren are, however, presented
below because mostly the work presented here in this PhD dissertation is based on these expert
recommendations where applicable:
a. The degree of the finite element shape functions for the displacements should be one order
higher than the thermal analysis due to the fact that the temperature field directly becomes the
thermal strain in the mechanical analysis.
An Overview of Historical Research Issues in Computational Weld Mechanics

b. Usually the same elements are used in the thermal and mechanical analysis. For the
determination of transient heat distribution, transient and residual strain and stresses a fine
mesh with linear elements is preferable than fewer elements of higher order because linear
quad, in two dimensions, and brick, in three dimensions, elements are the basic
recommendation in plasticity.
c. Using a linear element requires that the average temperature should be used to compute a
constant thermal strain in the mechanical analysis. It is also important to under integrate the
volumetric strain (as it gives constant volumetric strain) when using linear elements in order
to avoid locking.
d. The axis-symmetric model corresponds to welding the whole circumference at once and gives
larger deformation in axial direction than a moving heat source.
e. Models where the arc is traveling in the mesh usually have larger elements and it is
recommended that one take smaller time steps than the length of the elements along the weld
divided by the welding speed. Shorter time steps are required at the start and finish of the
f. Regarding addition of filler material, quiet or inactive elements can give the same results if
implemented carefully.
g. There is no longer any reason to use the analytical solutions for heat distribution in the model
because the numerical simulation of the thermal field is quite straightforward.
h. If the region near the arc is studied in detail, the double ellipsoidal model for prescribing the
heat input may be better. This should then be combined with about ten elements across the
axis of ellipsoid area of heat input.
i. The microstructure evolution is important to include in the material modeling of ferritic steels
since their properties can change a lot due to the phase transformations. The major influence
is on the thermal dilatation and yield stress.
j. The computational model needs at least some kind of experimental results in order to
determine net heat input. Using thermocouple to measure temperature is straightforward. It is
also wise to do more experiments in order to verify the computational model before one uses
the model to study the effect of different changes in the design and/or the welding procedure.

An Overview of Historical Research Issues in Computational Weld Mechanics


Measurement of transient thermo-mechanical history during welding process is of critical
importance, but proves to be prohibitively expensive and time consuming. It often fails to provide a
complete picture of temperature and stress/strain/deformation distribution in the weldment. On the
other hand, detailed experimental measurements of the residual elastic strain distributions in welded
parts are typically not feasible due to significant resource (man, machine material) consumption.
Mathematical modeling for residual stress evaluation provides a resource effective method in
comparison to the experimental methods. However, development of the modeling scheme again
demands a careful experimental data.
Many stress analyses have been carried out to understand the residual stress distributions induced
by welding processes analytically and numerically [29]. The type of model used and the sophistication
of the analysis has often hinged on the accuracy required and the type of computational resource
available to solve the problem. For simple geometries like plate welding, several simplified analytical
expressions for the determination of distortion and residual stress fields are available. Many
researchers [30, 31] summarize such formulas empirically and analytically for the longitudinal and
transverse shrinkage and angular distortion for butt-welded plates. In the simplified analytical
methods, it is assumed that the welding residual stresses are determined by the local plastic shrinkage
strains generated during cooling after welding. Hence, although the weldment undergoes many
complex physical changes, involving interactions between micro-structural, thermal and mechanical
changes during welding processes, the plastic strains accumulated during the final stages of cooling
largely determine the residual distortion. This also means that the temperature and corresponding
strain history present during the early stages of welding can be neglected. This simplified method is
sometimes referred to as Inherent Strain Method [32]. Vaidyanathan, Todato and Finnie evaluated
residual stress in piping components by imposing on a cylinder the residual stress profile generated by
a similar weld in a plate through using thin-shell theory [33]. The applicability of this technique is
limited to thin-walled pipes with one weld pass. Ueda et al. [32] developed the Inherent Strain
Method, which uses a combined experimental and analytical approach to determine the source of
residual stress by utilizing the characteristics of the distribution of inherent strains induced in a long
welded joint. Hill and Nelson [34] further developed this method. The inherent strain model assumes
an axis-symmetric condition. It is incapable of predicting the transient residual stress distributions
near the weld start and stop location [35].
After the advent of finite element based numerical simulation techniques for weld modeling, the
analytical approaches were replaced by numerical approaches. By using FEM, it is possible to account
for nonlinear effects like temperature-dependent convection and radiation to the surrounding medium,
plastic flow and volume expansion during possible final phase transformation. The temperature
An Overview of Historical Research Issues in Computational Weld Mechanics

history in the structure during welding and subsequent cooling and the corresponding buildup of
stresses can thus be followed closely. The prevailing simulation technique of welding process is
computationally very expensive and thus demands high computational resources and large data
storage. Due to the limitation of available computational power in the past some simplification in the
FE model remained inevitable to reduce computational expense. In most of the studies regarding
circumferential welds, two-dimensional finite element models have been analyzed by using the
assumptions of lateral and rotational symmetry. The experimental, analytical and numerical work
mostly focusing on the prediction/analysis of arc welding induced imperfections like deformation and
residual stresses is briefly discussed below. Attempt is also made to maintain the chronological order,
so that the development phases can be well understandable.
Another analytical study by Vaidyanathan et al. [36], regarding determination of residual stresses
in circumferential welds is among the initial work on determination of residual stresses in welded
cylindrical objects. This study described the methodology for determination of residual stresses in
thin-walled cylindrical shells welded by single passes full penetration welds. Same material was
assumed for base metal and weld metal. Difference in the calculated and measured stresses for
welding of plate is attributed to the differential thermal expansion of base and filler metals. An
experimental validation through electron beam welding of Aluminum sphere was also performed and
good agreement was achieved. Later they extended their study further for variety of the welding
conditions, including multi-pass weld, partial penetration of weld and different materials for base and
filler metals.
Rybicki et al. [8] presented a finite element study for girth welding of pipes. They developed a
two-dimensional axis-symmetric finite element model for simulation of two pass weld on stainless
steel pipes. Analytically determined temperature distribution was used during the finite element
solution of thermal stress analysis. The effect of 28 temperature sources was superimposed to match
the calculated temperatures with experimental data. In addition, temperature dependant material
properties without latent heat and phase transformation effects were utilized. The filler metal is added
in the model by using quiet elements technique and activation temperature was taken 1150C. For
experimental verification of the resulting distortions and residual stresses, two pipe pieces were butt-
welded in two passes. They used insert ring at the root gap and a spider assembly to align these
segments before welding. Calculated and experimentally measured results showed good agreement
between experimental and calculated values of axial stresses at the inner surface was found.
Comparisons between other residual stresses did not show as good agreement. However calculated
radial displacements near the weld centerline were found within a band of measured values. Rybicki
and Stonesifer extended their work of multi-pass welding of pipes and presented a study [37]
containing seven-pass and thirty-pass welds on the similar material as used in the above described
study. Same simulation methodology i.e., analytical temperature calculation followed by a FE
structural analysis, was also employed. They demonstrated a strategy to minimize computational time
An Overview of Historical Research Issues in Computational Weld Mechanics

by using a combination of lumping of welds and temperature envelops. In order to use symmetry
across the weld centerline they lump seven passes into four and thirty passes into nine layers (nine
welds). Lumping strategy gave good approximation for axial residual stresses on inner surface for
seven pass weld, while qualitative agreement for hoop stresses could be achieved. However, the
results for thirty pass weld did not correspond to the experimental measurements. Subsequently,
Rybicki and Stonesifer [38] presented a finite element methodology for prediction of residual stresses
during repairing welding of thick-walled pipes (vessel). New development was to achieve three-
dimensional effect from two-dimensional model. For this purpose they calculated stiffness from the
three-dimensional analysis and added stiffness orthogonal in two-dimensional analysis.
Rybicki and McGuire [39, 40] presented some work on stress reduction in welded pipes.
Computational technique for small diameter pipe was presented and induction heating was modeled as
mode of stress reduction- A sensitivity analysis was also performed to evaluate effect of induction
heating conditions on reduction of residual stresses. A numerical study of multi-pass welding
regarding the effect of pipe wall thickness on welding residual stresses by Rybicki et al. [41] is of
significant importance for relating residual stresses with geometrical size of the pipe. Lumping
strategy as described in [37] was also implemented and welded passes were combined in the form of
layers. Analysis procedure implemented in this study was kept the same as that followed in [8, 37, 38]
It was a parametric study in which basic FE model was validated for residual stresses with the
experimental measurements and subsequently validated FE model was used for different welding
parameters and geometrical dimensions of the pipe. The authors concluded tensile axial residual
stresses on the inner surface of 4 inch diameter pipe and found higher tensile stresses for relatively
thin pipe. As the thickness increased tensile stresses near the weld centerline decreased and
compressive stresses away from the weld centerline increased. Another conclusion was that the zone
of influence of stresses increased with both the diameter and the wall thickness.
S. Nair et al. [42] presented a numerical scheme based on an in-house finite element code Creep-
Plast for determination of residual stresses in AISI-304 steel and SAE 1020 Steel. This numerical
scheme was only valid for 2-D axis-symmetric and plane solid structures. Following the procedure
define by Rybicki et al. [8] he also utilized analytical model of Vidanathan et al. [33, 36] for
temperature distribution and stresses were calculated by FE analysis. He compared his calculated data
with earlier numerical and experimental work of Rybicki and found that his scheme was less accurate.
Predicted values were found within 10% of values calculated by Rybicki. The authors claimed that
though the scheme is less accurate but was computationally easy to implement and less expensive.
They also extended their work to the post weld heat treatment (stress relieving) for SAE 1020 Steel
pipe. For heat treatment they included creep in their material model and creep data for the mentioned
material was taken from literature. The authors predicted more than 50% stress reduction for their
suggested annealing cycle.
An Overview of Historical Research Issues in Computational Weld Mechanics

Andrew Debiccari [9] performed analytical and experimental work in calculation of welding
distortions and residual stresses. The author devised a multiple point heat source model for analytical
determination of temperature distribution in the weldments and its surrounding. His analytical
approached was based on the work of Tsai. A reasonable agreement was claimed with the
experimentally measured values. Towards the experimental measurement of welding residual
distortions he used laser interferometery method. Measurement taken with laser interferometery
equipment showed good agreement with the measurements taken by using dial gage. In another
experiment, though the agreement between transient responses of laser interferometery device was not
in good agreement with the measurement of an advanced profilometer, however, this provided a new
avenue to explore. In addition, he used a turn buckle as constraint to reduce radial contraction of the
pipe. Unfortunately the methodology was not proved to be effective because of low strength and in-
sufficient constraint but the author was convinced about the positive results by using some improved
design of turn-buckle. Scaramangas [43] presented finite difference scheme for sequentially coupled
thermo-mechanical analysis of multi-pass girth welded pipes. Through thickness residual stresses
were investigated for three different wall thicknesses including 9.1, 15.0 and 19.5 mm. The predicted
results were validated with experimental measurements. For experimental measurement of through
thickness residual stresses contour method was used. Essentially several gauged patches from the test
specimen were initially removed in the form of panel through overlapping drilling. By using EDM
process small blocks were obtained from the panels and successive layers were removed. For
determination of near surface residual stresses air abrasion hole drilling method was implemented.
Good agreement between the theoretical results and experimental measurements were achieved. After
getting his numerical technique validated, the author further compared his results (for sizes which did
not validated through experiments) with the available published data. On the basis of results presented
in this work and from the comparison with the previous data, empirical relationships (polynomial
expressions) for surfaces axial and bending stresses were proposed.
Jonsson and Josefson [44] conducted an experimental study for the determination of transient
strain and residual stresses in 8.8 mm thick 200 mm nominal diameter pipe. In-order to get rid of the
complexities due to tack weld and root bead they made a "V" groove on a pipe to a depth of 5.5 mm
and material below this depth was assumed to be an initial weld. The basic assumption made for this
study was no initial effect of root gap and tack weld. They got their specimen stress relieved before
application of final weld for justification of employed assumption. Metal inert gas (MIG) welding
with lower heal input was used for the application of physical weld. Transient strains at predefined
locations ware measured for the whole welding cycle including welding and cooling time. Residual
stresses were measured at different locations by using hole-drilling technique and measured data was
compared with different analytical solutions. The authors were of the opinion that during the last pass,
stress should be greatly influenced by phase changes and transformation plasticity effect but
An Overview of Historical Research Issues in Computational Weld Mechanics

surprisingly measured values were found in good agreement with analytical solution of Vaidyanathan
et al. [33] in which the effect of transformation plasticity was not included.
Josefson et al. [45] presented a comparative study of simplified FE theory, two-dimensional
rotationally symmetric model, three-dimensional solid element model, three-dimensional shell
element model and experimental technique implemented to find transient and residual stresses in a
single-pass butt-welded pipe. The objective of this work was to reach on some conclusion by
comparing different solution techniques of the same problem. FE models were taken from the
respective past studies. From comparison it was inferred that residual stresses, as a hole, are not axis-
symmetric. However, residual hoop stress was found relatively less sensitive to circumferential
position and to low temperature solid-state phase transformations. It was further concluded that all the
models predicted axial residual stresses with reasonable accuracy but residual hoop stresses were
found difficult to match, especially near the weld centerline. Near the weld centerline stress reversal
was reported and attributed to excessive volumetric expansion due to solid state transformations.
Lindgren and Karlsson [46] analyzed the deformation and stresses for butt welded thin walled
pipes. Analytical solution of Rosenthal [7] for moving line heat source for thin plates was used with
the assumptions that radius of the pipe was much larger than the wall thickness and temperature
through the thickness was uniform. The model presented was referred as the first complete three-
dimensional models for deformation and stresses in welded thin-walled pipes by followed researchers
[2]. Shell element, based on the degenerated-solid concept, was used to create model geometry. This
technique was implemented on two cases: butt-welded plate and butt-welded pipes. Analytical
solution presented by Rosenthal [7] for moving line heat source in thin plates was opted for thin shells
with the assumption that radius to thickness ratio was very high and temperature across the thickness
was uniform. Temperature dependant material model was taken from Karlsson and Josefson [47]
which included phase transformation effects. Strain hardening was not included in the model.
Relatively better agreement between experimental data (taken from literature) and calculated axial
residual stress on outer surface was achieved. However, hoop stress profile at outer surface near the
weld centerline had complete disagreement with experimental measurement and was attributed to
possible inadequacies in the material model.
Gareth A. Taylor, Michael Hughes, Nadia Strusevich, and Koulis Pericleous [48] present the
computational modeling of welding phenomena within a versatile numerical framework. The
framework embraces models from both the fields of Computational Fluid Dynamics (CFD) and
Computational Solid Mechanics (CSM). With regard to the CFD modeling of the weld pool fluid
dynamics, heat transfer and phase change, cell-centered Finite Volume (FV) methods are employed.
Alternatively, novel vertex-based FV methods are employed with regard to the elasto-plastic
deformation associated with the CSM. The FV methods are included within an integrated modeling
framework, PHYSICA, which can be readily applied to unstructured meshes. The modeling
techniques are validated against a variety of reference solutions. Figure 2.2 shows the model
An Overview of Historical Research Issues in Computational Weld Mechanics

employed along with the plot of residual effective stress field. In Figure 2.3 the comparative transient
temperature distribution after time 11.5 sec (left) and surface temperature distribution of the pipe
(right) are shown from the same research.

Figure 2.2 Effective residual stresses in girth welding of thin pipe [48]

Figure 2.3 Transient temperature profiles in girth welding of thin pipe [48]

Karlsson et al. [49] presented numerical study for butt welding of pipes by using shell elements.
Half of the model was analyzed considering lateral symmetry. The solid model (left) and FE model
(right) are shown in Figure 2.4. Temperature dependant material properties and analytical model for
temperature distribution were taken from the literature. However, several other simplifications like
absence of root gap and tack welds and planer symmetry were utilized to reduce computational load.
Predicted radial shrinkage was compared with experimentally determined values from manually
welded pipe and reasonable agreement was achieved. The difference in measured and calculated
values was attributed to manual welding procedure. Predicted residual stresses on outer surface were
compared with literature results [44] and good agreement was achieved. The authors concluded that
An Overview of Historical Research Issues in Computational Weld Mechanics

2D model may give adequate results for residual stresses but 3D model is essential for transient and
residual strains.

Figure 2.4 Solid model (left) and FE model (right) used in [49]

Following Debiccari [9], Haoshi Song [50] in his Ph.D. work attempted to calculate and control
in-process distortions. His main work was related to implementation of dynamic meshing in plate
welding, however, a part of his work was related to numerical two-dimensional simulation of welding
process on a circular ring stiffened cylinder and three-dimensional simulation of small weld on the
same cylinder. To the best of author's information this was a first attempt for three-dimensional
modeling of circumferential welding using brick elements. Temperature dependant material properties
were used. Quiet elements technique was used to model filler material and reactivation of inactive
elements was done at melting temperature. In his 2D model root gap and clamping effects were also
accounted for. Sequentially coupled analysis procedure consisting of thermal analysis followed by a
structural analysis was followed. Predicted deformation in 2D analysis were found over estimated and
attributed to the absence of friction between cylinder and stiffener in actual welding and out of plane
elastic constraint in 3D model. The second reason perhaps is more important because the solidified
weld metal behind the welding arc poses constraint against further deformation and this effect is not
available in 2D model where the whole weld is assumed to be laid at once. For trial 3D simulation,
calculated deformations of stiffener had qualitative agreement with the experimental measurement but
were found on lower side. Finally the use of 3D simulation was recommended to predict out of plane
deformations of the stiffener ring. The conclusion presented by the author is very much relevant to the
target study in the present research because the geometry of flange also overhangs the outer diameter
of pipe just like the stiffener ring. Similarly out of plane deformation are anticipated during welding
and may have significant effect on the functional performance of the structure.
Karlsson and Josefson [47] presented first time a complete three-dimensional FE model with solid
elements for welding simulation of a 100 mm nominal diameter pipe-pipe joint with 8 mm wall
thickness. Symmetry along the weld centerline was however utilized to reduce the computational
An Overview of Historical Research Issues in Computational Weld Mechanics

expense. Material for pipe and filler material was low carbon steel SIS 2172 (Swedish Standard).
Temperature dependant material properties for this material were partly investigated by the authors
themselves and partly collected from the numerous resources. They also included thermal dilatation
during low temperature solid-solid phase transformation. These material properties were subsequently
used by various authors for simulation of similar materials. Single-pass butt-weld Joint geometry was
modeled for simulation of MIG welding. Finite element code ADINAT was used to model
sequentially coupled thermal stress analysis. In thermal model, heat was generated as consistent nodal
heat flow corresponding to volume of internal heat generation. The power intensity was assumed to
increase linearly in the axial direction and radially inwards. Transient and residual strain and stresses
were presented and results were compared with the past experimental, analytical and numerical
results. Hoop stresses were found almost axis-symmetric except the weld start position. Due to the
solid-solid phase transformations taken in the material model, values of hoop stresses in close
proximity to the weld centerline were found significantly different from the past studies. However,
away from the weld centerline results were quite comparable.
Troive and Jonsson [51] investigated welding deformations in multi-pass butt-welded pipe-flange
joint by using two-dimensional axis-symmetric FE model. The joint is made up of two weld passes of
manual metal arc welding in double-J weld geometry. Temperature dependant material properties for
AISI-316 were collected from different previous studies. Bilinear kinematic hardening was accounted
far in material model. For thermal model validation, experimentally measured peak temperatures at
two points on the inner surface were matched with the simulation. Addition of filler metal was
modeled by using quiet elements technique with suppression method. Suppression temperature was
set to 1100C, above which plastic strain was assumed not to change with temperature. All the nodes
of deactivated elements (during structural analysis) were kept without any displacement constraint to
avoid excessive shape distortion and minimize numerical non-convergence. During experiment the
pipe and the flange were initially tacked at four locations but 2D model can't accommodate tack weld
effects. Predicted deformations including diametrical deflection of pipe, diametrical deflection of
flange and deflection of flange in axial direction, were usually on higher side and only qualitative
agreement with the measurements could be achieved. Poor agreement with the experimental data was
attributed to the inherent in-capabilities of 2D model and manually performed welding process. The
use of automatic welding process (GTAW / GMAW) was suggested for further improvement of
results. Significant variations in the measured deflection for different angular positions indicate that
welding deformations in circumferential joints are not axis-symmetric. Troive et al. [52] investigated
the effect of welding distortions and residual stresses on load carrying capacity of the pipe. Three-
dimensional FE model using thermo-elasto-plastic shell elements was developed for thin-walled pipe
with the same material model as used in Karlsson et al. [47]. Material response was assumed to be
elastic-perfectly plastic and without solid-state transformations. Rosenthal's analytical solution [7] for
a heat source with uniform strength along a line through the thickness of a plate was used to find
An Overview of Historical Research Issues in Computational Weld Mechanics

transient temperature history. Subsequently, FE thermal stress analysis was performed to determine
welding distortions and residual stresses. Welded pipe was then brought in contact with a rigid wall
and buckling behavior was studied due to the axial force. They concluded that load carrying capacity
of a welded pipe is mainly affected by the welding residual distortions, whilst the effect of residual
stresses is relatively small. Similarly the effect is more pronounced for thin pipes. Troive et al. [53]
further extended their previous work [52] of pipe-flange welding. In this study, the same model
geometry is investigated for two different structural materials i.e. stainless steel 2352 and Swedish
steel 2142, and two weld pass sequences for each type of material. Geometrical size of pipe and
flange is slightly greater than the one used in [52]. A new dental-foam technique was employed for
experimental measurement of axial distortions in flange. This newly introduced technique was
cumbersome and relatively less accurate due to involvement of thermal contraction of the foam.
However, the results presented had qualitative agreement with the experimental data. The author
himself accepted un-realistic prediction of distortion and presented certain recommendations.
Michaleris [54] presented a finite element study for multi-pass welding of pressure vessel and
piping components. Two-dimensional models (axis-symmetric FE model for girth weld and plane
strain formulation for longitudinal seam) were used for the welding of thin pipes (t = 0.25 inch) and
thick pipes (t = 1.3 inch). Goldak's double ellipsoidal heat source model [12] was used for FE thermal
analysis. Inactive element approach was used to model filler material. All the elements corresponding
to the weld passes yet to be laid were removed from the model and model change option was used to
update FE model for each weld pass. The displacement difference between the current and previous
pass was then accommodated through the use of multi-point constraint. This ensured zero initial stress
on the elements of current, pass due to previously deposited weld passes, Elastic-plastic material was
assumed with kinematic work hardening white solid-state transformations were not included. The
computational methodology was validated by comparing temperature and stress distribution with
experimentally measured data and good agreement was concluded. Blind hole-drilling technique
(center hole-drilling) was employed to determine residual stresses experimentally. Different
geometrical and loading parameters like thickness, radius to thickness ratio, weld joint geometry;
hydro-test and end restraint effects were studied. The author concluded that pipe wall thickness has
negligible effect for an r/t of 10. However, for r/t ratio of 25 and 50 with the same joint type (double
"V"), the axial residual stress at the inner diameter is higher in thin-walled pipes. Also, the magnitude
of axial residual stress at the inner diameter reduces in thin pipes as the r/t ratio increases, whereas
little variation in stress is observed in thick pipes. The authors also mentioned that in girth welds, the
hydro-test reduces the axial (transverse) residual stress at the inner diameter. The reduction is higher
in thin pipes than in thick pipes. The hydro-test has negligible effect on the transverse residual stress
of seam welds. Axial residual stress in girth welds of short length thick pipes with restrained ends are
of yield magnitude through thickness. Some sample results from referred research are shown in Figure
An Overview of Historical Research Issues in Computational Weld Mechanics


Figure 2.5 Through-thickness axial residual stress on girth welds of single V thin pipes (left) and Through-
thickness axial residual stress on girth welds of double V thin pipes (right) [54]

Ravichandran et al. Raghupathy et al. and Ganesan et al. [15], numerically solved the thermal
effects of TIG welding process for thin-walled structures including thin-cylinders and thin-spherical
dish ends. Bilinear degenerated shell elements are adapted in the analysis. Temperature dependence of
thermo-physical material properties, distributed heat source and convective & radiative heat losses
were fully accounted in the modeling. Parametric results for thin pipe to pipe, thin pipe to thin
spherical dish end and thin spherical dish end to thin spherical dish end are presented in detail for
circumferential and axial thermal profiles in various sections. For results comparison the wall
thickness, diameters and heat input in all the cases was kept the same. The most significant
conclusions presented were that the thermal results for all the three cases presented were same near
the weld line during the heating phase with non tolerable difference during the cooling phase. The
temperature profiles for the spherical end sides were slightly higher than the cylindrical ends. A
representative results case is shown in Figure 2.6 below.

Figure 2.6 Comparison of thermal cycles during welding for cylinder to cylinder and cylinder to
sphere presented by [15]
An Overview of Historical Research Issues in Computational Weld Mechanics

Dike, Ortega, and Cadden [55] presents modeling and experimental validation of residual stresses
in 304L stainless steel girth welds in small diameters (38 mm outer diameter) pipes. 3D decoupled
thermo-mechanical simulation with axial symmetry was used for autogenous Gas Tungsten Arc
Welding (GTAW). Volumetric heat source with filler metal addition along with the convection and
radiation losses was considered for the prediction of residual stresses. Mechanical material behavior at
elevated temperature was modeled by using Bammann-Chiesa-Johnson (BCJ) constitutive model.
Thermocouples were used to capture the temperature profiles at 6 mm and 12 mm from the weld line
for the thermal model validation and X-ray diffraction method for the measurement of residual
stresses was employed for validation of mechanical response. The comparison of predicted and
measured axial and hoop residual stresses on outer surface of the pipes agreed closely. FE mesh used
in the referred research work and the comparison of residual hoop stress against the axial distance
from the weld line at an angle of 35
from the weld start position are shown in Figure 2.7 for reference

Figure 2.7 FE model (left) and comparative residual hoop stresses (right) from previous research [55]

Mohr et al. [56] developed correlation between welding parameters and anticipated welding
residual stresses at the inner surface of the pipe, based on the data collected from different analytical,
numerical and experimental studies reported in the literature. They demonstrated that residual stresses
at the inner surface are the function of heat input in each pass and number of passes or nominal wall
thickness. Concentrating on the field welding it was assumed that nearly all circumferential welds
have some proportions to be welded in vertical and overhead position and these position limits the
heat input to maximum value of 50 kJ/in (excessive heating may cause fall of weld pool).
Dong et al. [57] performed a three-dimensional finite element simulation for austenitic stainless
steel pipe by using shell elements. Their FE model was similar to the Karlsson et al. [49] and
complete welding was supposed to be done in three passes. Several assumptions for simplification of
the process including symmetry about the weld centerline, stress relieving of previous weld passes by
An Overview of Historical Research Issues in Computational Weld Mechanics

the last weld pass and significant effect of transverse shear stress caused by thick-shell type thermal
mechanical response on the residual stress distribution were made. Temperature dependent material
properties were taken from the past studies. Blind hole-drilling technique was used for experimental
measurement of residual stresses at inner and outer surfaces. Qualitative agreement between
calculated and measured values of residual axial and hoop stresses was found on the outer surface.
However calculated stresses could not match with the measured stresses at the inner surface. In
addition to the above described work a parametric study was also performed to predict the effect of
wall thickness and welding speed. Two wall thicknesses (9.52 mm and 19.04 mm) were evaluated for
the effect of wall thickness on stress distributions and results on the inner surface showed a similar
stress variation trend on as presented by Rybicki et al. [41]. In addition, three different welding speeds
(4, 8 and 16 mm/s) were used to determine the effect of welding speed and it was concluded that arc
travel speed has minor effect on the stress distribution.
The contribution from Chapman et al. [58] is a new welding technique, referred to as Fine Line
Welding (FLW). The major advantage from the technique was significant reduction in tensile residual
stresses at the inner surface of the weldment. Further, benefits of the proposed technique were very
small tensile residual stresses and small heat affected zone (HAZ) resulting in significantly reduced
chances for occurrence of stress corrosion cracking. The authors used X-ray diffraction technique to
experimentally measure the residual stress fields. Chemical etching was used to view the size of HAZ
and boiling Magnesium chloride test was used to check occurrence of trans-granular stress corrosion
cracking. The results of FLW were compared with ordinary GTAW welding, manual metal arc
welding and narrow gap welding and FLW was claimed advantageous.
Yang and Lee [59] presented an axis-symmetric FE model for mechanical stress relieving in thin-
walled pipes. Same material was assumed for base and weld metals and temperature dependant
material properties for SUS 304 material were taken from literature. Generally acceptable sequentially
coupled thermal stress analysis was performed first to determine residual stresses. Subsequently the
model was subjected to three different stress relieving loading namely internal pressure, external
pressure and axial pull in the quantitative range of 30~90% of material yield strength (according to
thin-walled cylinder theory). The stress relieving behavior was studied and internal pressure was
concluded best because of its effectiveness. The hoop residual stress after post weld stress relieving
on pipe inner surface with different MSR loads equal to 90% of yield stress from the same research is
shown in Figure 2.8.

An Overview of Historical Research Issues in Computational Weld Mechanics


Figure 2.8 Hoop (left) and axial (right) stress distribution at pipe inner surface after
various MSR loadings of 90% yield stress [59]

Brickstad and Josefson [60] executed a parametric study for the determination of residual stresses
in multi-pass girth-butt-welded stainless steel pipes. The main focus of the work was to investigate
residual stresses sensitivity to variation in weld parameters. The predicted stresses were compared
with the anticipated stress state in ASME Section-IX and recommendations were proposed. In order to
reduce computational time, two dimensional FE models were studied by employing rotational
symmetry assumption (without lateral symmetry). Different pipe diameters (76.2 mm to 680 mm)
with corresponding wall thickness (7.1 mm to 40.0 mm) were studied. Number of weld passes varied
from 4 to 36 depending on the pipe wall thickness. Temperature dependant material properties with
bilinear kinematic hardening and without solid-state transformation were taken from literature (from
specifications by steel manufacturers). Von-Mises yield criteria and associated flow rules were
considered. In the thermal model line heat energy was averaged over the elements included in the
bead and applied as heat flux. Time step duration was chosen on the basis of welding speed and
fraction of circumference included in weld pass volume. Combined heat transfer coefficient was
defined to accommodate both the convection and radiation heat losses from the surface. Quiet
elements technique was chosen for the addition of the filler metal. They implemented temperature
truncation technique and effect of temperature rise above a so-called softening temperature (TSOFT)
was ignored. All the predicted temperatures above TSOFT were truncated to reduce computational
time in structural analysis. Elements corresponding to the weld bead being laid were considered born
when they cooled down to TSOFT after peak temperature. Newton iterative solver with line search
option to improve convergence was opted. Residual stress variation in radial direction resulting from
different weld parameters were discussed in detail and compared with the literature results.
Teng and Chang [61] presented sequentially coupled thermal stress analysis for the determination
of residual stresses. They presented parametric studies based on axis-symmetric FE models to
An Overview of Historical Research Issues in Computational Weld Mechanics

demonstrate the effects of pipe diameter and wall thickness on residual stresses. Temperature
dependant material properties were used for elastic-perfectly plastic material model without solid state
phase transformations. Their results exhibited self balancing behavior of residual stresses. Both the
axial and hoop stresses were tensile on the inner surface near the weld centerline and stress reversal
occurred in the regions away from the centerline. Increase in wall thickness reduced residual stresses
at the weld centerline and the author's finding was somewhat in agreement with Rybicki et al. [41].
However, the die out distance observed for both the axial and hoop stresses is larger than one found in
[41]. Increase in pipe diameter was reported for increase in zone of influence of residual stresses.
From the results presented it can be concluded that their accurate evaluation can help resolve
problems such as inter granular stress corrosion cracking (IGSCC) which has been observed in
the weld fusion lines or nearby, on the inside surfaces of the pipes of boiling water reactors. Teng
and Chang [62] enhanced their contribution [61] to three-dimensional shell element models and
analyze welding residual stresses on the same sized pipes to investigate the effect of pipe wall
thickness. Same material properties, simulation strategy and most of the welding parameters were
used. Only the pipe scheduled was changed and consequently welding speed is altered according to
the changed wall thickness of the pipes. Only two wall thicknesses i.e. 2.8 mm (schedule 5) and 3.8
mm (schedule 10) were studied. The contour plot of axial and hoop stresses are shown in Figure 2.9.
During the PhD studies most of the work of Henrik Runnemalm [63] was related to efficient finite
element modeling which included incorporation of graded elements for adaptive re-meshing.
However, a part of his work was related to the simulation of laser welding in AISI 304 stainless steel
pipe. He used a two-dimensional axis-symmetric model for simulation and compared the calculated
and measured elastic strains. Experimental elastic strain was measured by using neutron
diffractometer. Relatively better agreement was found for axial and normal strains but in the case of
hoop strains agreement was not good enough and the author suggested the use of three dimensional
models for realistic results. Dong and Zhang [64] investigated the effect of strength mismatch welds
on residual stress field and fracture behavior. Their work was pertaining to plate welding as well as
thick-walled (1.5 inch) pipe welding, however only circumferential welding part is included here. The
authors used temperature independent material properties with bilinear kinematic hardening. Two-
dimensional axis-symmetric FE model was used to model multi-pass girth welding of 304 austenitic
stainless steel pipes. Zhang et al. [65] presented a combined study of multi-pass welding simulation,
effect of specimen cutting from welded part and effect of welding residual stresses on crack
propagation. The finite element technique for each of these processes was presented with its
application on welding of BWR (Boiling water reactor) core shroud. Two dimensional axis-symmetric
FE model without lateral symmetry is employed to reduce computational expenses.
Basavaraju [66] presented a simplified analysis of shrinkage in butt-welded pipes by using two-
dimensional axis-symmetric models. Simple elastic-plastic analysis for the determination of weld
An Overview of Historical Research Issues in Computational Weld Mechanics

induced residual stresses and strains. Through analytical modeling elastic and plastic strains followed
residual stresses based on experimentally determined radial shrinkage at weld centerline from butt
welded pipes were predicted. Results revealed that elastic stresses greater than the yield strength of
the material are produced in elastic analysis approach and elastic-plastic analysis was considered to
produce better results. To incorporate the plasticity, bilinear approximations was used, the available
expressions for stress and strain formulation for axis-symmetric concentrated load on a thin-walled
cylinder were adjusted and materials Poisson ratio was artificially enhanced to a value of 0.5 from 0.3.
Finally, the results were compared with available published data with good agreement. The calculated
plastic strains were compared with published data and good agreement was reported.

Figure 2.9 Contour Plots of Axial and Hoop Stress from [62]

Engelhard et al. [67] experimentally evaluated various techniques for reduction of tensile residual
stresses at the inner surface of the pipes. First part in the work was related to in-process control of
residual stresses by using different welding techniques. The authors compared the results of
conventional and optimized GTA (gas tungsten arc) narrow gap welding. They concluded that the
width of fusion zone can be reduced to about 45% with optimized narrow gap welding technique. This
was found advantageous because of reduced axial shrinkage, less sensitized zone and relatively less
tensile residual stresses. Experimental determination of residual stresses was carried out by the ring
core method and XRD method. Subsequently, the effect of heat sink welding (HSW) either by using
water or CO
was also studied and recommended it a potential mean for converting tensile residual
stresses to compressive at the inner surface. As a post-weld stress mitigation technique, LPHSW (last
pass heat sink welding) was applied on a 4 inch schedule 40 pipe and it was concluded that tensile
residual stresses could be converted to compressive residual stresses of nearly equal magnitude.
Mochizuki et al. [68] compared residual stresses in welded pipes measured by experimental and
finite element techniques. Two geometries i.e. pipe-pipe weld and pipe-socket weld were investigated.
Two numerical techniques including inherent strain and thermal elastic plastic analysis were
An Overview of Historical Research Issues in Computational Weld Mechanics

implemented for the prediction of residual stresses. For application of inherent strain method, strain
gages were pasted on the welded geometry in a specific order. Comparison between different
analytical/numerical technique and experimental measurement both for internal stresses and surface
stresses showed good agreement. In most of the cases inherent strain method was found in better
agreement as compared to elastic plastic analysis. It was however, concluded that selection of suitable
method depends on purpose of measurement, evaluating location, and the operating condition of the
welded object to be evaluated.
Hedblom and Lindgren [69] presented comparative study of two different welding procedures,
including U-seam and OPTI-GAP, and their effect on the residual stress distribution profile with the
objective to reduce susceptibility of inter-granular stress corrosion cracking. Two-dimensional FE
model (axis-symmetric) was used with temperature dependent material properties taken from
literature. Gradient dependant automatic mesh refinement technique based on graded element was
incorporated. Quiet elements technique with a special procedure to readjust element shape at the time
of element activation during multi-pass welding was utilized for addition of filler material. Prescribed
temperature description in the weld puddle was implemented for the calculation of temperature
history. The predicted stresses were compared with the experimental results, reported in Hedblom
[70], carried for the target geometries and welding procedures. A reasonable agreement was achieved
between simulation and experiments. From the results it was concluded that OPTI-OAP produced
relative low tensile stresses in sensitive areas for thin-walled pipes and thus reduced the risk of
IOSCC. However, the effect of welding procedure was not found very decisive for thick pipes.
The study of numerical models for in-service welding of gas pipelines was presented by
Sabapathy et al. [14]. Efforts had been made to simulate a vertical down manual metal arc welding. A
three-dimensional FE model for a circumferential segment of complete geometry was developed by
using 8-node linear brick, elements. Temperature-based material properties for low strength steel were
considered without considering microstructure effects. In order to incorporate weave effect in heat
distribution double ellipsoidal heat source by Goldak et al. [12] was modified to a disk like heat
source. Forced convection at the inner surface of the pipe was considered and effective heat transfer
was calculated analytically. The predicted T
weld cooling times showed good agreement with the
experimental data. Effective reduction in thickness approach based on effective yield strength of a
section of pipe at elevated temperature was developed to predict bum through.
Frieke et al. [71] presented a three-dimensional finite element simulation of sequentially coupled
thermo-mechanical analysis of multi-pass welding of pipes. The study was aimed at the investigations
of weld induced residual stresses and three different techniques for the mitigation of theses residual
stresses. Out of these three techniques, two techniques application of narrow gap welding and last
pass heat sink welding were based on in-process control techniques and the remaining in-service
aging technique was based on the post weld stress mitigation technique. Temperature dependency of
material by using 8-node linear brick elements was incorporated. Results of simulation were
An Overview of Historical Research Issues in Computational Weld Mechanics

compared with previous experimental data and in general good agreement at discrete locations was
claimed. However, distribution of axial stresses along the circumferential direction did not exhibit
even qualitative agreement with the experimental results. On the basis of axial stress variation in the
circumferential direction (in both numerically predicted and experimentally determined) it was
concluded that residual stresses are by no means axis-symmetric. It was further revealed that tensile
axial residual stresses at the inner surface change drastically near the weld start and end locations as
shown in Figure 2.10.

Figure 2.10 Axial Residual Stresses at the inner side (0.2 mm from fusion line) after cooling to inter-pass
temperature [71]

Wen, Hilton and Farrugia [72] presented 2D and 3D multi-pass simulation of linearly seam
welded pipes joined by submerged arc welded method. Primary focus was to study the transient
temperature gradients in the fusion and heat affected zones. Additionally, the effects of various
welding and geometric parameters were also studies. The effects of residual stress fields due to
previously roll forming was not considered. Bang et al. [73] studied the sleeve repair welding on in-
service gas pipelines numerically. Circumferential welding of a sleeve on an existing pipe was
conducted to study cold cracking during welding. In this study a 2D model with planner and rotational
symmetry was used to represent a multi-pass fillet weld. Analytical approach was used to predict
maximum HAZ hardness to predict the occurrence of cold cracking.
Xiangyang [74] in his PhD thesis on titled Influence of Residual Stress on Fatigue Failure of
Welded Joints makes an effort to understand the influence of residual stresses on the fatigue failure of
butt and socket-welded piping joints. The study develops experimental fatigue failure data for butt and
socket welded pipe to flange joints and a numerical scheme for the analysis of these data. A set of
low-cycle fatigue tests of welded piping joints in the cantilever setup is conducted. A new observation
made in these tests is that the recorded strains near the weld toe ratchet continuously, which results in
the initiation of fatigue crack(s). Comparison of these ratcheting responses with those from the cyclic
pipe bending and material ratcheting experiments indicates that the residual stresses at the welded
An Overview of Historical Research Issues in Computational Weld Mechanics

joints may not relax to zero after a few inelastic cycles, as assumed in the fatigue design methods.
This observation is further supported by the symmetric strain response (no ratcheting) at the mid-pipe
length, which is located away from the welded joint. At this location, there are no residual stresses to
induce ratcheting. It is also observed that the fatigue cracks in all experiments occurred at the weld toe
location where the ratcheting strain is the largest. This indicates that the fatigue life of materials is
reduced in the presence of ratcheting. The recorded load responses demonstrate cyclic softening for
stainless steel, which is known as a cyclically hardening material. The research work presents a
detailed single and multi-pass butt and socket welded pipe-flange structures, showed good co-relation
between experimental and simulated results. Additionally some work pertaining to relieving of
residual stresses by the furnace stress relief heat treatment process is also presented. Some results of
interest from this research in the present thesis are shown in Figure 2.11.

Figure 2.11 Axial residual stresses (in Pa) distribution in four-pass girth weld (left) and Axial and Hoop
Residual Stresses in Circumferential Direction at Weld Toe from Welding start (right) taken from [74]

In their research work, Duranton et al. [75] presented used both two and three dimensional FE
models to simulate multi-pass welding of pipes. To enhance computational efficiency, the idea of
adaptive mesh refinement was used simulate inherently computationally intensive multi-pass welding
of 316L stainless steel pipes. In modeling approach, for heat input from the welding torch, double
ellipsoidal heat source model, temperature dependency of materials (without metallurgical
transformations) were incorporated in traditional sequentially coupled thermal and structural analysis
procedure. Automatic mesh refinement
algorithm, previously employed by
several researchers [21, 27], was used for
iso-parametric elements. During mesh
refinement, data transfer was
accomplished by mapping Gauss point
data of source element to Gauss points of
target elements. Three-dimensional FE

Figure 2.12 Three-dimensional FE model with varying
mesh density used in [75]
An Overview of Historical Research Issues in Computational Weld Mechanics

model as shown in Figure 2.12 was used to solve first five passes. Beyond this, the mesh size
increases to an unrealistic big size to be solved with computational efficiency. To cater for the issues,
complete multi-pass (thirteen weld passes) welding was accomplished by axis-symmetric two-
dimensional models. A good agreement between the predicted and experimentally measured axial
stresses was shown. In case of hoop residual stresses, two dimensional-models showed significantly
higher values in comparison to the experimental measured hoop stresses.
The structural analysis in arc welding simulations being highly non-linear transient problem is
computationally very expensive in terms of CPU time and storage. To overcome this problem,
Nishikawa Hiroyasu, Serizawa Hisashi and Murakawa Hidekazu [76] develop an interactive
substructure method to reduce the computational time in three-dimensional analysis. In this paper, in
order to confirm calculation efficiency for a large scale problem, calculation of the pipe model with
538,200 degree of freedoms is performed. Further, calculation of the residual distortion in engine
component is performed and its applicability for practical problems is demonstrated. Figure 2.13
shows the FEM model of the pipes used along with the temperature.

Figure 2.13 FEM Model and the Captured Temperature History in [76]

Another important contribution in circumferential welding is from Dean Deng and Hidekazu
Murakawa [77]. In this study, 3-D FE model and 2-D axis-symmetric FE model are developed to
analyze the temperature fields and the residual stress distributions for arc welding of stainless steel
(SS304) pipes. Results from numerical simulations were compared with the experimental
measurements. It has been proved that the proposed computational procedure is a very effective
method for predicting the thermal cycles and the welding residual stresses. According to the simulated
and experimental results, the authors portrayed that temperature distribution around the heat source is
very steady when the welding torch moves around the stainless steel pipe. Simulated results from the
3-D model showed that the residual stress around the circumferential direction almost has a
homogenous distribution except for the welding start part. The axial residual stresses on the insider
surface and the outside surface showed a contrary distribution. In weld zone and its vicinity, a tensile
axial residual stress is produced on the inside surface as shown in Figure 2.14(right), and compressive
axial stress at outside surface as shown in Figure 2.14(left). Away from the weld centerline, tensile
axial stress is formed on the outside surface, and compressive axial stress on the inside surface.
An Overview of Historical Research Issues in Computational Weld Mechanics


Figure 2.14 Axial residual stress plots on outer (left) and inner surface (right) presented by [77]

Further, to their work the same authors along with Liang [78] contributed a hybrid numerical and
experimental investigations of welding residual stresses in multi-pass butt-welded of medium thick-
walled austenitic stainless steel pipes. Initially multi-pass welding experiments were conducted to
examine the evolution of temperature and residual stress patterns. Two-dimensional axis-symmetric
FE models and similar modeling & simulation approach as presented in [77] was adopted and
metallurgical effects were not considered. The authors claimed a precise prediction of thermal and
stress strain fields in comparison with experimental measurements. The results also showed that the
yield strength of the weld metal has significant effect on the final welding residual stress especially in
the weld zone.
Another important and recent contribution in the field of arc welding simulations of girth welding
is presented by Siddique [79]. The PhD thesis primarily focuses on the prediction of weld induced
imperfections like distortions and residual stresses in pipe-flange joints. The effects of welding
parameters, welding procedures and applied mechanical constraints on residual stress built up and
transient deformations were presented in a systematic fashion. Some efforts were shown to suggest
the mitigation techniques to enhance the in-service life of welded pipe-flange structures. The effects
of number and locations of tack welds on the transient and residual stress/strains and deformations on
the pipe-flange joint are discussed in details. The results from some selected simulations are also
experimentally validated with close co-relation between the deformations and residual stresses. Figure
2.15 shows predicted and experimental residual stresses respectively from reference purpose [79].
In their research Sattari and Javadi [80] presents a three-dimensional thermo-mechanical analysis
for the investigation of the effects of welding sequence on welding deformations in pipe joints of AISI
stainless steel. Based on nine different welding sequences, parametric studies were conducted to using
single-pass TIG welding with "V" joint geometry in pipes having a diameter of 274 mm and a
thickness of 6.2 mm. Some experimental measurements based on conventional metrology were also
presented in order to verify the FE simulated data. The authors presented that by selection of a
An Overview of Historical Research Issues in Computational Weld Mechanics

suitable welding sequence, it is possible to optimize the weld induced deformations. The conclusions
of interest are:

Figure 2.15 Comparison of Actual and Predicted Residual Stress profiles (hoop on left) and (axial on right)
from [79]

Predicted diameteral deformations are in close agreement with experimental
measurements. Different ovality levels were obtained with different weld sequences.
It was shown that under certain conditions, welding with four segments may cause more
distortions than welding with one or two segments.
Considerable reduction in diameteral variation was shown by increasing the number of
segments from one to eight as shown by Figure 2.16.

Figure 2.16 Comparison of diameteral variation for different weld sequences form [80]

Lee and Chang [81] presented a 3D finite element simulation of circumferential welding of mild
steel pipes. Additionally, parametric studies with pipe inside radius to wall thickness ratio ranging
from 10 mm to 100 mm have been presented for the investigation of the effects of pipe diameters on
residual stresses. Thermo-physical and thermo-mechanical properties of mild steel were taken from
literature. Volumetric heat input through Gaussian distributed heat patterns to FE model was
incorporated in sequentially coupled manner. The simulated data was verified by previously measured
An Overview of Historical Research Issues in Computational Weld Mechanics

experimental data from literature. Axial and hoop residual stresses for different diameter to wall
thickness ratios are shown in Figure 2.17 for reference. Although the authors end up with some
important conclusions, the research work lacks two important aspects i.e. effects of tack welds and
effects of root openings on the corresponding residual stresses. The authors concluded that:
Three-dimensional FE models are mandatory to precisely capture the significant effects at
weld start and weld end locations.
Axial and hoop residual stresses were influenced by the pipe diameters in thin walled

Figure 2.17 Predicted axial (left) and hoop (right) residual stress distributions for the FE models with
different diameter to wall thickness ratios at circumferential location of 90
on inner surface [81]

Kermanpur et al. [82] in their research
work presented experimental and numerical
investigations on multi-pass, GTA butt-welded
Incoloy 800 pipes. Temperature dependence of
material properties along with birth and death
techniques is used. The focus of this
investigation was to investigate the thermal
effects of Gaussian distributed heat source on
circumferentially welded iron based Incoloy
800H super-alloy pipes to further investigate
the structural results. Temperature distribution
within the HAZ was measured by
thermocouples. Numerical models were
calibrated based on these temperature
measurements and weld bead geometries from
experimental micrograph as in Figure 2.18. Based on the FE model developed, parametric studies
were conducted to study the effects of heat input and heat source efficiency. Both surface and

An Overview of Historical Research Issues in Computational Weld Mechanics

volumetric heat flow functions were considered. The authors concluded that fully volumetric heat
flow provides best comparative results with experiments due to small thickness of the pipes used in
the studies (5 mm). Further, the authors showed that 100% increase in heat input results in 100
increase in maximum temperature of work-piece. For weld pool, the increased heat input results in
wider weld pool.
In recent years, Deng and Murakawa developed thermal-metallurgical-mechanical procedures to
simulate welding temperatures, microstructure and residual stresses in multi-pass butt-welded 2.25Cr-
1Mo steel pipes [83]. The authors emphasized on the influence of solid-state phase transformation to
predict the residual stress fields and the effects of volumetric changes due to this phase transformation
in two-dimensional axis-symmetric FE models was incorporated. Sequentially coupled thermo-
mechanical formulation by using ABAQUS code with incorporation of material non-linearties and
transformation induced plasticity was simulated. For simplification, the effects of work hardening
were neglected. To ensure the validity of the FE simulations, welding experiments were conducted by
the authors followed by strain residual stress measurements by using strain gage method. The
following were the significant conclusions:
The phase dependent material properties particularly yield strength is the key parameter
for precise results prediction.
On the steel under investigation, the volumetric and yield strength changes are shown to
have considerable effects on residual stresses. The simulated results closely co-related
with the experimentally measured axial and hoop residual stresses as shown in Figure
Phase transformation primarily affects the hoop tensile stresses on the outside surface.

Figure 2.19 Comparison of experimental axial (left) and hoop (right) residual stresses on inner surface with
simulated results by considering phase transformation (case C and D) from [83]

An Overview of Historical Research Issues in Computational Weld Mechanics


With enhanced computational resources and substantial development in more refined numerical
simulation tools, the use of Finite Element techniques in Computational Weld Mechanics (CVM)
focusing on structural response of welded structures has been progressively increasing. Welding
engineers around the globe feels comfortable with modeling & simulation tools to optimize the
welding parameters and subsequently their adverse consequences on in-service structural integrity of
welded structures rather than concentrating on highly resource intensive welding experiments. The
technique of CVM is reaching its maturity level and is being successfully employed in industrial
In the preceding discussion (in this chapter) efforts are made to encompass the evolution process
of this technique with reference to circumferential welding. It is evident that a lot of the work
pertaining to numerical simulation of circumferential welding relates to simplified two-dimensional
analysis by employing lateral and rotational symmetry assumptions. It is mainly because of very
limited computational power available. Though such simplification effectively reduce computation
time and give a reasonable judgment of residual stress profile in the weldments but at the same time it
makes the model oversimplified by elimination the effect of root pass and tack welds etc. The inherent
deficiency of such models is their incapability to predict change in deformation and residual stresses
in the circumferential direction. It is widely accepted fact that axial deformations are usually
overestimated by two-dimensional FE model. Deviation from rotational symmetry, especially near the
weld start and end position has been reported by several researchers [44, 49, 71]. Further, some
studies reported a considerable deviation of residual stresses predicted by axis-symmetric models as
compared to experimentally measured stresses, particularly hoop stresses [4, 75].
With the rapid development in the field of personal computers computational power and
necessary data storage capacity have increased drastically during the last decade. This development
attracted researchers to model welding phenomena more realistically by using three-dimensional FE
models. More realistic computational solutions through more realistic three-dimensional models have
become a well recognized need of industry. Even though the computational power has increased
substantially, but it's not the computational power alone which can make solving a three-dimensional
models handy and affordable. This situation demands further improvement of commercially available
finite element codes by adoption and incorporation of special techniques, such as adaptive mesh
management, more sophisticated solid shell elements and better hand-shaking parallel computing

An Overview of Historical Research Issues in Computational Weld Mechanics


[1] James, F., The Procedure Handbook of Arc Welding, 14th edition. Lincoln Arc Welding
Foundation, Cleveland, USA, 2000.
[2] M Abid, M Siddique and R A Mufti; Prediction of welding distortions and residual stresses in
a pipeflange joint using the finite element technique, Modeling Simulation .in Material
Science & Engineering, vol. 13, pp. 1-16, 2005.
[3] Lindgren, L.E., Finite Element Modeling and Simulation. Part 1: Increased Complexity,
Journal of Thermal Stresses, vol. 24, pp. 141-192, 2001.
[4] Lindgren, L.E., Finite Element Modeling and Simulation. Part 2: Improved Material Modeling,
Journal of Thermal Stresses, vol. 24, pp. 195-231, 2001.
[5] Lindgren, L.E., Finite Element Modeling and Simulation. Part 3: Efficiency and Integration,
Journal of Thermal Stresses, vol. 24, pp. 305-334, 2001.
[6] Yaghi, A., Becker, A., State of the ReviewWelding Simulations using Finite Element
Methods, Report No. FENET-UNOTT-DLE-08, School of Mechanical, Materials &
Manufacturing Engineering, University of Nottingham, UK, 2004.
[7] Rosenthal, D., The Theory of Moving Heat Source and its Application to Metal Treatment,
Transaction ASME, 1946.
[8] Rybicki, E.F., Schmueser, D.W., Stonesifer, R.W., Groom, JJ and Mishler, H.W., A Finite
Element Model for Residual Stresses and Deflections in Girth Butt Welded Pipes, Journal of
Pressure Vessel Technology, vol. 100, pp. 256-262, 1978.
[9] Debiccari, A., Control of Distortion and Residual Stresses in Girth Welded Pipes, Ph.D.
Thesis, Massachusetts Institute of Technology, USA, 1986.
[10] Seo, S.I., Yang, Y.H. and Jang C.D., Development of New Finite Element Method to Analyze
Deformation of Plate due to Line heating. Journal of Ship Production, vol. 17, pp. 1-7, 2001.
[11] Goldak, J., Zhou, J., Breiguine, V., and Montoya, F., Thermal Stress Analysis of Welds: From
melting Point to Room Temperature, JWRI, vol. 25, No. 2, pp. 1851-89,1996.
[12] Goldak, J Chakravarti, A. and Bibby, M., A new Finite Element Model for Heat Sources,
Metallurgical Transactions B, vol. 15 B, pp. 299-305, 1984.
[13] Goldak, J Bibby, M, Moore, J., House, R. and Patel, B Computer Modeling of Heat Flow in
Welds, Metallurgical Transactions B, vol. 17 B, pp. 587-600, 1986.
[14] Sabapathy, P.N., Wahab, M.A. and Painter, M.J., Numerical Models of In Service Welding of
Gas Pipeline, Journal of Materials Processing Technology, vol. 118. pp. 14-21, 2001.
[15] Ravichandran, G., Raghupathy, V.P. and Ganesan, N., Analysis of Temperature Distribution
During Circumferential Welding of Cylindrical and Spherical Components using the Finite
Element Method, Journal of Computer and Structures, vol. 59, No. 2, pp. 225-255, 1996.
An Overview of Historical Research Issues in Computational Weld Mechanics

[16] Ueda, Y., and Yamakawa, T., Analysis of Thermal-Elastic Stress and Strain During Welding
by Finite Element Method, JWRI, vol. 2, No. 2, 1971.
[17] Hibbitt, H.D., and Marcal, P.V., A Numerical Thermo-Mechanical Model for the Welding and
Subsequent Loading of Fabricated Structure, Journal of Computers and Structures, vol. 3, pp.
1145-1174, 1973.
[18] Friedman, E., Thermo-Mechanical Analysis of the Welding Process using the Finite Element
Method, ASME Journal of Pressure Vessel Technology, pp. 206-213, 1975.
[19] Andersson, B.A.B., Thermal Stresses in Submerged Arc Welded Joint Considering Phase
transformations, ASME Journal of Engineering Material and Technology, vol. 100, pp. 356-
362, 1978.
[20] McDill, J.M., Oddy, A.S. and Klien, M.E., Data Transfer for 3-D Adaptive Thermal-Elasto-
Plastic Finite Element Analysis, Simulation of Materials Processing: Methods and
Applications, Shen & Dawson (eds), 1995.
[21] McDill, J.M., Oddy, and Goldak, J.A., An Adaptive Mesh-management Algorithm for Three
Dimensional Automatic Finite Element Analysis, Transactions of CSME, vol. 15, No. I, pp.
57-70, 1991.
[22] McDill, J.M., Goldak, J.A., Oddy, A.S., and Bibby, M J., Isoparametric Quadrilaterals and
Hexahedrons for Mesh-Grading Algorithms, Communications in Applied Numerical Methods,
vol. 3, pp. 155-163, 1987.
[23] McDill, J.M and Oddy, A.S., A Non-Conforming Eight to 26-Node Hexahedron for Three
Dimensional Thermal-Elasto-Plastic Finite Element Analysis, Journal of Computers and
Structures, vol. 54, No. 2, pp. 183-189, 1995.
[24] McDill, J.M and Oddy, A.S., Arbitrary Coarsening for Adaptive Mesh Management in Three
Dimensional Automatic Finite Element Analysis, Mathematical Modeling and Science
Computing, vol. 2, pp. 1072-1077, 1993.
[25] Runnemalm, H. and Hyun, S., Three-Dimensional Welding Analysis using an Adaptive Mesh
Scheme, Computer Methods in Applied Mechanics and Engineering, vol. 189, pp. 515-523,
[26] McDill, J.M, Runnemalm, K.H. and Oddy, A.S., An 8-to-16 Node Solid Graded Shell
Elements for Far-Field Applications in 3-D Thermal Mechanical FEA, 12th International
Conference on Mathematical and Computer Modeling and Scientific Computing, Chicago,
[27] Lindgren L.E., Haggblad H.A., McDill J.M.J. and Oddy A.S., Automatic Re-meshing for
Three-Dimensional Finite Element Simulation of Welding, Computer Methods in Applied
Mechanics and Engineering, vol. 147, pp. 401-409, 1997.
An Overview of Historical Research Issues in Computational Weld Mechanics

[28] Hyun, S. and Lindgren, L.E., Smoothing and Adaptive Re-meshing Schemes for Graded
Elements, International Journal of Communications in Numerical Methods in Engineering.,
vol. 14, pp. 1-17, 2001.
[29] H.F. Junejo. Finite Element Modeling of Weld Based Rapid Prototyping, PhD Thesis, Ghulam
Ishaq Khan Institute of Engineering & Technology, Pakistan (2006).
[30] Leggatt, R.H., White, J.D., Predicting Shrinkage and Distortion in a Welded Plate, Proceedings
of the Conference on Residual Stresses in Welded Construction and their Effects, pp.119-132,
[31] M.S. Alam. Structural Integrity and Fatigue Crack Propagation Life Assessment of Welded
and Weld-Repaired Structures, PhD Thesis, Louisiana State University, 2005.
[32] Ueda, Y., Kim, Y.C., Measuring Theory of Three Dimensional Residual Stresses using a
Thinly Sliced Plate Perpendicular to Welded Line, Transactions of JWRI, Vol.14, No.2,
pp.151-157, 1985
[33] Vaidyanathan S., Todato A.T., and Finnic, I., Residual Stresses due to Circumferential Welds,
ASME J. Engineering Material and Technology, pp. 239-242, 1973.
[34] Hill, M.R. and Nelson, D.V., The Localized Eigenstrain Method for Determination of Triaxial
Residual Stress in Welds, PVP 373, pp.397-403, 1998
[35] Dong, Y., Hong, J.K., Tsai, C.L. and Dong, P., Finite Element Modeling of Residual Stresses
in Austenitic Stainless Steel Pipe Girth Welds, Welding Journal, Vol.76, No.10, pp.442s-449s,
[36] Vaidyanathan, S Weiss, H. and Finnic, I., A Further Study of Residual Stresses in
Circumferential Welds, ASME Journal of Engineering Material and Technology, pp. 239-242,
[37] Rybicki, E.F and Stonesifer, R.B., Computation of Residual Stresses Due to Multi-pass Welds
in Piping Systems, International Journal of Pressure Vessels Technology, vol. 101, pp. 149-
154, 1979
[38] Rybicki, E.P., and Stonesifer, R.B., An Analysis of Weld Repair Residual Stresses for an
Intermediate Test Vessel, Journal of Pressure Vessel Technology, vol. 102, pp. 323-331, 1980
[39] Rybicki, E.F and McGuire, P.A., "A Computational Model for Improving Weld Residual
Stresses in Small Diameter Pipes by Induction Heating," Journal of Pressure Vessel
Technology, vol. 103, pp. 294-299, August 1981.
[40] Rybicki, E.F and MeGuire, P.A., "The Effects of Induction Heating Conditions on
Controlling Residual Stresses in Welded Pipes," Journal of Engineering Materials and
Technology, vol. 104, pp. 267-273, 1982
[41] Rybicki, E.F., McGuire, P.A., Merrick, E., and Wert, B., "The Effect of Pipe Thickness on
Residual Stresses due to Girth Welds," Journal of Pressure Vessel Technology, vol. 104, pp.
204-209, 1982
An Overview of Historical Research Issues in Computational Weld Mechanics

[42] Nair, S., Pang, E. and Dix, R.C., Residual Stress Generation and Relaxation in Butt-Welded
Pipes, ASME Journal of Pressure Vessel Technology, vol. 104, pp. 188-192, 1982
[43] Scaramangas, A., Residual Stresses in Girth Butt Welded Pipes, Ph.D. Thesis, University of
Cambridge, 1984
[44] Jonsson, M. and Josefson, B.L., Experimentally Determined Transient and Residual Stresses
in the Butt-Welded Pipes, Journal of Strain Analysis, vol.23, no. I, pp. 25-31, 1988
[45] Josefson, L., Jonsson, M., Karlsson, L Karlsson, R., Karlsson, T. and Lindgren L.E.,
Transient and Residual Stresses in a Single-Pass Butt-Welded Pipe", ICRS2, Nancy, France,
Nov. 23-25. 1988
[46] Lindgren L.E. and Karlsson, L., Deformations and Stresses in Welding of Shell Structures, Int.
Journal of Numerical Methods in Engineering, vol. 25, pp. 635-655, 1988
[47] Karlsson, R.I. and Josefson, B.L, Three Dimensional Finite Element Analysis of Temperature
and Stresses in a Single-Pass Butt-Welded Pipe, ASME Journal of Pressure Vessel
Technology, vol. 112, pp. 76-84, 1990
[48] Gareth A. Taylor, Michael Hughes, Nadia Strusevich and Koulis Pericleous, Finite Volume
Methods applied to the Computational Modeling of Welding Phenomena, Centre for
Numerical Modeling and Process Analysis, University of Greenwich, London, UK
[49] Karlsson, L., Jonsson, M., Lindgren, L,E., Nasstrom, M. and Troive, L., Residual Stresses and
Deformations in a Welded Thin-Walled Pipe, Proceedings of ASME Pressure Vessel and
Piping Conference, Hawaii, 1989, Weld Residual Stresses and Plastic Deformation, PVP-Vol.
17.1, pp. 7-14, 1989
[50] Song, H., Application of Finite Element Techniques in the Simulation of Structure Interaction
during Welding, Ph.D. Thesis, Massachusetts Institute of Technology, USA, 1990
[51] Troive, L. and Jonsson, M Numerical and Experimental Study of Residual Deformations due
to Double-J Multi-Pass Butt-Welding of a Pipe-Flange Joint, Proceedings of IEMS'94 (1994
Annual International Conference on Industry, Engineering and Management Systems), Cocoa
Beach, Florida, USA, pp. 107-114, 1994
[52] Troive, L., Lindgren, L.E. and Jonsson, M., Axial Collapse Load of Girth Butt Welded Pipe,
Proceedings of First International Symposium on Thermal Stresses and Related Topics,
Shizuoka University, Hamamatsu, Japan, pp. 565-568, 1995
[53] Troive, L., Nsstrm, M. and Jonsson, M., Experimental and Numerical Study of Multi-Pass
Welding Process of Pipe-Flange Joint, ASME Pressure Vessel Technology, vol. 120, pp. 244-
251, 1998
[54] Michaleris, P., Residual Stress Distributions for Multi-Pass Welds in Pressure Vessel and
Piping Components, Proceedings of ASME Pressure Vessel and Piping Conference, Montreal,
Canada, 1996, Residual Stresses in Design Fabrication, Assessment and Repair, PVP-vol.
327, 1996
An Overview of Historical Research Issues in Computational Weld Mechanics

[55] Dike, J.J., Ortega, A.R. and Cadden, C.H., Finite Element Modeling and Validation of
Residual Stresses in 304L Girth Welds, 5th International Conference on Trends in Welding
Research, Pine Mountain, GA, June 1-5, 1998
[56] Mohr, W.C., Michaleris, P. and Kirk, M.T., An Improved Treatment of Residual Stresses in
Flaw Assessment of Pipes and Pressure Vessel Fabricated From Ferritic Steel, Proceedings of
ASME Pressure Vessel and Piping Conference, Orlando, Florida, 1997
[57] Dong, Y., Hong, J,K., Tasi, C.L. and Dong P., Finite Element Modeling of Residual Stresses in
Austenitic Stainless Steel Pipe Girth Welds, Welding Research Supplement, pp. 443s-449s,
[58] Chapman, T., Offer, H., Sanders, W. and Rusack, G., Reduced Stress Welding Process for
Nuclear Plant Piping, Nuclear Engineering and Design, vol. 170, pp. 81-88, 1997
[59] Yang, Y.S. and Lee, S.H., A study on Mechanical Stress Relieving in a Butt Welded Pipe,
International Journal of Pressure Vessels and Piping, vol. 73, pp. 175-182, 1997
[60] Brickstad, B. and Josefson, B.L A Parametric Study of Residual Stresses in Multi-Pass Butt-
Welded Stainless Steel Pipes, International Journal of Pressure Vessels and Piping, vol. 75, pp.
11-25, 1998
[61] Teng, T.L. and Chang, P.H., A Study of Residual Stresses in Multi-Pass Girth Butt Welded
Pipes, International Journal of Pressure Vessels and Piping, vol. 74, pp. 59-70, 1997
[62] Teng, T.L. and Chang, P.H Three-Dimensional Thermo-mechanical Analysis of
Circumferentially Welded Thin-Walled Pipes, International Journal of Pressure Vessels and
Piping, vol. 75, pp. 237-247, 1998.
[63] Runnemalm, H., Efficient Finite Element Modeling and Simulation of Welding, Ph.D. Thesis,
Lulea University of Technology, Sweden, 2000.
[64] Dong, P. and Zhang, J Residual Stresses in Strength-Mismatched Welds and Implication on
Fracture Behavior, Engineering Fracture mechanics, vol. 64, pp. 485-505, 1999.
[65] Zhang, J Dong, P., Brust, F.W., Shack, W.J., Mayfield, M.E. and McNeil, M., Modeling of
Weld Residual Stresses in Core Shroud Structures, Nuclear Engineering and Design, vol. 195,
pp. 171-187, 2000.
[66] Basavaraju, C., Simplified Analysis of Shrinkage in Pipe-Pipe Butt Welds, Nuclear
Engineering and Design, vol. 197, pp. 239-247, 2000.
[67] Engelhard, G., Halrip, L.M., Pellkofer, D., Schmidt, J. and Weber, J Optimization of
Residual Welding Stresses in Austenitic Stainless Steel Piping: Proof testing and Numerical
Simulation of Welding and Post Welding Process, Nuclear Engineering and Design, vol. 198,
pp. 141-151, 2000.
[68] Mochizuki, M Hayashi, M. and Hattori, T Numerical Analysis of Welding Residual Stress
and its Verification Using Neutron Diffraction Measurement. ASME Journal of Engineering
Material and Technology, vol. 122, pp. 98-103, 2000.
An Overview of Historical Research Issues in Computational Weld Mechanics

[69] Hedblom E, and Lindgren, L.E Simulation of Multi-pass Welding to Reduce IOSCC,
appended paper in Licentiate Thesis, Simulation of Multi-pass Welding to Reduce Inter-
granular Stress Corrosion Cracking, ISSN: 1402, Lulea University of Technology, Sweden,
[70] Hedblom, E Multi-pass Welding of Nuclear Reactor Components Measurements,
Technical Report, 2002:07, Lulea University of Technology, Sweden, 2002.
[71] Fricke, S Keim, E. and Schmidt, J Numerical Weld Modeling - A Method for Calculating
Weld-Induced Residual Stresses, Nuclear Engineering and Design, vol. 206, pp. 139-150,
[72] S. W. Wen, P. Hilton, and D. C. J. Farrugia; Finite Element Modeling of Submerged Arc
Welding Process, Journal of Materials Processing Technology, vol. 119, pp. 203-209, 2001.
[73] Bang, W., Son, Y.P., Oh, K.H., Kirn, Y.P. and Kim, W.S., Numerical Simulation of Sleeve
Repair Welding of In-Service Gas Pipelines, AWS Welding Journal, pp. 273s - 182s, 2002.
[74] Xiangyang, L., Influence of Residual Stress on Fatigue Failure of Welded Joints, PhD Thesis,
North Carolina State University, 2002.
[75] Duranton, P., Devaux, J., Robin, V., Gilles, P. and Bergheau, J.M., 3D Modeling of Multi-Pass
Welding of a 316L Stainless Steel Pipe, Journal of Material Processing Technology, vol. 153-
154, pp. 457-463, 2004.
[76] Nishikawa, H., Serizawa, H., and Murakawa, H. Actual Application of FEM to Analysis of
Large Scale Mechanical Problems in Welding. Journal of Science and Technology of Welding
& Joining, vol. 12(2), pp. 147-152, 2007.
[77] D, Deng and H, Murakawa. Numerical Simulation of Temperature Field and Residual Stresses
in Multi-Pass Welds in Stainless Steel Pipe and Comparison with Experimental Measurements.
Journal of Computational Materials Science, vol. 37, pp. 269-277, 2006.
[78] D, Deng, H, Murakawa and W. Liang. Numerical and Experimental Investigations on Welding
Residual Stress in Multi-Pass Butt-Welded Austenitic Stainless Steel Pipe. Journal of
Computational Materials Science, vol. 42, pp. 234-244, 2008.
[79] M. Siddique, Experimental and Finite Element Investigation of Residual Stresses and
Distortions in Welded Pipe-Flange Joints, PhD Thesis, Ghulam Ishaq Khan Institute of
Engineering & Technology, Pakistan (2005).
[80] I. Sattari-Far and Y. Jawadi. Influence of Welding Sequence on Welding Distortions in Pipes.
International Journal of Pressure Vessels and Piping (2007), doi:10.1016/j.ijpvp.2007.07.003
[81] C.H. Lee and K.H. Chang. Three-Dimensional Finite Element Simulation of Residual Stresses
in Circumferential Welds of Steel Pipe Including Pipe Diameter Effects. Journal of Material
Science & Engineering-A (2007), doi:10.1016/j.msea.2007.10.011.
An Overview of Historical Research Issues in Computational Weld Mechanics

[82] A. Kermanpur, M. Shamanian and V.E. Yeganeh. Three Dimensional Thermal Simulation and
Experimental Investigation of GTAW Circumferentially Butt-Welded Incoloy 800 Pipes.
Journal of Materials Processing Technology (2007), doi:10.1016/j.jmatprotec.2007.08.009.
[83] D, Deng and H, Murakawa. Finite Element Analysis of Temperature Field, Microstructure and
Residual Stresses in Multi-Pass Butt-Welded 2.25Cr-1Mo Steel Pipes. Journal of
Computational Materials Science, Article in Press, doi:10.1016/j.commatsci.2008.01.025.