Applied Mathematical Modelling 35 (2011) 688707

Contents lists available at ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Finite element modeling of welding processes

Andrs Anca, Alberto Cardona *, Jos Risso, Vctor D. Fachinotti
Centro Internacional de Mtodos Computacionales en Ingeniera (CIMEC-INTEC), Universidad Nacional del Litoral (UNL),
Consejo Nacional de Investigaciones Cientcas y Tcnicas (CONICET), Gemes 3450, S3000GLN, Santa Fe, Argentina

a r t i c l e i n f o a b s t r a c t

Article history: This work deals with the simulation of fusion welding by the Finite Element Method. The
Received 6 May 2009 implemented models include a moving heat source, temperature dependence of thermo-
Received in revised form 7 June 2010 physical properties, elasto-plasticity, non-steady state heat transfer, and mechanical anal-
Accepted 6 July 2010
ysis. The thermal problem is assumed to be uncoupled from the mechanical one, so the
Available online 21 July 2010
thermal analysis is performed separately and previously to the mechanical analysis at each
time step. The mechanical problem is based on the thermal history. A special treatment is
Keywords:
performed on mechanical elements during the liquid/solid and solid/liquid phase changes
Welding simulation
FEM
to account for stress states. The three-dimensional stress state of a butt-welded joint is
Solids mechanics obtained as an example of an application.
Thermal stresses 2010 Elsevier Inc. All rights reserved.

1. Introduction

The use of the Finite Element Method (FEM) in product development is now well established. Its use in manufacturing
processes is increasing and is part of the eld of new applications in computational mechanics. The most important reason
for this development is the industrial need to improve productivity and quality of products and to have better understanding
of the inuence of different process parameters. The modeled phenomena play an important role at various stages of the
production of steel parts, such as welding, heat treatment and casting, among others.
The importance of these applications lies in determining the evolution of stresses and deformations to predict, for exam-
ple, susceptibility to cracking and thus prevent failures during manufacturing or even service. Furthermore, this simulation
tool can be used to optimize different aspects of manufacturing processes.
Welding is dened by the American Welding Society (AWS) as a localized coalescence of metals or non-metals produced by
either heating of the materials to a suitable temperature with or without the application of pressure, or by the application of pres-
sure alone, with or without the use of ller metal [1]. Welding techniques are one of the most important and most often used
methods for joining pieces in industry. Any information about the shape, size and residual stress of a welded piece is of par-
ticular interest to improve quality.
The analysis of welding processes involves several branches of physics, and requires the coupling of different models ad-
dressed to describe the phenomenological behavior of a system. Many of these models have been implemented numerically
and are being used in an efcient way to solve the problems on an individual basis.
In the arc welding process, the energy required for metal fusion is produced by the Joule effect. This effect produces the
energy required to melt the base and ller metals, forming what is known as the liquid pool. The surface temperature of the

* Corresponding author.
E-mail addresses: aanca@intec.unl.edu.ar (A. Anca), acardona@intec.unl.edu.ar (A. Cardona), jrisso@intec.unl.edu.ar (J. Risso), vfachino@intec.unl.edu.ar
(V.D. Fachinotti).

0307-904X/$ - see front matter 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2010.07.026
 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 689

piece varies from 1700 K to 2500 K, depending on the material. In the liquid pool, convective effects take place that improve
the heat transport. Finally, after removing the heat sources, the metal solidies.
During the process, temperature changes in the alloy produce solid state transformations. These microstructural transfor-
mations cause changes in the material properties during the evolution of the process. Heating and cooling induce localized
volumetric changes. The thermal strains that occur in the vicinity of the welding zone are elasto-plastic and the resulting
stresses react, causing permanent distortions.
Distortions introduce residual deformation that complicates the assembly of welded structures and reduces their quality.
Also, in certain applications, this distortion may result in the structure being useless.
Since the beginning of the 70s, FEM has been used to simulate the mechanical problem in welding. Some well-known
references up to the 90s are the works of Ueda and Yamakawa [2], Marcal [3], Karlsson [4,5], Goldak et al. [6], Radaj [7]. Finite
element codes addressed to study this problem have been described in the papers of Dexter [8], Chandra [9], Ravichandran
et al. [10] and Karlsson [11], among many others. In most nite element models, it has been commonplace to assume a num-
ber of simplications, which can be described as follows:

 Several authors assume symmetries in the problem, in order to solve 2D models. This simplication can be found, for
instance, in the works of Song et al. [12] on welded joints, Branza et al. [13] on TIG welding for repairing parts, Cho
et al. [14] in laser welding, Hou et al. [15] for TIG butt welding, Hou et al. [16] for spot welding, and Deng and Murakawa
[17] for multiple-pass welding. The motivation (not always recognized) of this simplication is the high computational
cost of 3D models. However, Duranton et al. [18] and Gery et al. [19] reveal shortcomings in the 2D approach in longitu-
dinal welding when the heat ux in the direction of the weld is ignored.
 Concerning the mechanical description of the problem, it has been assumed that materials respond as an inviscid elasto-
plastic model in the works of Song et al. [12], Abid and Siddique [20], Hou et al. [16], Mochizuki [21], Mollicone et al. [22],
with isotropic hardening Branza et al. [13], Duranton et al. [18], Ferro et al. [23], Hyde et al. [15], Deng and Murakawa
[24,17], Mahapatra et al. [25], or as an elasto-viscoplastic material Nandan et al. [26], Hu and Richardson [27]. Alberg
and Berglund [28] compared plastic and viscoplastic models applied to welding simulations, and recommend using a sim-
ple plastic model in the initial stages of the study. However, it cannot be ignored that at high temperature, the viscous
effects have a noticeable effect on the behavior of metals, as mentioned by Fachinotti and Cardona [29].
Also, in general, welding involves heatingcooling cycles whose effect on the mechanical behavior of the material is rep-
resented only if kinematics hardening is taken into account. Therefore, the justication of using simpler models such as
the elasto-plastic without hardening or only with isotropic hardening (up to now, the most widely used) is a mere reduc-
tion of computational cost.

Very few models applied to simulate welding take into account the dynamics of the molten material. Che et al. [30] fo-
cuses on determining the seam shape. Taylor et al. [31], Pericleous et al. [32], Wang et al. [33], De and DebRoy [34], He et al.
[35], Roy et al. [36] model the convection in the liquid pool and its effect on heat transfer. This effect is usually approximated
in thermal analysis by increasing the metal conductivity. Detailed models of the weld pool allow to validate the augmented
conductivity parameters and achieve consistency in prediction.
Most works on welding validate results by comparison with welding tests. However, experimental tests have some uncer-
tainties in the obtained data, which make it difcult to debug coding errors in the implemented program. An alternative can
be to compare results with those obtained numerically by other authors. Nevertheless, welding is a complex process that
involves interactions of various physical and metallurgical phenomena. When dealing with this type of problem, several sim-
plications are assumed, and not all effects are taken into account in a rst approximation. It should be noted that several
papers in the literature that presented comparisons, did not model exactly the same physical phenomena or did not use the
same material model of the referenced work against which they are making the comparison. In addition, in the published
material, the lack of boundary conditions and welding parameters data makes it very difcult to exactly reproduce the test
cases.
In this paper, we use comparisons based on analytical and semi-analytical solutions to validate the thermal and mechan-
ical problems. Thus, we avoid the previously mentioned shortcomings. We believe that this methodology generates a solid
base for future developments. The following tests were performed:

 The thermal model of non-isothermal phase change has been validated against the analytical solution by zisik and Uzz-
ell 37].
 The thermal model with a moving heat source has been compared against experimental results of Christensen et al. [38].
 The mechanical model presents particularities when the solidication of molten metal is considered, and a proper vali-
dation of it is essential to achieve acceptable results in the application examples. In this paper, we validate the mechanical
problem against the semi-analytical solution of Weiner and Boley [39]. This test has been used by several authors to val-
idate their results in other applications such as continuous casting of steel.

The article is organized as follows. In Sections 2 and 3 we describe the developed model for the thermal and mechanical
problems, their hypotheses, the equations involved and the discretization procedure to obtain the set of equations in both
elds. Finally, in Section 5, the stress distribution in a butt-welded joint is modeled as an application example.
690 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

2. Thermal problem

In this section, a temperature-based nite element model to simulate unsteady conduction heat transfer problems in a 3D
media undergoing mushy phase change is described.
The analyzed domain is discretized using linear tetrahedral nite elements. Galerkin weighting functions are used.
During phase change, a considerable amount of latent heat is released or absorbed, causing a strong non-linearity in the
enthalpy function. In order to correctly model such phenomenon, we made an exact integration distinguishing between the
different one-phase subregions over those nite elements embedded into the solidication front.
Since contributions from different phases are integrated separately, the sharp variations of the material properties be-
tween phases are easily captured. This so-called discontinuous integration avoids the regularization of the phenomenon,
allowing the exact evaluation of the discrete non-linear governing equation, which is solved using a full NewtonRaphson
scheme, together with line-search.
We validated the performance of the thermal model by comparison with the exact solution of zisik and Uzzell [37].

2.1. Problem denition

Under the assumptions of incompressibility, negligible viscosity and dissipation, linear dependence of the heat ux on
temperature gradient (Fouriers law), and no melt ow during the solidication process, the energy balance for each subdo-
main Xi is governed by the classical energy balance equation
@H
q  r  jrT q 8x; t 2 Xi ; 1
@t
where T denotes the temperature, H the enthalpy (per unit volume), j = j(T) the material thermal conductivity, assumed
isotropic, and q = q(x,t) is the welding volume heat input (to be dened later in Section 2.4 in the context of welding anal-
ysis). Eq. (1) is supplemented by the following initial condition
T T0 8x 2 Xi ; t t 0
and the external boundary conditions on oX:

T T at @ XT ; 2
 jrT  n q
 at @ Xq ; 3
 jrT  n henv T  T env at @ Xc ; 4
with oXT, oXq and oXc being non-overlapping portions of the body boundary oX, with prescribed temperature, conductive
 refer to imposed temperature and heat ux elds, and Tenv
and convective heat ux, respectively (Fig. 1). In the above, T and q
is the temperature of the environment, whose lm coefcient is henv; n denotes the unit outward normal to oX.
Further, the following continuity conditions must hold at the interface(s) C:
T TC; 5
hHug jrT  gi 0; 6

Fig. 1. Thermal problem denition.

 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 691

where TC is a constant value (equal to the melting temperature for isothermal solidication, and either the solidus or liqui-
dus temperature otherwise), h*i denotes the jump of the quantity * in crossing the interface C, which is moving with speed u
in the direction given by the unit vector g. Note that the second equation states the jump energy balance at the interface.
In order to retrieve T as the only primal variable, enthalpy is dened as
Z T
HT c ds Lfl 7
T ref

with qc and qL being the unit volume heat capacity and latent heat, respectively, and Tref an arbitrary reference temperature;
fl is a characteristic function of temperature, called the volumetric liquid fraction, dened as
8
<0
> if T < T sol ;
fl T 0 6 flm T 6 1 if T sol 6 T 6 T liq ; 8
>
:
1 if T > T liq ;
where Tsol and Tliq denote the solidus and liquidus temperatures, respectively, i.e., the lower and upper bounds of the mushy
temperature range.

2.2. Finite element formulation

First, the weak or variational form of the balance equation (1) is derived, supplied by the boundary conditions Eqs. (2)(6),
using the weighted residuals method. The proper choice of weighting functions together with the application of Reynolds
transport theorem allows for cancellation of the terms arising from the interface conditions (6). Furthermore, using the def-
inition (7), a weak temperature-based form of the governing equation is obtained:
Z Z Z Z Z Z
@T @
W qc dV W qLfl dV jrW  rT dV  dS
Wq Whenv T  T env dS  Wq dV 0; 9
X @t @t X X @ Xq @ Xc X

where W is the weighting function.

In the nite element context, the unknown eld T is approximated as a linear combination of interpolation functions
Ni(x, y, z), the shape functions, as follows:
X
N
Tx; y; z Ni x; y; zT i 10
i

being Ti the temperature at each node i (i = 1, 2, . . ., N) arising from the discretization of the analyzed domain X.
We substitute T by its approximation (10) into Eq. (9). Then we adopt as weighting function W the interpolation function
N (W  Ni, Galerkin method). A non-linear system of N ordinary differential equations is set, stated in matrix form as
@T @L
WC KT  F 0; 11
@t @t
where T is the vector of unknown nodal temperatures, C the capacity matrix, L the latent heat vector, K the conductivity
(stiffness) matrix and F the external ux vector.
Each term of the residual vector W is given (in components) by:
Z
C ij qc Ni Nj dV;
ZX
Li qLfl Ni dV; 12
ZX Z
K ij jrNi  rNj dV henv N i Nj dS:
X @ Xc

On the other hand, the load vector F takes the form:

Z Z Z
Fi  N i dS
q henv T env Ni dS  qN i dV: 13
@ Xq @ Xc @X

The transient conduction problem in the absence of phase change has been widely discussed in the classic nite element
literature (see e.g. [40]).
We follow the discontinuous integration scheme given in Fachinotti et al. [41,42] that allows us to exactly solve the inte-
grals (12) in a relatively easy manner.

2.3. Solution scheme

Time integration is done with the unconditionally stable rst-order backward Euler method. This implicit scheme is ap-
plied on Eq. (11), which leads to a set of non-linear equations to be solved for the values of the temperatures at nite element
nodes, at the end of the time increment considered:
692 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

T n1  T n Ln1  Ln
Wn1 C n1 K n1 T n1  F n1 0 14
Dt Dt
The solution of the highly non-linear discrete balance equation (14) is achieved by means of the well-known Newton-
Raphson method.
At each new iteration i, W is approximated using a rst-order Taylor expansion,

WT i  WT i1 J T i1 DT i 0 15

with J = dW/dT being the Jacobian or tangent matrix of the equation system (14), and DT(i) = T(i)  T(i1) the incremental cor-
rection of temperatures. Thus, DT can be evaluated as follows:

DT i J T i1 1 WT i1 : 16

The Newton-Raphson scheme is efcient, provided that the initial guess T(0) lies within the convergence radius of the
solution T. Alternatively, the convergence can be improved using a line-search procedure.
Verication of the model has been performed comparing numerical and analytical results for a transient non-linear heat
transfer problem with an exact solution. This problem was solved exactly by zisik and Uzzell [37] and the numerical results,
in agreement with the corresponding analytical ones are reported in Anca et al. [43]

2.4. Heat source modeling in welding

The moving heat source is implemented as a typical transient formulation where the heat source moves along the part
with time. In order to model the heat source, the three-dimensional double ellipsoid proposed by [44] is studied (Fig. 2).
One characteristic of the double ellipsoid geometry is that it can be easily changed to model both the shallow penetration
arc welding processes as well as the deeper penetration laser and electron beam processes. The heat ux distribution is
Gaussian along the longitudinal axis. The front half of the source is a quadrant of one ellipsoidal source while the rear half
is a quadrant of another ellipsoidal source. Four parameters dene each ellipsoid. Physically, they correspond to the dimen-
sions of the molten zone. Knowing the cross-section of the molten zone from experiment, the heat source parameters can be
determined. As a rst approximation, [44] assume that it is reasonable to take the distance in front of the source equal to be
one half of the weld width and the distance behind the source equal to be twice the weld width.
In order to simulate arc heating effects, the equivalent heat input is assumed as a constant internal heat generation per
unit volume.
It is convenient to introduce a coordinate n, xed on the heat source and moving with it. The moving reference frame on
the heat source is related to the coordinate xed on the work piece by:
n z  v t  s; 17
where v is the welding speed and s is the time lag necessary to dene the position of the heat source at time t = 0. The
weighting fractions associated with the front and rear ellipsoids are denoted by ffand fr, respectively, and these fractions
are specied to satisfy ff + fr = 2. Let us denote q the power density in W/m3 within the ellipsoid, and let a , b and c denote
the semi-axes of the ellipsoid parallel to the x, y, n axes. Then, the power density distribution inside the front quadrant is
specied by
p !
6 3f f Q x2 y2 n2
qx; y; n p exp 3 2  3 2  3 2 18
abcf p p a b cf

Fig. 2. Goldak double ellipsoid heat source.

 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 693

and in the rear quadrant is specied by

p !
6 3fr Q x2 y2 n2
qx; y; n p exp 3 2  3 2  3 2 : 19
abcr p p a b cr

In these equations, Q is the heat available at the source. For an electric arc, the heat available is

Q gVI; 20
where g is the heat source efciency, V is the arc voltage, and I is the arc current. The parameters a, b, cf and cr are indepen-
dent, and can take different values to properly model the weld arc.
The spatial distribution of heat is calculated from Eqs. (18) and (19) and is applied as a volumetric heat generation.

2.5. Verication

In order to verify the developed moving heat source model, a 3D nite element model of a thick plate was generated.
Because of symmetry, one half of the welding plate was meshed. Fig. 3 shows the problem geometry used in the 3D FEM
analysis.
The model uses the same heat source distribution and material properties as those in Goldaks work, which are summa-
rized in Tables 13.

Fig. 3. Geometry used in the 3D FEM verication analysis.

Table 1
Arc parameters.

Arc parameter Symbol Value

Current I 1170 A
Voltage U 32.9 V
Efciency g 0.95
Total heat Q 36,538.35 W
Speed v 5 mm/s

Table 2
Goldaks parameters.

Parameter Value
a 20 mm
b 20 mm
cf 15 mm
cr 30 mm
ff 0.6
fr 1.4
694 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

Table 3
Material data for thermal problem.

Parameter Symbol Value

Density q 7200.0 kg/m3
Specic heat (solid) cs 680.0 J/kg K
Specic heat (liquid) cl 680.0 J/kg K
Latent heat L 291,660 J/kg
Conductivity (solid) js 34.0 W/m K
Conductivity (liquid) jl 2js
Solidus temp. Tsol 1774 K
Liquidus temp. Tliq 1763 K
Initial temp. T0 283 K

Fig. 4. Comparison of the present model solution with Christensen experimental results.

Fig. 4 plots the top surface temperature distribution along the x-axis 11.5 s after the arc has passed. The numerical results we
have computed are compared with the experimental test of [38]. Also, a comparison with an analytical solution for transient
temperature eld of a semi-innite body subjected to three-dimensional power density moving heat source was made [45].

3. Mechanical problem

During a thermal welding process, the weld site and immediate surrounding area experience different rates of heating/
cooling and thus expansion/contraction. This effect leads to considerable thermal strains. Due to the localized nature of heat
application, the expansion due to these strains is constrained by the cool material away from the site of the applied heat. The
physical and chemical properties of the material also change at the weld site and heat affected zone (HAZ), both during and
after the welding process. These changes affect mechanical material properties, and must be taken into account in the
mechanical analysis.
Due to the intrinsic three-dimensional nature of loads, boundary conditions, and geometries usually involved in welding
processes, a 3D mechanical model was implemented. It should be noted that the weld pool itself is not modelled in the
mechanical analysis. This is only a soft region serving as a support for heat input to the thermo-mechanical model. In this
sense, the existence of a cut-off or zero-strength temperature ZST was assumed. This is also the temperature above which no
further changes in material properties are accounted for in the mechanical analysis.
The thermoelastic material behavior is, in most cases, based on a hypoelastic version of Hookes law with inclusion of
thermal strains. The most important parameters for the analysis are the Youngs modulus and the thermal dilatation coef-
cient. Poissons ratio has a smaller inuence on the residual stresses and deformations [46]. The plastic material model used
in our tests for the solidied metal is a rate-independent model with an associative J2 von Mises law and isotropic hardening
[47]. We used rate-independent plasticity at high temperatures because of the involved time scales [48]: in the weld thermal
cycle the material has a high temperature during a relatively short time, and therefore the accumulated rate-dependent plas-
ticity may be neglected. Inertial effects are ignored in the momentum balance equations, according to the assumption of null
acceleration within the solid.

3.1. Meltingsolidifying behavior

It is recognized that nite element techniques, with standard constitutive material modelling, yield good solutions in
thermal stress analysis, even when the solid is subjected to temperatures near the solidication ones. However, there are
particular aspects in the formulation for melting/solidifying problems that need to be carefully considered.
 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 695

Three different congurations for every material point and its neighborhood (Fig. 5) are considered for the mechanical
simulation of this process:

i) The reference conguration (B), in which the particle label is assigned;

ii) The (intermediate) natural conguration (B0) which corresponds to that state where the material point has solidied
just below the zero-strength temperature (ZST), and started to develop mechanical strength;
iii) The current conguration (Bt).

Note that since the solidication time instant is not the same for all points in the domain of analysis, each material point
has its own (intermediate) natural conguration.
Let us dene uo as the displacement from the reference to the natural conguration, ut the displacement from the refer-
ence to the current conguration and u the displacement from the natural to the current conguration. Then, we can write:
ut u u o : 21
Usually, when a nite element procedure is used, the mesh is dened in the reference conguration (the set of points at time
t = 0) as depicted schematically in Fig. 5.
Let us consider that X and xo are coordinate systems in the reference and natural congurations, respectively. As a con-
sequence of the assumption of small deformations introduced to describe motion, and by assuming the existence of the
intermediate deformation gradient in the neighborhood of every point, we can approximate : rXxo  I. The same assumption
allows us to evaluate the strain e rsym
xo u, related to the motion from the natural to the nal congurations, by the following
approximation:
e rsym sym
xo u  rX u: 22
By taking gradients in Eq. (21) and using the assumptions stated above, we can verify the validity of the additive decompo-
sition of strains:
et e eo ; 23
where et rsym t
X u is the strain tensor at the actual conguration (time t) with respect to the reference conguration, and
eo rsym
X uo
is the strain at the natural conguration with respect to the reference one.
For a correct representation of the steel constitutive behavior in the zone of interest, it is mandatory to account for the
strong dependence of the material parameters on temperature.
Remarks:

 The strain eo is computed the rst time that all nodal temperatures of the considered element fall below the zero strength
temperature (ZST), and is stored at the nite element Gauss point as an additional tensorial internal variable. The Zero
Strength Temperature is dened as the minimum temperature amongst those at which strength is zero [49]. In this work,
ZST is set equal to the metal solidus temperature.
 In this model, the solid phase domain changes with time. This fact introduces some difculties concerning the mesh def-
inition for the FE analysis. The procedure we have implemented consists in dening a xed mesh that describes the com-
plete domain, including the liquid and mushy zones. Nodes in the liquid and mushy zones are initially xed. In
subsequent time steps, when the nodal temperatures fall below the ZST, nodal d.o.f.s are freed, and the stiffness contri-
bution of the solidied zone is taken into consideration (see Fig. 6).
 The described numerical model does not predict the real displacement eld of the zones that have melted because the
displacement eld of the natural conguration is not recorded.
 The technique of subtraction of the strain term eo from the total strain et must be emphasized for its simplicity and accu-
racy. The technique is validated in the examples by comparison with results obtained by using the semi-analytical for-

Fig. 5. Reference (B), natural (B0) and current (Bt) body congurations in the melting/solidication problem.
696 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

Fig. 6. Evolution of d.o.f.s and strains in melting/solidication problems.

mulation of Weiner and Boley [39]. Note also that this approach is less expensive than those used by other authors, which
are based on recording the ow strain for liquid elements [20,50].
 All elements internal variables, included the eo strain, are set to zero as soon as the element temperature rises above ZST.
 Mechanical elements have a special treatment during the liquid/solid and solid/liquid phase changes: while the temper-
ature is above the ZST, the corresponding elemental d.o.f.s are not included in the system of equations for the unknowns
of the mechanical problem. This methodology incorporates an advantage compared to other techniques found in litera-
ture (e.g. assigning a very low stiffness to the elements that are in liquid/mushy phases brings poor conditioning of the
algebraic equation system to be solved).

To avoid restructuring the stiffness matrix at each time step, liquid mechanical d.o.f.s are uncoupled from the rest by
zeroing the corresponding rows and columns, and putting a non-zero term in the diagonal entry of the stiffness matrix. Also,
the corresponding position in the residue is zeroed.
Fig. 7(a) shows the initial stiffness matrix distribution when the entire domain is solid. Fig. 7(b)(d) show an increase in
the uncoupled rows as the solid domain decreases its size.

3.2. Lagrangian formulation of the constitutive equations

According to the local state theory [51], for a given time instant t the thermodynamic state at any particle X of a material
medium is completely dened by the values of a certain number of state variables at this particle. Computations in inelastic
materials with strain-driven formulations, use as state variables the strain e and a set of phenomenological internal strain-
type variables describing material history, together with the temperature eld T. Since linear kinematics is assumed, the to-
tal strain can be additively decomposed as follows:

e ee ei ; 24
e i
e being the thermoelastic (reversible) strain and e the inelastic (irreversible) strain. Either term may play the role of an
internal variable, but ei is typically chosen (option followed in this work).
 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 697

(a) (b)

(c) (d)

Fig. 7. Matrix sparsity evolution.

We adopted the hypothesis of isotropic hardening, which is widely accepted in welding applications [48,52,53]. Isotropic
hardening is characterized using a single internal variable a, which in this case is the equivalent inelastic strain:

Z t r
2 i
a ke_ sk ds; 25
0 3
q
where e_ i is the inelastic strain rate and ke_ i k e_ iij e_ iij its L2-norm.
Although driving variables lie in the strain space, response functions (i.e. the yield criterion and the evolution laws) are
usually written in terms of their conjugated thermodynamic forces: the stress tensor r (dual of ee) and the isotropic hard-
ening variable in stress space, R = R(a).
The stress tensor r depends on e and ei through the decomposition (24). For linearly-elastic isotropic materials, the stress
is dened by the state law:

r jtree  3eTher  I 2l devee ; 26

|{z} |{z}
p s

where j = j(T) and l = l(T) are thermo-dependent material properties known as bulk and shear moduli, respectively, eTher is
the thermal strain, I the second-order unit tensor, tree eeii and dev(ee) = ee  tr(ee)I/3 are the trace and the deviator of the
second-order tensor ee, p = tr(r)/3 is the mean stress, and s = dev(r) is the stress deviator. Here, the thermal expansion is
dened by the thermal linear expansion (TLE) function:
698 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

Z T
eTher TLET aT s ds; 27
T ref

with aT the linear thermal expansion coefcient and Tref an arbitrary reference temperature.
The von Mises criterion is dened:
r
2
f ksk  rY Ra; 28
3
with rY denoting the initial yield stress and R(a) the hardening relation.
Associated to this yield criterion, the following J2 ow rule is considered:

e_ i cn; 29

with n = s/ksk being the normalized stress deviator dening the normal to the Von Mises yield surface f = 0 in the deviatoric-
stress space, and c P 0 the consistency parameter. For plastic materials, c is determined by means of the consistency
condition

cf_ 0: 30

Finally, having chosen the equivalent inelastic strain as hardening variable, the ow rule (29) completely denes the
hardening law:
r r
2 i 2
a_ ke_ k c: 31
3 3

3.3. Integration of the evolution equations

Following Simo and Hughes [47], the evolution laws (29) and (31) are discretized using the implicit backward-Euler -
nite-difference scheme. Then, given the total strain increment De at the particle X during the time interval [tn, tn+1],
tn+1 = tn + Dt, the material state at X is updated from the previous instant tn to the current one tn+1 by a standard return-map-
ping algorithm. The consistent tangent stiffness matrix was implemented, which is essential to achieve good numerical re-
sponse in the determination of equilibrium condition.

3.4. Finite element implementation

Incompressible material behavior may lead to some difculties in numerical simulation, such as volumetric locking, inac-
curacy of solution, checkerboard pattern of stress distributions, or occasionally, divergence. Mixed up elements, with both
displacements and hydrostatic pressure as primary unknown variables, are used to overcome these problems.
By ignoring inertial effects, the momentum balance equation can be written as:

r  r qb 0 in X; 32

subjected to the following boundary conditions

 on @ Xu ;
uu 33

r  n t on @ Xt : 34

The standard quasi-static boundary value problem in solid mechanics, stated in mixed up form [54], consists in nding
the displacement eld u and the mean-stress eld p that satisfy the integral equations
Z Z Z Z
s grad w dV p div w dV qb  w dV t  w dS; 35
X X X @ Xt
Z h
pi
tr eM  q dV 0; 36
X j

throughout the domain X, for all the admissible displacement and mean-stress weighting functions w and q, respectively.
Eq. (35) is the weak form of the momentum balance equations, where qb is the body-force (per unit volume) and t is the
traction prescribed over the portion oXt of the boundary (displacement boundary conditions over the complementary por-
tion oXu are assumed to hold a priori).
 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 699

The displacement and mean-stress trial functions are dened as follows:

X
nu
u Ni U i NU; 37
i1
Xnp
p Npj Pj N p P; 38
j1

Ni being the displacement shape function associated to the displacement node i = 1, 2, . . ., nu, Ui the nodal displacement, Npj
the mean-stress shape function associated to the mean-stress node j = 1, 2, . . ., np, and Pj the nodal mean stress.
An appropriate choice of the interpolation functions should be made in order to deal with the numerical difculties caused
by the inelastically-incompressible behavior of metals. We refer to the classical nite element literature [54,40] for a detailed
discussion of admissible u- and p-shape functions that guarantee the fulllment of the BabukaBrezzi stability conditions. In
our tests, we used linear interpolation functions for the displacements and constant functions for the pressure eld.
After replacing u and p by their respective nite element approximations, and by adopting the corresponding shape func-
tions as weight functions (Galerkin formulation), it yields

R1 F int  F ext 0; 39
R2 K Tp U  M p P 0; 40

where
Z
F int BT s dV K p P; 41
X
Z Z
F ext N T qb dV N T t dS; 42
X @ Xt
Z
Kp BT IN p dV; 43
X
Z
1 T
Mp N p N p dV: 44
X j
In the above equations, symmetric second-order tensors like the stress deviator s and the unit tensor I are mapped into vec-
tors, as is usually done in the nite element practice; B is the nite element matrix dening the straindisplacement kine-
matics relation et = BU.
Since the material is only nearly incompressible and the pressure variables are considered at the elemental level only, we
can statically condense out the pressure terms and express the element matrices in terms of displacements only.
As pointed out by Cifuentes and Kalbag [55] and Benzley et al. [56], hexahedral elements are superior to linear tetrahedral
elements, and also better than quadratic tetrahedron elements when plastic deformation occurs. Therefore, we used q1p0
hexahedral elements in our tests.
The discrete equilibrium equations are solved by using a standard Newton-Raphson method. The Jacobian matrix corre-
sponds to that obtained from an equivalent purely Lagrangian elasto-plastic quasi-static incremental problem (see for in-
stance Simo and Hughes [47]).

3.5. Validation of the model

Thermal stresses calculated by the nite element model developed in the present work have been compared with the
thermal stresses computed in the semi-analytical solution developed by Weiner and Boley [39] for one-dimensional solid-
ication of a semi-innite elastic-perfectly plastic body after a sudden decrease in surface temperature (Fig. 8).
The thermal problem they solved corresponds to the Neumann classical phase change 1D problem with uniform initial
temperature Ts (solidication temperature) and xed temperature Tw < Ts on one end, with To = Ts  Tw being dened as
the temperature drop.
They took an elastic-perfectly plastic material model for the mechanical problem, with a constant Youngs modulus E and
yield stress ry varying linearly with temperature:
Ts  T
ryT roy ; T 6 Ts; 45
To
where roy is the yield stress at temperature To. p
 p t, where t denotes time and the parameter p is given by:
The Neumann solution introduces a characteristic length y
r
p j T oCp
p 2 K s c; Ks ; c c2  1: 46
qC p 2L
The thermal diffusivity Ks is dened as the ratio of conductivity j to density q and specic heat Cp, while L is the latent
heat.
700 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

Fig. 8. Weiner and Boley [39] problem description.

Fig. 9. Finite element model for the 1-D numerical validation test.

^-coordinate. Comparison of semi-analytical [39] and FEM numerical solutions.

Fig. 10. Stress distribution along the y

Weiner et al. also introduced the dimensionless quantities:

y T  Ts
^ ;
y Tb ;

y To
1  mr 1  mry 1  mroy
r^ ; r^y m Tb ; m ;
aT ET o aT ET o aT ET o
where aT is the thermal expansion coefcient and m the Poisson ratio.
The problem has been solved in a prismatic domain by imposing Extended Plane Strain (EPS) conditions (ex = constant,
ez = constant) along transversal directions z and x to be consistent with the original hypothesis proposed by Weiner et al.
 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 701

The FE mesh consists of 200 hexahedral elements and the prism width, height and length are 1 mm, 1 mm and 25 mm,
respectively (Fig. 9). The EPS conditions are imposed via Lagrange multipliers, constraining all nodal x- and z-displacements
to be identical.
Fig. 10 compares the semi-analytical results with the numerical ones. The curves plot the non-dimensional stress com-
ponent r ^ xx r ^-line (longitudinal direction) for m = 0.097 and c = 0.465. The excellent agree-
^ zz along the non-dimensional y
ment of the numerical FE solution with the semi-analytical one is evident from the gure.
It should be noted that this semi-analytical solution gives a maximum tensile (compressive) stress which is constant in
time. This behavior is due to the particular thermal boundary conditions adopted (which correspond to a similarity solution)
in the mentioned semi-analytical study.

4. Thermalmechanical problem

Due to the weak nature of mechanical to thermal eld coupling, at each time step, the solution of the non-linear transient
problem is divided into two parts. First, a thermal analysis is performed to predict the temperature history of the whole do-
main, and then the temperature eld is applied as input for the subsequent mechanical analysis, in a staggered approach.
As pointed out in Section 3, hexahedral nite elements are used to solve the mechanical problem. On the other hand, the
thermal problem is solved using tetrahedral elements. In order to share the same nodes in both problems, rst a mesh of
hexahedra for the mechanical model is generated. Then by splitting each hexahedron in six tetrahedra, the thermal nite
element mesh is obtained.

5. Application examples

5.1. Residual stresses in fusion welding

Residual stresses associated with thermal welding processes are of particular importance because they are frequently of
the same order of magnitude of yield stresses. Fatigue failure of welded structures occurs usually at the weld site, along or
adjacent to the weld toe. Residual stresses and strains may also approach yield magnitude in the transverse direction to the
weld.
The magnitude of the residual stresses and their distribution depend on numerous factors, including the type of welding,
number of passes, material properties and degree of constraint during the process. A material that is rigidly constrained will
have greater residual stresses than one that is allowed to distort during the process.
During a thermal welding process, the weld site and immediate surrounding area experience different rates of heating/
cooling and thus expansion/contraction, leading to large thermal strains. Due to the localized nature of heat application, the
expansion is constrained by the cool material away from the site of the applied heat.
A stress rz parallel to the direction of the weld bead is called longitudinal residual stress. A high tensile longitudinal resid-
ual stress arises along the weld line near the weld toe, and then decreases to zero, nally becoming compressive, as distance
from the weld line on the ange increases.
A stress rx normal to the direction of the weld bead is known as transverse residual stress. Normally a tensile transverse
residual stress is produced at the surface of the specimen near the weld and decreasing close to zero as the distance from the
weld toe increases.

5.2. Bead-on-plate welding

In this example a heat source moves at constant speed and along a straight line over the welded plate, like in a bead-on-
plate weld without ller material. A 3-dimensional symmetrical model was used to estimate the residual stresses using the
nite element method.
The sample dimensions are shown in Fig. 11. The x, y and z-axes represent the width, thickness and length of the plate,
respectively. The coordinate y = 0 corresponds to the top surface of the sample and x = 0 corresponds to the symmetry plane.
The symmetric mesh model employs 49,200 eight-node hexahedral elements with 59,094 nodes. The heat affected zone
(HAZ) was rened locally.
The material properties used in the model correspond to a CMn steel. The plate is clamped at the corners opposite to the
location of the welding line. Natural convection in the top and bottom surfaces of the plate is assumed, with a heat transfer
coefcient of 45 W/m2 K. The welding parameters are summarized in Tables 4 and 5.

5.2.1. Simulation results

The process takes about 100 s to complete the weld. The time stepping has a minimum time step of 0.25 s during welding.
The complete analysis took about 110 hs of CPU, on a 2 Dual Core AMD Opteron 1.8 GHz processor with 8 GB of memory.
Fig. 12 shows the temperature distribution of the plate near the fusion zone at time 52.5 s. The locally concentrated heat
source produced a complete fusion through the thickness and made the temperature near the weld bead rapidly change with
the distance from the center of the heat source.
702 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

Fig. 11. Plate dimensions and mechanical boundary conditions.

Table 4
Arc parameters.

Arc parameter Symbol Value

Current I 250 A
Voltage U 23 V
Efciency g 0.825
Total heat Q 4743.75 W
Speed v 4.8 mm/s

Table 5
Goldaks parameters.

Goldak parameter Value

a 5 mm
b 3 mm
cf 5 mm
cr 10 mm
ff 0.6
fr 1.4

Fig. 12. Fusion zone temperature eld [K] at time t = 52.50 s.

A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 703

Fig. 13. rx stress [Pa] at t = 52.50 s.

Fig. 14. rz stress [Pa] at time t = 52.50 s.

Fig. 15. rx stress distribution of along z for different times.

704 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

Plots of the stress distribution near the weld center are shown in Figs. 13 and 14 at time 52.50 s after the welding has
started. In Figs. 1517, 18 changes in the longitudinal and transverse stresses at the mid-thickness of the workpiece at dif-
ferent times are plotted.
The rx stress map (Fig. 13) shows that a compressive stress is formed in front of the fusion zone, whereas a tensile stress is
appearing at the back of it, mainly due to the thermal expansion effects. The longitudinal stress rz has a similar behavior, as
shown in Fig. 14.

Fig. 16. rz stress distribution of along z for different times.

Fig. 17. rx stress distribution of along x for different times.

Fig. 18. rz stress distribution of along x for different times.

 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 705

Fig. 19. Residual stress distribution at t = 300 s in a longitudinal section.

Fig. 20. Residual stress distribution at t = 300 s in a transversal section.

A compressive stress eld is formed near the high temperature region (see rx1 and rz1 in Figs. 1518), whereas a tensile
stress is formed in the relatively low temperature region.
As time progresses, the temperatures in the hot weld center line region decreases; whereas temperatures in the sur-
rounding cold region start to increase due to heat conduction from the fusion zone. This causes the metal close to the weld
central line region begin to contract, whereas in the surrounding region the metal expands. A gradually increasing tensile
stress is formed (see rx3 and rz3 in Figs. 1518).
For the transverse stresses (Figs. 17 and 18), one can see that compressive stresses are predicted at the centerline (rx1)
with tensile stresses appearing in the newly solidied metal (rx3).
Figs. 19 and 20 show calculated longitudinal (rz) and transverse (rx) residual stresses. The stress distributions are plotted
along the mid-thickness line in the x = 0 mm and y = 0.25 mm planes. A tensile stress state in the longitudinal direction (rz4)
is shown in Figs. 19 and 20. The maximum stresses are in the longitudinal direction, and have the same order of magnitude
as the yield stress of the material. The transverse residual stress (rx4) proles show a considerable gradient near the weld
line ends, mainly due to boundary effects.

6. Conclusions

In the present article, nite element models have been used to analyze the thermal and mechanical phenomena observed
in welding processes. Thermal histories and residual stresses have been predicted.
The described methods provide a powerful means to optimize process parameters off-line. The welding model can be
used to predict and minimize deformations due to changes in the weld sequence during the product development stage.
A mathematical model was implemented to represent the power density distribution of the external heat source. Results
obtained in this work using the heat source model are in close agreement with the original results obtained by the exper-
imental data of Christensen et al. [38].
A standard nite element formulation limited to the solid domain has been adopted for the mechanical analysis, proving
to be effective. Regarding the material description, a standard model with isotropic hardening has proven to well character-
ize the behavior of steel at high temperatures.
706 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707

In relation to the methodology described in Section 3.1 the melting/solidifying treatment of mechanical elements has pro-
ven to be satisfactory in validating the thermo-mechanical analysis of phase-change problems with the semi-analytical solu-
tion of Weiner and Boley [39] as described in Section 3.5. This is a solid base from which to continue software development
with the ability to accurately simulate complex welding processes.
The nite element models described were applied to solve the nal residual stress state of a butt-welded joint. The results
observed have good agreement with those obtained by Camilleri and Gray [57].
Simulation of welding problems is very cpu time consuming and requires important computer resources. The process can
be seen as a multi-scale one, which involves the description of localized phenomena coupled with phenomena at the large
scale. Future research will be addressed to the implementation of strategies to deal both with the multi-physics and multi-
scale characters of the problem in a parallel computer environment, as already proposed by Cross et al. [58].

Acknowledgements

We acknowledge support from European Community, Contract AST 5-CT 2006-030953, Project RAPOLAC (Rapid Produc-
tion of Large Aerospace Components), CONICET (Consejo Nacional de Investigaciones Cientcas y Tcnicas, Argentina), and
UNL (Universidad Nacional del Litoral, Argentina).

References

[1] J. Mackerle, Finite element analysis and simulation of welding-an addendum: a bibliography (19962001), Model. Simul. Mater. Sci. Eng. 10 (2004)
295318.
[2] Y. Ueda, T. Yamakawa, Analysis of thermal elasticplastic stress and strain during welding, Trans. Jpn. Weld. Soc. 2 (2) (1971) 90100.
[3] P. Marcal, Weld Problems, Structural Mechanics Programs, University Press, Charlottesville, 1974. pp. 191206.
[4] L. Karlsson, Thermal stresses in welding, in: R.B. Hetnarski (Ed.), Thermal Stresses, vol. 1, Elsevier Science Publishers, 1986, p. 300.
[5] L. Karlsson, Thermomechanical fnite element models for calculation of residual stresses due to welding, in: Hauk et al. (Eds.), Residual Stresses,
DGMInformationsgesellschaft Verlag, 1993, p. 33.
[6] J. Goldak, A. Oddy, M. Gu, W. Ma, A. Mashaie, E. Hughes, Coupling heat transfer, microstructure evolution and thermal stress analysis in weld
mechanics, in: IUTAM Symposium on the Mechanical Effects of Welding, Springer-Verlag, 1992, pp. 130.
[7] R. Radaj, Finite element analysis of welding residual stresses, in: Proceedings of 2nd International Conference on Residual Stresses (ICRS-2), 1988, p.
510.
[8] R. Dexter, Residual Stress Analysis in Reactor Pressure Vessel Attachments Review of Available Software, Electric Power Research Institute, EPRI NP-
7469s, 1991.
[9] U. Chandra, Determination of residual stresses due to girth-butt welds in pipes, ASME J. Pressure Vessel Technol. 107 (1985) 178184.
[10] G. Ravichandran, V.P. Raghupathy, N. Ganesan, R. Krishnakumar, Prediction of axis shift distortion during circumferential welding of thin pipes using
the nite element method, Weld. J. (1997) 39s55.
[11] L. Karlsson, Modeling in Welding, Hot Powder Forming, and Casting, ASM International, 1997.
[12] J. Song, J. Peters, A. Noor, P. Michaleris, Sensitivity analysis of the thermomechanical response of welded joints, Int. J. Solids Struct. 40 (2003) 4167
4180.
[13] T. Branza, A. Duchosal, G. Fras, F. Deschaux-Beaume, P. Lours, Experimental and numerical investigation of the weld repair of superplastic forming dies,
J. Mater. Process. Technol. (2004) 16731680.
[14] C. Cho, G. Zhao, S.-Y. Kwak, C.B. Kim, Computational mechanics of laser cladding process, J. Mater. Process. Technol. (2004) 494500.
[15] T.H. Hyde, A.A. Becker, Y. Song, W. Sun, Failure estimation of TIG butt-welded Inco718 sheets at 620 C under creep and plasticity conditions, Comput.
Mater. Sci. 35 (2006) 3541.
[16] Z. Hou, I.-S. Kim, Y. Wang, C. Li, C. Chen, Finite element analysis for the mechanical features of resistance spot welding process, J. Mater. Process.
Technol. 185 (13) (2007) 160165.
[17] D. Deng, H. Murakawa, Numerical simulation of temperature eld and residual stress in multi-pass welds in stainless steel pipe and comparison with
experimental measurements, Comput. Mater. Sci. 37 (2006) 269277.
[18] P. Duranton, J. Devaux, V. Robin, P. Gilles, J.M. Bergheau, 3d modelling of multipass welding of a 316l stainless steel pipe, J. Mater. Process. Technol.
(2004) 457463.
[19] D. Gery, H. Long, P. Maropoulos, Effects of welding speed, energy input and heat source distribution on temperature variations in butt joint welding, J.
Mater. Process. Technol. 167 (2005) 393401.
[20] M. Abid, M. Siddique, Numerical simulation to study the effect of tack welds and root gap on welding deformations and residual stresses of a pipe-
ange joint, Int. J. Pressure Vessels Piping 82 (2005) 860871.
[21] M. Mochizuki, Control of welding residual stress for ensuring integrity against fatigue and stress-corrosion cracking, Nucl. Eng. Des. 237 (2) (2007)
107123.
[22] P. Mollicone, D. Camilleri, T.F. Gray, T. Comlekci, Simple thermo-elasticplastic models for welding distortion simulation, J. Mater. Process. Technol.
172 (2006) 7786.
[23] P. Ferro, H. Porzner, A. Tiziani, F. Bonollo, The inuence of phase transformations on residual stresses induced by the welding process 3D and 2D
numerical models, Model. Simul. Mater. Sci. Eng. 14 (2006) 117136.
[24] D. Deng, H. Murakawa, Prediction of welding residual stress in multi-pass butt-welded modied 9Cr 1Mo steel pipe considering phase transformation
effects, Comput. Mater. Sci. 37 (2006) 209219.
[25] M.M. Mahapatra, G.L. Datta, B. Pradhan, N.R. Mandal, Three-dimensional nite element analysis to predict the effects of SAW process parameters on
temperature distribution and angular distortions in single-pass butt joints with top and bottom reinforcements, Int. J. Pressure Vessels Piping 83 (10)
(2006) 721729.
[26] R. Nandan, G.G. Roy, T.J. Lienert, T. DebRoy, Numerical modelling of 3D plastic ow and heat transfer during friction stir welding of stainless steel, Sci.
Technol. Weld. Join. 11 (5) (2006) 526537.
[27] B. Hu, I.M. Richardson, Mechanism and possible solution for transverse solidication cracking in laser welding of high strength aluminium alloys,
Mater. Sci. Eng. A 429 (2006) 287294.
[28] H. Alberg, D. Berglund, Comparison of plastic, viscoplastic, and creep models when modelling welding and stress relief heat treatment, Comput.
Methods Appl. Mech. Eng. 192 (2003) 51895208.
[29] V. Fachinotti, A. Cardona, Constitutive models of steel under continuous casting conditions, J. Mater. Process. Technol. 135 (2003) 3043.
[30] J. Che, S.L. Ceccio, G. Tryggvason, Computations of structures formed by the solidication of impinging molten metal drops, Appl. Math. Model. 28
(2004) 127144.
 A. Anca et al. / Applied Mathematical Modelling 35 (2011) 688707 707

[31] G. Taylor, M. Hughes, N. Strusevich, K. Pericleous, Finite volume methods applied to the computational modelling of welding phenomena, Appl. Math.
Model. 26 (2002) 309320.
[32] K. Pericleous, M. Hughes, N. Strusevich, Modelling the metal arc weldpool, in: D. Stefanescu, J. Warren, M. Jolly, M. Krane (Eds.), Proceedings of the
Tenth International Conference on Modeling of Casting, Welding and Advanced Solidication Processes, 2003, pp. 523530.
[33] W. Wang, Y. Shi, Y. Lei, Z. Tian, FEM simulation on microstructure of DC ash butt welding for an ultra-ne grain steel, J. Mater. Process. Technol. 161
(2005) 497503.
[34] A. De, T. DebRoy, Improving reliability of heat and uid ow calculation during conduction mode laser spot welding by multivariable optimisation, Sci.
Technol. Weld. Join. 11 (2) (2006) 143153.
[35] X. He, J.T. Norris, P.W. Fuerschbach, T. DebRoy, Liquid metal expulsion during laser spot welding of 304 stainless steel, J. Phys. D: Appl. Phys. 39 (2006)
525534.
[36] G.G. Roy, J.W. Elmer, T. DebRoy, Mathematical modeling of heat transfer, uid ow, and solidication during linear welding with a pulsed laser beam, J.
Appl. Phys. 100 (034903) (2006).
[37] M. zisik, J. Uzzell, Exact solution for freezing in cylindrical symmetry with extended freezing temperature range, J. Heat Transfer 101 (May) (1979)
331334.
[38] N. Christensen, V.d.L. Davies, K. Gjermundsen, Distribution of temperatures in arc welding, British Weld. J. 12 (1965) 5475.
[39] J. Weiner, B. Boley, Elasto-plastic thermal stresses in a solidifying body, J. Mech. Phys. Solids 11 (1963) 145154.
[40] O. Zienkiewicz, R. Taylor, The Finite Element Method, The Basis, 5th ed., vol. 1, Butterworth-Heinemann, 2000.
[41] V. Fachinotti, A. Cardona, A. Huespe, A fast convergent and accurate temperature model for phase-change heat conduction, Int. J. Numer. Methods Eng.
44 (1999) 18631884.
[42] V. Fachinotti, A. Cardona, A. Huespe, Numerical simulation of conductionadvection problems with phase change, Latin Am. Appl. Res. 31 (1) (2001)
3136.
[43] A. Anca, A. Cardona, J. Risso, 3D thermo-mechanical simulation of welding processes, in: G. Buscaglia, E. Dari, O. Zamonsky (Eds.), Mecnica
Computacional, XIV Congreso de MTtodos NumTricos y sus Aplicaciones (Enief 2004), vol. XXIII, 2004, pp. 23012318.
[44] J. Goldak, A. Chakravarti, M. Bibby, A new nite element model for welding heat sources, Metall. Trans. B 15 (1984) 299305.
[45] V. Fachinotti, A. Anca, A. Cardona, Analytical solutions of the thermal eld induced by moving double-ellipsoidal and double-elliptical heat sources in a
semi-innite body, Communications in Numerical Methods in Engineering, in press. doi:10.1002/cnm.1324.
[46] P. Tekriwal, J. Mazumder, Transient and residual thermal strain-stress analysis of GMAW, ASMEJ Eng. Mater. Technol. 113 (1991) 336343.
[47] J. Simo, T. Hughes, Computational Inelasticity, Springer-Verlag, New York, 1998.
[48] L.-E. Lindgren, Finite element modelling and simulation of welding. Part 2: Improved material modelling, J. Therm. Stress. 24 (3) (2001) 195231.
[49] T. Nakagawa, T. Umeda, J. Murata, Y. Kamimura, N. Niwa, Deformation behavior during solication of steels, ISIJ Int. 35 (6) (1995) 723729.
[50] A. Yaghi, A. Becker, State of the Art Review Weld Simulation Using Finite Element Methods, Tech. Rep., University of Nottingham, UK.
[51] J. Lematre, J. Chaboche, Mechanics of Solid Materials, Cambridge University Press, 1994.
[52] M. Francis, S. Rahman, Probabilistic analysis of weld cracks in center-cracked tension specimens, Comput. Struct. 76 (2000) 483506.
[53] P. Chang, T. Teng, Numerical and experimental investigations on the residual stresses of the butt-welded joints, Comput. Mater. Sci. 29 (4) (2004) 511
522.
[54] T. Hughes, The Finite Element Method, Linear Static and Dynamic Finite Element Analysis, Prentice-Hall, 1987.
[55] A. Cifuentes, A. Kalbag, A performance study of tetrahedral and hexahedral elements in 3-d nite element structural analysis, Finite Elem. Anal. 12
(1992) 313318.
[56] S. Benzley, E. Perry, K. Merkley, B. Clark, G. Sjaardama, A comparison of all hexagonal and all tetrahedral nite element meshes for elastic and elasto-
plastic analysis, in: Proceedings of 14th Ann. Int. Meshing Roundtable, Albuquerque, USA, 1995.
[57] D. Camilleri, T.G.F. Gray, Computationally efcient welding distortion simulation techniques, Model. Simul. Mater. Sci. Eng. 13 (2005) 13651382.
[58] M. Cross, T. Croft, A. Slone, A. Williams, N. Christakis, M. Patel, C. Bailey, K. Pericleous, Computational modelling of multi-physics and multi-scale
processes in parallel, Int. J. Comput. Methods Eng. Sci. Mech. 8 (2) (2007) 6374.