You Sung Han1

e-mail: yousung_han@kaist.ac.kr

Kyehyung Lee1
e-mail: mech9676@kaist.ac.kr
Department of Mechanical Engineering,
KAIST,
Science Town,
DaeJeon 305-701, South Korea

Myoung-Soo Han
Industrial Application R&D Institute,
Daewoo Shipbuilding and Marine Eng. Co. Ltd.,
1 Ajoo-dong,
Geoje-si,
Gyoungnam 656-220, South Korea
e-mail: mshan@dsme.co.kr

Hyunchil Chang
Department of Machinery Design Research,
Hyundai Maritime Research Institute,
1 Jeonha-dong,
Dong-gu,
Ulsan 682-792, South Korea
e-mail: lucky7@hhi.co.kr

Kanghyouk Choi
Department of CEM R&E Project,
700 Gumho-dong,
Gwangyang-si,
Jeonnam 545-711, South Korea
e-mail: khchoi76@posco.com

Seyoung Im2

Finite Element Analysis of

Welding Processes by Way of
Hypoelasticity-Based Formulation
Welding is one of the most important joining processes, and the effect of welding residual
stresses in a structure has a great deal of influence on its quality. In spite of such a key
interest, the analysis of a welding process has not been successful as in a structural
analysis. This is partially because welding involves complex phenomena that are manifested by the phase evolution and by thermomechanical processes as well. In the present
study, a hypoelasticity-based formulation is applied to welding processes to determine
residual deformation and stresses. Algorithmic consistent moduli for elastoplastic deformations including transformation plasticity are also obtained. Leblonds phase evolution
equation, coupled with the energy equation, is employed to calculate the phase volume
fraction; this plays an important role as a constitutive parameter reflecting phase fraction
effects in a mechanical constitutive equation. Furthermore, transformation plasticity is
taken into account for an accurate evaluation of stress. The influence of the phase transformation and the transformation plasticity on residual stress is investigated by means of
numerical analyses using metallurgical parameters in Leblonds phase evolution equation that are adjusted with respect to various cooling rates in a CCT-diagram. Coding
implementation is conducted by way of the ABAQUS user subroutines, DFLUX, UEXPAN, and
UMAT. The numerical examples demonstrated that the phase transformation and the
transformation plasticity have a significant effect on the residual stress of a welded
structure. DOI: 10.1115/1.4003099
Keywords: finite element analysis, welding, residual stress, transformation plasticity,
phase transformation, hypoelasticity-based plasticity model, ABAQUS user subroutine,
UMAT

Department of Mechanical Engineering,

KAIST,
Science Town,
DaeJeon 305-701, South Korea
e-mail: sim@kaist.ac.kr

Introduction

An analysis of a welding process poses a challenge to researchers as welding involves complex phenomena such as the interaction of thermal, mechanical, and metallurgical behaviors. For example, phase transformation in steels is governed by the
temperature history. Conversely, during a phase transformation,
the latent heat influences the actual temperature field. Thermal
loading applied to a structure induces deformation. In addition,
each phase transformation is a source of deformation, mainly due
to density variations and the degree of transformation plasticity.
In any steel, iron atoms are arranged in a certain cell structure.
This arrangement can be a body-centered cubic bcc structure or
a face-centered cubic fcc structure. Ferritic steel -phase:
ferrite/pearlite, bainite, or martensite has a bcc structure at room
temperature. On the other hand, austenite -phase, which exists
1

These authors contributed equally to this work.

Corresponding author.
Contributed by the Materials Division of ASME for publication in the JOURNAL OF
ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received September 16, 2009;
final manuscript received September 23, 2010; published online March 3, 2011.
Assoc. Editor: Somnath Ghosh.
2

above the critical eutectoid temperature, 727 C for plain-carbon

steel, is arranged in a fcc structure. In ferritic steel, carbon atoms
can be dissolute between iron atoms or can have a direct chemical
binding to iron atoms. Over a certain temperature, the iron atoms
switch into an austenitic arrangement, and the iron atoms in the
ferritic steel are therefore rearranged into a fcc structure. During
cooling back to the original temperature, austenitic steel transforms into the phases of ferrite/pearlite, bainite, and martensite
depending on the cooling rate. For high cooling rates, all carbon
atoms are trapped during the transformation into ferrous cells,
with very few carbon atoms able to diffuse out of these cells. This
type of ferrous cell is thus maximally distorted and maximally
hard. This phase is termed martensite. For lower cooling rates, on
the other hand, the carbon atoms essentially diffuse out of the
ferrous cell depending on the cooling rate. They result in the
ferrite-pearlite and/or bainite phases, again depending on the cooling rate. Austenite has a higher specific volume than the ferrous
phases. Therefore, phase transformation has to be taken into account 1.
During the transformation from austenite to ferrous phases, under mechanical loading, plasticity can occur at a stress level much

Journal of Engineering Materials and Technology

Copyright 2011 by ASME

APRIL 2011, Vol. 133 / 021003-1

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

lower than the macroscopic yield stress of the phases. This effect
is known as transformation plasticity. It is attributed to two different mechanisms 2.
1. The GreenwoodJohnson mechanism: The volume difference between two coexisting phases generates microscopic
plasticity in the weaker phase that with the lower yield
stress; this permits macroscopic plastic flow in the presence
of an external load even if this load alone would be insufficient to induce macroscopic classical plasticity.
2. The Magee mechanism for martensitic transformations: If
a martensitic transformation takes place under an external
loading, martensite plates are formed with a preferred orientation. This affects the overall shape of the body. In other
words, when a small region is transformed, its shape
changes, and these microscopic shape changes do not average out to be zero when macroscopic stress is applied.
In recent decades, a number of works have been conducted to
predict welding residual stress. Leblond and Devaux 3 proposed
a kinetic model for metallurgical transformation in steel using a
semi-inverse method based on a CCT-diagram of the material.
This drew an amount of substantial attention in the welding community because it retained a well-founded micromechanics basis
for phase evolution.
Regarding the martensitic transformation, Koistinen and Marburger 4 proposed that the martensite proportion depends only
on the temperature and that this is valid for most steels. Goldak
et al. 5 suggested a double ellipsoidal distribution model and
showed that it was superior to other heat source models by comparing the temperature field results with the experimental results.
Papazoglou and Masubuchi 6 considered the effect of metallurgical phase transformation on the residual stress using only the
average thermal expansion coefficients of austenite, bainite, and
martensite. However, this method could not ensure an accurate
solution yet as the transformation plasticity was not considered.
An alternative method proposed by Leblond et al. utilized a
thermo-elastoplastic model with transformation plasticity 7,8,
and this approach is deemed to accurately predict residual stresses
in welding.
Recently, Dean and Hidekazu 9 investigated the influence of a
martensitic transformation on the evolution of the residual stress
distribution in a butt-welded pipe using a thermo-elastoplastic FE
model that took into account the phase transformation employing
the KoistinenMarburger equation. However, transformation plasticity and other phase transformations such as austenitic, bainitic,
and ferritic transformations, which have an influence on the temperature and stress field during the welding process, were not
taken into account in their study. Kim et al. 10 conducted a
numerical implementation of the thermo-elastoplastic constitutive
equation proposed by Leblond and Devaux 2 and Leblond 11
based on a hyperelastic assumption considering transformation
plasticity. The implementation was verified through comparisons
using the commercial software SYSWELD.
The present study focuses on obtaining a systematic finite element implementation of the constitutive equation of Leblond considering transformation plasticity. Borrowing the key idea in the
hypoelastic formulation based upon additive decomposition, an
efficient implementation of the stress-update procedure is proposed, and a calculation of tangent moduli accounting for the
constitutive equation that includes transformation plasticity is formulated. The hypoelastic formulation is often employed in finite
element analyses with small elastic strain mainly due to its conceptual simplicity 12. It is accurate enough to represent Hookes
law up to the leading order as long as the elastic strains remain
small. In addition, the influence of the phase transformation and
the transformation plasticity on the residual stress has been investigated using the metallurgical parameters in Leblonds phase evolution equation adjusted with respect to various cooling curves in
a CCT-diagram.
021003-2 / Vol. 133, APRIL 2011

The outline of the present paper is as follows: First, a brief

review is given in Sec. 2 of metallurgical, mechanical, and thermal analyses during the welding process. A kinetic equation of
metallurgical transformation 3, the energy equation, the constitutive equation by Leblond et al. 7,8, and a double ellipsoidal
heat model by Goldak et al. 5 are also discussed in this section.
In Sec. 3, the hypoelastic formulation that considers transformation plasticity is discussed along with the associated finite element
implementation including numerical integration and the consistent
tangent modulus. Furthermore, calibration with the CCT-diagram
is given for the metallurgical analysis. This will make it possible
to reflect the actual metallurgical phenomenon by considering the
phase transformation with respect to various cooling rates. In Sec.
4, numerical examples are provided to demonstrate the validation
of the present implementation and to investigate the influence of
the phase transformation and the transformation plasticity induced
on the residual stress during the welding process.

2 Formulation of a Welding Process for ThermoMetallurgical and Mechanical Analyses

2.1 Metallurgical Analysis. In this section, a metallurgical
analysis of the welding process is summarized briefly. Kim et al.
10 contains more detailed information.
The kinetic equation of metallurgical transformation in steel
proposed by Leblond and Devaux 3 is applied to calculate the
phase volume fraction. This is expressed by the following equations:
N

pi =

AijT,T,

i = 1,2, . . . ,N

j=1,ji

Here, pi is the volumetric proportion of phase i, N is the total

number of phases, and T is the Celsius temperature. The rate of
phase proportion i transformed from phase j is written as
AijT , T. In this study, it is assumed that up to four different
phases can be present in the material. In addition, p1, p2, p3, and
p4 represent the phase portion of ferrite-pearlite, bainite, martensite, and austenite, respectively. At any instant in time, the sum of
all phase volume fractions should be unity:
4

p =1
i

for t 0

i=1

Therefore, i and j in Eq. 1 both range from 1 to 4. Details related

to AijT , T can be found in Leblond and Devaux 3. The trapezoidal rule is employed to integrate Eq. 1, and the obtained
phase volume fractions are used to solve the energy equation and
the mechanical constitutive equation.
For a martensitic transformation, the KoistinenMarburger
model may be used as follows:
p3T = 1 expaT3,S T

T T3,S

lnp1T3,S + p2T3,S
with a =
T3,S T3,F

here T is used to denote the current temperature in the Celsius

scale, while T3,s , T3,F are the Celsius temperatures to start and
finish the transformation from austenite into martensite, respectively. In this formula, the martensite proportion depends only on
the current temperature.
The energy equation taking into account phase transformation
and the thermal boundary conditions can be expressed as follows:

p c dt + p H = p T
dT

i i

i i

Transactions of the ASME

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 1 The values of the functions f and g these functions

should be linearly interpolated between the indicated points
8

fza

gza

0
0.125
0.25
0.5
0.75
1.0

0
0.044
0.125
0.391
0.668
1

0
2.53
4
2.76
1.33
1

ent plastic flow rules are prescribed separately: transformation

plasticity y, where is equivalent stress and macroscopic
plasticity = y. See Refs. 2,11 for additional details. For
transformation plasticity y, the flow rules can be written by
introducing the internal variable p, as given below:

p =

p =

Note: fz and gz are the modification functions in Eqs. 9 and 12, respectively.

T
=q
n

on Sq

T
= 1T T0 + 2T4 T40
n

on S

2.2 Mechanical Analysis. The thermo-elastoplastic constitutive equations including transformation plasticity proposed by
Leblond et al. 7,8 are introduced in this section. They divide a
mixture of phases in steel into two categories: the weak phase
-phase and the austenitic phase in steel, as indicated by subscript 1, and the hard phase -phase containing ferrite, bainite,
or martenisite, as indicated by subscript 2. Despite the different
microstructures, it is normal practice to assume an isotropic material behavior for simplicity in computational welding mechanics.
In the welding process, the thermo-metallurgical strain thm is
induced during the phase transformation. It is expressed as follows:

thm

= 1

zth
1 T

z2th
2 T

th
i

Here,
and z denote the thermal strains of each phase and the
phase volume fraction of hard phase 2, respectively. The thermal
strain th
i of each phase is calculated from the secant value of the
thermal expansion coefficient specified as a function of
temperature.
The yield strength of the hard phase -phase is calculated as
the average of the yield strengths of the individual ferritic phases
with a law of linear weighting, i.e.,

2y T =

piiy

eff
eff
1 ,2 ,T

= 1

fz1y T,eff
1

fz2y T,eff
2

Here, fz represents a modification function for the nonlinear

mixture rule Table 1, and 1y and 2y are the yield strengths of the
weak and the hard phase, respectively.
Leblonds model distinguishes between two regimes according
to whether the macroscopic equivalent stress does or does not
reach the yield stress of the phase mixture. Therefore, two differJournal of Engineering Materials and Technology

3s
2

10

3 1 z eff
1 s
2 1y
eff
1

11

3 gz
s
2 E
12

z
z eff
eff
+ eff
2 =
z 1
z 2

13

= 23 s:s

14

th
12

is the thermal strain difference between the two

Here,
eff
phases, s indicates the deviatoric stress, hs , eff
1 ,
2 is a correction function that considers the nonlinearity of the stress, as
given in Appendix B, gz is a modification function of 1 / z to
consider small z values Table 1, E is the Youngs modulus, 1
and 2 are the thermal expansion coefficients of phases 1 and 2,
respectively, is the memory coefficient, which describes the
recovery effect it takes a value of 1 for the martensitic transformation and 0 for the other transformations. Additional details
regarding are available in Leblond 11.
For macroscopic plasticity = y, the following equations
hold:

p =

3 p
s
2

15

= y

16

eff
p
1 =

17

z
z
eff
2p eff
+ eff
2 =
z 1
z 2

18

eff
Here, y depends upon eff
1 , 2 , z, and T, as indicated in Eq. 9.

2.3 Heat Source Modeling. In this study, the heat from the
welding arc is applied as a volumetric heat source with the double
ellipsoidal model proposed by Goldak et al. 5. Hence, the power
density distribution inside the front quadrant becomes
Qx,y,z,t =

where pi and iy denote the phase portion and the yield strength of
phase i, respectively. The yield strength of the phase mixture is
expressed with a law of nonlinear mixture rule as follows:

with n =

21 2z ln z
T
1z

where i is the density of phase i, and ci, Hi, and i are the specific
heat, the enthalpy, and the thermal conductivity of phase i, respectively. Sq is the surface where the external heat flux is applied, and
S is the surface where the convection or the radiation boundary
condition is prescribed. Moreover, q is the heat flux on Sq, and 1,
2, and T0 denote the convection heat transfer coefficient, the
StephanBoltzmann constant multiplied by emissivity, and the
surrounding Kelvin temperature, respectively. Note that the
Kelvin temperature is employed here due to the radiation boundary condition, while the Celsius temperature will be used otherwise.

3
n
2

th
212
eff
eff
ln zzhs,
1 ,
2 +
1z

eff
1 =

3 p
s=
2

63f fQ

a f bc

exp

3x2
a2f

3y 2 3z2
2
b2
c

19

Similarly, for the rear quadrant of the heat source, the power density distribution inside the ellipsoidal becomes
Qx,y,z,t =

63f rQ

arbc

exp

3x2
ar2

3y 2 3z2
2
b2
c

20

where x, y, and z are the local coordinates of the double ellipsoidal model, and ff and fr are parameters that give the fraction of the
heat deposited in the front and the rear parts, respectively. In the
present study, we choose f f + f r = 2; f f is assumed to be 1.4, and f r
is assumed to be 0.6 5. This choice is adopted because the temperature gradient in the front part is steeper than that in the rear
APRIL 2011, Vol. 133 / 021003-3

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

and strain tensors are known at the time tn. To satisfy the principle
of material frame indifference, a rotation-neutralized strain measure is introduced by way of a co-rotational approach. In this
approach, the rotation-neutralized rate of deformation is expressed
as follows:
= RT D R = 1 U
U1 + U1 U

D
2

22

is the rotation-neutralized rate of deformation.

In this formula, D
Additionally, R and U denote the rotation tensor and the right
stretch tensor for the incremental deformation from the previous
configuration n to the current configuration n+1, respectively.
These values can be obtained from the polar decomposition of the
incremental deformation of gradient Finc, as expressed by

xn+1
= Rn+1 Un+1
xn

Finc =

Fig. 1 Double ellipsoidal heat source configuration 5

part 5. Q represents the power of the welding torch. It can be

calculated from the welding current, the arc voltage, and the welding efficiency. The parameters a f , ar, b, and c are related to the
characteristics of the welding heat source, and they are described
in Fig. 1; these parameters can be adjusted to create a desired
temperature history by comparison with the experimental data. In
the absence of experimental data, Goldak et al. 5 suggested that
it is reasonable to take the distance a f in front of the heat source
equal to one-half of the weld width and the distance ar behind the
heat source equal to twice the width. These suggestions are implemented in this study.

3.1 Basic Formulation. In this study, small elastic strains,

compared with the plastic strains, are assumed. Thus, the hypoelastic formulation can adequately describe the realistic elastic
material responses. In addition, only an isotropic material is considered, and the J2 flow model based on the von Mises yield
surface is applied. It is also assumed that the elastic properties
apart from the thermal expansion coefficients remain invariant
with respect to the phase transformation. This leads to efficiency
and simplicity. As discussed in Eq. 7, for thermal expansion
coefficients, the mixture rule is applied to take into account the
phase transformation. This enables us to treat a mixture of the
weak and hard phases as an equivalent single homogeneous phase.
Constitutive equations within the framework of thermoelastoplasticity are used under the additive decomposition of the
rate of deformations in conjunction with the hypoelastic formulation. In the hypoelastic formulation, the trial stress is to be defined, and a rotation neutralization procedure is required to satisfy
material objectivity. The first step concerns the decomposition of
the strain rate into elastic, plastic, and thermo-metallurgical parts.
The additive decomposition of the strain rate tensor is expressed
as follows:
21

where , , , and represent the total strain rate, the

elastic strain rate, the transformation plasticity strain rate, and the
conventional plastic strain rate, respectively. Furthermore, thm
denotes the thermo-metallurgical strain, as defined in Eq. 7.
It is initially assumed that the entire incremental deformation is
a purely elastic process and that the end state obtained under this
assumption is the elastic trial state, which is denoted by superscript trial. Then, it is assumed that the relaxation from the trial
state occurs due to a plastic flow such that the final stress state
may satisfy the yield condition. It is assumed that integration has
been conducted up to the previous time step, tn. Therefore, stress
tot

el

tp

The rotation-neutralized strain increment RN between the reference configuration n and the current configuration n+1 is now
obtained by integrating the rotation-neutralized rate of deforma in the time interval t , t , as follows:
tion D
n n+1

dt = RN
D

24

tn

In this study, the stress-update procedure is conducted, utilizing

the rate form or the hypoelastic type, from the known strain increment during the incremental configuration. First, it is assumed
that pl = 0, which implies an elastic process. Hence, the trial
stress of the elastic predictor is expressed as follows:
trial
n+1
= n + Cel: RN + C:el
n

25

is the trial stress at the current time tn+1, referred to

where
the previous configuration n; Cel, RN, and el
n denote the
fourth-order elastic tangent moduli, the total rotation-neutralized
strain increment, and the elastic strain at the time tn, respectively.
Here, in the welding process the temperature increment is high
enough to affect the material properties, and both of the two terms
C : and C : should be considered in the calculation of the
trial stress.
The plastic deformation is divided into two categories, as discussed in Sec. 2.2; one is the conventional macroscopic plasticity
and the other is the transformation plasticity. Regardless of the
type of the plasticity, the first step to integrate the constitutive
equation is to obtain the trial stress of Eq. 25. The subsequent
step is now to account for the effect of the plastic relaxation, and
this should be implemented in a different way depending on
whether the type of plastic deformation at hand is the conventional macroscopic plasticity or the transformation plasticity.
3.2 Conventional Macroscopic Plasticity. In the conventional plasticity, material behaviors for the elastoplastic constitutive model are classified into two regimes: elastic deformation
trial
trial
and plastic deformation y n+1
, where y is the
y n+1
trial
yield stress of the mixture, as defined in Eq. 9, and n+1
is the
trial
trial
y
n+1. For elastic deformation n+1
, the
von Mises stress of
following equations apply:
trial
n+1 =
n+1

cp

021003-4 / Vol. 133, APRIL 2011

tn+1

trial
n+1

Integration of the Constitutive Equation

tot = el + thm + tp + cp

23

ipl = 0

pl
pl
i.e., n+1,i
= n,i
i = 1,2

26
27

trial
n+1

n+1 denote the trial stress in Eq.

and
In these equations,
25 and the stress tensor at time tn+1 with reference to the previpl
and pl
ous configuration n. Furthermore, n+1,i
n,i represent the
plastic strain tensors of the hard or the weak phase at tn+1 and tn,
respectively.
trial
, radial-return mapping 13
For plastic deformation y n+1
pl
is used to find and i , where and pl
i denote the algoTransactions of the ASME

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

rithmic plastic parameter and the plastic strain increment of the

hard or the weak phase, respectively. The procedure for finding
is covered in the next paragraph. These values are used to
update the stress and strain, as in the expressions
n+1 =

trial
n+1

Cn+1:

pl
pl
n+1,1
= pl
1,
n,1 +

28

trial
trial
/s n+1
= s n+1/s n+1
Nn+1 = s n+1

31

32

where R is the rotation tensor obtained by Eq. 23.

The first step toward finding starts from taking the inner
product of each side of Eq. 28 with Nn+1, given by Eq. 30. This
leads to the following expression:
trial
eff
eff
s n+1 = s n+1
2n+1n+1 = 2/3y
1 n+1,
2 n+1

33
where the internal variables
follows:

and

eff

2 n+1

are updated as

2
eff
eff

1 n+1 =
1 n + 3 n+1
zn+1
zn+1
eff

2 n+1 =
zn+1
1+
zn+1
1+

2
n+1 +
3

34

zn+1 eff
n

zn+1 1
zn+1
1+
zn+1

eff

2 n +

21 2zn+1 ln zn+1
3 gz
T
sn+1 sn +
2 E
1 zn+1
39

eff

2 n+1 =

zn+1 eff
z
zn+1
1 n n+1
eff

2 n
zn+1
zn+1
zn+1
eff
+

1 n+1
zn+1
zn+1
1+
1+
zn+1
zn+1
40

where the temperature and the phase proportions are the known
variables from the thermo-metallurgical analysis. Note that
s n+1 = sn+1 indicates the magnitude of the deviatoric stress, and
eff
it is now given in terms of
1 n+1 from Eqs. 36, 37, and
eff
40. This is combined with Eq. 39 to find sn+1 and
1 n+1.
For this, the iterative solution procedure of the Newton type is
now applied to the resulting nonlinear equation.
3.4 Consistent Tangent Modulus. It is assumed that transformation plasticity takes place only during the cooling process 2.
Therefore, all cases are categorized into either the conventional
macroscopic plasticity or transformation plasticity cases. The tangent moduli are discussed for each case separately. Following the
standard linearization process, we can show that the consistent
tangent moduli from the hypoelastic formulation are written as
follows.
For macroscopic plasticity,

n+1
= I I + 2 62
C
n+1
n+1
n+1 trial
n+1
2
6n+1

35
Radial-return mapping is used to update the stress and strain, and
the Euler backward scheme is utilized to integrate the other internal state variables.
3.3 Transformation Plasticity. The phase transformation
taking place during welding processes induces transformation
plasticity. This phenomenon stands out in the elastic regime
wherein the stress has not reached the yield stress, while it is often
neglected in the elastic-plastic regime as it is relatively small compared with the conventional macroscopic plasticity.
trial
n+1
, as given
For the transformation plasticity, the trial stress
by Eq. 25, is relaxed due to the transformation plasticity to reach
the final stress state inside the yield surface. Accordingly, the first
equality in Eq. 33 is valid for the transformation plasticity as
well; i.e., we have
trial
eff
eff
2n+1n+1 2/3y
s n+1 = s n+1
1 n+1,
2 n+1

36
Note that the yield condition was employed to determine the plastic parameter for the conventional macroscopic plasticity, as
depicted in Sec. 3.2. However, the yield condition is not fulfilled,
as shown in Eq. 36, for the transformation plasticity. Instead, we
need to employ the constitutive equations for the transformation
plasticity, as given by Eqs. 1114,
Journal of Engineering Materials and Technology

th
212
eff
ln zn+1hsn+1,
1 n+1z
1 zn+1

30

pl
pl
n+1,2
= R n+1,2
RT

eff

1 n+1

38

29

where n+1 denotes the shear modulus of a material and Nn+1 is a

trial
and s n+1
vector normal to the von Mises yield surface, and s n+1
indicate the deviatoric part of the elastic trial stress and the deviatoric stress at time n + 1. The term 2n+1n+1Nn+1 represents the
plastic corrector, which adjusts the trial stress onto a suitably up n+1 and the
dated yield surface. After finding the stress tensor
pl
pl
plastic strain tensors n+1,1
, n+1,2
, the transformation law presents
them in the current configuration as follows:

pl
pl
n+1,1
= R n+1,1
R T,

eff

3
1 n+1
eff
K
1 n+1 = 1 zn+1 y
2
eff
1
1 n+1
eff

1 n+1 =

pl
pl
n+1,2
= pl
2
n,2 +

n+1 R
n+1 = R

37

where

pl

trial
n+1 = n+1

2n+1n+1Nn+1

eff
n+1 = K
1 n+1sn+1

pl

y
pl + 3n+1
n+1

For transformation plasticity,

1
II I I
3

pl
n+1

trial Nn+1 Nn+1 41

n+1

2n+1A
1
= I I + 2
C
Nn+1
n+1
n+1 II I I 2n+1
3
1 + 2n+1A

Nn+1

2
4n+1
trial n+1
s n+1

1
II I I Nn+1 Nn+1
3

42

where n+1 and n+1 denote the shear modulus and the bulk
modulus, and A in Eq. 42 is expressed as follow:
A=

h
1
+b
a
G sn+1

K 1y
K

sn+1 + K
eff
eff
1y

1
1

43

where
G=1a

a =

h
zn+1/zn+1
eff n+1
1 1 + zn+1/zn+1

th
212
n+1
ln zn+1zn+1
1 zn+1

b=

3 gzn+1
2 En+1

44

45

46

APRIL 2011, Vol. 133 / 021003-5

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

kij =

PeqT
,

lij =

1 PeqT

50

where PeqT and are the parameters to be determined; the

former is the equilibrium proportion, which is reached at a constant temperature after an infinitely long time, having a value
between 0 and 1, and the latter represents the characteristic time
of transformation, which is necessarily positive.
With the assumption that PeqT may be written as the ratio of
the temperature increment from Ts to the temperature interval
T f , Ts, the following equation holds:
PeqT =

21 2n+1zn+1 ln zn+1
c = bsn +
Tn+1
1 zn+1

Cijkl = RimR jnRkpRlqC

mnpq

48

Here, R is the rotation tensor obtained by the polar decomposition, as in Eq. 23.
3.5 Calibrating for a CCT-Diagram. It is well known that
the phase transformation affects the thermal and mechanical
analyses in a welding simulation. Hence, it is important to perform the metallurgical analysis accurately. AijT , T in Eq. 1 denotes the algebraic proportion of phase i, which is transformed
into phase j per unit of time Aij 0 for an i j transformation
and Aij 0 for a j i transformation; A ji = Aij 3. Leblond and
Devaux 3 assumed that the i j transformation is not affected
by phases k, k i and k j, and is given in terms of pi and p j by
an following expression:

Aij =

i j transf.,

k jiTp j + l jiTpi
if k jiTp j l jiTpi 0

j i transf.,

0 if kijTpi lijTp j 0
and k jiTp j l jiTpi 0
no transformation between phases i and j

49

Here, kij and lij are the metallurgical parameters, and they are
required to be calibrated with a CCT-diagram for the metallurgical
analysis. This section discusses a means of calibrating the metallurgical parameters kij and lij in Eq. 1.
Figure 2 shows a CCT-diagram for EH36-TM steel, a commonly used material in the shipbuilding industry. The diagram
shows the phase portions of ferrite, bainite, and martensite in a
ferritic transformation with respect to the temperature for various
cooling rates. Leblond and Devaux 3 defined the metallurgical
parameters kij and lij as
021003-6 / Vol. 133, APRIL 2011

dP Peq P
=
dt

47

For additional details, one may refer to Appendix A, and Kim et

al. 10 is a good source of additional information regarding the
derivation of the moduli from the hyperelastic formulation.
The tangent modulus in Eq. 41 or Eq. 42 is defined in the
reference configuration n. The tangent modulus Cijkl referred to
the current configuration n+1 is now obtained from the following
transformation:

if kijTpi lijTp j 0

51

Here, Ts and T f denote the temperatures to start and finish the

ferritic transformation, respectively. Here, can denote the time
necessary for the phase to reach an equilibrium proportion. The
derivative of the phase may be defined as Eq. 52, and the phase
at time tn+1 can be approximated using Taylors expansion in the
form of Eq. 53,

Fig. 2 CCT-diagram for EH36 steel

kijTpi lijTp j

Ts T
Ts T f

Pn+1 = Pn +

dP
dt

dT
dt

T = Pn +

52

Peq P dT

dt

T
53

In these equations, Pn+1 and Pn are the phases at the time tn+1 and
tn, respectively, dT / dt is the cooling rate, and T is the temperature difference between tn+1 and tn. Calibration with the CCTdiagram was conducted by finding by an iterative computation
of Eq. 53 such that the phase at the temperature T f may be in
agreement with that from the diagram. To assess the accuracy of
calibrating the metallurgical parameters, comparisons of the results from the numerical simulation utilizing the calibrated parameters with those directly read from the CCT-diagram are discussed
in Sec. 4.

Numerical Results

This section discusses the numerical results from the welding

simulation. The algorithm discussed above was implemented in
the user subroutines DFLUX, UEXPAN, and UMAT for the ABAQUS
code: DFLUX is employed to define the position, velocity, and heat
flux of the welding torch. The thermal-metallurgical strain considering the phase transformation is defined in UEXPAN. UMAT is
implemented to define a materials mechanical behavior from the
hypoelastic formulation in consideration of the transformation
plasticity.
4.1 Calibration Test. To explore the accuracy of the calibrated metallurgical parameters discussed in Sec. 3.5, the numerical simulation results using the parameters were compared with
those from the CCT-diagram. The parameters PeqT and were
calibrated from the CCT-diagram of EH36 steel in Fig. 2 with
respect to the 11 different given cooling rates. Tables 2 and 3
illustrate the functions of the ferrite-pearlitic and the bainitic
transformations. PeqT can be obtained from Eq. 51 Tables 2
and 3. As discussed above, this study uses Eq. 3 for the martensitic transformation. Tables 46 show the temperatures to start
and finish the phase transformation as obtained from the numerical simulation and the diagram. The tables show that the present
results are in good agreement with those from the diagram, showing that the maximum temperature differences from the two results are less than 5 C. Table 7 shows the phase volume fractions
at the final state in both cases, showing the good agreement between the two results as well.
Transactions of the ASME

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 2 Metallurgical parameter for ferritic transformation

during cooling

Cooling rate
C / s

Ts a
C

Tf b
C

0.98
1.74
2.58
5.08
10.71
16.72
20.46
29.35
41.67
65.29
84.39

699.8
689.0
678.9
651.2
624.1
612.7
603.3
587.8
567.9
c

581.2
570.4
567.8
562.2
555.5
545.8
550.1
546.7
543.7

0.95
0.001
0.01
0.007
0.9775
1.05
1.95
1.8
1.055

Temperature to start the ferritic transformation.

Temperature to finish the ferritic transformation.
No ferritic transformation occurs in the given cooling rate.

b
c

Table 5 The temperatures to start and finish the bainitic transformation, which are obtained from the numerical simulation
and the CCT-diagram of EH36-TM steel
Cooling rate
C / s

Present
resulta Ts

CCT-diagramb
Ts

Present
resultc Tf

CCT-diagramd
Tf

0.98
1.74
2.58
5.08
10.71
16.72
20.46
29.35
41.67
65.29
84.39

581.2
570.4
567.8
562.2
555.5
545.8
550.1
546.7
543.7
537.0
526.4

581.9
571.4
569.8
564.7
556.1
552.6
551.0
547.5
544.0
538.9
526.8

508.0
501.2
494.2
484.0
461.7
446.3
424.2
454.3
452.0
448.9
451.1

508.0
501.2
494.2
478.9
461.7
446.3
424.0
453.1
451.4
451.4
450.0

The temperature to start the transformation, obtained from the numerical simulation.
The temperature to start the transformation from a CCT-diagram in Fig. 2.
The temperature to finish the transformation, obtained from the numerical simulation.
d
The temperature to finish the transformation from a CCT-diagram in Fig. 2.
b
c

4.2 Cantilever Beam Under Transformation Plasticity. In

this example, to assess the performance of the numerical implementation, nodal shear forces are applied to the tip of a cantilever
beam so that the beam is subjected to transformation plasticity
undergoing a stress level below the macroscopic flow stress.
These results were compared with those from the commercial
Table 3 Metallurgical parameter for bainitic transformation
during cooling

Cooling rate
C / s

Ts a
C

Tf b
C

0.98
1.74
2.58
5.08
10.71
16.72
20.46
29.35
41.67
65.29
84.39

581.2
570.4
567.8
562.2
555.5
545.8
550.1
546.7
543.7
537.0
526.4

508.0
501.2
494.2
484.0
461.7
446.3
424.2
454.3
452.0
448.9
451.1

1010
1010
1010
1010
0.005
0.01
1010
0.44
0.91
0.787
1.82

Temperature to start the bainitic transformation.

Temperature to finish the bainitic transformation.

code, SYSWELD. Figure 3 shows the geometry of the model and the
thermal history in which the entire domain is applied. While the
thermal strain in SYSWELD is computed based on the temperature
at the corresponding integration point, ABAQUS uses the average
temperature in an element to calculate the thermal strain 1,14.
Hence, in this example, the temperature at all nodes is prescribed;
see Fig. 3b for the comparison. The material properties are given
in Tables 812, and the metallurgical parameters are for A 508
cl.3 steel, which can be found in Leblond and Devaux 3. Forty
elements were employed, termed here as C3D8T, which is an
eight-node trilinear element for a fully coupled thermalmechanical analysis 14. The boundary conditions are given as
follows:
ux = u y = uz = 0
ty = 500 N

at x = 0 m
at x = 10 m

Figure 4 shows the phase evolution results of the given temperature history. It was found that the present results are in good
agreement with those from SYSWELD. The stress history in element
A, plotted in Fig. 5, also shows excellent agreement between the
two results.

Table 4 The temperatures to start and finish the ferritic transformation, which are obtained from the numerical simulation
and the CCT-diagram of EH36-TM steel

Table 6 The temperatures to start and finish the martensitic

transformation, which are obtained from the numerical simulation and the CCT-diagram of EH36-TM steel

Cooling rate
C / s

Present
resulta Ts

CCT-diagramb
Ts

Present
resultc Tf

CCT-diagramd
Tf

Cooling rate
C / s

Present
resulta Ts

CCT-diagramb
Ts

Present
resultc Tf

CCT-diagramd
Tf

0.98
1.74
2.58
5.08
10.71
16.72
20.46
29.35
41.67
65.29
84.39

699.8
689.0
678.9
651.2
624.1
612.7
603.3
587.8
567.9
e

700.3
690.0
679.6
652.1
624.7
614.5
604.2
588.6
568.2

581.2
570.4
567.8
562.2
555.5
545.8
550.1
546.7
543.7

581.9
571.4
569.8
564.7
556.1
552.6
551.0
547.5
544.0

0.98
1.74
2.58
5.08
10.71
16.72
20.46
29.35
41.67
65.29
84.39

452.8
452.0
448.9
451.1

453.1
451.4
451.4
450.0

326.5
303.8
283.1
271.6

326.1
303.8
283.1
271.2

The temperature to start the transformation, obtained from the numerical simulation.
The temperature to start the transformation from a CCT-diagram in Fig. 2.
c
The temperature to finish the transformation, obtained from the numerical simulation.
d
The temperature to finish the transformation from a CCT-diagram in Fig. 2.
e
No ferritic transformation occurs in the given cooling rate.
b

Journal of Engineering Materials and Technology

The temperature to start the transformation, obtained from the numerical simulation.
The temperature to start the transformation from a CCT-diagram in Fig. 2.
The temperature to finish the transformation, obtained from the numerical simulation.
d
The temperature to finish the transformation from a CCT-diagram in Fig. 2.
e
No martensitic transformation occurs in the given cooling rate.
b
c

APRIL 2011, Vol. 133 / 021003-7

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 7 The phase volume fractions of EH36-TM steel at the final state, which are obtained
from the numerical simulation and the CCT-diagram
Ferrite-pearlite

Bainite

Martensite

Cooling rate
C / s

Present result

CCT-diagram

Present result

CCT-diagram

Present result

CCT-diagram

0.98
1.74
2.58
5.08
10.71
16.72
20.46
29.35
41.67
65.29
84.39

0.990
0.993
0.994
0.999
0.841
0.708
0.450
0.302
0.227
0.000
0.000

0.990
0.995
0.990
0.990
0.840
0.712
0.450
0.305
0.227
0.000
0.000

0.010
0.003
0.003
0.001
0.159
0.292
0.549
0.594
0.480
0.519
0.159

0.000
0.005
0.010
0.010
0.160
0.288
0.550
0.598
0.479
0.521
0.160

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.104
0.293
0.481
0.840

0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.097
0.294
0.479
0.840

4.3 Butt Welding of a Thin Plate. This example is a simulation of a butt welding process. To inspect the influence of the
phase transformation and the transformation plasticity on the residual stress, two different cases are studied: one considering the
phase transformation and transformation plasticity and the other
disregarding them. A total of 9400 C3D8T elements are used, as
shown in Fig. 6, and the material properties are identical apart
from the metallurgical parameters between the two simulations.
The parameters in Tables 13, 2, and 3 are employed in this example because these properties can reflect the phase transformation with respect to various cooling rates. Due to the symmetry

along the welding line, one-half of the model is taken into account
for the computational domain, and the symmetric thermal and
mechanical boundary conditions are imposed on the symmetric
surface. The welding heat inputs are given as Eqs. 19 and 20.
The speed of the welding torch, the welding current, the arc voltage, and its efficiency are 3 mm/s, 1170 A, 32.9 V, and 0.95,
Table 8 Thermal properties
T
C

Specific heat
J kg1 C1

Conductivity
W mm1 C1

Density
106 kg mm3

0
15
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
850
900
950
1000
1050
1100
1510

480

483
496
512
525
544
565
586
611
642
682
726
768
816
1139
1190
843
693

648
652

0.0519
0.0519
0.0515
0.0511
0.0498
0.0486
0.0465
0.0444
0.0435
0.0427
0.0410
0.0393
0.0377
0.0356
0.0339
0.0318
0.0285
0.0259
0.0260
0.0264
0.0268
0.0272
0.0280
0.0285
0.0120

7.863
7.859
7.849
7.834
7.819
7.803
7.787
7.775
7.753
7.736
7.718
7.699
7.679
7.659
7.635
7.617
7.625
7.624
7.616
7.604
7.602
7.548
7.522
7.496
7.493

Table 9 Youngs modulus and Poissons ratio

Fig. 3 a The model of the deflection test. b The thermal

history of the deflection test.

021003-8 / Vol. 133, APRIL 2011

T
C

Youngs modulus
GPa

Poissons ratio

0
300
500
600
625
875
1000
1510

206
179
158
118
108
5.9
5.1
0.1

0.3
0.354
0.39
0.408
0.413
0.458
0.48

Transactions of the ASME

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 10 Yield stress y and thermal strain th

T
C

yf a
MPa

by a
MPa

my a
MPa

ay a
MPa

th
1

20
700
1300

425
110
12.5

425
110
12.5

750
210
12.5

220
98
12.5

0.025

th
2

0.0115

0.0225

Note: f = ferrite-pearlite, b = bainite, m = martensite, and a = austenite.

Subscripts 1 and 2 denote the austenitic phase and the ferritic phase, respectively.

respectively. The radiation and the convection boundary conditions are given on the external surface, with a surrounding temperature of T0 = 20 C, a convection heat transfer coefficient of
1 = 2.0 105 W / mm2, and the StephanBoltzmann constant
multiplied by the emissivity of 2 = 2.84 1014 W / mm2. Only

Table 11 Hardening coefficient H pl , T of the austenitic

phase

pl

0.1

H MPa T = 20 C
H MPa T = 700 C
H MPa T = 1300 C

0
0
0

118.2
16.6
8.75

Table 12 Hardening coefficient H pl , T of the ferritic phase

pl

0.1

H MPa T = 20 C
H MPa T = 700 C
H MPa T = 1300 C

0
0
0

290
200
8.75

the rigid body motion is restricted without any external force.

Element A in Fig. 6 is chosen to investigate the influence of the
phase transformation and the transformation plasticity on the residual stress in the welding processes. The temperature and the
phase evolution history at element A are plotted in Fig. 7 The
histories of the von Mises stress and the major stress component
S22 at element A for each of the two cases, wherein the phase
transformation and the transformation plasticity are considered
and neglected, respectively, are compared in Fig. 8 The stress
differences between two results are found to occur upon the instigation of the phase transformation. Figure 9 shows the von Mises
stress and the major stress component S22 at the final state along
the welding line. Whether the phase transformation and the transformation plasticity are accounted for or not, the stress along the
welding line is nearly 46% different in the von Mises stress and
41% in the major stress component S22 on average. Figure 10
shows the von Mises stress and the major normal stress component S22 in the direction perpendicular to the welding line at the
final state. Stress differences between considering and neglecting
the two effects remarkably occur in the heat affected zone. The

Fig. 4 The phase evolution of the given temperature history: a pearlite-ferrite, b bainite, c
martensite, and d austenite

Journal of Engineering Materials and Technology

APRIL 2011, Vol. 133 / 021003-9

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 5 The stresses in the element A: a stress S11, b stress S12, c stress S13, d stress S22,
e stress S23, and f stress S33

phase distribution when the time reaches 64 s is plotted over

ferrite-pearlite and austenite in Fig. 11. The part contoured with
different colors indicates the regions in which phase transformation takes place, and these regions correspond to where the stress
differences occur in Fig. 10. This implies that the phase transformation and transformation plasticity induced by it have a significant influence on the residual stress. For this reason, these factors
should be taken into account in welding simulation for accuracy
and reliability.

evolution equation were adjusted with a CCT-diagram with respect to various cooling rates. Furthermore, the influence of a
phase transformation and the transformation plasticity induced by
it on the residual stress was investigated by means of a numerical
example. The example demonstrated that these two factors have a
significant effect on the residual stress of a welded structure.

Acknowledgment
5

Concluding Remarks

In this study, a numerical implementation of a thermoelastoplastic constitutive equation taking into account the phenomenon of transformation plasticity was conducted for a welding
simulation. The framework of a hypoelastic formulation was employed based on the additive decomposition of the Eulerian strain
rate. The consistent tangent modulus was derived in consideration
of the transformation plasticity. In order to reflect actual metallurgical phenomena, the metallurgical parameters in Leblonds phase
021003-10 / Vol. 133, APRIL 2011

The study in this paper is part of a two-year research project

from Daewoo Shipbuilding and Marine Engineering DSME entitled The development of solid-shell element and its implementation into ABAQUS for analyzing welding-residual deformation in
large scale structures. The authors thank DSME for the financial
support. This research was also supported by the National Research Laboratory NRL program through the National Research
Foundation of Korea NRF funded by the Ministry of Education,
Science and Technology Grant No. R0A-2007-000-20115-0.
Transactions of the ASME

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 13 Metallurgical parameters kij , lij for austenitic transformation during heating
T
C

k14 , k24 , k34 a s1

l14 , l24 , l34 a s1

750

800

900

1
0.8

2
0.2

10
0

a
Subscripts 1, 2, 3, and 4 denote ferrite-pearlite, bainite, martensite, and austenite,
respectively.

Appendix A: The Consistent Tangent Modulus for the

Transformation Plasticity
For transformation plasticity, one can recall Eq. 37, and we
have the following relations:

sn+1
n+1 K
=
sn+1 +
K

A1

Fig. 6 The finite element meshes of the butt welding process

and element A

Fig. 7 a The temperature history in element A. b The phase

evolution history in element A.

Journal of Engineering Materials and Technology

Fig. 8 The history of stresses with considering the phase

transformation and transformation induced plasticity TRIP or
not. a The von Mises stress in element A. b The major stress
component S22 in element A.

APRIL 2011, Vol. 133 / 021003-11

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 9 The history of stresses at the final state, considering

the phase transformation and transformation induced plasticity
or not. a The region that is considered along the welding
line. b The von Mises stress. c The major stress component
S22.

eff
eff
K
K
K
K 1y
1
1
=
=
eff
eff
eff y
1
1
1
1

After some algebraic manipulations, one can obtain
021003-12 / Vol. 133, APRIL 2011

Fig. 10 The history of stresses at the final state, considering

the phase transformation and transformation induced plasticity
or not. a The region that is considered perpendicular to the
welding line. b The von Mises stress. c The major stress
component S22.

eff

h
sn+1
1
1
+b
a
=
G sn+1

A2

A3

where
Transactions of the ASME

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

heff
1 n+1,sn+1

if z 0.03

if

1 + 3.5

23 sn+1

eff
1 n+1

1
2

if

3
1
sn+1

2 yeff
2

1 n+1

3
sn+1
1
2 yeff
1 n+1
B1

References

Fig. 11 The phase distribution at t = 64 s. a The phase fraction of ferrite-pearlite. b The phase fraction of austenite.

G=1a

h
h zn+1/zn+1

a
eff

eff
1
2 1 + zn+1/zn+1

A4

By plugging Eqs. A2 and A3 into Eq. A1, one can obtain

sn+1
n+1
=A


where
A=

h
1
+b
a
G sn+1

A5

K
K 1y

sn+1 + K A6
eff
eff
1y

1
1

Appendix B: The Definition of h

The definition of h in Eq. 12 is defined as

Journal of Engineering Materials and Technology

1 ESI Group, 2009, SYSWELD Reference Manual.

2 Leblond, J. B., and Devaux, J. C., 1989, Mathematical Modeling of Transformation Plasticity in Steels I: Case of Ideal-Plastic Phase, Int. J. Plast., 5, pp.
551572.
3 Leblond, J. B., and Devaux, J. C., 1984, A New Kinetic Model for Anisothermal Metallurgical Transformations in Steels Including Effect of Austenite
Grain Size, Acta Metall., 321, pp. 137146.
4 Koistinen, D. P., and Marburger, R. E., 1959, A General Equation Prescribing
the Extent of the Austenite-Martensite Transformation in Pure Iron-Carbon
Alloys and Plain Carbon Steels, Acta Metall., 7, pp. 5960.
5 Goldak, J., Chakravarti, A., and Bibby, M., 1984, A New Finite Element
Model for Welding Heat Sources, Metall. Trans. B, 15, pp. 299305.
6 Papazoglou, V. J., and Masubuchi, K., 1982, Numerical Analysis of Thermal
Stress During Welding Including Phase Transformation Effects, ASME J.
Pressure Vessel Technol., 104, pp. 198203.
7 Leblond, J. B., Mottet, G., and Devaux, J. C., 1986, A Theoretical and Numerical Approach to Plastic Behavior of Steels During Phase
TransformationsI. Derivation of General Relations, J. Mech. Phys. Solids,
344, pp. 395409.
8 Leblond, J. B., Mottet, G., and Devaux, J. C., 1986, A Theoretical and Numerical Approach to Plastic Behavior of Steels During Phase
TransformationsII. Study of Classical Plasticity for Ideal-Plastic Phases, J.
Mech. Phys. Solids, 344, pp. 411432.
9 Dean, D., and Hidekazu, M., 2005, Prediction of Welding Residual Stress in
Multi-Pass Butt-Welded Modified 9Cr-1Mo Steel Pipe Considering Phase
Transformation Effects, Comput. Mater. Sci., 37, pp. 209219.
10 Kim, J., Im, S., and Kim, H. G., 2005, Numerical Implementation of a
Thermo-Elastic-Plastic Constitutive Equation in Consideration of Transformation Plasticity in Welding, Int. J. Plast., 21, pp. 13831408.
11 Leblond, J. B., 1989, Mathematical Modeling of Transformation Plasticity in
Steels II: Coupling With Strain Hardening Phenomena, Int. J. Plast., 5, pp.
573591.
12 Simo, J. C., and Hughes, T. J. R., 1998, Computational Inelasticity, Springer,
New York, pp. 269275.
13 Belytschko, T., Liu, W. K., and Moran, B., 2000, Nonlinear Finite Elements
for Continua and Structures, Wiley, West Sussex, pp. 215308.
14 Hibbitt, Karlsson & Sorensen, Inc., 2008, ABAQUS/STANDARD Users Manual,
Vol. 2, Version 6.8, p. 23.1.4-2.

APRIL 2011, Vol. 133 / 021003-13

Downloaded 03 Mar 2011 to 143.248.58.30. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm