European Journal of Mechanics A/Solids 54 (2015) 120e131

Contents lists available at ScienceDirect

European Journal of Mechanics A/Solids

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

Comparison of two models for anisotropic hardening and yield surface

evolution in bcc sheet steels
T. Clausmeyer a, b, *, B. Svendsen c, d
a

Institute of Mechanics, TU Dortmund University, Leonhard-Euler Strae 5, D-44227 Dortmund, Germany

Institute of Forming Technology and Lightweight Construction, TU Dortmund University, Baroper Strae 303, D-44227 Dortmund, Germany
c
Material Mechanics, RWTH Aachen University, Schinkelstrae 2, D-52062 Aachen, Germany
d
Microstructure Physics and Alloy Design, Max-Planck-Institut fr Eisenforschung GmbH, Max-Planck Strae 1, D-40237 Dsseldorf, Germany
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 23 June 2014
Accepted 29 May 2015
Available online 9 June 2015

The purpose of the current work is the investigation and comparison of aspects of the material behavior
predicted by two models for anisotropic, and in particular cross, hardening in bcc sheet steels subject to
non-proportional loading. The rst model is the modied form (Wang et al., 2008) of that due to Teodosiu and Hu (1995, 1998). In this (modied) Teodosiu-Hu model (THM), cross hardening is assumed to
affect the yield stress and the saturation value of the back stress. The second model is due to Levkovitch
and Svendsen (2007) and Noman et al. (2010). In the Levkovitch-Svendsen model (LSM), cross hardening
is assumed to affect the ow anisotropy. As clearly demonstrated in a number of works applying the THM
(e.g., Boers et al., 2010; Bouvier et al., 2005, 2003; Hiwatashi et al., 1997; Li et al., 2003; Thuillier et al.,
2010; Wang et al., 2008) and the LSM (e.g., Clausmeyer et al., 2014, 2011b; Noman et al., 2010), both of
these are capable of predicting the effect of cross hardening on the stress-deformation behavior observed
experimentally in sheet steels. As shown in the current work, however, these two models differ
signicantly in other aspects, in particular with respect to the development of the yield stress, the back
stress, and the yield surface. For example, the THM predicts no change in the shape of the yield surface
upon change of loading path, in contrast to the LSM and crystal plasticity modeling of bcc sheet steels
(Peeters et al., 2002). On the other hand, the LSM predicts no hardening stagnation after cross hardening
as observed in experiments, in contrast to the THM. Examples are given.
2015 Elsevier Masson SAS. All rights reserved.

Keywords:
Material modeling
Cross hardening
Yield surface

1. Introduction
Finite-element-based modeling and simulation of the material
and structural behavior of sheet metal parts in various stages of
design and manufacture is today standard. In general, one aim of
this is to benet from the predictive capability of such simulations
(Zienkiewicz et al., 2010). In this regard, Wagoner et al. (2013)
emphasize the importance of improving material models to account for the loading path-dependent behavior of metals during
sheet metal forming. When subject to complex non-proportional
loading processes such as those found in many technological applications, a number of metals exhibit hardening behavior which is
more complex than isotropic and kinematic hardening alone.

* Corresponding author. Institute of Forming Technology and Lightweight

Construction, TU Dortmund University, Baroper Strae 303, D-44227 Dortmund,
Germany. Tel.: 49 231 755 8429; fax: 49 231 755 2489.
E-mail address: till.clausmeyer@iul.tu-dortmund.de (T. Clausmeyer).
http://dx.doi.org/10.1016/j.euromechsol.2015.05.016
0997-7538/ 2015 Elsevier Masson SAS. All rights reserved.

Observed effects in this regard include cross hardening and hardening stagnation during orthogonal loading (e.g., tension to shear).
Cross hardening is observed to occur for example in a number of
steels such as austenitic fcc tube steels (e.g., SUS304: Ishikawa,
1997; Wu, 2003), ferritic bcc tube steels (e.g., S355: Kowalewski
and Sliwowski, 1997), multi-phase tube steels (e.g., X100:
Shinohara et al., 2010), or ferritic bcc sheet steels (e.g., LH800:
Ghosh and Backofen, 1973; Noman et al., 2010). Systematic studies
(Bouvier et al., 2005, 2006a, 2003) of interstitial free (IF), highstrength low-alloyed (HSLA), transformation-induced plasticity
(TRIP), and dual-phase (DP), sheet steels, found signicant kinematic hardening, hardening stagnation, as well as cross hardening,
the latter especially in IF sheet steels. In these investigations, the
material was subjected to monotonic shear, reverse shear, as well as
orthogonal tension-shear, loading. Clausmeyer et al. (2012); van
Riel and van den Boogaard (2007); Wang et al. (2008) have documented these effects in the IF sheet steel DC06 with the help of
monotonic tension, reverse shear, and orthogonal tension-shear,

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

tests, all under plane-strain conditions. In particular, cross hardening occurs during discontinuous (e.g., tension-shear: Bouvier
et al., 2005, 2006a, 2003) and continuous (e.g., tension-shear:
Noman et al., 2010; van Riel and van den Boogaard, 2007; Wang
et al., 2008) orthogonal tension-shear tests. Similar results were
obtained by Verma et al. (2011) in a series of tension and
compression tests on ultra-low carbon IF sheet steel in which the
tension or compression direction changed from rolling to transverse. In these tests, cross hardening was correlated with a change
of the tension axis. As attested to in particular by the continuous
orthogonal tension-shear test results (Noman et al., 2010; van Riel
and van den Boogaard, 2007; Wang et al., 2008), cross hardening is
transient and strongly depends on the rate of transition. Its
occurrence and strength are strongly inuenced by the particular
path taken in stress space in changing from one loading direction to
another.
Generally speaking, anisotropic hardening in sheet steels may
be inuenced by the grain and dislocation (micro)structures. In
particular, the former is related to the grain orientation distribution
(texture). The inuence of texture on the hardening behavior of IF
sheet steel was investigated by Bacroix and Hu (1995) and
Nesterova et al. (2001a,b) using two-stage loading tests (e.g., shear
to reverse shear, tension to shear). In particular, Bacroix and Hu
(1995) concluded that, at least up to moderate strains, the inuence of texture evolution on hardening in the specimens investigated was small compared to that of dislocation structure
evolution. This conclusion was substantiated by later crystal plasticity modeling (e.g., Peeters et al., 2002). Related to this are more
recent EBSD investigations on DC06 (Boers et al., 2010; Clausmeyer
et al., 2012), which imply that the rolling-induced texture in this
steel does not change considerably for strains lower than 35% in
simple tension. This may be the case in other ferritic steels (e.g.,
LH800: Clausmeyer et al., 2012; Noman et al., 2010) as well. These
results imply that it is sufcient to account for the effect of the
initial (e.g., rolling) texture on the anisotropic hardening and ow
behavior in the material model.
Although texture evolution in this sense may be secondary,
grain orientation (i.e., glide-system orientation) in relation to
loading direction certainly inuences dislocation structure development. The development of certain characteristic dislocation
structures related to cross hardening have been observed during
quasi-static loading of mild steels such as DC06 at room temperature (e.g., Rauch and Schmitt, 1989; Rauch and Thuillier, 1993;
Thuillier and Rauch, 1994). These include for example dense
dislocation wall structures. The morphology and orientation of such
walls depends for example on grain orientation, the type of loading,
and the loading direction in relation to the grain orientation (e.g.,
Clausmeyer et al., 2012; Nesterova et al., 2001a,b; Thuillier and
Rauch, 1994). A change in loading direction or type activates new
glide systems for which existing walls act initially as obstacles,
resulting in cross hardening.
One of the rst phenomenological models accounting in
particular for cross hardening is the Teodosiu-Hu model (THM: e.g.,
Hu et al., 1992; Teodosiu and Hu, 1995, 1998). In the THM, cross
hardening is assumed to affect the yield stress sY in the yield
function fY. The THM has been employed in a number of works
(e.g., Bouvier et al., 2005, 2003; Haddadi et al., 2006; Hiwatashi
et al., 1997; Li et al., 2003; Thuillier et al., 2010) to model anisotropic ow and hardening behavior in sheet metals. This has
motivated similar work on models for anisotropic hardening in the
continuum (Barlat et al., 2013; Butuc et al., 2011; Carvalho Resende
et al., 2013; Clausmeyer et al., 2014; Pietryga et al., 2012; Shi and
Mosler, 2012; Tarigopula et al., 2008, 2009) and crystal plastic
(Peeters et al., 2002; Viatkina et al., 2007) contexts. More recently,
the THM has been modied, extended and generalized to deal with

121

arbitrary changes of loading path by Wang et al. (2008). This

modied version of the THM is that considered in the current work.
A second model for cross hardening was introduced by
Levkovitch and Svendsen (2007) and Noman et al. (2010). This
model has been referred to by Shi and Mosler (2012) as the LevkovitcheSvendsen model (LSM), who discussed related models for
distortional hardening and the strength differential effect in magnesium alloys. In the LSM, cross hardening is assumed to inuence
the ow anisotropy through the corresponding tensor A determining fY. Common to both the THM and the LSM is the constitutive form

fY

p
M  X$A M  X  sY

(1)

for fY with respect to the intermediate (local) conguration. Here,

M is the Mandel stress (e.g., Mandel, 1971, 1974), and X is the back
stress. To be more precise, in the THM, sY is assumed to depend on
both isotropic and cross hardening, and A is assumed constant. On
the other hand, in the LSM, cross hardening is assumed to inuence
the evolution of A , and sY is assumed to depend only on isotropic
hardening. As shown in the previous works discussed above, both
models are capable of quantitatively predicting experimentally
observed cross hardening. The question arises as to how the THM
and the LSM compare in other respects. To this end, in the current
work, a direct comparison of these two has been carried out. To this
end, both models have been identied from the same data set for
the ferritic sheet steel DC06 (for comparison, the ferritic-pearlitic
steel LH800 is also briey discussed). As the current comparison
of the THM and the LSM shows, the two models are not equivalent
in other respects. Among these, the prediction of yield surface
evolution is perhaps the most prominent.
The current work begins with a brief review of the formulation
of the two models in Section 2. This is carried out within the
framework of the multiplicative decomposition of the deformation
gradient and the assumption of small elastic strain relevant to
metal inelasticity. Again, for a meaningful comparison, the two
models are identied in Section 3 using the same test data sets for
the ferritic bcc sheet steel DC06. The identied THM and LSM are
then compared on the basis of their respective predictions for yield
and back stress evolution in Section 4 as well as yield surface
development in Section 5. The latter results are also compared
qualitatively with analogous results from the crystal plasticity
model of Peeters et al. (2002) for IF steel. Lastly, these two models
are compared in the context of their application to the modeling of
sheet metal forming during non-proportional loading in Section 6.
The work ends with a summary and discussion in Section 7.
2. Model formulation
2.1. Notation
In this work, Euclidean vectors (i.e., rst-order Euclidean tensors) are represented by lower-case bold italic characters a,b,; in
particular, let i1,i2,i3 represent the Cartesian basis vectors. Likewise,
upper-case bold italic characters A,B, represent second-order
Euclidean tensors; in particular, let I represent the second-order
identity tensor. Such tensors are dened in this work as linear
mappings between (three-dimensional) Euclidean vectors. In other
words, Ab is a vector for all A and all b. Let I,A and dev A A(I,A)
I/3 represent the trace and deviatoric part, respectively, of any A.
Likewise, let sym A:(A AT)/2 and skw A:(A  AT)/2 represent
the symmetric and skew-symmetric parts, respectively, of any A.
Fourth-order tensors are represented by upper-case calligraphic
characters A ; B ; , in this work. Interpreting these as linear
mappings between second order tensors, A B represents a second

122

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

order tensor for all A and all B. The dyadic product C A5B of
any A and any B is a fourth-order tensor with Cartesian components
Cijkl AijBkl. The scalar product of two tensors A and B of any order
is represented by A,B AijBij via the summation convention.
The (Euclidean) norm of any A is dened by jAj:(A,A)1/2.
2.2. Relations common to both models
The formulation of both models for cross hardening and yield
surface evolution under consideration in this work is carried out in
the framework of the standard multiplicative decomposition

F FEFP

(2)

of the deformation gradient F (e.g., Lee, 1969) into elastic FE and

inelastic FP parts, respectively. Note that such a decomposition
arises naturally in the context of the modeling of FP as a change of
local reference conguration (Svendsen, 2001). More precisely, the
current formulation is based on evolution relations derived from
(2) and the right polar decomposition FE REUE of FE in the context
of small elastic strain as follows. To begin, note that the time derivative of F E FF 1
P RE U E results in the relation

R_ E U E RE U_ E RE U E LP LRE U E

(3)

_ 1 and its inelastic counterbetween the velocity gradient L : FF

part LP : F_ P F 1
.
Multiplying
(3)
on
the
right by U 1
P
E and on the left
by RTE R1
,
we
obtain
E
1
T
RTE R_ E U_ E U 1
E U E LP U E RE LRE :

(4)

Restrict the formulation now to metals and small elastic strain

as gauged or measured by (the magnitude of) the right logarithmic
stretch ln UE, i.e., jln UEj1. In this case,

2 



U 1
E Iln U E O ln U E  ;

(5)

and so

U E LP U 1
E LP Ojln U E j;

_
U_ E U 1
E ln U E Ojln U E j;
(6)

then hold. Here, O(jAjn) represents all terms depending explicitly

on all jAjm for m  n. Substituting (6) into (4), rearrangement yields

_
RTE R_ E ln U E LP RTE LRE Ojln U E j:

(7)

_
Lastly, since RTE RE I implies RTE RE 2 sym RTE R_ E 0, the
symmetric and skew-symmetric parts of (7) then imply the (smallelastic-strain-based) evolution relations

_
ln U E RTE DRE  DP ;

(8a)

R_ E WRE  RE W P ;

(8b)

to O(jln UEj) for ln UE and the elastic rotation RE, respectively. Here,
D sym L is the continuum rate of deformation, DP sym LP its
inelastic counterpart, W skw L the continuum spin, and
WP skw LP the plastic spin. In particular, the usual associated ow
rule

DP a_ P

vfY
vM

is assumed in both models for DP in terms of the yield function fY

from (1), with aP the equivalent inelastic deformation. Since various
investigations (e.g., Bacroix and Hu, 1995; Boers et al., 2010) suggest
that the inuence of grain orientation evolution on hardening in
ferritic bcc sheet steels such as DC06 is not signicant in the range
of deformation relevant to the forming processes considered in this
work, WP is neglected here for simplicity. Consequently, the evolution relation (8b) for RE depends only on the continuum spin W,
and as such is purely kinematic. In addition, attention is restricted
here to forming below the forming limit, and damage or any other
process resulting in inelastic volume changes are assumed negligible. In this case, plastic incompressibility I,DP 0 pertains, and
DP dev DP is purely deviatoric as usual.
Since the elastic range and elastic strain are small, the effect of
any elastic anisotropy on the material behavior is assumed negligible here. In this case, the isotropic form

M kI$E E I 2 m dev E E

(10)

holds for the Mandel stress to lowest order in the small elastic
strain measure EE:ln UE from (5), with k the bulk modulus, and m
the shear modulus. Likewise in the context of small elastic strain,

K RE MRTE

(11)

holds between the Kirchhoff K and Mandel M stresses. In this case,

note that the trace I,K I,M is independent of RE. Lastly, both
models are based on common saturation-based forms

r_ cr sr  ra_ P ;

(12a)

X_ cx sx N P  Xa_ P ;

(12b)

for the evolution of isotropic and kinematic hardening, respectively.

Here, r represent the contribution of isotropic hardening to sY. In
addition, cr and cx represent the respective growth rates, sr and sx
are the respective saturation values, and NP:DP/jDPj is the direction of inelastic ow.
Since the THM (e.g., Hu et al., 1992; Teodosiu and Hu, 1995, 1998;
Wang et al., 2008) and the original version (e.g. Levkovitch and
Svendsen, 2007; Noman et al., 2010) of the LSM1 are limited to
rate-independence, the current model comparison is carried out on
this basis. To this end, attention is restricted in this work to quasistatic loading rates at which the rate-dependent behavior of the
steels in question is small. As is well-known, the rate-independent
formulation is based in particular on the concepts of an elastic
range, yield surface and yield conditions, resulting in the so-called
KaresheKuhneTucker conditions: f_ Y  0, fY  0, fY a_ P 0. In this
context, aP is determined as usual by the consistency condition:
f_ Y 0 when fY 0. For the case of rate-dependence, of course, the
yield condition generalizes to an activation condition, and the
consistency condition no longer applies.

2.3. THM for cross hardening

Besides the common ones just discussed, the THM (Teodosiu
and Hu, 1995, 1998; Wang et al., 2008) is based in particular on
the constitutive relations

sY sY0 r f jS j;

(13a)

(9)
1

This model was generalized to rate-dependence recently in Barthel et al. (2013).

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

A A 0;

(13b)

for the yield stress sY and ow anisotropy tensor A , respectively,

appearing in (1). In (13b), A 0 is the initial value2 of A determined
by the initial grain structure (i.e., rolling texture). As discussed in
the introduction, for the current case of ferritic bcc sheet metals,
the grain (orientation) structure is not observed to vary signicantly in the loading range of interest, resulting in (13b). In (13a),
sY0 is the initial (i.e., aP 0) yield stress, and f is a material constant
determining the fraction of oriented dislocation structures
contributing to isotropic hardening. In addition, S is a (symmetric
traceless) fourth-order structure tensor, the magnitude of which
represents the effective strength of (persistent) oriented dislocation structures and their effect on the level of yield. The evolution of
S as a standard internal variable3 is modeled by the relation


S_ cd hp ss N




 hp hx S d a_ P  cl jS l =ss jnl S l a_ P

(14)

in the back-rotated frame with N P N P 5N P . Here, S d represents

the strength of the dislocation structures (e.g., walls) associated
with the currently active slip systems, and S l S  S d is that
associated with the latent (i.e., currently inactive) slip systems.
Note that cd (hp hx) determines the effective saturation rate, and
hp ss/(hp hx) the effective saturation magnitude, associated with
S d . By analogy, cl jS l =ss jnl is the effective saturation rate. These
quantities contain the material constants cd, cl, ss, and nl. hp represents a further material function depending on S d , the direction
NP, a further material constant np, and the symmetric traceless
second-order tensor-valued internal variable P with constitutive
relation

P_ cp N P  Pa_ P

(15)

which accounts for the effect of dislocation polarization on cross

hardening (Teodosiu and Hu, 1998; Wang et al., 2008). cp is the
corresponding saturation rate. The effect of hardening stagnation is
accounted for via material function hx depending in particular on
NP and X (Teodosiu and Hu, 1998; Wang et al., 2008). Lastly, the
saturation magnitude sx of X is given by the constitutive relation

q
sx x0 1  f jS d j2 mjS l j2

(16)

depending on the directional strength of dislocation structures.

Here, x0 represents the initial (i.e., jS d j 0 jS l j) value of sx. The
parameter m determines the inuence of latent as opposed to
current dislocation structures on the saturation magnitude sx of the
back stress X.
This model contains 21 material parameters. These include the
elastic bulk k and shear m moduli, the initial yield stress sY0, the six
parameters F, G, H, L, M, and N determining the initial ow
anisotropy tensor A 0, the ve saturation rates cr, cx, cd, cl and cp, the
three saturation values sr, x0, and ss, the two exponents nl and np, as
well as f and m.

2
A 0 is often assumed to be of the type due to Hill (1948). Other choices (such as
in Cazacu and Barlat, 2004; Shi and Mosler, 2012) for the representation of the
initial ow anisotropy in non-bcc metals can be combined with the presented
approaches for anisotropic hardening.
3
In the original form of the THM due to Teodosiu and Hu (1995, 1998), S is not
modeled as such a variable. To this end, and for the purpose of modeling continuous
loading path changes, the model was reformulated by Wang et al. (2008), resulting
in (14) for S_ .

123

2.4. LSM for cross hardening

As just discussed, the effect of cross hardening on the material
behavior is accounted for in the THM and (13) by the dependence of
the yield stress sY on S . Instead of this, the effect of cross hardening
on the material behavior is accounted for in the LSM by a dependence of the ow anisotropy tensor A on a structure tensor H
formally analogous to S . Instead of (13), then, the LSM for cross
hardening is based on the constitutive relations

sY sY0 r;
A A

(17a)

H;

(17b)

for the yield stress and ow anisotropy tensor, respectively, in (1).

Since the effect of cross hardening on the material behavior is then
accounted for in the modeling (17b) of A here, that (17a) of sY
accounts only for dislocation processes resulting in isotropic
hardening. As discussed in detail elsewhere (Levkovitch and
Svendsen, 2007; Noman et al., 2010), the form

H_ cd hd N

 H d a_ P cl fhl I dev  N P  H l ga_ P

(18)

of the evolution relation for H (and so for A since A 0 is constant)

in the LSM is motivated by that (14) for S in the THM. Here, I dev
I  13 I5I is the fourth-order deviatoric identity. Further,
H d N P $H N P is the part of H parallel, and H l H  H d that
perpendicular, to N P . The idea here is that currently active dislocation structures oriented parallel to NP persist after an orthogonal
loading-path in inactive form (e.g., as cell-block boundaries). This
strengthens existing obstacles to glide-system activation in the
new loading direction.
Comparison of (13) and (17) shows clearly the qualitative difference between the two models in regards to the assumed effect of
cross hardening on the material behavior. In addition, note that
isotropic and kinematic hardening are coupled to cross hardening
in the THM, but basically decoupled from each other in the LSM. As
it turns out, the LSM is an example of the class of models distinguished by a yield function of the general form introduced in Baltov
and Sawczuk (1965). As will be shown in detail in what follows, one
result of this is that cross hardening results in a change in shape of
the yield surface in the LSM. This is completely analogous to the
association of isotropic hardening with a change in size, and kinematic hardening with a change in center, of this surface. Before
delving into this, however, we rst summarize briey the identication of the above models upon which their subsequent comparison is based.
3. Parameter identication
For completeness, the results of the model parameter identication are briey summarized and discussed in this section. For this
study, DC06 steel sheets (DIN EN 10130) of 1.0 mm thickness supplied by ThyssenKrupp Steel were tested in uniaxial tension, planestrain tension, monotonic shear, shear to reverse-shear, and planestrain tension to shear. The advantage of shear tests is that larger
strains can be achieved with common testing facilities (Bouvier
et al., 2006b; Klepaczko et al., 1999; Miyauchi, 1987) than generally possible in uniaxial tension tests. The steel DC06 is often used
in auto-body parts4 due to its high ductility and formability. The
mechanical tests involving shear and plane-strain deformation

4
For example,
Tiefziehstaehle.pdf.

see

http://incar.thyssenkrupp.com/download/Broschueren/

124

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

models is hardly surprising and certainly to be expected. As

investigated in more detail in the next section, another difference
between the two models regards the yield surface evolution predicted by each, to which we turn after a more detailed analysis of
the contribution of isotropic as well as kinematic hardening.

were carried out using the biaxial testing facility at the Faculty of
Engineering Technology, University Twente, The Netherlands. Details of the mechanical testing procedure are given elsewhere (e.g.,
Clausmeyer et al., 2011b; Noman et al., 2010; van Riel and van den
Boogaard, 2007). The region of the specimen subject to deformation is initially 45.0 mm wide and 3.0 mm high. The strain eld is
homogeneous in the range of deformation investigated. As an unambiguous and transparent measure of the state of deformation,
the deformation gradient F is used here; let Fij ii,Fij represent its
Cartesian components. For the tests, the tension/rolling direction is
oriented in the i2 direction, and the shear/transverse direction in
the i1 direction. Consequently, in the case of plane-strain testing,
F22 is the deformation component for tension, and F12 the deformation component for shear. Given plastic incompressibility and
small elastic strain, the Cauchy stress T K/detF is wellapproximated by the Kirchhoff stress K; let Kij ii,Kij represent
its Cartesian components in what follows.
Among the material parameters, note that k, m, sY0, F, G, H, L, M,
N, cr, sr, cx, and x0 are common to both models. The Hill parameters
F, G, H, L, M, N are computed from average r-values (see Clausmeyer
et al., 2011b). The r-values are obtained from uniaxial tension tests
performed at 0 , 45 and 90 with respect to the rolling direction. In
particular, for DC06 at room temperature, we have k151 GPa,
m69.6 GPa, sY0 132 MPa, F 0.252, G 0.302, H 0.698,
N 1.36, L 1.5, and M 1.5. The remaining parameters values
identied for all models are shown in Table 1. The hardening parameters cr, sr, cx, and x0 are identied from uniaxial tension, planestrain tension, shear-reverse shear, and plane-strain tension to
simple shear, tests (Haddadi et al., 2006; Noman et al., 2010; Wang
et al., 2008). Since for example the modeling of kinematic hardening is different in the two models, note that the corresponding
material parameter values in Table 1 are different. In particular, sx is
constant in the LSM and variable in the THM.
The identied models for DC06 are compared with experimental data for the case of plane-strain tension to simple shear
loading in Fig. 1 (left). To document the applicability of the THM and
the LSM to steels other than DC06, and for comparison, analogous
results for the ferritic-pearlitic steel LH800 (thickness 0.7 mm)
from Noman et al. (2010) and Noman (2010) are also shown in Fig. 1
(right). Clearly, for both DC06 and LH800, cross hardening is
captured well by both models; this hardening is particularly pronounced in the former material (i.e., more than 50 MPa; Fig. 1, left).
Also more pronounced in DC06 than in LH800 is hardening stagnation after cross hardening and change of loading direction (e.g.,
from tension to shear). Such stagnation is accounted for in the THM
(via (15) and (16)) but not in the LSM. This is the reason why, after
loading path change and cross hardening, the THM-based result
(red dashed curve, in the web version) in Fig. 1 (left) lies below the
LSM-based one (green dashed curve, in the web version) and closer
to the data (crosses). In any case, this difference between the two

4. Comparison of isotropic and kinematic hardening

modeling
As discussed in Section 2, the relations (13a) and (17a) for sY in
the THM and the LSM, respectively, differ in the assumed inuence
of cross hardening via S on the former. Note that sY0 and the
evolution relation (12a) for r are the same in both models. As well,
the evolution relation (12b) for X is of exactly the same form in both
the THM and the LSM. On the other hand, the saturation magnitude
sx in (12b) is treated differently in the two models, i.e., dependent
on S and variable in the THM via (16), and constant in the LSM.
As an example of the consequences of these model differences,
consider the orthogonal loading of DC06 in discontinuous tensionshear. In particular, this loading path consists of (i) loading in tension to F22  1 0.1, (ii) unloading, and (iii) reloading in shear to
F22  1 F12 0.5. Consider rst the results for aP and sY for this
loading case displayed in Fig. 2. Although the relation (12a) for the
evolution of r is the same in both models, the coupled nature of
model identication results in different values for the same material coefcients in Table 1. For example, note that cr is a factor of 3
smaller, and sr is an order of magnitude larger, in the LSM than in
the THM. This is the reason why, as shown in Fig. 2 (right), isotropic
hardening saturates much more quickly in the THM than in the
LSM. The much lower level of r in the THM is due of course to the
additional contribution from S to isotropic hardening not assumed
in the LSM. Given the coupled nature of model identication, from a
quantitative point of view, the contribution of X to fY in (1) will also
inuence the level of isotropic hardening in both models.
Consider next the development of the components of the back
b : RE XRT in the current conguration during tensionstress X
E
shear loading displayed in Fig. 3. During the tension stage, the
b , X
b and X
b is qualitaevolution of the normal components X
11
22
33
tively similar for both models. The reduction in these components
predicted by both models during shear (re)loading after tension
(pre)loading evident in Fig. 3 is due to the fact that the corresponding components of NP vanish in the shear stage, resulting in a
zero saturation level for the evolution of these components as
determined by (12b). As is the case for DC06 in Fig. 3, due to the
increase of jS j (i.e., jS d j; jS l j) and its contribution to sx via (16), the
level of kinematic hardening predicted by the THM will generally
be higher than that of the LSM, in which sx is assumed constant.
This together with the effect of hardening stagnation on S is also
b 12 in Fig. 3
the reason for qualitative differences in the results for X
in the shear stage of discontinuous tension-shear loading. Indeed,

Table 1
Material parameter values for DC06 determined for the THM (above) and the LSM (below) from room-temperature uniaxial tension, monotonic shear, cyclic shear, and
orthogonal tension-shear, test data (Clausmeyer et al., 2011b). See text for details.
THM
parameter

cr

sr

units
value

cx

MPa
20.1

x0

cd

ss

MPa

20.8

499.0

8.1

cl

nl

cp

np

4.1

1.0

4.0

90.0

0.2

1.6

MPa
8.4

221.0

LSM
parameter

cr

sr

6.64

192.0

units
value

cx

sx

33.1

56.0

MPa

cd

hd

cl

hl

23.9

0.0

87.3

-0.447

MPa

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

125

Fig. 1. Comparison of model and experimental results for shear stress as a function of deformation in DC06 (left) and LH800 (right: Noman et al., 2010; Noman, 2010). Both
materials were subject to plane-strain tension in the rolling direction (up to 10% in DC06 and 13% in LH800) followed by an orthogonal loading path change to simple shear in the
transverse direction. See text for discussion and web version for color.

Fig. 2. Development of aP (left) and isotropic hardening (right) predicted by the THM and the LSM during discontinuous tension-shear loading. This loading path consists of (i)
loading in tension to F22  1 0.1, (ii) unloading, and (iii) reloading in shear to F22  1 F12 0.5. See text for details.

Fig. 3. Back stress evolution predicted by the THM (left) and the LSM (right) during discontinuous tension-shear loading. Here, the material is (i) loaded in tension to F22  1 0.1,
(ii) unloaded, and (iii) reloaded in shear to F22  1 F12 0.5. See text for details and web version for color.

in the LSM (Fig. 3, right), sx is constant (e.g., 56 MPa for DC06 in

b from the LSM has the
Table 1). Consequently, the result for X
12
typical Voce-based growth-saturation form as expected from (12b).
In contrast, as shown in Fig. 3 (left), the coupling of the evolution of
X to S via sx in the THM results in qualitative deviations of the
b
result for X
12 from ideal Voce form. In particular, the effect of
hardening stagnation and the increase of sx are evident in Fig. 3
(left). As well, in the case of DC06 at least, the much larger value

b in
of cx for the THM in Table 1 results in much faster growth of X
12
the THM (Fig. 3, left) than in the LSM (Fig. 3, right). Note the simb in Fig. 3 (left) and K12 in Fig. 1 (left).
ilarity of X
12
5. Comparison of yield surface modeling
In this section, the THM and the LSM identied for DC06 via the
model parameter values in Table 1 are used to model yield surface

126

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

development in DC06. In the current model context and relation (1)

for the yield function fY, the yield surface represents the boundary
fY 0 in stress space of the so-called elastic range of the material,
i.e., those stress states for which fY < 0 holds at xed inelastic
microstructure. In the current phenomenological context, (the effect of) such microstructure (on the material behavior) is modeled
by (evolution of) the internal variables. In particular, given the
relation (11) between the Kirchhoff K and Mandel M stresses, fY
from (1) can be expressed as a function f(K,I)fY of K and a modeldependent set I of internal variables; in particular,
ITHM fRE ; r; X; S g via (13) and ILSM fRE ; r; X; H g via (17). The
yield condition f(K,I)fY 0 then represents an implicit relation
for K as a function of I. In a given loading test along the yield surface, the initial value of I evolves to a new value depending in
general on loading path. For simplicity, attention is restricted to
biaxial monotonic and cyclic loading paths involving tension,
compression, and shear (or different combination of these) in the
(K12,K11) or (K12,K22) planes. These loading paths are analogous to
those used in the identication procedure. In what follows, nondimensional stress components

K ij : Kij sY0 i j;

K ij :

p .
3Kij sY0 isj;

(19)

are employed to display yield surface results.

Consider again discontinuous tension-shear loading. The corresponding yield surface development predicted by the two models
is shown in Fig. 4. In particular, as shown in Fig. 4 (left), the yield
surface development predicted by the THM in the tension-shear
case is characterized by a change in size and a change in center,
now in both the tension and shear directions, but no change in
shape. Again, this is directly related to (13b). With increasing shear
loading, the yield surface saturates towards the one (THM3) given
for the forward shear case in Fig. 4 (left). On the other hand, on the
basis of (17b), the LSM predicts a change in shape as well due to
cross hardening, now in both the tension and shear directions.
Indeed, as shown in Fig. 4 (right), this change is in K12 direction
during the tension part, and in the K22 direction during the shear
part, of the tension-shear loading path.
Lastly, yield surface results from the THM and the LSM are
compared with those of the more sophisticated Peeters et al. (2002)
model (PM) based on crystal plasticity. Application to an IF steel
subject to simple tension, simple shear, and tension-shear, loading

paths demonstrated that the PM predicts the experimentally

observed isotropic and cross hardening, as well as the Bauschinger
effect and hardening stagnation, in this material. The PM accounts
for the hardening due to interactions of currently active dislocations with persistent dislocation structures. The single crystal
model was combined with a full constraints Taylor model for the
polycrystalline IF steel. The initial texture of the IF steel was
determined by x-ray diffraction. A set of 250 orientations was used
to model the behavior of the polycrystalline IF steel. Modeling of
yield surface development in Peeters et al. (2002) was based in
particular on texture simulation and the critical resolved shear
stresses on each slip system in each grain. Their results imply that,
for the IF steel they considered, the effects of texture development
on the yield surface were minimal in comparison to that of the
dislocation structures such as (dislocation) cell block boundaries.
As it turns out, the IF steel investigated by Peeters et al. (2002)
exhibits very similar hardening behavior in monotonic shear to
the DC06 material investigated in this study.
This is also implied by the comparison in Fig. 5 of yield surface
results for DC06 from the THM and the LSM with those for IF steel
from the PM in the cases of simple tension loading (left) and simple
shear loading (right).
As shown by these results, in contrast to the THM and the PM,
very good qualitative agreement exists between the LSM and the
PM. In particular, with respect to the curvature and the widening of
the yield surfaces in directions orthogonal to the current loading
directions, the agreement is even quantitatively very good. On the
other hand, as shown in Fig. 5, the LSM and the PM differ in regards
to the predicted translation of the yield surface. As well, in directions which correspond to a loading reversal, there are differences in the level of yield stress predicted by the LSM and the PM.

6. Model comparison in the context of metal forming

simulation
Generally speaking, during sheet metal forming, different parts
of the workpiece experience different loading and deformation
histories. In particular, this results in a heterogeneous distribution
of strength, hardening and residual stress states in the formed
workpiece which can have a strong inuence on its subsequent
behavior. In this regard, cross hardening for example was shown to
have a signicant inuence on the springback behavior of hat

Fig. 4. Yield surface evolution during discontinuous tension-shear loading predicted by the THM (left) and by the LSM (right). The surfaces THM1 and LSM1 are those at the end of
the tension stage at F22  1 0.1. THM2 and LSM2 represent the mixed state at F22  1 F12 0.14, while THM3 and LSM3 are at the end of shear loading at F22  1 F12 0.5.

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

127

Fig. 5. Comparison of yield surface development in K 12 ; K 11 space predicted by the THM and the LSM for DC06, and by the PM for IF steel, after uniaxial tension in the rolling
direction (left) and simple shear in the rolling direction (right), up to 10% von Mises equivalent strain. As done for the results from the THM and the LSM via (19), those for K from
the PM were normalized using the initial yield stress of the material as determined by Peeters et al. (2002). See text for details and web version for color.

proles (consisting of mild steel and DP steel: Haddag et al., 2007),

split rings from a drawn cup (consisting of DC06: Wang et al., 2008),
or even more complex parts (e.g., channel die and S-rail consisting
of DC06: Clausmeyer et al., 2014). Depending on factors such as the
geometry of the workpiece, different types of loading path changes
occur locally in the structure.
For example, in the case of deep drawing with a cross-shaped
die, nite-element (FE) simulations (e.g., Clausmeyer et al., 2011a)
show that, during this process, certain regions in the sides of the
workpiece structure experience a loading path change from planestrain tension to uniaxial tension. Stress-deformation results for
DC06 subject to such loading are shown in Fig. 6. As shown, both
models predict a moderate increase of cross hardening with
increasing (pre)tension. In particular, after 10% (pre)tension (Fig. 6,
right), the level of cross hardening predicted by the THM is larger
than predicted by the LSM. This is in contrast to the tension-shear
case in Fig. 1. Although less pronounced than in the tension-shear
case, cross hardening in the tensionetension case is also transient
and tends to disappear with increasing loading after the loading
path change. Consequently, for this case as well, dislocation structures responsible for cross hardening appear to break down and
disappear after change of loading path, i.e., as along as the loading
path direction remains constant. The results in Fig. 6 imply that the
level of (pre)deformation during the rst stage of a multi-stage
loading path inuences the amount of resulting cross hardening.

As shown in Fig. 7, such a dependence is also reected in the

concomitant yield surface development. As before, the cross
hardening evident in stress-deformation results like those in Fig. 6
results in the LSM (Fig. 7, right), but not in the THM (Fig. 7, left), in a
change in shape (and orientation) of the yield surface. In particular,
as evident in Fig. 7 (right), the LSM predicts a rotation of the major
axes of the (in the K 11 ; K 22 plane) elliptic yield surface away from
the equibiaxial tension orientation towards K22 with increasing
deformation.
Compared to the case of plane-strain tension followed by simple
shear (Fig. 1), the amount of cross hardening in the case of planestrain tension in the rolling direction followed by uniaxial tension
(Fig. 6) is generally much smaller (e.g., Clausmeyer et al., 2011a).
This is correlated with the degree of orthogonality of the loadingpath change. According to Schmitt et al. (1994), a two-stage
loading path is characterized in this regard by the angle
q arccos(NP1,NP2) determined by the direction of the rate of inelastic ow NPi during the rst (i 1) and second (i 2) stages of
the two-stage path. In terms of this angle, plane-strain tension to
simple shear is characterized by q 90 , and plane-strain tension to
uniaxial tension by q 30 . As has been discussed elsewhere (e.g.,
Wang et al., 2008), this angle can be used to characterize the expected degree of orthogonal loading-path change and similarly
represent the strength of possible cross hardening as a function of
position in sheet metal parts subject to forming operations like

Fig. 6. Development of K11 in DC06 during loading in plane-strain tension (in the rolling direction) to 5% (left) and 10% (right) deformation followed by unloading and reloading in
uniaxial tension (again in the rolling direction) predicted by the THM and the LSM. For comparison, experimental and model results are also shown for monotonic uniaxial tension
(Clausmeyer et al., 2011a) and web version for color.

128

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

Fig. 7. Yield surface development in the K 11 ; K 22 plane from the THM (left) and the LSM (right) due to plane-strain tension in the rolling direction to different levels of deformation.
The symbols x, and * mark stress states on the yield surface at F11 1 0.05, F11  1 0.10, and F11  1 0.15, respectively.

deep drawing. An extension to this concept is the recently proposed

cross hardening indicator (Clausmeyer et al., 2014) which considers
the degree of loading-path change and the material specic tendency to exhibit cross hardening.
As mentioned above and shown in Fig. 6 (right), more cross
hardening is predicted by the THM than the LSM after 10% pretension. To examine this in more detail, different contributions to
the general hardening level are displayed in Fig. 8. As evident, the
THM predicts a transient increase in X11 (Fig. 8, left) not predicted
by the LSM (Fig. 8, right) resulting from the coupling of sx in (16) to
S . For this loading path change, note that the change in the shape
of the yield surface predicted by the LSM in the tensionetension
case (Fig. 7, right) is not as pronounced as for the tension-shear case
(Fig. 4, right).
Lastly, consider the FE-based simulation of the tension-bending
of a DC06 sheet metal workpiece. For this purpose, the THM and the
LSM were implemented into LS-DYNA (LSTC, 2012) via the user
material interface. Implicit global time integration was used for the
solution of the initial boundary value problem. The specimen of
dimension 10 mm  10 mm  1 mm was discretized with a regular
mesh of 2000 (20  20  5) tri-linear hexahedral nite elements
(see Fig. 9) with reduced integration and hourglass control (LSTC,
2012). Two tension-bending cases are considered: (i) plane strain
tension in e1 (rolling direction), and (ii) equibiaxial tension in e1
and e2 , both followed by bending in the e2 direction (see Fig. 11
below). After predeformation in tension, a bending moment is
applied on the lower right hand edge of the specimen, and the left

Fig. 9. Initial dimensions and FE discretization of sheet for tension-bending

simulations.

edge is xed. The sheet is bent until a bending angle of 90 is obtained. Bending moment-angle results are shown in Fig. 10; the
deformed sheet and nal spatial stress distribution are displayed in
Fig. 11.
As evident in Fig. 10, bending-moment results based on the THM
and the LSM are qualitatively similar for both loading cases.
Consideration of hardening stagnation in the THM results in faster
saturation of the bending moment after about 60 in particular for
the equibiaxial (pre)tension. The expected trend in stress state and

Fig. 8. Development of r and X11 predicted by the THM (left) and the LSM (right) during loading in plane-strain tension (in the rolling direction to 10%) to uniaxial tension (Fig. 6).

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

129

Fig. 10. Bending moment as a function of bending angle during tension-bending of DC06 predicted by FE simulations based on the THM and the LSM. Left: plane-strain (pre)tension
to F11 1.2. Right: equibiaxial (pre)tension to F11 F22 1.1. See text for details.

Fig. 11. Deformed sheet geometry (bending angle 90 ) after equibiaxial-tension-bending and spatial distribution of the rst principle component KI of the Kirchhoff stress predicted
by the THM (left) and the LSM (right). Refer to the web version for color.

stress distribution from tension (above) to compression (below) the

midplane of the workpiece is clearly predicted by both models. Inplane inhomogeneity of the stress eld perpendicular to this is
related to the nite dimensions of the specimen. As indicated by the
bending moment results in Fig. 10, the THM predicts more hardening in the specimen than the LSM for both plane-strain tension
and equibiaxial tension followed by bending. This is consistent with
the results for KI shown in Fig. 11, i.e., the stress level predicted by
the THM is higher than that predicted by the LSM. Since bending
results in a transition from tension to compression, the above observations for plane strain tension to uniaxial tension loading can
be transferred to the bending case. Up to a bending angle of about
8 bending of the sheet is mainly governed by elastic deformation
because the release of the displacement constraints in the e1 - and
e2 directions is only partly compensated by the incipient straining
in the e1 -direction due to bending. Similar to the plane strain
tension to uniaxial tension case, a transient hardening regime follows. Due to the inhomogeneous nature of the bending deformation, the duration of this regime (in terms of the bending angle) is
larger compared to the homogeneous plane strain tension to uniaxial tension case. However, the same tendencies are observed. The
level of stress predicted by the THM is larger compared to the LSM.
Consequently, the resulting bending moment is also larger. For
larger bending angles, the differences between the two models
become smaller. Note, that the transition between the elastic
regime and the plastic transient hardening regime is sharper for the
LSM compared to the THM. This is in agreement with the results for
plane strain tension to uniaxial tension.
7. Summary and discussion
In the current work, two models for anisotropic, and in particular cross, hardening in bcc sheet steels subject to non-proportional

loading have been compared with each other in detail. The rst
model is the modied form (Wang et al., 2008) of that due to
Teodosiu and Hu (1995, 1998). In this (modied) Teodosiu-Hu
model (THM), cross hardening is assumed to affect (i) the yield
stress and (ii) the saturation value of the back stress. The second
model is due to Levkovitch and Svendsen (2007) and Noman et al.
(2010). In this Levkovitch-Svendsen model (LSM), cross hardening
is assumed to affect the ow anisotropy. As attested to by numerous
applications of the THM (e.g., Boers et al., 2010; Bouvier et al., 2005,
2003; Hiwatashi et al., 1997; Li et al., 2003; Thuillier et al., 2010;
Wang et al., 2008) and the LSM (e.g., Barthel et al., 2013;
Clausmeyer et al., 2014, 2011b; Noman et al., 2010), both models
are able to account for the observed effect of cross hardening on
experimental stress-deformation data. As investigated and documented in the current work, there are otherwise a number of differences between the two. For example, in contrast to the THM, the
LSM predicts no hardening stagnation after cross hardening as
observed in experiments. On the other hand, in contrast to the LSM,
the THM predicts no change in yield surface shape during nonproportional loading. As documented in the current work, this is
in contrast to experimental results (for ferritic tube steel:
Kowalewski and Sliwowski, 1997) as well as to predictions of more
sophisticated micromechanical models (e.g., Holmedal et al., 2008;
Peeters et al., 2002).
As discussed in many previous works as well as here, in the
context of a yield function fY of the form (1), the (ow) anisotropy
tensor A determines the shape of the yield surface. In the THM, this
shape is attributed to the effect of an (initial) rolling texture, i.e., the
grain (orientation) structure, on the ow behavior. On the other
hand, in the LSM, A is associated with both the grain orientation
and dislocation structures, i.e., A A gra H . In the yield surface
context and model form (1) for fY, then, (13) and (17) imply the
correspondence

130

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131


q

 p

S$ A gra H S S$A gra S  f S






(20)

between H and S , where SM  X. Consequently, in the THM, the

coupling between the state of stress M on the yield surface and the
cross hardening tensor S is linear and scalar. As in the case of
standard isotropic hardening itself, then, it is not surprising that
this model decouples any change in form of the yield surface from
cross hardening effects. This is in contrast to the LSM, which couples M and the contribution H from cross hardening to the ow
anisotropy tensor non-linearly and tensorially. In this case, a
directional dependence is generic and inherent. As implied by the
comparisons in Section 5 and Fig. 5, micromechanically based
models like the PM (Peeters et al., 2001b,a; 2002) appear to support
the idea of a change in yield surface shape due to cross hardening as
assumed in the LSM. Being physically based, the PM provides at
least indirect physical evidence for the LSM, since it directly incorporates the physical knowledge of the contribution of dislocation interaction to the evolution of the yield surface.
With respect to the hardening modeling in general, note that
isotropic, kinematic, and cross, hardening are decoupled in the
LSM, but coupled in the THM. From the point of view of the geometry and morphology of the yield surface, then, only in the LSM
are isotropic, kinematic, and cross, hardening uniquely related to
the size, center, and shape, respectively, of the yield surface. As
implied by the current comparison, however, to account for processes such as hardening stagnation, coupling of hardening types
such as that assumed in the THM may be necessary. Indeed, as
implied by the results of the current work as well, except in the case
of the coupling of yield surface morphology and development to
anisotropic hardening in the sense of a change of yield surface
shape, however, the THM is more sophisticated and micromechanical in nature.
The authors are not aware of comprehensive experimental
evidence which shows the evolution of the yield surface of DC06
for larger (pre)strains (z10%). Experiments on tubular specimens
have shown that there is an evolution of the yield surface (Phillips
et al., 1974). In the case of other steels (e.g., SUS304: Ishikawa,
1997), experimentally observed changes in the yield surface
shape during tension-torsion loading are accounted for by the
modeling of cross hardening in the LSM as based on H and related
to dislocation structure evolution. Kuwabara et al. (2000) and
Kuwabara (2007) investigated the effect of uniaxial, plane strain
and biaxial tension on the yield surface of IF steel up to 1.0% strain.
This work focused on the initial yield surface alone. In any case,
more experimental work on sheet steels along these lines is
clearly needed and required, and will hopefully be available in the
future.

Acknowledgments
The authors thank the reviewers of the rst version of this work
for their constructive criticism and comments which have lead to
major improvement. The authors would like to thank Ton van den
Boogaard from the Faculty of Engineering Technology, University
Twente, The Netherlands, for providing the opportunity to use the
biaxial tester, and Alper Gner from the Institute of Forming Technology and Lightweight Construction, TU Dortmund University,
Dortmund, Germany, for providing the uniaxial tension test data.
Partial nancial support from the German Research Foundation
(DFG) under contract PAK 250 Identikation und Modellierung der
Werkstoffcharakteristik fr die Finite-Element-Analyse von Blechumformprozessen - TP4 is gratefully acknowledged. The material
investigated was provided by ThyssenKrupp Steel Europe AG.

References
Bacroix, B., Hu, Z., 1995. Texture evolution induced by strain path changes in low
carbon steel sheets. Metall. Mater. Trans. A 26 (3), 601e613.
Baltov, A., Sawczuk, A., 1965. A rule of anisotropic hardening. Acta Mech. I (2),
81e92.
cio, J.J., Lee, M.-G., Rauch, E.F., Vincze, G., 2013. Extension of
Barlat, F., Ha, J., Gra
homogeneous anisotropic hardening model to cross-loading with latent effects.
Int. J. Plast. 46, 130e142.
Barthel, C., Klusemann, B., Denzer, R., Svendsen, B., 2013. Modeling of a thermomechanical process chain for sheet steels. Int. J. Mech. Sci. 74, 46e54.
Boers, S.H.A., Schreurs, P.J.G., Geers, M.G.D., Levkovitch, V., Wang, J., Svendsen, B.,
2010. Experimental characterization and model identication of directional
hardening effects in metals for complex strain path changes. Int. J. Solids Struct.
47 (10), 1361e1374.
Bouvier, S., Alves, J., Oliveira, M., Menezes, L., 2005. Modelling of anisotropic workhardening behaviour of metallic materials subjected to strain-path changes.
Comput. Mater. Sci. 32, 301e315.
Bouvier, S., Gardey, B., Haddadi, H., Teodosiu, C., 2006a. Characterization of the
strain-induced plastic anisotropy of rolled sheets by using sequences of simple
shear and uniaxial tensile tests. J. Mater. Process. Technol. 174 (1e3), 115e126.
e, P., Teodosiu, C., 2006b. Simple shear tests: experiBouvier, S., Haddadi, H., Leve
mental techniques and characterization of the plastic anisotropy of rolled
sheets at large strains. J. Mater. Process. Technol. 172, 96e103.
Bouvier, S., Teodosiu, C., Haddadi, H., Tabacaru, V., 2003. Anisotropic workhardening behaviour of structural steels and aluminium alloys at large
strains. J. Phys. IV Fr. 105, 215e222.
Butuc, M.C., Teodosiu, C., Barlat, F., Gracio, J.J., 2011. Analysis of sheet metal formability through isotropic and kinematic hardening models. Eur. J. Mech. A/Solids
30 (4), 532e546.
Carvalho Resende, T., Bouvier, S., Abed-Meraim, F., Balan, T., Sablin, S.S., 2013.
Dislocation-based model for the prediction of the behavior of b.c.c. materials e
grain size and strain path effects. Int. J. Plast. 47, 29e48.
Cazacu, O., Barlat, F., 2004. A criterion for description of anisotropy and yield differential effects in pressure-insensitive metals. Int. J. Plast. 20 (11), 2027e2045.
Clausmeyer, T., Bargmann, B., Gershteyn, G., Schaper, M., Svendsen, B., 2011a. Finiteelement Simulation of the Evolution of Plastic Anisotropy. Steel Research International ICTP 2011 Special Issue, pp. 836e841.
Clausmeyer, T., Gerstein, G., Bargmann, S., Svendsen, B., van den Boogaard, A.,
Zillmann, B., 2012. Experimental characterization of microstructure development during loading path changes in bcc sheet steels. J. Mater. Sci. 48 (2),
674e689.
Clausmeyer, T., Gner, A., Tekkaya, A.E., Levkovitch, V., Svendsen, B., 2014. Modeling
and nite element simulation of loading-path-dependent hardening in sheet
metals during forming. Int. J. Plast. 63, 64e93.
Clausmeyer, T., van den Boogaard, A.H., Noman, M., Gershteyn, G., Schaper, M.,
Svendsen, B., Bargmann, S., 2011b. Phenomenological modeling of anisotropy
induced by evolution of the dislocation structure on the macroscopic and
microscopic scale. Int. J. Mater. Form. 4 (2), 141e154.
Ghosh, A.K., Backofen, W.A., 1973. Strain-hardening and instability in biaxially
stretched sheets. Metall. Trans. 4, 1113e1123.
Haddadi, H., Bouvier, S., Banu, M., Maier, C., Teodosiu, C., 2006. Towards an accurate
description of the anisotropic behaviour of sheet metals under large plastic
deformations: modelling, numerical analysis and identication. Int. J. Plast. 22
(12), 2226e2271.
Haddag, B., Balan, T., Abed-Meraim, F., 2007. Investigation of advanced strain-path
dependent material models for sheet metal forming simulations. Int. J. Plast. 23
(6), 951e979.
Hill, R., 1948. A theory of the yielding and plastic ow of anisotropic metals. Proc. R.
Soc. Lond. Ser. A. Math. Phys. Sci. 193 (1033), 281e297.
Hiwatashi, S., van Bael, A., van Houtte, P., Teodosiu, C., 1997. Modelling of plastic
anisotropy based on texture and dislocation structure. Comput. Mater. Sci. 9,
274e284.
Holmedal, B., van Houtte, P., An, Y., 2008. A crystal plasticity model for strain-path
changes in metals. Int. J. Plast. 24 (8), 1360e1379.
Hu, Z., Rauch, E.F., Teodosiu, C., 1992. Work-hardening behavior of mild steel under
stress reversal at large strains. Int. J. Plast. 8 (7), 839e856.
Ishikawa, H., 1997. Subsequent yield surface probed from its current center. Int. J.
Plast. 13 (6e7), 533e549.
Klepaczko, J.R., Nguyen, H.V., Nowacki, W.K., 1999. Quasi-static and dynamic
shearing of sheet metals. Eur. J. Mech. A/Solids 18 (2), 271e289.
Kowalewski, Z.L., Sliwowski, M., 1997. Effect of cyclic loading on the yield surface
evolution of 18G2A low-alloy steel. Int. J. Mech. Sci. 39 (1), 51e68.
Kuwabara, T., 2007. Advances in experiments on metal sheets and tubes in support
of constitutive modeling and forming simulations. Int. J. Plast. 23 (3), 385e419.
Kuwabara, T., Kuroda, M., Tvergaard, V., Nomura, K., 2000. Use of abrupt strain path
change for determining subsequent yield surface: experimental study with
metal sheets. Acta Mater. 48 (9), 2071e2079.
Lee, E., 1969. Elastic-plastic deformation at nite strains. J. Appl. Mech. 36, 1e6.
Levkovitch, V., Svendsen, B., 2007. Accurate hardening modeling as basis for the
realistic simulation of sheet forming processes with complex strain-path
change. In: Cesar de Sa, J.M.A. (Ed.), 9th International Conference on Numerical Methods in Industrial Forming Processes (NUMIFORM 2007), AIP Conference Proceedings, Porto, Portugal, vol. 2, pp. 1331e1336.

T. Clausmeyer, B. Svendsen / European Journal of Mechanics A/Solids 54 (2015) 120e131

Li, S., Hoferlin, E., van Bael, A., van Houtte, P., Teodosiu, C., 2003. Finite element
modeling of plastic anisotropy induced by texture and strain-path change. Int. J.
Plast. 19, 647e674.
LSTC, 2012. LS-DYNA. Keyword User's Manual, vol. I. Livermore Software Technology Corporation, Livermore, USA.
 Classique et Viscoplasticite
. CISM Course Lecture Notes
Mandel, J., 1971. Plasticite
97. Springer, Berlin, Germany.
Mandel, J., 1974. Thermodynamics and Plasticity. In: Foundations of Continuum
Thermodynamics Delgado Domingos. MacMillan.
Miyauchi, K., 1987. On the simple shear deformation. Sci. Pap. Inst. Phys. Chem. Res.
81, 57e67.
Nesterova, E.V., Bacroix, B., Teodosiu, C., 2001a. Experimental observation of
microstructure evolution under strain-path changes in low-carbon IF steel.
Mater. Sci. Eng. A309e310, 495e499.
Nesterova, E.V., Bacroix, B., Teodosiu, C., 2001b. Microstructure and texture evolution under strain-path changes in low-carbon interstitial-free steel. Metall.
Mater. Trans. 32A, 2527e2538.
Noman, M., 2010. Characterization and model identication for the simulation of
t Maschinenbau, TU
the forming behavior of ferritic steels (Ph.D. thesis). Fakulta
Dortmund University.
tink, J., van Riel, M., 2010.
Noman, M., Clausmeyer, T., Barthel, C., Svendsen, B., Hue
Experimental characterization and modeling of the hardening behavior of the
sheet steel LH800. Mater. Sci. Eng. A 527 (10e11), 2515e2526.
Peeters, B., Bacroix, B., Teodosiu, C., van Houtte, P., Aernoudt, E., 2001a. Workhardening/softening behaviour of b.c.c. polycrystals during changing strain:
part i. an integrated model based on substrucure and texture evolution, and its
prediction of the stress-strain behaviour of an IF steel during two-stage strain
paths. Acta Mater. 49 (9), 1607e1619.
Peeters, B., Kalidindi, S.R., Teodosiu, C., van Houtte, P., Aernoudt, E., 2002. A theoretical
investigation of the inuence of dislocation sheets on evolution of yield surfaces
in single-phase b.c.c. polycrystals. J. Mech. Phys. Solids 50 (4), 783e807.
Peeters, B., Seefeldt, M., van Houtte, P., Aernoudt, E., 2001b. Taylor ambiguity in bcc
polycrystals: a non-problem if substructural anisotropy is considered. Scr.
Mater. 45 (12), 1349e1356.
Phillips, A., Tang, J.L., Ricciuti, M., 1974. Some new observations on yield surfaces.
Acta Mech. 20 (1), 23e39.
Pietryga, M.P., Vladimirov, I.N., Reese, S., 2012. A nite deformation model for
evolving ow anisotropy with distortional hardening including experimental
validation. Mech. Mater. 44, 163e173.
Rauch, E.F., Schmitt, J.H., 1989. Dislocation substructures in mild steel deformed in
simple shear. Mater. Sci. Eng. A 113, 441e448.
Rauch, E.F., Thuillier, S., 1993. Rheological behaviour of mild steel under monotonic
loading conditions and cross-loading. Mater. Sci. Eng. A 164, 255e259.
Schmitt, J.H., Shen, E.L., Raphanel, J.L., 1994. A parameter for measuring the
magnitude of a change of strain path: validation and comparison with experiments on low carbon steel. Int. J. Plast. 10 (5), 535e551.

131

Shi, B., Mosler, J., 2012. On the macroscopic description of yield surface evolution by
means of distortional hardening models: application to magnesium. Int. J. Plast. 44,
1e22.
Shinohara, Y., Madi, Y., Besson, J., 2010. A combined phenomenological model for
the representation of anisotropic hardening behavior in high strength steel line
pipes. Eur. J. Mech. A/Solids 29 (6), 917e927.
Svendsen, B., 2001. On the modeling of anisotropic elastic and inelastic material
behaviour at large deformation. Int. J. Solids Struct. 38, 9579e9599.
Tarigopula, V., Hopperstad, O.S., Langseth, M., Clausen, A.H., 2008. Elastic-plastic
behaviour of dual-phase, high-strength steel under strain-path changes. Eur. J.
Mech. A/Solids 27 (5), 764e782.
Tarigopula, V., Hopperstad, O.S., Langseth, M., Clausen, A.H., 2009. An evaluation of a
combined isotropic-kinematic hardening model for representation of complex
strain-path changes in dual-phase steel. Eur. J. Mech. A/Solids 28 (4), 792e805.
Teodosiu, C., Hu, Z., 1995. Evolution of the intragranular microstructure at moderate
and large strains: modelling and computational signicance. In: Shen, S.F.,
Dawson, P.R. (Eds.), Simulation of Materials Processing: Theory, Methods and
Applications. Balkema, Rotterdam, pp. 173e182.
Teodosiu, C., Hu, Z., 1998. Microstructure in the continuum modelling of plastic
anisotropy. In: Proceedings of 19th Ris International Symposium on Material's
Science: Modelling of Structure and Mechanics of Materials from Microscale to
Product. Ris National Laboratory, Roskilde, Denmark, pp. 149e168.
Thuillier, S., Manach, P.Y., Menezes, L.F., 2010. Occurrence of strain path changes in a
two-stage deep drawing process. J. Mater. Process. Technol. 210 (2), 226e232.
Thuillier, S., Rauch, E.F., 1994. Development of microbands in mild steel during cross
loading. Acta Metall. Mater. 42, 1973e1983.
van Riel, M., van den Boogaard, A.H., 2007. Stress-strain responses for continuous
orthogonal strain path changes with increasing sharpness. Scr. Mater. 57 (5),
381e384.
Verma, R.K., Kuwabara, T., Chung, K., Haldar, A., 2011. Experimental evaluation and
constitutive modeling of non-proportional deformation for asymmetric steels.
Int. J. Plast. 27 (1), 82e101.
Viatkina, E.M., Brekelmans, W.A.M., Geers, M.G.D., 2007. Modelling of the internal
stress in dislocation cell structures. Eur. J. Mech. A/Solids 26 (6), 982e998.
Wagoner, R.H., Lim, H., Lee, M.-G., 2013. Advanced issues in springback. Int. J. Plast.
45, 3e20.
tink, J., van Riel, M., 2008. On
Wang, J., Levkovitch, V., Reusch, F., Svendsen, B., Hue
the modeling of hardening in metals during non-proportional loading. Int. J.
Plast. 24 (6), 1039e1070.
Wu, H.-C., 2003. Effect of loading-path on the evolution of yield surface for
anisotropic metals subjected to large pre-strain. Int. J. Plast. 19 (10), 1773e1800.
Zienkiewicz, O.C., Taylor, R.L., Zhu, J.Z., 2010. The Finite Element Method e its Basis
& Fundamentals, sixth ed. Elsevier e Butterworth Heinemann, Amsterdam.