Você está na página 1de 38

Constitutive modeling for anisotropic/asymmetric

hardening behavior of magnesium alloy sheets


Myoung-Gyu Lee
a,1
, R.H. Wagoner
b
, J.K. Lee
c
,
K. Chung
d
, H.Y. Kim
e,
*
a
Eco-Materials Research Center, 66 Sangnam, Korea Institute of Machinery and Materials,
Changwon, Kyungnam 641-010, Republic of Korea
b
Department of Materials Science and Engineering, 2041 College Road, Ohio State University,
Columbus, OH 43210, USA
c
Department of Mechanical Engineering, Scott Laboratory, 201 West 19th Avenue,
Ohio State University, Columbus, OH 43210, USA
d
School of Material Science and Engineering, Seoul National University, 56-1, Shinlim-Dong,
Kwanak-Ku, Seoul 151-742, Republic of Korea
e
Division of Mechanical Engineering and Mechatronics, Kangwon National University, 192-1 Hyoja 2-Dong,
Chunchon, Gangwon-Do 200-701, Republic of Korea
Received 8 March 2007; received in nal revised form 25 May 2007
Available online 14 June 2007
Abstract
Magnesium alloy sheets have been extending their eld of applications to automotive and elec-
tronic industries taking advantage of their excellent light weight property. In addition to well-known
lower formability, magnesium alloys have unique mechanical properties which have not been thor-
oughly studied: high in-plane anisotropy/asymmetry of yield stress and hardening response. The rea-
son of the unusual mechanical behavior of magnesium alloys has been understood by the limited
symmetry crystal structure of HCP metals and thus by deformation twinning. In this paper, the
phenomenological continuum plasticity models considering the unusual plastic behavior of magne-
sium alloy sheet were developed for a nite element analysis. A hardening law based on two-surface
model was further extended to consider the general stressstrain response of metal sheets including
Bauschinger eect, transient behavior and the unusual asymmetry. Three deformation modes
observed during the continuous in-plane tension/compression tests were mathematically formulated
with simplied relations between the state of deformation and their histories. In terms of the anisotropy
0749-6419/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijplas.2007.05.004
*
Corresponding author. Tel.: +82 33 250 6317; fax: +82 33 242 6013.
E-mail address: khy@kangwon.ac.kr (H.Y. Kim).
1
Department of Materials Science and Engineering, Ohio State University, Columbus, OH 43210, USA.
Available online at www.sciencedirect.com
International Journal of Plasticity 24 (2008) 545582
www.elsevier.com/locate/ijplas
and asymmetry of the initial yield stress, the DruckerPragers pressure dependent yield surface was
modied to include the anisotropy of magnesium alloy. The numerical formulations and character-
ization procedures were also presented and nally the correlation of simulation with measurements
was performed to validate the proposed theory.
2007 Elsevier Ltd. All rights reserved.
Keywords: Magnesium alloy sheet; Asymmetry; Two-surface model; Bauschinger eect; Modied Drucker
Prager model
1. Introduction
Over the last several decades, increasing attention has been paid to understand the
deformation behavior of magnesium alloys due to the lowest density of all metallic con-
structional materials (Roberts, 1960; Mordike and Ebert, 2001). In addition to their excel-
lent light property, magnesium alloys have several other advantages: high specic strength,
good welding capability and corrosion resistance. Wrought magnesium alloys have been
reported to have better mechanical properties including tensile and fatigue resistance than
the casting counterparts (Duygulu and Agnew, 2003; Agnew and Duygulu, 2005; Bettles
and Gibson, 2005; Easton et al., 2006; Lou et al., 2007) and therefore more promising
for the potential applications. Taking advantage of these benets, magnesium alloys have
great potential for high performance automotive applications such as steering wheels,
seats, gear box housing and so on. Due to better mechanical properties, resistance to
aging, electrical and thermal conductivity, magnesium alloys have also been substituting
the polymeric materials in the electronic devices industries. In sheet metal forming appli-
cation with magnesium alloys, the lower formability and high springback due to the lower
elastic property (Youngs modulus = 45 GPa, Cubberly et al., 1979; Lou et al., 2007) at
room temperature are major hurdles by which magnesium alloys have limited applications
(Gradinger and Stolg, 2003). Therefore, the studies on the process optimization based on
accurate characterization of the material properties of magnesium alloys are signicantly
necessary to broaden their applications.
In spite of potential applications of magnesium alloys in various areas, limited
researches have been performed, especially for their unusual mechanical properties.
Magnesium alloys have unique plastic behavior compared to other alloys. First, magne-
sium alloys have signicant dierence in initial yield stress for tension and compression.
The initial yield stress for tension is much higher than that for compression (Ball and
Prangnell, 1994). For example, the yield stress of AZ31B magnesium sheet in the trans-
verse direction is 192 MPa, while 110 MPa for the compression (Lou, 2005; Lou et al.,
2007). The dierence in tensile and compressive yield stresses is often denoted as (yield)
asymmetry. In addition to the yield asymmetry, magnesium alloys show asymmetry in
the plastic ow stress (ow asymmetry). Literature survey on the plastic behavior of
magnesium alloys with strong basal texture reveals the following common mechanistic
behavior.
v During the uniaxial tensile deformation from undeformed state, hardening curve is nor-
mal concave-down shape. Basal slip dominates during the deformation with other con-
tributions of non-basal slip and twinning to maintain local compatibility. Therefore, the
546 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
conventional constitutive equations can contribute to the slip deformation. That is, the
stressstrain curve during the slip deformation can be tted by exponential or power-
law type hardening equations.
v Unusual concave-up or S-shape hardening curves are observed during in-plane com-
pression or tension following compression. The unique inected ow curves are caused
by the activation of twinning or exhaustion of twinning. When the loading is reversed
from twinning, untwining initiates due to the contraction of twinned regions without
nucleation. For more details, refer to Lou et al. (2007).
Fig. 1 shows the schematic view of typical stressstrain curves of cubic crystal alloys
and textured magnesium alloy sheets. The gure shows reduced yield stress during reverse
loading, which is associated with the Bauschinger eect (Bauschinger, 1886) for both mate-
rials. For the cubic metals such as aluminum alloys, the magnitude of yield stress in ten-
sion and compression is assumed to be same and ow stress curve during tension and
compression is also symmetric, while strong asymmetries in yield stress and ow curves
are shown for the textured magnesium alloy sheets. In sheet metal forming applications,
the non-monotonous deformation is especially important because reverse loading is com-
monly observed when sheet element moves through the tool radii and draw beads. Also,
when sheet parts are removed from tools after forming, material elements experience elas-
tic unloading and springback.
Two main approaches have been made to describe the reverse loading behavior in
the continuum phenomenological plasticity: one based on kinematic hardening involv-
ing shifting of a single-yield surface and the other involving multiple yield surfaces
(Khan and Huang, 1995). The simplest one in the former group is based on linear
kinematic hardening proposed by Prager (1956), Ziegler (1959) and Hodge (1957)
to describe the Bauschinger eect. Note that these phenomenological models ignore
the microplastic eects and introduce mathematical rule of translation of yield sur-
face. To add the transient behavior, the linear model was modied to nonlinear forms
by Amstrong and Frederick (1966) and Chaboche (1991) by introducing an additional
term to Pragers linear kinematic hardening model. Nonlinear and smooth deforma-
tion during loading and reverse loading were reproduced by introducing additional
back stress term which makes total back stress decrease gradually with deformation.
Several nonlinear kinematic hardening models based on AmstrongFredrick model
have been emerged by introducing multiple back-stress terms (Ohno and Wang,
1993a,b) and translating limiting surface (Ohno and Kachi, 1986; Bower, 1989).
The Chaboche model was further generalized recently as a combined type model, uti-
lizing a non-quadratic anisotropic yield function and the Ziegler kinematic hardening
model (1959), based on the plastic work equivalence principle modied for kinematic
hardening to properly dene eective (or equivalent) quantities in stress and plastic
strain rate (Chung et al., 2005; Lee et al., 2005a,b). This modied Chaboche model
only accounts for the Bauschinger and transient behavior, not the permanent soften-
ing (Kim et al., 2005). The nonlinear kinematic hardening model has been further
modied to incorporate the permanent softening as well as the Bauschinger eect
and transient behavior (Geng and Wagoner, 2002; Geng et al., 2002; Chun et al.,
2002a,b).
Another frequently used family of phenomenological hardening models is multi-
surface model. The classical Mroz model involves multiple numbers of yield surfaces
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 547
in which piecewise linear variation of hardening is dened (Mroz, 1967) according to
Pragers kinematic hardening model. On the other hand, two-surface models indepen-
dently proposed by Krieg (1975) and Dafalias and Popov (1976) dene the continuous
variation of hardening between two yield surfaces. In the original multi-surface model
proposed by Mroz (1967), the predicted stressstrain curve is piecewise linear because
Fig. 1. Schematic view of typical stressstrain curves of: (a) cubic crystal alloy sheets which show symmetric yield
stress in tension and compression as well as reduced reverse yield stress (Bauschinger eect) and (b) textured
magnesium alloys which show asymmetric yield stress in tension and compression, Bauschinger eect and
asymmetry in ow curves with three deformation modes.
548 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
of the constant plastic moduli (Khan and Huang, 1995). Therefore, an innite number of
yield surfaces are needed to predict a smooth nonlinear curve as proposed by Mroz and
Niemunis (1987), while the two-surface model can represent realistic smooth hardening
with the continuous plastic modulus. However, the two-surface model may lead to the
discontinuous change of elasto-plastic stiness after partial unloading and also may pro-
duce unrealistic behavior such as too strong ratcheting as addressed by Hashiguchi
(1997) and Chaboche (1986). Several multi/two-surface models have been proposed later
(Mroz et al., 1979; Hashiguchi, 1981; McDowell, 1985; Hashiguchi, 1988; Khoei and
Jamali, 2005; Lee et al., 2007) to analyze the one-dimensional cyclic behavior of solid
structures at small strains.
In the continuum plasticity, in order to describe initial yield anisotropy, mathematical
representations of the yield surfaces are combined with hardening behavior for numerical
modeling. For the FCC and BCC materials which do not have strong asymmetry in yield-
ing, several yield criteria have been proposed from isotropic (von Mises, 1928; Tresca,
1864; Hosford, 1972) to anisotropic yield surfaces (Hill, 1948; Hill, 1979; Barlat et al.,
1991; Barlat et al., 2003).
Although considerable eorts have been made for the constitutive equations which can
reproduce more realistic mechanical behavior of metallic materials, few phenomenological
models have been reported for the asymmetric materials including magnesium alloys (Li,
2006; Kim et al., 2007a). Modeling constitutive equations for the magnesium alloys
requires understanding of plastic behavior of more complex deformation paths including
reverse loading condition. The challenge is met by their unusual hardening curves with
large asymmetry and considerable Bauschinger eect. Other attempts to explain yielding
asymmetry and anisotropy of HCP metals have also been made by crystal plasticity.
For example, self-consistent (SC) methods were adopted to simulate tension and compres-
sion behavior of textured HCP alloys (Lebenson and Tome, 1993; Brown et al., 2005) and
nite element analysis with crystal plasticity model was performed to incorporate defor-
mation twinning (Kalidindi, 1998). More recently, a crystal-mechanics-based constitutive
model to account for the twinning was developed for polycrystalline magnesium alloy
AZ31B (Staroselsky and Anand, 2003). Although the texture or crystal plasticity models
were partially successful in predicting deformation behavior of magnesium alloys, it is nec-
essary for these models to contain large number of orientations at each integration point.
Thus, these approaches are computationally inecient in forming application with nite
element method.
In the present study, constitutive models for the sheet magnesium alloys are devel-
oped based upon the phenomenological continuum plasticity, which is capable of
describing the yielding asymmetry and anisotropy in stressstrain response. For the
hardening model, two-surface model is modied to include ability to model the three
deformation modes: slip, twinning and untwining. In terms of asymmetry in the initial
yield stress and anisotropy, the modied DruckerPrager model is adopted with aniso-
tropic coecients under plane stress conditions. Based on the developed theory, charac-
terization procedures for the material parameters from continuous in-plane uni-axial
tension/compression tests are presented. The models have been implemented into a
nite element program, ABAQUS/Standard with user material subroutine that success-
fully reproduces the main features of the experimental results for the unusual stress
strain responses of O-tempered AZ31B and AZ31B without tempering magnesium alloy
sheets.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 549
2. Mechanical behavior of magnesium alloys
2.1. Anisotropy/asymmetry
In the phenomenological continuum plasticity, the yield surface is one of the main con-
stitutive equations which complete the response of materials. Magnesium (Mg) alloy sheet
and HCP metals show high anisotropy and strong eccentricity in yielding. In other words,
the yield stress shows directional dierence (anisotropy) and dierence in tension and com-
pression (asymmetry or eccentricity). It has been well known that the yielding asymmetry
in in-plane tension and compression is associated with the strong basal texture and the
activation of twinning during compression. For example, the tensile yield stress in textured
magnesium alloy is much larger than that of compressive stress (Hosford, 1993; Lou et al.,
2007).
Two dierent ways have been proposed to represent the eccentricity. The rst method is
to introduce the pressure dependent or invariant of stress into the existing anisotropic
yield functions. For example, Hosford (1966, 1972) added linear stress terms into Hills
48 yield surface (Hill, 1948), while Cazacu and Barlat (2001, 2004) added third invariant
of stress deviator into the orthotropic yield surface which was modied from the isotropic
Drucker yield criterion (Drucker, 1949). The other method is to introduce initial transla-
tion of yield surface by assuming non-zero back-stress in the combined isotropickine-
matic hardening model. Yoon et al. (1998) represented dierent in-plane tension and
compression yield stresses of aluminum alloy 2008-T4 by using asymmetry model incorpo-
rated into Barlats96 anisotropic yield function (Barlat et al., 1997) and the earring predic-
tion was in good accordance with the measurement. More recently, Li (2006) also used
non-zero initial back stresses along with the von Mises isotropic yield function and proper
translation rule consistent with texture evolution of magnesium alloys to simplify the
formulation.
2.2. Hardening behavior
Besides the yielding asymmetry and high in-plane anisotropy, magnesium alloy sheets
have unique hardening behavior during the plastic deformation. Inected stressstrain
curves during in-plane compression and subsequent tension tests are reported for the mag-
nesium alloys. These unusual behaviors are due to the twinning and untwining deforma-
tion which induce abrupt grain reorientation, creation and disappearance of twin
boundaries. As schematically shown in Fig. 1b, the typical stressstrain curves of magne-
sium alloy sheets have dierent behavior for the following three deformation modes: (a)
in-plane tension, (b) in-plane compression and (c) compression followed by tension. The
tensile response is similar to the behavior of FCC or BCC metals which have strong sym-
metry in slip activities and at least ve independent slip systems exist. The dislocation slip
mechanism is dominant and the shape of stressstrain curve is normal concave-down in
this deformation mode. For the in-plane compression mode, twinning mechanism is pre-
dominant with abrupt texture change and exhausted with continuous compression. When
the twinning is exhausted, the slip initiates again and the ow stress rises rapidly. There-
fore, the stressstrain curve during in-plane compression shows unusual concave-up shape.
For the tension following compression, the deformation is similar to the in-plane compres-
sion. During the early tensile deformation, untwining mode dominates so that the stress
550 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
strain curve in this region shows concave-up shape. When the untwining exhausts with fur-
ther tensile deformation and no further texture change exists, the dislocation slip mode
takes over again. Therefore, the shape of stressstrain curve during in-plane tension fol-
lowing compression is unusual S-shape. The summary of the stressstrain responses and
evolution of microstructure for the magnesium alloy sheet is well presented in the previous
work by Lou et al. (2007).
3. Theory
3.1. Summary of the two-surface theory
In the two-surface model, two yield stress surfaces are employed as shown in Fig. 2: (1)
the loading surface on which the current stress at point a in Fig. 2 is dened and (2)
bounding surface on which a corresponding stress at point A in Fig. 2 is dened. In gen-
eral, these two surfaces have the same shape and translate and/or expand at the same time.
The correspondence between points a and A is dened by the common yield surface nor-
mal directions. The gap between the current and corresponding stresses determines the
hardening rate, which is initially steep but becomes smaller as the gap decreases. Since
the two surfaces can make contact only at points sharing the same normal direction, con-
tact is imposed to occur at the current stress and corresponding stress points in the model.
For the detailed description of the two-surface model and its problematic issues are
reported in the previous article by Lee et al. (2007).
The general yield function for the loading surface is
U(r a) r
iso
= 0: (1)
Fig. 2. A schematic view of the two-surface model and two gap distances aA and bB.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 551
Here, r is the Cauchy stress and a is the back-stress, which denes the central position of
the current yield stress surface (with a = 0 initially). Also, r
iso
is the eective stress, a mea-
sure of the size of the yield surface. Note that the function used in the present paper is rst-
order homogeneous and will be described in the later section.
From Eq. (1) and the plastic work equivalence principle,
oU
o(r a)
dr
oU
o(r a)
da h
iso
(r a)
r
iso
de
p
= 0 (2)
where h
iso
=
d r
iso
de
_ _
is the slope of the isotropic hardening curve, r
iso
, as a function of the
eective plastic strain e =
_
de
_ _
, while
da =
da
r
iso
(m)
m =
da
de
de
_ _
m
r
iso
(m)
: (3)
where the relation U(da) = U(dm) = dc
1
r
iso
(m) = da is utilized for the rst-order homoge-
neous function and the translational direction of the loading surface m ~ (r a) is
adopted.
Considering linear elasticity and the additive decomposition of the strain increment,
dr = C de
e
= C (de de
p
) (4)
where C is the elastic modulus, while de and de
e
are total and elastic strain increments,
respectively. The plastic strain increment is dened by the normality rule as
de
p
= dk
oU
o(r a)
= dk
o r
iso
o(r a)
= de
o r
iso
o(r a)
; (5)
After some manipulations,
de =
o r
iso
o(ra)
C de da
o r
iso
o(ra)
C
o r
iso
o(ra)
h
iso
: (6)
Therefore, for a given strain increment de prescribed at every time increment, Eqs. (5) and
(6) determine the plastic strain increment, while the stresses are updated by Eqs. (4) and (5)
on the loading surface as Jaumann increments.
Similarly to the loading surface, the bounding surface is described as
P(R A) R
iso
= 0 (7)
where R and A are the stress and back-stress of the bounding surface, respectively. Also,
R
iso
represents the size of the bounding surface and it is pre-determined by the proper
decomposition. Since the bounding surface P shares the same shape with the loading sur-
face U and the corresponding stress on the bounding surface R shares the same normal
direction with the current stress r at the loading surface,
R A =
R
iso
r
iso
(r a) (8)
As for the back-stress evolution, the following condition is imposed in addition to Eq. (3):
dA da = dl(R r) or dA = da dl(R r) = dA
1
dA
2
(9)
552 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
The condition in Eq. (9) species that two surfaces relatively translate along the line be-
tween the current and corresponding stresses, which ensures that contact between two sur-
faces occur at current and corresponding stresses. For the second term of the last term of
Eq. (9),
dA
2
=
dA
2
r
iso
(R r) (10)
where dA
2
= r
iso
(dA
2
) and r
iso
= r
iso
(R r) as similarly done in Eq. (3).
In the two-surface model, the expansion and translation of the two surfaces are param-
eterized such that the Bauschinger, transient and permanent softening behaviors are prop-
erly represented and are matched to the 1-D reference state. For the reference state, the
stress relationship becomes
R = r

d or
dR
de
=
d r
de

d

d
de
_ _
(11)
where

d is the gap between R and r, the stress states at the bounding and loading surfaces,
respectively. Initially, the hardening of the bounding surface is prescribed for expansion,
R
iso
(e), and translation, A(e) (or A
2
(e)) with proper separation. Then, r(e) is obtained to
account for the transient behavior every time reverse loading occurs, considering the pre-
scribed gap function

d(e), which is dependent on

d
in
, the initial gap distance measured at
the start of reverse loading. The separation of r into the expansion and translation, r
iso
(e)
and a(e), is executed to properly account for the Bauschinger eect. For the hardening
model (d r > 0), the simple decomposition as following,
d r = m
l
d r (1 m
l
)d r = d r
iso
da (12)
dR = m
b
dR (1 m
b
)dR = dR
iso
dA (13)
where m
l
and m
b
are the ratios of the isotropic hardening for the loading and bounding
surfaces, respectively, which are the functions of the accumulative plastic strain in general,
however, constant values are assumed here for simplicity. When the loading path is not
proportional, caution should be made for the separation of the isotropic hardening and
the kinematic hardening in the loading surface. The gap

d is the distance between the cur-
rent stress on the loading surface and the corresponding stress on the bounding surface
(marked a and A in Fig. 2), while the gap

n is the distance between the corresponding stres-
ses (marked b and B in Fig. 2) aligned with the line connecting two centers of the loading
and bounding surfaces. Note that premature contact at b and B should be avoided in the
two-surface model with the proper separation of the isotropic and kinematic hardening in
order not to penetrate the bounding surface.
The scalar parameter

d to measure the gap between the current stress at the loading sur-
face and the corresponding stress at the bounding surface is dened here as

d = r
iso
(R r) = U(R r) (14)
which is the eective stress value obtained by replacing r a with R r.
In order for the current two-surface model to be used in a practical way for the
plane stress problem, it is ecient to properly dene the reverse loading criterion such
that the new initial gap distance

d
in
is updated only when the reverse loading criterion is
satised. Fig. 3 shows the reverse loading criterion introduced here, in which h
d
is the
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 553
angle between two subsequent stresses on the loading surface, while h
r
is a prescribed
reference angle for reverse loading: the reverse loading condition that (06) h
r
6 h
d
(6p)
where
h
d
= cos
1
r a
[r a[

new

r a
[r a[

old
_ _
= cos
1
(d
new
d
old
): (15)
The hardening update which involves the new gap function and reverse loading condi-
tion is performed considering the following linear combination associated with new and
previous initial gaps and the parameters c
a
and c
b
, which are the functions of the angle
h
d
:

d = (1 c
a
)

d(

d
old
in
; e
i
= (1 c
b
)e
old
)
c
a


d(

d
new
in
; e
i
= (1 c
b
)e
old
); 0 6 c
a;b
6 1

d =

d(

d
new
in
; e
i
= 0); c
a;b
> 1
(16)
Here,

d
old
in
and

d
new
in
are initial gap distances for the previous and the current loading curves,
respectively, while e
i
and e
old
are the initial plastic strain for the reversed hardening curve
and the strain at the load reversal from the previous loading curve, respectively. The
parameters are supposed to be experimentally determined, however, c
a
= c
b
= [h
d
/h
r
[ is as-
sumed here for simplicity.
3.2. Hardening behavior of Mg alloys
Although the conventional two-surface models can eectively represent realistic hard-
ening behavior such as Bauschinger eect, transient behavior and permanent softening,
it is still necessary for new or modied constitutive models developed to represent unusual
asymmetric stressstrain response in in-plane tension, compression and continuous ten-
sioncompression (or compressiontension). As discussed in Section 2.2, the magnesium
alloys (or HCP metals) have three dierent deformation modes: slip mode, twinning mode,
and untwining mode. Therefore, the gap distance

d should be updated for dierent defor-
mation modes in addition to the initial gap distance

d
in
when the loading is reversed. The
r
d
old
d
new
d
reverse loading criterion

Fig. 3. Reverse loading criterion for the two-surface model.


554 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
twinning deformation is known to be closely related with c/a ratio (Roberts, 1960; Yoo,
1981) in hexagonal close packed (HCP) alloys. The most activating twin system in magne-
sium alloys is 10

1210

1) where c/a (=1.624) is less than

3
_
. Therefore, this twin
mode is a tensile twin which induces the extension of the c-axis in magnesium alloy. Note
that the extension of the c-axis in rolled sheet alloy is equivalent to the in-plane compres-
sion. Twinned magnesium alloys may undergo disappearance of twin density by subse-
quent deformation, which is called as untwining. In order to simplify the model and for
the practical application in the sheet metal forming analysis, the following assumptions
are made in the present study:
(1) The stress through the thickness is negligible so that plane stress state can be applied.
(2) The initial texture of magnesium alloy is perfect basal texture where all crystal c-axes
align parallel to the sheet normal (thickness direction) and all a-axes are randomly
distributed in the sheet plane. This assumption is veried for the annealed AZ31B
magnesium alloy by measuring pole gures in the previous research (Lou et al.,
2007).
(3) There is no texture evolution during the slip mode. On the other hand, the texture
evolution for twinning and untwining is same but reverse direction each other.
The critical strain for the complete rotation of c-axis during twinning and untwining
is assumed to be known and the angle of rotation is 90.
With the above assumptions, the modied hardening model considering the three defor-
mation modes is proposed. The gap distance as a general form is

d =

d(

d
in
; c
old
; c
new
; e
p
) (17)
where c
old
and c
new
are the vectors of c-axes before and after rotation, respectively. Eq. (17)
denotes that the gap distance is updated by considering the initial gap distance, the history
of texture evolution and the magnitude of plastic deformation. The details on the criteria
to determine the three hardening modes are as following.
As well reported in the previous article (Roberts, 1960), deformation twinning occurs
under both compressive and tensile deformations. However, for magnesium alloy which
has c=a <

3
_
, the tensile twin that elongates the c-axis during the deformation dominates.
Therefore, the criterion for the twinning mode includes the stress state which causes
elongation of the current direction of c-axis. Here, current means the c-axis before
rotation and usually thickness direction of sheet plate. This deformation mode can be also
called as thickening mode. In the plane stress state, the thickening mode is determined
by the principal increment of plastic strain rate. During the deformation, the yield stress
surface will translate and expand according to the isotropickinematic hardening rule,
which is shown in Fig. 4a. By the normality rule, the plastic strain increment is normal
to the yield surface at the corresponding stress point. Therefore, the two components of
principal plastic strain increment can be produced as shown in Fig. 4(b). This is often
called as strain rate potential. The thickening or twinning mode is called when the fol-
lowing conditions are satised:
De
p
1
De
p
2
P0 (18)
where De
p
1;2
= _ e
p
1;2
Dt are principal values of plastic strain increments during the time step
Dt. Therefore, under plane stress condition,
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 555
De
p
1;2
=
De
p
xx
De
p
yy
2

De
p
xx
De
p
yy
2
_ _
2
De
p
xy

(19)
due to relation De
p
1
De
p
2
= De
p
xx
De
p
yy
. During the twinning deformation, the c-axes ro-
tate onto the loading plane. The amount of texture evolution (or rotation) may be ex-
pressed by the functional relation with the critical strain, e
twin;cr
for the evolution to be
completed. Linear proportional factor is introduced in this study:
r
twin
=
1
e
twin
e
twin;cr
(e
twin
6e
twin;cr
)
0 (e
twin
> e
twin;cr
)
_
; (20)
where e
twin
is accumulated plastic strain during the twinning deformation and r
twin
denotes
residual of c-axis remaining in the original direction. Thus, when e
twin
reaches e
twin;cr
, all of
the c-axes completely rotate on the sheet plane.
The criterion for the untwining mode is similar to the twinning mode. However,
the criterion should consider the history of deformation before initiating untwining
Fig. 4. (a) Yield stress surface considering translation and expansion by the isotropickinematic hardening rule
and corresponding and (b) locus of principal plastic strain increment and criterion for the thickening and
thinning modes.
556 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
process because the untwining can only occur for the material which experienced the twin-
ning process in advance. Therefore, in order for the untwining mode to occur, the follow-
ing conditions should be satised. First, the loading history before untwining should
satisfy the twinning criterion. For the simplied rst-order approach, the threshold
(r
+
twin
) of residual of c-axis is introduced. Thus, only when the texture residue is less than
this threshold during the previous twinning process, the untwining can be initiated. Since
the untwining mode occurs by the reverse process of twinning mode, the deformation
accompanies thinning mode:
r
twin
6 r
+
twin
and De
p
1
De
p
2
< 0: (21)
In terms of the texture evolution during the untwining process, the similar method as done
for the twinning mode is adopted:
r
untwin
=
1
e
untwin
e
untwin;cr
(1r
twin
)
= 1
e
untwin
e
/
untwin;cr
(e
untwin
6e
/
untwin;cr
)
0 (e
untwin
> e
/
untwin;cr
)
_
; (22)
where e
untwin;cr
and e
untwin
are critical strain for the complete untwining from the perfect
twinned deformation state (r
twin
= 0) and accumulated plastic strain during the untwin-
ing deformation, respectively. Similarly, the threshold r
+
untwin
is used to determine the
initiation of twinning mode when the loading is revered from the untwining
deformation.
The criterion for the slip mode occurs when the deformation state does not satisfy
either the twinning mode or untwining mode. For example, when tensile loading (or
thinning mode) is applied from the initial undeformed state or when the texture evolu-
tion exhausts during the twinning or untwining modes or when the residue of c-axis dur-
ing the previous twinning (or untwining) is larger than the prescribed threshold value,
the slip modes are activated. Note that there is no rotation of c-axis during the slip
mode.
To characterize the hardening behavior of three dierent modes from the experimen-
tally obtained stressstrain responses, the 1-D ow curves are generalized with a

0
1 =
1 0 <
t

=
1
1 0 <

Fig. 5. Schematic illustration of S-shaped loading curve and corresponding bounding curve.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 557
mathematical description. Under the current two-surface scheme, two hardening curves
need to be dened: bounding curve (R) and loading curve ( r). In order to account for
asymmetry in tension and compression, two dierent hardening curves or gap distances
are assigned to satisfy the plastic work rate equivalence. If the thinning mode is chosen
for a reference hardening curve which is usually obtained by the uni-axial tension
test from the undeformed state, the gap distance between bounding and loading curves
is

d = U(R
xx
r
xx
) = R
xx
r
xx
= R r = R ( r
t
#(R r
t
)) = (1 #)(R r
t
)
(23)
where R
xx
and r
xx
are uni-axial tensile stresses on the loading and bounding curves,
respectively, r
t
is the reference stress as indicated in Fig. 5, and # is a sigmoid function
with S-shape, which gives a value between 0 and 1.
On the other hand, in terms of the hardening data (eective value of gap distance) for
the thickening mode or twinning mode, the curve is obtained from the uni-axial compres-
sive test with the following conditions:

d =
r
xx;0
r
C
xx;0
d
C
e
p;C
=
r
C
xx;0
r
xx;0
e
_ _
(24)
where d
C
is measured gap distance during the twinning (compressive) deformation which is
similar for dened in Eq. (23),
r
xx;0
r
C
xx;0
is the asymmetry ratio which shows dierence in initial
tensile and compressive stresses, and superscript C denotes the compressive
deformation.
The Eq. (23) can be re-written as a rate form:
d

d
de
=
d#
de
(R r
t
) (1 #)
d r
t
de
#
dR
de
: (25)
The material parameters for the functions r
t
and # are characterized by tting the stress
strain curves of continuous in-plane tension/compression (or compressiontension) tests
of magnesium alloys with dierent pre-strains (Boger et al., 2005). Note that these param-
eters are functions of initial gap distance after load reversal in the conventional two-sur-
face model as well as the history of deformation which decides three dierent modes.
Therefore,
r
t
= r
t
(

d
in
; c
old
; c
new
; e
p
) (26)
# = #(

d
in
; c
old
; c
new
; e
p
) (27)
3.3. Yield function of the anisotropic/asymmetric materials
In order to consider yielding asymmetry in tension and compression, classical
approaches have introduced hydrostatic pressure. For instance, the Coulomb-Mohr yield
criterion modies the Trescas isotropic yield criterion by adding the eect of mean stress
on yielding, while the DruckerPrager criterion is based on the von Mises yield criterion
558 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
with hydrostatic stress term. The models can reasonably represent the yielding behavior of
pressure sensitive materials like porous medium.
More recently, the yield criterion for the HCP metals has been developed by Cazacu
et al. (2004) where the linear transformation of the deviatoric Cauchy stress is adopted.
The yield surface is dened as
r = (U
c
=b)
1
a
; U
c
= ([S
1
[ kS
1
)
a
([S
2
[ kS
2
)
a
([S
3
[ kS
3
)
a
(28)
where r is eective stress and S
1~3
are the principal values of the following matrix under
the plane stress condition:
~s = C
/
s; C
/
=
C
/
11
C
/
12
C
/
13
0
C
/
12
C
/
22
C
/
23
0
C
/
13
C
/
23
C
/
33
0
0 0 0 C
/
66
_
_
_
_
_
_
_
_
_
_
(29)
where C
/
is a fourth-order tensor containing anisotropy coecients and s is the deviatoric
Cauchy stress. Therefore, the principal values of ~s becomes
S
1;2
=
~s
xx
~s
yy
2

~s
xx
~s
yy
2
_ _
2
~s
2
xy

; S
3
= ~s
zz
(30)
The coecient of anisotropy and other material constants can be obtained by the tension
and compression tests along three dierent directions. For more details, refer to the article
by Cazacu et al. (2004).
In the present study, to further simplify the criterion by reducing the material param-
eters under the plane-stress condition, the DruckerPrager model is slightly modied by
adding the anisotropic coecients. Note that the criterion in Eq. (28) can be also quadratic
form with a = 2. The criterion adopted in the paper is
U = p(r
2
xx
b
2
r
xx
r
yy
b
2
2
r
2
yy
3b
2
3
r
2
xy
)
1=2
q(r
xx
b
4
r
yy
) r
iso
= 0; (31)
where r
iso
denotes the size of the yield surface and the ve parameters p, q, b
2
, b
3
and b
4
are the material constants to be determined experimentally. Note that the proposed yield
criterion is a rst-order homogeneous function of the stress tensor for the eective stress.
The yield criterion reproduces the von Mises yield criterion if b
2
= b
3
= 1, q = 0 and the
classical DruckerPrager yield criterion if b
2
= b
3
= b
4
= 1 and q ,= 0. The above yield cri-
terion was applied to predict the bending behavior of polymeric composite material which
has high anisotropy and asymmetry (Kim et al., 2007b).
Besides the yield criterions ability in describing anisotropy and asymmetry (or eccen-
tricity), it can also prescribe shear yield stress independently. Therefore, ve material
parameters can be determined from the two tensile yield stresses r
T
xx
, r
T
yy
, two compressive
yield stresses r
C
xx
, r
C
yy
in the x, y directions, and the shear yield stress r
xy
or tensile yield
stress r
T45
xx
in the 45 direction.
With tensile and compressive yield stresses and Eq. (31):
pr
T
xx
qr
T
xx
= r
iso
pr
C
xx
qr
C
xx
= r
iso
(32)
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 559
where the reference value is the tensile yield stress in the x-direction, r
iso
= r
T
xx
and T and
C denote tension and compression, respectively. For the tensile and compressive tests in
the y direction, Eq. (31) gives
p[b
2
[r
T
yy
qb
4
r
T
yy
= r
iso
p[b
2
[r
C
yy
qb
4
r
C
yy
= r
iso
(33)
From the experimentally measured yield stresses in tension and compression, the param-
eters p, q, b
2
and b
4
are obtained.
When the shear stress r
xy
is known, Eq. (31) gives

3
_
p[b
3
[r
xy
= r
iso
(34)
Since the pure shear stress is hardly measured with accuracy, an alternate 45 tension test
can be utilized to determine b
3
by solving the following equation:
p
r
T45
xx
2
_ _
2
b
2
r
T45
xx
2
_ _
2
b
2
2
r
T45
xx
2
_ _
2
3b
2
3
r
T45
xx
2
_ _
2
_ _
1=2
q
r
T45
xx
2
b
4
r
T45
xx
2
_ _
= r
iso
(35)
The solutions of Eqs. (32)(35) are given as follows:
p =
1
2
1
r
T
xx
r
C
xx
_ _
(36)
q =
1
2
1
r
T
xx
r
C
xx
_ _
(37)
b
2
=
(r
T
yy
r
C
yy
):r
T
xx
:r
C
xx
(r
T
xx
r
C
xx
):r
T
yy
:r
C
yy
(38)
b
4
=
(r
C
yy
r
T
yy
):r
T
xx
:r
C
xx
(r
C
xx
r
T
xx
):r
T
yy
:r
C
yy
(39)
b
3
=
2

3
_
r
T
xx
:r
C
xx
(r
T
xx
:r
C
xx
)r
xy
or (40)
b
3
=
1

3
_
1
p
2
2
r
iso
r
T45
xx
q(1 b
4
)
_ _
2
(1 b
2
b
2
2
)
_ _
1=2
3.4. Numerical implementation
For the numerical formulation, the incremental deformation theory (Chung and
Richmond, 1993) was applied to the elasto-plastic formulation based on the materially
embedded coordinate system. For a given total strain increment De and the other state
variables from the previous time step, the numerical formulation provides increments of
elastic and plastic strain, Cauchy stress and back stress. The stored state variables at
560 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
the end of the previous time step are stress, back stress, plastic strain, and the informa-
tion on the orientation of c-axes along with their residues. During the incremental step,
the magnitudes of all of these variables are functions of the incremental eective strain
De only. Therefore, the following nonlinear equation for De is valid for the loading
surface:
U(r
0
a
0
Dr(De) Da(De)) = r
iso
(e
0
De) (41)
along with two hardening curves, r
iso
(e) and a(e), which are prescribed in advance once
reloading criterion is satised and the deformation mode is determined as discussed in
the previous section. After De is obtained as a solution of Eq. (41), updated stress and
back-stress are obtained from r
iso
(e) and a(e), respectively, while the new conguration
of the bounding surface is also updated at the end of each step considering R
iso
(e),
A
2
(e) and a(e) for De. The two hardening curves of the loading surface, r
iso
(e) and a(e),
are newly updated every time reloading occurs, considering the new initial gap distance

d
in
, corresponding gap function and deformation history which decides one of three defor-
mation modes.
The predictorcorrector scheme was used to solve De in Eq. (41) for the loading surface;
i.e.,
F = U(r
n1
a
n1
) r
iso
(De) = 0; (42)
where
r
n1
= r
T
n1
DeC
o r
iso
o(r
nb
a
nb
)
(43)
and
a
n1
= a
n
Da(De)
(r
nb
a
nb
)
r
iso
; (44)
where 0 6 b 6 1.
In Eq. (43), the superscript T stands for a trial state and the subscript denotes the pro-
cess time step. Therefore,
r
T
n1
= r
n
C De: (45)
Eq. (42) is a nonlinear equation to solve for De, when De is given. Then, linearization of
Eq. (42) leads to
d(De)
k1
= F
k
_
oF
oDe
_ _
k
(46)
for the kth iteration and
oF
oDe
=
oF
or
n1
or
n1
oDe

oF
oa
n1
oa
n1
oDe

oF
o r
iso;n1
o r
iso;n1
oDe
; (47)
where
or
n1
oDe
= C
o r
iso
o(r
nb
a
nb
)
(48)
oa
n1
oDe
=
oDa
oDe

r
nb
a
nb
r
iso
(49)
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 561
oF
or
n1
=
oF
oa
n1
=
o r
iso
o(r
n1
a
n1
)
(50)
oF
o r
iso
= 1: (51)
After e
n1
(therefore, along with r
n+1
and a
n+1
) is obtained for the loading surface, the cur-
rent stress on the bounding surface R
n+1
and its center A
n+1
(or DA
2
) are obtained from
the following two conditions:
A
n1
= A
n
Da
n
DA
2
(De
n1
)
(R
n1
r
n1
)
r
iso
(R
n1
r
n1
)
(52)
R
n1
A
n1
=
R
iso
(De
n1
)
r
iso
(De
n1
)
(r
n1
a
n1
) (53)
which are two simultaneous equations for the stress on the bounding surface and the cen-
ter of the surface: R
n+1
and A
n+1
. Therefore, adding Eqs. (52) and (53), the following non-
linear equation is obtained for the unknown quantity R
n+1
:
U = R
n1

R
iso
(De
n1
)
r
iso
(De
n1
)
(r
n1
a
n1
) A
n
Da
n
DA
2
(De
n1
)
(R
n1
r
n1
)
r
iso
(R
n1
r
n1
)
= 0
(54)
Linearizing Eq. (51) for the NewtonRaphson method provides, for the kth iteration,
dR
k1
n1
=
U
k
oU
oR
n1
_ _
k
; (55)
where
oU
oR
n1
= I
DA
2
(De
n1
)
r
iso
(R
n1
r
n1
)
I
DA
2
(De
n1
)
r
2
iso
(R
n1
r
n1
)
o r
iso
oR
n1
: (56)
Here, I is the second-order identity tensor. After solving R
n+1
from Eq. (56), A
n+1
is ob-
tained from Eq. (52).
4. Model calibration
4.1. Characterization of material parameters of AZ31B magnesium alloy sheets
One-dimensional continuous in-plane tensioncompression and compressiontension
tests are performed in order to verify the implementation of formulations developed in
Table 1
Chemical compositions of Mg alloys
Al Cu Mn Zn Si Mg
AZ31B-O tempered 3.0 0.2 1.0 Balance
AZ31B 3.0 0.05 0.2 1.0 0.1 Balance
562 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
the previous sections. Two model materials are adopted to investigate the eect of temper-
ing on the cyclic behavior of magnesium alloys: O-tempered AZ31B magnesium alloy with
3.2 mm thickness and AZ31B magnesium alloy with 2 mm thickness. The experimental
data for the O-tempered AZ31B were reproduced from the previous work by Lou et al.
(2007) and as-received AZ31B alloy was newly adopted in the present work. The chemical
compositions of the two materials are listed in Table 1. The stressstrain responses of
O-tempered AZ31B in tensioncompressiontension (TCT) and compressiontension
(CT) are shown in Fig. 6a and b, respectively, while in Fig. 7a and b for AZ31B without
tempering. For the continuous tension/compression tests, the specially designed device
which prevents the sheet from being buckled during the compression was utilized. Note
that the stressstrain curves shown in Figs. 6 and 7 are corrected curves after considering
friction between sheets and clamping device and bi-axial eect from clamping force. For
more details on the experimental procedure, refer to Boger et al. (2005). From these g-
ures, three important unique features are observed, which should be properly considered
with the current constitutive models.
Strain
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
T-C-T
(Lou et al., 2007)
strain
-0.12 -0.08 -0.04 0.00 0.04 0.08 0.12
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
C-T
(Lou et al., 2007)
b
a
Fig. 6. The stressstrain responses of O-tempered AZ31B magnesium alloy sheets in (a) tensioncompression
tension (TCT) and (b) compressiontension (CT) (Lou et al., 2007).
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 563
strain
-0.08 -0.04 0.00 0.04
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B, 2mm thick
C-T
Strain
-0.10 -0.05 0.00 0.05 0.10 0.15
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B, 2mm thick
T-C-T
a
b
Fig. 7. The stressstrain responses of AZ31B Mg alloy sheets (no tempering, 2 mm thickness) in (a) tension
compressiontension (TCT) and (b) compressiontension (CT).
Table 2
Initial yield stresses in uni-axial tension and compression of Mg alloy (unit: MPa)
r
T
xx
r
C
xx
r
T
yy
r
C
yy
r
45T
xx
AZ31B-O tempered
a
192 110 164 104 180
AZ31B 220 120 250 140 210
a
Lou et al. (2007).
Table 3
Anisotropic parameters for the modied DruckerPrager yield surface
p q b2 b3 b4
AZ31B-O tempered 1.373 0.373 1.099 1.03 0.906
AZ31B 1.318 0.318 1.138 0.994 0.943
564 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
(1) Strong asymmetry in tensile and compressive yield stresses- the ratios of initial
yield stresses in tension and compression are 192 MPa: 110 MPa for O-tempered
AZ31B and 220 MPa:120 MPa for AZ31B.
(2) Unusual concave-up shape during compression and S-shape during tension fol-
lowing compression, which is explained by deformation twinning.
Normalized Yield Stress at TD
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0
N
o
r
m
a
l
i
z
e
d

Y
i
e
l
d

S
t
r
e
s
s

a
t

R
D
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
Calculated (no offset)
fs
Measured (no offset)
Calculated (0.2% offset)
Calculated (0.2% offset)
AZ31B-O tempered
Normalized Yield Stress at RD
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
N
o
r
m
a
l
i
z
e
d

Y
i
e
l
d

S
t
r
e
s
s

a
t

T
D
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
C
AZ31B
t=2mm
a
b
Calculated (0.2% ffset)
Measured (0.2% ffset)
Fig. 8. Initial yield locus of (a) AZ31B-O tempered (3.2 mm thick): Closed squares are experimentally measured
yield stress by no-oset and 0.2% oset method. The uni-axial tests data were reproduced from Lou et al. (2007)
and balanced bi-axial test data was reproduced from Jain and Agnew (2005). (b) AZ31B (no tempering, 2 mm
thick) with modied DruckerPrager model. Zero shear stress is assumed.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 565
(3) Signicant reduction in the size of elastic range during compression and compres-
sion following tension.
The yield parameters for the modied DruckerPrager yield surface are calculated
from the measured data and Eqs. (36)(40). Five initial yield stresses for the tension
and compression of dierent loading directions (Table 2) are utilized to calculate aniso-
tropic coecients (Table 3). Considering uni-axial tension tests as reference state, the
yield surfaces normalized by initial tensile yield stress are shown in Fig. 8. Zero shear
stress is assumed for both model materials. In case of O-tempered AZ31B sheet, both
criteria with zero and 0.2% oset methods in determining yield stress are shown in the
gure. For the yield surface of O-tempered AZ31B Mg sheet shown in Fig. 8a, the uni-
axial test data were reproduced from Lou et al. (2007), which are used for the calcu-
lation of yield surface, and balanced bi-axial yield stress is used from Jain and Agnew
(2005) for the verication. The gures show that the strong anisotropy and asymmetry
in the initial yield stress are well represented by the current modied yield surface for
both materials.
The hardening parameters dened in Section 3.2 are characterized from the measured
data by applying specic phenomenological forms for the Eqs. (26) and (27). Since the
bounding surface is arbitrarily dened, constant bounding surface with large enough ini-
tial size is assumed to simplify the characterization procedure. That is, dR = 0 in Eq. (25).
Therefore, the bounding surface does not expand or move during the deformation and the
gap distance is measured from the static bounding curve (with constant value) and loading
curve. Therefore,
R = R
+
: (57)
Here, for the model material AZ31B alloy sheets, constant value R
+
= 520 MPa is chosen.
By considering the common shape of bounding and loading surface and the asymmetry of
the loading surface, the constant bounding values in tension and compression are 520 MPa
and 300 MPa for O-tempered AZ31B and 520 MPa and 284 for AZ31B.
Acummulated strain
0.00 0.02 0.04 0.06 0.08 0.10


(
M
P
a
)
150
200
250
300
350
AZ31B O-tempered, measured
AZ31B O-tempered, fitted
AZ31B, measured
AZ31B, fitted
Fig. 9. Fitting curve of gap distance from the measured uni-axial tensile curves (closed circle: AZ31B-O
tempered, open square: AZ31B without tempering).
566 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
As for the loading surface, combined isotropickinematic hardening is assumed with
the gap distance for dierent deformation modes. The gap distance during the slip defor-
mation from undeformed state is calculated from the measured stressstrain curve and
bounding surface dened in Eq. (57). The gap function for slip mode is well dened with
the following function as used for the conventional alloy sheets such as aluminum alloy
(Lee et al., in press):

d = a
s
(

d
in
) b
s
(

d
in
) exp(c
s
(

d
in
)e
l
); (58)
where a
s
, b
s
, c
s
are material parameters which depend on initial gap distance and e
l
is the
plastic strain whose value is re-initialized for each reverse loading. From the experimental
data, the material parameters in Eq. (58) are obtained by considering gap distances be-
tween bounding and loading stresses. Due to the limited availability of the experimental
data, constant parameters are assumed and obtained by curve tting hardening curves
during the initial tension. The tted curve is shown in Fig. 9 for two materials and the

in
(MPa)
300 350 400 450 500 550
a
U
T

(
M
P
a
)
0
1000
2000
3000
4000
5000
6000
AZ31B
AZ31B-O tempered

in
(MPa)
300 350 400 450 500
b
U
T

(
M
P
a
)
0
20
40
60
80
100
120
140
AZ31B
AZ31B-O tempered
a
b
Fig. 10. Fitting parameters for r
t
: (a) a
UT
, (b) b
UT
, (c) a
T
, (d) b
T
with varying d
in
(closed circle: AZ31B-O
tempered, open square: AZ31B without tempered).
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 567
constant parameters are a
s
= 216.2 MPa, b
s
= 115.4 MPa, and c
s
= 16.2, for O-tempered
AZ31B and a
s
= 186.0 MPa, b
s
= 114.0 MPa, and c
s
= 15.4 for AZ31B.
In terms of hardening curve during the compression or compression following tension,
the sigmoid or S-shaped function is introduced to represent particular stressstrain
curve of magnesium alloys during deformation twinning. The two functions, r
t
and #
are characterized by tting the stressstrain curves of continuous TCT and CT with
respect to several dierent initial gap distances. For the function r
t
, the following linear
function with accumulated plastic strain is used.
r
t
= a
T or UT
(

d
in
) b
T or UT
(

d
in
)e
l
(59)
where a
T or UT
and b
T or UT
are material parameters which depend on the initial gap dis-
tance with load reversal. The subscripts T and UT denote the twinning and untwining
modes, respectively. Fig. 10 shows the linear dependence of the two parameters on the ini-
tial gap for the untwinning mode (Fig. 10a,b) and twining mode (Fig. 10c,d).

in
(MPa)
120 140 160 180 200 220
a
T

(
M
P
a
)
0
100
200
300
400
500
600
AZ31B
AZ31B-O tempered

in
(MPa)
100 120 140 160 180 200 220
b
T

(
M
P
a
)
0
50
100
150
200
AZ31B
AZ31B-O tempered
d
c
Fig. 10 (continued)
568 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
For the function #, the following sigmoid type curve tting is utilized:
# = y
0;T or UT
(

d
in
)
c
T or UT
(

d
in
)
1 exp
e
l
x
0;TorUT
(

d
in
)
d
T or UT
(

d
in
)
_ _
(60)
where y
0,T or UT
, c
T or UT
, x
0,T or UT
, and d
T or UT
are material parameters and obtained by
considering initial gap distance. The dependence of these four parameters on the initial
gap stress is assumed piecewise linear as shown in Fig. 11a for the untwining and
Fig. 11b for the twinning mode, respectively. Note that Eqs. (57)(60) are empirically ob-
tained forms which conform the cyclic behavior of the present two materials in room
temperature.
Other common features in the cyclic stressstrain curves of sheet alloys are early re-
yielding (or Bauschinger eect) and rapid work hardening rate (or transient behavior)
when the material undergoes change of loading path. The Bauschinger eect and transient
behavior could be modeled with combined isotropickinematic hardening laws (Chung
et al., 2005). Proper measurement of the size of yield surface for the isotropic hardening

in
(MPa)
350 400 450 500
y
0
,
U
T
,

c
U
T
,
d
U
T
,

x
0
,
U
T
0.00
0.05
0.10
0.20
0.40
0.60
0.80
c
UT
x
0,UT
y
0,UT
d
UT

in
(MPa)
120 140 160 180 200 220
y
0
,
T
,

c
T
,
d
T
,

x
0
,
T
0.0
0.1
0.2
1.0
2.0
3.0
c
T
x
0,T
y
0,T
d
T
a
b
Fig. 11. Material parameters of # (Eq. (60)) for: (a) Untwining, (b) Twinning with d
in
(closed circle: AZ31B-O
tempered, open square: AZ31B without tempering).
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 569
(MPa)
(MPa)
(MPa)
0 20 40 60 80 100 120

f

-

r
|
i
s
o
,
e
x
p

(
M
P
a
)

f

-

r
|
i
s
o
,
e
x
p

(
M
P
a
)

f

-

r
|
i
s
o
,
e
x
p

(
M
P
a
)
0
50
100
150
200

R
a
t
i
o
,

f

-

r
|
i
s
o
,
e
x
p
/

f

-

r
|
p
u
r
e

i
s
o

R
a
t
i
o
,

f

-

r
|
i
s
o
,
e
x
p
/

f

-

r
|
p
u
r
e

i
s
o

R
a
t
i
o
,

f

-

r
|
i
s
o
,
e
x
p
/

f

-

r
|
p
u
r
e

i
s
o
0.0
0.2
0.4
0.6
0.8
1.0
Ratio
Pure isotropic
Measured
avg: 0.87
avg: 0.88
0 20 40 60 80 100 120 140
-300
-200
-100
0
100
200
300
400
-20
-15
-10
-5
0
5
10
Ratio
Pure isotropic
Measured
avg: -6.98
avg: -5.96
0 20 40 60 80 100 120 140
0
100
200
300
400
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Ratio
Pure isotropic
Measured
avg: 0.84
avg: 0.81
a
b
c
Fig. 12. Measured size change and the ratio of isotropic change to the assumed pure isotropic case for various
pre-strains or dierent stress increments from initial yield stress: (a) during the slip mode, (b) during the twinning
deformation without assumption of instant shrinkage, and (c) during the twinning deformation with instant
shrinkage (closed circle: AZ31B-O tempered, open square: AZ31B without tempering).
570 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
ratio, which is often called as Bauschinger ratio (Boger, 2006) is needed. The size change of
yield surface during deformation is measured from the experimental data in Fig. 12. For
the assumed (pure) isotropic hardening case, the stress increment during deformation com-
pletely contributes to the size change of the yield surface. Therefore, the size change for the
pure isotropic D[r
f
r
r
[
pure iso
becomes
D[r
f
r
r
[
pureiso
= Dr(1 f
a
) (61)
where Dr is the stress increment and f
a
is the factor of asymmetry of the initial yield
surface. f
a
is assumed as
r
C
x
r
T
x
for tension and
r
T
x
r
C
x
for compression. For the symmetric
material with f
a
= 1, size change becomes 2Dr. Note that there is no size change for
assumed (pure) kinematic hardening. The size change during tensile deformation from
undeformed state is measured and compared with that of the assumed pure isotropic
hardening in Fig. 12a. For several dierent pre-strains or dierent stress increments
from initial yield stress, the following ratio of size change to the assumed (pure) iso-
tropic increment is calculated:
m
l
= D[r
f
r
r
[
iso;measured
=D[r
f
r
r
[
pure iso
(62)
The variations of the ratio during the slip deformation are very small for both mate-
rials and their average value are 0.87 and 0.88, respectively. Therefore, constant ratios
of isotropic hardening in Eq. (12) are adopted using the average values for the slip
mode.
In terms of the size change of yield surface during the twinning or compression from
initial material state, similar procedure is applied to calculate the assumed pure isotropic
change and size change in measured stressstrain curves of tension following compression
(CT curves in Figs. 6b and 7b) with several pre-strains. In case of O-tempered AZ31B, for
the pre-strain around 2%, the size of yield surface (or size of linear region) is approxi-
mately 100 MPa, while initial size of yield surface before compression (or initial size of
Fig. 13. Back stress of loading surface after instant shrinkage.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 571
True strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Compression-Tension
Pre-strain: -0.023
True strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Compression-Tension
Pre-strain: -0.045
True strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Compression-Tension
Pre-strain: -0.075
a
b
c
Fig. 14. Comparisons of calculated uni-axial compressiontension (CT) curves of AZ31B Mg alloy sheet with
measurements for various initial compressive strains: pre-strains are: (a) 0.023, (b) 0.045, and (c) 0.075; Lines and
squares are calculated results and measured data reproduced from Lou et al. (2007), respectively.
572 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
yield surface) is 302 MPa. Therefore, the conventional isotropic hardening model with
phenomenological yield surface cannot explain the size reduction or softening behavior.
As shown in Fig. 12b, the average ratios of size change to the assumed pure isotropic
increment are negative and the absolute value are decreasing for two materials. This means
that the yield surface shrinks abruptly just after compression and expands again as the
compressive deformation proceeds.
In order to consider the eect of softening during twinning mode, the initial shrinkage
of the yield surface is introduced in the current model. Here, for the AZ31B magnesium
alloy sheets, the shrinkage ratios are assumed as 0.325 (=99/302) for O-tempered sheet
and 0.54 (=185/340), respectively. With the new initial sizes of yield surfaces, the size
change and corresponding ratio are re-calculated in Fig. 12c. The gure shows that the
yield surface expands almost isotropically up to 5% of plastic strain and then isotropic
kinematically. The average ratios 0.84 and 0.81 which are very similar to that of slip modes
are adopted as ratios of isotropic hardening in Eq. (12). As for the size change of yield
True strain
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Tension-Compression-Tension
Pre-strain: 0.017

True strain
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Tension-Compression-Tension
Pre-strain: 0.042
a
b
Fig. 15. Comparisons of calculated uni-axial tensioncompressiontension (TCT) curves of AZ31B Mg alloy
sheet with measurements for various initial tensile strains: pre-strains are (a) 0.017, (b) 0.045, (c) 0.075, and (d)
0.165; Lines and squares are calculated results and measured data reproduced from Lou et al. (2007), respectively.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 573
surface during the untwining mode or tension following compression in Figs. 6 and 7, the
same shrinkage ratios and the ratios of isotropic hardening are assumed. The center of
yield surface or back stress needs to be determined once the instant softening (or shrink-
age) occurs. Since the current stress shares the same position on the loading surfaces
before and after instant softening (marked a in Fig. 13), the back stress of loading surface
can be calculated considering the stress on the bounding surface (marked A in Fig. 13) and
relation in Eq. (8):
a
/
= r
r
/
iso
R
iso
(R A) (63)
where a
/
and r
/
iso
are back-stress and size of the yield surface after instant softening,
respectively.
4.2. Correlation of uni-axial cyclic tests
The developed constitutive models based on two-surface hardening law and asym-
metric yield surface are implemented into a commercial nite element program ABA-
True strain
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Tension-Compression-Tension
Pre-strain: 0.075
True strain
-0.10 -0.05 0.00 0.05 0.10 0.15 0.20
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B-O tempered
Tension-Compression-Tension
Pre-strain: 0.165
c
d
Fig. 15 (continued)
574 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
QUS/Standard which allows the new constitutive models to be implemented by UMAT
subroutine. In terms of nite element simulations for the uni-axial tension or compres-
sion tests, a four-node shell element with reduced integration, S4R is utilized. To sim-
ulate a uniform stressstrain response which is consistent with experimental procedure,
proper boundary conditions were applied to the one shell element. Note that the cur-
rent analysis can be carried out without use of nite element method. However, in
order to validate the developed constitutive laws and their numerical implementation,
nite element procedure was adopted in this work. Fig. 14 shows the comparison of
the measured continuous uni-axial compressiontension (CT) tests shown in Fig. 6b
to the results calculated from the nite element simulations with developed models
for O-tempered AZ31B sheet. Three dierent pre-strains are applied before unloading;
0.023, 0.045, and 0.075. In general, the developed constitutive models are able to
reproduce the main features of the tension following compression curves. Especially,
True Strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B
Compression-Tension
Pre-strain: -0.02
True Strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B
Compression-Tension
Pre-strain: -0.04
a
b
Fig. 16. Comparisons of calculated uni-axial compressiontension (CT) curves of AZ31B Mg alloy sheet (no
tempering, 2 mm thick) with measurements for various initial compressive strains: pre-strains are (a) 2%, (b) 4%,
(c) 6%, and (d) 8%; Lines are calculated.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 575
the unusual concave-up shape of ow curves during the initial compression (or twin-
ning mode) and sigmoid shape (S-shape) during the tension following compression
are well reproduced for each stressstrain curve with various pre-strains. Also, the con-
stant parameter used for the instant shrinkage (or softening) of the initial yield surface
and constant ratio of isotropic hardening during the subsequent plastic deformation
can be well veried by the good agreement between the results of model and measure-
ments. The largest discrepancy is shown in the transient region of the reloading curve
as illustrated in Fig. 14c. The measurement curve shows smooth transient from elastic
to plastic, while almost linear in the simulated curve.
Similar comparisons are made for the continuous uni-axial tensioncompressionten-
sion (TCT) tests in Fig. 15 for O-tempered AZ31B sheet. The test involves all three
dierent deformation modes explained in the previous sections: slip mode during the
initial tension, twinning mode during the compression following tension, and untwining
mode during the tension following compression. Four dierent pre-strains before
unloading from the initial tension are chosen: 0.017, 0.045, 0.075, and 0.165. For all
four stressstrain curves, excellent agreements are shown with the developed constitu-
True Strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B
Compression-Tension
Pre-strain: -0.06
True Strain
-0.10 -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
AZ31B
Compression-Tension
Pre-strain: -0.08
c
d
Fig. 16 (continued)
576 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
tive models. It is shown that the constant averaged ratio of isotropic hardening well
represents the isotropic expansion of the yield surface during the slip mode. Also,
the rapid decrease and subsequent expansion of the loading surface during the twinning
(or untwining) deformation by introducing instant shrinkage ratio and the ratio of iso-
tropic hardening are well veried from good agreements in the compression curves and
sizes of the linear regions during the reloading. Note that these two parameters are
characterized from the compressiontension curves shown in Fig. 6b. The same proce-
dures have been applied to AZ31B sheet for the compressiontension and tensioncom-
pressiontension tests. Similarly, the current material models are able to reproduce the
experimentally observed behavior except for the transient regions as shown in Figs. 16
and 17. Note that the current model has limited validity only to the measured stress
strain curves with limited strain range. However, the present constitutive equations may
be eectively utilized to predict the forming and springback behavior in a typical
stamping process where the strain ranges are moderate.
True Strain
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
500
AZ31B
Tension-Compression-Tension
Pre-strain: 0.04
True Strain
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
500
AZ31B
Tension-Compression-Tension
Pre-strain: 0.06
a
b
Fig. 17. Comparisons of calculated uni-axial tensioncompressiontension (TCT) curves of AZ31B Mg alloy
sheet (no tempering, 2 mm thick) with measurements for various initial tensile strains: pre-strains are: (a) 4%,
(b) 6%, (c) 8%, and (d) 10%; Lines are calculated results.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 577
5. Conclusions
The constitutive modeling for the magnesium alloy sheets was developed to practically
represent their unusual mechanical properties. The developed constitutive equations
include the modied anisotropic yield function and advanced hardening model. First of
these is the modied yield criterion with pressure dependent term and can represent high
directional dierences in the initial yield stress (anisotropy) and also high asymmetry in
tension and compression. The unusual hardening behavior during the non-monotonous
deformation was well represented by the practical two-surface model. The summary of
the works done in the present paper is as follows:
1. A practical two-surface plasticity model developed for symmetric materials has been
further extended to represent hardening behavior of magnesium alloy sheets. The
two-surface model is based on classical Dalias/Popov and Krieg concepts and is able
to represent complex hardening eects for non-monotonous loading such as Bauschin-
ger eect and transient.



True Strain
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
500
AZ31B
Tension-Compression-Tension
Pre-strain: 0.08
True Strain
-0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12
S
t
r
e
s
s

(
M
P
a
)
-300
-200
-100
0
100
200
300
400
500
AZ31B
Tension-Compression-Tension
Pre-strain: 0.1
c
d
Fig. 17 (continued)
578 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
2. To represent anisotropy and asymmetry between the yield stresses in tension and com-
pression, the classical yield surface, DruckerPrager criterion, was modied by intro-
ducing the coecients of anisotropy in the plane stress condition. The yield surface
is a conical surface with additional hydrostatic stress term in the three-dimensional
space, thus can represent dierence in tension and compression in the plane stress state.
3. The developed theory was numerically formulated for the nite element analysis and
implemented into the commercial nite element program ABAQUS/Standard with user
material subroutine UMAT.
4. In order to characterize material parameters for the developed constitutive equations,
two AZ31B magnesium alloy sheets were utilized: one with O-tempered condition
and the other without tempering. Continuous in-plane uni-axial cyclic tests were
adopted for the characterization purpose. As reported previously, three deformation
modes are observed during the tests. Uni-axial tensile deformation is dominated by
the slip mode and the ow curve is normal concave-down shape. The deformations dur-
ing the in-plane compression and tension following compression are dominated by the
twinning and untwining modes which show unusual concave-up or S-shape stress
curves. Besides the dierent type of hardening behavior, the size of elastic region is also
unusual for each deformation mode. The size of linear region increases from the initial
state during the slip mode, while abrupt reduction of the linear region and subsequent
increase are observed for the twinning and untwining modes.
5. Five experimentally measured initial yield stresses were utilized for the anisotropic/
asymmetric yield surface: tensile yield stresses in the rolling and transverse directions,
compressive yield stresses in the rolling and transverse directions and tensile yield stress
in the 45.
6. Since the hardening behavior is updated every time when the reversal of loading direc-
tion occurs in the two-surface model, dierent hardening curves (or gap distance) are
used for each deformation mode. For the slip mode, normal exponential type gap func-
tion was used. On the other hand, for the twinning and untwining modes, sigmoid type
function which represents S-shape hardening was adopted.
7. To eectively consider the abrupt reduction of linear elastic region during the twinning
or untwining deformation, constant shrinkage ratio of the yield surface was introduced
and then the yield surface increases with constant ratio.
8. Calculated stressstrain curves for the uni-axial compressiontension (CT) tests and
tensioncompressiontension (TCT) tests with various pre-strains were compared
with measurements. In general, the model could reproduce the experimental behavior
with great accuracy. Both the unusual hardening curve during the twinning mode
and size change of yield surface were well predicted although small discrepancy in
the transient behavior was observed.
9. Finally, based on the promising results of the present constitutive modeling work, fur-
ther research on the prediction of real forming and springback behavior needs to be
explored as an application part of the current paper.
Acknowledgements
This work was supported by the sabbatical program of KNU, the Int. Joint R&D Pro-
gram by MCIE (10028109), by the SRC/ERC Program of MOST/KOSEF (R112005
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 579
065) in Korea and by the National Science Foundation (DMI0355429). MGL also appre-
ciates the partial support from the center for Advanced Materials Processing of the 21
st
Century Frontier R&D Program by MOST.
References
ABAQUS, 2006. Users Manual for Version 6.5.1. Hibbit, Karlson & Sorensen Inc.
Agnew, S.R., Duygulu, O., 2005. Plastic anisotropy and the role of non-basal slip in magnesium alloy AZ31B.
Int. J. Plasticity 21, 11611193.
Amstrong, P.J., Frederick, C.O., 1966. A Mathematical Representation of the Multiaxial Bauschinger Eect.
G.E.G.B. Report RD/B/N 731.
Ball, E.A., Prangnell, P.B., 1994. Tensile-compression yield asymmetries in high strength wrought magnesium
alloys. Scripta Metall. Mater. 31, 111116.
Barlat, F., Lege, D.J., Brem, J.C., 1991. A six-component yield function for anisotropic materials. Int. J.
Plasticity 7, 693712.
Barlat, F., Becker, R.C., Hayashida, Y., Maeda, Y., Yanagawa, M., Chung, K., Brem, J.C., Lege, D.J., Matsui,
K., Murtha, S.J., Hattori, S., 1997. Yielding description of solution strengthened aluminum alloys. Int. J.
Plasticity 13, 185.
Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Choi, S.H., Pourboghrat, F., Chu, E., Lege, D.J.,
2003. Plane stress yield function for aluminum alloy sheets. Int. J. Plasticity 19, 1297.
Bauschinger, J., 1886. On the change of the elastic limit and the strength of iron and steel, by drawing out, by
heating and cooling, and by repetition of loading. In: Minutes of Proceedings of the Institution of Civil
Engineers with Other Selected and Abstracted Papers LXXXVII, p. 463.
Bettles, C.J., Gibson, M.A., 2005. Current wrought magnesium alloys: strengths and weakness. JOM 57, 46.
Boger, R.K., 2006. Non-monotonic Strain Hardening and its Constitutive Representation. Ph.D. Thesis, Ohio
State University, USA.
Boger, R.K., Wagoner, R.H., Barlat, F., Lee, M.-G., Chung, K., 2005. Continuous, large strain, tension/
compression testing of sheet material. Int. J. Plasticity 21, 2319.
Bower, A.F., 1989. Cyclic hardening properties of hard-drawn copper and rail steel. J. Mech. Phys. Solids 37, 455.
Brown, D.W., Agnew, S.R., Bourke, M.A.M., Holden, T.M., Vogel, S.C., Tome, C.N., 2005. Internal strain and
texture evolution during deformation twinning in magnesium. Mater. Sci. Eng. A 399, 112.
Cazacu, O., Barlat, F., 2001. Generalization of Druckers yield criterion to orthotropy. Math. Mech. Solids 6,
613630.
Cazacu, O., Barlat, F., 2004. A criterion for description of anisotropy and yield dierential eects in pressure-
insensitive metal. Int. J. Plasticity 20, 20272045.
Chaboche, J.L., 1991. Time independent constitutive theories for cyclic plasticity. Int. J. Plasticity 2, 149.
Chun, B.K., Jinn, J.T., Lee, J.K., 2002a. Modeling the Bauschinger eect for sheet metals-part I: theory. Int. J.
Plasticity 18, 569.
Chun, B.K., Kim, H.Y., Lee, J.K., 2002b. Modeling the Bauschinger eect for sheet metals-part II: applications.
Int. J. Plasticity 18, 597.
Chung, K., Richmond, O., 1993. A deformation theory of plasticity based on minimum work paths. Int. J.
Plasticity 9, 907.
Chung, K., Lee, M.G., Kim, D., Kim, C., Wenner, M.L., Barlat, F., 2005. Spring-back evaluation of automotive
sheets based on isotropickinematic hardening laws and non-quadratic anisotropic yield functions, Part I:
theory and formulation. Int. J. Plasticity 21, 861.
Cubberly, W.H., Baker, H., Benjamin, D., Unterweiser, P.M., Kirkpatrick, C.W., Knoll, V., Nieman, K. (Eds.),
1979. Properties and selection: Nonferrous alloys and pure metals. In: Metals Handbook, ninth ed. American
Society for Metals.
Dafalias, Y.F., Popov, E.P., 1976. Plastic internal variables formalism of cyclic plasticity. J. Appl. Mech. ASME
98, 645.
Drucker, D.C., 1949. Relation of experiments to mathematical theories of plasticity. J. Appl. Mech. 16,
349357.
Duygulu, O., Agnew, S.R., 2003. The eect of temperature and strain rate on the tensile properties of texture
magnesium alloy AZ31B sheet. In: Kaplan, H. (Ed.), Magnesium Technology 2003. TMS, San Diego, CA,
USA, pp. 237242.
580 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582
Easton, M., Song, W., Abbott, T., 2006. A comparison of the deformation of magnesium alloys with aluminum
and steel in tension, bending, and buckling. Mater. Des. 27, 935946.
Geng, L., Wagoner, R.H., 2002. Role of plastic anisotropy and its evolution on springback. Int. J. Mech. Sci. 44
(1), 123.
Geng, L., Shen, Y., Wagoner, R.H., 2002. Anisotropic hardening equations derived from reverse-bend testing.
Int. J. Plasticity 18, 743.
Gradinger, R., Stolg, P., 2003. Magnesium wrought alloys for automotive applications. In: 2003 TMS Annual
Meeting: Magnesium Technology 2003. TMS, San Diego, CA, USA, 231.
Hashiguchi, K., 1981. Constitutive equations of elastoplastic materials with anisotropic hardening and elastic
plastic transition. J. Appl. Mech. ASME 48, 297.
Hashiguchi, K., 1988. A mathematical modication of two surface model formulation in plasticity. Int. J. Solids
Structures 24, 987.
Hashiguchi, K., 1997. The extended ow rule in plasticity. Int. J. Plasticity 13, 37.
Hill, R., 1948. A theory of the yielding and plastic ow of anisotropic metals. Proc. Roy. Soc. Lond. A 193, 281
297.
Hill, R., 1979. Theoretical plasticity of textured aggregates. Math. Proc. Cambridge Philos. Soc. 85, 179191.
Hodge, P., 1957. A new method of analyzing stresses and strains in work hardening plastic solids. J. Appl. Mech.
ASME 24, 482.
Hosford, W.F., 1966. Textured strengthening. Met. Eng. Q. 6, 1319.
Hosford, W.F., 1972. A generalized isotropic yield criterion. Trans. ASME J. Appl. Mech. 39, 607.
Hosford, W.F., 1993. The Mechanics of Crystals and Textured Polycrystals. Oxford University Press, New York.
Kalidindi, S.R., 1998. Incorporation of deformation twinning in crystal plasticity models. J. Mech. Phys. Solids
46, 267290.
Khan, A.S., Huang, S., 1995. Continuum Theory of Plasticity. John Wiley & Sons, Inc., NY, USA.
Khoei, A.R., Jamali, N., 2005. On the implementation of a multi-surface kinematic hardening plasticity and its
applications. Int. J. Plasticity 21, 1741.
Kim, J., Lee, W., Kim, D., Kong, J., Kim, C., Wenner, M.L., Chung, K., 2005. Eect of Hardening Laws and
Yield Function Types on Spring-Back Simulations of Dual-Phase Steel Automotive Sheets, Metal and
Materials (submitted).
Kim, J., Ryou, H., Kim, D.G., Kim,D., Chung, K., Hong, S-H., 2007a. Constitutive law of AZ31 Mg alloy sheet
and nite element simulation of three point bending. Int. J. Mech. Sci., submitted.
Kim, J.H., Lee, M.G., Ryou, H., Chung, K., Youn, JR, Kang, TJ., 2007b. Development of nonlinear constitutive
laws for anisotropic and asymmetric ber reinforced composites. Polymer Composites, in press.
Krieg, R.D., 1975. A practical two surface plasticity theory. J. Appl. Mech. ASME 42, 641.
Lebenson, R.A., Tome, C.N., 1993. A self-consistent anisotropic approach for the simulation of plastic
deformation and texture development in polycrystals-application to zirconium alloys. Acta Metall. 41, 2611
2624.
Lee, M.G., Kim, D., Kim, C., Wenner, M.L., Wagoner, R.H., Chung, K., 2005a. Spring-back evaluation of
automotive sheets based on isotropickinematic hardening laws and non-quadratic anisotropic yield
functions, Part II: characterization of material properties. Int. J. Plasticity 21, 883.
Lee, M.G., Kim, D., Kim, C., Wenner, M.L., Chung, K., 2005b. Spring-back evaluation of automotive sheets
based on isotropic-kinematic hardening laws and non-quadratic anisotropic yield functions, Part III:
applications. Int. J. Plasticity 21, 915.
Lee, M.G., Kim, D., Kim, C., Wenner, M.L., Wagoner, R.H., Chung, K., 2007. A practical two-surface plasticity
model and its application to springback prediction. Int. J. Plasticity 23, 11891212.
Li, M., 2006. Constitutive Modeling of Slip, Twinning, and Untwining in AZ31B Magnesium. Ph.D. Thesis, Ohio
State University, USA.
Lou, X.Y., 2005. Hardening Evolution of AZ31B Mg Sheet. MS Thesis, Ohio State University, USA.
Lou, X.Y., Li, Min., Boger, R.K., Agnew, S.R., Wagoner, R.H., 2007. Hardening evolution of AZ31B Mg sheet.
Int. J. Plasticity 23, 44.
McDowell, D.L., 1985. A two surface model for transient nonproportional cyclic plasticity: Part 1: development
of appropriate equations. J. Appl. Mech. ASME 52, 298.
Mordike, B.L., Ebert, T., 2001. Magnesium: propertiesapplicationspotential. Mater. Sci. Eng. A 302, 3745.
Mroz, Z., 1967. On the description of anisotropic work-hardening. J. Mech. Phys. Mech. 15, 163.
Mroz, Z., Niemunis, A., 1987. On the description of deformation anisotropy of materials. In: Boehler, J.P., Proc.
IUTAM Symp.: Plasticity and damage of anisotropic solids. Grenoble, Mech. Eng. Press.
M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582 581
Mroz, Z., Norris, V.A., Zienkiewicz, O.C., 1979. Application of an anisotropic hardening model in the analysis of
deformation of solids. Geotechnique 29, 1.
Ohno, N., Kachi, Y., 1986. A constitutive model of cyclic plasticity for nonlinear hardening materials. J. Appl.
Mech. 53, 395.
Ohno, N., Wang, J.D., 1993a. Nonlinear hardening rules with critical state of dynamic recovery. Part I:
formulation and basic features for ratcheting behavior. Int. J. Plasticity 9, 375.
Ohno, N., Wang, J.D., 1993b. Nonlinear hardening rules with critical state of dynamic recovery. Part II:
application to experiments of ratcheting behavior. Int. J. Plasticity 9, 391.
Prager, W., 1956. A new method of analyzing stresses and strains in work hardening. Plastic Solids. J. Appl.
Mech. ASME 23, 493.
Roberts, C.S., 1960. Magnesium and its Alloys. John Wiley, New York and London.
Staroselsky, A., Anand, L., 2003. A constitutive model for hcp materials deforming by slip and twinning:
application to magnesium alloy AZ31B. Int. J. Plasticity 19, 18431864.
Tresca, H., 1864. On the yield of solids at high pressure. Comptes Rendus Academie Science 59, 754.
von Mises, R.V., 1928. Mechanik der plastishen formanderung von kristallen. Zeitschrift fur Angewandte
Mathematik und Mechanik 8, 161185.
Yoo, M.H., 1981. Slip, twinning, and fracture in hexagonal close-packed metals. Metall. Mater. Trans. A 12A,
409418.
Yoon, J.W., Barlat, F., Chung, K., 1998. Inuence of initial back stress on the earing prediction of drawn cups for
planar anisotropic aluminum sheets. J. Mater. Proc. Tech., 8081.
Ziegler, H., 1959. A modication of Pragers hardening rule. Quart. Appl. Math. 17, 55.
582 M.-G. Lee et al. / International Journal of Plasticity 24 (2008) 545582

Você também pode gostar