Caz Acu 2006

International Journal of Plasticity 22 (2006) 11711194
www.elsevier.com/locate/ijplas
Orthotropic yield criterion for hexagonal

closed packed metals
a,*
Oana Cazacu , Brian Plunkett a, Frederic Barlat b
a
Department of Mechanical and Aerospace Engineering, University of Floridas Graduate
Engineering and Research Center, Shalimar, FL 32579-1163, USA
b
Materials Science Division, Alcoa Inc., Alcoa Technical Center, 100 Technical Drive, Alcoa Center,
PA 15069-0001, USA
Received 2 April 2005

Available online 11 October 2005
Abstract
In this paper, a macroscopic orthotropic yield criterion, which can describe both the aniso-
tropy of a material and the yielding asymmetry between tension and compression, is intro-
duced. The yield function is expressed in terms of the principal values of the stress deviator,
ensuring insensitivity to the hydrostatic pressure. Application of the proposed criterion to
magnesium and titanium alloy sheets shows that this model can capture very well the complex
behavior of these materials.
2005 Elsevier Ltd. All rights reserved.
Keywords: Plastic anisotropy; Twinning; Tensioncompression asymmetry; Hexagonal closed packed

metals
1. Introduction
Hexagonal closed packed (hcp) metals and alloys are important to nearly every
modern industry, ranging from biomedical applications, such as Ti implants and
NiTi superelastic catheter guides; to new Sn-based lead-free solders under explora-
*
Corresponding author. Tel.: +1 850 833 9350; fax: +1 850 833 9366.
E-mail address: [email protected] (O. Cazacu).
0749-6419/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijplas.2005.06.001
1172 O. Cazacu et al. / International Journal of Plasticity 22 (2006) 11711194
tion by the microelectronics industry; to Ti (or TiAl) alloy jet engine compressor
components. These advanced materials have excellent properties for their target
applications (e.g., Ti implants exhibit biocompatibility, light weight, high strength,
and low stiness); however, their use is still limited because of problems associated
with poor formability and consequently high manufacturing costs (see for example,
Avedesian and Baker, 1999; Takuda et al., 1999). Traditional explanations of these
phenomena relate to the fact that most cold-rolled hcp alloy sheets have basal or
nearly basal textures, i.e., the c-axis is aligned normal to the plane of the sheet. A
degree of spread from this ideal texture by up to 20 about the transverse direction
is observed for magnesium while for alpha titanium and zirconium alloys the spread
is up to 40 (see Hosford, 1966). Depending on the c/a ratio, either prism slip or basal
slip along the closed-packed basal directions a generally dominates the plastic re-
sponse. Slip on pyramidal planes along the a-directions is signicantly harder, and
even more dicult on the pyramidal planes along the c + a directions. However,
basal slip does not produce any elongation or shortening parallel to the c-axis. Thus,
only twinning or pyramidal slip can accommodate inelastic shape changes in the c-
direction. For most hcp metals, the most easily activated twinning mode is the tensile
twin f1 0 1 2gh1 0 1 1i, which is not activated by tension in the plane of the sheet. Be-
cause of the directionality of twinning, at low strain levels a very pronounced
strength dierential (SD) eect is observed: the compressive strengths are much low-
er than the tensile strengths (Hosford, 1993; Tome et al., 2001; etc.). At high strain
levels, when twinning has practically ceased, this asymmetry is markedly diminished.
Since the pyramidal slip and compression twinning are much harder than the pri-
mary deformation modes of basal slip and tension twinning, most hcp sheets display
a strong resistance to thinning. For example, for textured magnesium sheets values
of 24.5 of the ratio r of the width strain-to-thickness strain are commonly observed
(Agnew and Duygulu, 2005; Kaiser et al., 2003) while for titanium values as high as 9
have been reported (see Hosford, 1993).
A general framework for description of yielding anisotropy and its evolution with
accumulated deformation is oered by polycrystal viscoplasticity. Direct polycrystal
viscoplasticity implementation in which a representation of the texture is embedded
within every nite element and the stress and stiness needed are computed using the
equations for the single crystal response and the associated interaction law (linking
hypothesis) has the advantage that it follows the evolution of anisotropy due to tex-
ture changes. However, the correct modeling of the strong asymmetry between ten-
sion and compression due to deformation twinning remains a challenge. As discussed
by Van-Houtte (1978) and Tome et al. (1991), a major obstacle in extending the crys-
tal plasticity framework to include deformation twinning is the diculty in handling
the large number of orientations created by twinned regions. Although progress has
been made and models that track the evolution of the twinned regions in the grain
and account for predominant twin reorientation (e.g., Tome and Lebensohn,
2004) or intergranular mechanisms (e.g., Staroselsky and Anand, 2003) have been
proposed, the use of such models is still limited because the calculations are compu-
tationally very intensive even for problems that do not require a ne spatial resolu-
tion. Phenomenological descriptions of plastic anisotropy are convenient and time-
O. Cazacu et al. / International Journal of Plasticity 22 (2006) 11711194 1173
ecient for sheet forming process simulations. However, in contrast to the recent
progress in the formulation, numerical implementation, and validation of macro-
scopic plasticity models for materials with cubic structure (e.g., Barlat et al., 2003,
2004, 2005; Cazacu and Barlat, 2003; Bron and Besson, 2004, etc.), modeling of
hcp materials is less developed. Due to the lack of adequate macroscopic criteria
for hcp materials, hcp sheet forming FEM simulations are still performed using clas-
sic anisotropic formulations for cubic metals such as Hill (1948) (see for example,
Takuda et al., 1999; Kuwabara et al., 2001). There is a clear need to develop phys-
ically based macroscopic level anisotropic formulations that could describe both the
anisotropy and asymmetry in yielding (tension vs. compression) of hcp metals and that
can be easily implemented in FE element codes and thus applicable routinely to the
simulation of sheet forming processes.
To account for initial plastic anisotropy or to describe an average material re-
sponse over a certain deformation range, rigorous methods have been proposed in
the literature (e.g., the linear transformation approach proposed by Barlat et al.
(1991) and Karallis and Boyce (1993); the generalized invariants approach pro-
posed by Cazacu and Barlat (2001, 2003)). The major diculty encountered in for-
mulating analytic expressions for the yield functions of hcp metals is related to the
description of the SD eect (tension vs. compression asymmetry).
The purpose of this paper is to propose a general, physically based and numeri-
cally robust macroscopic formulation that describes with accuracy both the asymme-
try and anisotropy in yielding exhibited by hcp materials. The approach that will be
adopted is: (1) to develop macroscopic isotropic yield function that could describe
in a realistic manner non-symmetrical yielding eects (tension vs. compression)
based on single and polycrystalline viscoplasticity; (2) to extend the isotropic formu-
lation such as to account for orthotropy. The validity of the proposed model is dem-
onstrated through comparison between calculated and experimental yield loci for
sheets of textured polycrystalline binary MgTh and MgLi alloys (data after Kelley
and Hosford, 1968) and a Titanium (data after Lee and Backofen, 1966).
2. Proposed isotropic yield criterion for description of asymmetric yielding
If a material only deforms by a reversible shear mechanism such as slip, yielding

depends only on the magnitude of the resolved shear stress. Thus, the yield locus in
the deviatoric p plane (plane which passes through the origin and is perpendicular to
the hydrostatic axis) must have sixfold symmetry. The shape of the yield surface of
such an isotropic material is dictated by the requirements of: (a) pressure-indepen-
dence; (b) invariance with respect to any transformation belonging to the orthogonal
group; (c) equal yield stresses in tension and compression. The size of the yield locus
is determined quite eectively by observing when yielding occurs in a few tests, such
as a tension test or a pure shear test. Yielding of such materials is thus represented by
an even function of the principal values of the stress deviator, S.
Twinning and martensitic shear are directional deformation mechanisms, and if
they occur, yielding will depend on the sign of the stress (Hosford, 1993). Early
simulations results by Chin et al. (1969), who analyzed deformation by mixed slip
and twining in fcc crystals, predicted a yield stress in uniaxial tension 25% lower than
that in uniaxial compression. Hosford and Allen (1973) extended the calculations to
other types of loading. Based on the simulation results they concluded that yield loci
with a strong asymmetry between tension and compression should be expected in
any isotropic pressure insensitive material that deforms by twinning or directional
slip. Based on the shape of the yield locus obtained through polycrystalline simula-
tions by Hosford and Allen (1973), Cazacu and Barlat (2004) proposed an isotropic
yield criterion of the form
3
f J 2 2 cJ 3 s3Y ; 1
2 3
where J2 = trS /2 and J3 = trS /3 are the second and third invariants of the stress 3
P
3
deviator S, respectively (tr denotes the trace operator), i.e., trA Akk ; sY is the
k1
yield stress in pure shear and c is a material constant expressible solely in terms of rT
and rC the uniaxial yield stresses in tension and compression
p
3 3r3T r3C
c . 2
2r3T r3C
Note that for equal yield stresses in tension and compression c = 0, hence the pro-
posed criterion reduces to the von Mises yield criterion. For the yield function (1)
to be pconvex,
p the constant c is limited to a given numerical range:
c 2 3 3=2; 3 3=4.
For plane stress conditions, the yield locus is given by
3=2
1 2 c 3
r1 r1 r2 r22 2r1 2r32 3r1 r2 r1 r2 s3Y ; 3
3 27
where r1 and r2 denote the principal stresses. For any c 6 0, the yield function is
homogeneous of degree 3 in stresses and Eq. (3) represents a triangle with rounded
corners. As an example, in Fig. 1 are shown the plane stress yield loci (3) correspond-
ing to rT/rC = 2/3, 1 (von Mises), and 3/2, respectively.
In this paper, we introduce a new isotropic yield criterion of the form
jS 1 j kS 1 a jS 2 j kS 2 a jS 3 j kS 3 a F ; 4
where Si, i = 1, . . . ,3 are the principal values of the stress deviator. At dierence with
the yield criterion (3), the proposed yield function (4) is a homogeneous function of
degree a in stresses. The exponent a is considered to be a positive integer. Also, in (4)
k is a material constant, while F gives the size of the yield locus and may be consid-
ered either constant (rigid plastic response) or function of the accumulated plastic
strain (isotropic hardening). The physical signicance of the material parameter k
may be revealed from uniaxial tests. Indeed, according to the proposed criterion
(4), the ratio of tensile to compressive uniaxial yield stress is given by
( a 1 a )1a
2
rT 1 k 2 1 k
32 a 3 a 5a
rC 3
1 k 2 13 1 k
1.00
T /C
2/3
0.50 1
3/2
2/T
0.00
-0.50
-1.00
-1.00 -0.50 0.00 0.50 1.00
1/T
Fig. 1. Plane-stress yield loci according to Cazacu and Barlat (2004) criterion for dierent values of the
ratio rrCT between the yield stress in tension and compression, in comparison with the Von Mises yield locus
(r1 and r2 are the principal values of the Cauchy stress).
or

rT
1h r
k C 5b
rT
1h rC
with
2 a 31a

a
2 2 rrCT
rT 6 7
h 4 a 5. 5c
rC rT
2 rC 2
Hence, for a xed, the parameter k is expressible solely in terms of the ratio rrCT (see
(5b) and (5c)). Note that for any value of a P 1,a integer if k = 0, there is no dier-
ence between the response in tension and compression. In particular, for k = 0 and
a = 2, the proposed criterion reduces to the von Mises yield criterion. From (5b) fol-
lows that for a given exponent a, for the parameter k to be real,
1a rT a1
2a 6 62a . 6
rC
Specically:
1a
for 2 a 6 rrCT 6 1 ) 1 6 k 6 0,
a1
for 1 6 rrCT 6 2 a ) 0 6 k 6 1.
As an example, in Fig. 2 are shown the representation

p of the plane stress yield loci (4)
corresponding to a = 2 (xed) and rT =rC 2; 1:26; 1:13 and 1 (von Mises),
respectively (i.e., corresponding to k = 1; 0.4; 0.2; 0, respectively). Note that the
higher the ratio between the yield stress in tension and compression, the greater is
the departure from the von Mises ellipse; for the highest admissible value for k,
the yield function (4) represents a triangle with rounded corners.
Furthermore, since h1x hx
1
, it follows that krrCT krrCT (see Eqs. (5)). To
illustrate this property of the proposed yield function, in Fig. 3 are represented
the plane stress yield loci (4) corresponding to rT/rC = 1.13 (k = 0.2) and rT/
rC = 1/1.13 (k = 0.2). It is clearly seen that a change in the sign of k results in a
mirror image of the yield surface. The variation of rrCT with k is illustrated in Fig. 4
a1
for dierent values of the exponent a. If k = 1 then rrCT 2 a so for a = 1: rT = rC,
1a
while for a ! 1, rrCT ! 2: if k 1: rrCT 2 a , so for a = 1 there is no dierence be-
tween tension and compression, while if a ! 1, then rrCT ! 12. For any integer a P 1
and for 1 6 k 6 1, the yield function (4) is convex (for the proof, see Appendix A).
For combined tension and torsion conditions where the uniaxial tensile stress is
set equal to r, the shear stress is set equal to s, and all other stress components
are zero, the proposed yield criterion becomes
k=1 k = 0.2
k = 0.4 k = 0 (von Mises)
1.5
a=2
1
0.5
2/T
-0.5
-1
-1.5
-1.5 -1 -0.5 0 0.5 1 1.5
1/T
Fig. 2. Plane stress yield loci according to the proposed criterion (4) for dierent values of the ratio rT/rC
between the yield stress in tension and compression, in comparison with the Von Mises yield locus.
1.5
a=2
1
0.5
2/ T
-0.5
-1
k = 0.2
k = -0.2
-1.5
-1.5 -1 -0.5 0 0.5 1 1.5
1/T
Fig. 3. Plane stress yield loci (4) corresponding to rT/rC = 1.13(k = 0.2) and rT/rC = 1/1.13 (k = 0.2).
2.25
2 a=100
a=5
1.75
a=3
1.5 a=2
C
/
1.25
T

0.75
0.5
0.25
-1 -0.5 0 0.5 1
k
Fig. 4. The inuence of the value of the parameter k on the ratio rT/rC of the uniaxial yield stress in
tension and compression, for various values of the exponent a.
r!a r!a
r r2 r r2 r a
s2 1 ka s2 1 ka 1 ka F .
6 4 6 4 3
7
Fig. 5 shows the representation in the tensiontorsion plane (r/rT,s/rT) of the pro-
posed yield loci corresponding to a xed value of the parameter a (a = 2) and several
dierent values of k. Note the clear deviation from both Tresca and Mises criteria for
k dierent from zero. Fig. 6 shows the representations in the deviatoric p plane
(plane normal to the hydrostatic axis r1 = r2 = r3) of the proposed yield loci (4)
for various values of the coecient k between 0 and 1 and a = 2 (xed), along with
the von Mises and Tresca yield loci for comparison. As k increases, the ratio rT/rc is
increasing and the yield loci depart drastically from the circular Von Mises locus.
As already mentioned, the analytic expression of the proposed yield criterion (4)
was constructed based on the shape of the yield loci for randomly oriented cubic
crystals deforming solely by twinning reported by Hosford and Allen (1973). These
authors calculated these yield loci using an extension of the Bishop and Hill (1951)
model, thus, assuming that the plastic strain of all crystals within a polycrystal is
equal to the macroscopic strain. In view of comparison with the proposed analytic
criterion (4), we calculated yield loci for randomly oriented cubic and hcp polycrys-
tals using the one-site viscoplastic self-consistent polycrystal (VPSC) model of
0.7
k=0 (von Mises)
k=0.2
0.6
k=0.4
0.5
0.4
/
k=1
Tresca
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
/
Fig. 5. Projections in the tensiontorsion plane of the proposed yield loci (4) for various k-values and
a = 2 (xed), in comparison with Tresca and von Mises (k = 0, a = 2) loci.
3
k=0 (von Mises)
Tresca
k=0.2
k=0.4
k=1
2
1
Fig. 6. Projection in the deviatoric p plane of the yield loci (4) for a = 2 and various values of k in
comparison to Von Mises and Tresca loci.
Lebensohn and Tome (see Lebensohn and Tome, 1993) which assume a less rigid
interaction between each grain and its surroundings (i.e., each grain is treated as
an anisotropic, viscoplastic, ellipsoidal inclusion embedded in a uniform matrix
which has the average constitutive behavior of the polycrystal).
First, let compare the yield loci obtained using the proposed criterion (4) with the
yield loci for randomly oriented fee polycrystals deforming solely by f1 1 1gh1 1 2i
twinning calculated using the VPSC model. The proposed yield condition (4) in-
volves 2 parameters: the exponent a and the parameter k, which for a xed is express-
ible solely in terms of the rT/rc ratio (see Eq. (5)). The VPSC model predicts a ratio
of 0.83 between the yield stress in tension and compression (Hosford and Allen, 1973
reported a value of 0.78 for the same ratio). Assuming a = 2, we obtain k = 0.3098.
Fig. 7(a) and (b) shows the yield stresses (open circles) obtained using the VPSC
model and the projection of the yield locus predicted by the proposed criterion (4)
for k = 0.3098 (solid line) in the biaxial plane and in the p plane, respectively. It
is clearly seen that the proposed isotropic criterion describes very well the polycrys-
talline results. On the same gure are shown the comparison between the yield loci
obtained with the VPSC model for randomly oriented bcc polycrystals deforming so-
lely by f1 1 2gh1 1 1i twinning (solid circles) and the yield loci according to the pro-
posed criterion (4) with a = 2 and for k = 0.3098 (which correspond to a ratio
between the yield stress in tension and compression which is the reciprocal of the va-
lue corresponding to fcc polycrystals). Fig. 8(a) and (b) shows a comparison between
the yield loci obtained using the proposed criterion (for a = 3 and k = 0.0645) with
the yield loci for randomly oriented hcp Zr polycrystals deforming solely by tensile
1.5
zz
1
0.5
yy/T
-0.5
-1 xx yy
-1.5
-1.5 -1 -0.5 0 0.5 1 1.5
a xx/T b
Fig. 7. Comparison between the VPSC yield locus for randomly oriented fcc (open circles) and bcc (closed
circles) polycrystals deforming solely by twinning and the predictions of the proposed criterion (4): (a) in
the biaxial plane; (b) in the p-plane.
1.5
zz
1
0.5
yy/T
-0.5
-1
xx yy
-1.5
-1.5 -1 -0.5 0 0.5 1 1.5
a xx/ T b
Fig. 8. Comparison between the VPSC yield locus for randomly oriented hcp Zr polycrystals deforming
solely by twinning (open rectangles) and the predictions of the proposed criterion (4): (a) in the biaxial
plane; (b) in the p-plane.
twinning f1 0
1 2gh1 0
1 1i and compressive twinning f1 1 2 2gh1 1 2 3 i calculated using
the VPSC model. Again, the strength dierential eect is very well captured.
3. Extension of the proposed isotropic yield criterion to include anisotropy
To describe both the asymmetry between tension and compression and the anisot-
ropy observed in hcp metal sheets, we extend the proposed isotropic criterion (4) to
orthotropy.
For the description of incompressible plastic anisotropy, Cazacu and Barlat

(2001, 2003) introduced a general and rigorous method which is based on the theory
of representation of tensor functions (Wang, 1970; Liu, 1982). It consists in substi-
tuting in the expression of any given isotropic criterion, the second and third invar-
iants of the stress deviator with generalizations of these invariants compatible with
the symmetry group of the material considered. However, with this approach, con-
vexity is reinforced only numerically. For this reason, a particular case of this general
theory, which is based on applying a 4th-order linear transformation operator on the
Cauchy stress tensor, has received more attention (Sobotka, 1969; Barlat et al., 1991;
Karallis and Boyce, 1993; etc.). It is worth noting that by using the linear transfor-
mation approach, the convexity of the resulting anisotropic extension is automati-
cally satised (Rockafellar, 1972).
To extend to orthotropy the proposed criterion (4), we use a linear transformation
on the stress deviator S, i.e., in the expression of the isotropic criterion (4), the prin-
cipal values of the Cauchy stress deviator s1,s2,s3 are substituted by the principal val-
ues of the transformed tensor R dened as
R CS; 8
where C is a constant 4th-order tensor. Thus, the proposed orthotropic criterion is of
the form
jR1 j k R1 a jR2 j k R2 a jR3 j k R3 a F ; 9
where R1,R2,R3 are the principal values of R. The only restrictions imposed on the
tensor C are: (i) to satisfy the major and minor symmetries and (ii) to be invariant
with respect to the orthotropy group. Thus, for 3-D stress conditions the orthotropic
criterion involves 9 independent anisotropy coecients; it reduces to the isotropic
criterion (4) for C equal to the 4th-order identity tensor. It is worth noting that
although the transformed tensor is not deviatoric, the orthotropic criterion is insen-
sitive to hydrostatic pressure and thus the condition of plastic incompressibility is
satised (proof is given in Appendix B). For k 2 [1,1] and any integers a P 1, the
anisotropic yield function is convex in the variables R1,R2,R3 (principal transformed
stresses).
Let (x,y,z) be the reference frame associated with orthotropy. In the case of a
sheet, x, y, and z represent the rolling, transverse, and the normal directions. Rela-
tive to the orthotropy axes (x,y,z), the tensor C is represented by
2 3
C 11 C 12 C 13
6 C 12 C 22 C 23 7
6 7
6 7
6 C 13 C 23 C 33 7
C 6 6
7.
7 10
6 C 44 7
6 7
4 C 55 5
C 66
Since in the case of a thin sheet, the only non-zero stress components are the in-plane
stresses (rxx,ryy,rxy), the principal values of R are:

q
1 2
R1 Rxx Ryy Rxx Ryy 4R2xy ;
2

q
1 11
R2 Rxx Ryy Rxx Ryy 2 4R2xy ;
2
R3 Rzz ;
where

2 1 1 1 2 1
Rxx C 11 C 12 C 13 rxx C 11 C 12 C 13 ryy ;
3 3 3 3 3 3

2 1 1 1 2 1
Ryy C 12 C 22 C 23 rxx C 12 C 22 C 23 ryy ;
3 3 3 3 3 3

2 1 1 1 2 1
Rzz C 13 C 23 C 33 rxx C 13 C 23 C 33 ryy ;
3 3 3 3 3 3
Rxy C 66 rxy .
If rT0 and rC0 dene the yield stress in tension and compression along the rolling
direction x, according to the proposed orthotropic criterion (9) it follows that:
1a
F
rT0 ;
jU1 j kU1 a jU2 j kU2 a jU3 j kU3 a
1a
F
rC0 a a a ;
jU1 j kU1 jU2 j kU2 jU3 j kU3
where

2 1 1
U1 C 11 C 12 C 13 ;
3 3 3

2 1 1
U2 C 12 C 22 C 23 ; 12
3 3 3

2 1 1
U3 C 13 C 23 C 33 .
3 3 3
Similarly, if rT90 and rC90 are tensile and compressive yield stresses in the transverse
direction, y, then:
1a
F
rT90 a a a ;
jW1 j kW1 jW2 j kW2 jW3 j kW3
1a;
C F
r90
jW1 j kW1 a jW2 j kW2 a jW3 j kW3 a
where

1 2 1
W1 C 11 C 12 C 13 ;
3 3 3

1 2 1
W2 C 12 C 22 C 23 ; 13
3 3 3

1 2 1
W3 C 13 C 23 C 33 .
3 3 3
Furthermore, yielding under pure shear parallel to the orthotropy axes occur when
rxy is equal to
1a
0 F
s a a . 14
jC 66 j kC 66 jC 66 j kC 66
Yielding under equibiaxial tension occurs when rxx and ryy are both equal to
1a
T F
rb ; 15
jX1 j kX1 a jX2 j kX2 a jX3 j kX3 a
while yielding under equibiaxial compression occurs when rxx ryy rCb ,
1a
F
rCb a a a ; 16
jX1 j kX1 jX2 j kX2 jX3 j kX3
where

1 1 2
X1 C 11 C 12 C 13 ;
3 3 3

1 1 2
X2 C 12 C 22 C 23 ;
3 3 3

1 1 2
X3 C 13 C 23 C 33 .
3 3 3
Furthermore, we assume that the plastic potential coincides with the yield function.
Let denote by rh the Lankford coecients (width to thickness strain ratios) under
uniaxial tension in a direction at angle h with the rolling direction. According to
the proposed orthotropic criterion, it follows that:
a a
1 k Ua1 a1 a1
1 W1 1 k U2 W2 U3 W3
rT0 a a ;
1 k U1a1 W1 U1 1 k U2a1 W2 U3a1 W3 Ua2 Ua3
1 ka Wa1 a a1 a1
2 U2 1 k W1 U1 W3 U3
rT90 a a ;
1 k Wa1 a1 a1
2 U2 W2 1 k W1 U1 W3 U3 W1 W3
a a
a a
1 k Ua1 a1 a1
1 W1 1 k U2 W2 U3 W3
rC0 a a ;
1 k U1a1 W1 U1 1 k U2a1 W2 U3a1 W3 Ua2 Ua3
a a
1 k Wa1 a1 a1
2 U2 1 k W1 U1 W3 U3
rC90
1 ka Wa1 a a1 a1 a
2 U2 W2 1 k W1 U1 W3 U3 W1 W3
a
17
with U1 to U3 given by (12), W1 to W3 given by (13) and the superscripts T and C

designating tensile and compressive states, respectively.
In the following, the proposed anisotropic criterion will be applied to the descrip-
tion of the anisotropy and asymmetry of the yield loci of textured polycrystalline
magnesium and binary MgTh and MgLi alloys (data reported in Kelley and Hos-
ford, 1968) and a (hcp) titanium (data after Lee and Backofen, 1966).
4. Applications to magnesium alloys
Kelley and Hosford (1968) reported the results of an experimental investigation

into the anisotropy and asymmetry in yielding of textured polycrystalline binary
MgTh (0.5% Th) and MgLi (4% Li) alloys. The data consists of the results of uni-
axial compression tests in the rolling, transverse, and normal directions, respectively,
uniaxial tensile tests in the rolling and transverse directions, as well as plane strain
compression tests. Based on these data, the experimental yield loci corresponding
to several constant levels (1%, 5%, and 10%) of the largest principal strain were re-
ported (see Figs. 9 and 10 where experimental data are represented by symbols). Due
to the strong basal pole alignment in the thickness direction, f1 0 1 2g twinning is eas-
ily activated by compression perpendicular to this direction, but is not active in ten-
sion within the plane. The eect of f1 0 1 2g twinning is clearly evident in the low
compressive strengths at 1%. At 10% strain, the third quadrant strengths are compa-
300
200
100
10% 5% 1%
yy
-100
-200
-300
-300 -200 -100 0 100 200 300
xx
Fig. 9. Comparison between the plane stress yield loci (rxy = 0) for a Mg0.5% Th sheet predicted by the
proposed theory (solid lines) and experiments (symbols). Data after Kelley and Hosford (1968). Stresses in
MPa.
200
100
10% 5% 1%
yy
-100
-200
-200 -100 0 100 200

xx
Fig. 10. Comparison between the plane stress yield loci (rxy = 0) for a Mg4% Li sheet predicted by the
proposed theory (solid lines) and experiments (symbols). Data after Kelley and Hosford (1968). Stresses in
MPa.
rable to the rst quadrant owing to the reorientations caused by f1 0 1 2g twinning

which act to hinder further deformation processes. Fig. 9 shows the section of the
theoretical yield loci (Eq. (9)) with r3 = 0 for MgTh together with the experimental
data reported in Kelley and Hosford (1968). The anisotropy coecients involved in
the expression of the theoretical yield loci for biaxial stress states as well as the con-
stant k were determined by least square t using Eqs. (12), (13), and (15)(17) and
the data corresponding to the given strain level. The obtained values of these param-
eters corresponding to the 1%, 5%, and 10% surfaces are given in Table 1. No data
were available for determination of the shear coecients, C44, C55, and C66. Note
that the proposed theory reproduces very well the observed asymmetry in yielding.
The experimental yield loci for the MgLi alloy sheets are similar in shape to
those for the MgTh alloy, but with much reduced yield stresses due to the occur-
rence of prism slip and to the weaker crystallographic texture. The eect of
f1 0
1 2g twinning is evident in the low compressive strengths at 1% and 5% strains.
Fig. 10 shows the theoretical yield loci for MgLi along with the data reported by
Table 1
MgTh coecients
k C12 C13 C22 C23 C33
1% 0.3539 0.4802 0.2592 0.9517 0.2071 0.4654
5% 0.2763 0.3750 0.0858 0.9894 0.0659 0.1238
10% 0.0598 0.6336 0.2332 1.4018 0.5614 0.7484
Table 2
MgLi coecients
k C12 C13 C22 C23 C33
1% 0.2026 0.5871 0.6975 0.9783 0.2840 0.1497
5% 0.2982 0.6103 0.8056 1.0940 0.5745 0.1764
10% 0.1763 0.5324 0.8602 1.0437 0.8404 0.2946
Kelley and Hosford (1968). The coecients involved in the expressions of the theo-
retical yield loci are given in Table 2.
5. Applications to titanium 4A11/4 O2
In the following, we apply the proposed orthotropic criterion to the description of

the anisotropy and tensioncompression asymmetry of 4A11/4 O2 textured a (hcp)
titanium alloy (data after Lee and Backofen, 1966). True stressstrain curves were
reported for dierent loading paths: uniaxial tension in the x-direction (rolling direc-
tion); uniaxial compression in the z-direction (through-thickness compression), and
plane strain compression in the z and y (transverse) directions. The material has
nearly ideal basal texture with a deviation of about 25 from the sheet normal toward
the transverse direction. Based on these data, the experimental yield loci correspond-
ing to several constant levels of the largest principal strain were reported (see Fig. 11,
1500
1000
500
4% 1%
yy
0
0.2%
-500
-1000
-1500
-1500 -1000 -500 0 500 1000 1500
xx
Fig. 11. Comparison between the plane stress yield loci (rxy = 0) for a 4A11/4 O2 sheet predicted by the
proposed theory (solid lines) and experiments (symbols). Data after Lee and Backofen (1966). Stresses in
MPa.
experimental data are represented by symbols). Due to the strong basal pole align-
ment in the direction of the normal to the sheet, f1 0 1 2g twinning was activated
by compression perpendicular to this direction, but is no twinning was revealed in
tension testing within the plane (see Lee and Backofen, 1966). The eect of
f1 0
1 2g twinning is clearly evident in the low compressive strengths in the rolling
and transverse directions.
Fig. 11 also shows the theoretical yield loci along with the experimental data. The
coecients involved in the expressions of the theoretical yield loci are given in Table
3. Note the ability of the proposed criterion to correctly describe the asymmetry in
yielding of 4Al1/4 O2.
Table 3
4A11/4 O2 coecients
k C12 C13 C22 C23 C33
0.2% 0.1556 0.2285 0.0374 1.2967 0.2439 0.3422
1% 0.1868 0.0431 0.3369 0.9562 0.3139 1.0861
4% 0.2577 0.2178 0.3635 1.0422 0.3754 0.8825
Note: For all levels of eective plastic strain a = 2 and C11 = 1.0. The coecients C44, C55, C66 were not
determined.
1500
1000
500
yy
4% 1%
500
0.2%
1000 0.2% data

1% data
4% data
1500
1500 1000 500 0 500 1000 1500

xx
Fig. 12. Comparison between the plane stress yield loci (rxy = 0) for a 4A11/4 O2 sheet predicted by
Hosfords (1966) modied Hill criterion (solid lines) and experiments (symbols). Data after Lee and
Backofen (1966) (stresses in MPa).
In order to account for the eccentricity of the yield surfaces of titanium and its
alloys, Hosford (1966) proposed a modication of the Hill criterion to include terms
linear in stress
Arx Bry B Arz F ry rz 2 Grz rx 2 H rx ry 2 1. 18
Hosfords (1966) yield function given by Eq. (18) was applied to the same 4A11/
4 O2 textured a (hcp) titanium alloy (see Fig. 12). Comparison between theoretical
and experimental yield loci show that the proposed criterion (9) describes with great-
er accuracy the behavior near the biaxial tension state.
6. Summary and discussion
In this paper, a yield function describing both the tension/compression asymmetry

and the anisotropic behavior of hcp metals and alloys in the full stress space was pro-
posed. Anisotropy was introduced through a linear transformation on the deviatoric
stress tensor. This yield function is convex, thus numerically robust and easily appli-
cable to 3-D FE simulations of forming processes. For isotropic conditions, it reduces
to a criterion with threefold symmetry which is expressed in terms of the principal val-
ues of the stress deviator and involves two parameters: parameter a, which gives the
degree of homogeneity of the yield function and parameter k, which is expressible in
terms of the ratio between the tensile and compressive yield strengths. If a is odd, for
the yield function to be convex, the constant k must belong to a given numerical
range: [1,1], while when a is even, the surface is convex irrespective of the values
of k. This isotropic criterion reduces to von Mises when k = 0 and a = 2. It was shown
that the isotropic criterion describes polycrystal yielding well.
For full 3-D stress conditions, the proposed orthotropic yield function involves 11
parameters, including 9 anisotropy coecients along with k and a. The experimental
characterization needed to obtain the material coecients for this new yield function
is relatively routine. It was shown that the proposed yield function appears to be
suitable for the description of the strong asymmetry and anisotropy observed in tex-
tured binary MgTh and MgLi alloy sheets (data after Kelley and Hosford, 1968)
and for 4A11/4 O2 titanium sheet (data after Lee and Backofen, 1966). Based on
these results, it is believed that a very good description of the behavior for large
strains can be obtained by using the proposed yield function for describing initial
yielding and a hardening law that closely models the evolution of texture.
Acknowledgments
The authors thank Dr. Ricardo Lebensohn of Los Alamos National Laboratories
for providing the VPSC code and for many helpful comments. Partial nancial
support for this work from NSF through Grant DMII-0322730 is gratefully
acknowledged.
Appendix A. Proof of the convexity of the proposed isotropic yield function
The proposed isotropic yield condition is

a a a
f S 1 ; S 2 ; S 3 jS 1 j kS 1 jS 2 j kS 2 jS 3 j kS 3 ; A:1
where Si are the principal values of the stress deviator S. We shall prove that for
k 2 [1,1] and any integer a P 1, this yield function is convex.
For the yield function to be convex, its Hessian matrix ought to be positive semi-
denite.
Let H the Hessian matrix, i.e.,
o2 f
H ij ; A:2
ori orj
where i,j = 1 . . . 3 and ri are the principal stresses. Isotropy dictates threefold symme-
try of the yield surface, thus, it is sucient to prove its convexity for stress states cor-
responding to r1 P r2 P r3.
If r1 P r2 P r3, it follows that:
r
J2
S 1 2 cosa1 ;
3

r
2p J2
S 2 2 cos a1 ; A:3
3 3

r
2p J2
S 3 2 cos a1 ;
3 3
where a1 is the angle satisfying 0 6 3a1 6 p and whose cosine is given by:
3
cos3a1 J23 J32 2 , where J2 and J3 are the second and third invariants of the stress
deviator (see Malvern, 1969).
For 0 6 a1 < p/6: S1 > 0, S2 < 0, S3 < 0, and:
aa 1
H 11 f41 ka S 1a2 1 ka 1a S 2a2 S a2
3 g;
9
aa 1
H 22 f1 ka S 1a2 1 ka 1a 4S 2a2 S a2
3 g;
9
aa 1
H 33 f1 ka S 1a2 1 ka 1a S 2a2 4S a2
3 g;
9 A:4
aa 1
H 12 f21 ka S a2
1 1 k a
1 a
2S a2
2 S a2
3 g;
9
aa 1 a a a
H 13 f21 k S a21 1 k 1 S a2 2 2S 3a2 g;
9
aa 1 a a a
H 23 f1 k S 1a2 1 k 1 2S 2a2 2S 3a2 g.
9
P3
Note that H ij 0, for any i = 1 . . . 3. Thus, the determinant of H is zero and its
j1
principal values are k1, k2, and k3 = 0. Furthermore:
6aa 1 a a a
trH k1 k2 1 k S a2
1 1 k 1 S a2 2 S a2
3 ;
9
a2 a 1 h
2
tr2 H k1 k2 1 k2a S 21 a2 1 k2a S 22 a2
9
a a2 a2 a a2 a2
1 k 2 S 1 S 2 31 k 2 S 1 S 3
i
31 k2a S 2 a2 S 3 a2 .
Since S1 > 0, S2 < 0, S3 < 0, it follows that for k 2 (1,1) and any integer a P 1:
tr(H) = k1 + k2 P 0 and tr2(H) = k1k2 P 0, i.e., the Hessian is always positive
semi-denite.
For p/6 < a1 < p/3:
aa 1 a a a a2
H 11 f1 k 4S 1a2 S a2
2 1 k 1 S 3 g;
9
aa 1 a a a a2
H 22 f1 k S 1a2 4S a2
2 1 k 1 S 3 g;
9
aa 1 a a a a2
H 33 f1 k S 1a2 S a2
2 41 k 1 S 3 g;
9 A:5
aa 1 a a2 a2 a a a2
H 12 f21 k S 1 S 2 1 k 1 S 3 g;
9
aa 1 a a a
H 13 f1 k 2S 1a2 S 2a2 21 k 1 S a2
3 g;
9
aa 1 a a a a2
H 23 f1 k S 1a2 2S a2
2 21 k 1 S 3 g.
9
P3
It follows that H ij 0, for any i = 1, . . . ,3. Thus, the determinant of H is zero and
j1
6aa 1 a a a a2
trH 1 k S 1a2 S a2
2 1 k 1 S 3
9
a2 a 1 h
2
a a2 2a a2
tr2 H k1 k2 1 k 2 S 22 1 k S 22
9
2a a2 a2 2a a2 a2
1 k S 1 S 2 31 k S 2 S 3
i
2a a2 a2
31 k S 1 S 3 .
Since S1 > 0, S2 > 0, S3 < 0, it follows that for k 2 (1,1) and any integer a P 1:
tr(H) P 0 and tr2(H) P 0.
Thus, for k 2 [1,1] and any integer a P 1 the yield function is convex.
Appendix B. Proof of the insensitivity of the proposed orthotropic yield function to

hydrostatic pressure
The proposed orthotropic yield condition is

a a a
gR1 ; R2 ; R3 jR1 j kR1 jR2 j kR2 jR3 j kR3 ; B:1
where Ri are the principal values of the transformed stress tensor R dened as
R CS;
where C is a constant 4th-order tensor and S is the deviator of the Cauchy stress ten-
sor. Let (x, y, z) be the reference frame associated with orthotropy. Thus,
og og oRm oRij
; i; j; m 1 . . . 3; B:2
op oRm oRij op
where p = rmm denotes the mean stress and the convention of summation of repeated
indices is adopted.
oR
We shall prove that opij 0, hence og
op
0, i.e., the condition of plastic incompress-
ibility is satised.
Indeed, the transformed stress tensor R can be expressed as:
R CS CTr; B:3
where T denotes the 4th-order deviatoric projection that transforms a 2nd-order ten-
sor in its deviator. Thus,
oRij
Lijkl dkl Lijkk ; i; j; k 1; . . . ; 3; B:4
op
where L = CT is the 4th-order orthotropic tensor that relates the transformed tensor
R to the Cauchy stress r.
Relative to (x,y,z), the tensor C is represented by
2 3
C 11 C 12 C 13
6C 7
6 12 C 22 C 23 7
6 7
6 C 13 C 23 C 33 7
C6 6
7;
7 B:5
6 C 44 7
6 7
4 C 55 5
C 66
while T is given by
2 3
2 1 1
6 1 2 1 7
6 7
6 7
16 1 1 2 7
T 6 7.
366 3 7
7
6 7
4 3 5
3
In (B.5) we used the simplied contracted indices convention of Voigt
not not not
(C 1111 C 11 ; C 1122 C 12 ; C 13 C 1133 , etc.). It follows that the non-zero components
of the 4th-order tensor L are:
L11 2C 11 C 12 C 13 ;
L12 C 11 2C 12 C 13 ;
L13 C 11 C 12 2C 13 ;
L21 2C 12 C 22 C 23 ;
L22 C 12 2C 22 C 23 ;
L31 C 32 2C 31 C 33 ;
L32 C 31 2C 32 C 33 ;
L33 C 31 2C 33 C 32 .
Hence, we obtain
8
>
> L11 L12 L13 0;
<
L21 L22 L23 0; B:6
>
>
:
L31 L32 L33 0.
Remarks:
1. Although R is not deviatoric, the three restrictions on the components of L ensure

oRij
that op
0 for any i,j = 1, . . . ,3.
2. Lijkl 6 Lklij. Hence, it can be concluded that the transformation considered is
more general than the transformation adopted by Karallis and Boyce (1993)
in which major symmetries of the linear operator are reinforced despite
the fact that such conditions are not necessary for ensuring plastic
incompressibility.
3. To capture anisotropy, Barlat et al. (2005) have proposed the following transfor-
mation on the stress deviator:
~
R CS; B:7
where the orthotropic tensor Ce is given by

2 3
0 Ce 12 Ce 13
6 7
6 Ce 21 0 Ce 23 7
6 7
6 Ce 31 C
e 32 0 7
Ce 6 6
7
7.
6 e 44
C 7
6 7
6 e 55 7
4 C 5
e 66
C
Note that the transformation proposed by Barlat et al. (2005) is equivalent to
the transformation proposed in this paper. Indeed, using Eqs. (B.3) and (B.7)
we obtain:
e 12 C 11 C 12 ;
C
e 13 C 11 C 13 ;
C
e 21 C 22 C 12 ;
C
B:8
e 23 C 22 C 23 ;
C
e 31 C 33 C 13 ;
C
e 32 C 33 C 23 .
C
References
Avedesian, M.M., Baker, H. (Eds.), 1999. Magnesium and Magnesium Alloys. ASM Speciality
Handbook. ASM International, Metal Parks, OH.
Agnew, S.R., Duygulu, O., 2005. Plastic anisotropy and the role of non-basal slip in magnesium alloy
AZ31B. Int. J. Plasticity 21, 11611193.
Barlat, F., Lege, D.J., Brem, J.C., 1991. A six-component yield function for anisotropic materials. Int. J.
Plasticity 7, 693712.
Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Lege, D.J., Pourboghrat, F., Choi, S.-H., Chu,
E., 2003. Plane stress yield function for aluminum alloy sheet Part I: Theory. Int. J. Plasticity 19,
12971319.
Barlat, F., Cazacu, O., Zyczkowski, M., Banabic, D., Yoon, J.W., 2004. Yield Surface Plasticity and
Anisotropy. In: Continuum Scale Simulation of Engineering Materials. Wiley-VCH, Weinheim, pp.
145178.
Barlat, F., Aretz, H., Yoon, J.W., Karabin, M.E., Brem, J.C., Dick, R.E., 2005. Linear transformation-
based anisotropic yield functions. Int. J. Plasticity 21, 10091039.
Bishop, J.W.F., Hill, R., 1951. A theoretical deviation of the plastic properties of a polycrystalline face-
centered metal. Phil. Mag. Ser. 42, 12981307.
Bron, F., Besson, J., 2004. A yield function for anisotropic materials application to aluminum alloys. Int.
J. Plasticity 20, 937963.
Cazacu, O., Barlat, F., 2001. Generalization of Druckers yield criterion to orthotropy. Math. Mech.
Solids 6, 613630.
Cazacu, O., Barlat, F., 2003. Application of representation theory to describe yielding of anisotropic
aluminum alloys. Int. J. Eng. Sci. 41, 13671385.
Cazacu, O., Barlat, F., 2004. A criterion for description of anisotropy and yield dierential eects in
pressure-insensitive metals. Int. J. Plasticity 20, 20272045.
Chin, G.Y., Mammel, W.L., Dolan, M.T., 1969. Taylor analysis for {1 1 1} 1 1 2 twinning and {1 1 1}
1 1 0 slip under conditions of axisymmetric ow. Trans. TMS-AIME 245, 383388.
Hill, R., 1948. A theory of the yielding and plastic ow of anisotropic metals. Proc. Roy. Soc. London
A193, 281297.
Hosford, W.F., 1966. Texture strengthening. Met. Eng. Quarterly 6, 1319.
Hosford, W.F., Allen, T.J., 1973. Twining and directional slip as a cause for strength dierential eect.
Met. Trans. 4, 14241425.
Hosford, W.F., 1993. The Mechanics of Crystals and Textured Polycrystals. Oxford University Press, New
York.
Kaiser, F., Bohlen, J., Deitmar, K., Kainer, K.U., 2003. Anisotropic properties of magnesium sheet AZ31.
Mat. Sci. Forum 419422, 315320.
Karallis, A.P., Boyce, M.C., 1993. A general anisotropic yield criterion using bounds and a
transformation weighting tensor. J. Mech. Phys. Solids 41, 18591886.
Kelley, E.W., Hosford, W.F., 1968. Deformation characteristics of textured magnesium. Trans. TMS-
AIME 242, 654661.
Kuwabara, T., Katami, C., Kikuchi, M., Shindo, T., Ohwue, T., 2001. Cup drawing of pure titanium
sheet-nite element analysis and experimental validation. In: Proceedings of the Seventh International
Conference on Numerical Methods in Industrial Forming Processes, Toyohashi, Japan, 1820 June, p.
781.
Lebensohn, R.A., Tome, C.N., 1993. A self-consistent anisotropic approach for the simulation of plastic
deformation and texture development of poly crystals: application to zirconium alloys. Acta Metall. et
Mater. 41, 26112624.
Lee, D., Backofen, W.A., 1966. An experimental determination of the yield locus for titanium and
titanium-alloy sheet. TMS-AIME 236, 10771084.
Liu, S.I., 1982. On representations of anisotropic invariants. Int. J. Eng. Sci. 20, 10991109.
Malvern, L.E., 1969. Introduction to the Mechanics of a Continuous Medium. Prentice-Hall Inc.,
Englewood Clis, NJ.
Rockafellar, R.T., 1972. Convex Analysis. Princeton University Press, Princeton, NY.
Sobotka, Z., 1969. Theorie des plastischen Fliessens von anisotropen Korpern. Zeit. Angew. Math. Mech.
49, 2532.
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.
Takuda, H., Yoshii, T., Hatta, N., 1999. Finite-element analysis of the formability of a magnesium-based
alloy AZ31 sheet. J. Mater. Process. Techn. 8990, 135140.
Tome, C.N., Lebensohn, R.A., Kocks, U.F., 1991. A model for texture development dominated by
deformation twinning: application to zirconium alloys. Acta Mater. 39, 26672680.
Tome, C.N., Maudlin, P.J., Lebensohn, R.A., Kaschner, G.C., 2001. Mechanical response of zirconium-I.
Derivation of a polycrystal constitutive law and nite element analysis. Acta Mater. 49, 30853096.
Tome, C.N., Lebensohn, R.A., 2004. Self consistent homogenization methods for texture and anisotropy.
In: Continuum Scale Simulation of Engineering Materials. Wiley-VCH, Weinheim, pp. 473497.
Van-Houtte, P., 1978. Simulation of the rolling and shear texture of brass by the Taylor theory adapted
for mechanical twinning. Acta Metall. 26, 591604.
Wang, C.C., 1970. A new representation theorem for isotropic functions, Part I and II. Arch. Rat. Mech.
An. 36, 166223.

Caz Acu 2006

Uploaded by

Copyright:

Available Formats

Caz Acu 2006

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Caz Acu 2006

Uploaded by

Copyright:

Available Formats

International Journal of Plasticity 22 (2006) 11711194

Orthotropic yield criterion for hexagonal

Received 2 April 2005

Keywords: Plastic anisotropy; Twinning; Tensioncompression asymmetry; Hexagonal closed packed

2. Proposed isotropic yield criterion for description of asymmetric yielding

If a material only deforms by a reversible shear mechanism such as slip, yielding

As an example, in Fig. 2 are shown the representation

3. Extension of the proposed isotropic yield criterion to include anisotropy

For the description of incompressible plastic anisotropy, Cazacu and Barlat

with U1 to U3 given by (12), W1 to W3 given by (13) and the superscripts T and C

4. Applications to magnesium alloys

Kelley and Hosford (1968) reported the results of an experimental investigation

-200 -100 0 100 200

rable to the rst quadrant owing to the reorientations caused by f1 0 1 2g twinning

5. Applications to titanium 4A11/4 O2

In the following, we apply the proposed orthotropic criterion to the description of

1000 0.2% data

6. Summary and discussion

In this paper, a yield function describing both the tension/compression asymmetry

Appendix A. Proof of the convexity of the proposed isotropic yield function

The proposed isotropic yield condition is

Appendix B. Proof of the insensitivity of the proposed orthotropic yield function to

The proposed orthotropic yield condition is

1. Although R is not deviatoric, the three restrictions on the components of L ensure

where the orthotropic tensor Ce is given by

You might also like