Anisotropic and Asymmetrical Yielding and Its distorted-IJP2016 PDF

International Journal of Plasticity 82 (2016) 127e158
Contents lists available at ScienceDirect
International Journal of Plasticity

journal homepage: www.elsevier.com/locate/ijplas
Anisotropic and asymmetrical yielding and its distorted

evolution: Modeling and applications
H. Li a, *, X. Hu a, b, H. Yang a, **, L. Li a
a
State Key Laboratory of Solidication Processing, School of Materials Science & Engineering, Northwestern Polytechnical University,
Xi'an, 710072, China
b
Automobile Steel Research Institute, Research Institute of Baosteel, Shanghai, 201900, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 6 October 2015
Received in revised form 1 February 2016
Available online 19 March 2016
Characterizing the anisotropy/asymmetry-induced distortional yielding and subsequent

evolution is still a challenge for potential usages of hard-to-deform materials. From
perspective of multiple mechanisms, two types of yield functions are classied, viz., the
principal shear stress-based models (SSM) and the stress invariants-based models (SIM);
then a unied continuum-based discontinuous (CBD) framework is constructed, in which
SSM and SIM are introduced to capture the distorted shape of the yielding, and an interpolation approach is adopted to smoothly present the nonlinear evolution of the distorted
plasticity in the full stress space. Taking the CPB06 (Cazacu et al., 2006) and Yoon's criteria
(Yoon et al., 2014) as typical SSM and SIM, the CBD framework is implemented in the
explicit 3D-FE platform for practical usages by combining implicit algorithm and interpolation approach, and the Nelder-Mead (N-M) method and the genetic algorithm (GA)
approach are evaluated for calibrating of CBD related to convergence, overlapping and
accuracy. The evaluation proves that the GA-based method is suitable for CBD, and the SIM
seems to be feasible for embedding into the CBD framework because of its solid physical
basis and numerical robustness. Taking high strength titanium alloy tube (HSTT) as a case,
the distorted plasticity evolution of the HSTT with six typical initial textures are characterized, then the correlations among initial textures, distorted behaviors and inhomogeneous deformation are quantitatively established to improve the multi-defect constrained
formability in uniaxial tension/compression and mandrel bending.
2016 Elsevier Ltd. All rights reserved.
Keywords:
A. Yield condition
B. Anisotropic material
B. Constitutive behaviour
C. Finite elements
C. Numerical algorithms
1. Introduction
The urgent needs for lightweight and high-performance components in many industries require the precision forming of
hard-to-deform materials with complex structures. The precision forming specic to these types of components depends on
accurate and efcient modeling of their plastic behaviors under complex loading conditions. While coordinated by multiple
mechanisms such as twinning and the Non-Schmid effect (Patra et al., 2014; Kabirian et al., 2015; Tuninetti et al., 2015), many
hard-to-deform materials, not only HCP structured polycrystalline aggregates but also some BCC or even FCC structured ones,
tend to present pronounced anisotropy/asymmetry behaviors. In particular, the microstructure variation during the
* Corresponding author. Tel./fax: 86 29 88495632.

** Corresponding author. Tel./fax: 86 29 88495632.
E-mail addresses: liheng@nwpu.edu.cn (H. Li), yanghe@nwpu.edu.cn (H. Yang).
http://dx.doi.org/10.1016/j.ijplas.2016.03.002
0749-6419/ 2016 Elsevier Ltd. All rights reserved.
128
H. Li et al. / International Journal of Plasticity 82 (2016) 127e158
successive deformation induces the distorted evolution of yield loci in sizes and shapes. Additionally, the forming of these
components generally involves complex loadings such as tension, compression, torsion, internal pressure or their combinations. The inherent unique behaviors and external boundary conditions easily induce inhomogeneous deformation and
further result in the dramatic evolution of strong texture reorientation and much distorted plasticity, which may affect the
formability of these materials. Thus, to excavate the forming potentials of these high strength and lightweight materials, it is
imperative and fundamental to construct suitable constitutive models to describe the distortional yielding and subsequent
evolution and then to evaluate their applicability in practical processes (Gawad et al., 2015).
Based on understanding of the unique plastic behaviors such as anisotropy, to advance the constitutive modeling of newly
developed materials is the ultimate frontiers (Chaboche, 2008; Banabic, 2010; Horstemeyer and Bammann, 2010; Lee and
Barlat, 2014; Chang and Kochmann, 2015; Smith et al., 2015). The multi-scale modeling blueprint prevails for thoroughly
characterizing the materials' constitutive features at the atomistic scope, meso scope and continuum scale (McDowell, 2010;
Zhang et al., 2014). Atomistic-scale modeling is used to reveal underlying mechanisms, such as non-Schmid effects (Bassani
and Racherla, 2011). Mesoscopic modeling, seen in the Taylor-Bishop-Hill polycrystal model and the Visco-plastic selfconsistent (VPSC) crystal plasticity model, is used to relate the macroscale plastic deformation to the mesoscale microstructures (Agnew and Duygulu, 2005; Graff et al., 2007; Choi et al., 2009; Kabirian et al., 2015; Patra et al., 2014; Cyr et al.,
2015), and both anisotropy and asymmetry can be predicted. However, the intensive computation costs strongly limit the
applications of these approaches in practice. Thus, developing the constitutive formulations at the macroscopic level is the
preferable way to achieve accurate and efcient simulation of complex forming processes (Lee and Barlat, 2014).
At the macro scale, regarding the yield criteria, ow rules and hardening laws, many continuum-based constitutive models
have been proposed and numerically implemented into FE platforms (Banabic, 2010; Xiao et al., 2012; Lee and Barlat, 2014).
To characterize the texture-induced anisotropy, as shown in Fig. 1, many anisotropic models have been proposed. To cover the
abnormal anisotropy of aluminum alloys, several anisotropic yield functions have been developed, extended and applied;
the typical ones include Karallis-Boyce model, YLD91, YLD96, YLD2000-2d, YLD2004-18p, Banabic model, the homogeneous
polynomials (Karallis and Boyce, 1993; Barlat et al., 1991, 1997, 2003, 2005; Banabic, 2010; Soare et al., 2008; Bron and
Besson, 2004; Iadicola et al., 2008). In addition to the anisotropy, as shown in Fig. 1, asymmetry yield is observed for HCP
structured materials and even BCC ones. The root cause is still under intensive exploration, but includes possibilities such as
porosity deformation, the polar nature of twinning and non-Schmid law (Cazacu and Stewart, 2009; Bassani and Racherla,
2011; Mohr et al., 2013). Several efforts have been undertaken to describe the yield asymmetry aside from plastic anisotropy (Cazacu and Barlat, 2004; Cazacu et al., 2006, 2010; Plunkett et al., 2008; Cazacu and Stewart, 2009; Ghaffari et al., 2014;
Tuninetti et al., 2015).
The above studies focus on describing the initial anisotropy or asymmetry behaviors for certain materials, and the evolution of the yield surface is largely described using combinations of isotropic and kinematic hardening laws (Wegener and
Schlegel, 1996; Lee et al., 2008; Choi and Pan, 2009). Due to the interaction of multiple deformation mechanisms associated
with complex microstructures, the distorted yielding and nonlinear hardening in full stress states during deformation have
been frequently observed (Barlat et al., 2005; Choi et al., 2009; Khan et al., 2009). Within the Mises criterion framework,
Franois (2001) introduced a distorted stress to replace the usual stress deviator to obtain the egg-shaped yield surface for
an aluminum alloy used for both proportional and non-proportional tension-torsion loading paths. Taking combined
isotropic, kinematic and distortional hardening into account, Shutov and Ihlemann (2012) proposed a rheological model to
describe the distortion of the yield surface for an annealed aluminum alloy. By introducing three material parameters, a
modied Franois model (2001) based on egg-shaped subsequent yield surfaces has been developed to describe the change
in the shape of the yield surface of the 1100 Al (Yue et al., 2014). Until now, modeling distorted plasticity and its evolution in
full stress space still remains a challenge for practical metal forming.
This study focuses on accurately and efciently modeling the distorted plasticity and its evolution of hard-to-deform
materials for practical usage. First, we conduct a critical review of the methodologies for developing macroscopic
Fig. 1. Distorted behaviors induced by anisotropy and asymmetry.
129
constitutive models. Second, a unied continuum-based discontinuous (CBD) framework is constructed to characterize the
initial distorted yielding and subsequent evolution in a full stress space. Third, the above discontinuous models are
numerically implemented in a commercial 3D-FE platform, and the calibration methods for the CBD are discussed. Next,
several CBD models are evaluated regarding capturing capabilities on anisotropy/asymmetry coupled distorted behaviors.
Then, because titanium tubes currently present the most potential for bleeding transformation in many industries, such as
aerospace, nuclear, chemical and healthcare (Yang et al., 2012; Banerjee and Williams, 2013), taking high-strength titanium
alloy tubes (HSTT) of Ti-3Al-2.5V (SAE, 2010) as the case, the evaluation and application of the above CBD constitutive models
are conducted, and new transferable knowledge about the correlations among initial textures, distorted behaviors and
inhomogeneous deformation is quantitatively claried during several practical processes such as uniaxial tension/
compression and mandrel bending.
2. Critical review of developing continuum-based constitutive models
From viewpoint of underlying physical mechanisms, the most typical constitutive models are revisited to nd a methodology for constructing a continuum-based model with sound physical meaning and numerical robustness to describe
distorted plasticity and its evolution in hard-to-deform materials for practical usage.
2.1. Physical deconstruction of the yield functions
The yield criterion for any form of the equations can be physically decomposed. The shear stress is the basic element for all
yield criteria. This behavior is based on the observation that plastic strain occurs by crystallographic gliding under shear stress
when the maximum shear stress reaches a critical value. The earliest proposed Tresca criterion (also called the maximum
shear stress criterion) was established according to the above theory. By adding another two principal shear stresses, the
Mises criterion was constructed to extend the single shear stress-based Tresca model to the full shear stress-based model for
smooth and convex description of the yield surface. By introducing the anisotropy parameters, the most widely used
anisotropic yield criterion, Hill'48 quadratic yield function (Hill, 1948), was constructed to describe the anisotropy of the
material. It is worth noting that the Mises and Hill'48 criteria can be reformulated as the form of the second invariant of the
stress tensor J2. From the mechanism of Schmid glide, the above full shear stress-based model (SSM) is equivalent to the J2based model. In light of the physical meanings of principal shear stresses and the stress invariants, most of the advanced yield
functions have been proposed considering various micro mechanisms-induced behaviors. The developed yield criteria can be
called the SSM or the stress invariants-based criterion (SIM).
2.1.1. Shear stress-based criterion (SSM)
By replacing the xed exponent 2 with a variable exponent a, Hershey (1954) introduced a non-quadratic formulation, in which the principal shear stresses are the major elements. Hosford (1972) extended Hershey's model
allowing for a continuous transition between Mises and Tresca formulations. Barlat et al. (1991) rewrote the Hosford
criterion in a form containing the deviator principal stresses. Karallis and Boyce (1993) further generalized the
Hosford criterion.
By introducing the anisotropy constants, the above isotropy criteria were further developed to describe the anisotropy. To
describe the anomalous behavior of aluminum alloys, the Hill'48 criterion was extended to a non-quadratic function (Hill,
1979). Hosford (1979) generalized his own isotropy criterion to anisotropy yielding. Barlat et al. (1997) proposed a more
general anisotropy expression of the yield function introduced by himself in 1991. The YLD2004-18p was constructed with
the capability to predict the occurrence of six and eight ears in cup drawing processes (Barlat et al., 2005).
In addition to the above anisotropy behaviors, when describing the asymmetry in HCP alloys, Cazacu and Barlat (2004)
proposed an isotropic yield function, as shown in Eq. (1), by introducing an asymmetry factor k. Next, via the linear
transformation of the stress tensor, as shown in Eq. (2), this isotropic criterion was extended to an anisotropic formulation
CPB06, which can describe both the anisotropy and the asymmetry (Cazacu et al., 2006). By adding another linear
transformation, the CPB06ex2 yield criterion (Plunkett et al., 2008) was constructed for a more accurate description of the
plastic ow and the anisotropy during both tension and compression.
f s jjS1 j kS1 ja jjS2 j kS2 ja jjS3 j kS3 ja taY 0
(1)
where S1,S2,S3 are the principal values of the stress deviator, a is a positive integer, a1.
X
Xa X
Xa X
Xa

f s k$
k$
k$
taY 0
1
where L is a constant fourth-order tensor, and
P P P
P
L[s].
1, 2, 3 are the principal values of.
(2)
130
2.1.2. Stress invariants-based criterion (SIM)

C By adding the third invariants of the deviatoric stress tensor, Drucker (1949) proposed an isotropy criterion to represent
the experimental data located between the Tresca and Mises yield surfaces. By introducing the rst invariant of the
Cauchy stress, the Drucker-Prager yield criterion (1952) was developed to represent the plastic deformation of soils,
polymers, foams and other pressure-dependent materials. Based on Hill'48 (1948) criterion and the Drucker-Prager
model (1952), Liu et al. (1997) proposed an asymmetric yield function for plastically orthotropic materials.
C Based on the theory of the representation of tensor functions, Cazacu et al. (2001) developed a method for the
generalization of the invariants of the stress deviators J2 and J3, and proposed an extension of the Drucker isotropic
yield criterion to orthotropy by replacing J2 and J3 with J20 and J30 .
C To capture the asymmetry specic to alloys with HCP structures, Cazacu and Barlat (2004) proposed an isotropic yield
function in the form of Eq. (3). Similarly, using the linear transformation approach and replacing the Cauchy stress with
the L[s], the above isotropic and asymmetry criterion was extended to an anisotropic formulation, as shown in Eq.
(4) (Nixon et al., 2010). In the I1-J2-J3 framework, the most widely used Gurson-Tvergaard-Needleman porous plasticity
was extended to include the effects of hydrostatic stress and the third invariant of stress on the matrix materials (Gao
et al., 2011). Independently, using J2 and J3, Khan and Yu (2012) proposed a yield criterion with product formulation as
shown in Eq. (5) to describe both the anisotropy and asymmetry of Ti-6Al-4V metals, and J3 was implicitly included by
the Lode parameter x. Considering the full stress invariants, I1, J2 and J3, and assuming a linear dependence of yielding
on the rst invariant, as shown in Eq. (6), Yoon et al. (2014) proposed an orthotropic yield model to describe the
anisotropy and asymmetry of pressure sensitive metals.
f s J2 s3=2 c$J3 s t3Y 0
(3)
where
c is the asymmetry coefcient and expressed in terms of uniaxial yield stresses intension sT and compression sC as
p
c 3 3s3T s3C =2s3T s3C .

h i3=2
f s J2
c$J3 t3Y 0
(4)
f f s$gs 1
(5)
where f(s) refers to Hill'48 criterion, g(s) ec(x
2
f s bI1 4 J2 A
!!3=2
J3
1)
, x cos3q 27J3 =23J2 3=2 .
!31=3
tY 0
5
B
(6)
where J2 A is the second stress invariant of the transformed stress tensor of A, J3 B is the third stress invariant of
another transformed stress tensor of B; A LA s, B LB s, LA and LB are two distinct fourth-order linear transformation tensors (Barlat et al., 1991).
2.2. Explicit and implicit modeling evolution of plasticity

Given the yield criteria, accurately and efciently modeling the evolution of the yield loci is another imperative issue that
depends on the ow rules and hardening laws.
2.2.1. Flow rules
To describe the relationship between the applied stress and the plastic strain increment, the concept of plastic potential was proposed. If the plastic potential is different from the yield function, g s f, this ow rule is called nonassociated ow rule (non-AFR). Otherwise, when g f, this ow rule is called associated ow rule (AFR, also called
normality rule). Whether to use the AFR or non-AFR depends on the distinguishable physical meanings of the consequent
yield functions and plastic potential. In crystalline plasticity, the yield function depends on the resolved shear stress on
several intersecting slip planes, while the ow potential depends on which slip plane activates. In this case, the derivative
of the consequent yield functions with respect to the stress determines the value of the plastic strain increment, while the
derivative of the plastic potential determines the ow direction, which is the direction of the plastic strain increment.
Thus, according to the different physical meanings, the selection of ow rules relies on whether the stress states have the
same effects on hardening and plastic ow.
For most metallic materials, the AFR has been conrmed to be accurate when establishing a constitutive relationship.
However, non-AFR must be used for pressure sensitive materials (Spitzig et al., 1975; Stoughton and Yoon, 2004) because of
the different effects of pressure stress on the yielding behaviors and the ow direction. The most general way to construct the
131
plastic potential is to follow the form of the corresponding consequent yielding functions; the different material constants
should be calibrated by considering the experimentally obtained plastic ow characteristics as comprehensively as possible.
However, the preferable way to describe the plastic ow is to directly use the non-AFR.
2.2.2. Hardening laws
Compared with the initial yielding formulation, the subsequent yield locus may evolve with an obvious change in sizes,
shapes or even locations along the straining (Ishikawa, 1997). Eq. (7) explicitly presents the general consequent yield
equations, in which the evolution models of the internal variables, such as aij, Aij and R, should be constructed from the
measurements for the descriptions of the expansion, translation and distortion of the yield loci.

F sij ; aij ; Aij ; R 0
(7)
where aij is the back stress related to the changes in the loading paths, Aij is the coefcient matrix describing the anisotropy or
asymmetry behaviors, and R is the hardening stress.
In most cases, the isotropic hardening models, such as the Hollomon and Swift power laws, J-C model, KHL model (Khan
et al., 2009) and physical model (Haddadi et al., 2006), are used to describe the proportional expansion of the yield surface in
which aij equals zero, Aij is constant and R is generally related to the effective plastic strain, temperature and strain rate.
By introducing the back stress, the kinematic hardening is used to describe the special phenomena upon reversal loading
such as the Bauschinger effect, smooth elastoplastic transient behavior, permanent softening and stagnation behavior
(Bruschi et al., 2014). The kinematic hardening model describes the yield surface translating by using aij in the stress space
without changing the form or size, in which Aij and R are constant. Several kinematic hardening models have been developed
such as the linear kinematic hardening laws (Prager, 1956; Ziegler, 1959) and nonlinear kinematic hardening laws (Armstrong
and Frederick, 1966; Chaboche, 1986; Yoshida and Uemori, 2002; Lee et al., 2008; Choi and Pan, 2009; Xiao et al., 2012; Mohr
et al., 2013). Because reverse loading-related phenomena usually occur during slight plastic deformations, the kinematic
hardening models were used for modeling forming processes with small plastic strains.
The other non-isotropic hardening rules, such as the distorted hardening models, in which Aij is a nonlinear function of
effective plastic strain, are used to represent the irregular uniform evolution of the yield surfaces because of variations in
microstructure, such as dislocation and texture (Abedrabbo et al., 2006; Gao et al., 2011). Yeganeh (2007) simultaneously
considered the deformation-induced anisotropy, kinematic and isotropic hardening to develop a constitutive model incorporating the yield surface distortion. Treating the coefcient of the J3 term as a function of accumulated equivalent plastic
strain, the distortion hardening was introduced in the asymmetry yielding function (Zhai et al., 2014). Regarding the three
different deformation modes and considering the evolution of the anisotropy coefcients and asymmetry parameters with
the locally accumulated plastic strains, the consequent yielding loci were established based on the CPB06ex2 yield function
(Muhammad et al., 2015).
It is noted that, for some hard-to-deform materials, such as titanium and magnesium alloys, the multiple mechanisms of
slipping and twinning cause more severe non-uniform evolution of the yield surface shape with plastic deformation. This
makes the curve tting unsuitable or even unable to obtain the coefcient matrix in the explicit hardening law as in Eq. (7).
Plunkett et al. (2006) uses the CPB06 yield function and interpolation technique to capture the anisotropy evolution as a result
of the evolving textures in high-purity zirconium metals. The nonlinear hardening laws are implicitly included by the
different yield loci.
Fig. 2. Unied CBD constitutive framework for describing the evolution of plasticity.
132
Fig. 3. Schematics of the SIM in different coordinates reecting multiple physical mechanisms.
3. Unied continuum-based discontinuous (CBD) framework

Based on the above brief reference on the development of methodologies for continuum-based constitutive models, it
becomes possible to collect ideas for constructing a feasible constitutive framework to model distorted plasticity and the
evolution of hard-to-deform materials in the full stress space for practical usage.
As shown in Fig. 2, considering the difculty that the explicit hardening law has in capturing the irregular evolution of
yield loci, especially at large deformation, a unied CBD framework is constructed to discontinuously describe the distorted
plasticity and its evolution in the full stress space. The distorted shapes of the yielding at different incremental deformation
stages can be described by the SSM, the SIM or the combined SSM SIM; an interpolation approach is adopted to smoothly
present the nonlinear evolution of distorted plasticity in the full stress space based on the established individual yield loci.
The newly developed CPB06 model (Cazacu et al., 2006) and Yoon's model (Yoon et al., 2014) can be considered as two typical
full SSM and full SIM that describe the anisotropy/asymmetry-induced distorted yield loci under different strains. The nonAFR is used, and the hardening laws are implicitly included in the CBD constitutive framework. The major features of the CBD
constitutive model are described below.
As shown in Fig. 3, from the perspective of the physical meaning, J2 can be used to describe the reversible shear mechanism
that obeys Schmid law, J3 can be used to characterize the irreversible (directional) deformation modes, such as twinning and
non-Schmid effects, and I1 has the capability to reect the effects of the pressure stress. While, principal shear stress cannot
represent the irreversible mechanisms to reect the asymmetry. As mentioned in Section 2, the effects of principal shear
stress on yielding are equivalent to those of J2. Thus, compared with the SSM, the SIM presents a more sound physical
background as mentioned in Section 2.
For SIM, within the I1-J2-J3 plasticity model framework, through two different fourth-order linear transformation tensors
LA and LB to the stress invariants, any anisotropy and asymmetry induced distortion plasticity may be conveniently
achieved without considering the convex any further as shown in Fig. 3. For cold rolled sheet metals, the pressure
insensitivity can be assumed for yielding and hardening. Thus, in Eq. (6), the material constant b is set to zero, and this SIM
can be reduced to the following form:
!!3=2
f s
J2
!
B
J3
t3Y 0
(8)
where the associated linear transformations on the stress for a 3D case are
a3 a2 =3
6 a3 =3
6
6 a2 =3
LA 6
6
0
6
4
0
0
a3 =3
a2 =3
0
a1 =3
a1 a3 =3
0
a1 =3
a1 a2 =3 0
0
0
a4
0
0
0
0
0
0
0
0
0
0
a5
0
3
0
07
7
07
7
07
7
05
a6
2
6
6
6
LB 6
6
6
4
b3 b2 =3
b3 =3
b2 =3
0
0
0
b3 =3
b2 =3
b1 b3 =3
b1 =3
b1 =3
b1 b2 =3
0
0
0
0
0
0
0
0
0
b4
0
0
0
0
0
0
b5
0
133
3
0
07
7
07
7
07
7
05
b6
For SSM as in Eq. (2), aside from the linear transformation L in Eq. (9), the asymmetry behaviors may be described by
adding an asymmetry factor k or even more linear transformations such as CPB06exn. When dealing with CPB06 model, to
satisfy the convex, a 1 and 1 k 1, and for the asymmetry behaviors, the related material factors should meet special
constraints as shown in Eq. (10). This formula causes more difculty in the numerical implementation such as the derivation and calibration of material parameters.
L CT
2
C11
6 C12
6
6 C13
C6
6 0
6
4 0
0
(9)
C12
C22
C23
0
0
0
C13
C23
C33
0
0
0
0
0
0
C44
0
0
0
0
0
0
C55
0
2
3
2=3 1=3 1=3 0 0
0
6 1=3 2=3 1=3 0 0
0 7
6
7
6
0 7
7 T 6 1=3 1=3 2=3 0 0
6 0
0
0
1 0
0 7
6
7
4 0
0
0
0 1
0 5
0
0
0
0 0
C66
sT
jF1 j kF1 a jF2 j kF2 a jF3 j kF3 a
sC
jF1 j kF1 a jF2 j kF2 a jF3 j kF3 a
3
0
07
7
07
7
07
7
05
1
1=a
(10)
where F1, F2 and F3 can be found in Appendix A.

This CBD framework has high exibility in selecting the yield functions. As shown in Fig. 2, for this method, only limited
individual yield loci at different intervals of straining need to be established to describe the irregular evolution of the
subsequent yielding. The sole SSM or SIM can be reduced and replaced by a mixed pattern of the different models, such as
Hill'48 Yoon and Hill'48 CPB06exn. At the small strain stages, the Hill'48 yield equations can be constructed to describe
the evolution of anisotropy, and the SIM or SSM can be used to capture the distortional hardening in latter deformation
stages. Additionally, in light of the correlation between dominant deformation mechanisms and yielding behaviors, the
above model is conveniently reduced to only anisotropy or asymmetry forms if necessary.
Although several experimental procedures are available for testing materials under different stress states, it is not always
possible to probe them all. As mentioned in the Introduction, if the experimental data are not available for a given strain
path, the VPSC crystal plasticity framework can be used to replace the missing experimental data for material calibration of
several yield loci (Gawad et al., 2015).
4. Numerical implementation of CBD constitutive models

Taking the CPB06exn and Yoon's criterion as the typical SSM or SIM, the above CBD framework is numerically implemented into the explicit 3D-FE platform for practical usages by combining the implicit algorithm and interpolation approach.
4.1. Implicit integration algorithm for stress updating
A general constitutive equation adapted to small strain and rate-independent elastoplasticity is used, which is reduced
from the general formula of the nite deformation elastoplasticity (Simo and Hughes, 1998; Belytschko et al., 2014):
8
s_ C : _ e C : _ _ p
>
>
>
> _ p l_ rs; q
>
<
q_ l_ hs; q
>
>
> f_ fs : s_ fq ,q_ 0
>
>
:_
l 0; f 0; l_ f 0
(11)
134
where C is the fourth-order elastic modulus tensor, fs and fq are the derivation of the yield function, l_ is the plastic multiplier, r
is the direction of the plastic ow (the AFR is assumed, i.e., r ~ fs along the normal direction of the current yield surface in
stress space), q is the internal variable (such as effective plastic strain), h is the gradient of the internal variable, and the last
equation represents the loading-unloading condition, the linear elastic behavior is assumed.
The magnitude of the plastic multiplier l_ can be determined by Eq. (12):
l_
fs : C : _
fq ,h fs : C : r
(12)
For a strain-driven FE formula, at the beginning of increment step n 1, n ; pn ; qn and strain increment D Dt _ are
p
given. The integration algorithm is used to get n1 ; n1 ; qn1 to meet consistency condition. Accordingly, the stress rate,
plastic strain rate and gradient of an internal variable can be calculated.
To overcome the shortcomings of the explicit algorithm, such as the rst order forward Euler integration method, the
implicit integration one can be derived based on Euler's backward integration and return mapping technique (Simo and
Hughes, 1998). Eq. (13) shows the integration equations. Compared with the explicit algorithm, the implicit one is unconditionally stable. The updating variables are determined using the results of the last increment. This avoids some impractical
values resulting in pseudo-unloading.
8
n1 n D
>
>
>
> pn1 pn Dln1 rn1
<
qn1 qn
Dln1 hn1
>
>
>
s
C : n1 pn1
n1
>
:
fn1 f sn1 ; qn1 0
(13)
where Dln Dt l_ n .
A numerical scheme based on the elastic trial stress, plastic corrector and returning mapping technique is applied. Fig. 4
shows the updating algorithm with a geometric interpretation.
First, we obtain the plastic strain increment:
Dpn1 pn1 pn Dln1 rn1

By substituting Eq. (14) into Eq. (13), we can get Eq. (15)
Fig. 4. Stress updating algorithm using the closest projective method.
(14)
sn1 C : n1 pn Dpn1 C : n Dn pn Dpn1

p
sn C : Dn C : Dn1 strial
n1 Dln1 C : r n1
135

(15)
where strial
n1 sn C : Dn is the elastic trial stress tensor, Dln 1C:rn 1 is the plastic correction along the plastic ow
direction of increment step n 1. The state of elastic prediction is driven by the increment of total strain, and the plastic
correction process is driven by the plastic multiplier l_ .
In the elastic prediction process, from Eq. (15), we obtain
Dsn1 C : Dpn1 Dln1 C : r n1
(16)
During the plastic correction process, the total strain is constant, and the linearization relates to the increment of the
plastic multiplier Dl. The Newton iteration is used to solve the nonlinear algebraic equations. As Simo and Hughes (1998)
observed, the process of the Newton iteration based on the classication of linear equations is essentially the same as the
closest point projection. In the rest of this paper, we will omit the corner mark n 1 for the increment of stress and time in
the equations, so all the values are the (n 1)th increment step unless specically illustrated. We rewrite Eq. (13) in the
form of Eq. (17).
8
< a p pn Dlr 0
b q qn Dlh 0
:
f f s; q 0
(17)
Linearizing this equation system, the following can be obtained
8
k
k
1
>
< ak C : Dsk Dl Drk dl r k 0
k
k
bk Dqk Dl Dhk dl hk 0
>
: f k f k : Dsk f k $Dqk 0
s
q
(18)
where
k
k
k
k
k
k
k
Drk rk
hk
s : Ds r q $Dq ; Dh
s : Ds hq $Dq
(19)
The corner mark s and q denote the derivative.

From Eq. (18), we can get Ds(k), Dq(k) and dl(k) simultaneously.
By substituting Eq. (19) into Eq. (18), we obtain
h
i1 k
o
n
o
n
Ds
~ k dlk ~r k
Ak
a
k
Dq
(20)
where
h
i1
1
C Dlrs
Ak
Dlhs
Dlrq
I Dlhq
k n
o k n
o k
~ k ak ; ~r k r k
; a
b
h
(21)
The stress and internal variables can be determined as
Dsk
Dqk
in
o
h
in
o
h
~ k dlk Ak ~rk
Ak a
(22)
Substituting this into Eq. (18), dl(k) can be obtained
dlk
~
f k vf k Ak a
k
vf k Ak ~r
where we use the mark:
(23)
136

vf fs
fq
(24)
Lastly, obtain the updating values of plastic strain, internal variables and plastic multiplier.
8 pk1
pk Dpk pk C 1 : sk
<
qk1 qk Dqk
: k1
Dl
Dlk dlk
(25)
The detailed algorithms for the SSM and SIM-based CBD models are stated in Appendix A.
4.2. General interpolation approach describing evolution of distorted plasticity
Based on the constructed discontinuous yield loci, a general interpolation approach is provided to smoothly calculate the
critical shear stress at arbitrary strain for the CBD constitutive model.
By using the yield stress data through experiments and VPSC computations, the constitutive coefcients in the initial and
subsequent yield surfaces corresponding to several certain equivalent plastic strains (p1 < p2 < / < pm ) are obtained.
The effective stress si sfsi ; LA pi ; LB pi g and the critical shear stress tY i tY pi corresponding to each of the individual strain levels are calculated.
For any given level p (pi < p < pi1 ), the distorted yield function can be determined:
a
f s; p Fs; p Fp 0
(26)
where a is the exponent of the homogeneous yield functions, a equals 3 for Yoon's model, a equals 2 for CPB06 model in this
study.

Fs; p xp F s; pi 1 xp F s; pi1

Fp xp tY ai 1 xp tY ai1
1=a
(27)
(28)
By considering a linear interpolation scheme, the weighting parameter appearing in Eqs. (27) and (28) is dened as
xp pi1 p
.

p
p
i1 i
(29)
such that xpi 1 and xpi1 0.

Thus, the observed distortions in the shapes of the yield loci could be described.
5. Calibration methods for the CBD constitutive model
The parameter identication procedure is crucial for potential usages of the complex constitutive models in forming
processes when the material parameters involved increase to capture the unique plastic behaviors. The constructed CBD
framework involves the calibration of not only one yield locus but also couples discontinuous yield loci as shown in Fig. 2. Two
corresponding bottlenecks should be addressed: rst, establishing the objective function to comprehensively cover the plastic
deformation behaviors of the materials, and then, optimizing the identication procedure to ensure the convergence of each
yield locus and avoid the overlapping of the neighboring calibrated yield loci.
5.1. Objective function for parameter identication
The fundamental data, such as yield stress and plastic ow, should be obtained through physical experiments or virtual
experiments, i.e., x-y in-plane deformation, x-z through-thickness deformation and y-z through-thickness deformation. Here,
the equations associating yield stress and plastic ow with the material parameters are presented for the SIM and SSM-based
CBD constitutive models.
Tension and compression yield stress for the SSM (Yoon's model)-based CBD
In the x-y plane, the uniaxial yield stresses denoted as Txy(q) and Cxy(q) are measured from in-plane uniaxial tensile and
compressive tests in a direction with an angle of q from the rolling. The stress components of T(C)xy(q) for tension and
compression in the orthotropic coordinate are obtained:
sxx TCxy qcos2 q; syy TCxy qsin2 q; sxy TCxy qsin q cos q
137
(30)
The uniaxial tensile and compressive yield stress in the x-y plane are
Txy q tY
Cxy q tY

3=2
00
1=3
00
00
3K102 K202
2K1 K12 K22
(31)
00
1=3
00
00
2K1 K12 K22
(32)

3K102 K202
3=2
where
r
h
.
i. 2

a2 2a3 cos2 q a1 2a3 sin2 q 6 a4 sin q cos q2
K10 a2 cos2 q a1 sin2 q 6; K20
r
h
.
i. 2

00
00
2
b2 2b3 cos2 q b1 2b3 sin2 q 6 b4 sin q cos q
K1 b2 cos2 q b1 sin2 q 6; K2
Similarly, Txz(q) and Cxz(q) are the uniaxial tensile and compressive yield stresses in the x-z plane with an angle of q from
the transverse direction:
Txz q tY

3=2
00
1=3
00
00
3M102 M202
2M1 M12 M22
(33)
Cxz q tY

3=2
00
1=3
00
00
3M102 M202
2M1 M12 M22
(34)
where
r
.
i. 2

h
2
2
0
a1 cos q a3 sin q 6; M2
a1 2a2 cos2 q a3 2a2 sin2 q 6 a5 sin q cos q2
r
h
.
i. 2

00
00
2
2
b1 2b2 cos2 q b3 2b2 sin2 q 6 b5 sin q cos q2
M1 b1 cos q b3 sin q 6; M2
M10
Similarly, Tyz(q) and Cyz(q) are the uniaxial tensile and compressive yield stresses in the y-z plane with an angle of q from
the vertical direction:
Tyz q tY
Cyz q tY
00
1=3
00
00
2N1 N12 N22
(35)

3=2
00
1=3
00
00
3N102 N202
2N1 N12 N22
(36)

3N102 N202
3=2
where
r
h
.
i. 2

2
2a1 a3 cos2 q 2a1 a2 sin2 q 6 a6 sin q cos q
N10 a3 cos2 q a2 sin2 q 6; N20
r
.
i. 2

h
00
00
N1 b3 cos2 q b2 sin2 q 6; N2
2b1 b3 cos2 q 2b1 b2 sin2 q 6 b6 sin q cos q2
Tension and compression yield stress for SSM (CPB06)-based CBD
Similarly, in the x-y plane:

1=a
Txy q tY jK1 j kK1 a jK2 j kK2 a jK3 j kK3 a
(37)
138

1=a
Cxy q tY jK1 j kK1 a jK2 j kK2 a jK3 j kK3 a
with

K1
Kxx Kyy
(38)

q
q

2
2
2; K2 Kxx Kyy
2; K3 Kzz
Kxx Kyy 4K 2xy
Kxx Kyy 4K 2xy
where
Kxx F1 cos2 q J1 sin2 q; Kyy F2 cos2 q J2 sin2 q

Kzz F3 cos2 q J3 sin2 q; Kxy C44 sin q cos q
2
3
1
3
1
3
2
3
1
3
1
3
2
3
1
3
1
3
J1 C12 C11 C13 ; J2 C22 C12 C23 ; J3 C23 C13 C33

In the x-z plane:

1=a
Txz q tY jM1 j kM1 a jM2 j kM2 a jM3 j kM3 a
(39)

1=a
Cxz q tY jM1 j kM1 a jM2 j kM2 a jM3 j kM3 a
(40)
with

M1
Mxx Mzz

q
q
Mxx Mzz 2 4M2xz
2; M2 Mxx Mzz Mxx Mzz 2 4M 2xz 2; M3 Myy
where
Mxx F1 sin2 q P1 cos2 q; Myy F2 sin2 q P2 cos2 q

Mzz F3 sin2 q P3 cos2 q; Mxz C55 sin q cos q
2
3
1
3
1
3
2
3
1
3
1
3
2
3
1
3
1
3
P1 C13 C11 C12 ; P2 C23 C12 C22 ; P3 C33 C13 C32

In the y-z plane:

1=a
Tyz q tY jN1 j kN1 a jN2 j kN2 a jN3 j kN3 a
(41)

1=a
Cyz q tY jN1 j kN1 a jN2 j kN2 a jN3 j kN3 a
(42)
with

N1
Nyy Nzz

q
q

2
2
2; N2 Nyy Nzz
2; N3 Nxx
Nyy Nzz 4N2yz
Nyy Nzz 4N2yz
where
Nxx J1 cos2 q P1 sin2 q; Nyy J2 cos2 q P2 sin2 q

Nzz J3 cos2 q P3 sin2 q; Nyz C66 sin q cos q
Plastic ow for the CBD constitutive models
For both the SIM and SSM-based CBD models, according to the AFR, the anisotropic exponent Rq is obtained in the x-y
plane:
Rq
dq90+
dzz
vF
vF
vF
sin2 q
cos2 q
sin q cos q
vsxx
vsyy
vsxy

vF
vF
vsxx vsyy

(43)
Error function
For calibration purposes, using the above equations, an error function as shown in Eq. (44) is constructed based on predicted tensile stress ratios and plastic ow ratios. The coefcients related to the anisotropy/asymmetry parameters are obtained by minimizing the error function.
Error
X
2
sexp
spre
1
i
i
2
X exp . pre
Rj
Rj 1
139
(44)
where the superscript pre and exp represent the predicted value and the experimental data, respectively. This objective
function is constructed based on the Weight-Sum approach where the weight factors for all experimental values are set
to 1.
The SIM-based model (Yoon's model), as shown in Eq. (8), involves 12 coefcients in the fourth-order tensors LA and LB in
the full stress space. Among them, eight coefcients (a1, a2, a3, a4, b1, b2, b3, b4) are related to the in-plane plastic deformation
behavior. They are calibrated using four uniaxial tensile yield stresses in the rolling direction (RD), the direction at a 45 angle
to the RD, the transverse direction (TD) and the equibiaxial tension, denoted as T0, Txy45, T90 and Tb, and four uniaxial
compressive yield stresses along the same orientations denoted as C0, Cxy45, C90 and Cb. With the material constants identied
above, the proposed yield function could describe the plastic behavior of metals under the x-y plane stress condition.
However, it cannot be used to model anisotropic/asymmetric plastic deformation under 3D loading because the throughthickness plastic behavior related parameters a5, a6, b5 and b6 are not fully calibrated yet. These material constants are
computed based on the yield stress and plastic ow data in the x-z and y-z planes. Here, we use the uniaxial tensile and
compressive yield stresses in the x-z and y-z planes along a 45 angle from the rolling direction denoted as Tzx45, Tyz45, Czx45
and Cyz45.
For the SSM (CPB06)-based model, as shown in Eqs. (2) and (9), nine independent anisotropy related parameters and one
asymmetry factor k should be calibrated. Here, eight yield stresses in the x-y plane, T0, Txy45, T90, Tb, C0, Cxy45, C90 and Cb, and
two yield stresses in the x-z plane and the y-z plane, Tzx45 and Tyz45, are employed.
5.2. Optimization algorithms
To minimize the above objective function Eq. (44), an efcient optimization algorithm should be used to identify the
parameters combined with the available experimental data. The iterative and heuristic methods are two currently used
optimization approaches. The typical iterative algorithms include the simplex method, the penalty function method, the
generalized Lagrange multiplier method and gradient-based algorithms. The typical heuristic methods are the intelligent
algorithms, such as the genetic algorithm (GA), the simulation annealing algorithm, the ant colony algorithm, the particle
swarm algorithm and the immune algorithm. Although the iterative algorithms usually have solid mathematical foundations,
the complicated derivation is difcult to solve in many problems. Even if the derivation can be obtained, an incorrect choice of
initial values may lead to severe local minima. Relatively, the heuristic algorithms depend less on the properties of the
objective functions and can provide approximate solutions with better robust stabilization despite the lack of strict mathematical descriptions and expensive in terms of CPU time. Due to the high nonlinearity of the above objective function,
instead of the iterative methods, two types of heuristic algorithms, adv., the Nelder-Mead (N-M) method (Nelder and Mead,
1965; Mathews and Fink, 2004) and GA (Lin and Yang, 1999), are selected as the candidates, and a comparison study is
conducted to develop a suitable calibration method.
The N-M method is a technique for minimizing a nonlinear objective function of N variables in an N-dimensional space
without constraints (Nelder and Mead, 1965). This method generates a new test position by extrapolating the behavior of the
objective function measured at each test point arranged as a simplex, which is a special polytope of N 1 vertices (P0, P1, Pn)
in N dimensions. The algorithm identies the point of the greatest function value and replaces this point with a better one
with a smaller function value, which is obtained through reection, expansion, contraction and reduction according to the
evaluation of the function value of every point to get a new simplex. This allows the constantly updated simplex to shrink to
the optimum solution. The original points of the simplex are used to dene a set of ablique axes with co-ordinates xi, then the
points may be taken as a N (N 1)Hessian matrix.
GA mimics the natural selection process, such as inheritance, selection, mutation and crossover (Lin and Yang, 1999). The
quality of the GA search is governed by genetic representation, population size, population initialization, tness functions, the
number of generations and the probabilities and operators of selection, crossover and mutation. For the GA method, the
options used for this research are as follows: representation type is double vector; population size is 1000; tness scaling
option is ranked; the algorithm adopts stochastic uniform method to select parents based on the scaled values calculated by
the rank scaling function; ve of the selected parents will be elite; other than elite children, 95% will be produced through
scatter recombination and 5% will be produced by mutation; number of generations is 500.
6. Results and discussions
By taking uniaxial tension/compression and mandrel bending of the HSTT as the case, the above CBD constitutive
models have been identied, implemented, evaluated and nally applied to establish the quantitative correlations
among initial textures, distorted behaviors and inhomogeneous deformation as well as multi-defect constrained
formability.
140
Tube plug
Tube
12
12
50
0.9
35
180
(a)
(b)
Fig. 5. Dimensions of the specimens: (a) Tension; (b) Compression.
6.1. Case material and forming procedures

As previously mentioned, cold rolling may produce titanium tubes with strong variations in their initial textures, which
causes great uctuations of their mechanical properties, and in turn, affects the formability. The preliminary study has
conrmed that the irregular yielding and nonlinear hardening behaviors along monotonic loadings are pronounced. However, due to the hollow structure's limitation of tubular materials, generally only the axial tension along the rolling direction
can be used to identify the plastic response, and the Hill'48 criterion is often employed with the assumption of normal
anisotropy and planar isotropy (Dick and Korkolis, 2015). Compared with sheet metals, the modeling of tubular materials lags
far behind (Kuwabara, 2007); this is especially true for titanium tubes. Subjected to the complex loading conditions such as
bending, inhomogeneous tension/compression deformation and multiple forming defects may occur (Yang et al., 2012),
which are closely related to the constitutive characteristics (Corona et al., 2006; Cazacu et al., 2013). Here, HSTT of Ti-3Al-2.5 V
(SAE, 2010) is taken as a case material. The initial texture of the as-received HSTT is the near radial crystallographic
reorientation.
Three types of experiments are conducted for the HSTT, viz., a uniaxial tension test, a uniaxial compression test and
mandrel bending. As shown in Fig. 5, for the tension test, the tube is clamped and tensioned by inserting a tube plug into a
piece of tube specimen. The compression test is performed with the tubular specimen. Fig. 6 shows that, upon bending, nonuniform tension and compression inevitably occur at the extrados and intrados of the bent tube to accommodate the bending
deformations, viz., wall thinning, wall thickening, cross-section attening and springback, which affects the service performance. Thus, mandrel bending should be an ideal process to assess the CBD constitutive models. The bending radius Rd is
2.0D, the bending angle is 135 , the relative pushing speed Vp/V is 100%; The mandrel diameter d is designed as 9.94 mm, the
mandrel extension length e is assigned as 2 mm.
6.2. Suitable identication methods for CBD constitutive models

According to the equivalent work principle, and by taking the uniaxial tension data along the RD as a reference, as seen in
Table 1, the yield stresses of HSTT for the individual equivalent plastic strains (0.2%, 2.5%, 5%, 10%, 20% and 30%) under different
paths are calculated. By comparing the calibration capability of N-M method and GA method regarding convergence, overlapping and accuracy, the suitable method for the identication of numerous parameters of the CBD framework is developed,
viz., the GA-based method, and some key factors such as initial values are provided. The Yoon's criterion is considered here as
Vp/V
Mandrel extension length e
Pressure die
Extrados of tube
Intrados of tube
Tangent point
Tube
Rd
D'
Wiper die
A-A
t'
Mandrel ball
A
Mandrel shank
t'
Bend die
t
Clamp die
Fig. 6. Mandrel bending principle and non-uniform tension and compression deformation.
141
Table 1
Yield stress for individual equivalent plastic stain under different strain paths (MPa).
p
0.2%
2.5%
5.0%
10%
20%
30%
T0
T90
Tb
C0
C90
Cb
Txy45
Tyz45
Tzx45
Cxy45
Cyz45
Czx45
739
877
992
725
828
988
869
978
978
861
945
983
846
989
1085
828
948
1071
983
1075
1081
976
1034
1084
886
1021
1122
862
992
1110
1019
1108
1144
1015
1072
1123
928
1041
1151
907
1050
1165
1061
1133
1236
1046
1115
1166
974
1069
1166
1006
1154
1244
1098
1151
1302
1078
1210
1198
1002
1097
1179
1070
1221
1298
1127
1154
1344
1104
1264
1214
a typical SIM to be incorporated into the CBD constitutive framework for the identication procedure. Tables 2 and 3 show the
nally calibrated material parameters for each level of plastic strain by the N-M method and the GA method, respectively. The
discontinuous yield loci can then be depicted as shown in Fig. 7. The predicted yield loci obtained by both optimization
methods pass through all the corresponding data points.
In the N-M method, the successful calibration is very sensitive to the initial parameters of the coordinates for 13 points
associating with the Yoon's criterion-based SIM. By using a 13 12 matrix as the initial values at the beginning of the
optimization, the material parameters of the CBD constitutive model corresponding to the rst strain increment of 0.2% are
iteratively obtained. Then, the above rst set of parameters is used to obtain the material parameters of the yield equation
corresponding to the next level of the equivalent strain such as 2.5%. The material parameters under the other strain levels are
calculated to calibrate the entire discontinuous model. We found that, in the N-M method, the non-convergence problem is
likely to occur, especially for initial small strains. It is difcult to determine the initial values of the 13 points using the N-M
method. Finally, the 13 12 matrix (1,0,0, , 0), (0, 1,0, ,0), , (0,0,0, ,1), (1,1,1, ,1) is determined as the initial values to
obtain the material parameters listed in Table 2. Unfortunately, as shown in Fig. 7(a), even the convergence problem is solved,
another abnormal phenomenon, viz., the overlapping of the neighboring yield loci under certain levels of equivalent plastic
strains, usually occurs, especially at the larger strain stages with much more pronounced distortional hardening. This
Table 2
Material coefcients under various levels of equivalent plastic strain by the N-M.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
0.5888
0.8944
1.1206
0.8877
0.7122
0.6706
1.2593
0.0164
1.0448
0.0081
0.0439
0.3225
0.6344
0.9258
1.0933
0.8881
0.7475
0.7200
1.2488
0.0464
1.0937
0.0066
0.0417
0.3541
0.6485
0.9310
1.0941
0.8955
0.7455
0.7348
1.2146
0.0552
1.1230
0.0044
0.0061
0.3341
0.6671
0.9290
1.0914
0.9058
0.7324
0.7537
0.7363
0.0024
0.8085
0.0001
1.5968
0.2809
0.6943
0.9187
1.0488
0.9212
0.7469
0.7633
0.7194
0.3132
1.1534
0.0640
0.7237
0.0088
0.7048
0.9151
1.0210
0.9251
0.7592
0.7740
0.8097
0.3872
1.2951
0.0283
0.7725
0.0167
Table 3
Material coefcients under various levels of equivalent plastic strain by the GA.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
0.5866
0.8950
1.1218
0.8878
0.7124
0.6697
0.2825
0.1627
0.8807
0.3120
0.1241
0.3456
0.6347
0.9256
1.0928
0.8878
0.7478
0.7203
0.3159
0.2655
0.6909
0.2582
0.1216
0.4046
0.6501
0.9293
1.0951
0.8957
0.7450
0.7356
0.2452
0.3226
0.6838
0.2150
0.2951
0.3837
0.6669
0.9294
1.0917
0.9059
0.7323
0.7535
0.2162
0.4649
0.4684
0.5014
0.8570
0.3042
0.6969
0.9181
1.0478
0.9214
0.7465
0.7657
1.2161
0.2336
0.8255
0.2489
0.7660
0.4533
0.7077
0.9158
1.0195
0.9246
0.7581
0.7810
1.3460
0.3246
0.9434
0.2250
0.8365
0.5638
142
Fig. 7. Calibrated yield loci using different calibration methods for the as-received tube: (a) N-M method; (b) GA method.
indicates that the N-M based method may converge to non-stationary points on problems that are related to the selection of
the initial values; thus, the reliability of the material coefcients optimized by the N-M method remains uncertain.
The GA-based method is relatively more robust and exible. In essence, the convergence result can be obtained if the
appropriate initial values are determined as mentioned in Section 5 for the GA-related algorithm. Similarly, the initial parameters are used to obtain the convergence results for the small strain. Then, to solve the overlapping problem, the
143
10
20
18
Thickening (N-M)
Thickening (GA)
14
12
10
Thinning (N-M)
Thinning (GA)
Flattening (N-M)
Flattening (GA)
2
0
20
40
60
80
100
Angles from initial bending plane to tangent section
Section flattening degree (%)
Wall changing degree (%)
16
120
140
Fig. 8. Comparison of the predicted bending deformations with the material parameters calibrated using the N-M and GA methods.
previously obtained parameters are used as a reference to obtain the material parameters for the next level of equivalent
plastic strain. This method can provide a better chance to converge to the global minimum and completely overcome the
difculty of choosing correct starting values for the constants in traditional optimization techniques.
Because both the non-convergence for each yield locus and the overlapping of the neighboring calibrated yield loci can be
solved, the above GA-based method is used to identify the material parameters in this study to describe strong anisotropy and
asymmetry behaviors. In particular, the material parameters obtained using N-M based method and the GA-based method are
used in the 3D-FE simulation of the HSTT bending. Fig. 8 shows that there are obvious discrepancies between the prediction
results by the N-M based method and the GA-based one.
6.3. Preferable yield functions for CBD constitutive model
As mentioned in Section 3, the CBD methodology has high exibility in selecting the yield functions for each level of plastic
strain, and the CPB06exn, Yoon's model and the mixed model of Hill'48 Yoon are taken as the typical SSM and SIM candidates to be implemented into the CBD framework. The type of yield criterion preferable for describing the evolution of the
distortional hardening should be conrmed. Other typical criteria, viz., the Mises, normal anisotropy Hill'48 and full Hill'48
function, are also considered into the CBD for comparison.
6.3.1. Yield loci evolution by different yield functions
Based on Fig. 7(b), the normalized yield loci are presented. Fig. 9 shows the dramatic change in the yield shape with
deformation for the as-received HSTT in either normal stress plane or shear stress plane. Fig. 10 shows the predicted yield loci
Fig. 9. Normalized yield loci for the as-received tubes.
144
Fig. 10. Comparison of the yield loci evolutions for the as-received tube using the different yield functions: (a) CPB06; (b) full Hill'48; (c) normal anisotropy
Hill'48; (d) Mises.
for the as-received tube constructed using the CPB06-based SSM, Mises, normal anisotropy Hill'48 and full Hill'48 yield
functions. For the full Hill'48 criterion, the parameters at different interval strains are calibrated using the GA method based
on Table 1. For the normal anisotropy Hill'48, the constant R of 1.51 is directly calculated using uniaxial tension tests along the
axial direction of the tube.
From Fig. 10, the greatest discrepancy between the calibrated yield loci and the experimental data can be found for the
Mises criterion and the normal anisotropy Hill'48 criterion. For the full Hill'48 yield criterion, the deformation behaviors can
be accurately depicted within the plastic strain of less than 10%, and further distorted yield behaviors cannot be described due
to its obvious asymmetry in the quadratic equation. The CPB06 presents a more accurate description than the full Hill'48 yield
equation. However, there is a certain discrepancy against the experimental data in the biaxial tension/compression region
with the equivalent plastic strain of larger than 10%. As shown in Fig. 7(b), the Yoon's criterion-based SIM provides the most
accurate description of the anisotropic/asymmetry-induced distortional plasticity and evolution.
Additionally, as shown in Fig. 11, by comparing the yield surfaces constructed by the different yield functions, it is found
that, with small deformations of less than 10%, the yield loci by all three yield functions, i.e., the full Hill'48, CPB06-based SSM
and Yoon's criteria-based SIM, are the same, while with plastic deformations of larger than 10%, the yield loci shapes by the
above three yield criteria vary greatly, although they pass through nearly all of the experimental data.
6.3.2. Prediction capability of inhomogeneous deformation
The above calibrated yield loci are introduced into the CBD framework. Considering the multiple forming indexes, a
thorough comparison of the predictions by the different CBD constitutive models is conducted against the experimental
results. The detailed FE modeling issues for mandrel bending can be found in the literature (Li et al., 2012). In the FE models,
the Young's modulus of the HSTT equals 102.662 GPa.
Fig. 12 shows the comparison of the predicted stressestrain curves with the experimental ones for uniaxial tension/
compression. The predictions by the SSM (CPB06) and SIM (Yoon's criterion)-based CBD models agree with the experimental
data of uniaxial loading. However, for each forming index of the tube bending, the predictions by the different CBD models
vary greatly. Fig. 13 shows that, as for the plastic strain distribution by the Mises criterion, hardly any plastic deformation
occurs near the neutral layer of the tube, which causes the largest springback angle, as shown in Fig. 14(a), Whereas for the
145
Fig. 11. Comparison of the yield surfaces constructed by the different yield functions: (a) 10%; (b) 30%.
Yoon's criterion-based SIM, CPB06-based SSM, Hill'48 Yoon's mixed criterion and Hill'48 model, the predicted elastic strain
has a lower share of the total strain, resulting in a smaller springback. However, the predictions of the Hill'48 and the Mises
criteria are much closer to the experimental ones. The reason for this interesting result may be attributed to the assumptions
of both linear elastic behavior and isotropy plastic behavior for the Mises criterion, which will be further conrmed in future
research. It has been observed that the nonlinear elastic recovery phenomenon is obvious for HSTT during unloading, In this
study, we focus on modeling of anisotropy and asymmetric induced plasticity and revealing the non-ignorable effects of the
distorted yielding on springback.
Fig. 14(b) shows that the cross-section attening degrees predicted by Yoon's criterion, CPB06, Hill'48 Yoon mixed
criterion and full Hill'48 based-CBD models present remarkable uctuations along the entire bending region. Yoon's criterion,
Hill'48 Yoon mixed criterion and full Hill'48-based CBD models produce the largest values of the cross-section attening,
which is much closer to the experimental results. Fig. 14(c) and (d) show that the similar tendencies of the wall thickening and
the wall thinning degree are predicted by all CBD constitutive models. Except for the Mises criterion and CPB06-based CBD,
both the wall thinning and wall thickening predictions by the other models all agree with the experimental results. In
summary, the Yoon's criterion and Hill'48 Yoon mixed criterion-based CBD models can provide the most accurate predictions of the above forming indexes with the exception of the springback.
As mentioned in Section 6.3.1, the CPB06 presents an accurate description of the distortional hardening. However, when
the discontinuous yield loci by the CPB06 are implemented into the CBD framework, the unequal deformation during bending
cannot be accurately predicted. Why does this interesting phenomenon happen? It is known that the prediction accuracy of
the constitutive model depends on whether the local deformation characteristics under special stress states closely related to
the forming indexes are captured by the proposed model or not.
Here, the stress states along the inner and outer crest line of tube and geometrical center line of the tube are depicted as
shown in Fig. 15. It is found that there simultaneously exist distinguishable tension and compression regions during tube
bending. The forming quality is closely related to these local stress states. Especially, as shown in Fig. 6, because the cross-
146
1050
1000
True stress (MPa)
950
900
850
Exp (compression)
Exp (tension)
SIM (Yoon's model)-based CBD (compression)
SIM (Yoon's model)-based CBD (tension)
SIM (CPB06)-based CBD (compression)
SIM (CPB06)-based CBD (tension)
800
750
700
0.00
0.05
0.10
Plastic strain
0.15
0.20
Fig. 12. Comparison between numerical and experimental stressestrain curves.
section attening is subjected to the resultant force of tension and compression stress, it is considered to be more sensitive to
the distorted plasticity and evolution of the tubular materials.
As shown in Fig. 11, with the small strain of less than 10%, the shapes of all three yield loci coincide, while the shapes of the
three yield loci gradually become different as the plastic deformation increases. At the strain of 30%, the distinction is most
obvious in all four quadrants. This distinction in yield loci shapes may be a nontrivial reason why the predictions by the Yoon's
criterion and Hill'48-based CBD models are in good agreement with the experimental data, whereas the predictions by the
CPB06-based CBD have an apparent discrepancy.
It is known that the prediction accuracy of the plastic deformation for the constitutive model is determined by the reliable
computation of both the yielding and the plastic ow direction. Because the AFR rule is used, the plastic ow direction is
directly related to the yield function. As shown in Fig. 16, the abovementioned shape discrepancy of the CPB06-based CBD
thus results in obvious different predictions from the ones using Yoon's criterion and the Hill'48-based CBD models. According to Eqs. (A-5) and (A-8), the r and rs at different strain increments can be calculated based on the established
discontinuous yield functions. Here, considering the stress state of s [s0,s0,0,0,0,0]1, the detailed r and rs with p of 30% are
calculated as below for different yield models including the CPB06, Yoon's criterion and Hill'48. It is found that both values of
the r and rs are different for the three yield functions. It is noted that the spatial allocations of the r and rs for the CPB06 are
most dramatic.
8
9
0:788 >
>
>
>
>
>
>
0:212 >
>
>
>
>
<
=
1
r
s0 ;
>
> 0 >
>
>
>
>
>
0
>
>
>
>
:
;
0
8
9
0:489 >
>
>
>
>
> 0:511 >
>
>
>
>
>
<
=
1
s0 ;
r
0
>
>
>
>
>
>
>
>
0 >
>
>
>
:
;
0
8
9
0:687 >
>
>
>
>
>
>
0:313 >
>
>
>
>
<
=
1
r
s0 ;
0 >
>
>
>
>
>
>
>
0 >
>
>
>
:
;
0
rs
rs
rs
0:388
6 0:030
6
6 0:418
6
6
0
6
4
0
0
2
0:326
6 0:170
6
6 0:162
6
6
0
6
4
0
0
2
0:420
6 0:240
6
6 0:180
6
6
0
6
4
0
0
0:030
0:082
0:112
0
0
0
0:418
0:112
0:530
0
0
0
0
0
0
1:537
0
0
0
0
0
0
0:952
0
0:170
0:332
0:168
0
0
0
0:162
0:168
0:342
0
0
0
0
0
0
0:726
0
0
0
0
0
0
0:892
0
0:240
0:320
0:180
0
0
0
0:180
0:080
0:260
0
0
0
0
0
0
1:010
0
0
0
0
0
0
0:900
0
3
0
0 7
7
0 7
7s0 CPB06
0 7
7
0 5
1:807
3
0
0 7
7
0 7
7s0 Yoon0 s criterion
0 7
7
0 5
0:491
3
0
0 7
7
0 7
7s0 Hill0 48
0 7
7
0 5
0:390
(45)
(46)
(47)
As mentioned in Section 3, the accuracy of CPB06 in describing the anisotropy and asymmetry depends on the careful
selection of the values of a, k and the linear transformation matrix. In this study, the CPB06 yield function is used with an a of
147
Fig. 13. Comparison of effective plastic strain predicted by the different constitutive models.
2. In the report (Cazacu et al., 2006), it was suggested that, by modifying the value of a, the plastic deformation of different
types of materials can be described. As shown in Fig. 17, for different values of a, the yield loci of CPB06 all pass through the six
experimental data. However, the shape of the yield loci becomes very distorted. This uncertainty of the yield surfaces by the
CPB06 can be further improved by increasing the linear transformation matrix as CPB06ex2, CPB06ex3 or CPB06exn (Plunkett
et al., 2008). However, it is unfortunate that the initial physical meaning may be lost away, more uncertainties may be
induced, and the total number of the involved material parameters increases dramatically, which causes the yield function to
be unfeasible for analyzing the practical forming processes.
In addition to the prediction accuracy, the computation cost is also evaluated. The simulations in the paper are all performed on the personal computer with Inter(R) Core (TM) i3-3220 CPU, 3.30 GHz and 3.47 GB memory. The actual bending
5.5
(a)
5.0
D=
D D
D
(b)
100%
4.5
6
5
4
3
2
4.0
3.5
3.0
Exp
SIM(Yoon's model)
Hill'48+SIM(Yoon's model)
SSM(CPB06)
Hill'48
Hill'48R
Mises
2.5
2.0
1.5
1
1.0
0
0
Exp
SIM(Yoon) Hill+YoonSSM(CPB06) Hill'48
Hill'48R
25
t=
t
t
Mises
60
80
100
t=
20
Wall thinning degree (%)
20
Wall thickening degree (%)
40
120
25
(c)
100%
20
t
t
140
(d)
100%
15
15
Exp
SIM(Yoon's model)
SSM(CPB06)
Hill'48
Hill'48R
Mises
10
10
Exp
SIM(Yoon's model)
SSM(CPB06)
Hill48
Hill48R
Mises
0
0
20
40
60
80
100
120
140
20
40
60
80
100
120
140
Fig. 14. Comparison of bending deformations for different constitutive models: (a) springback angle; (b) wall thickening; (c) wall thinning; (d) section attening.
148
Fig. 15. Stress states of the bending tube.
Fig. 16. Schematic of the different plastic ow for various yield surfaces.
time is 3.5 s, and the computation times for the SIM (Yoon's model), Hill'48 Yoon mixed criterion, SSM (CPB06) and Hill'48based CBD models are 6.5 h, 6.3h, 10.1 h and 4.3 h, respectively. Additionally, Yoon's yield equation associated with a Swift
hardening model is established as the case of the traditional continuous model with the assumption of isotropic hardening. As
shown in Fig. 18, the expected discrepancy is observed for different forming indexes. However, the computation cost for the
continuous model is 6.2 h, and the difference in the computation efciency is minor. Thus, the computation efciency is
validated. This further indicates that the CBD constitutive framework is suitable for practical usages in complex forming
processes.
In summary, from the perspective of capturing capabilities and efciencies of these models, we can conclude that the CBD
framework provides a practical model for the characterization of the anisotropy and asymmetry related inhomogeneous
deformations. The Yoon's criterion-based SIM proves to be preferable when embedded into the CBD framework for the
prediction of the distorted plasticity-induced unequal deformation. In the following, we used the SIM-based CBD constitutive
model to explore the role of distorted plasticity in the formability.
6.4. Application of the CBD constitutive models

Because strong texture variations of the HSTT may occur during fabrication processing, it is fundamental to have an insight
into which type of textures can achieve improved formability for subsequent processes. By taking the SIM (Yoon's criterion)based CBD constitutive models as the bridge, the quantitative correlations among the initial textures, distorted behaviors and
inhomogeneous deformation as well as multi-defect constrained formability are established under uniaxial tension/
compression as well as mandrel bending.
149
Fig. 17. The yield surfaces calibrated by CPB06-based SSM at different a.
20
14
18
12
14
Wall changing degree (%)
16
10
12
10
Thickening (continuous)
Thickening (discontinuous)
Thinning (continuous)
Thinning (discontinuous)
8
6
8
6
4
4
Flattening (continuous)
Flattening (discontinuous)
2
0
20
40
60
80
100
120
140
Fig. 18. Comparison of the bending deformations predicted using the different material models.
6.4.1. 3D characterization of tubes with different initial textures

In addition to the near radial texture of the as-received tubes, another ve typical initial textures of the HSTT are
considered, viz., rolling texture, tangential texture, radial texture, bimodal texture and random texture. Here, the distorted
plasticity and its evolution in HSTT with these initial textures are characterized by SIM (Yoon's criterion)-based CBD in 3D
space, as shown in Fig. 19. The detail calibrated parameters by GA-based method are shown in Appendix B.
It can be seen that the diverse yielding behaviors and their evolution features are presented for the tubes with different
initial textures. The yield loci for the random texture are an elliptical shape with sound symmetry and isotropy, which can be
depicted exactly using the Mises yield function. Similar to the as-received tube, with the exception of random texture, the
yield loci present a nonlinear and distorted evolution in shapes. Of these initial textures, the yield loci for the bimodal texture
are most distorted with the highest strength under in-plane equibiaxial loading. The yield loci for the as-received texture
seem more similar to the combinations of the loci for the radial texture and the tangent texture.
6.4.2. Role of distorted plasticity in uniaxial tension/compression
The detailed tension and compression processes are mentioned in Section 6.1. For the uniaxial tension, as shown in Fig. 20,
the ow stress and elongation of the HSTT with different initial textures differ greatly from each another. The tube with the
rolling texture has the best tensile properties for both strength and elongation. The reason is attributed to the parallel
relationship between the grain orientation (c-axis of the grain) and the loading direction, which causes increased deformation resistance along the axial direction and more uniform deformation. In contrast, for the tube with an initial bimodal
texture, due to harder deformation resistance along both the normal and tangent directions compared to that along the axial
direction, the deformation along the axial direction cannot be effectively accommodated by the other deformation sources.
150
Fig. 19. Calibrated yield loci from SIM-based CBD model with different initial textures: (a) rolling texture; (b) tangent texture; (c) radial texture; (d) bimodal
texture; (e) random texture.
This causes decreased strength and poorer formability with uniaxial tension along axial direction. For the tubes with initial
radial and tangent textures, the worst elongation is observed, in which necking occurs after a short stretching.
For uniaxial compression with 30% height reduction, as shown in Fig. 21, different deformation behaviors occurs. For the
tubes with initial rolling, random and tangential textures, severe inhomogeneous deformation, viz., the cross section
distortion of the ring specimens, is observed. The reason is that the deformation resistance along the thickness direction of
the tube is less than the resistance along the axial direction. However, as previously mentioned, the tubes with initial rolling,
random and tangential textures all present preferential behaviors under tension conditions. For the tubular materials with
initial bimodal, radial and as-received textures, homogenous deformation can be observed with only slight attening phenomena. The reason is attributed to the sufcient deformation coordination capability along the thickness direction of the
tube.
6.4.3. Role of distorted plasticity in bending formability
The details of the mandrel bending are mentioned in Section 6.1. Fig. 22 represents the distribution contours of the
equivalent plastic strain for bent tubes with different initial textures. The distribution of plastic strain in tubes with bimodal
and as-received textures is more homogeneous and thus has a greater share of the total strain. As a result, the springback
angles of the tubes with bimodal textures and as-received textures are the lowest as shown in Fig. 23(a). The springback of the
tubes with bimodal textures is approximately 50% less than that of the as-received tube and 74% less than that of the tubes
151
1200
True stress (MPa)
1000
800
600
As-recieved
Rolling
Tangential
Radial
Bimodal
Random
400
200
0
0.000
0.025
0.050
0.075
0.100
True strain
0.125
0.150
0.175
Fig. 20. True stressestrain curves of tubes with different initial textures during uniaxial tension.
with an initial rolling texture, which present the largest springback angle. Regarding the cross-section attening, as shown in
Fig. 23(b), there are obvious uctuations along the entire bending regions for tubes with initial bimodal textures and asreceived textures, adv., apparent minimal value and maximal value are observed. However, these deformation characteristics for the tubes with other textures show little change along the bending regions. Because of difcult deformation along the
normal direction of tube coordinated by a harder pyramidal slip and compression twinning, both the wall thinning degree
and the wall thickening degree of the tubes with the initial bimodal texture and the as-received texture are much smaller than
those with the other textures, as shown in Fig. 23(c) and (d).
The comparison of the above forming indexes has claried that the initial bimodal texture can signicantly improve the
bending formability, i.e., reduced cross-section attening, wall thinning/thickening and springback. This means that a higher
formability with smaller bending radius can be achieved if the near bimodal texture was tailored during cold rolling by
dedicatedly coordinating the plastic ows along the axial, normal and hoop directions.
7. Conclusions and remarks
This study focuses on accurately and efciently modeling of anisotropy/asymmetry-induced distorted plasticity and
nonlinear evolutions of hard-to-deform materials. The constructed CBD framework is numerically implemented, fully evaluated and practically applied to provide transferable knowledge of the correlations among initial textures, distorted plasticity,
inhomogeneous deformation and multi-defect constrained formability. The conclusions and remarks are as follows:
(1) We try to divide the existing yield criteria into two categories from the viewpoint of the physical meanings, viz., the
principal shear stress-based models (SSM) and the stress invariants-based models (SIM). Considering the multimechanism coordinated distortional yielding and nonlinear evolution of the materials, a unied CBD constitutive
framework is constructed, in which both the discontinuous SSM, the SIM or their combinations are used to characterize
the distorted shape of the yielding, and an interpolation approach is adopted to present the nonlinear evolution of the
distorted plasticity in the full stress space. The microstructure evolutions (such as texture variation) caused nonlinear
hardening and ow rules are implicitly included in this discontinuous framework.
(2) Taking CPB06 and Yoon's criterion into the CBD constitutive framework, both the SSM- and SIM-based CBD models are
successfully numerically implemented into the explicit 3D-FE platform by combining implicit algorithm and interpolation approach to simulate complex forming processes. Via a comparison of the N-M and GA methods in terms of
convergence, overlapping and accuracy, the GA-based approach is proposed to accurately and efciently calibrate the
numerous parameters of the CBD framework, and some key factors such as the initial values are provided. Regarding
the capturing capability of anisotropy/asymmetry coupled distorted behaviors, the SIM model seems to be preferable
for embedding into the CBD framework because of its sound physical basis and numerical robustness.
(3) Taking the uniaxial tension/compression and mandrel bending as the cases, the distorted plasticity and evolution of
the HSTT with six typical initial textures are characterized. The correlations among the initial textures, distorted
behaviors and inhomogeneous deformation are quantitatively established to improve the formability of the hard-todeform materials. Different texture characteristics and plastic deformation behaviors are required to improve the
formability in different forming processes. The anisotropy/asymmetry characteristics specic to HSTT greatly affect the
inhomogeneous deformation and related defects. For tension, the tubes with rolling textures present the best tensile
properties for strength and elongation. For compression, a more uniform deformation is observed for the tubes with
152
Fig. 21. Compression deformation of tubes with different initial textures: (a) cross section; (b) half of the ring specimen.
the as-received texture and bimodal texture. For complex bending, the bimodal texture is identied as most preferable
for the best bendability in terms of cross-section attening, wall thickening, wall thinning and springback.
(4) The currently developed materials with more complex microstructures may cause the transition from elasticity to
plasticity to be more distorted in the full stress space, particularly accompanying with interaction of the multiple
complex mechanisms, such as reversible/irreversible, hardening/softening and cyclic/monotonic behaviors. This situation demands further extensive evaluation, extension and application of the present CBD constitutive framework for
describing greater nonlinear plastic behaviors in the case of complex forming conditions with spatial, abrupt strain
path changes or thermal-mechanical coupling loadings, such as cold/warm sheet/bulk sheet forming. In any event,
153
Fig. 22. Comparison of effective plastic strain for tubes with different initial textures.
5.5
(a)
5.0 (b)
Section flattening degree(%)
9
7
6
5
4
3
2
1
0
Wall thinning degree (%)
Wall thickening degree (%)
3.5
3.0
2.5
20
15
As-received
Rolling
Tangential
Radial
Bimodal
Random
10
5
0
20
40
60
80
100
120
140
()
As-recieved
Rolling
Tangential
Radial
Bimodal
Random
2.0
1.5
1.0
0.5
30
(c)
4.0
20
40
60
80
100
120
140
Angles from initial bending plane to tangent section ( )
As-receivedRolling Tangential Radial Bimodal Random
30
25
4.5
As-received
Rolling
Tangential
Radial
Bimodal
Random
(d)
25
20
15
10
5
0
20
40
60
80
100
120
140
()
Fig. 23. Comparisons of the bending deformations: (a) springback angle; (b) section attening; (c) wall thickening; (d) wall thinning.
efciently calibrated and robustly implemented constitutive models with fewer parameters, greater accuracy and
sound physical backgrounds are required when analyzing practical forming processes which are subjected to complex
loading boundaries.
154
Supplementary video related to this article can be found at http://dx.doi.org/10.1016/j.ijplas.2016.03.002.

Acknowledgments
The authors would like to thank the reviewers for their constructive suggestions, Dr. R. Lebensohn of Los Alamos National
Laboratory for the access of VPSC code and Dr. J. Ma of the Northwestern Polytechnical University for gure preparation. This
research was conducted under the nancial support of the National Science Fund for Excellent Young Scholars (51522509),
the National Natural Science Foundation of China (51275415), Fundamental Research Funds for the Central Universities
(3102014KYJD001), the EU Marie Curie Actions e MatProFuture Project (FP7-PEOPLE-2012-IRSES-318968) and the 111 Project
(B08040).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.ijplas.2016.03.002
Appendix A. Stress updating algorithm
According to the effective theory, the effective stress can be dened as
s A1 tY
(A-1)
where for the SIM-based yield function (which here refers to Yoon's model)
A1 3
a22 a23 a2 a3
3=2
1=3
b2 b3 b2 b3
(A-2)
and for the SSM-based yield function (which here refers to CPB06 model)

1=a
A1 jF1 j kF1 a jF2 j kF2 a jF3 j kF3 a
(A-3)
where. F1 23 C11 13 C12 13 C13 ; F2 23 C12 13 C22 13 C23 ; F3 23 C13 13 C32 13 C33
Therefore, the gradient of the internal variable h can thus be deduced
s:r
s
(A-4)
For the AFR, the direction of plastic ow can be determined by
vf
fr1 r2 r3 r4 r5 r6 gT
vs
(A-5)
ri
3 A 1=2 A
B
J
J2;i J3;i
2 2
(A-6)
and for the SSM-based yield function (which here refers to CPB06 model)
ri

vf vS1
vf vS2
vf vS3
vf vS1 vH1 vS1 vH2 vS1 vH3
vS1 vsi vS2 vsi vS3 vsi

vS1 vH1 vsi vH2 vsi vH3 vsi


(A-7)
where S1, S2, S3 is the principal value of S, viz, the three

PSi S3i 3H1 S2i 3H2 Si 2H3 0, and H1, H2, H3 is the invariants of S:

H1 Sxx Syy Szz 3
.
H2 S2yx S2zx S2yz Syy Szz Szz Sxx Sxx Syy 3
.

H3 2Syx Szx Syz Sxx Syy Szz Syy S2zx Szz S2xy Sxx S2yz 2
roots
of
the
characteristic
equation
155
r

q
H 21 H2 cos
H1
3
r
q 4p
H1
S2 2
H 21 H2 cos
3
r
q 2p

H1
S3 2
H 21 H2 cos
3
3=2
q arccos q p
; p H12 H2 ; q H13 3=2H1 H2 H3
S1 2
Especially, S2i 2H1 Si H2 0, Si H1
q
H 21 H2
Then, the derivative of r and h can be determined:
r11
6 r21
6
vr 6
r31
rs
6
vs 6
6 r41
4 r51
r61
r12
r22
r32
r42
r52
r62
r13
r23
r33
r43
r53
r63
r14
r24
r34
r44
r54
r64
r15
r25
r35
r45
r55
r65
3
r16
r26 7
7
r36 7
7
r46 7
7
r56 5
r66
(A-8)
rij
3 A 1=2 A A 3 A 1=2 A
B
J
J2;i J2;j J2
J2;ij J3;ij
4 2
2
(A-9)
for the SSM-based yield function (which here refers to CPB06 model)
rij
v2 f
v2 f vS1 vS1
vf v2 S1
v2 f vS2 vS2
vf v2 S2
v2 f vS3 vS3
vf v2 S3
vsi vsj vS21 vsi vsj vS1 vsi vsj vS22 vsi vsj vS2 vsi vsj vS23 vsi vsj vS3 vsi vsj
(A-10)
Then, we obtain
hs
s : rs
1
h
r

s
s atY a
rq 0; hq 0
(A-11)
(A-12)
Then, the plasticity parameters can be determined:
A
Q :C
Dlhs : Q : C

1

r
0
~ g C : R ; fvf g
; fa
a1
atY
1
0
(A-13)
where
Q I DlC : r s
1
p
; R sn1 strial
n1 DlC : r n1 ; H vtY =v
Substituting Eq. (A-13) into Eq. (23), the increment of the plastic multiplier can be determined:

f r aHta1
Y $Dlhs : Q : R

dl
r aHta1
aHhta1
Y
Y $Dlhs : Q : C : r
(A-14)
156
Appendix B. Material coefcients for the CBD constitutive models
Table B-1
Material coefcients for SIM (Yoon's criterion) with rolling textured tubes.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
1.2160
1.0019
1.0006
0.9443
0.9445
1.1518
0.2931
0.2913
0.1384
1.5326
0.1536
0.4091
1.1042
0.9959
0.9964
0.9175
0.9174
1.0749
0.0024
0.6390
0.3164
0.1545
0.0001
0.4326
1.1272
1.0057
1.0064
0.9323
0.9321
1.0944
0.0024
0.4552
0.2795
0.4020
0.5127
0.3858
1.1789
1.0285
1.0290
0.9631
0.9630
1.1326
0.0010
0.5283
0.7067
0.5147
0.4430
0.3300
1.2207
1.0408
1.0410
0.9892
0.9888
1.1473
0.0006
0.5283
0.8896
0.3924
0.3017
0.0002
1.2448
1.0408
1.0410
1.0056
1.0057
1.1625
0.0002
0.5291
0.8909
0.3916
0.3017
0.0001
Table B-2
Material coefcients for SIM (Yoon's criterion) with tangential textured tubes.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
0.9020
1.0954
0.9015
0.8510
1.0373
0.8509
0.1588
0.2119
0.2031
0.2356
0.3808
0.5947
0.9465
1.0518
0.9477
0.8736
1.0230
0.8728
0.6143
0.0024
0.2500
0.1467
0.3790
0.0006
0.9420
1.0566
0.9430
0.8737
1.0263
0.8736
0.3325
0.0024
0.3444
0.4267
0.3652
0.4194
0.9316
1.0673
0.9308
0.8718
1.0256
0.8722
0.5782
0.0006
0.5325
0.4231
0.3002
0.4408
0.9189
1.0775
0.9206
0.8740
1.0135
0.8730
0.5747
0.0006
0.6518
0.3121
0.0010
0.2925
0.9110
1.0865
0.9098
0.8791
1.0164
0.8799
0.5744
0.0002
0.6422
0.3073
0.0009
0.2910
Table B-3
Material coefcients for SIM (Yoon's criterion) with radial textured tubes.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
0.9017
0.9018
1.0951
1.0366
0.8506
0.8515
0.1992
0.1719
0.2236
0.3760
0.3535
0.2383
0.9474
0.9472
1.0518
1.0235
0.8730
0.8730
0.6246
0.2215
0.0013
0.3601
0.1455
0.0005
0.9424
0.9428
1.0568
1.0264
0.8737
0.8737
0.2350
0.4685
0.0014
0.3586
0.5080
0.3583
0.9316
0.9310
1.0668
1.0256
0.8717
0.8723
0.6536
0.4676
0.0001
0.2988
0.3985
0.4700
0.9209
0.9182
1.0787
1.0133
0.8732
0.8737
0.7844
0.4674
0.0001
0.0001
0.2661
0.3457
0.9099
0.9112
1.0868
1.0162
0.8793
0.8790
0.7823
0.4620
0.0001
0.0001
0.2631
0.3427
Table B-4
Material coefcients for SIM (Yoon's criterion) with bimodal textured tubes.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
0.5068
0.9827
1.0651
0.8376
0.8301
0.8377
0.2508
1.1385
0.6222
0.1298
0.9282
0.0000
0.4787
0.9758
1.0503
0.8091
0.7938
0.8695
0.2304
0.9196
0.4001
0.0000
1.2405
0.0000
0.4306
0.9750
1.0391
0.8085
0.7922
0.8663
0.8841
0.5917
0.1442
0.6243
0.4998
0.0000
0.3170
0.9719
1.0049
0.8252
0.7891
0.8547
0.9936
0.1738
0.3576
0.4165
0.3959
0.6262
0.3249
0.9515
0.9839
0.8611
0.7235
0.8743
1.4887
0.1877
0.2729
0.3751
0.0000
0.4121
0.3403
0.9462
1.0011
0.8891
0.7216
0.8880
1.5319
0.1894
0.2953
0.1708
0.0000
0.0000
157
Table B-5
Material coefcients for SIM (Yoon's criterion) with random textured tubes.
p
0.2%
2.5%
5.0%
10%
20%
30%
a1
a2
a3
a4
a5
a6
b1
b2
b3
b4
b5
b6
1
1
1
1
1
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
0
0
0
0
1
1
1
1
1
1
0
0
0
0
0
0
Table B-6
Material coefcients for SIM (Yoon's criterion) with the as-received tubes.
p
0.2%
2.5%
5.0%
10%
20%
30%
C11
C22
C33
C23
C13
C12
C44
C55
C66
k
5.313
3.017
4.440
3.811
3.915
4.260
0.805
0.871
1.095
0.011
5.764
4.525
5.003
4.192
4.901
5.513
0.853
0.916
1.089
0.005
5.120
2.793
3.177
4.308
3.708
3.807
0.874
0.895
1.096
0.000
2.319
4.537
2.894
2.924
3.597
3.566
0.905
0.859
1.112
0.006
0.665
0.028
0.497
0.552
0.306
0.673
0.979
0.833
1.122
0.104
0.696
0.062
0.457
0.521
0.313
0.661
1.033
0.813
1.120
0.147
References
Abedrabbo, N., Pourboghrat, F., Carlsey, J., 2006. Forming of aluminum alloy at elevated temperatures-part 2: numerical modeling and experimental
verication. Int. J. Plast. 22, 342e373.
2005. Plastic anisotropy and the role of non-basal slip in magnesium alloy AZ31B. Int. J. Plast. 21, 1161e1193.
Agnew, S.R., Duygulu, O.,
Armstrong, P.J., Frederick, C.O., 1966. A Mathematical Representation of the Multiaxial Bauschinger Effect. Report RD/B/N731. CEGB, Central Electricity
Generating Board, Berkeley,UK.
Banabic, D., 2010. Sheet Metal Forming Processes: Constitutive Modeling and Numerical Simulation. Springer.
Banerjee, D., Williams, J.C., 2013. Perspectives on titanium science and technology. Acta Mater. 61 (3), 844e879.
Barlat, F., Lege, D.J., Brem, J.C., 1991. A six-component yield function for anisotropic materials. Int. J. Plast. 7, 693e712.
Barlat, F., Maeda, Y., Chung, K., Yanagawa, M., Brem, J.C., Hayashida, Y., Lege, D.J., Matsui, K., Murtha, S.J., Hattori, S., Becker, R.C., Makosey, S., 1997. Yield
function development for aluminum alloy sheets. J. Mech. Phys. Solids 45 (11/12), 1727e1763.
Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Lege, D.J., Chu, E., 2003. Plane stress yield function for aluminum alloy sheets-part 1: theory. Int. J. Plast.
19 (9), 1297e1319.
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. Plast. 21, 1009e1039.
Bassani, J.L., Racherla, V., 2011. From non-planar dislocation cores to non-associated plasticity and strain bursts. Prog. Mater. Sci. 56 (6), 852e863.
Belytschko, T., Liu, W.K., Moran, B., Elkhodary, K., 2014. Nonlinear Finite Elements for Continua and Structures, second ed. John Wiley & Sons. Inc.
Bron, F., Besson, J., 2004. A yield function for anisotropic materials: application to aluminum alloys. Int. J. Plast. 20, 937e963.
Bruschi, S., Altan, T., Banabic, D., Bariani, P.F., Brosius, A., Cao, J., Tekkaya, A.E., 2014. Testing and modelling of material behaviour and formability in sheet
metal forming. CIRP Annals Manuf. Technol. 63 (2), 727e749.
Cazacu, O., Plunkett, B., Barlat, F., 2001. Generalization of Drucker's yield criterion to orthotropy. Math. Mech. Solids 6, 613e630.
Cazacu, O., Barlat, F., 2004. A criterion for description of anisotropy and yield differential effects in pressure-insensitive metals. Int. J. Plast. 20, 2027e2045.
Cazacu, O., Plunkett, B., Barlat, F., 2006. Orthotropic yield criterion for hexagonal closed packed metals. Int. J. Plast. 22, 1171e1194.
Cazacu, O., Stewart, J.B., 2009. Analytic plastic potential for porous aggregates with matrix exhibiting tension-compression asymmetry. J. Mech. Phys. Solids
57 (2), 325e341.
Cazacu, O., Ionescu, I.R., Yoon, J.W., 2010. Orthotropic strain rate potential for the description of anisotropy in tension and compression of metals. Int. J. Plast.
26 (6), 887e904.
Cazacu, O., Revil-Baudard, B., Barlat, F., 2013. New interpretation of monotonic swift effects: role of tension-compression asymmetry. Mech. Mater. 57,
42e52.
Chaboche, J.L., 1986. Time independent constitutive theories for cyclic plasticity. Int. J. Plast. 2, 149e188.
Chaboche, J.L., 2008. A review of some plasticity and viscoplasticity constitutive theories. Int. J. Plast. 24 (10), 1642e1693.
Chang, Y.R., Kochmann, D.M., 2015. A variational constitutive model for slip-twinning interactions in hcp metals: application to single- and polycrystalline
magnesium. Int. J. Plast. 73, 39e61.
Choi, S.H., Kim, D.H., Lee, H.W., Seong, B.S., Piao, K., Wagoner, R.R., 2009. Evolution of the deformation texture and yield locus shape in an AZ31 Mg alloy
sheet under uniaxial loading. Mater. Sci. Eng. A 526, 38e49.
Choi, K.S., Pan, J., 2009. A generalized anisotropic hardening rule based on the Mroz multi-yield-surface model for pressure insensitive and sensitive
materials. Int. J. Plast. 25, 1325e1358.
Corona, E., Lee, L.H., Kyriakides, S., 2006. Yield anisotropy effects on buckling of circular tubes under bending. Int. J. Solids Struct. 43, 7099e7118.
Cyr, E.D., Mohammadi, M., Mishra, R.K., Inal, K., 2015. A three dimensional (3D) thermo-elasto-viscoplastic constitutive model for FCC polycrystals. Int. J.
Plast. 70, 166e190.
Dick, C.P., Korkolis, Y.P., 2015. Anisotropy of thin-walled tubes by a new method of combined tension and shear loading. Int. J. Plast. 71, 87e112.
Drucker, D.C., 1949. Relation of experiments to mathematical theories of plasticity. J. Appl. Mech. 16, 349e357.
Drucker, D.C., Prager, W., 1952. Soil mechanics and plastic analysis for limit design. Q. Appl. Math. 10 (2), 157e165.
158
Franois, M., 2001. A plasticity model with yield surface distortion for non proportional loading. Int. J. Plast. 17 (5), 703e717.
Gao, X.S., Zhang, T.T., Zhou, J., Graham, S.M., Hayden, M., Roe, C., 2011. On stress-state dependent plasticity modeling: signicance of the hydrostatic stress,
the third invariant of stress deviator and the non-associated ow rule. Int. J. Plast. 27 (2), 217e231.
Gawad, J., Banabic, D., Van Bael, A., Comsa, D.S., Gologanu, M., Eyckens, P., Van Houtte, P., Roose, D., 2015. An evolving plane stress yield criterion based on
crystal plasticity virtual experiments. Int. J. Plast. 75, 141e169.
Ghaffari Tari, D., Worswick, M.J., Ali, U., Gharghouri, M.A., 2014. Mechanical response of AZ31B magnesium alloy: experimental characterization and
material modeling considering proportional loading at room temperature. Int. J. Plast. 55, 247e267.
Graff, S., Brocks, W., Steglich, D., 2007. Yielding of magnesium: from single crystal to polycrystalline aggregates. Int. J. Plast. 23 (12), 1957e1978.
Haddadi, H., Bouvier, S., Banu, M., Maier, C., Teodosiu, C., 2006. Towards an accurate description of the anisotropic behaviour of sheet metals under large
plastic deformations: modelling, numerical analysis and identication. Int. J. Plast. 22 (12), 2226e2271.
Hershey, A.V., 1954. The plasticity of an isotropic aggregate of anisotropic face-centered cubic crystals. J. Appl. Mech. Trans. ASME 21 (3), 241e249.
Hill, R., 1948. A theory of the yielding and plastic ow of anisotropic metals. Proc. Roy. Soc. Lond. 193, 281e297.
Hill, R., 1979. Theoretical plasticity of textured aggregates. Math. Proc. Camb. Philos. Soc. 85 (1), 179e191.
Horstemeyer, M.F., Bammann, D.J., 2010. Historical review of internal state variable theory for inelasticity. Int. J. Plast. 26 (9), 1310e1334.
Hosford, W.F., 1972. A generalized isotropic yield criterion. J. Appl. Mech. 39, 607e609.
Hosford, W.F., 1979. On yield loci of anisotropic cubic metals. In: Proc. 7th North American Metalworking Conf. SME, Dearborn, MI.
Iadicola, M.A., Foecke, T., Banovic, S.W., 2008. Experimental observations of evolving yield loci in biaxially strained AA5754-O. Int. J. Plast. 24 (11),
2084e2101.
Ishikawa, H., 1997. Subsequent yield surface probed from its current center. Int. J. Plast. 13 (6e7), 533e549.
Kabirian, F., Khan, A.S., Gn
aupel-Herlod, T., 2015. Visco-plastic modeling mechanical responses and texture evolution of extruded AZ31 magnesium alloy in
various loading conditions. Int. J. Plast. 68, 1e20.
Karallis, A.P., Boyce, M.C., 1993. A general anisotropic yield criterion using bounds and a transformation weighting tensor. J. Mech. Phys. Solids 41,
1859e1886.
Khan, A.S., Kazmi, R., Pandey, A., Stoughton, T., 2009. Evolution of subsequent yield surfaces and elastic constants with nite plastic deformation. Part-I: a
very low work hardening aluminum alloy (Al6061-T6511). Int. J. Plast. 25 (9), 1611e1625.
Khan, A.S., Yu, S.J., 2012. Deformation induced anisotropic responses of Ti-6Al-4V alloy. Part I: experiments. Int. J. Plast. 38, 1e13.
Kuwabara, T., 2007. Advances in experiments on metal sheets and tubes in support of constitutive modeling and forming simulations. Int. J. Plast. 23, 385e419.
Lee, M.G., Wagoner, R.H., Lee, J.K., Chung, K., Kim, H.Y., 2008. Constitutive modeling for anisotropic/asymmetric hardening behavior of magnesium alloy
sheets. Int. J. Plast. 24 (4), 545e582.
Lee, M.G., Barlat, F., 2014. Modeling of plastic yielding, anisotropic ow, and the Bauschinger effect. Compr. Mater. Process. 235e260.
Li, H., Yang, H., Song, F.F., Zhan, M., Li, G.J., 2012. Springback characterization and behaviors of high-strength Ti-3Al-2.5V tube in cold rotary draw bending. J.
Mater. Process. Technol. 212, 1973e1987.
Lin, J., Yang, J., 1999. GA-based multiple objective optimisation for determining viscoplastic constitutive equations for superplastic alloys. Int. J. Plast. 15 (11),
1181e1196.
Liu, C., Huang, Y., Stout, M.G., 1997. On the asymmetric yield surface of plastically orthotropic materials: a phenomenological study. Acta Mater. 45 (6),
2397e2406.
Mathews, J.H., Fink, K.D., 2004. Numerical Methods Using MATLAB. Prentice Hall. Inc.
McDowell, D.L., 2010. A perspective on trends in multiscale plasticity. Int. J. Plast. 26 (9), 1280e1309.
Mohr, D., Chevin, M.A., Greve, L., 2013. Deformation behavior of magnesium extrusions with strong Basal texture: experiments and modeling. J. Appl. Mech.
80 (6), 061002.
Muhammad, W., Mohammadi, M., Kang, J., Mishra, R.K., Inal, K., 2015. An elasto-plastic constitutive model for evolving asymmetric/anisotropic hardening
behavior of AZ31B and ZEK100 magnesium alloy sheets considering monotonic and reverse loading paths. Int. J. Plast. 70, 30e59.
Nelder, J.A., Mead, R., 1965. A simplex method for function minimization. Comput. J. 7, 308e313.
Nixon, M.E., Cazacu, O., Lebensohn, R.A., 2010. Anisotropic response of high-purity a-titanium: experimental characterization and constitutive modeling. Int.
J. Plast. 26, 516e532.
Patra, A., Zhu, T., McDowell, D.L., 2014. Constitutive equations for modeling non-Schmid effects in single crystal bcc-Fe at low and ambient temperatures.
Int. J. Plast. 59, 1e14.
Plunkett, B., Lebensohn, R.A., Cazacu, O., Barlat, F., 2006. Evolving yield function of hexagonal materials taking into account texture development and
anisotropic hardening. Acta Mater. 54, 4159e4169.
Plunkett, B., Cazacu, O., Barlat, F., 2008. Orthotropic yield criteria for description of the anisotropy in tension and compression of sheet metals. Int. J. Plast. 24
(5), 847e866.
Prager, W., 1956. A new method of analysing stresses and strains in work-hardening plastic solids. ASME J. Appl. Mech. 23, 493e496.
SAE, 2010. Titanium Alloy Tubing, Seamless, Hydraulic 3Al-2.5V, Texture Controlled, Cold Worked, Stress Relieved. SAE Aerospace Material Specication SAE
AMS4946C(2002e10).
Shutov, A.V., Ihlemann, J., 2012. A rheological model for arbitrary symmetric distortion of the yield surface. Int. J. Plast. 39, 152e167.
Simo, J.C., Hughes, T.J.R., 1998. Computational Inelasticity. Springer.
Smith, J., Liu, W.K., Cao, J., 2015. A general anisotropic yield criterion for pressure-dependent materials. Int. J. Plast. 75, 2e21.
Soare, S., Yoon, J.W., Cazacu, O., 2008. On the use of homogeneous polynomials to develop anisotropic yield functions with applications to sheet forming.
Int. J. Plast. 24 (6), 915e944.
Spitzig, W.A., Sober, R.J., Richmond, O., 1975. Pressure dependence of yielding and associated volume expansion in tempered martensite. Acta Metal. 23 (7),
885e893.
Stoughton, T.B., Yoon, J.W., 2004. A pressure-sensitive yield criterion under a non-associated ow rule for sheet metal forming. Int. J. Plast. 20 (4), 705e731.
Tuninetti, V., Gilles, G., Milis, O., Pardoen, T., Habraken, A.M., 2015. Anisotropy and tension-compression asymmetry modeling of the room temperature
plastic response of Ti-6Al-4V. Int. J. Plast. 67, 53e68.
Wegener, K., Schlegel, M., 1996. Suitability of yield functions for the approximation of subsequent yield surfaces. Int. J. Plast. 12, 1151e1177.
Xiao, Y.Z., Chen, J., Cao, J., 2012. A generalized thermodynamic approach for modeling nonlinear hardening behaviors. Int. J. Plast. 38, 102e122.
Yang, H., Li, H., Zhang, Z.Y., Zhan, M., Liu, J., Li, G.J., 2012. Advances and trends on tube bending forming technologies. Chin. J. Aeronaut. 25 (1), 1e12.
Yeganeh, M., 2007. Incorporation of yield surface distortion in nite deformation constitutive modeling of rigid-plastic hardening materials based on the
Hencky logarithmic strain. Int. J. Plast. 23 (12), 2029e2057.
Yoon, J.W., Lou, Y.S., Yoon, J., Glazoff, M.V., 2014. Asymmetric yield function based on the stress invariants for pressure sensitive metals. Int. J. Plast. 56,
184e202.
Yoshida, F., Uemori, T., 2002. A model of large-strain cycle plasticity describing the Bauschinger effect and work hardening stagnation. Int. J. Plast. 18,
661e686.
Yue, Z.M., Badreddine, H., Saanouni, K., 2014. A new model describing plastic distortion fully coupled with ductile damage. Procedia Eng. 81, 1234e1239.
Zhai, J., Gao, X., Sobotka, J.C., Webler, B.A., Cockeram, B.V., 2014. Modeling the tension-compression asymmetric yield behavior of b-treated Zircaloy-4. J.
Nucl. Mater. 451 (1), 292e299.
Zhang, K., Holmedal, B., Hopperstad, O.S., Dumoulin, S., Gawad, J., Van Bael, A., Van Houtte, P., 2014. Multi-level modelling of mechanical anisotropy of
commercial pure aluminium plate: crystal plasticity models, advanced yield functions and parameter identication. Int. J. Plast. 66, 3e30.
Ziegler, H., 1959. A modication of Prager's hardening rule. Quart. Appl. Math. 17, 55e65.

Anisotropic and Asymmetrical Yielding and Its distorted-IJP2016 PDF

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Anisotropic and Asymmetrical Yielding and Its distorted-IJP2016 PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

International Journal of Plasticity 82 (2016) 127e158

Contents lists available at ScienceDirect

International Journal of Plasticity

Anisotropic and asymmetrical yielding and its distorted

Characterizing the anisotropy/asymmetry-induced distortional yielding and subsequent

* Corresponding author. Tel./fax: 86 29 88495632.

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

Fig. 1. Distorted behaviors induced by anisotropy and asymmetry.

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

f s jjS1 j kS1 ja jjS2 j kS2 ja jjS3 j kS3 ja taY 0

where L is a constant fourth-order tensor, and

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

2.1.2. Stress invariants-based criterion (SIM)

f s J2 s3=2 c$J3 s t3Y 0

where f(s) refers to Hill'48 criterion, g(s) ec(x

, x cos3q 27J3 =23J2 3=2 .

2.2. Explicit and implicit modeling evolution of plasticity

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

3. Unied continuum-based discontinuous (CBD) framework

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

where F1, F2 and F3 can be found in Appendix A.

4. Numerical implementation of CBD constitutive models

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

Dpn1 pn1 pn Dln1 rn1

Fig. 4. Stress updating algorithm using the closest projective method.

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

sn1 C : n1 pn Dpn1 C : n Dn pn Dpn1

Dsn1 C : Dpn1 Dln1 C : r n1

Linearizing this equation system, the following can be obtained

The corner mark s and q denote the derivative.

The stress and internal variables can be determined as

Substituting this into Eq. (18), dl(k) can be obtained

where we use the mark:

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

such that xpi 1 and xpi1 0.

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

Kxx F1 cos2 q J1 sin2 q; Kyy F2 cos2 q J2 sin2 q

J1 C12 C11 C13 ; J2 C22 C12 C23 ; J3 C23 C13 C33

Mxx F1 sin2 q P1 cos2 q; Myy F2 sin2 q P2 cos2 q

P1 C13 C11 C12 ; P2 C23 C12 C22 ; P3 C33 C13 C32

Nxx J1 cos2 q P1 sin2 q; Nyy J2 cos2 q P2 sin2 q

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

Fig. 5. Dimensions of the specimens: (a) Tension; (b) Compression.

6.1. Case material and forming procedures

6.2. Suitable identication methods for CBD constitutive models

Mandrel extension length e

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

Angles from initial bending plane to tangent section

Section flattening degree (%)

Wall changing degree (%)

Fig. 9. Normalized yield loci for the as-received tubes.

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

True stress (MPa)

Fig. 12. Comparison between numerical and experimental stressestrain curves.

H. Li et al. / International Journal of Plasticity 82 (2016) 127e158

f s J2 s3=2 c$J3 s t3Y 0