materials

Article
The Anisotropic Distortional Yield Surface
Constitutive Model Based on the Chaboche
Cyclic Plastic Model
Jianyun Chen 1 , Keshi Zhang 1 , Zheng Kuang 1 , Guijuan Hu 2 , Qiao Song 1 and
Yanjun Chang 1, *
1 College of Civil Engineering and Architecture, Key Laboratory of Disaster Prevention and Structural Safety
of Ministry of Education, Guangxi Key Laboratory of Disaster Prevention and Engineering Safety,
Guangxi University, Nanning 530004, China; jianyun_chen@126.com (J.C.); zhangks@gxu.edu.cn (K.Z.);
kuangzheng256@163.com (Z.K.); guxier@163.com (Q.S.)
2 School of Landscape Architecture, Zhejiang A & F University, Hangzhou 311300, China; hgj1973@sina.com
* Correspondence: yanjun_chang@126.com




Received: 7 January 2019; Accepted: 9 February 2019; Published: 12 February 2019


Abstract: Considering the cross effect in the evolution of subsequent yield surfaces for metals,
an anisotropic distortional yield surface constitutive model is developed. By introducing an
anisotropic distortional hardening function into the isotropic hardening part of the classical Chaboche
rate-dependent constitutive model, the plastic-deformation-induced distortional and anisotropic
hardening behaviors of subsequent yield surfaces are characterized. The experimental data of
distortional yield surfaces for T2 pure copper under three different loading paths, including
pre-tension, pre-torsion, and pre-tension-torsion proportional loading of 45-degree, are simulated
by implementing the models into a numerical user defined material (UMAT) procedure based on
the ABAQUS finite element package. To validate the anisotropic plastic model, the simulated yield
surfaces are compared with experimental observations and predicted results for a crystal plasticity
model and good agreement are noted. The simulations demonstrate that the proposed model can
accurately capture the characteristics of the distortional yield surface and the anisotropic hardening
process of the yield surface. Moreover, the distortional shapes of experimental subsequent yield
surfaces in loading direction and opposite direction can be better revealed by the anisotropic plastic
constitutive model than the crystal plastic constitutive model.

Keywords: anisotropic yield; constitutive model; subsequent yield surface; distortional yield surface

1. Introduction
It has been long observed that plastic deformations induce anisotropy in initially isotropic
materials [1,2], which has been a challenge for the traditional plastic constitutive theory to describe.
The study of the evolution law of yield surface relies on the establishment of a plastic constitutive
model, which serves as a basis for the analysis of macroscopic test data and verification of the
applicability and validity of the model by multiaxial tests. As loading applied to the material increases,
further deformation takes place, the stress state reaches the subsequent yield point, and there exists
a subsequent yield surface for three-dimensional loading in the stress space. The evolution forms
of a subsequent yield surface are decomposed into expansion and movement of the yield surface
in classical plasticity theory, which correspond to isotropic hardening and kinematic hardening,
respectively. Isotropic hardening and kinematic hardening can be characterized by a yield surface
radius and a back stress in the von Mises theory. Until now, researchers have conducted many studies
on plastic behaviors of metals [3–7].

Materials 2019, 12, 543; doi:10.3390/ma12030543 www.mdpi.com/journal/materials

Materials 2019, 12, 543 2 of 19
Materials 2018, 11, x FOR PEER REVIEW 2 of 19
Materials 2018, 11, x FOR PEER REVIEW 2 of 19
Chaboche [8–10] proposed a more general cyclic plasticity constitutive model on the basis of
Chaboche [8–10] proposed a more general cyclic plasticity constitutive model on the basis of
Armstrong
Chabocheand [8–10]
Frederick (A-F model)
proposed a more [11], which can better
general reflect the cyclic model
hardening and cyclic
Armstrong and Frederick (A-F model) [11], whichcyclic plasticity
can better reflectconstitutive
the cyclic hardening on the
andbasis
cyclicof
softening
Armstrong behaviors of
and Frederick materials
(A-Funder under
model) proportional
[11], whichloading. loading.
can better The experiments on various metals
softening behaviors of materials proportional Thereflect the cyclic
experiments hardening
on various and from
metals cyclic
from Phillips
softening [12], Wuof[13],
behaviors and Khan
materials [14]proportional
under showed distortion loading. of The
the yield surface characterized
experiments on various by a
metals
Phillips [12], Wu [13], and Khan [14] showed distortion of the yield surface characterized by a region of
region
from of high[12],
Phillips curvature
Wu along
[13], and the
Khan direction
[14] of loading
showed and aofregion
distortion the of flattening
yield surface on the opposite
characterized by a
high curvature along the direction of loading and a region of flattening on the opposite side. Moreover,
side.
regionMoreover,
of high the distortion of the subsequent yield surface varies for different materials. Hu
the distortion of curvature alongyield
the subsequent the direction
surface variesof loading and a region
for different of flattening
materials. Hu Guijuan on the
andopposite
Zhang
Guijuan and Zhang
side.[15–17]
Moreover, Keshi [15–17]
the distortion of the systematically
the subsequent studied
yield the evolution
surface varies for behaviors
different of the yield
materials. Hu
Keshi systematically studied evolution behaviors of the yield surfaces for copper, steel, and
surfaces for
Guijuan and copper, steel, and aluminum under tension-shear loading. It was shown that the shape
aluminum underZhang Keshi [15–17]
tension-shear loading.systematically
It was shown that studied the evolution
the shape behaviors
of the measured yield ofsurface
the yield
is
of the measured
surfaces for yieldsteel,
copper, surface
and isaluminum
related tounderthe offset strain of the
tension-shear yield point.
loading. It was The
shownsubsequent
that the yield
shape
related to the offset strain of the yield point. The subsequent yield surface defined by a small offset
surface defined byyield
of thedevelops
measured a small offset
surface strain develops
is related an obvious “sharp point”. The shape of the yield
strain an obvious “sharp point”.toThe the shape
offset strain
of the of thesurface
yield yield point. The
is close tosubsequent yield
the cylindrical
surface is
surfacewith close
defined to the cylindrical
by a small offset surface
strain with increases
develops of
an obvious offset strain, which is consistent with the
surface increases of offset strain, which is consistent with “sharp
the circlepoint”.
yieldThe shape
surface of the yield
described by
circle yield
surface surface
is close described
to the by Mises
cylindrical yield
surface withcriterion.
increases Typical distortional subsequent yield surfaces
Mises yield criterion. Typical distortional subsequent yieldofsurfaces
offset strain,
underwhich is consistent
pretension with the
and pretorsion
under
circle pretension
yield surfaceand pretorsion
described by loadings
Mises arecriterion.
yield shown inTypical
Figuresdistortional
1 and 2 [17]subsequent
for 45 steel.yield surfaces
loadings are shown in Figures 1 and 2 [17] for 45 steel.
under pretension and pretorsion 500 loadings are shown in Figures 1 and 2 [17] for 45 steel.
500
400 -4
Σstiffness
offset =2x10
-3 Σstiffness
offset =6x10
400 -4
300 Σstiffness
offset =2x10
-3 Σstiffness
offset =6x10
300
200
200
100
τ /MPa

100 Test points

/MPa

0
Symmetric points
Test points
√3τ √3

-1000
Symmetric points
-100
-200
Initial
-200
-300
Initial
-300
-400
-400 Unloading to zero
-500
Unloading to elastic
-500 Unloading
-400 -300 -200 -100 0 100
to zero 200 300
Unloading 400 500
to elastic
-500
σ /MPa
-500 -400 -300 -200 -100 0 100 200 300 400 500
σ /MPa
Figure 1. Schematics of subsequent yield surface of 45 steel on the stress plane under unloading to
Figure
Figure1.
elastic 1.Schematics
range ofofsubsequent
after preloading
Schematics along yield
subsequent yieldsurface
tension ofof45[17].
direction
surface 45steel
steelononthe
thestress
stressplane
planeunder
underunloading
unloadingtoto
elastic range after preloading along tension direction [17].
elastic range after preloading along tension direction [17].
500

500 Σstiffness -3 -4
400 offset =2x10 Σstiffness
offset =2x10
stiffness -3
400 Σoffset =2x10
300 Σstiffness=2x10-4
offset
300
200 Symmetric Test
points points
200 Symmetric Test
√3τ /MPa

100
points points

√3τ
/MPa

100
0

-1000

-100
-200

-200
-300 Initial
-300
-400 Initial
Unloading to elastic Unloading to zero
-400
-500
-500 Unloading
-400 -300 -200 -100 0 100 Unloading
to elastic 200 300 to
400 500
zero
-500 σ /MPa
-500 -400 -300 -200 -100 0 100 200 300 400 500
Figure 2. Schematics of subsequent yield surface of σ /MPa
45 steel on the stress plane under unloading to
Figure 2. Schematics of subsequent yield surface of 45 steel on the stress plane under unloading to
elastic range after preloading along torsion direction [17].
elastic
Figure 2. Schematics of subsequent yield surface of [17].
range after preloading along torsion direction 45 steel on the stress plane under unloading to
elastic range after preloading along torsion direction [17].
Since the initial and subsequent yield surfaces described by the von Mises theory are cylindrical
surfaces in the
Since principal stress
initial and space, theyield
subsequent traditional models
surfaces cannot
described by describe the cross
the von Mises effect
theory are (the sharp
cylindrical
surfaces in principal stress space, the traditional models cannot describe the cross effect (the sharp
Materials 2019, 12, 543 3 of 19

Since the initial and subsequent yield surfaces described by the von Mises theory are cylindrical
surfaces in principal stress space, the traditional models cannot describe the cross effect (the sharp
point in pre-deformation direction and the relatively flat bottom at its opposite direction) of a yield
surface accurately. Many researchers have studied the anisotropic evolution of a yield surface
by the macroscopic method [18–21] and the mesoscopic method [22,23], respectively. In classical
plasticity theory, the anisotropy is reflected by the translation and distortion of the yield surface.
Francois [24] modified the classical von Mises yield criterion by replacing the usual stress deviator
with distortional stress tensor Sd and described the yield surface as an “egg shape” to reduce the
difference between simulation and experiment. Shi Baodong et al. [25] improved the theoretical model
by supplementing the distortion factor with the classical plastic constitutive theory. Fu Qiang et al. [26]
established a slip element model to simulate the subsequent yield surface evolution based on the
mechanism of slip deformation. Chen Yuan et al. [27] proposed the mathematical models for
various deformation mechanisms of slip, twinning, and detwinning to capture the Bauschinger
effect under cyclic loading. Wen et al. [28] used the self-consistent model to determine the yield
characteristics of polycrystalline materials under different loading conditions. The crystal plasticity
method is powerful enough to simulate the inhomogeneous deformation and anisotropy induced by
pre-deformation [16,29–31]. The crystal plasticity constitutive model follows continuum mechanics
and considers the slip mechanisms at grain-level by introducing resolved shear stresses along different
slip orientations and the nonlinear hardening of back stress. Zhang Keshi et al. [16,17] adopted the
Voronoi polycrystalline representative volume element to compute the subsequent yield surfaces of
pure copper and aluminum under different proportional pre-loadings and tension-compression cyclic
loading. The results showed that the anisotropic distortion of the yield surface can be characterized
with the polycrystalline crystal plastic model. Due to the higher computational cost of polycrystalline
analysis of subsequent yield behavior of metals, it is difficult to apply this approach directly in
engineering structural analysis. A constitutive model presenting the anisotropic evolution of yield
surface under the hypotheses of continuous and uniform deformation will be more accurate and
efficient in engineering application.
The anisotropic distortion of the subsequent yield surface has an obvious difference in metals,
including different work hardening aluminum alloy [18,32]. Taking the subsequent yield of T2 pure
copper behavior into account, a macroscopic constitutive model reflecting the distortional yield surface
induced by plastic deformation is put forward under the hypothesis of continuous uniform in material.
The main research work includes the following: (1) a distortional hardening multiplier in conjunction
with an anisotropic degradation factor are introduced into the isotropic hardening term of the yield
function, and an anisotropic distortional yield surface model considering cross effect is proposed;
(2) the yield surfaces simulated by the anisotropic plastic constitutive model are compared with the
experimental results for T2 pure copper under pre-deformation, the simulations of the Chaboche
cyclic plastic model and the crystal plasticity model. The model parameters are calibrated and the
reasonability of the anisotropic model is verified.

2. Theoretical Model
The strain tensor can be decomposed into elastic εe and plastic ε p parts under a small strain
assumption as:
ε = εe + ε p . (1)

According to Hooke’s law, the relationship of stress and strain is expressed as:

σ = D : εe , (2)

where the fourth-order tensor D represents the material stiffness. The plastic constitutive model
concentrates on the calculation of plastic deformation, and the incremental theory considers the
Materials 2019, 12, 543 4 of 19

.p
influence of loading history on present deformation. The plastic strain rate ε of classical plastic theory
in regards to rate-independent model is given by:
r
.p 3. 3 . ∂Fy
ε = pn = p , (3)
2 2 ∂σ
.
where p is the accumulated equivalent plastic strain rate, Fy is the loading potential function, and a
second-order tensor n is the plastic flow direction, which is calculated by the partial differential of the
loading potential function according to Equation (22).
. .p
For a rate-dependent model, the equivalent plastic strain rate p and plastic strain rate ε can be
expressed as:  m
. Fy
p= , (4)
K

3 Fy m ∂Fy
r  
.p 3.
ε = pn = · , (5)
2 2 K ∂σ
where the symbol < > is the MacCauley bracket, K, n are the parameters reflecting the material viscosity.
Fy is the loading potential function, which is same as the yield function on the premise of the associative
flow. The second-order tensor above n is the plastic flow direction, as noted above.
The yield function is the foundation in plastic hardening and flow analysis, thus a proper yield
function should be developed and chosen to characterize most features of the subsequent yield surface
due to the diversity yield behavior of the numerous materials. Based on the classical von Mises
criterion, the yield function of mixed hardening mode is defined by:

f = ks − αk − R − σy = 0, (6)

where s, α are the deviatoric stress and back stress, respectively, σy represents the yield stress
corresponding to zero plastic strain, and R is the isotropic hardening function. For mixed hardening,
the subsequent yield surfaces evolve as the expansion (i.e., isotropic hardening) and translation
(i.e., kinematic hardening due to the evolution of back-stress) constitutive model.
The most common kinematic rule is the linear one due to Prager [3], given as:

2
dα = Cdε p . (7)
3
A practical way to describe a nonlinear kinematic hardening—giving rise to a correct modelization
of cyclic loops—consists of introducing an evanescent memory of the plastic strain path, as initially
proposed by Armstrong and Frederick [4]. This nonlinear kinematic effect appears in the internal
stress equation:  
2 p
dα = C adε − αdp , (8)
3
where the first term corresponds to the Prager’s linear rule and the second term to the evanescent
strain memory, which will gradually weaken and disappear as the cumulative plastic strain increases;
the cumulated plastic length is defined, for example, by:
r
2 p
dp = dε : dε p . (9)
3

Chaboche et al. [8] proposed a very good continuous description of cyclic loading of materials,
obeying a combination of isotropic and nonlinear kinematic rules by introducing many strain memory
items. The back stress is decomposed into several individual variables, and the general form is:
Materials 2019, 12, 543 5 of 19

M  
. 2 .p .
α= ∑ αi , αi = Ci a ε − αi p ,
3 i
(10)
i =1

where M denotes the number of back stress components. Considering the description ability and
conciseness of the model, M = 2 is adopted in this paper. ai and Ci are the material constants
responsible for the evolution of the back stress.
The isotropic hardening function R of Equation (6) is given as:

R = Q[1 − exp(−b · p)], (11)

where Q is used to describe the saturation value of an isotropic hardening function, b is a parameter
describing the relationship between yield surface radius and plastic strain, and p is the accumulated
equivalent plastic strain.
Subjected to the initial conditions:

αi (0) = 0, p(0) = 0, R(0) = 0. (12)

Vincent [19] established a general cyclic plastic model by introducing 25 distortional variables
to replace the classical back stress components. Considering the rotation of principal stress axes in
the loading direction, this model is suitable for exhibiting the ratcheting behavior under multi-axial
loading. Based on the contribution of Feigenbaum [20], the distortional hardening model of metal
was proposed by implementing the fourth-order anisotropic evolution tensor into the classical Hill
orthogonal anisotropic yield criterion in thermodynamic framework. The yield function is defined as:

s−α
   
3
f = (s − α ) : I+ : α A : (s − α) − k2 = 0, (13)
2 ks − α k

where s, α are deviatoric stress and back stress of yield function, respectively, I is the fourth-order unit
stress deviatoric tensor, and A is the fourth-order anisotropic distortional hardening tensor.
Francois [24] replaced the deviatoric stress tensor with the distortional stress tensor in the yield
function of the classical mixed hardening mode, and the distortional yield function can be defined as:

f = ksd (s, α, R) − αk − R − σy = 0, (14)

where s, α are deviatoric stress and back stress of yield function, respectively, σy represents the yield
stress corresponding to zero plastic strain, and R is the isotropic hardening function. sd is the distortion
stress tensor calculated by deviatoric stress s, back stress α, and isotropic hardening function R.
Introducing the anisotropic distortional function H into the yield function of classical mixed
hardening model by combining the isotropic hardening and the kinematic hardening, which
characterize, respectively, the expansion and translation of subsequent yield surface, the anisotropic
yield function can be expressed as:

Fy = ks − αk − (σ0 + R) H (w, d), (15)

where s, α are deviatoric stress and back stress of yield function. σ0 represents the yield stress
corresponding to zero plastic strain. H is the anisotropic distortional function determined by
anisotropic distortional factor d and the anisotropic degradation factor w. In this presentation,
the nonlinear kinematic hardening rate of yield surface, which determines the evolution of back
stress (center of yield surface), is described as Equation (10), and the isotropic hardening of yield
surface R is given by Equation (11). The values for the corresponding coefficients ai and Ci are indicated
in Table 1.
Materials 2019, 12, 543 6 of 19

Table 1. Material parameters of anisotropic plastic model for T2 copper.

Elastic Kinematic Hardening Anisotropic Hardening

Isotropic Hardening Parameters
Constants Parameters Parameters
E ν b Q σ0 K m a1 C1 a2 C2 β1 β2 w
unit GPa MPa MPa MPa MPa MPa
Anisotropic
97 0.326 15 50 28 8 100 31 3000 51 7 0.3 0.1 4
model
Chaboche
97 0.326 15 50 28 8 100 31 3000 51 7 0 0 0
model

The anisotropic distortional function H in Equation (15) can be summarized as:



 1, Preloading and unloading
kα−α k kα−αunloading k

d + (1 − d) w·kα unloadingk ,

H (d, w) = Reloading and w·kαunloading k
≤1 , (16)
unloading
 kα−αunloading k
≥1

 1,
 Reloading and w·kαunloading k

where α is the back stress during reloading, and it is constant when the stress state is within the yield
surface. The material yields with the increase of reloading load and back stress α evolves with the
following Equation (10). The fourth-order tensor αunloading is the back stress corresponding to the
unloading stress followed by preloading in subsequent yield surface test, which is related to preloading
path and preloading load. αunloading remains the constant during elastic deformation.


 0, Preloading

 α
unloading , Unloading
αunloading = , (17)
 αunloading ,
 Reloading without unloading

αnew unloading , New unloading


where w is the anisotropic degradation factor, which describes the evolution from anisotropy to
quasi-isotropy with regard to yield function. It can be calibrated by the evolution test data of yield
surface under different offset strains. If the anisotropic degradation factor w is very big, the subsequent
yield surface presents only the proportional expansion of the initial anisotropic yield surface, i.e., the
anisotropy of yield keeps stable. On the other hand, the initial anisotropic yield surface evolves
toward the isotropic yield surface when the anisotropic degradation factor w is a smaller value, i.e., the
anisotropy of yield degrades. An appropriate anisotropic degradation factor w is attempted based on
the comparison of the evolution progress between the subsequent yield surface simulated and the
experimental results for different given offset strains (confer the data in Figure 3). The anisotropic
distortional function H varies from d to 1 with the increase of reloading load. When the back
stress reaches a certain critical value, which means kα − αunloading k = w · kαunloading k and H = 1,
the anisotropic distortional yield surface evolves into a yield surface without distortion. √
The shape of subsequent yield surface under reloading tends to be circular in σ − 3τ stress
space with the increase of the offset strain [15,17]. The subsequent yield surfaces of T2 pure copper
under different offset strains [16] are shown in Figure 3. The results show that the anisotropy of the
subsequent yield surfaces degrades gradually, and the subsequent yield surfaces approach a circle.
α − α unloading = w ⋅ α unloading and H = 1 , the anisotropic distortional yield surface evolves into a yield
surface without distortion.
The shape of subsequent yield surface under reloading tends to be circular in σ − 3τ stress
space with the increase of the offset strain [15,17]. The subsequent yield surfaces of T2 pure copper
Materials 2019, 12, 543 7 of 19
under different offset strains [16] are shown in Figure 3. The results show that the anisotropy of the
subsequent yield surfaces degrades gradually, and the subsequent yield surfaces approach a circle.
120 Experiment
Pre-ten-strain 1% residual
Εoffset =
80 1x10-4
5x10-4
1.0x10-3
40

0 O1

-40

-80 Circle

-120
-80 -40 0 40 80 120 160
σ /MPa
√
Figure3.
Figure Theexperimental
3. The experimentalsubsequent
subsequentyield
yieldsurfaces
surfacesof
of T2
T2 copper
copper on the σσ −− 33τ
on the τ stress
stressplane
planewith
with
1% pretension equivalent strain and unloading to 61.8 MPa
1% pretension equivalent strain and unloading to 61.8MPa [16].[16].

Noting the cross effect and anisotropic characteristics of metal subsequent yield surface test in
Figure 3, the anisotropic distortional factor d in anisotropic distortional function H can be given as:
h ih i
d = 1 − β 1 + β 1 (cosθ )3 1 + β 2 (sinθ )2 , (18)

where β 1 , β 2 are the parameters of the cross effect, which can be calibrated by the characteristic of
distortional subsequent yield surface corresponding to the small offset strain under tension and torsion
loading (cf. Section 3.4). Anisotropic distortional yield surface degenerates into the circular mixed
hardening yield surface without cross effect in the condition of β 1 = β 2 = 0. θ is the angle between
the preloading direction and the reloading direction, and it is defined by the deviatoric stress and
back stress:
s−α
 
α
θ = arccos · . (19)
ks − α k k α k
Defining the two flow tensors as:

s−α α
n1 = , n2 = . (20)
ks − α k kαk

Substituting Equation (20) into (19), Equation (19) can be simplified as:

θ = arccos(n1 · n2 ). (21)

Following the orthogonal flow rule, the plastic flow direction of anisotropic yield surface can be
described by the partial derivative of Equation (15):

∂Fy s−α ∂H
n= = − (σ0 + R) . (22)
∂σ ks − α k ∂σ

Because H is 1 and ∂H
∂σ = 0 under preloading and unloading [given in Equation (16)], the plastic
flow direction is orthogonal to the cylindrical yield surface during the two stages, i.e., n = n1 .
Similarly, ∂H
∂σ = 0 and n = n1 are established when anisotropy disappears in the process of reloading,
i.e., kα − αunloading k ≥ w · kαunloading k. The yield surface evolves from anisotropy to quasi-isotropy
gradually in the condition of kα − αunloading k ≥ w · kαunloading k under reloading, as displayed in
Figure 3. Combining Equations (16), (20), and (22), the plastic flow direction n can be expressed as:
!
kα − αunloading k ∂d
n = n1 − (σ0 + R) 1 − . (23)
w · kαunloading k ∂σ
Materials 2019, 12, 543 8 of 19

∂d
Based on the definition of anisotropic distortional factor d in Equation (18), ∂σ can be obtained by:

∂d ∂d   ∂(cosθ)
= = 3β 1 (1 + β 2 )cos2 θ − 2β 2 cosθ − 5β 1 β 2 cos4 θ . (24)
∂σ ∂ (s − a) ∂ (s − a)

Due to the definition of n2 in Equation (20), which is irrelevant to s − a, the expression can be
deduced as:
∂n2
= 0. (25)
∂ (s − a)
Therefore,
∂(cosθ) ∂(n1 : n2 ) ∂n1
= = : n2 . (26)
∂ (s − a) ∂ (s − a) ∂ (s − a)
Integrating Equations (23), (24), and (26), the plastic flow direction n can be summarized as :



 n1 , Preloading and unloading

n1 − (σ0 + R) 3β 1 (1 + β 2 )cos2 θ − 2β 2 cosθ − 5β 1 β 2 cos4 θ



n= 
kα−α k

kα−αunloading k , (27)

 · 1 − w·kα unloadingk J : n2 , Reloading and w·kαunloading k
≤1

 unloading
 kα−αunloading k
n Reloading and ≥1

1 w·kαunloading k

where the stress tensors n1 and n2 are normalized according to back stress tensor α and stress tensor
s − α in Equation (20). J is a fourth-order tensor which can be defined as:

I − n1 ⊗ n1
J= . (28)
ks − α k

3. Simulation Procedure, Model Calibration, and Data Processing

3.1. Specimen Geometry and Finite Element Model

The deformation of a thin-walled circular tube specimen, which had a gauge length of 50 mm, a
wall thickness of 1 mm, and an outer diameter of 16.5 mm, was simulated. The full-scale geometric
model was established and divided into 64640 elements with 77952 nodes, as shown in Figure 4. Five
elements were discretized along the direction of wall thickness during the gauge segment. The element
type was C3D8R three-dimensional eight-node reduced integration element, which could avoid shear
locking and improved the computational efficiency. The model was constrained and loaded by
coupling the chuck elements (cf. the red region in Figure 4) with the displacement of the reference
point, as shown in Figure 4. The first, third, fourth, and sixth degrees of freedom in the loading end
were set to zero, and the axial and rotational displacements were performed on the second and fifth
degrees of freedom to realize the combination loading of tension and torsion. All degrees of freedom
at the fixed end were set to zero. Based on the finite element platform ABAQUS, the subroutine of user
defined material (Umat) was applied to simulate the coupling behavior under tension-torsion loading.

3.2. Loading Path for Yield Stress Test

The subsequent yield surfaces of metals were tested by the single-specimen method and the
multiple-specimen method. The yield stress was determined by the offset-strain method, which
required the elastic modulus calculated from the linear segment of stress-strain curve. Then, the stress
corresponding to the specified plastic strain (offset strain) was obtained. The test and the simulation
of polycrystalline plastic model showed that the yield point test of single-specimen method was
affected by the accumulated plastic deformation under different loading paths and was closely related
to the reloading sequence. The dispersion of test data from multiple-specimen method could be
controlled by sufficient specimens even though the properties of the material were scattered between
specimens. Therefore, the test results based on the multiple-specimen method were used to compare
 elements were discretized along the direction of wall thickness during the gaug
element type was C3D8R three-dimensional eight-node reduced integration elemen
avoid shear locking and improved the computational efficiency. The model was c
loaded by coupling the chuck elements (cf. the red region in Figure 4) with the disp
Materials 2019, 12, 543 reference point, as shown in Figure 4. The first, third, fourth, 9and of 19sixth degrees of
loading
Materials 2018, 11, x FOR PEER REVIEW
end were set to zero, and the axial and rotational displacements
9 of 19
were pe
second and fifth degrees of freedom to realize the combination loading of tension
with the simulation results of the anisotropic model in this paper. With the multiple-specimen method,
degrees
Figure 4. The FEM (Finite Elementof freedom at theoffixed
Method) model end were
thin cylinder set toand
specimen zero.
the Based on the finite element plat
experimental
the single specimen was used to determine the different yield stresses on the three offset strains along
tension-torsion strainsthe
acquisition method.of user defined material (Umat) was applied to simulate the cou
subroutine
one reloading direction.
under tension-torsion loading.
3.2. Loading Path for Yield Stress Test
The subsequent yield surfaces of metals were tested by the single-specimen method and the
multiple-specimen method. The yield stress was determined by the offset-strain method, which
required the elastic modulus calculated from the linear segment of stress-strain curve. Then, the
stress corresponding to the specified plastic strain A
(offset
B strain) was obtained. The test and the
simulation of polycrystalline plastic model showed that the yield point test of single-specimen
method was affected by the accumulated plastic deformation under different loading paths and was
closely related to the reloading sequence. The dispersion of test data from multiple-specimen
method could be controlled by sufficient specimens even though the properties of the material were
scattered between specimens. Therefore, the test results based on the multiple-specimen method
were used to compare with the simulation results of Athe anisotropic model in this paper. With the
B
multiple-specimen method, the single specimen was used to determine the different yield stresses
on the three4.offset
Figure The FEMstrains along
(Finite one reloading
Element direction.
Method) model of thin cylinder specimen and the experimental
In order to investigate
tension-torsion the applicability
strains acquisition method. of anisotropic distortional yield surface model under
different preloading paths, the subsequent yield surfaces of thin-walled tube specimens of pure
In order to investigate the applicability of anisotropic distortional yield surface model γ under
different preloading paths, the subsequent yield surfaces of thin-walledεtube specimens of pure 3copper = 1%
copper under three typical pre-deformation conditions (pre-tension = 1% , pre-torsion √ ,
under three typical pre-deformation conditions (pre-tension √
ε = 1%, pre-torsion γ/ 3 = 1%, and
γ 2 2
pre-tension-torsion proportional loading ε = √εγ == 200= ) were predicted.
200 ) were predicted.
after plastic3deformation
3
and pre-tension-torsion
As the unloading process proportional
of metals loading was nonlinear, and the nonlinearity
As the unloading process of metals after plastic
induced by plastic pre-deformation under loading and unloading process deformation waswas nonlinear, and the
more prominent,
nonlinearity
the test data of induced by plastic
elastic stage pre-deformation
in unloading section was under loading
selected. The and unloading
equivalent stressprocess was more
of the unloading
prominent,
point was 61theMPatest(nominal
data of elastic
stress)stage in unloading
corresponding to thesection was selected.
equivalent strain of The
1%equivalent stress of
under preloading.
The displacement boundary and force boundary were used to realize the strain control and under
the unloading point was 61 MPa (nominal stress) corresponding to the equivalent strain of 1% stress
preloading.
control, The displacement
respectively. boundary
The specimens andfirstly
were force subjected
boundary to were
theused to realize
equivalent the strain
strain of 1% control
under
and stress control,
preloading respectively.
and unloaded to theThe specimens
equivalent wereoffirstly
stress 61 MPa subjected
and thento the equivalent
reloaded alongstrain
0, 30,of 1%
60,
under preloading and unloaded to the equivalent stress of 61 MPa and then
90, 105, 120, 135, 150, 165, and 180 angles, as depicted in Figure 5. By adopting the continuation reloaded along 0, 30, 60,
90, 105, [16],
method 120, the
135,measured
150, 165, andyield180 angles,
stresses as depicted
along 0, 30, 60,in 90,Figure 5. By
105, 120, 135,adopting
150, 165,theandcontinuation
180 angles
were symmetrically extended to the opposite side of the pre-deformation line. Consequently,180
method [16], the measured yield stresses along 0, 30, 60, 90, 105, 120, 135, 150, 165, and theangles
yield
were symmetrically
points whose reloading extended to the
paths were opposite
along −30, − side
60, −of90,the−pre-deformation
105, −120, −135, line.−150,Consequently,
and −165 angles the
yield points
could whoseeasily.
be obtained reloading paths were along −30, −60, −90, −105, −120, −135, −150, and −165 angles
could be obtained easily.
3τ Reloading along different directions

C
Pretorsion-unloading:
O→C→O3
O3 B
Pre-tension-torsion-unloading:
O2 O→B→O2

Pretension-unloading:
O→A→O1
σ
O O1 A

Figure 5. Schematic
Figure 5. Schematic ofof probing
probing path
path for
for subsequent
subsequent yield
yield surface
surface under
under pretension,
pretension, pretorsion,
pretorsion, and
and
pre-tension -torsion loading.
pre-tension -torsion loading.

3.3. Data Processing Method

The stress during the gauge segment of a thin wall circular pipe was distributed uniformly
under the tension-torsion loading. The displacements of the two nodes set in the middle of the
Materials 2019, 12, 543 10 of 19

3.3. Data Processing Method

The stress during the gauge segment of a thin wall circular pipe was distributed uniformly under
the tension-torsion loading. The displacements of the two nodes set in the middle of the gauge segment
were simulated with the output of the extensometer, as shown in Figure 4. The Python script file
of ABAQUS was edited to deal with and output the variables. The axial displacement, torsional
displacement, axial force, and torque of the two reference points in each increment were recorded, and
the total strain, plastic strain, direct stress, shear stress, equivalent strain, and equivalent stress were
calculated and output by the state variables from the program. Due to the limitation of increments
in ABAQUS standard simulation, the yield stresses under different offset strains were automatically
obtained using a linear interpolation program by a few results output on different increments. The axial
stress, shear stress, and torsional strain in simulation can be calculated as follows:

4F
σ= , (29)
π ( D 2 − d2 )

16TD
τ= , (30)
π ( D 4 − d4 )
U2A − U2B
ε= , (31)
AB
R UR2A − UR2B


γ= , (32)
AB
where F and T denote axial force and torque, respectively, D is the outer diameter, and d is the internal
diameter of the thin-walled tube specimens of the gauge section. U2A , U2B are the axial displacements of
points A and B, and UR2A , UR2B are the torsional displacements of points A and B, as shown in Figure 4.
R is the mean radius in gauge segment, and AB is corresponding to the gauge length of extensometer.
Based on Equations (29)–(32), the equivalent stress and strain under the von Mises theory can be
defined as: r
γ2
ε eq = ε2 + . (33)
3
Using Hooke’s law, plastic strains can be approximately expressed as:

σ
εp = ε − , (34)
E
τ
γp = γ − , (35)
G
s
p ( γ p )2
ε eq = ( ε p )2 + , (36)
3
where E and G are elastic modulus and shear modulus, respectively.

3.4. Calibration of Model

3.4.1. Calibration of Anisotropic Parameters

It can be analyzed from Equations (18) and (21) that the anisotropic distortional factor d changes
with n1 · n2 . The anisotropic hardening model in this paper is calibrated by the tension-torsion tests.
Assume the stress state of unloading after preloading is σ ∗ and the reloading path is σ − σ ∗ , σ ∗ can be
simplified to O0 under tension-torsion stress state, which is the center of the subsequent yield line (cf.
the blue solid line in Figure 6), and the preloading-unloading path is O → A0 → O0 . The reloading
→
path σ − σ ∗ corresponds to the vector O0 E on two-dimensional stress plane. The angle between
preloading path and reloading path can be denoted as ∠ A0 O0 E, as shown in Figure 6. Observe that
Materials 2018, 11, x FOR PEER REVIEW 11 of 19

the subsequent yield line (cf. the blue solid line in Figure 6), and the preloading-unloading path is

O → A′ → O′ . The reloading path σ − σ * corresponds to the vector O ′E on two-dimensional
Materials 2019, 12,
stress plane. The between preloading path and reloading path can be denoted as ∠A′O11
543 angle ′Eof, as
19

shown in Figure 6. Observe that the angle θ between preloading path and reloading path equals
the angle
angle θbetw een npreloading
between 1 , n2 defined
√ path
byand reloading
Equation (22)path 11 − 3s12the
in sequals angle between
deviatoric n1 , n2 defined
stress space. Therefore,by
β ,β
Equation (21) in s11 − 3s12 deviatoric stress space. Therefore, the cross effect parameters β
the cross effect parameters 1 2 can be calibrated by the characteristic of distortional subsequent
,
1 2β can be
calibrated by the characteristic of distortional subsequent yield surface under tension-torsion loading.
yield surface under tension-torsion loading.
Initial yield surface
Subsequent yield surface
3.5
Distortional surface
3.0 B′ A′
2.5 E d =1
Reloading stage O′E θ
2.0 O′
1− 2β1
1.5 Preloading stage OO′
C′ D′
1.0
B A
0.5
A′C ′ = AC = BD = 2
0.0 O
O′B′ = O′D′
-0.5 = (1− β1 ) ⋅ (1+ β 2 )
-1.0 C D
Isotropic surface
-1.5
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Normal stress σ11 /MPa
√
Figure 6. Schematic of anisotropic distortion subsequent yield surface on σ − 3τ stress plane.
Figure 6. Schematic of anisotropic distortion subsequent yield surface on σ − 3τ stress plane.

If the distortional yield surface exhibits prominent cross effect (the sharp point in pre-deformation
If the
direction anddistortional
the relatively yield surface atexhibits
flat bottom its oppositeprominent
direction),cross effect surface
the yield (the sharp
can bepoint in
roughly
pre-deformation direction and thepreloading
relatively flat bottominatσ its √
symmetric with the proportional direction − opposite
3τ stressdirection),
space. Asthe theyield surface
subsequent
_
can be roughly symmetric with the proportional preloading direction in σ − 3τ stress space. As the
yield surface A0 B0 C 0 D 0 is symmetric with respect to A0 C 0 in Figure 6, O0 B0 = O0 D 0 = (1 − β 1 )(1 + β 2 ).
 A′0Cand
′ in
O
subsequent yield surface A′B′C ′D′ isyield
0 , which is the center of the distortional surface, is with
symmetric the intersection
respect of to B0 D A0 C 0Figure
, therefore
6,
O 0
O′A 0
B′ ==O1, ′ =C(1−=β11 )(
′DO 0 0 2ββ12. )B and
−1+ 0 D 0 are the points of maximum distance from the symmetry axis.
. O′ , which is the center of the distortional yield surface, is the
The anisotropic model parameters β 1 , β 2 can be calibrated by the measured yield surface as:
′ ′
intersection of B ′D ′ and A′C ′ , therefore O ′A′ = 1 , O C = 1− 2 β1 . B′ and D′ are the points of
0 0 0 0 B0 D 0
maximum distance from the symmetry β1 =
O A −O C
axis. The, β =
anisotropic − 1model
. β ,β
parameters 1 2 can(37) be
2
2 · O0 A0 A0 C 0
calibrated by the measured yield surface as:
The cross effect parameters β 1 , β 2 of distortional yield surface can describe the characteristics
O′A′ − O′C ′ B′D ′ （38
of anisotropic yield. The directionalβdistortion 1 = β 2 = effect−1
and,cross of subsequent
. yield surface are more
2 ⋅ O′A′ A′C ′ ）
distinct as β 1 increases. In particular, the yield surface expands in the direction perpendicular to the
preloading path as β 2 > 1, the yield β ,surface
β shrinks in the direction perpendicular to the preloading
The cross effect parameters 1 2 of distortional yield surface can describe the characteristics
path as β 2 < 1, and the inner diameter of the yield surface keeps constant in the direction perpendicular
of anisotropic yield. The directional distortion and cross effect of subsequent yield surface are more
to the preloading path as β 2 = 1.
β
distinct as 1 increases. In particular, the yield surface expands in the direction perpendicular to the
3.4.2. A Representative
preloading path as 2
β >Volume
1 Element for the Calibration of Parameters in Anisotropic
, the yield surface shrinks in the direction perpendicular to the preloading
Constitutive Model
path as 2
β < 1 , and the inner diameter of the yield surface keeps constant in the direction
The above anisotropic constitutive model proposed was implemented as a user defined material
perpendicular to the into
preloading path as 2
β =1.
subroutine (UMAT) the ABAQUS/Standard module. In order to rapidly determine the model
parameters, a representative volume element (1 mm × 1 mm × 1 mm) was adopted, as shown in
Figure 7. The external surface of the representative element kept plane after deformation due to the
specific boundary condition. By fitting the computationally obtained cyclic stress-strain curve to the
experimentally measured one, the material parameters of anisotropic plastic model for T2 copper
under monotonic loading and cyclic loading are estimated as Table 1.
model parameters, a representative volume element (1 mm × 1 mm × 1 mm) was adopted, as shown
in Figure 7. The external surface of the representative element kept plane after deformation due to
the specific boundary condition. By fitting the computationally obtained cyclic stress-strain curve to
the experimentally measured one, the material parameters of anisotropic plastic model for T2 copper
Materials 2019, 12, 543 12 of 19
under monotonic loading and cyclic loading are estimated as Table 1.

Figure 7. Representative volume element for the simulation with unique element.
Figure 7. Representative volume element for the simulation with unique element.
To validate the precision of the anisotropic plastic model, the simulated stress-strain curve of T2
pure copperTableunder uniaxialparameters
1. Material tension and ofthe hysteresis
anisotropic loop model
plastic under for
uniaxial tension-compress with
T2 copper.
symmetric strain amplitudes were compared with the experimentally measured ones in Figures 8
Elastic
and 9, respectively. The resultshardening
indicate that Kinematic
the macroscopic Hardening
mechanical Anisotropic
behaviors Hardening
of the materials
Isotropic parameters
constants Parameters
under monotonic loading and cyclic loading simulated by the anisotropic constitutive model wereParameters
in good agreement
ν bwith the σ0
Q experimental results. mThe equivalent
a1 C1 stress-strain
a2 C2 curves
β1 obtained
β2 wby
Ε K
the uniaxial tensile test and the torsion test with thin-walled circular pipe specimen were different,
unit as indicated
GPa in Figure 8. All
MPamodelMPaparameters
MPain this paper
MPa were calibrated
MPa by the pre-tension test.
The equivalent stress-strain curve simulated by the anisotropic plastic model was fully consistent with
Anisotrop
the experimental
97 0.326 one15under50 tension, but
28 the equivalent
8 100 stress-strain
31 curve51
3000 simulated
7 by0.3 the anisotropic
0.1 4
ic modelplastic model was slightly higher than the experimental data under torsion of thin-walled circular
Chaboche Materials 2018, 11, x FOR PEER REVIEW
pipe specimen.
97 0.326 15 50 28 8 100 31 3000 51 7 0 0 0
model
150
140 Strain amplitude 0.3%
To validate the precision of the anisotropic plastic model, the simulated stress-strain curve of T2
120 100
pure copper under uniaxial tension and the hysteresis loop under uniaxial tension-compress with
Equivalent stress /MPa

symmetric strain amplitudes were100

compared with the experimentally measured ones in Figure 8
50
Stress /MPa
and Figure 9, respectively. The results indicate that the macroscopic mechanical behaviors of the
materials under monotonic loading and cyclic loading simulated by the anisotropic constitutive
80
0
model were in good agreement with the experimental results. The equivalent stress-strain curves
60
obtained by the uniaxial tensile test and the torsion test with thin-walled circular pipe specimen
-50
were different, as indicated in Figure
40 8. All model parameters in this paper were calibrated by the
pre-tension test. The equivalent stress-strain curveExperimental
simulated by uniaxial tension
the anisotropic plastic model was
20
fully consistent with the experimental one underExperimental
tension, butpure torsion
the -100
equivalent stress-strain curve
Uniaxial tension simulated
simulated by the anisotropic plastic model was slightly higher than the experimental data under
0
torsion of thin-walled circular pipe -150
0.00specimen.
0.01 0.02 0.03 0.04 0.05
-0.004 -0.003 -0.002 -0.001 0.000
Equivalent strain
Strain

Figure 8. The stress-strain curves of T2 pure copper under uniaxial tension and pure torsion by test [33]
and simulation with the anisotropic
Figure 8.model.The stress-strain curves of T2 Figure 9. Simulation for t
pure copper under uniaxial tension and hysteresis loop [33] of u
pure torsion by test [33] and simulation compression with the aniso
with the anisotropic model.

4. Simulation of Anisotropic Subsequent Yield Surface

The subsequent yield surface of T2 pure copper under pretension loading e
anisotropy [34], as shown in Figure 10. According to the test method describ
initial yield surface with the offset strain of 5 × 10−5 was nearly circular when t
firstly under pretension loading [cf. the point A in Figure 10(a,b)]. T2 pure copp
 Materials 2018, 11, x FOR PEER REVIEW 13 of 19
Materials 2019, 12, 543 13 of 19
150
x FOR PEER REVIEW 140 13 of
Strain amplitude 19
0.3%

120 150 100

Strain amplitude 0.3%
Equivalent stress /MPa

100
50

Stress /MPa
100
80
0

Stress /MPa
50
60

-50
40
0
Experimental uniaxial tension
20 Experimental pure torsion -100
-50 tension simulated
Experiment
Uniaxial Anisotropy simulation
0
Experimental uniaxial -150
0.00 tension
0.01 0.02 0.03 0.04 0.05
-0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004
Experimental pure torsion Equivalent
-100 strain
Experiment Strain
Uniaxial tension simulated Anisotropy simulation
-150
01 0.02 0.03 0.04 0.05
-0.004 -0.003 -0.002 -0.001 0.000 0.001 0.002 0.003 0.004
Equivalent strain
Figure 8. The stress-strain curves of T2 Figure 9. Simulation for the experimental
Strain
pure copper under uniaxial tension and hysteresis loop [33] of uniaxial tension-
pure torsion by test [33] and simulation compression with the anisotropic model.
with the anisotropic model.
Figure
The stress-strain curves Simulation for the
of9.T2 experimental
Figure hysteresis
9. Simulation forloop
the[33] of uniaxial tension- compression with
experimental
the anisotropic
per under uniaxial tension and model. hysteresis loop [33] of uniaxial tension-
on by test [33] and simulationof Anisotropiccompression
4. Simulation Subsequentwith Yield the anisotropic model.
Surface
nisotropic model. 4. Simulation of Anisotropic Subsequent Yield Surface
The subsequent yield surface of T2 pure copper under pretension loading exhibited significant
The subsequent
anisotropy [34], as shown yield surface10.
in Figure of According
T2 pure copperto theunder pretension
test method loading
described exhibited
in Section significant
3, the initial
anisotropy [34], as shown in Figure 10. According
− 5 to the test method
yield surface with the offset strain of 5 × 10 was nearly circular when the material yielded firstly described in Section 3, the
initialpretension
under yield surface with[cf.
loading thetheoffset
pointstrain of 5 × 10
A in Figure
−5 wasT2
10a,b]. nearly
purecircular when to
copper began theharden
material
when yielded
the
of Anisotropic Subsequent
firstly under Yield Surface
pretension loading [cf. the point A in Figure 10(a,b)]. T2 pure copper began to harden
equivalent pre-tensile strains increased to 1% and 5% [cf. the point B and point C in Figure 10a,b],
when
and the the equivalent
subsequent pre-tensile
yield strains
the increased
surfacepretension
[cf. blueloading to 1%
triangle andandred5% [cf. the point
rectangular B and point
in Figure 10a] wasC inlarger
Figure
equent yield surface of T2 pure copper under exhibited significant
10(a,b)], and the subsequent yield surface [cf. the blue triangle and red rectangular in Figure 10(a)]
4], as shown in than
Figurethe 10.
initial yield surface
According to the [cf.test
the method
black circle in Figurein
described 10a]. The subsequent
Section 3, the yield surface was
was larger
characterized than the
by 5a regioninitial yield
of high surface
curvature [cf.
along the black circle
the direction in Figure
of preloading 10(a)]. The subsequent
and a region of flattening yield
urface with the offset strain of × 10 −5 was nearly circular when the material yielded
onsurface was characterized
the opposite side. The yield bysurface
a region of high by
calculated curvature along the
the anisotropic modeldirection
in this of preloading
paper and a
[cf. the blue,
pretension loadingregion
[cf. the point A inonFigure 10(a,b)]. T2 The
pureyieldcopper began to harden
red, and black curves in Figure 10a] were in good agreement with the experimental results, whichin
of flattening the opposite side. surface calculated by the anisotropic model
valent pre-tensile this
strains
paperincreased to 1%red, andand5% [cf. thecurves
pointin B Figure
and point C in Figure
indicates that[cf.
thisthe blue,
model can describe black
the anisotropic 10(a)]
distortional were in good agreement
characteristics with the
of the subsequent
he subsequent yield surface
experimental
yield
[cf. the blue triangle and red rectangular in Figure
surface. results, which indicates that this model can describe the anisotropic distortional
10(a)]
an the initial yieldcharacteristics
surface [cf. the of theblack circle in
subsequent Figure
yield 10(a)]. The subsequent yield
surface.
haracterized by a region
100 of high curvature along the direction of preloading and a
Experimental initial yield surface
ening on the opposite side.
75 The yield surface calculated by the anisotropic
Experimental subsequent yield surface of pretension=1%
Experimental subsequent yield surface of pretension=5%
140 model in
the blue, red, and black
50
curves in Figure 10(a)] were in good agreement
120 with the C
results, which indicates that this model can describe the anisotropic 100 distortional
25 B
Tensile stress /MPa

of the subsequent yield surface. 80

0
O A B C 60
erimental initial yield surface
-25
erimental subsequent yield surface of pretension=1% 140
erimental subsequent yield surface of pretension=5% 40
A
-50 120 20 C
Simulated initial yield surface
-75 100
Simulated subsequent yield surface of pretension=1%
0
O
Tensile stress /MPa

Simulated subsequent yield surface of pretension=5% B

-100 80 -20
-0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
-50 -25 0 25 50 75 100 125 150
O A B C Tensile strain
σ 60
/ MPa
(a)
40 (b)
A
Figure
Figure 10.10.(a)(a)The
Theexperimental
experimental
20 yieldsurfaces
yield surfaceswith
withdifferent
differentpretension
pretensionstrains
strainsofof1%1%and
and5%
5%[34]
[34]
mulated initial yield surface versus
versus thethe results
results simulated;
simulated; (b)(b)
thethe stress-strain
stress-strain relationship
relationship corresponds
corresponds to to
thethe path
path O-A-B-C.
O-A-B-C.
0
mulated subsequent yield surface of pretension=1%
mulated subsequent yield surface of pretension=5% O
-20
-0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
0 25 50 75 100 125 150
/ MPa Tensile strain

(a) (b)
Materials 2018, 11, x FOR PEER REVIEW 14 of 19
Materials 2019, 12, 543 14 of 19

The simulation results of the anisotropic distortional yield surface model were compared with
the experimental
The simulationdataresults
and the simulation
of the anisotropic results of polycrystalline
distortional yield surfaceplastic
model model [16] to discuss
were compared with the
subsequent yield data
the experimental and and
hardening process
the simulation of of
results metal materials.plastic
polycrystalline In particular,
model [16]the yield the
to discuss stresses
determined
subsequentby theand
yield very small offset
hardening strain
process showed
of metal significant
materials. dispersion,
In particular, which
the yield gradually
stresses decreased
determined
asby
thetheoffset
very small
strainoffset strain showed
increased. Thus the significant
typical dispersion,
offset strains which
of 1gradually
× 10−4, 5decreased
× 10−4, andas the
1 ×offset
10−3 were
strain increased. Thus the typical offset strains of 1 × 10 −4 , 5 × 10−4 , and 1 × 10−3 were adopted to
adopted to investigate the subsequent yield behavior.
investigate the subsequent
The subsequent yield behavior.
yield surfaces determined by unloading and proportional reloading after three
The subsequent yield surfaces
directions of pre-deformations (tensile, determined by unloading
torsion, and proportional
and tension-torsion) reloading
are shown in after three11–13
Figures
directions of pre-deformations (tensile, torsion, and tension-torsion) are shown in Figures 11–13
respectively, which includes yield stresses corresponding to the typical offset strains of 1−4× 10−4
respectively, which includes yield stresses corresponding to the typical offset strains of 1 × 10
[Figure 11(a), Figure 12(a) and Figure 13(a)], 5 × 10−4 [Figure 11(b), Figure 12(b) and Figure 13(b)], 1 ×
[Figures 11a, 12a and 13a], 5 × 10−4 [Figures 11b, 12b and 13b], 1 × 10−3 [Figures 11c, 12c and 13c].
10−3 [Figure 11(c), Figure 12(c) and Figure 13(c)]. Since the classical Chaboche plastic constitutive
Since the classical Chaboche plastic constitutive model only includes kinematic hardening and isotropic
model only the
hardening, includes kinematic
subsequent hardening
yield surface and by
calculated isotropic
Chaboche hardening, the distortional
model without subsequenthardening
yield surface
calculated by Chaboche
is a circle with the center model without
of back stress, distortional
as shown hardening
in Figures 11–13. is a circle with the center of back
stress, as shown in Figures 11–13.
Anisotropy simulation -4 Anisotropy simulation
Εresidual -4 Εresidual
offset =5x10
offset =1x10 Chaboche simulation Chaboche simulation
[16]
Experiment[16] Experiment
Crystal plasticity RVE[16] Crystal plasticity RVE[16]

100 11(a) 100 11(b)

80 80
60 60
40 40
20 20
0 0
-20 -20
-40 -40
-60 -60
-80
-80
Pre-ten-strain 1% Unloading to 61.8MPa
-100 Pre-ten-strain 1% Unloading to 61.8MPa
-100
-100 -80 -60 -40 -20 0 20 40 60 80 100 120
-100 -80 -60 -40 -20 0 20 40 60 80 100 120
σ /MPa
σ /MPa
-3 Anisotropy simulation
Εresidual
offset =1x10 Chaboche simulation
[16]
Experiment Crystal plasticity RVE[15]

100 11(c)
80
60
40
20
0
-20
-40
-60
-80
-100 Pre-ten-strain 1% Unloading to 61.8MPa

-100 -80 -60 -40 -20 0 20 40 60 80 100 120

σ /MPa
√
Figure11.
Figure The simulations
11. The simulationsand experiments
and of subsequent
experiments yield surfaces
of subsequent on the σ −
yield surfaces on 3τ σ − plane
thestress 3τ stress
with preloading of 1% proportional pretension equivalent strain and unloading to equivalent stress
plane with preloading of 1% proportional−4pretension equivalent strain −and unloading to equivalent
of 61.8 MPa: (a) the offset strain is 1 × 10 ; (b) the offset strain is 5 × 10 4 ; (c) the offset strain is
stress of−61.8 MPa: (a) the offset strain is 1 × 10 ; (b) the offset strain is 5 × 10 ; (c) the offset strain is 1
−4 −4
1 × 10 3 .
× 10−3.
 Materials 2019, 12, 543 15 of 19
Materials 2018, 11, x FOR PEER REVIEW 15 of 19
Materials 2018, 11, x FOR PEER REVIEW 15 of 19
Anisotropy simulation -4 Anisotropy simulation
Εresidual -4 Εresidual
offset =5x10
offset =1x10 Chaboche simulation Chaboche simulation
Experiment[16] Anisotropy simulation Experiment[16]
Εresidual -4 Anisotropy simulation [16]
Crystal plasticity RVE
Εresidual=1x10-4 Crystal plasticity RVE[16] offset =5x10 Chaboche simulation
offset Chaboche simulation 120 Experiment[16]
120 Experiment[16] Crystal plasticity RVE[16] 12(b) Crystal plasticity RVE[16]
12(a) 120
100 120 100
12(a) 12(b)
100 100
80
80
80 6080
60
60 4060
40 40
40 20
20 20
20 0
0 0
0 -20
-20 -20
-20 -40
-40 -40
-40 -60
-60
-60 -60 -80
-80 Pre-tor-strain
Pre-tor-strain 1%
1% Unloadingtoto60.9MPa
Unloading 60.9MPa
-80 -80
Pre-tor-strain 1%
Pre-tor-strain 1% Unloading to 60.9MPa
Unloading to 60.9MPa -100
-100
-100 -100 -100
-100-80
-80 -60
-60 -40
-40 -20
-20 00 20
20 40
40 6060 8080100
100
-100 -80 -60
-100 -80-40
-60 -20 0 20
-40 -20 40 40
0 20 60 60
80 80
100100
σ /MPa
/MPa
σ /MPa
σ /MPa
residual -3 Anisotropysimulation
Anisotropy simulation
Εoffset
Εresidual =1x10
=1x10-3
offset Chabochesimulation
Chaboche simulation
[16]
Experiment
Experiment[16]
Crystalplasticity RVE[15]
plasticity RVE [15]
Crystal
120
120 12(c)
12(c)
100
100
80
80
60
60
40
40
20
20 0
0 -20
-20 -40
-40 -60
-60 -80
Pre-tor-strain 1% Unloading to 60.9MPa
-80-100
Pre-tor-strain
-100 -80 -60 1%-40 -20 0Unloading
20 40 to6060.9MPa
80 100
-100
-100 -80 -60 -40 -20σ /MPa
0 20 40 60 80 100
σ /MPa
Figure 12. The simulations and experiments of subsequent yield surfaces on the √ σ − 3τ stress
Figure 12. The simulations and experiments of subsequent yield surfaces on the σ − 3τ stress plane
Figureplane with
with12. The preloading
preloadingsimulations of and
1% proportional
experiments
of 1% proportional
pretorsion
pretorsionof
equivalent
subsequent
equivalent yield
strain
strain and unloading
surfaces
and unloading on the σ to−equivalent
to equivalent
3τ stress
stress
stress of 60.9 MPa: (a) the offset strain is 1−×4 10−4; (b) the offset strain is 5 × 10−4; (c) the offset strain is 1
−4 ;unloading
plane
of with
60.9 preloading
MPa: (a) the of 1% strain
offset proportional
is 1 × 10pretorsion
; (b) the equivalent
offset strain strain
is 5 × and
10 (c) the to strain
offset equivalent
is
× 10−3−. 3
stress
1× 60.9. MPa: (a) the offset strain is 1 × 10 ; (b) the offset strain is 5 × 10 ; (c) the offset strain is 1
of10 −4 −4

× 10−3. Anisotropy simulation -4 Anisotropy simulation

Εresidual
offset =1x10
-4 Εresidual
offset =5x10
Chaboche simulation Chaboche simulation
Experiment[16] Experiment[16] Crystal plasticity RVE[16]
Crystal plasticity RVE[16]
Anisotropy simulation Anisotropy simulation
Εresidual
offset =1x10
-4 Εresidual=5x10-4
offset13(b)
100 13(a) Chaboche simulation 100 Chaboche simulation
Experiment[16] Experiment[16] Crystal plasticity RVE[16]
Crystal plasticity RVE[16] 80
80

100 13(a)
60 10060 13(b)

80 40 8040
20 20
60 60
0 0
40 40
-20 -20
20 20
-40 -40
0 0
-60 -60
-20 -20
-80 -80
-40 -100 Pre-ten-tor-strain 1% -40
-100 Pre-ten-tor-strain 1% Unloading to 58MPa
Unloading to 58MPa
-60 -60 -100 -80 -60 -40 -20 0 20 40 60 80 100
-100 -80 -60 -40 -20 0 20 40 60 80 100
-80 σ /MPa -80 σ /MPa
Pre-ten-tor-strain 1% -100 Pre-ten-tor-strain 1% Unloading to 58MPa
-100 Unloading to 58MPa
Figure 13. Cont.
-100 -80 -60 -40 -20 0 20 40 60 80 100 -100 -80 -60 -40 -20 0 20 40 60 80 100
σ /MPa σ /MPa
Materials 2018, 12,
Materials 2019, 11, 543
x FOR PEER REVIEW 16 of
16 of 19
19

-3 Anisotropy simulation
Εresidual
offset =1x10 Chaboche simulation
Experiment[16] Crystal plasticity RVE[16]

100 13(c)
80
60
40
20
0
-20
-40
-60
-80
-100 Pre-ten-tor-strain 1% Unloading to 58MPa

-100 -80 -60 -40 -20 0 20 40 60 80 100

σ /MPa
√
Figure 13. The simulations and experiments of subsequent yield surfaces on the σ − 3τ stress plane
Figure
with 13. The of
preloading simulations and experiments
1% proportional of subsequent
pre-tension-torsion equivalentyield surfaces
strain on the σto
and unloading
− 3τ stress
equivalent
plane of
stress with preloading
58 MPa: of 1%strain
(a) the offset proportional
is 1 × 10pre-tension-torsion equivalent
−4 ; (b) the offset strain is 5 × 10strain
−4 ; (c) and unloading
the offset strain to
is
equivalent
−
1 × 10 . 3 stress of 58 MPa: (a) the offset strain is 1 × 10−4; (b) the offset strain is 5 × 10−4; (c) the offset

strain is 1 × 10−3.
The anisotropic distortional yield surface expanded continuously with the increase of offset strain,
whichThe was anisotropic
consistentdistortional yield surface
with the experimental expanded
data and thecontinuously
simulation results with the increase
of the crystalofplastic
offset
strain, which was consistent with the experimental data and the simulation
model. The cross effect of the anisotropic yield surface was relatively prominent with the offset strain results of the crystal
plastic
of 1 × 10 model.
−4 . The The crosspoint”
“sharp effect appeared
of the anisotropic yield surface
in the preloading was relatively
direction, and the yield prominent
surface with the
was flat
offset
in the strain
opposite of 1direction.
× 10 . TheThe
−4 “sharp
crosspoint”
effect appeared in the preloading
of the distortional yield surfacedirection, and the yield
was gradually surface
weakened
was flat in the opposite direction. The cross− 4 effect of the
− 4 distortional
when the offset strain increased from 1 × 10 to 5 × 10 and eventually reached 1 × 10 , and the yield surface was − gradually
3
weakened when
anisotropic the offset
distortional yieldstrain increased
surface from 1 ×approaching
was gradually 10−4 to 5 × 10the −4 and eventually reached 1 × 10−3,
isotropic circular yield surface
and the anisotropic distortional yield surface was gradually
computed by the Chaboche model (cf. the purple solid lines in Figures 11–13), approaching the isotropic circular yield
which complied with
surface computed
the experimental data. by the Chaboche model (cf. the purple solid lines in Figures 11–13), which
complied
Comparingwith the theexperimental
results of the data.crystal plastic constitutive model (cf. the green lines in Figures 11–13)
Comparing the results of
with the results of anisotropic distortional the crystalyield
plastic constitutive
surface model shows modelthat (cf.both
the of green
themlines
could indescribe
Figures
11–13) with the results of anisotropic distortional yield surface model shows
the expansion, shift, and distortion of subsequent yield surface, but the results of the anisotropic model that both of them could
describe
in this paper the were
expansion,
generally shift, andthan
better distortion
those of of thesubsequent
crystal plastic yield surface, model.
constitutive but theThe results of the
simulation
anisotropic model in this paper were generally better than those
results of the anisotropic model were very close to the experimental data under pre-tension, which of the crystal plastic constitutive
model.
was The simulation
superior resultsresults
to the simulation of theofanisotropic modelmodel,
crystal plasticity were asvery close tointhe
presented experimental
Figure 11. The results data
under
of pre-tension,
the anisotropic which
model werewas superior
slightly to the
higher thansimulation
the experimental results values
of crystalin theplasticity model, as
case of pre-torsion
presented in Figure 11. The results of the anisotropic model were
and pre-tension-torsion, and the crystal plastic model was more accurate along the torsional direction. slightly higher than the
experimental
However, the values in theperpendicular
yield stress case of pre-torsion to theand pre-tension-torsion,
preloading direction forand thethe crystal
crystal plastic
plastic model
model in
was more accurate along the torsional direction.
Figures 12 and 13 was not as precise as that of the anisotropic model. However, the yield stress perpendicular to the
preloading directionof
The parameters for the crystal
kinematic plastic model
hardening in Figures
and isotropic 12–13 was
hardening partsnotof as precise asyield
anisotropic thatmodel
of the
anisotropic
in Table 1 aremodel.the same as those in Chaboche model. The diameter of the anisotropic distortional yield
The parameters
surface (cf. the red solid of kinematic hardening
lines in Figures 11–13)and wasisotropic
smaller than hardening
that ofparts of anisotropic
the circular yield surface yield
model in Table 1 are the same as those in Chaboche model.
simulated by Chaboche model under small offset strains, and the ratio of the two diameters along The diameter of the anisotropic
distortional
the yield surface
loading direction was R (cf. the red solid lines in Figures 11–13) was smaller than that of the
distortion : Rcircle = 1 : (1 + β 1 ), as shown in Figures 11–13. Therefore,
circular yield surface simulated
the subsequent yield surface derived by Chaboche
from the model undermodel
Chaboche small offset
was the strains,
sameand thesubsequent
as the ratio of the
yield surface simulated byloading
the anisotropic yield Rdistortion : Rcircle = 1 : (1+ βgradually
1)
two diameters along the direction wasmodel when anisotropy , as showndegenerated
in Figures to
isotropy. The diameter
11–13. Therefore, of the circular
the subsequent yield
yield surface
surface simulated
derived frombythe Chaboche
Chaboche model
model could
wasbethe reduced
same to as
fit the experimental curve by adjusting the parameters of the model,
the subsequent yield surface simulated by the anisotropic yield model when anisotropy gradually and the simulation results were
obviously
degenerated better than the curves
to isotropy. with the of
The diameter parameters
the circular adopted
yield in this paper.
surface However,
simulated the non-circular
by Chaboche model
characteristic of the experimental yield surface could not be well
could be reduced to fit the experimental curve by adjusting the parameters of the model, and the captured by Chaboche model.
simulation results were obviously better than the curves with the parameters adopted in this paper.
However, the non-circular characteristic of the experimental yield surface could not be well
captured by Chaboche model.
Materials 2019, 12, 543 17 of 19

Due to the different numbers of the experimental data of the three pre-loading conditions, the error
of each yield surface determined by different offset strain under various pre-deformation conditions is
defined based on the benchmark of the experimental yield stresses as:
   k 
N kσaniso − σtest k2 k
  N  σ −

1 1 cryst σ test
= ∑  ∑

err aniso ij , errcryst ij = 2
 
 , (38)
N k =1 
σtest − σunloading
k
 N k =1 

σtest − σunloading
k

2 2

where the err aniso ij and err aniso ij represent, respectively, the mean errors of the anisotropic model and
the crystal plastic model with the i-th offset strain and the j-th pre-deformation type, the N is the total
number of reloading directions, the k indicates the k-th reloading direction, and kk2 expresses the
2-norm (Euclid norm) of the stress tensor.
The mean value of every error for the anisotropic model and the crystal plastic model are given
respectively by:
1 3 1 3
err aniso = ∑ err aniso ij , errcryst = ∑ errcryst ij , (39)
9 i,j=1 9 i,j=1

where the err aniso and errcryst are the mean error of the anisotropic model and the crystal plastic model,
and the total numbers of i and j are both three.
The standard deviation is calculated with fundamental statistics knowledge by the error and the
mean error. The errors and the standard deviation of the yield surface with the anisotropic model and
the crystal plastic model under every offset strain and pre-formation are listed in Table 2. There was
an obvious difference in the predicted accuracy between various offset strains and pre-formations
conditions. In terms of the pre-torsion loading, the mean error and the standard deviation of the
anisotropic model were higher than those of the crystal plastic model, which was attributed to the
shear hardening model in slip process. On the whole, however, the mean error and the standard
deviation of the anisotropic model, 11.54% and 8.23%, were both smaller than those of the crystal
plastic model, as shown in the last column of Table 2.

Table 2. The errors and the standard deviation of the yield surface with the anisotropic model and the
crystal plastic model.

Pre-Tension Loading Pre-Torsion Loading Pre-Ten-tor Loading Mean

Error (%) (Standard
1 × 10−4 5 × 10−4 1 × 10−3 1 × 10−4 5 × 10−4 1 × 10−3 1 × 10−4 5 × 10−4 1 × 10−3 Deviation)
Crystal 18.64 17.1 12.67 14.0 12.61 11.5 15.5 15.16 15.29 14.72
plasticity RVE (11.09) (10.73) (7.99) (14.33) (8.61) (10.19) (5.41) (8.8) (7.28) (9.38)
Anisotropy 9.91 8.24 8.08 20.83 13.05 11.28 14.1 9.74 8.65 11.54
simulation (6.98) (9.42) (10.07) (12.93) (10.95) (11.2) (4.66) (5.32) (2.56) (8.23)

5. Discussion and Conclusions

Compared with the experimental data and the simulated results of crystal plastic model,
the anisotropic distortional yield surface model proposed in this paper can better exhibit the anisotropic
yield characteristic after preloading and the cross effect of subsequent yield surface. The anisotropic
distortional yield surface model can characterize the nonlinear hardening process that anisotropic yield
surface expands to isotropic yield surface under reloading. The predicted accuracy of the anisotropic
model is higher than the crystal plastic model, with the mean error and the standard deviation as
11.54% and 8.23%, which proves the reasonability and validity of the anisotropic distortional yield
surface model to capture the anisotropic distortional features.
Considering the cross effect, the anisotropic yield model can accurately simulate the experimental
yield stresses, which are used to calibrate the parameters of the anisotropic model (here, it is calibrated
by the experimental results under pre-tension loading). Nevertheless, there was a certain deviation
between the yield surface predicted and the experimental data under other preloading conditions.
The anisotropic yield model was developed based on the Chaboche model within the framework
Materials 2019, 12, 543 18 of 19

of the von Mises theory. Meanwhile, the equivalent stress-strain curve under pure torsion loading
was lower than that of uniaxial tension (shown in Figure 8). Therefore, the yield stresses predicted
by the anisotropic yield model were higher than those of the experimental ones under pre-torsion
and pre-tension-torsion loading. The cross effect of subsequent yield surface varies with preloading
conditions according to the experimental data and simulation results are shown in Figures 11–13.
It should be pointed out that some combinations of the two cross effect parameters β 1 , β 2 in
anisotropic yield model can describe the concave yield surface, which is consistent with the concavity
of experimental yield surface under pre-deformation strain of 5% for T2 pure copper [34]. However, it
is contrary to the convexity hypothesis of the loading surface deduced by the fundamental Drucker
postulate within the classical plastic theory framework, which implies that further study in relation to
the concavity of yield surface needs to be performed.

Author Contributions: Conceptualization, Y.C.; Data curation, Z.K.; Funding acquisition, K.Z.; Investigation, J.C.;
Methodology, J.C. and Y.C.; Project administration, K.Z.; Software, Z.K.; Supervision, K.Z.; Visualization, Z.K.;
Writing—original draft, J.C.; Writing—review & editing, G.H. and Q.S.
Funding: This research was funded by [the National Natural Science Foundation of China] grant numbers
[11472085, 51675110 and 11632007], Key Laboratory Project of Disaster Prevention and Engineering Safety of
Guangxi (Grant 2016ZDX07).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Ohashi, Y.; Kawashima, K.; Yokochi, T. Anisotropy due to plastic deformation of initially isotropic mild steel
and its analytical formulation. J. Mech. Phys. Solids 1975, 23, 277–294. [CrossRef]
2. Cheng, S.; Krempl, E. Experimental determination of strain-induced anisotropy during nonproportional
straining of an A1/Mg alloy at room temperature. Int. J. Plast. 1991, 7, 827–846. [CrossRef]
3. Prager, W. Recent developments in the mathematical theory of plasticity. J. Appl. Phys. 1949, 20, 235–241.
[CrossRef]
4. Ziegler, H. A modification of prager’s hardening rule. Q. Appl. Math. 1959, 17, 55–65. [CrossRef]
5. Mróz, Z. On the description of anisotropic workhardening. J. Mech. Phys. Solids 1967, 15, 163–175. [CrossRef]
6. Hecker, S.S. Constitutive equations in viscoplasticity:Computational and engineering aspects. Am. Soc. Mech.
Eng. 1976, 1, 1–33.
7. Ohno, N.; Wang, J.D. Transformation of a nonlinear kinematic hardening rule to a multisurface form under
isothermal and nonisothermal conditions. Int. J. Plast. 1991, 7, 879–891. [CrossRef]
8. Chaboche, J.L.; Dang Van, K.; Cordier, G. Modelization of the strain memory effect on the cyclic hardening
of 316 stainless steel. In Proceedings of the Transactions of the 5th International Conference on Structural
Mechanics in Reactor Technology, Berlin, Germany, 13–17 August 1979.
9. Chaboche, J.L. On some modifications of kinematic hardening to improve the description of ratchetting
effects. Int. J. Plast. 1991, 7, 661–678. [CrossRef]
10. Chaboche, J.L. Modeling of ratchetting: Evaluation of various approaches. Eur. J. Mech. A-Solids 1994, 13,
501–518.
11. Armstrong, P.; Frederick, C.O. A mathematical representation of the multiaxial Bauschinger effect. High Temp.
Technol. 2007, 24, 1–26.
12. Phillips, A.; Tang, J.L. The effect of loading path on the yield surface at elevated temperatures. Int. J. Solids
Struct. 1972, 8, 464–474. [CrossRef]
13. Wu, H.C.; Yeh, W.C. On the experimental determination of yield surfaces and some results of annealed 304
stainless steel. Int. J. Plast. 1991, 7, 803–826. [CrossRef]
14. Khan, A.S.; Wang, X.W. An experimental study on subsequent yield surface after finite shear prestraining.
Int. J. Plast. 1993, 9, 889–905. [CrossRef]
15. Hu, G.J.; Huang, S.H.; Lu, D.M.; Zhong, X.C.; Li, Z.H.; Brocks, W.; Zhang, K.S. Subsequent yielding of
polycrystalline aluminum after cyclic tension–compression analyzed by experiments and simulations. Int. J.
Solids Struct. 2015, 56–57, 142–153. [CrossRef]
Materials 2019, 12, 543 19 of 19

16. Zhang, K.S.; Huang, S.H.; Liu, G.J.; Lu, D.M. Measuring subsequent yield surface of pure copper by crystal
plasticity simulation. Chin. J. Theor. Appl. Mech. 2017, 49, 870–879.
17. Hu, G.J. Plastic Behavior of Metals under Tension-Torsion Loading-Exprerimental and Numerical Research
on Yield Surface Evolution. Ph.D. Thesis, Guangxi University, Nanning, China, 2012.
18. Khan, A.S.; Pandey, A.; Stoughton, T. Evolution of subsequent yield surfaces and elastic constants with finite
plastic deformation. Part II: A very high work hardening aluminum alloy (annealed 1100 Al). Int. J. Plast.
2010, 26, 1421–1431. [CrossRef]
19. Vincent, L.; Calloch, S.; Marquis, D. A general cyclic plasticity model taking into account yield surface
distortion for multiaxial ratchetting. Int. J. Plast. 2004, 20, 1817–1850. [CrossRef]
20. Feigenbaum, H.P.; Dafalias, Y.F. Directional distortional hardening in metal plasticity within thermodynamics.
Int. J. Solids Struct. 2007, 44, 7526–7542. [CrossRef]
21. Feigenbaum, H.P.; Dugdale, J.; Dafalias, Y.F.; Kourousis, K.I.; Plesek, J. Multiaxial ratcheting with advanced
kinematic and directional distortional hardening rules. Int. J. Solids Struct. 2012, 49, 3063–3076. [CrossRef]
22. Shi, Y.K.; Zhang, K.S.; Hu, G.J. Subsequent yield and plastic flow analysis of polycrystalline copper under
biaxial loading. Acta Metall. Sin. 2009, 45, 1370–1377.
23. Tadano, Y.; Yoshihara, Y.; Hagihara, S. A crystal plasticity modeling considering volume fraction of
deformation twinning. Int. J. Plast. 2016, 84, 88–101. [CrossRef]
24. Francois, M. A plasticity model with yield surface distortion for non proportional loading. Int. J. Plast. 2010,
17, 703–717. [CrossRef]
25. Shi, B.D.; Yan, P.; Chong, Y.; Pan, F.S.; Cheng, R.J.; Peng, Q.M. Loading path dependent distortional hardening
of Mg alloys: Experimental investigation and constitutive modeling. Int. J. Plast. 2017, 90, 76–95. [CrossRef]
26. Fu, Q.; Liu, F.; Zhang, J. A physically motivated model for the evolution of subsequent yield surfaces. J. Theor.
Appl. Mech. 2010, 42, 880–888.
27. Chen, Y.; Hu, G.J.; Lan, Y.T.; Zhang, K.S.; Cai, G.W. Constitutive modeling of slip, twinning and detwinning
for mg alloy and inhomogeneous evolution of microstructure. Acta Mech. Solida Sin. 2018, 31, 493–511.
[CrossRef]
28. Wen, W.; Borodachenkova, M.; Tomé, C.N.; Vincze, G.; Rauch, E.F.; Barlat, F.; Grácio, J.J. Mechanical behavior
of low carbon steel subjected to strain path changes: Experiments and modeling. Acta Mater. 2016, 111,
305–314. [CrossRef]
29. Zhang, K.S.; Wu, M.S.; Feng, R. Simulation of microplasticity-induced deformation in uniaxially strained
ceramics by 3-D Voronoi polycrystal modeling. Int. J. Plast. 2005, 21, 801–834. [CrossRef]
30. Zhang, K.S.; Shi, Y.K. Anisotropy of yielding/hardening and microinhomogeneity of deforming/rotating for
a polycrystalline metal under cyclic tension-compression. Acta Metall. Sin. 2011, 47, 1292–1300.
31. Zhang, K.S.; Shi, Y.K.; Ju, J.W. Grain-level statistical plasticity analysis on strain cycle fatigue of a FCC metal.
Mech. Mater. 2013, 64, 76–90. [CrossRef]
32. Khan, A.S.; Kazmi, R.; Pandey, A.; Stoughton, T. Evolution of subsequent yield surfaces and elastic constants
with finite plastic deformation. Part-I: A very low work hardening aluminum alloy (Al6061-T6511). Int. J.
Plast. 2009, 25, 1611–1625. [CrossRef]
33. Huang, S.H. Test and Crystal Plasticity Numerical Study on the Evolution of Metals Subsequent Yield
Surface. Ph.D. Thesis, Guangxi University, Nanning, China, 2018.
34. Liu, G.L.; Huang, S.H.; Shi, C.S.; Zeng, B.; Zhang, K.S.; Zhong, X.C. Experimental Investigations on
Subsequent Yield Surface of Pure Copper by Single-Sample and Multi-Sample Methods under Various
Pre-Deformation. Materials 2018, 11, 277. [CrossRef] [PubMed]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).