You are on page 1of 31

International Journal of Plasticity xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Plasticity


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

Crystal plasticity based constitutive model of NiTi shape


memory alloy considering different mechanisms
of inelastic deformation
Chao Yu a, Guozheng Kang a,⇑, Qianhua Kan b
a
State Key Laboratory of Traction Power, Southwest Jiaotong University, Chengdu, Sichuan 610031, PR China
b
School of Mechanics and Engineering, Southwest Jiaotong University, Chengdu, Sichuan 610031, PR China

a r t i c l e i n f o a b s t r a c t

Article history: To comprehensively describe the deformation behaviors of polycrystalline NiTi shape
Received 21 March 2013 memory alloy under various thermo-mechanical loading conditions, a micromechanical
Received in final revised form 18 August constitutive model is constructed based on crystal plasticity. At the scale of single crystal,
2013
24 martensite variants are introduced. Different mechanisms of inelastic deformation in
Available online xxxx
the NiTi shape memory alloy, including martensite transformation, martensite reorienta-
tion and detwinning, dislocation slipping in the austenite and twinning in the martensite,
Keywords:
are considered in the proposed model. The Helmholtz free energy for the representative
NiTi shape memory alloy
Constitutive model
volume element of a single crystal is constructed and the thermodynamic driving forces
Crystal plasticity of internal variables are obtained from the dissipative inequalities. The evolution equations
Martensite transformation of internal variables are deduced in power-law forms. The differences of elastic properties
Reorientation between the austenite and martensite phases, as well as the restraint effect of twinning in
Detwinning the martensite on the reverse transformation, are considered. A simplified explicit scale-
Plasticity transition rule is adopted to extend the single crystal model to a polycrystalline version.
Finally, the capability of proposed model to describe the various thermo-mechanical defor-
mation behaviors of polycrystalline NiTi alloy is verified by comparing the simulated
results with the experimental ones.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction

NiTi shape memory alloy has been widely used in aeronautic, microelectronic and biomedical industries due to its unique
super-elasticity and shape memory effect. It is well known that such features are determined by its thermo-elastic martens-
ite transformation, and martensite reorientation and detwinning.
When the test temperature T is higher than its austenite transformation finish temperature Af , the NiTi shape memory
alloy consists of high symmetric austenite phase originally. The austenite phase will transform to martensite phase when
the applied stress is higher than its martensite transformation stress, and a transformation strain is produced. The transfor-
mation strain will be fully recovered during the unloading if the applied stress is not high enough to cause any plastic defor-
mation in the induced martensite. This is the well-known super-elastic effect of NiTi shape memory alloy. However, more
and more experiments show that the plastic deformation occurs in the NiTi shape memory alloy under certain loading con-
ditions. For example, if the NiTi alloy is subjected to a mechanical loading at a temperature high enough (e.g., at 363 K), the
plastic deformation of austenite phase occurs before the martensite transformation starts, which has been observed by Shaw

⇑ Corresponding author. Tel.: +86 28 87603794; fax: +86 28 87600797.


E-mail addresses: guozhengkang@126.com, guozhengkang@home.swjtu.edu.cn (G. Kang).

0749-6419/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijplas.2013.08.012

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
2 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

and Kyriakides (1995) and Qian et al. (2006). This temperature is called as the critical temperature of austenite plasticity, Ad .
Furthermore, the plastic deformation of martensite phase will occur when the applied stress is high enough (e.g., higher than
1200 MPa), even at the test temperature lower than Ad . McKelvey and Ritchie (1999) reported that the plastic deformation of
induced martensite could remarkably restraint the reverse transformation from the induced martensite to original austenite
phase. Similar results were also observed by Shaw and Kyriakides (1995) and Miyazaki et al. (1981).
On the other hand, when the test temperature T is lower than its martensite transformation finish temperature M f , the
NiTi shape memory alloy consists of a low symmetric twinned martensite phase originally. The reorientation and detwinning
of twinned martensite phase will take place during the external loading, and then a kind of inelastic strain occurs. Also, the
inelastic strain can be fully recovered by increasing the test temperature if no plastic deformation is caused in the detwinned
martensite by the applied stress. This is the so-called shape memory effect of NiTi shape memory alloy. Similar to the super-
elastic effect, if the applied stress is high enough, the plastic deformation of detwinned martensite phase can occur, as ob-
served by Kang et al. (2009). The plastic deformation can not be fully recovered during the subsequent heating process.
Based on the experimental results, many constitutive models were constructed in the last two decades. These models can
be classified mainly into two groups, i.e., macroscopic phenomenological models and micromechanical ones. The macro-
scopic phenomenological models are constructed only from the macroscopic experimental data, and some of them have been
implemented into the finite element code to perform the numerical analysis for the structure components manufactured by
NiTi shape memory alloys. Representatives of the macroscopic phenomenological constitutive models can be referred to
those done by Brinson (1993), Tanaka et al. (1995), Lubliner and Auricchio (1996), Abeyaratne and Kim (1997), Auricchio
et al. (1997, 2003, 2007), Bo and Lagoudas (1999a,b,c), Lagoudas and Bo (1999), Lexcellent et al. (2000), Lagoudas and Ent-
chev (2004), Panico and Brinson (2007), Zaki and Moumni (2007a,b), Reese and Christ (2008), Savi et al. (2008), Saint-Sulpice
et al. (2009, 2012), Kan and Kang (2010), Hartl et al. (2010), Saleeb. and Kumar. (2011), Morin et al. (2011), Arghavani et al.
(2011), Chemisky et al. (2011), Zhou (2012), Lagoudas et al. (2012), Zaki (2010, 2012) and Peng et al. (2012). However, the
macro-phenomenological models can not reasonably reflect the microscopic physical natures of thermo-mechanical defor-
mation of NiTi shape memory alloys. Thus, many micromechanical constitutive models were developed to describe the
super-elasticity and shape memory effect of NiTi shape memory alloys by addressing the microscopic physical natures.
Among them, the models based on crystal plasticity are the most popular, since 24 martensite variants with different mor-
phological features and their evolution laws during the thermo-mechanical deformation of NiTi shape memory alloy are rea-
sonably considered. With the help of finite element methods (FEM) or homogenization methods (e.g., self-consistent
method), a single crystal constitutive model can be extended into a polycrystalline version. The typical constitutive models
can be referred to those developed by Patoor et al. (1996, 2006), Tokuda et al. (1999), Gall and Sehitoglu (1999), Gall et al.
(2000), Huang et al. (2000), Gao et al. (2000), Anand and Gurtin (2003), Lagoudas et al. (2006), Thamburaja (2005), Tham-
buraja and Anand (2001, 2002, 2003), Thamburaja et al. (2005, 2009) and Pan et al. (2007).
It is noted that the above-mentioned crystal plasticity based micromechanical constitutive models only focus on the mar-
tensite transformation or the reorientation and detwinning of twinned martensite occurred in the thermo-mechanical defor-
mation of NiTi shape memory alloy. More recently, the crystal plasticity based micromechanical constitutive models (e.g.,
Patoor et al., 1996, 2006; Huang et al., 2000; Gao et al., 2000) were extended to describe the plastic deformation of NiTi shape
memory alloy by considering the dislocation slip in the austenite or both the dislocation slip and twinning in the martensite
phase, as done by Wang et al. (2008), Manchiraju and Anderson (2010) and Yu et al. (2012). Furthermore, in order to describe
the cyclic deformation (including the transformation ratchetting) of super-elastic NiTi shape memory alloy presented in the
cyclic thermo-mechanical loading (observed by Lagoudas and Bo, 1999; Kang et al., 2009; Moumni et al., 2009; Saint-Sulpice
et al., 2012; Khalil et al., 2013), Yu et al. (2013) constructed a crystal plasticity based constitutive model by introducing 24
friction slip systems at austenite–martensite interfaces. However, the extended micromechanical constitutive models by
Wang et al. (2008), Manchiraju and Anderson (2010) and Yu et al. (2012, 2013) only addressed the martensite transforma-
tion and the plasticity in austenite or induced martensite phase, and were mainly suitable for describing the thermo-
mechanical deformation of super-elastic NiTi shape memory alloy. The cyclic deformation of the NiTi shape memory alloy
originally consisting of twinned martensite phase (i.e., for the shape memory effect) observed by Kang et al. (2009) cannot
be reasonably described by such models (i.e., Wang et al., 2008; Manchiraju and Anderson, 2010; Yu et al., 2012, 2013). To

Table 1
Summary for the existing crystal plasticity based constitutive models.

Model Inelastic mechanisms


A B C D E F
p p
Thamburaja (2005)    
p p p
Thamburaja et al. (2005)   
p p
Wang et al. (2008)    
p p
Manchiraju and Anderson (2010)    
p p
Yu et al. (2012)    
p
Yu et al. (2013)     

Notes: A, martensite transformation; B, reorientation of twinned martensite; C, detwinning of twinned martensite; D, dislocation slip in austenite; E,
twinning in martensite; F, restraint of plasticity to reverse transformation.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 3

sum up, the capability of the existing crystal plasticity based constitutive models to describe the thermo-mechanical defor-
mation of NiTi shape memory alloy is listed in Table 1. It can be found from Table 1 that the crystal plasticity based consti-
tutive model comprehensively considering the inelastic mechanisms of NiTi shape memory alloy, such as the martensite
transformation, the reorientation and detwinning of twinned martensite, and the plasticity in the austenite and martensite
phases, has not been proposed.
Therefore, in this work, a crystal plasticity based micromechanical constitutive model of NiTi shape memory alloy is con-
structed by considering the martensite transformation, the reorientation and detwinning of twinned martensite phase, as
well as the plasticity caused by the dislocation slip in the austenite at relatively higher temperature and by the twinning
in the martensite under relatively higher applied stress together. The Helmholtz free energy in a representative volume ele-
ment of NiTi shape memory single crystal is established, which contains the elastic strain energy, the energy related to the
martensite transformation, reorientation and detwinning, respectively, and the one related to the plastic deformation. An
explicit scale-transition rule originally proposed by Cailletaud and Pilvin (1994) and recently simplified by Yu et al.
(2012, 2013) is employed to obtain the polycrystalline constitutive model from the single crystal version. Firstly, the pro-
posed model is verified by comparing the simulations with the experimental results of polycrystalline NiTi shape memory
alloys under uniaxial loading conditions at different ambient temperatures. Then, the deformation behaviors of polycrystal-
line NiTi shape memory alloys under the uniaxial multi-stepped loading conditions are simulated. Finally, the capability of
the proposed model to predict the non-proportionally multiaxial deformation behaviors of polycrystalline NiTi shape mem-
ory alloy is discussed.

2. Outline of experimental results

To illustrate each unique phenomenon of NiTi shape memory alloy dominated by the above-mentioned inelastic mech-
anisms and keep the integrity of the content, some typical experimental results are outlined in this section (Shaw and
Kyriakides, 1995; Miyazaki et al., 1981; Kang et al., 2009) and additional experimental observations (for the non-proportion-
ally multiaxial loading cases) have also been made.

2.1. Uniaxial loading–unloading tests at different temperatures

Cited from Shaw and Kyriakides (1995), the critical temperature parameters of the used NiTi shape memory alloy are: the
martensite transformation finish temperature M f = 203 K and start temperature M s = 272 K; the austenite transformation
start temperature As = 302.5 K and finish temperature Af = 335 K.

2.1.1. Material being initially twinned martensite


Fig. 1(a) and (b) show the stress–strain curves obtained in the uniaxial tension-unloading tests at 283 K and with different
maximum strains. Since the specimen was first cooled to 77 K and then warmed up to 283 K before it was tested, the mate-
rial consisted of the twinned martensite phase initially even if the test temperature (283 K) was higher than M f . It is seen
from Fig. 1(a) that the reorientation and detwinning of twinned martensite phase occur when the applied stress increases
progressively. Then, the twinned martensite variants transform into a single variant which has the highest stability under
the applied stress, which results in a large inelastic strain (about 5% in Fig. 1(a)). When the applied stress is high enough
(e.g., higher than 1200 MPa), the plastic strain occurs in the detwinned martensite phase due to its twinning deformation
(as commented by Otsuka and Ren, 2005), as shown in Fig. 1(b). Moreover, Kang et al. (2009) reported that such plastic strain
could accumulate cyclically during the cyclic loading–unloading as shown in Fig. 1(c) for the load case 300 ± 300 MPa, and
such accumulation of plastic strain was dependent on the applied stress levels. The details can be referred to Kang et al.
(2009).

2.1.2. Material being initially austenite


When the test temperature is higher than 335 K, the material consists of austenite phase originally. Fig. 2(a) and (b) show
that at the temperatures equal to or lower than 343 K, as stated in the Introduction, after certain applied stress, the thermo-
elastic transformation from austenite to martensite phase occurs, and a large transformation strain is produced; while dur-
ing the unloading, when the applied stress becomes lower than the reverse transformation start stress, the induced martens-
ite will reversely transform to the austenite phase, and the transformation strain is gradually recovered. No residual strain is
remained after completely unloading, because no plastic yielding takes place in the original austenite phase. It implies that
the transformation stress is still lower than the yielding stress of austenite phase within such temperature range (6343 K),
even if the transformation stress increases with the increasing temperature.
When the test temperature is higher than 343 K, an apparently residual strain is observed after completely unloading as
shown in Fig. 2(c), which is caused by the plastic yielding of some austenite single crystal grains before the transformation
from austenite to martensite phase occurs. Further, the plastic yielding of austenite phase becomes more and more remark-
able with the increasing temperature and then results in larger and larger residual strain as shown in Fig. 2(c) and (d). Such
plastic deformation comes from the dislocation slip in the austenite phase.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
4 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

(a) 800
Experiment of Shaw and Kyriakides (1995) (c) 800
700 Simulation Load case 300 ±300MPa
600 600

Stress σ , MPa
Stress, MPa
500

400 400

300
200
200

100
0
0 0 2 4 6 8 10 12
0 2 4 6 8
Strain ε , %
Strain, %

(b) 1600 Experiment of Shaw and Kyriakides (1995) (d) 1600


Simulation 4%
1200 7%

Stress, MPa
1200 9.5%
12%
Stress, Mpa

800

800
400

0
400 280
Te 300320 12
14
m 340 10
pe 6
8
ra 360
in , %
tu 380 4
2
re Stra
0 0
0 2 4 6 8 10 12 14 , K 400
Strain, %
Fig. 1. Stress–strain responses of the NiTi shape memory alloy (initially twinned martensite) in the uniaxial tension-unloading tests at 283 K and with
different maximum strains: (a) 7%; (b) 12%; (c) cyclic accumulation of plastic strain (from Kang et al., 2009); (d) simulated shape memory effect.

It should be noted that, since the applied stresses for the tests shown in Fig. 2(c) and (d) are much lower than the plastic
yielding stress of martensite phase (about 1200 MPa), the plastic deformation observed in Fig. 2(c) and (d) are only caused by
the dislocation slip in the austenite phase. However, from the result shown in Fig. 2(e), which is obtained at 343 K and with a
high stress (about 1400 MPa), it is seen that the plastic deformation of induced martensite occurs, and then the reverse
transformation from the induced martensite to austenite phase is greatly restrained by it. Only a small part of transformation
strain can be recovered by the reverse transformation after the applied stress is unloaded to zero. The residual strain consists
of residual transformation strain (which is represented by the residual martensite volume fraction in the proposed model)
and martensite plastic strain.

2.2. Multi-stepped uniaxial loading–unloading-heating tests

The experimental data of uniaxial multi-stepped loading–unloading-heating conditions are cited from Miyazaki et al.
(1981). The critical temperature parameters of used NiTi shape memory alloy are: M s = 190 K, Af = 221 K. The experiments
are performed at 243 K (i.e., the alloy consists of the austenite phase initially) and 173 K (i.e., the alloy consists of the
twinned martensite phase initially), respectively, and the results are shown in Figs. 3(a) and 4(a). In order to make the resid-
ual martensite phase recovered, the specimens are heated to 373 K after each loading–unloading cycle, and then cooled to
243 K or 173 K, respectively.
It is seen from Fig. 3(a) that for the super-elastic NiTi shape memory alloy (originally austenite phase at 243 K), the effect
of heating process on the thermo-mechanical deformation of the NiTi alloy just can be observed obviously after six thermo-
mechanical cycles, since the applied strain becomes high enough to cause the plastic yielding of induced martensite (which
will restrain the reverse transformation from induced martensite to austenite phase during the unloading). The residual
strain is about 3.4% (Point a) after 6th cycle, but it becomes to be about 2.4% at the start point of 7th cycle (i.e., Point b).
The recovered strain (about 1.0%) is caused by the recovery of residual martensite phase occurred during the subsequent
heating to 373 K. The remained strain (about 2.4%) is caused by the plastic strain of induced martensite. Also, Fig. 3(a) shows
that when the maximum applied strain for each cycle is not large enough (e.g., <8%), no plastic deformation of induced mar-
tensite occurs; however, when the maximum applied strain becomes large enough (e.g., >8%), the plastic deformation occurs
apparently.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 5

600 800
(a) (c) Experiment of Shaw and Kyriakides (1995)
Simulation
500
600
400

Stress, MPa
Stress, MPa

300 400

200
200
100
Experiment of Shaw and Kyriakides (1995)
Simulation
0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Strain, % Strain, %
800
(b) (d) 800

600 600
Stress, MPa

400 Stress, MPa 400

200 200 Experiment of Shaw and Kyriakides (1995)


Simulation
Experiment of Shaw and Kyriakides (1995)
Simulation
0 0
0 1 2 3 4 5 6 7 0 2 4 6 8
Strain, % Strain, %
(e) 1600
Experiment of Shaw and Kyriakides (1995)
Simulation
1200
Stress, Mpa

800

400

0
0 2 4 6 8 10 12

Strain, %
Fig. 2. Stress–strain curves of the NiTi shape memory alloy (initially austenite) in the uniaxial tension-unloading tests at different temperatures: (a) 333 K;
(b) 343 K; (c) 353 K; (d) 363 K; (e) 343 K (with high applied stress).

From the results obtained at 173 K and shown in Fig. 4(a), it is also seen that the apparent plastic yielding of detwin-
ned martensite just occurs after four cycles, when the applied strain becomes high enough. When the applied strain is
relatively lower (for example, lower than 8% as shown in Fig. 4(a)), no apparent plastic deformation occurs in the
detwinned martensite, and the inelastic strain caused by the orientation and detwinning of twinned martensite can
be recovered by the subsequent heating process (to 373 K). In Fig. 4(a), Point a represents the residual strain after four
cycles, and Point b represents the strain at the start point of fifth cycle. The strain difference between these two points
represents the strain caused by the reorientation and detwinning of twinned martensite, which is recoverable by the
subsequent heating process. The remained strain after the heating (i.e., the strain at Point b) represents the plastic strain
of detwinned martensite.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
6 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

(a) 800 Experiment of Miyazaki et al. (1981)


1st cycle
2nd cycle
600 3th cycle
4th cycle

Stress, MPa
5th cycle
400 6th cycle
7th cycle

200

b a
0
0 3 6 9 12 15 18
Strain, %

(b) 800
1st cycle Simulation
2nd cycle
3th cycle
600
4th cycle
Stress, MPa

5th cycle
6th cycle
400 7th cycle

200

b a
0
0 3 6 9 12 15 18
Strain, %

Fig. 3. Stress–strain responses in the multi-stage loading–unloading-heating–cooling test of polycrystalline NiTi shape memory alloy at 243 K: (a)
experiment (from Miyazaki et al., 1981); (b) simulation.

2.3. Non-proportionally multiaxial loading tests

The non-proportionally multiaxial loading tests, as well as the uniaxial tension-unloading one, are performed in this work
for the super-elastic NiTi shape memory alloy micro-tubes (Ni, 55.9at%, from Jiangyin Materials Development Co. Ltd., Chi-
na). Four different non-proportional paths, i.e., square, hourglass, butterfly and octagon paths (as shown in Fig. 5), are used.
The NiTi alloy at the test temperature (300 K) originally consists of austenite phase. The cylindrical specimens with an outer
diameter of 2.01 mm and inner diameter of 1.68 mm are cut from the as-received micro-tubes. The total length of the spec-
imen is 60 mm and the gauge length is 30 mm. All the tests are performed under the stress-controlled loading conditions
using the test machine of MTS858-5KN. The applied stress rate is set as 10 MPa/s in both axial and torsional directions.
The minimum axial stresses are all set as zero and the maximum ones are set as 565.6, 565.6, 565.6 and 800 MPa, respec-
tively; while the minimum (negative) and maximum (positive) equivalent shear stresses are set to be the same in magni-
tude, i.e., 282.8, 282.8, 282.8 and 400 MPa, respectively. It means that a symmetrical stress-controlled loading condition
is prescribed in the torsional direction.
Fig. 6 shows the experimental results of the uniaxial tension-unloading, which is used to understand the basic perfor-
mance of the NiTi shape memory alloy micro-tubes. It can be found from Fig. 6 that the start and finish stresses of forward
transformation (i.e., from austenite to martensite phase) are about 400 MPa and 500 MPa, respectively; while the start and
finish stresses of reverse transformation (i.e., from induced martensite to austenite phase) are about 100 MPa and 10 MPa,
respectively. After the unloading, a small residual strain is observed. It is mainly caused by the dislocation slip in parent aus-
tenite phase, since the applied stress is much lower than the plastic yielding stress of induced martensite phase.
The non-proportionally multiaxial experimental results are shown in Figs. 7–10 Responded strain points, i.e., o0 , a0 , b0 , c0 ,
o00 , d0 , e0 , and a0 shown in Fig. 7(a) are corresponding to the applied stress points o, a, b, c, o, d, e, and a shown in Fig. 5(a)
(where o–a–b–c–o–d–e–a forms a closed loading cycle) for the square path. In the loading segment o–a, almost no shear
deformation occurs, because the applied shear stress is still zero. When the applied stress reaches to point a, the transfor-
mation from austenite to martensite phase has been finished as shown in Fig. 7(b). In the loading segment a–b, the shear
deformation of induced martensite phase occurs, but the axial strain keeps almost unchanged. In the loading segment

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 7

(a) 1000 Experiment of Miyazaki et al. (1981)

800 1st cycle


2nd cycle
3th cycle

Stress, MPa
600 4th cycle
5th cycle

400

200
a
b
0
0 2 4 6 8 10 12 14 16 18 20
Strain, %

(b) 1000 1th cycle


Simulation
2th cycle
800 3th cycle
4th cycle
5th cycle
Stress, MPa

600

400

200
a
b
0
0 2 4 6 8 10 12 14 16 18 20
Strain, %

Fig. 4. Stress–strain responses in the multi-stage loading–unloading-heating–cooling test of polycrystalline NiTi shape memory alloy at 173 K: (a)
experiment (from Miyazaki et al., 1981); (b) simulation.

b–c, the shear strain increases, as well as the axial strain decreases progressively, since the shear stress keeps unchanged but
the axial stress decreases gradually to zero in this segment. The reverse transformation from induced martensite to austenite
phase partially occurs. In the loading segment c–o, further reverse transformation takes place, which leads to the decrease in
both axial and shear strains. After a closed applied stress loop, i.e., o–a–b–c–o, only small residual axial and shear strains are
resulted in as shown in Fig. 7. In the loading segment o–d, negative shear strain occurs due to the negative shear stress, but
the axial strain keeps almost unchanged due to the zero axial stress and no transformation is caused as shown in Fig. 7(c).
The transformation from austenite to martensite phase (caused by compressive stress) just occurs again in the segment d-e
when the applied stress (compressive stress) becomes to be higher than the transformation stress, as shown in Fig. 7(b). In
the segment e–a, the shear strain decreases apparently with the decreasing shear stress, but the axial strain increases slightly
due to the constant axial stress, as shown in Fig. 7.
Furthermore, from the results shown in Figs. 8–10, it illustrates that the non-proportionally multiaxial stress–strain re-
sponses of super-elastic NiTi shape memory alloy depend on the shapes of loading paths. It should be noted that, for the
octagon path after a half cycle (i.e., o–a–b–c–d–o), both the axial and shear stresses come back to point o, i.e., return to zero.
However, the axial strain (point o00 in Fig. 10(a)) does not return to zero due to the plastic deformation of austenite phase.

3. Constitutive model

3.1. Inelastic mechanisms and framework of thermodynamics

Based on the hypothesis of small deformation, the total strain tensor in the single crystal of NiTi shape memory alloy can
be decomposed into three parts, i.e., the elastic strain tensor ee , strain tensor em caused by the thermo-elastic martensite
transformation and the reorientation and detwinning of twinned martensite, and plastic strain tensor ep :

e ¼ ee þ em þ ep ð1Þ
m tr
As mentioned above, the strain tensor e should consist of three parts, i.e., the martensite transformation strain e , reori-
entation strain ere and detwinning strain ede :

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
8 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

(a) (c)

Torsion equivalent stress 3 τ , MPa


300
Torsion equivalent stress 3 τ , MPa
300 b b
c d
200 200

1/2
1/2

100 100

o o
0 a
0 a

-100 -100

-200 -200
d e
-300 e -300 c

0 100 200 300 400 500 600 0 100 200 300 400 500 600
Axial stress σ, MPa Axial stress σ, MPa

(b) (d)
400

Torsion equivalent stress 3 τ , MPa


Torsion equivalent stress 3 τ , MPa

300
b c
c 300 b
d
200

1/2
1/2

200
100
100
o a
0 0 o a

-100
-100
-200
-200 g
e -300 e
d
-300 f
-400
0 100 200 300 400 500 600 0 100 200 300 400 500 600 700 800
Axial stress σ, MPa Axial stress σ, MPa

Fig. 5. Shapes of non-proportionally multiaxial loading paths: (a) square (o-a-b-c-o-d-e-a); (b) hourglass (o-a-b-c-a-d-e-a); (c) butterfly (o-a-b-c-o-d-e-a);
(d) octagon (o-a-b-c-d-o-e-f-g-a).

600
Experiment
Simulation
500
Axial stress σ, MPa

400

300

200

100

0
0 1 2 3 4 5 6
Axial strain ε, %

Fig. 6. Stress–strain curves of the super-elastic NiTi shape memory alloy in the uniaxial tension-unloading test.

em ¼ etr þ ere þ ede ð2Þ


It is well-known that there are 24 martensite variants in the transformed martensite of NiTi shape memory alloy. If the
volume fraction of each martensite variant in the NiTi single crystal is known, the transformation strain etr caused by the
transformation from austenite to martensite phase can be obtained as:

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 9

(a)
3

γ,%
c'
2

-1/2
Torsion equivalent strain 3
Experiment
1 b'
Simulation
o' o''
0
a'
-1
d'
-2 e'
a''

-3

0 1 2 3 4 5 6 7
Axial strain ε, %

(b) 600 e' a''


Experiment
500 Simulation b'
a'
Axial stress σ, MPa

400

300

200
o'

100 d' c'


o''
0

0 1 2 3 4 5 6 7
Axial strain ε, %

(c)
300
Torsion equivalent stress 3 τ , MPa

b'

200 Experiment
c'
Simulation
1/2

100 o'
a''
0 a'
o''
-100

e'
-200

-300 d'

-4 -3 -2 -1 0 1 2 3 4
-1/2
Torsion equivalent strain 3 γ , %

Fig. 7. Experimental and simulated stress–strain results of the super-elastic NiTi shape memory alloy with the non-proportional square path: (a) axial
strain–shear strain response, (b) axial stress–strain response, (c) shear stress–strain response.

X
24
etr ¼ natr Ka ð3-aÞ
a¼1

1 tr a
Ka ¼ g ðm  na þ na  ma Þ ð3-bÞ
2

where, natr and Ka are the volume fraction and strain tensor of the ath martensite variant produced by the martensite trans-
formation, respectively. g tr is the magnitude of shearing deformation caused by the transformation. ma and na are the trans-
formation orientation and habit plane normal vectors, respectively. The details for 24 variants are listed in Table 2.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
10 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

(a) 4 c'

γ,%
b'
3

-1/2
2

Torsion equivalent strain 3


1 o'
0
a' a''
-1 Experiment
Simulation
-2
e' d'
-3

-4

0 1 2 3 4 5 6 7
Axial strain ε, %

(b) 600 b'

500 Experiment d'


Simulation
Axial stress σ, MPa

400
a''
300

a'
200

100
o'
c'
0 e'

0 1 2 3 4 5 6 7
Axial strain ε, %

(c)
300 b'
Torsion equivalent stress 3 τ , MPa

200 Experiment
Simulation
1/2

c'
100
a'
a''
o'
0

-100
e'
-200

-300 d'

-4 -3 -2 -1 0 1 2 3 4
-1/2
Torsion equivalent strain 3 γ , %

Fig. 8. Experimental and simulated stress–strain results of the super-elastic NiTi shape memory alloy with non-proportional hourglass path: (a) axial
strain–shear strain response, (b) axial stress–strain response, (c) shear stress–strain response.

In the work done by Thamburaja (2005) and Thamburaja et al. (2005), the reorientation and detwinning of twinned mar-
tensite variants were considered separately. In each martensite variant, there are two sub-variants. During the external load-
ing, the movement of the interface between every two variants is defined as the reorientation of twinned martensite. The
detwinning of twinned martensite represents the movement of the interfaces between the two sub-variants of each mar-
tensite variant. It should be noted that Thamburaja (2005) and Thamburaja et al. (2005) constructed the models in the
framework of finite deformation. Thus, the definitions of inelastic deformation gradients caused by the reorientation and
detwinning of twinned martensite in Thamburaja (2005) and Thamburaja et al. (2005), are changed to the definitions of
inelastic strains as shown in Eq. (2). The strain caused by the reorientation of twinned martensite can be written as:

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 11

(a) 2

b'

γ,%
1
Experiment

-1/2
Simulation

Torsion equivalent strain 3


d'

0
o'' a'
o'
-1 a''

-2
e'
c'
-3
0 1 2 3 4 5 6 7
Axial strain ε, %

(b) 600 a'


b'
Experiment
500
Simulation a''
Axial stress σ, MPa

400 e'

300

200

100
o' d' o''

0 c'

0 1 2 3 4 5 6 7
Axial strain ε, %

(c)
300 b'
Torsion equivalent stress 3 τ , MPa

d'

200 Experiment
Simulation
1/2

100
a''
0 a'
o'
-100

-200

-300 c' e'

-3 -2 -1 0 1 2 3
-1/2
Torsion equivalent strain 3 γ , %

Fig. 9. Experimental and simulated stress–strain results of the super-elastic NiTi shape memory alloy with non-proportional butterfly path: (a) axial strain–
shear strain response; (b) axial stress–strain response; (c) shear stress–strain response.

X
24 X
23
ere ¼ kij Sij ð4Þ
j>i i¼1

where, kij represents the amount transited from jth to ith martensite variant due to the reorientation of twinned martensite
variants. It is noted that j > i in order to avoid counting repeatedly. Sij is the orientation tensor which is defined by Thambu-
raja (2005) and Thamburaja et al. (2005) as:

Sij ¼ Ki  Kj ð5Þ

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
12 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

(a) 4
d'
3 c'

γ,%
2

-1/2
o' o'' b'
1

Torsion equivalent strain 3


0

-1 Experiment a'

-2 Simulation
e'
-3 a''

-4
g'
-5
f'
-6
0 1 2 3 4 5 6 7 8
Axial strain ε, %

(b) 800 a'


a''
700 Experiment g'
Simulation b'
600
Axial stress σ, MPa

500

400 c'

300 f'

200
e'
100 d'
o'
0 o''
0 1 2 3 4 5 6 7 8
Axial strain ε, %

(c) c'
400
Torsion equivalent stress 3 τ , MPa

300 d'
Experimention b'
Simulation
1/2

200

100 a'' o'


0 o''
a'
-100

-200
g' e'
-300

-400 f'
-5 -4 -3 -2 -1 0 1 2 3 4
-1/2
Torsion equivalent strain 3 γ , %

Fig. 10. Experimental and simulated stress–strain results of the super-elastic NiTi shape memory alloy with non-proportional octagon path: (a) axial
strain–shear strain response; (b) axial stress–strain response; (c) shear stress–strain response.

If the kij is obtained, the volume fraction of ath martensite variant nare formed by the reorientation of twinned martensite
aij
can be calculated by the following equation with the help of interaction matrix k (Thamburaja, 2005; Thamburaja et al.,
2005):
X
24 X
23
aij
nare ¼ k kij a ¼ 1; 2; . . . ; 24 ð6Þ
j>i i¼1

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 13

Table 2
Crystallographic data for the martensite transformation systems in NiTi shape memory alloy (from Thamburaja et al., 2005).

a ma1 ma2 ma3 na1 na2 na3

1 0.8889 0.4044 0.2152 0.4114 0.4981 0.7633


2 0.4044 0.8889 0.2152 0.4981 0.4114 0.7633
3 0.8889 0.4044 0.2152 0.4114 0.4981 0.7633
4 0.4044 0.8889 0.2152 0.4987 0.4114 0.7633
5 0.8889 0.4044 0.2152 0.4114 0.4981 0.7633
6 0.4044 0.8889 0.2152 0.4981 0.4114 0.7633
7 0.8889 0.4044 0.2152 0.4114 0.4981 0.7633
8 0.4044 0.8889 0.2152 0.4981 0.4114 0.7633
9 0.2152 0.8889 0.4044 0.7633 0.4114 0.4981
10 0.2152 0.8889 0.4044 0.7633 0.4114 0.4981
11 0.2152 0.4044 0.8889 0.7633 0.4981 0.4114
12 0.2152 0.4044 0.8889 0.7633 0.4981 0.4114
13 0.2152 0.8889 0.4044 0.7633 0.4114 0.4981
14 0.2152 0.8889 0.4044 0.7633 0.4114 0.4981
15 0.2152 0.4044 0.8889 0.7633 0.4981 0.4114
16 0.2152 0.4044 0.8889 0.7633 0.4981 0.4114
17 0.8889 0.2152 0.4044 0.4114 0.7633 0.4981
18 0.8889 0.2152 0.4044 0.4114 0.7633 0.4981
19 0.4044 0.2152 0.8889 0.4981 0.7633 0.4114
20 0.4044 0.2152 0.8889 0.4981 0.7633 0.4114
21 0.8889 0.2152 0.4044 0.4114 0.7633 0.4981
22 0.8889 0.2152 0.4044 0.4114 0.7633 0.4981
23 0.4044 0.2152 0.8889 0.4981 0.7633 0.4114
24 0.4044 0.2152 0.8889 0.4981 0.7633 0.4114

8
<1
> if a ¼ i
aij
k ¼ 1 if a > i and a ¼ j ð7Þ
>
:
0 otherwise
Since the reorientation of twinned martensite represents the transition between every two twinned variants, the total
volume fraction of martensite should be kept as constant during the reorientation of twinned martensite, and the constraint
P _a
equation 24 a¼1 nre ¼ 0 should be satisfied as proposed by Thamburaja (2005) and Thamburaja et al. (2005).
Considering both the contributions of martensite transformation and reorientation to the volume fraction of ath martens-
ite variant na yields:
na ¼ natr þ nare ð8Þ
In the process of martensite detwinning, the detwinning strain is defined as the sum of transition amount occurred be-
tween every two sub-variants in each martensite variant. Following this consideration, the detwinning stain can be written
as:
X
24
ede ¼ na ðka  ka0 ÞPade ð9Þ
a¼1

where, the volume fractions of two sub-variants in ath martensite variant are ka and 1  ka , respectively. ka0 is the initial va-
lue. Thus, the term ka  ka0 represents the amount of transition occurred between the two sub-variants. Pade is the orientation
tensor for ath detwinning system and is obtained from Thamburaja (2005) and Thamburaja et al. (2005), i.e.,
1 a
Pade ¼ ða  wa þ wa  aa Þ ð10Þ
2
where aa is the unit normal vector of detwinning plane, and wa is the vector denoting the shear direction. The details about
ka0 ; aa and wa are listed in Table 3.
Different from the work done by Thamburaja (2005) and Thamburaja et al. (2005), the plastic deformation of NiTi shape
memory alloy is considered in this paper. The plasticity is assumed to be caused by the dislocation slipping in the austenite
and the twinning in the martensite phase. Thus, the plastic strain can be decomposed into two parts, i.e.,
ep ¼ epA þ epM ð11Þ
As a body-centered cubic (BCC) crystal, the austenite phase of NiTi shape memory alloy possesses 12 primary slip sys-
tems. The primary slip planes are {110}, and the slip directions are h111i. Only the primary slip systems are activated when
the temperature is not so high, e.g., lower than 373 K. So the total plastic strain caused by the dislocation slip in the austenite
phase, e_ pA , can be written as:

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
14 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

Table 3
Crystallographic data for the martensite detwinning systems in NiTi shape memory alloy (from Thamburaja et al., 2005).

a wa1 wa2 wa3 aa1 aa2 aa3 ka0

1 0.5846 0 0.8133 0 0.2804 0 0.271


2 0 0.5846 0.8133 0.2804 0 0 0.271
3 0.5846 0 0.8133 0 0.2804 0 0.271
4 0 0.5846 0.8133 0.2804 0 0 0.271
5 0.5846 0 0.8133 0 0.2804 0 0.729
6 0 0.5846 0.8133 0.2804 0 0 0.729
7 0.5846 0 0.8133 0 0.2804 0 0.729
8 0 0.5846 0.8133 0.2804 0 0 0.729
9 0.8133 0.5846 0 0 0.2804 0 0.729
10 0.8133 0.5846 0 0 0 0.2804 0.729
11 0.8133 0 0.5846 0 0.2804 0 0.271
12 0.8133 0 0.5846 0 0.2804 0 0.729
13 0.8133 0.5846 0 0 0.2804 0 0.729
14 0.8133 0.5846 0 0 0 0.2804 0.729
15 0.8133 0 0.5846 0 0.2804 0 0.271
16 0.8133 0 0.5846 0 0.2804 0 0.729
17 0.5846 0.8133 0 0 0 0.2804 0.729
18 0.5846 0.8133 0 0 0 0.2804 0.271
19 0 0.8133 0.5846 0.2804 0 0 0.271
20 0 0.8133 0.5846 0.2804 0 0 0.729
21 0.5846 0.8133 0 0 0 0.2804 0.729
22 0.5846 0.8133 0 0 0 0.2804 0.271
23 0 0.8133 0.5846 0.2804 0 0 0.271
24 0 0.8133 0.5846 0.2804 0 0 0.729

X
12
e_ pA ¼ ð1  nÞ c_ bA PbA ð12-aÞ
b¼1

1 b b b
PbA ¼ ðs  lA þ lA  sbA Þ ð12-bÞ
2 A

X
24
n¼ na ð12-cÞ
a¼1

where, c_ bA is the dislocation slipping rate of bth slip system in the austenite phase, PbA is the orientation tensor, sbA is the slip
b
direction, and lA is the slip plane normal of bth slip system. n is the total volume fraction of martensite phase. The term
ð1  nÞ in Eq. (12-a) implies that the dislocation slipping only occurs in the austenite phase.
Since only one slip system [1 0 0](0 0 1) of the martensite exists due to its low symmetry, its contribution to the plastic
deformation can be neglected for simplicity. The plastic deformation of martensite phase mainly comes from the twinning
deformation. Otsuka and Ren (2005) summarized all of the 11 twinning systems for the martensite phase of NiTi shape
memory alloy. Based on their work, Wang et al. (2008) constructed a micromechanical model to describe the martensite
plasticity of NiTi shape memory alloy occurred under high external stress. More details about the twinning systems can
be referred to Otsuka and Ren (2005) and Wang et al. (2008). The plastic strain epM occurred in the martensite phase can
be written as:

X
11
e_ pM ¼ n f_ bM1 KbM1 ð13-aÞ
b1 ¼1

1 b1
KbM1 ¼ g ðmb1  nbtwin
1
þ nbtwin
1
 mbtwin
1
Þ ð13-bÞ
2 twin twin
where, f_ bM1 is the rate of twinning volume fraction for the b1th twinning system of martensite phase, and KbM1 is the strain of
b1th twinning system. mbtwin 1
and nbtwin
1
are the twinning orientation and the twinning plane normal, respectively. g btwin
1
is the
magnitude of shearing deformation of b1th twinning system and the details can be found in Table 4. Different from Otsuka
and Ren (2005) and Wang et al. (2008), the volume fraction of martensite phase n is assumed to be contributed by two mech-
anisms in this work, i.e., martensite transformation and reorientation, as shown in Eqs. (8) and (12-c).
Based on the inelastic mechanisms mentioned above, the framework of thermodynamics for the thermo-mechanical
behaviour of NiTi shape memory single crystal is constructed as following:
Considering a representative volume element (RVE) of NiTi shape memory single crystal, its Helmholtz free energy can be
written as:

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 15

Table 4
Crystallographic data for the martensite twinning systems in NiTi shape memory alloy (from Wang et al., 2008).

b mb1 mb2 mb3 nb1 nb2 nb3 g btwin

1 0.3280 0.3280 0.8859 0.7071 0.7071 0 0.3096


2 0.3280 0.8859 0.3280 0.7071 0 0.7071 0.3096
3 0.6894 0.2227 0.6894 0.7071 0 0.7071 0.1422
4 0.6894 0.6894 0.2227 0.7071 0.7071 0 0.1422
5 0.6181 0.7861 0 0 0 1 0.2804
6 0.6181 0 0.7861 0 1 0 0.2804
7 0 0 1 0.5870 0.8096 0 0.2804
8 0 1 0 0.5870 0 0.8096 0.2804
9 1 0 0 0 0.7071 0.7071 0.2385
10 0 0.7071 0.7071 1 0 0 0.2385
11 0.3333 0.6667 0.6667 0.9428 0.2357 0.2357 0.425

w ¼ we þ wm þ wp ð14Þ
Similar to the definition of total strain tensor, the Helmholtz free energy is also decomposed as three parts: i.e., the elastic
energy of RVE we ; the energy related to martensite transformation, reorientation and detwinning wm ; and the one related to
the plasticity wp . The details about the energy parts will be discussed in the following paragraphs.
The explicit expression for the elastic energy of RVE we can be given as:
1 e
we ðee ; nÞ ¼ e : CðnÞ : ee ð15Þ
2
where, CðnÞ is the fourth-ordered elasticity tensor, which is isotropic and dependent on the volume fraction of martensite.
For simplicity, CðnÞ is set as a linear function of martensite volume fraction n, i.e., CðnÞ ¼ ð1  nÞCA þ nCM . CA and CM are the
elasticity tensors of austenite and martensite phases, respectively.
The energy related to the martensite transformation, reorientation and detwinning wm can be newly expressed as:
   X Z tX
T 24
1X 24
2
24  
wm ðT; na ; tÞ ¼ c ðT  T 0 Þ  T ln þ lðT  T 0 Þ na þ Hðna Þ þ na X aM k_ a dt ð16Þ
T0 a¼1
2 a¼1 0 a¼1

where, the first term in the right side of the equation is the internal energy related to the temperature, and c is the heat
capacity. The second term is the energy caused by the entropy difference between the austenite and martensite phases. l
is a constant, named as the coefficient of entropy difference. This term is also used to describe the dependence of martensite
transformation on the temperature. That is to say, as the temperature increases, the free energy increases. It implies that the
potential barrier necessary to be overcome during the transformation is enhanced, thus the martensite transformation be-
comes more difficult at higher temperature. Thus, the transformation start stress from the austenite to martensite phase in-
creases with the increasing test temperature. Eq. (16) is newly obtained from the original version only containing the first
two terms in the right side of Eq. (16) and used by Lagoudas and Entchev (2004), Thamburaja et al. (2005) and Wang et al.
(2008). The last two terms currently proposed in this work are to describe the hardening/softening effects observed during
the martensite transformation, reorientation and detwinning (Nemat-Nasser and Guo, 2006; Kang et al., 2009). The third
term is called as the hardening/softening energy describes the transformation and reorientation hardening/softening fea-
tures presented during the martensite transformation and reorientation. H is the hardening/softening modulus. The fourth
term in the right side of Eq. (16) describes the hardening/softening effect presented in the process of martensite detwinning
and X aM is the detwinning resistance of ath detwinning system. Since the detwinning deformation only occurs in the mar-
tensite phase, this energy is related to the volume fraction of martensite na .
The Helmholtz free energy related to the plastic deformation of NiTi shape memory alloy is given in the rate form, i.e.,
X
12   X
11
w_ p ¼ ð1  nÞ RbA c_ bA  þ n Q bM1 f_ bM1 ð17Þ
b¼1 b1 ¼1

The first term in the right side of Eq. (17) describes the hardening/softening caused by the dislocation slip in the austenite
phase, and RbA is the isotropic slipping resistance of bth slip system. The second term describes the hardening/softening
caused by the twinning in the martensite phase. Q bM1 is the twinning resistance of b1th twinning system. Since the dislocation
slipping only occurs in the austenite phase and the twinning occurs only in the martensite phases, the two energy terms are
related to 1  n and n.
It is well-known that the Clausius dissipative inequality can be written as:
q  rT
C ¼ w_  w_  gT_  P0 ð18Þ
T

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
16 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

where, w_ is the external power. C is the total dissipation, g is the entropy. T is the absolute temperature and q is the heat flux.
It should be noted that the effects of transformation latent heat and the heat from internal dissipation on the inelastic defor-
mation of NiTi shape memory single crystal are neglected for simplicity, since only the quasi-static loading conditions are
considered here. So, the temperature in the single crystal can be taken as homogeneous. Then, the reduced dissipative
inequality can be written as:

C ¼ w_  w_  gT_ P 0 ð19Þ
Generally, the internal power is equal to the external power which can be written as:
_ ¼ r : e_
w ð20Þ
Substituting Eqs. (1), (2), (11) and (15)–(17) into Eq. (19) yields:
    X24 @we @wm 
@we @wm _ @wm
C¼ r : _e  g þ
e T þ r : _ p  w_ p þ r : e_ m 
e a þ a n_ a  P0 ð21Þ
@e e @T |fflfflfflfflfflfflffl{zfflfflfflfflfflfflffl} a ¼1 @n @n @t
|fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflffl} plasticity |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
elastic entropy martensite transformation; reorientation and detwinning

Eq. (21) is the total dissipation which consists of the elastic, entropy and plastic parts, as well as the parts related to the
martensite transformation, reorientation and detwinning. Since the elastic and entropy dissipation parts are always zero, it
yields
@we
r¼ ¼ CðnÞ : ee ð22-aÞ
@ ee
 
@wm T
g¼ ¼ ln þ c ln ð22-bÞ
@T T0
Thus, the dissipative inequality can be re-written as:
X24 @we @wm  m
C ¼ Cm þ Cp ¼ r : e_ m  þ _ a  @w þ r : e_ p  w_ p P 0
n ð23Þ
a¼1 @n a a |fflfflfflfflfflfflffl{zfflfflfflfflfflfflffl}
@n @t
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} plasticity
Martensite transformation; reorientation and detwinning

Stricter constraints for the dissipative inequality are introduced as:


Cm P 0 ð24-aÞ

Cp P 0 ð24-bÞ

3.2. Evolution equations for internal variables

In this section, the evolution equations of internal variables are given in the power-law forms with the help of corre-
sponding thermodynamic driving forces. The detailed deductions for the driving forces can be found in Appendix A. Firstly,
the thermodynamic driving forces and the conjugated variables for each inelastic mechanism are listed as follows:

Transformation : patr $ n_ atr ð25-aÞ


 ij 
Reorientation : p  $ k_ ij sign ðpij Þ ð25-bÞ
re re

Detwinning : pade $ k_ a sign ðr : Pade Þ ð25-cÞ

Austenite plasticity : F bAslip $ c_ bA sign ðr : PbA Þ ð25-dÞ

Martensite plasticity : F bMtwin


1
$ f_ bM1 ð25-eÞ
The constraint of thermodynamics on the martensite transformation can be written as:

n_ atr ¼ 0 no constraint on patr ð26-aÞ

n_ atr > 0 patr > 0 ð26-bÞ

n_ atr < 0 patr < 0 ð26-cÞ

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 17

It is noted that if no transformation takes place for the ath martensite variant, the value of n_ atr is zero. So, the dissipative
inequality, Eq. (A.6), is satisfied automatically and the value of patr is arbitrary. During the forward transformation (from the
austenite to martensite phase), the value of n_ atr is positive. The dissipative inequality, Eq. (A.6), requires the patr to be positive.
Similarly, during the reverse transformation (from the martensite to austenite phase), the value of n_ atr is negative, and the
dissipative inequality, Eq. (A.6), requires the patr to be negative (see Table 5).
Thus, considering the constraint of thermodynamics, the evolution equations of martensite volume fraction can be ob-
tained in the power-law form, i.e.,
 a ntr
p 
n_ atr ¼  trtr  whenever patr P 0 0 6 natr 6 1 0 6 na 6 1 0 6 n 6 1 ð27-aÞ
K
 ntr
 patr 

_a
ntr ¼  tr  whenever patr < 0 0 6 natr 6 1 0 6 na 6 1 0 6 n 6 1 ð27-bÞ
K þ AepM 

n_ atr ¼ 0 other conditions ð27-cÞ


tr
where, K is the transformation resistance. ntr represents the rate sensitivity of martensite transformation, while a rate-inde-
pendent response can be approximately
  described by setting ntr to be high enough (e.g., ntr = 30 in this work as listed in
Tables 6–8). An additional term AepM  is introduced into the evolution equation (Eq. (27-b) for the reverse transformation.
It implies that the reverse transformation will be restrained by the plastic deformation occurred in the induced martensite
phase. A is a material parameter.
The evolution equation for the amount transited from jth to ith martensite variants during the martensite reorientation
can be also obtained in the power-law form, i.e.,
 nre
 ij 
_kij sign ðpij Þ ¼ pre  ð28Þ
re
K re 

Then, multiplying both sides of Eq. (28) by signðpijre Þ, the final evolution equation can be obtained as:
 nre
 ij 
_kij ¼ pre  signðpij Þ whenever 0 6 ni 6 1 0 6 nj 6 1 ð29-aÞ
K re  re

k_ ij ¼ 0 other conditions ð29-bÞ


The evolution equations for other inelastic mechanisms can be obtained in the same way. Here, for simplicity, only the
final forms of evolution equations are given as follow:
For martensite detwinning:

nde
pade
k_ a ¼ sign ðr : Pade Þ whenever na > 0 ð30-aÞ
K de

k_ a ¼ 0 other conditions ð30-bÞ


de
where, K is the detwinning resistance. nde represents the rate sensitivity of martensite detwinning. hxi is McCauley bracket.
When x > 0, hxi ¼ x; when x 6 0, hxi ¼ 0.
For dislocation slip in austenite phase:
* +nAslip
F bAslip
c_ bA ¼ sign ðr : PbA Þ whenever n < 1 ð31-aÞ
K Aslip

c_ bA ¼ 0 other conditions ð31-bÞ


Aslip
where, K and nAslip are the deformation resistance and rate sensitivity of dislocation slipping in the austenite phase,
respectively.
For twinning in martensite phase:
* +nMtwin
F bMtwin
1

f_ bM1 ¼ whenever n > 0 ð32-aÞ


g btwin
1
K Mtwin

f_ bM1 ¼ 0 other conditions ð32-bÞ


where, ntwin and g btwin
1
K Mtwin are the deformation resistance and rate sensitivity of twinning deformation in the martensite
phase, respectively.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
18 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

Table 5
Summary of the proposed model.

Main equations

e ¼ ee þ em þ ep em ¼ etr þ ere þ ede ep ¼ epA þ epM ein ¼ em þ ep

X
24 X
24 X
23 X
24
etr ¼ natr Ka ere ¼ kij Sij ede ¼ na ðka  ka0 ÞPade
a¼1 j>i i¼1 a¼1

X
12 X
11
e_ pA ¼ ð1  nÞ c_ bA PbA e_ pM ¼ n f_ bM1 KbM1
b¼1 b1 ¼1
Thermodynamic forces

1
patr ¼ r : ½Ka þ ðka  ka0 ÞPade   lðT  T 0 Þ  Hna  ee : DC : ee
2

X
24
aij
X
24
aij
pijre ¼ r : ½Sij þ ðka  ka0 Þk Pade   k Hna
a¼1 a¼1

   
pade ¼ r : Pade   X aM F bAslip ¼ r : Pb   Rb
A A F bMtwin
1
¼ r : KbM1  Q bM1
Evolution equations for phase
transformation  a ntr
 
_na ¼ ptr  whenever patr P 0 0 6 natr 6 1 0 6 na 6 1 0 6 n 6 1
tr K tr 

 ntr
 patr 
_na ¼   whenever patr < 0 0 6 natr 6 1 0 6 na 6 1 0 6 n 6 1
tr
K tr þ AepM 

n_ atr ¼ 0 other conditions


Evolution equations for martensite
reorientation and detwinning  nre
 ij 
_kij ¼ pre  signðpij Þ whenever 0 6 ni 6 1 0 6 nj 6 1
K re  re

k_ ij ¼ 0 other conditions

nde
pade
k_ a ¼ de
signðr : Pade Þ whenever na > 0
K

k_ a ¼ 0 other conditions
Evolution equations for plasticity
8 b
nAslip
< F Aslip sign ðr : PbA Þ whenever n < 1
_ bA
c ¼ K Aslip
:
0 other conditions
8 b

nMtwin
< 1
F Mtwin
whenever n > 0
f_ bM1 ¼ b 1 K Mtwin
g twin
:
0 other conditions

Scale-transition rule

r ¼ R þ DðEin  ein Þ

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 19

Table 6
Material parameters for the alloy used by Shaw and Kyriakides (1995).

EA ¼ 60 GPa; EM ¼ 40 GPa; m ¼ 0:33; M s ¼ 272 K; g tr ¼ 0:1308;


b ¼ 0:5 MPa=K; A ¼ 4 GPa;
ntr ¼ 30; K tr ¼ 9 MPa; nre ¼ 30; K re ¼ 25 MPa; H ¼ 2 MPa;
nde ¼ 30; K de ¼ 70 MPa; hde ¼ 1 GPa;
slip
nAslip ¼ 30; K Aslip ¼ 420 MPa; hA ¼ 3 GPa; qAslip ¼ 0:5;
Mtwin twin
nMtwin ¼ 30; K ¼ 530 MPa; hM ¼ 120 MPa; qMtwin ¼ 0:5;
D ¼ 2 GPa;

Table 7
Material parameters for the alloy used by Miyazaki et al. (1981).

EA ¼ 32:5 GPa; EM ¼ 20 GPa; m ¼ 0:33; M s ¼ 190 K; g tr ¼ 0:1308;


b ¼ 0:38 MPa=K; A ¼ 120MPa;
ntr ¼ 30; K tr ¼ 8 MPa; nre ¼ 30; K re ¼ 15 MPa; H ¼ 2 MPa
nde ¼ 30; K de ¼ 30 MPa; hde ¼ 5 GPa;
twin
nMtwin ¼ 30; K Mtwin ¼ 260 MPa; hM ¼ 20 MPa; qMtwin ¼ 0:5; D ¼ 2 GPa

Table 8
Material parameters for the alloy tested under the multiaxial loading conditions.

EA ¼ 45 GPa; EM ¼ 30 GPa; m ¼ 0:33; Ms ¼ 243 K; gtr ¼ 0:10;


b ¼ 0:375 MPa=K;
ntr ¼ 30; K tr ¼ 10:5 MPa; nre ¼ 30; K re ¼ 11 MPa; H ¼ 2 MPa
nde ¼ 30; K de ¼ 22 MPa; hde ¼ 5 GPa;
slip
nAslip ¼ 30; K Aslip ¼ 300 MPa; hA ¼ 3 GPa; qAslip ¼ 0:5;
D ¼ 2 GPa;

3.3. Hardening rules for inelastic deformation

During the thermo-mechanical deformation of NiTi shape memory alloy, the plastic deformation resistance RaA of each slip
system in the austenite phase, the resistance Q aM of each twinning system and the detwinning deformation resistance X aM are
not constant, but evolve progressively. The evolution equation of the resistance RaA for each slip system in the austenite can
be described as:

X
12
ab  _ b 
R_ aA ¼ HAslip cA ð33-aÞ
b¼1

where, HaAslip
b
is the hardening coefficient matrix containing self- and latent-hardening, and can be written as (Hutchinson,
1976):
slip
HaAslip
b ab
¼ hA qAslip ð34-aÞ

ab
qAslip ¼ qAslip þ ð1  qAslip Þdab ð34-bÞ

slip
where, hA is the hardening modulus of dislocation slipping in the austenite phase, and qAslip is a constant determining the
magnitude of latent-hardening during the plastic deformation in the austenite phase.
Similarly, the evolution equation of the resistance Q aM for each twinning system can be obtained as:
X
11
Q_ aM ¼ ab
HMtwin f_ bM ð35-aÞ
b¼1

twin
HaMtwin
b ab
¼ hM qMtwin ð35-bÞ

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
20 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

ab
qMtwin ¼ qMtwin þ ð1  qMtwin Þdab ð35-cÞ
ab twin
where, HMtwin is the hardening coefficient matrix, hM
is the hardening modulus of twinning deformation in the martensite
phase, and qMtwin is a constant determining the magnitude of latent-hardening.
The hardening equation for the martensite detwinning can be written as:
 
X_ aM ¼ hde k_ a  ð36Þ
hde is the hardening modulus of detwinning.

3.4. Explicit scale-transition rule

The constitutive equations mentioned above are used to describe the thermo-mechanical deformation of NiTi shape
memory single crystal. To deduce the polycrystalline version of NiTi shape memory alloy, an effective scale-transition rule
is required. Hill (1965) established a well-known self-consistent model to obtain certain polycrystalline responses from that
of single crystal. However, the self-consistent model requires an iterative algorithm due to its implicit expression, which is
very time-consuming. Comparing to implicit self-consistent model, the explicit scale-transition model proposed by Kröner
(1961) is much easier for application. However, the Kröner’s model overestimates the interaction between the grains in the
polycrystalline aggregates and makes the predicted stress–strain curves higher than the experimental ones. Then, Cailletaud
and Pilvin (1994) formulated a new explicit scale-transition rule denoted as b-rule, with which the stress–strain responses of
polycrystalline aggregates can be easily obtained from that of single crystal. The b-rule contains two material parameters
and has been used to describe the complicated ratchetting of polycrystalline materials by Cailletaud and Sai (2008) and Kang
et al. (2011), and also adopt by Yu et al. (2012, 2013) to describe the thermo-mechanical properties of NiTi shape memory
alloy without considering the dynamics recovery term.
Even if it sounds too simple to predicti the complex thermo-mechanical deformation of NiTi shape memory alloy, the sim-
plified b-rule is used to obtain the thermo-mechanical properties of polycrystalline NiTi shape memory alloy in this work,
With the assumption of uniform isotropic elasticity for the polycrystalline aggregates, the local stress tensor r of certain
grain, which is assumed to be uniform within one specified grain, can be obtained from the applied uniform macro stress
tensor R by using following equations:
r ¼ R þ DðEin  ein Þ ð37Þ

ein ¼ em þ ep ð38Þ
where, Ein ¼ ½ein  represents the inelastic strain of polycrystalline aggregates obtained by the volume average of each single
crystal grain. D is a material parameter controlling the heterogeneity of stress field in the inter-granular scale. The predicted
heterogeneity in polycrystalline aggregates increases with the increasing D. If D = 0, the heterogeneity in the inter-granular
scale disappears and then a uniform stress field is obtained in the whole polycrystalline aggregates. It should be noted that
an explicit integration algorithm is adopted in the numerical implementation of the proposed model for simplification, since
we only focus on the stress integration, rather than the implicit finite element implementation (where the consistent tangent
modulus tensor should be deduced by using an implicit stress integration algorithm). The outline of explicit stress integra-
tion algorithm can be found in Appendix B.

4. Verification and discussion

4.1. Determination of material parameters

All the material parameters used in the proposed model of single crystal can be determined from four kinds of typical
experimental data of NiTi shape memory single crystals, as shown in Fig. 11(a) and (b). Since the viscosity is not obvious
for the polycrystalline NiTi shape memory alloy, the parameters ntr ; nre ; nde ; nAslip and nMtwin used in the single crystal model
can be set to be a large number, i.e., 30. The parameters reflecting the latent hardening (qAslip and qMtwin Þ in the scale of single
crystal are all set as 0.5 for simplicity. Furthermore, since the elastic responses of polycrystalline aggregates are assumed to
be uniform in the simplified explicit scale-transition rule, the elastic moduli EA and EM can be obtained directly from the cor-
responding experimental data of polycrystalline NiTi shape memory alloy such as that shown in Fig. 2(a). v is set as 0.33 by
referring to Thamburaja et al. (2005).

4.1.1. Determination of K tr ; l; H and T 0


From the solid stress–strain curve in Fig. 11(a), it can be found that the start stress of the forward and reverse transfor-
mation processes are ram and rma1 , respectively. It is noted that the parameter l controls the start stress of forward trans-
formation at a specific temperature; K tr controls the width of hysteresis loop; H controls the hardening of transformation.
Thus, the three parameters can be determined from the experimental data of ram ; rma1 and Htr , respectively. As mentioned
above, T 0 is the referential temperature. Similar to the work done by Lagoudas and Entchev (2004), T 0 can be written as:

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 21

pm
(a) Af<T<Ad σpm H

EA H
de

Stress
tr
H

σam σma1

σma2
EM

Strain

(b)
HpA
σpA
Stress

T<Ms
σre T>Ad

Strain

Fig. 11. Illustration for the determination of material parameters from the supposed stress–strain data of NiTi shape memory single crystals: (a) under
different stress levels at temperature Af < T < Ad ; (b) at temperature T < Ms and T > Ad .

K tr
T 0 ¼ Ms þ ð39Þ
l
where, M s is the start temperature of martensite transformation. Thus, the material parameter T 0 can be obtained from other
parameters.

twin
4.1.2. Determination of K Mtwin ; hM and A
From the dash-dot stress–strain curve shown in Fig. 11(a), it is seen that the start stress of twinning in martensite phase is
rpm . Since the plastic deformation can restrain the reverse transformation from the induced martensite to austenite phase,
twin
the start stress of reverse transformation shall be changed as rma2 . Thus, the parameters K Mtwin ; hM and A can be deter-
mined from the experimental data of rpm ; Hpm and rma2 , respectively.

4.1.3. Determination of K re ; K de and hde


From the solid stress–strain curve shown in Fig. 11(b), the start stress of martensite reorientation is rre , and the param-
eter K re controls its value. So, K re can be determined from the experimental data of rre . Thamburaja (2005) reported that the
start stress of martensite detwinning was higher than that of martensite reorientation, and then the value of K de is assumed
to be lager than that of K re according to its physical nature (i.e., it occurs after the reorientation of twinned martensite). The
dash-dot stress–strain curve in Fig. 11(a) also gives the modulus of detwinning, i.e., Hde . Thus, hde can be obtained by fitting
with Hde .

slip
4.1.4. Determination of K Aslip and hA
From the dash-dot stress–strain curve in Fig. 11(b), it can be observed that when the test temperature is higher than Ad ,
the dislocation slip in austenite phase occurs in certain extent before the martensite transformation. The start stress of dis-
slip
location slip in austenite is rpA , and the modulus is HpA . Thus, the parameter K Aslip and hA can be obtained from this stress–
strain curve by considering the residual strain after completely unloading.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
22 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

It should be noted that the above procedure for determining the material parameters can only be accomplished using the
corresponding experimental data of NiTi shape memory single crystals, as supposed in Fig. 11(a) and (b). However, such
experimental data now are unavailable yet. Therefore, in fact, all the material parameters used in the proposed model are
determined directly from the corresponding experimental data of polycrystalline NiTi shape memory alloy by the trial-
and-error method and keeping the physical nature of each parameter unchanged.

4.2. Simulations and predictions in tension-unloading and at different temperatures

The proposed model is used to simulate the experimental data obtained by Shaw and Kyriakides (1995) under the ten-
sion-unloading conditions and at different temperatures (i.e., 283, 333, 343, 353 and 363 K, respectively). Total of 40 ran-
domly oriented single crystal grains are used to represent the polycrystalline aggregates. The material parameters used in
the proposed model are determined from only the experimental results shown in Figs. 1(a) and 2(b), (d) and (e) and are listed
in Table 6. Other experimental results shown in Figs. 1 and 2 are predicted by the proposed model. It should be noted that,
for the twinned martensite at 283 K, the initial volume fraction of each martensite is set as 1/24 and the macroscopic stain is
zero.
The simulated and predicted uniaxial stress–strain responses of the polycrystalline NiTi shape memory alloy by the pro-
posed model at different temperatures are shown in Figs. 1 and 2. It can be found that the simulations and predictions agree
with the experimental ones very well. Furthermore, the shape memory effect of the NiTi alloy is also predicted by the pro-
posed model as shown in Fig. 1d. When the applied stress is low (e.g., <800 MPa), a perfect shape memory effect is observed
and the inelastic strain caused by the reorientation and detwinning of twinned martensite is completely recovered in the
subsequent heating process, because no twinning deformation occurs in the martensite. However, when the applied stress
is higher than 1000 MPa, only a part of the inelastic strain is recovered by the subsequent heating, because the plastic strain
is produced by the twinning of martensite. It is noted that the magnitude of recoverable strain decreases with the increasing
applied stress due to the restraint of martensite plasticity to the reverse transformation (i.e., from martensite to austenite
phase). The cyclic deformation of NiTi shape memory alloy shown in Fig. 1(c) will be discussed in Section 4.5.

4.3. Simulations and predictions under multi-stepped loading

The proposed model is used to simulate and predict the experiments obtained by Miyazaki et al. (1981). Since the test
temperatures are not high enough to cause the plastic deformation of austenite phase, the dislocation slipping in the aus-
tenite is neglected here. The material parameters used in the proposed model are listed in Table 7, which are obtained from
the stress–strain curves of the 3rd and 7th cycles in Fig. 3(a) and 3rd cycle in Fig. 4(a). Forty randomly oriented single crystal
grains are used as the same as that in Section 4.2. Fig. 3b shows the simulated and predicted results of the polycrystalline
NiTi shape memory alloy obtained in the multi-stepped loading–unloading test at 243 K, where the alloy originally consists
of austenite phase; while Fig. 4(b) gives the corresponding results obtained at 173 K (originally martensite phase), and the
initial volume fraction of each martensite variant is set as 1/24. It is seen that the simulations and predictions are in good
agreement with the experiment ones, since the martensite transformation, the reorientation and detwinning of twinned
martensite, and the plastic deformation presented in the induced martensite are considered in the proposed model.

4.4. Simulations and predictions under non-proportionally multiaxial loading

In Sections 4.2 and 4.3, the capability of the proposed model to describe the uniaxial thermo-mechanical responses of NiTi
shape memory alloy is verified. Here, the model is checked again by simulating and predicting the non-proportionally mul-
tiaxial stress–strain responses of super-elastic NiTi shape memory alloy.
100 randomly oriented single crystal grains are used to represent the polycrystalline aggregates. Since the maximum ap-
plied stress is not high enough to cause the plastic deformation of induced martensite, the twinning deformation of martens-
ite phase is neglected in the simulations and predictions. For this kind of super-elastic NiTi shape memory alloy, the material
slip
parameters EA ; EM ; v ; K tr ; l; H; K Aslip and hA can be determined from the uniaxial stress–strain curve shown in Fig. 6. The
re de
parameters K ; K and hde are obtained from the experimental data of multiaxial response with square loading path, since
the uniaxial stress–strain curve of the NiTi alloy has not tested at the temperature T < M f yet. The material parameters used
in the proposed model are listed in Table 8. The experimental results obtained with other non-proportionally multiaxial
loading paths are predicted by the proposed model.
Shown in Figs. 7–10, the non-proportionally multiaxial stress–strain responses of super-elastic NiTi shape memory alloy
under the stress-controlled tension–torsion cyclic loading conditions with different loading paths are simulated and pre-
dicted by the proposed model. The simulated and predicted results are in good agreement with the experimental ones. As
mentioned above, the axial strain (point o00 in Fig. 10(a)) does not return to zero due to the plastic deformation of austenite
phase. This phenomenon is also described well by the proposed model by considering the plasticity in the austenite.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 23

(a) 300
Load case
(b) 450 Load case
400
250 120±120MPa 200±200MPa
350

200 300

Stress, MPa
Stress, MPa

250
150
200
50th cycle
100 150

50th cycle 100


50
50

0 0
0 2 4 6 0 2 4 6 8
Strain, % Strain, %

(c) 500 (d) 700


Load case Load case
235±235MPa 600 300±300MPa
400
500
Stress, MPa

Stress, MPa
300
400

200 300

50th cycle 200


100
100
50th cycle
0 0
0 2 4 6 8 0 2 4 6 8 10 12 14
Strain, %

(e) 800
Load case (f) 700
Load case
350±350MPa 600 425±225MPa
600
500
Stress, MPa

Stress, MPa

400
400
300

200 200

100 50th cycle


50th cycle
0 0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16
Strain, % Strain, %
Strain, %

(g) 700
Load case
600 550±100MPa

500
Stress, MPa

400
50th cycle
300

200

100

0
0 2 4 6 8 10 12 14 16
Strain, %

Fig. 12. Predicted results for the ratchetting behaviors of NiTi shape memory alloy (initially twinned martensite): (a) 120 ± 120 MPa; (b) 200 ± 200 MPa; (c)
235 ± 235 MPa; (d) 300 ± 300 MPa; (e) 350 ± 350 MPa; (f) 425 ± 225 MPa; (g) 550 ± 100 MPa.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
24 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

4.5. Discussions

The proposed model has been verified by describing various thermo-mechanical experimental results of NiTi shape mem-
ory alloy, including that obtained in the tests with the uniaxial and non-proportionally multiaxial loading paths. In this sub-
section, the cyclic stress–strain responses of NiTi shape memory alloy under the stress-controlled cyclic loading conditions
are predicted, and then the capability of the proposed model is discussed by comparing the predicted results with some
existing experimental data (Kang et al., 2009).
For the NiTi shape memory alloy with initially twinned martensite phase (i.e., addressing the shape memory effect), all
the parameters (listed in Table 6) and the aggregates of single crystal grains used in the predictions are simply set to be the
same as that used in Section 4.3. We do not determine all the material parameters used in the proposed model from the
practical experimental data done by Kang et al. (2009). The predicted results are shown in Fig. 12 for various stress-con-

(a) 7.0
Experiments for 120±120MPa
6.5 Predictions for 120±120MPa
Ratchetting strain, %

Predictions for 200±200MPa


6.0

5.5

5.0

4.5

4.0
0 10 20 30 40 50
Number of cycles

(b) 16
14
Experiments for 235±235MPa
Experiments for 300±300MPa
Ratchetting strain, %

12
Experiments for 350±350MPa

10
Predictions for 235±235MPa
Predictions for 300±300MPa
8 Predictions for 350±350MPa

6
0 10 20 30 40 50
Number of cycles

(c) 15
Experiments for 425±225MPa
Experiments for 550±100MPa
14
Ratchetting strain, %

13

12 Predictions for 425±225MPa


Predictions for 550±100MPa
11

10

9
0 10 20 30 40 50
Number of cycles

Fig. 13. Predicted and experimental results for the ratchetting strain vs. number of cycles: (a) for the cyclic tension-unloading cases with peak stress lower
than 475 MPa; (b) for the cyclic tension-unloading cases with peak stress higher than 475 MPa; (c) for the cyclic tension–tension cases.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 25

600 Peak stress=600MPa

500

400

Stress, MPa
300

200
50th cycle
100

0
0 2 4 6
Strain, %
Fig. 14. Predicted results for the cyclic stress–strain responses of NiTi shape memory alloy (initially austenite) in the cyclic tension-unloading case with the
peak stress of 600 MPa.

trolled cyclic loading cases, i.e., 120 ± 120, 200 ± 200, 235 ± 235, 300 ± 300, 350 ± 350, 425 ± 225 and 550 ± 100 MPa, respec-
tively. The loading case, such as 550 ± 100 MPa means the applied mean stress is 550 MPa and stress amplitude is 100 MPa.
The prescribed number of cycles is 50.
From Figs. 12 and 13, it is seen that: (1) A measurable ratchetting behaviour is predicted by the proposed model for the
NiTi shape memory alloy, especially for the cases with higher applied stress levels (as shown in Fig. 12(c)–(f)), and the evo-
lution of predicted ratchetting is in a good agreement with the corresponding experimental results obtained by Kang et al.
(2009) (such as a typical one shown in Fig. 1(c)). However, the magnitude of predicted ratchetting strain is somewhat dif-
ferent from the experimental ones (as shown in Fig. 13) due to the different materials used in the prediction (from Miyazaki
et al., 1981) and experiment (from Kang et al., 2009). (2) When the applied peak stress (i.e., the sum of applied mean stress
and stress amplitude) is relatively low (e.g., lower than 400 MPa, which is approximately the finish stress of martensite reori-
entation and detwinning), the plastic deformation of martensite phase does not occur. The ratchetting is mainly caused by
the viscosity of martensite reorientation and detwinning. However, when the applied peak stress becomes to be equal to
400 MPa, the ratchetting does not occur because the martensite reorientation and detwinning is ended and the martensite
deforms elastically, as shown in Fig. 12(b). When the applied peak stress further becomes to be 475 MPa, which is higher
than the yielding stress of martensite phase (about 450 MPa), a measurable ratchetting occurs again (as shown in
Fig. 12(c)) due to the visco-plasticity of martensite phase caused by the twinning mechanism. When the applied peak stress
is higher than 450 MPa, the predicted ratchetting strain increases with the increasing applied peak stress (as shown in
Fig. 12(d) and (e)). This predicted evolution rule of ratchetting agrees with the experimental result very well, as shown in
Fig. 13. (3) As shown in Fig. 12(f) and (g), apparent ratchetting also occurs in the cyclic tension–tension loading cases due
to the visco-plasticity of martensite phase, and the predicted results are in fairly good agreement with the experimental ones
even if the materials are not exactly the same.
To sum up, the proposed model can provide a reasonable prediction to the cyclic deformation of the NiTi shape memory
alloy (which is originally martensite phase) presented under the stress-controlled cyclic tension-unloading and tension–ten-
sion tests, which cannot be described by the micromechanical model proposed by Yu et al. (2013).
For the NiTi shape memory alloy being originally austenite phase, its cyclic deformation, especially the transformation
ratchetting presented under the stress-controlled cyclic loading conditions, was reasonably described by the micromechan-
ical constitutive model proposed by Yu et al. (2013). In the previous model (Yu et al., 2013), two kinds of microscopic mech-
anisms for the transformation ratchetting of super-elastic NiTi shape memory alloy, i.e., the accumulated residual martensite
and the cyclic accumulation of transformation-induced plasticity caused by the irreversible friction at the interfaces between
austenite and martensite phases during the cyclic transformation, were reasonably considered. However, since the two
mechanisms of transformation ratchetting mentioned above are not included in the proposed model in this work, the trans-
formation ratchetting of the NiTi shape memory alloy (originally austenite phase) observed by Kang et al. (2009) cannot be
reasonably described, as shown in Fig. 14. In the prediction, the material parameters and the aggregates of single crystal
grains are set to be the same as that used in Section 4.4. The predicted stress–strain responses for the cyclic tension-unload-
ing cases are overlapped from 2nd to 50th cycles, and no cyclic accumulation of peak and valley strains occurs. It implies that
certain new mechanisms related to the transformation ratchetting should be additionally included to extend the proposed
model to describe the transformation ratchetting of super-elastic NiTi shape memory alloy in the future work.

5. Conclusion

A crystal plasticity based micromechanical constitutive model of NiTi shape memory alloy is constructed by considering
the martensite transformation, martensite reorientation and detwinning and the plasticity occurred in the austenite and

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
26 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

martensite phases. The Helmholtz free energy for the RVE of single crystal is divided into three parts, i.e., the elastic energy,
the energy related to the martensite transformation, reorientation and detwinning, and the energy related to the dislocation
slipping in the austenite and twinning in the martensite, respectively. Based on thermodynamics, the dissipative inequality
for every one of inelastic deformation mechanisms is obtained by the proposed Helmholtz free energy. The evolution equa-
tion of each internal variable is obtained in the power-law form. Also, the elastic property difference between the austenite
and martensite phases, and the restraint of martensite plasticity to the reverse transformation are considered in the pro-
posed model. The polycrystalline model is extended from the single crystal one by a simplified explicit scale-transition rule.
The proposed constitutive model reasonably describes the thermo-mechanical responses of NiTi shape memory alloy, such
as super-elasticity, shape memory effect, dislocation slip in the austenite and twinning in the martensite. The predicted re-
sults are in good agreement with the corresponding uniaxial and non-proportionally multiaxial experimental data, as well as
the uniaxial ratchetting of NiTi shape memory alloy being originally martensite phase. The model should be further im-
proved to predict the transformation ratchetting of super-elastic NiTi shape memory alloy in the future work.

Acknowledgments

Financial supports by National Natural Science Foundation of China (11025210; 11202171), the special fund for Sichuan
Provincial Youth Science and Technology Innovation Team (2013, China), the Cultivation Foundation of Excellent Doctoral
Dissertation of Southwest Jiaotong University (2013) and the Fundamental Research Funds for the Central Universities (Chi-
na) are appreciated.

Appendix A. Thermodynamics driving forces

In this appendix, the derivation of driving force for each inelastic mechanism is given.

A.1. Driving forces for martensite transformation, reorientation and detwinning

Substituting Eqs. (3-a), (4), (9), (15) and (16) into Eq. (24-a) yields:
24 
X  X
24 X
23
1
Cm ¼ r : Ka  lðT  T 0 Þ  Hna  ee : DC : ee n_ atr þ ðr : Sij Þk_ ij
a¼1
2 j>i i¼1

24 
X  24 n
X  o
1
 lðT  T 0 Þ þ Hna þ ee : DC : ee n_ are þ r : ½n_ a ðka  ka0 ÞPade þ na k_ a Pade   na X aM k_ a  ðA:1Þ
a¼1
2 a¼1

where DC ¼ CM  CA . It is noted that in the process of reorientation the total volume fraction of martensite keeps being a
constant, which leads to:
X
24
n_ are ¼ 0: ðA:2Þ
a¼1
P24 P24 a
Thus, the term a¼1 lðT  T 0 Þ þ 12 ee : DC : ee n_ are ¼ lðT  T 0 Þ þ 12 ee : DC : ee _
a¼1 nre ¼ 0. And then, Eq. (A.1) can be reduced
as:
24 
X  X
24 X
23 X
24
1
Cm ¼ r : Ka  lðT  T 0 Þ  Hna  ee : DC : ee n_ atr þ ðr : Sij Þk_ ij  Hna n_ are
a¼1
2 j>i i¼1 a¼1
X
24  
þ fr : ½n_ a ðka  ka0 ÞPade þ na k_ a Pade   na X aM k_ a g ðA:3Þ
a¼1

Considering Eqs. (6) and (8) and combining the terms with the same rates such as n_ atr ; k_ ij and k_ a , Eq. (A.3) can be re-written
as:
X24 1

Cm ¼ r : ½Ka þ ðka  ka0 ÞPade   lðT  T 0 Þ  Hna  ee : DC : ee n_ atr
a¼1 2
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
transformation
X24 X23 n h X24 aij
i X24
aij
o
þ j>i i¼1
r : Sij þ a¼1
ðka  ka0 Þk Pade  a¼1
k Hna k_ ij
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
reorientation
X24 h  i
þ a¼1 ðr : Pade Þna k_ a  na X aM k_ a  P 0 ðA:4Þ
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
detwinning

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 27

Three terms in Eq. (A.4) represent the dissipation parts related to the martensite transformation, reorientation and detwin-
ning, respectively. More strictly, it is assumed that each dissipation part is non-negative. Thus, it yields
24
X 
1
Ctr ¼ r : ½Ka þ ðka  ka0 ÞPade   lðT  T 0 Þ  Hna  ee : DC : ee n_ atr P 0 ðA:5-aÞ
a¼1
2
( " # )
X
24 X
23 X
24
aij
X
24
aij
Cre ¼ r : Sij þ ðka  ka0 Þk Pade  k Hna k_ ij P 0 ðA:5-bÞ
j>i i¼1 a¼1 a¼1

24 h
X  i
Cde ¼ ðr : Pade Þna k_ a  na X aM k_ a  P 0 ðA:5-cÞ
a¼1

Eq. (A.5-a) is the dissipative inequality satisfied during the martensite transformation. Furthermore, it is assumed that the
dissipation for each martensite variant is non-negative. So Eq. (A.5-a) can be divided into 24 equations, i.e.,

1
Catr ¼ r : ½Ka þ ðka  ka0 ÞPade   lðT  T 0 Þ  Hna  ee : DC : ee n_ atr P 0 a ¼ 1; 2; . . . ; 24 ðA:6Þ
2
Thus, by Eq. (A.6), the general thermodynamic force conjugated to n_ atr is:
1
patr ¼ r : ½Ka þ ðka  ka0 ÞPade   lðT  T 0 Þ  Hna  ee : DC : ee ðA:7Þ
2
Eq. (A.5-b) is the dissipative inequality for the martensite reorientation. Similarly, it is assumed that the dissipation for
each reorientation system is non-negative. So Eq. (A.5-b) can be divided into 276 equations, i.e.,
( " # )
X
24
aij
X
24
aij
r : Sij þ ðka  ka0 Þk Pade  k Hna k_ ij P 0 i ¼ 1; 2 . . . 23; j ¼ i þ 1; i þ 2 . . . 24 ðA:8Þ
a¼1 a¼1

Thus, by Eq. (A.8), the thermodynamic driving force conjugated to k_ ij is:


" #
X
24
aij
X
24
aij
pijre ¼ r : Sij þ ðka  ka0 Þk Pade  k Hna ðA:9Þ
a¼1 a¼1

By using jXj = Xsign(X), Eq. (A.8) can be rewritten as:


 ij  ij
p k_ signðpij Þ P 0 ðA:10Þ
re re
 
 
Thus, pijre  can be taken as the thermodynamic driving force conjugated to k_ ij signðpijre Þ.
Eq. (A.5-c) is the dissipative inequality satisfied during the detwinning of martensite. It is noted that the volume fraction
of each martensite variant satisfies. Thus 0  na  1, Eq. (A.5-c) can be re-written as:
24 h
X  i
Cde ¼ ðr : Pade Þk_ a  X aM k_ a  P 0 ðA:11Þ
a¼1

Also, it is assumed that the dissipation for each detwinning system is non-negative. So Eq. (A.11) can be divided into 24 equa-
tions, i.e.,
 
Cade ¼ ðr : Pade Þk_ a  X aM k_ a  P 0 ðA:12Þ
In order to satisfy the constraint of thermodynamics to the detwinning of martensite, it is assumed that:

signðk_ a Þ ¼ signðr : Pade Þ ðA:13Þ


Substituting Eq. (A.13) into Eq. (A.12) and using jXj = Xsign(X), we can obtain:
  

Cade ¼ r : Pade   X aM k_ a signðr : Pade Þ P 0 a ¼ 1; 2; . . . 24 ðA:14Þ
 
where, pa
de ¼ r : Pade   X aM is the general thermodynamic force conjugated to k_ a signðr : Pade Þ.

A.2. Driving forces for austenite and martensite plasticity

Substituting Eq. (12-a), (13-a) and (17) into Eq. (24-b), it yields
X12 h  i X11
Cp ¼ ð1  nÞ b¼1
ðr : PbA Þc_ bA  RbA c_ bA  þ n b ¼1 ðr : KbM1  Q bM1 Þf_ bM1 P 0 ðA:15Þ
1
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
austenite martensite

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
28 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

Eq. (A.15) is the total dissipative inequality for the dislocation slipping in the austenite and twinning in the martensite phase.
More strictly, it is assumed that the dissipation for each term is non-negative. Thus, the dissipative inequality can be sepa-
rated as:
12 h
X  i
CAp ¼ ð1  nÞ ðr : PbA Þc_ bA  RbA c_ bA  P 0 ðA:16aÞ
b¼1

X
11
CM
p ¼ n ðr : KbM1  Q bM1 Þf_ bM1 P 0 ðA:16bÞ
b1 ¼1

Similar to the detwinning of martensite, in order to satisfy the constraint of thermodynamics, it is assumed that:

signðc_ bA Þ ¼ signðr : PbA Þ ðA:17Þ


Using the same procedure as that for the detwinning of martensite, the thermodynamic driving forces for c_ bA signðr : PbA Þ and
f_ bM1 can be given as:
 
F bAslip ¼ r : PbA   RbA ðA:18-aÞ

F bMtwin
1
¼ r : KbM1  Q bM1 ðA:18-bÞ

Appendix B. Explicit numerical integration procedure

In this appendix, the explicit numerical integration procedure is given. Considering the interval from step n to n + 1 with
the time increment Dt. For a grain in the polycrystalline aggregates, it is assumed that the stress tensor rn , strain tensor en ,
temperature T n and other internal variables such as xn ¼ fðnatr Þn ; ðkij Þn ; ðka Þn ; ðcbA Þn ; ðfMb1 Þn ; ðX aM Þn ; ðRbA Þn ; ðQ bM1 Þn g and
einn ; ðepA Þn ; ðepM Þn in the step n are all known. It is noted that only the quasi-static loading conditions are considered in this
work. The temperature is uniform in the polycrystalline aggregates. For a given macroscopic stress tensor increment DR
and temperature increment DT, the macroscopic stress tensor and temperature in the step n + 1 can be written as:
Rnþ1 ¼ Rn þ DR; T nþ1 ¼ T n þ DT. The unknown quantities are: xnþ1 ¼ fðepM Þnþ1 ; ðnatr Þnþ1 ; ðkij Þnþ1 ; ðka Þnþ1 ; ðcbA Þnþ1 ; ðfMb1 Þnþ1 ;
ðX aM Þnþ1 ; ðRbA Þnþ1 ; ðQ bM1 Þnþ1 g and enþ1 ; ein p
nþ1 ; ðeA Þnþ1 .

Step 1: Calculating the local stress


By the simplified explicit scale-transition rule, it is obtained:

Ein in
n ¼ ben c ðB:1Þ

rnþ1 ¼ Rnþ1 þ DðEinn  einn Þ ðB:2Þ


Step 2: Calculating the elastic strain tensor and the driving forces

eenþ1 ¼ C1 ðnn Þ : rnþ1 ðB:3Þ

1 e
ðpare Þnþ1 ¼ rnþ1 : fKa þ ½ðka Þn  ka0 Pade g  lðT nþ1  T 0 Þ  Hnn  e : DC : eenþ1 ðB:4Þ
2 nþ1
( )
X
24
aij
X
24
aij
ðp ij
re Þnþ1 ¼ rnþ1 : ij
S þ ½ðka Þn  ka0 k Pade   k Hðna Þn ðB:5Þ
a¼1 a¼1

 
ðpade Þnþ1 ¼ rnþ1 : Pade   ðX aM Þn ðB:6Þ
 
ðF bAslip Þnþ1 ¼ rnþ1 : PbA   ðRbA Þn ðB:7Þ

ðF bMtwin
1
Þnþ1 ¼ rnþ1 : KbM1  ðQ bM1 Þn ðB:8Þ

Step 3: Integrating the evolution equations


(1) Martensite transformation
 a 
ðp Þ ntr
Dnatr ¼  tr trnþ1  Dt whenever ðpatr Þnþ1 P 0 0 6 ðnatr Þn 6 1 0 6 ðna Þn 6 1 0 6 ðnÞn 6 1 ðB:9-aÞ
K

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 29

 ntr
 ðpatr Þnþ1 
a   p  Dt
Dntr ¼  tr whenever ðpatr Þnþ1 P 0 0 6 ðnatr Þn 6 1 0 6 ðna Þn 6 1 0 6 ðnÞn 6 1 ðB:9-bÞ
K þ AðeM Þn 

Dnatr ¼ 0 other conditions ðB:9-cÞ


(2) Martensite reorientation
 
ðpij Þ nre
ij re nþ1 
Dk ¼   sign½ðpijre Þnþ1 Dt whenever 0 6 ðni Þn 6 1 0 6 ðnj Þn 6 1 ðB:10-aÞ
 K re 

Dkij ¼ 0 other conditions ðB:10-bÞ


(3) Martensite detwinning

nde
ðpade Þnþ1
Dka ¼ signðrnþ1 : Pade ÞDt whenever ðna Þn > 0 ðB:11-aÞ
K de

Dka ¼ 0 other conditions ðB:11-bÞ


(4) Austenite plasticity
* b +nAslip
ðF Aslip Þnþ1
DcbA ¼ signðrnþ1 : PbA ÞDt whenever nn < 1 ðB:12-aÞ
K Aslip

DcbA ¼ 0 other conditions ðB:12-bÞ


(5) Martensite plasticity
* b +nMtwin
ðF Mtwin
1
Þnþ1
DfMb1 ¼ Dt whenever nn > 0 ðB:13-aÞ
g btwin
1
K Mtwin

DfMb1 ¼ 0 other conditions ðB:13-bÞ


Step 4: Calculating the hardening variables
X
12
ab  
DRaA ¼ HAslip DcbA  ðB:14Þ
b¼1

X
11
DQ aM ¼ ab
HMtwin DfMb ðB:15Þ
b¼1

DX aM ¼ hde jDka j ðB:16Þ

Step 5: Updating the variables

xnþ1 ¼ xn þ Dx ðB:17Þ

X
24
etrnþ1 ¼ ðnatr Þnþ1 Ka ðB:18Þ
a¼1

X
24 X
23
erenþ1 ¼ ðkij Þnþ1 Sij ðB:19Þ
j>i i¼1

X
24
ede
nþ1 ¼ ðna Þnþ1 ½ðka Þnþ1  ka0 Pade ðB:20Þ
a¼1

X
12
ðepA Þnþ1 ¼ ðepA Þn þ ð1  nnþ1 Þ DcbA PbA ðB:21Þ
b¼1

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
30 C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx

X
11
ðepM Þnþ1 ¼ ðepM Þn þ nnþ1 DfMb1 KbM1 ðB:22Þ
b1 ¼1

p p
einnþ1 ¼ etrnþ1 þ erenþ1 þ ede
nþ1 þ ðeA Þnþ1 þ ðeM Þnþ1 ðB:23Þ

enþ1 ¼ eenþ1 þ einnþ1 ðB:24Þ

References

Abeyaratne, R., Kim, S.J., 1997. Cyclic effects in shape memory alloys: a one-dimensional continuum model. Int. J. Solids Struct. 34, 3273–3289.
Anand, L., Gurtin, M.E., 2003. Thermal effects in the superelasticity of crystalline shape-memory materials. J. Mech. Phys. Solids 51, 1015–1058.
Arghavani, J., Auricchio, F., Naghdabadi, R., 2011. A finite strain kinematic hardening constitutive model based on Hencky strain: general framework,
solution algorithm and application to shape memory alloys. Int. J. Plasticity 27, 940–961.
Auricchio, F., Taylor, R.L., Lubliner, J., 1997. Shape-memory alloys: macromodelling and numerical simulations of the superelastic behavior. Comput.
Methods Appl. Mech. 146, 281–312.
Auricchio, F., Marfia, S., Sacco, E., 2003. Modelling of SMA materials: training and two way memory effects. Comput. Struct. 81, 2301–2317.
Auricchio, F., Reali, A., Stefanelli, U., 2007. A three-dimensional model describing stress-induced solid phase transformation with permanent inelasticity. Int.
J. Plasticity 23, 207–226.
Bo, Z., Lagoudas, D.C., 1999a. Thermomechanical modelling of polycrystalline SMAs under cyclic loading. Part I: Theoretical derivations. Int. J. Eng. Sci. 37,
1089–1140.
Bo, Z., Lagoudas, D.C., 1999b. Thermomechanical modelling of polycrystalline SMAs under cyclic loading. Part III: Evolution of plastic strains and two-way
memory effect. Int. J. Eng. Sci. 37, 1141–1173.
Bo, Z., Lagoudas, D.C., 1999c. Thermomechanical modelling of polycrystalline SMAs under cyclic loading. Part IV: Modelling of minor hysteresis loops. Int. J.
Eng. Sci. 37, 1174–1204.
Brinson, L.C., 1993. One-dimensional constitutive behavior of shape memory alloys: thermomechanical derivation with non-constant material functions
and redefined martensite internal variable. J. Intel. Mater Syst. Struct. 4, 229–242.
Cailletaud, G., Pilvin, P., 1994. Utilisation de modeles polycristallins pour le calcul par elements finis. Revue Eur. Elements Finis. 3, 515–541.
Cailletaud, G., Sai, K., 2008. A polycrystalline model for the description of ratchetting: effect of intergranular and intragranular hardening. Mater. Sci. Eng. A
480, 24–39.
Chemisky, Y., Duval, A., Patoor, E., Ben Zineb, T., 2011. Constitutive model for shape memory alloys including phase transformation, martensitic
reorientation and twins accommodation. Mech. Mater. 43, 361–376.
Gall, K., Sehitoglu, H., 1999. The role of texture in tension–compression asymmetry in polycrystalline Ni–Ti. Int. J. Plasticity 15, 69–92.
Gall, K., Lim, T.J., McDowell, D.L., Sehitoglu, H., Chumlyakov, Y.I., 2000. The role of intergranular constraint on the stress-induced martensitic transformation
in textured polycrystalline NiTi. Int. J. Plasticity 16, 1189–1214.
Gao, X.J., Huang, M., Brinson, L., 2000. A multivariant micromechanical model for SMAs. Part 1. Crystallographic issues for single crystal model. Int. J.
Plasticity 16, 1345–1369.
Hartl, D.J., Chatzigeorgiou, G., Lagoudas, D.C., 2010. Three-dimensional modeling and numerical analysis of rate-dependent irrecoverable deformation in
shape memory alloys. Int. J. Plasticity 26, 1485–1507.
Hill, R., 1965. Continuum micromechanics of elastoplastic polycrystals. J. Mech. Phys. Solids 13, 89–101.
Huang, M., Gao, X.J., Brinson, L., 2000. A multivariant micromechanical model for SMAs. Part 2: Polycrystal model. Int. J. Plasticity 16, 1371–1390.
Hutchinson, J.W., 1976. Elastic-Plastic behavior of polycrystalline metals and composites. Proc. R. Soc. Ser. A 319, 247–272.
Kang, G.Z., Kan, Q.H., Qian, L.M., Liu, Y.J., 2009. Ratchetting deformation of superelastic and shape memory NiTi Alloys. Mech. Mater. 41, 139–153.
Kan, Q.H., Kang, G.Z., 2010. Constitutive model for uniaxial transformation ratchetting of super-elastic NiTi shape memory alloy at room temperature. Int. J.
Plasticity 26, 441–465.
Kang, G.Z., Bruhns, O.T., Sai, K., 2011. Cyclic polycrystalline visco-plastic model for ratchetting of 316L stainless steel. Comput. Mater. Sci. 50, 1399–1405.
Khalil, W., Saint-Sulpice, L., Arbab-Chirani, S., Bouby, C., Mikolajczak, T., Zineb, B., 2013. Experimental analysis of Fe-based shape memory alloy behavior
under thermo-mechanical cyclic loading. Mech. Mater., 10.1016/j.mechmat. 2013.04.002.
Kröner, E., 1961. Zur plastischen verformung des vielkristalls. Acta Metall. 9, 155–161.
Lagoudas, D.C., Bo, Z., 1999. Thermomechanical modelling of polycrystalline SMAs under cyclic loading. Part II: Material characterization and experimental
results for a stable transformation cycle. Int. J. Eng. Sci. 37, 1205–1249.
Lagoudas, D.C., Entchev, P.B., 2004. Modelling of transformation-induced plasticity and its effect on the behaviour of porous shape memory alloys. Part I:
Constitutive model for fully dense SMAs. Mech. Mater. 36, 865–892.
Lagoudas, D.C., Entchev, P.B., Popov, P., Patoor, E., Brinson, L.C., Gao, X.J., 2006. Shape memory alloys. Part II: Modelling of polycrystals. Mech. Mater. 38,
430–462.
Lagoudas, D., Hartl, D., Chemisky, Y., Machado, L., Popov, P., 2012. Constitutive model for the numerical analysis of phase transformation in polycrystalline
shape memory alloys. Int. J. Plasticity 32–33, 155–183.
Lexcellent, C., Leclerq, S., Gabry, B., Bourbon, G., 2000. The two way shape memory effect of shape memory alloys: an experimental study and a
phenomenological model. Int. J. Plasticity 16, 1155–1168.
Lubliner, J., Auricchio, F., 1996. Generalized plasticity and shape memory alloys. Int. J. Solids Struct. 33, 991–1003.
Manchiraju, S., Anderson, P.M., 2010. Coupling between martensitic phase transformations and plasticity: a microstructure-based finite element model. Int.
J. Plasticity 26, 1508–1526.
McKelvey, A.L., Ritchie, R.O., 1999. On the temperature dependence of the superelastic strength and the predicition of the theoretical uniaxial
transformation strain in Nitinol. Philos. Mag. A 80, 1759–1768.
Miyazaki, S., Otsuka, K., Suzuki, Y., 1981. Transformation pseudoelasticity and deformation behavior in a Ti–50.6at%Ni alloy. Scr. Metall. 53, 287–292.
Morin, C., Moumni, Z., Zaki, W., 2011. A constitutive model for shape memory alloys accounting for thermomechanical coupling. Int. J. Plasticity 27, 748–
767.
Moumni, Z., Zaki, W., Maitournam, H., 2009. Cyclic behavior and energy approach to the fatigue of shape memory alloys. J. Mech. Mater. Struct. 4, 395–411.
Nemat-Nasser, S., Guo, W., 2006. Superelastic and cyclic response of NiTi SMA at various strain rates and temperatures. Mech. Mater. 38, 463–474.
Otsuka, K., Ren, X., 2005. Physical metallurgy of Ti–Ni-based shape memory alloys. Prog. Mater. Sci. 50, 511–678.
Pan, H., Thamburaja, P., Chau, F.S., 2007. An isotropic-plasticity-based constitutive model for martensitic reorientation and shape-memory effect in shape-
memory alloys. Int. J. Solids Struct. 44, 7688–7712.
Panico, M., Brinson, L.C., 2007. A three-dimensional phenomenological model for martensite reorientation in shape memory alloys. J. Mech. Phys. Solids 55,
2491–2511.
Patoor, E., Eberhardt, A., Berveiller, M., 1996. Micromechanical modelling of superelasticity in shape memory alloys. J. Phys. IV 6. C1-277–C1-292.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012
C. Yu et al. / International Journal of Plasticity xxx (2013) xxx–xxx 31

Patoor, E., Lagoudas, D.C., Entchev, P.B., Brinson, L.X., Gao, X., 2006. Shape memory alloys. Part I: General properties and modeling of single crystals. Mech.
Mater. 38, 391–429.
Peng, X.H., Chen, B., Chen, X., Wang, J., Wang, H.Y., 2012. A constitutive model for transforamtion, reorientation and plastic deformation of shape memory
alloys. Acta Mech. Solid. Sin. 25, 285–298.
Qian, L.M., Sun, Q.P., Xiao, X.D., 2006. Role of phase transition in the unusual microwear behavior of superelastic NiTi shape memory alloy. Wear 260, 509–
522.
Reese, S., Christ, D., 2008. Finite deformation pseudo-elasticity of shape memory alloys. Constitutive modeling and finite element implementation. Int. J.
Plasticity 24, 455–482.
Saint-Sulpice, L., Arbab-Chirani, S., Calloch, S., 2009. A 3D superelastic model for shape memory alloys taking into account progressive strain under cyclic
loadings. Mech. Mater. 41, 12–26.
Saint-Sulpice, L., Arbab-Chirani, S., Calloch, S., 2012. Thermomechanical cyclic behavior modeling of Cu–Al–Be SMA materials and structures. Int. J. Solids
Struct. 49, 1088–1102.
Saleeb, A.F., Padula II., S.A., Kumar, A., 2011. A multi-axial, multimechanism based constitutive model for the comprehensive representation of the
evolutionary response of SMAs under general thermomechanical loading conditions. Int. J. Plasticity 27, 655–687.
Savi, M.A., Sá, M.A., Paiva, A., Pacheco, P.M., 2008. Tensile-compressive asymmetry influence on shape memory alloy system dynamics. Chaos Soliton. Fract.
36, 828–842.
Shaw, J.A., Kyriakides, S., 1995. Thermomechanical aspects of NiTi. J. Mech. Phys. Solids 43, 1243–1281.
Tanaka, K., Nishimura, F., Hayashi, T., Tobushi, H., Lexcellent, C., 1995. Phenomenological analysis on subloops and cyclic behaviour in shape memory alloys
under mechanical and/or thermal loading. Mech. Mater. 19, 281–292.
Thamburaja, P., 2005. Constitutive equations for martensitic reorientation and detwinning in shape-memory alloys. J. Mech. Phys. Solids 53, 825–856.
Thamburaja, P., Anand, L., 2001. Polycrystalline shape-memory alloys: effect of crystallographic texture. J. Mech. Phys. Solids 49, 709–737.
Thamburaja, P., Anand, L., 2002. Superelastic behaviour in tension-torsion of an initially-textured Ti–Ni shape-memory alloy. Int. J. Plasticity 18, 1607–1617.
Thamburaja, P., Anand, L., 2003. Thermo-mechanically coupled superelastic response of initially-textured Ti–Ni sheet. Acta. Mater. 51, 325–338.
Thamburaja, P., Pan, H., Chau, F.S., 2005. Martensitic reorientation and shape-memory effect in initially-textured polycrystalline Ti–Ni sheet. Acta Mater. 53,
3821–3831.
Thamburaja, P., Pan, H., Chau, F.S., 2009. The evolution of microstructure during twinning: constitutive equations, finite-element simulations and
experimental verification. Int. J. Plasticity 25, 2141–2168.
Tokuda, M., Ye, M., Takakura, M., Sittner, P., 1999. Thermomechanical behaviour of shape memory alloy under complex loading conditions. Int. J. Plasticity
15, 223–239.
Wang, X.M., Xu, B.X., Yue, Z.F., 2008. Micromechanical modelling of the effect of plastic deformation on the mechanical behaviour in pseudoelastic shape
memory alloys. Int. J. Plasticity 24, 1307–1332.
Yu, C., Kang, G.Z., Song, D., Kan, Q.H., 2012. Micromechanical constitutive model considering plasticity for super-elastic NiTi shape memory alloy. Comput.
Mater. Sci. 56, 1–5.
Yu, C., Kang, G.Z., Kan, Q.H., Song, D., 2013. A micromechanical constitutive model based on crystal plasticity for thermo-mechanical cyclic deformation of
NiTi shape memory alloys. Int. J. Plasticity 44, 161–191.
Zaki, W., 2010. An approach to modeling tensile–compressive asymmetry for martensitic shape memory alloys. Smart Mater. Struct. 19, 025009.
Zaki, W., 2012. An efficient implementation for a model of martensite reorientation in martensitic shape memory alloys under multiaxial nonproportional
loading. Int. J. Plasticity 37, 72–97.
Zaki, W., Moumni, Z., 2007a. A three-dimensional model of the thermomechanical behavior of shape memory alloys. Int. J. Mech. Phys. Solids 55, 2455–
2490.
Zaki, W., Moumni, Z., 2007b. A 3-D model of the cyclic thermomechanical behaviour of shape memory alloys. Int. J. Mech. Phys. Solids 55, 2427–2454.
Zhou, B., 2012. A macroscopic constitutive model of shape memory alloy considering plasticity. Mech. Mater. 48, 71–81.

Please cite this article in press as: Yu, C., et al. Crystal plasticity based constitutive model of NiTi shape memory alloy considering different
mechanisms of inelastic deformation. Int. J. Plasticity (2013), http://dx.doi.org/10.1016/j.ijplas.2013.08.012

You might also like