Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 114 (2015) 776 – 783

1st International Conference on Structural Integrity

Shape memory alloys behaviour: A review

Paulo Silva Loboa,b,∗, João Almeidaa , Luı́s Guerreiroa
a Department
of Civil Engineering, Architecture and Georesources, CEris, Instituto Superior Técnico, University of Lisbon, Av. Rovisco Pais,
1049-001 Lisbon, Portugal
b Centre of Exact Sciences and Engineering, University of Madeira, Campus Universitário da Penteada, 9000-390 Funchal, Madeira, Portugal

Abstract
Shape memory alloys are used in a variety of fields, such as medical or aeronautical. Other fields of knowledge have been
researching these materials, attracted by their capacity to dissipate energy through high-strain hysteretic cycles without significant
residual strains. Because of these interesting properties for seismic protection, an example of the possible beneficiaries of these
materials are civil engineering structures. This paper reports a bibliographic review on the characteristics and uniaxial macroscale
constitutive models for shape memory alloys, of interest for a significant number of applications, most often based on wires and
bars. The constitutive model assessment focuses on mechanical and kinetic laws, as well as on the energy balance law, of relevance
for dynamic loadings. Some characteristics of these materials are still not sufficiently well known, especially those related to
ageing. With regard to behaviour prediction, the most frequently used uniaxial constitutive models result in similar responses.
©©2015
2015TheThe Authors.
Authors. Published
Published by Elsevier
by Elsevier Ltd.
Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of INEGI - Institute of Science and Innovation in Mechanical and Industrial Engineering.
Peer-review under responsibility of INEGI - Institute of Science and Innovation in Mechanical and Industrial Engineering
Keywords: Shape memory alloys; Ageing; Residual strain; Uniaxial models; Self-heating

1. Introduction

Shape memory alloys (SMAs) may be found in two state phases: austenite and martensite. Austenite is called the
parent phase, from which a transformation process occurs into the softer product phase, martensite. This solid-solid
transformation is referred to as non-diffusive, a reference to the transformation behaviour of the crystalline structure,
with the atoms taking small-length ordered displacements, the opposite of diffusive. Transformation from the parent
phase may be induced by a temperature reduction or a stress increase, each resulting in a characteristic type of
martensite: twinned or multi-variant martensite in the first case and detwinned or single-variant in the latter. This
terminology results from the crystallographic geometry of the formed martensite. Macroscopically, SMAs behaviour
is characterised by the superelasticity and by the shape memory effect. The former is associated with the recovery from
loading-originated strains of up to 8% or 10%, without significant residual strains. The latter effect is characterized
by the capacity of recovering residual strains developed after cyclic loading with a temperature variation.
Stress-free, SMAs are characterized by four transition temperatures: M f , M s , A s and A f , from the lowest to the
highest. M referes to martensitic and A to austenitic state phases. f and s are references to the finish and start

∗ Corresponding author.
E-mail address: paulo.lobo@ist.utl.pt

1877-7058 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of INEGI - Institute of Science and Innovation in Mechanical and Industrial Engineering
doi:10.1016/j.proeng.2015.08.025
 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783 777

temperatures of the transformation process. When the temperature, T , is lower than M f , martensite is stable. For
T values higher than A f , the stable phase is austenite. For temperature values between A s and M s , both phases are
stable. From the start to the finish temperatures, martensitic, forward, and austenitic, reverse, transformations occur.
There are SMAs that do not follow this order [1], however, those materials, which do not belong to the ”Type I”
group, are not considered here. If the temperature is lower than M f , single-variant martensite may be obtained from
multi-variant martensite by the application of stress, which occurs due to a reorientation of the martensite crystals.
After unloading, significant residual strains remain. In order to recover multi-variant martensite, it is necessary to
transform the material into austenite, by raising the temperature of the SMA higher than A f , and then back to single-
martensite as the temperature is lowered. The superelastic effect is characterized by the single-variant martensite
transformation from the parent phase due to loading at a temperature higher than A f . This effect begins with elastic
austenite, after which the transformation into single-variant martensite is initiated at a critical stress value. The elastic
loading of martensite begins after the transformation is completed. Increasing loading leads to the SMA plastification.
Unloading the elastic martensite initiates the reverse transformation, because at that range of T values only austenite is
stable in the stress-free state. The unloading path presents lower stresses at the transformation phase than the loading
one, what allows for energy dissipation associated with cyclic loadings. After complete unloading, the SMA presents
virtually null residual strains. For temperatures higher than Md , SMA materials present an elastic-plastic behaviour.
Although SMAs are highly sensitive to variations in material composition and thermomechanical treatments, they
can generally be divided in nickel-titanium (NiTi) and Cu-based alloys. Near-equiatomic NiTi with cold working
and annealing treatments are considered most suitable for a number of engineering applications, given its better
superelastic properties, higher temperature variations stability and higher resistance to corrosion and fatigue [2–4].
The behaviour of SMAs can be predicted with constitutive models that are available, which should be adapted in
order to take into account all the characteristics of the specific material to be applied. Even though three-dimensional
macroscale constitutive models for SMAs exist, one-dimensional models are adequate for the simulations involving
wires and bars, which are the majority of cases. Also, the majority of the constitutive laws are phenomenological
in nature. Most uniaxial constitutive models of SMAs have a mechanical law, relating stress, strain and martensite
fraction, and a kinetic law governing the martensitic transformation as a function of temperature and stress. These
models are known as temperature dependent models. If an energy balance law is added, it is possible to obtain a strain
rate dependent constitutive law. SMA uniaxial constitutive relations are discussed in Chapter 3.

2. Characterization of shape memory alloys

Miyazaki et al. [2] pointed out the change in the stress-strain curve with loading cycles as the most important
factor inhibiting the use of NiTi SMAs. Based on 0.5 mm diameter wire of Ti-50.5 at.%Ni testing, they reported that
residual strains result from dislocations in the martensitic phase, which develop a stress field at the grain boundaries,
reducing the externally applied-stress required to commence the transformation [2]. The same authors found that, in
order to stabilize the superelasticity, it is necessary to raise the critical stress for slip, attainable by ageing treatments
and by annealing. The combination of these treatments accompanied by training was found to be highly efficient for
this purpose. The benefits of cyclic training was advocated by other authors [5–7]. Auguet et al. [8], based on test of
2.46 mm diameter wires, reaching a deformation of more than 8% concluded that SMAs residual strain is due to both
a recoverable and a permanent contribution. McCormick et al. [9] tested coupons extracted from the center and the
edge of 25.4, 12.7 and 6.35 mm diameter bars of hot rolled Ti-50.90 at.%Ni, and also tested a 12.7 and a 6.35 mm
bar. These authors concluded that the decrease in bar diameter seems to be related to a decrease in the residual strain.
A different trend had been reported previously [10]. NiTi wires of 0.254 mm and bars of 12.7 mm diameter were
tested. The residual strains of the wires were close to 1% after 6% strain cycles, while the bars registered values close
to 0.2%. The chemical composition of the materials were not reported and the specimens were subjected to different
treatments, which impairs the evaluation of the results. Nevertheless, the test of NiTi bars with very low residual
strain is noted. Residual strain development is accompanied by a reduction in the energy dissipated in each hysteresis
loop [5,9]. This is a core issue with regard to damping applications. Researchers have come to conclusions which,
in some cases, are contradictory. This may result from the limited number of specimens analysed and to different
chemical compositions and treatments of the tested materials. Even though efficient thermomechanical treatments for
stabilizing the superelasticity of Ni-rich alloys have been reported [2], this is an issue that requires further research.
778 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783

Differences in tension/compression behaviour have been found in tests on martensitic and austenitic NiTi SMAs,
due to different microstructural deformation mechanisms [11,12]. DesRoches [12] tested austenitic Ti-55.95 wt.%Ni
bars and reported good superelastic behaviour, with a flag shape hysteresis and good re-centring capacity, under both
compression and tension. These tests showed different transformation plateau stress and initial elastic modulus, of
about 25% and 30% lower in tension, respectively. Jalaeefar and Asgarian [13] also studied the tension/compression
behaviour of superelastic Ti-53.5 wt.%Ni bars, of 8 mm diameter, subjected to cycles of up to 6% strain. Almost sim-
ilar stress-strain curves in tension and compression were observed, with higher stresses for the compression loadings,
following the trend reported by the previously cited authors. An increased initial elastic modulus was not reported.
Despite the above, the hypothesis that the same behaviour in tension and compression is followed is usually assumed
[10,14,15]. Nevertheless, if, for a given SMA, this matter is found to be relevant, different material parameters may
be used for tension and compression [13]. Moreover, the review of the bibliography indicates that this is an issue that
depends on the material composition and one that should be further investigated.
Dolce et al. [3] reported significant variations in stress-strain profile, energy dissipation and stiffness of SMAs due
to temperature changes. Reductions of the equivalent damping ratio close to 40% for temperature variations from
0 ◦ C to 40 ◦ C were reported. Noting that temperature variations related to annual and daily cycles are inevitable and
that high-rate loadings, because of self-heating, discussed below, induce significant temperature changes, this rela-
tionship, defined by the Clausius-Clapeyron equation, becomes important. It follows experimental evidence, which
demonstrates that the transformation stresses increase linearly with increasing temperature, and it is characterized by
the Clausius-Clapeyron coefficient (CCC), usually assumed to be equal for both martensitic and austenitic transfor-
mation (C A = C M = CCC) and constant throughout the entire temperature range [1]. Previous reports claimed that
NiTi and other frequently used SMAs were essentially strain rate independent [16], but later research indicates other-
wise [17,18]. DesRoches and Smith [19] noted that research on this subject has led to contradictory conclusions on
the effects of strain rate. There are reports which demonstrate that the dissipation capacity increases with the increase
of the strain rate and others that show the exact opposite. Different responses were found for bars and wires [10].
This may be due to the use of different strain ranges, sample sizes, testing conditions and properties of the materi-
als [19]. Self-heating, due to the latent heat originating at the solid-solid transformation phases [20], and, with less
significance, to hysteretic dissipation, has been identified as the main source of SMA strain rate dependency. Because
the superelastic behaviour is temperature dependent, when the temperature of the material changes so does the critical
transformation stresses, σAs , σAf , σ M
s and σ f (the subscript and superscripts definitions are presented above). This
M

modifies the damping capacity of SMAs, with boundary cases for adiabatic, typical of high-strain rate loadings, and
isothermal conditions, associated with low-strain rate loadings response.
Although the atomic displacements associated with the solid-solid transformations of SMAs are essentially dis-
placive, Auguet et al. [21] reported measurements of diffusive effects in NiTi. They focused on temperature effects
on ageing, involving tests at constant strain and at room temperature for periods of nearly one month. At the parent
phase, the stress needed to proceed with cycling was increased roughly proportionally to the previously applied stress.
On this matter, Torra et al. [7] also concluded, based on wire tests, that for low prestress of about 200 MPa in the
parent phase, even for temperatures resulting from direct sun exposure, the variation of stress cannot be considered
as relevant. At the coexistence phase, the stress required to proceed with the transformation was also increased (prior
relaxation was observed), with this stress step reaching 20 MPa [8], due to an increase in the austenite fraction [21].
This test was also conducted at room temperature and with only one month duration, which led Torra et al. [7] to
state later that for significant periods of times this variation may become highly relevant. For the martensite phase,
the strain was maintained for different time intervals, after which an inner loop was imposed [8]. The most relevant
observation is that, after the period in which the strain was fixed, the martensite transformation stress of the inner loop
was lower than the initial transformation stress and this difference increased with longer fixed strain time intervals.
Dolce et al. [3] conducted 20 days time interval tests and reported negligible relaxation effects. Auguet et al. [4]
studied NiTi wires of 0.5 and 2.46 mm diameter, involving ageing at 100 ◦ C for time intervals of up to 270 days. This
research showed that NiTi cannot be exposed to direct sunlight for long periods of time without occurring progressive
and spontaneous increase in the transformation stresses, of over 90 MPa. Furthermore, Torra et al. [7] evaluated the
effect of eight months ageing at 100 ◦ C of 2.46 mm NiTi wires subjected to a high prestrain of 6.8%. The results show
that the flag shape of the hysteresis cycle changed to a narrow “S” shape curve and that the maximum stress for the
cyclic tests with maximum strain of about 8% changed from about 600 MPa to close to 1000 MPa. These relevant
 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783 779

changes in SMAs under temperature and stress require further investigation [7]. With the current state of knowledge,
the use of permanently prestressed NiTi may be dangerous, given the previously mentioned ageing effects, which may
result in unpredicted changes in the stress-strain response [21].
Structural fatigue, also known as classical mechanical fatigue, is related to the formation of microcracks, inducing
material rupture [22]. A series of cyclic loading tests, with 2.46 mm diameter wires, was carried out [23]. The results
indicate that high-cycle fatigue is significant, associated with the vibration due to rain and/or wind actions. Given the
small number of cycles expected in earthquakes damping - less than 200 -, the decay of the energy dissipation due
to this issue does not go beyond 2% [23]. Dolce et al. [3], based on experimental tests, also reported considerable
low-cycle fatigue resistance of NiTi alloys, of hundreds of 6% to 8% strain cycles. Therefore, unlike for other cases,
for seismic applications substitution of SMAs is not required [24].

3. Uniaxial constitutive models for shape memory alloys

A significant number of the current macroscale models are based on the work by Tanaka et al. [25]. This model
combines a mechanical and a kinetic law which governs the martensitic fraction of the material. It is a temperature
dependent phenomenological model with the crystallographic structural changes taken into account through an in-
ternal state variable, the martensite fraction, ξ, characterizing the extent of the transformation. Although a general
theory was developed, Tanaka et al. [25] applied it to the uniaxial behaviour description of alloys in the process of
stress-induced transformation, thus modelling the superelasticity effect, resulting in

σ̇ = E ε̇ + ΘṪ + Ωξ̇, with E = E A + (E M − E A )ξ and Θ = ΘA + (Θ M − ΘA )ξ (1)

where σ is the normal stress, ε represents the strain, Ω is the phase transformation tensor, E is the elasticity modulus
and Θ is the modulus of thermoelasticity. The subscripts A and B are a reference to the pure austenitic and martensitic
phases. An overdot denotes a time derivative. When integrated, taking E and Θ as constants, the previous equation
becomes Δσ = EΔε+ΘΔT +ΩΔξ. Using the concept of maximum residual strain, εL , which reflects the residual strain
of a SMA with 100% austenite and with temperature lower than A s that is completely transformed into martensite
through loading and after it is unloaded, the relation Ω = −εL E was obtained. This model is strain rate independent.
Liang and Rogers [26] developed a model based on the previous work. For uniaxial behaviour, taking the material
functions as constants, this model is the same as the former. Similar to the Tanaka et al. model, the Liang and Rogers
model cannot predict the shape memory effect if any temperature-induced martensite exists. On this matter, Brinson
[1] presented a constitutive model considering the martensite fraction composed of one part related to the martensite
induced by temperature and another part related to the stress-induced martensite, given by ξ = ξS + ξT . S and T are
references to the stress-induced and temperature-induced martensite, respectively. Moreover, experimental evidence
shows that taking the material functions, E, Ω and Θ, as constants is not accurate. In particular, the Young’s modulus
shows a significant dependence on the martensite fraction [1]. Thus, this author adopted the relation expressed in
the second equation in (1), which may be obtained with the Voigt scheme. If the Reuss scheme had been used, the
mechanical law obtained would have been identical [25]. Also, the relation Ω(ξ) = −εL E(ξ) was derived. Given its
insignificance for the overall response, presenting a value five orders of magnitude lower than E(ξ) [1], Θ(ε, ξ, T ) was
assumed constant. The constitutive relation with material functions varying linearly with ξ, which allows for good
agreement with experimental observations [1], delivers

σ − σo = E(ξ)ε − E(ξo )εo + Ω(ξ)ξS − Ω(ξo )ξS o + Θ(T − T o ) (2)

The subscript o is a reference to the initial conditions. This mechanical law is the backbone of most current SMA
constitutive models. It is a temperature dependent model, capable of tracing both the superelastic and the shape
memory effect at all temperatures, but it is strain rate independent.
Lubliner and Auricchio [27] developed a fully three-dimensional model which was later specialized to model
SMAs behaviour [16,27,28]. The one-dimensional mechanical law derived [16,27] is the same adopted by Brinson
[1] considering only single-variant martensite and isothermal conditions, σ = Eεe , in which the e superscript is a
reference to the elastic strain, given by εe = ε − εL ξS , which may be obtained by the substitution in Equation (2)
of null initial values, σo = 0, εo = 0 and ξS o = 0, assuming that the material always starts with 100% austenite,
780 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783

with T = T o , or simply neglecting the contribution of this term, knowing that it is significantly smaller than the
transformation strain [29]. This mechanical law is equal to the simplified constitutive equation presented by Brinson
and Huang [29]. Furthermore, the Young’s modulus was considered constant [27]. Auricchio and Sacco [30] proposed
a uniaxial model capable of simulating only the superelastic effect, but they also addressed the E variation with the
transformation process. The mechanical law adopted is the same as proposed previously [16,27]. For the evolution
of the elastic modulus, three homogenization methods were addressed: the Mori-Tanaka scheme, the Reuss scheme
and the Voigt scheme. A similar model was presented by Fugazza [15], with the same mechanical and linear kinetic
law, differing only in the adoption of a constant Young’s modulus. Later, Auricchio and Lubliner [28] proposed
an extension to their previous work, consisting of a more general uniaxial constitutive model for SMAs, making it
possible to reproduce the shape memory effect and the superelastic effect, with the inclusion of two internal variables,
as did Brinson [1]. The same phase transformation zones presented by Brinson [1] were adopted and the flow rules
were defined including parameters that measure the rates at which the transformations occur.
Piecewise linear constitutive relations have been proposed for superelastic analysis [31]. These proposals aim to
reduce the computational effort, but their application is limited, regarding the consideration of constant temperature.
Other more complex models, such as the micromechanical models proposed by Patoor et al. [32] and by Mirzaeifar
et al. [33], for studying the superelastic response of polycrystalline SMAs, in which the transformation strain is related
to the crystallographic data, have been developed. Their review is beyond the scope of this paper.
There are numerous possibilities for the kinetic law. The most relevant seem to be the exponential law adopted by
Tanaka et al. [25], the cosine laws developed by Liang and Rogers [26] and adapted by Brinson [1], and the linear
flow rule adopted by Auricchio and Sacco [30]. Reports from these authors show that any of these flow rules are able
to satisfactorily model the behaviour of SMA materials.
The cosine laws deliver, in the case of superelastic SMAs,
1 − ξo   σ  1 + ξo ξo    σ  
ξ= cos a M T − M f − + and ξ = cos aA T − A s − +1 (3)
2 CM 2 2 CA
for forward and reverse transformation, respectively. ξo is the fraction of martensite existing before the current trans-
formation, a M = π/(M s − M f ) and aA = π/(A f − A s ). Note that, as in the Brinson [1] kinetic law, the ξo memory
term has to be maintained for the other rules, because the transformation is dependent on the previous martensite
fraction. Taking this parameter into consideration, small cycles are intrinsically modelled. Thus, the exponential flow
law adopted by Tanaka et al. [25], expressed by Koistinen and Marburger [34] and Wang and Inoue [35], delivers
ξ = (ξo − 1)e[aM (Ms −T )+bM σ] + 1 and ξ = ξo e[aA (As −T )+bA σ] (4)
for forward and reverse transformations, with a M = −2ln(10)/(M s − M f ), b M = a M (M s − M f )/(Δσ M ), aA =
2ln(10)/(A f − A s ) and bA = aA (A f − A s )/(ΔσA ), in which Δσ M and ΔσA are the width along the stress axis of
the martensitic and austenitic transformation strips. Given the nature of the exponential function, for the determi-
nation of the previous constants the transformation was considered completed when ξ is equal to 0.99, which is a
common criterion in metallurgy [25]. This results in a discontinuity in the constitutive law, which is not compatible
with nonlinear equations solution methods that may be employed. It is resolved by dividing the right-hand member
of the kinetic rule by the above mentioned value. This does not change the nature of the kinetic function and has no
relevant influence on the solution. The linear kinetic law is defined by

|σ| − σ M
s
|σ| − σAf
ξ = (1 − ξo ) + ξo and ξ = ξo (5)
σMf − σ M
s σAs − σAf

Given the dependency of the superelasticity effect on the temperature, a more complex model is required for general
behaviour prediction. If an energy balance law is added, it is possible to obtain a strain rate dependent constitutive
model, capable of tracing the behaviour of SMAs for both quasi-static and dynamic loading [17,20]. The heat equation
can be written as ρc p Ṫ + div (q) + γ(T − T o ) = b, in which ρ is the density of the material, c p is the specific heat, div is
the divergence operator, q is the heat flux, γ is a heat exchange parameter linking the temperature difference between
the surface of the element and the surroundings and b is the heat source [20,36].
Given the high thermal conductivity of SMAs and the uniform distribution of heat sources, for elements with small
sections, such as wires and bars, a uniform temperature inside the body may be assumed and conduction resistance
 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783 781

may be neglected [15,20,36]. The validity of this assumption is verified by the calculation of the Biot number, defined
as the ratio between the conductive and the convective resistance, in which case it should be lower than 0.10 [17].
b is due to two contributions: the heat production associated with internal friction, i.e., hysteretic dissipation, and
the contribution from the transformation latent heat. It may be defined as b = cL ρξ̇ + Ẇ, in which cL is the latent heat
and W is the energy dissipated by hysteresis [17,18,20], whose influence may be neglected when compared to latent
heat contribution [20,37]. The energy balance law can then be written as −ρc p V Ṫ + hAi (T − T o ) = bV, in which V is
the specimen volume, h is the convection heat transfer coefficient and Ai is the interface area of the SMA [17,20,37].
The results obtained with different macro-scale models depend greatly on the kinetic rule adopted, because the
mechanical laws are essentially similar, with any differences being associated with the Young’s modulus of the trans-
forming material. Nearly identical results were obtained with different models based on the schemes of Voigt and
Reuss when the same kinetic law was used. Because the transformation strain is high when compared to the elastic
strain, the differences between mechanical laws are mainly lost [29].
As seen, the majority of the uniaxial mechanical laws can be derived from the Tanaka et al. rate model [29]. For
superelastic behaviour, this mechanical law, in the time continuous form, is written as

σ = E(ξ)(ε − εL ξ) + Θ(T − T o ) with E(ξ) = E A + (E M − E A )ξ (6)

This equation is written with σ in an implicit form, thus an iterative procedure must be applied, in which case the
complete Newton-Raphson method may be chosen. The mechanical law may be written in the time discrete evolutive
form as
∂ε(σnj ) j+1
εnj+1 = εnj + (σn − σnj ) (7)
∂σ
in which n refers to the instant and j refers to the iteration step. In the transformation zone, the derivative present in
Equation (7) has to be computed according to the kinetic law adopted. The kinetic laws can then be treated separately.
Figure 1 shows a generic isothermal stress-strain curve for each of the kinetic laws considered.

Fig. 1: General SMA stress-strain curves for different kinetic laws

The heat balance law may now be coupled with the model. It can be integrated with regard to time, by means of
the backward Euler method [17,38], and introducing the iterations regarding the fixed-point method results in
b hA[T nj − T o ] 
n
T nj+1 = T n−1 + Δt − (8)
ρc p ρc p V
where Δt is the time interval. Thus, before performing the material state determination, the temperature of the material
must be computed. As a result, according to the Clausius-Clapeyron equations, different critical stresses are consid-
ered for each time instant. If the strain steps are sufficiently small, ξ̇n may be taken approximately equal to ξ̇n−1 , with
insignificant error in the case of the cosine and the exponential laws and without error in the case of the linear kinetic
law. It is not possible to generally define an analytical equation relating strain with time, thus a numerical procedure,
such as the finite difference approximations method, may be adopted to obtain the martensitic fraction derivative with
respect to time.
Figure 2 shows a general response of a NiTi SMA subjected to two complete cycles of 6.5% strain, considering
coupling between the mechanical, the kinetic and the heat balance equations.
782 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783

(a) Stress-strain curve (b) Tempearture variation

Fig. 2: General SMA behaviour considering self-heating; (a) and (b) in the image caption

4. Conclusions

A bibliographic review of the behaviour of SMAs has been presented. The review focused on the characterization of
SMAs and on uniaxial models for prediction of the response of these materials. It was concluded that residual strains
may be of relevance for real design projects. The extent of this effect has shown to be a function of the composition of
the alloy, the specimen size and the thermomechanical treatments, that resulted in contradictory reports. Nevertheless,
research indicates that ageing treatments and annealing together with cyclic training are highly efficient in stabilizing
the superelasticity of NiTi, commonly considered the most suitable SMA for a number of engineering solutions. Even
though different tension/compression behaviour of SMAs has been observed, researchers most often consider the
same SMA behaviour in tension/compression. The temperature dependence of SMAs was also addressed. Together
with self-heating, it should be accounted for in the response prediction. Prestress and temperature ageing may also
be important for actual design. Research has shown that significant changes in the material behaviour may occur for
SMAs subjected to permanent prestress. Structural fatigue is not of concern for seismic applications, unlike for rain
and/or wind vibration control. Generally, it was found that there is a deficit in research in aspects of the behaviour of
SMAs fundamental to many applications. The constitutive models used for predicting the behaviour of SMAs were
also reviewed. The focus on uniaxial models was due to the fact that the majority of applications are based on wires
and bars, in which case these models are adequate. They generally combine a mechanical and a kinetic law, which
governs the martensitic fraction of the material. The majority of the mechanical laws are similar, with differences
arising from the Young’s modulus considered, which nevertheless are limited. Differences result mainly from the
adopted kinetic rule. Nevertheless, implementation indicates that the cosine and the linear laws deliver very similar
stress-strain curves. Even though the exponential kinetic law is the one which differs the most from the previous two,
the differences are moderate. A strain rate dependent model is obtained when the mechanical and kinetic laws are
coupled with an energy balance law. The general response of a SMA dynamic test is also presented.

References

[1] L. C. Brinson, One-dimensional constitutive behavior of shape memory alloys: Thermomechanical derivation with non-constant material
functions and redefined martensite internal variable, Journal of Intelligent Material Systems and Structures 4 (1993) 229–242.
[2] S. Miyazaki, T. Imai, Y. Igo, K. Otsuka, Effect of cyclic deformation on the pseudoelasticity characteristics of Ti-Ni alloys, Metallurgical
Transactions A 17 (1986).
[3] M. Dolce, D. Cardone, R. Marnetto, Implementation and testing of passive control devices based on shape memory alloys, Earthquake
Engineering and Structural Dynamics 29 (2000) 945–968.
[4] C. Auguet, A. Isalgue, F. Lovey, J. Pelegrina, S. Ruiz, V. Torra, Metastable effects on martensitic transformation in SMA Part III. Tentative
temperature effects in a NiTi alloy, Journal of Thermal Analysis and Calorimetry 89 (2007) 537–542.
[5] Z. Moumni, A. V. Herpen, P. Riberty, Fatigue analysis of shape memory alloys: Energy approach, Smart Materials and Structures 14 (2005)
287–292.
[6] A. Isalgue, V. Torra, A. Yawny, F. Lovey, Metastable effects on martensitic transformation in SMA Part VI. The Clausius-Clapeyron relation-
ship, Journal of Thermal Analysis and Calorimetry 91 (2008).
 Paulo Silva Lobo et al. / Procedia Engineering 114 (2015) 776 – 783 783

[7] V. Torra, A. Isalgue, C. Auguet, G. Carreras, F. C. Lovey, P. Terriault, Damping in civil engineering using SMA Part 2 - particular properties
of NiTi for damping of stayed cables in bridges, Canadian Metallurgical Quarterly 52 (2013) 81–89.
[8] C. Auguet, A. Isalgue, V. Torra, F. Lovey, J. Pelegrina, Metastable effects on martensitic transformation in SMA Part VII. Aging problems in
NiTi, Journal of Thermal Analysis and Calorimetry 92 (2008) 63–71.
[9] J. McCormick, J. Tyber, R. DesRoches, K. Gall, H. Maier, Structural engineering with NiTi. II: mechanical behavior and scaling, Journal of
Engineering Mechanics, ASCE 133 (2007) 1019–1029.
[10] J. McCormick, R. DesRoches, D. Fugazza, F. Auricchio, Seismic vibration control using superelastic shape memory alloys, J. Eng. Mater.
Technol. 128 (2006) 294–301.
[11] Y. Liu, Z. Xie, J. V. Humbeeck, L. Delaey, Asymmetry of stress-strain curves under tension and compression for niti shape memory alloys,
Acta Materialia 46 (1998) 4325–4338.
[12] R. DesRoches, Development of Innovative SMA Connections and Dampers for Frame Structures Under Extreme Loads, SESM Report 68-1,
Georgia Institute of Technology, School of Civil & Environmental Engineering, Atlanta, 2005.
[13] A. Jalaeefar, B. Asgarian, A simple hybrid damping device with energy-dissipating and re-centering characteristics for special structures, The
Structural Design of Tall and Special Buildings 23 (2014) 483–499.
[14] B. Andrawes, R. DesRoches, Comparison between shape memory alloy seismic restrainers and other bridge retrofit devices, Journal of Bridge
Engineering, ASCE 12 (2007) 700–709.
[15] D. Fugazza, Shape-Memory Alloy Devices in Earthquake Engineering: Mechanical Properties, Constitutive Modelling and Numerical Simula-
tions, Masters dissertation, Rose School, European School of Advanced Studies in Reduction of Seismic Risk, 2003.
[16] F. Auricchio, R. L. Taylor, J. Lubliner, Shape-memory alloys: Macromodelling and numerical simulations of the superelastic behavior,
Computer Methods in Applied Mechanics and Engineering 146 (1997) 281–312.
[17] A. Vitiello, G. Giorleo, R. E. Morace, Analysis of thermomechanical behaviour of nitinol wires with high strain rates, Smart Materials and
Structures 14 (2005) 215–221.
[18] V. Torra, C. Auguet, A. Isalgue, F. Lovey, A. Sepulveda, H. Soul, Metastable effects on martensitic transformation in SMA Part VIII. Temper-
ature effects on cycling, Journal of Thermal Analysis and Calorimetry 102 (2010) 671–680.
[19] R. DesRoches, B. Smith, Shape memory alloys in seismic resistant design and retrofit: A critical review of their potential and limitations,
Journal of Earthquake Engineering 08 (2004) 415–429.
[20] D. Entemeyer, E. Patoor, A. Eberhardt, M. Berveiller, Strain rate sensitivity in superelasticity, International Journal of Plasticity 16 (2000)
1269–1288.
[21] C. Auguet, A. Isalgue, F. Lovey, F. Martorell, V. Torra, Metastable effects on martensitic transformation in SMA Part 4. Thermomechanical
properties of CuAlBe and NiTi observations for dampers in family houses, Journal of Thermal Analysis and Calorimetry 88 (2007) 537–548.
[22] L. Janke, C. Czaderski, M. Motavalli, J. Ruth, Applications of shape memory alloys in civil engineering structures - Overview, limits and new
ideas, Materials and Structures 38 (2005) 578–592.
[23] G. Carreras, A. Isalgue, V. Torra, F. Lovey, H. Soul, Metastable effects on martensitic transformation in SMA Part V. Fatigue-life and detailed
hysteresis behavior in NiTi and Cu-based alloys, Journal of Thermal Analysis and Calorimetry 91 (2008) 575–579.
[24] M. Dolce, D. Cardone, F. C. Ponzo, C. Valente, Shaking table tests on reinforced concrete frames without and with passive control systems,
Earthquake Engineering and Structural Dynamics 34 (2005) 1687–1717.
[25] K. Tanaka, S. Kobayashi, Y. Sato, Thermomechanics of transformation pseudoelasticity and shape memory effect in alloys, International
Journal of Plasticity 2 (1986) 59–72.
[26] C. Liang, C. Rogers, One-dimensional thermomechanical constitutive relations for shape memory materials, Journal of Intelligent Material
Systems and Structures 1 (1990) 207–234.
[27] J. Lubliner, F. Auricchio, Generalized plasticity and shape memory alloys, International Journal of Solids and Structures 33 (1996) 991–1003.
[28] F. Auricchio, J. Lubliner, A uniaxial model for shape-memory alloys, International Journal of Solids and Structures 34 (1997) 3601–3618.
[29] L. C. Brinson, M. S. Huang, Simplifications and comparisons of shape memory alloy constitutive models, Journal of Intelligent Material
Systems and Structures 7 (1996) 108–114.
[30] F. Auricchio, E. Sacco, A one-dimensional model for superelastic shape-memory alloys with different elastic properties between austenite and
martensite, International Journal of Non-Linear Mechanics 32 (1997) 1101–1114.
[31] X. Du, G. Sun, S. Sun, Piecewise linear constitutive relation for pseudo-elasticity of shape memory alloy (SMA), Materials Science and
Engineering: A 393 (2005) 332–337.
[32] E. Patoor, D. C. Lagoudas, P. B. Entchev, L. C. Brinson, X. Gao, Shape memory alloys, Part I: General properties and modeling of single
crystals, Mechanics of Materials 38 (2006) 391–429.
[33] R. Mirzaeifar, R. DesRoches, A. Yavari, K. Gall, A micromechanical analysis of the coupled thermomechanical superelastic response of
textured and untextured polycrystalline NiTi shape memory alloys, Acta Materialia 61 (2013) 4542–4558.
[34] D. Koistinen, R. Marburger, A general equation prescribing the extent of the austenite-martensite transformation in pure iron-carbon alloys and
plain carbon steels, Acta Metallurgica 7 (1959) 59–60.
[35] Wang, Inoue, Analyses of temperature, structure and stress during quenching of steel with consideration for stress dependence of transformation
kinetics, J Soc Mater Sci 32 (1983) 991–996.
[36] F. Auricchio, D. Fugazza, R. Desroches, Rate-dependent thermo-mechanical modelling of superelastic shape-memory alloys for seismic
applications, Journal of Intelligent Material Systems and Structures 19 (2008) 47–61.
[37] S. Seelecke, Modeling the dynamic behavior of shape memory alloys, International Journal of Non-Linear Mechanics 37 (2002) 1363–1374.
[38] C. Cismaşiu, F. P. A. dos Santos, Numerical simulation of superelastic shape memory alloys subjected to dynamic loads, Smart Materials and
Structures 17 (2008) 25–36.