International Journal of Solids and Structures xxx (2015) xxxxxx

Contents lists available at ScienceDirect

International Journal of Solids and Structures

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

Analytical investigation of structurally stable congurations in shape

memory alloy-actuated plates
Edwin A. Peraza Hernandez a, Bjoern Kiefer b, Darren J. Hartl a,, Andreas Menzel b,c, Dimitris C. Lagoudas a
a
Department of Aerospace Engineering, Texas A&M University, College Station, TX 77843, United States
b
Institute of Mechanics, Department of Mechanical Engineering, TU Dortmund, 44227 Dortmund, Germany
c
Division of Solid Mechanics, Lund University, P.O. Box 118, 221 00 Lund, Sweden

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

Article history: Strains produced by active materials embedded in plates have been extensively used to manipulate the
Received 17 February 2015 shape of surface-like engineering structures. Shape memory alloys (SMAs) are active materials that pro-
Received in revised form 23 April 2015 vide a signicant amount of strain under large stresses, a characteristic of great utility in such morphing
Available online xxxx
structures. In this work, an analytical approach to approximate the deformation of plates with SMA con-
stituents is developed via the RayleighRitz method. An additive set of kinematically admissible displace-
Keywords: ment elds with unknown coefcients is used to describe the plate displacement eld. The total potential
Shape memory alloys
energy is then calculated using the displacement eld, loading conditions, and constitutive relations for
Morphing structures
Gaussian curvature
the plate layer(s) composed of SMA wire meshes, dense SMA lms, and/or elastic material. The unknown
RayleighRitz method coefcients are then found via minimization of the total potential energy. This approach provides
Plate theory closed-form expressions for the approximate deformation of the plates including multistable congura-
tions. The response of circular SMA-based plates is studied herein. The results show that temperature
elds with a linear variation in the radial direction induce multistable congurations in which the plate
Gaussian curvature is determined by the direction of the temperature gradient. An alternative inversion
of the proposed approach is used to directly compute the temperature eld required to morph a plate
towards a prescribed goal shape. The obtained closed-form expressions show good agreement with
detailed non-linear nite element analysis simulations. The proposed approach provides analysts with
a low computational cost and relatively simple implementation to assess the potentially stable congu-
rations of SMA-based plates under given loading conditions. Knowledge of such stable congurations is
very valuable in the design of SMA-based morphing structures.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Manipulating the shape of surface-like structures (e.g., plates)
and conceiving morphing structures is an active research subject
in engineering (Seffen and Maurini, 2013; Fernandes et al., 2010).
Active materials, those that convert various forms of energy into
mechanical work (Srinivasan and McFarland, 2001; Leo, 2008),
provide the means to induce controllable inelastic (usually recov-
erable) strains that can exert such manipulations. The design of
morphing surface-like structures with intrinsic material actuation
is a challenging problem. It entails two main difculties: (i) taking
into account non-linear material and geometrical effects when
designing the structure; and (ii) conceiving appropriate and espe-
cially efcient actuation techniques to morph the structure among
different goal shapes.
Corresponding author.
E-mail address: darren.hartl@tamu.edu (D.J. Hartl).
For example, the art and theory of origami have provided
numerous solutions to the aforementioned challenges during the
past decades (Lang, 2007). Origami provides novel methods to
assemble and morph structures (Bassik et al., 2009; Fernandes
and Gracias, 2012; Tolley et al., 2014; Kuribayashi et al., 2006;
Pandey et al., 2013; Cheng et al., 2013; Peraza-Hernandez et al.,
2013b,f, etc.). Furthermore, self-folding structures, those that have
the capability of folding and/or unfolding among different shapes
without requiring externally applied mechanical loads, have been
developed by leveraging the use of active materials as agents of
fold generation (Peraza-Hernandez et al., 2014c).
The mechanism for morphing these origami-based structures
has been primarily folding (i.e., bending). Bending deformation is
isometric and conformal with respect to the neutral surface of a
plate structure (i.e., there are no changes in lengths and angles
along such a surface resulting from this deformation mode).
Deformation via bending provides for large global deections and
rotations; however, such a deformation mode is based primarily

http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
0020-7683/ 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
2 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx

on disallowing changes in the Gaussian curvature of the structure.

Gaussian curvature is the product of the two principal curvatures
at any region of a surface (g jI jII ) (Calladine, 1989) where the
principal curvatures jI and jII correspond to the maximum and
minimum values of the local curvatures (Pressley, 2010). Such
principal curvatures are oriented orthogonally to each other and
aligned in the principal directions v jI and v jII , respectively.
Considering a surface subdomain sufciently small so as to be
effectively approximated by a surface patch of uniform Gaussian
curvature, four cases can be distinguished: (i) if both jI and jII
are zero, the region is planar and has zero Gaussian curvature
(Fig. 1(a)); (ii) if either jI or jII is zero and the other is non-zero,
the region is parabolic and has zero Gaussian curvature (as in a
bend; Fig. 1(b)); (iii) if jI and jII are both < 0 or both > 0, the
region is elliptic and has positive Gaussian curvature (Fig. 1(c));
and (iv) If jI and jII are both non-zero and of opposite sign, the
region is hyperbolic and has negative Gaussian curvature Fig. 1. Principal curvatures and Gaussian curvature of different surfaces.
(Fig. 1(d)), Pressley, 2010. Deformation via bending (not allowing
changes in Gaussian curvature) largely limits the spectrum of
shapes that can be obtained by a structure. Furthermore, regions
of non-zero Gaussian curvature from initially planar structures
are often needed for aerodynamic, aesthetic, or load bearing MARTENSITE
reasons.
One approach for the achievement of changes in Gaussian cur-
vature in plate structures is through the generation of inhomoge-
neous in-plane extensional/shear strain elds throughout their CM CA
middle surface (Seffen and Maurini, 2013; Fernandes et al., 2010;
Yu et al., 2000; Liu et al., 2004; Ryu and Shin, 2006). Relatively
moderate to large rotations in plate structures entail a coupling AUSTENITE
between bending and in-plane extension/shear. For thin plate
structures, there is potentially a choice for engineers in the mode
of actuation, the so-called actuation paradigm mentioned by Mf Ms As Af
Modes et al., 2011: whether in-plane extensional/shear actuation
can provide a feasible alternative to bending actuation. The usage Fig. 2. Schematic of an SMA stresstemperature phase diagram.
of in-plane extensional/shear strains and bending strains (i.e.,
those varying linearly through the plate thickness), either sepa-
rately or in combination, can provide for an extensive set of possi- classical plasticity) that depends on stress and temperature (see Eq.
ble shapes that can be attained by a plate structure (Seffen and (15)). Such a transformation function captures the essential features
Maurini, 2013). of the phase diagram shown in Fig. 2. The form of the transformation
Shape memory alloy (SMA)-based plate structures are consid- function is selected depending on whether the SMA is transforming
ered herein. Shape memory alloys are active materials that from austenite to martensite (forward transformation) or from
undergo solid-to-solid phase transformations induced by appropri- martensite to austenite (reverse transformation). Transformation
ate temperature and/or stress changes during which they can gen- strains associated with the evolution of the martensite volume
erate or recover seemingly permanent strains (Lagoudas, 2008). fraction may be generated during forward transformation and are
These characteristics allow them to have several existing and recovered during reverse transformation.
potential applications in diverse elds such as aerospace (Hartl For the present application, SMA domains initially in stress-free
and Lagoudas, 2007), biomedical (El Feninat et al., 2002), and oriented martensite states are considered (i.e., having
others (Mohd Jani et al., 2014). Shape memory alloys exhibit the pre-generated transformation strains, denoted as pre-strains in
highest actuation energy density of all active materials. This indi- this work). During reverse transformation, which is the case con-
cates that SMAs can provide a signicant amount of strain under sidered herein, the direction and magnitude of the transformation
large stresses, a characteristic of great utility in morphing struc- strain evolution are dened such that all transformation strains
tures for use in realistic conditions. existing at the cessation of forward transformation (the
A schematic of the stresstemperature phase diagram for SMAs pre-strain in the cases considered here) are fully recovered when
is shown in Fig. 2. It can be described by the transformation tem- the SMA transforms fully back into austenite (see Eq. (13)). Thus,
peratures at zero stress: martensite start M s , martensite nish by heating the SMA and moving from the austenite start boundary
Mf , austenite start As , and austenite nish Af . The slopes of the towards the austenite nish boundary in the stresstemperature
transformation boundaries in the stresstemperature hyperspace space (Fig. 2), the pre-strains in the SMA are recovered and actua-
are called the stress inuence coefcients C M and C A (Lagoudas, tion is produced.
2008)1. For the present constitutive model described in detail in The SMA-based plate layups considered in this work are shown
Section 2.3, the evolution of the martensite volume fraction is deter- in Fig. 3. The plates may be comprised of a single SMA lm that is
mined via a transformation function (analogous to a yield function in pre-strained homogeneously in the in-plane directions (Fig. 3(a)).
The lm can provide half of the maximum recoverable uniaxial
strain by the SMA (denoted H) in any two orthogonal directions
1
Generally, the slopes of the transformation boundaries in the stresstemperature
in the plane co-planar to the plate (as shown in Fig. 3(a)). The
phase diagram are not constant and thus measured at a reference stress level plates may also be comprised of a three-layer laminate with two
(Lagoudas et al., 2012). outer layers of SMA wire meshes separated by a middle compliant

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx
Fig. 3. Plate layup for (a) single SMA lm, and (b) three layer laminate with two outer meshes of SMA wires separated by a middle elastomeric layer. (c) Examples of periodic
SMA wire meshes.
and thermally insulating layer such as an elastomer
(Peraza-Hernandez et al., 2013e, Fig. 3(b)). The role of the
elastomer interlayer is to serve as a thermal insulator between
the two SMA mesh layers when either side is being heated and
actuation of the opposite side is unwanted, as in folding. This
characteristic is not considered in this work, rather uniform
through-thickness temperature is applied. Instead, the elastomer
layer provides for continuity of the displacements throughout the
plate, analogous to a matrix in a ber composite if perfect bonding
is maintained. The outer layer meshes are composed of woven SMA
wires installed in periodic arrangements (see Fig. 3(c) for exam-
ples). This periodicity allows for the effective modeling of such
structural layers as continuum layers. Details on the modeling of
such layers are provided in Section 2.4. In a periodic SMA wire
mesh conguration, these active layers can provide strains with a
magnitude of H in each direction collinear to the alignment of
the wires (as indicated in the left mesh of Fig. 3(c)). After actuation
and possible deformation towards non-planar congurations, the
SMA-based plates will return to their initially planar stress-free
congurations upon cooling the SMA below M f , unless restricted
by external mechanical loads. This is due to the recovery of elastic
strains generated during deformation of the plate and the stable
two-way shape memory effect (Otsuka and Wayman, 1999) pro-
cessed or trained in most commercially available SMA wires.
The objective of this work is to provide a low computational
cost and easily implemented RayleighRitz analytical tool for the
design of SMA-based, possibly multistable morphing structures
of simple geometrical form. The inherent ability of this approach
for capturing non-linear multistable responses with relatively little
computational effort makes it an appealing choice for the initial
analysis of these plate-like structures (Eckstein et al., 2014;
Pirrera et al., 2012). Commonly available nite element analysis
(FEA) tools have no built-in ability to determine the possible stable
solution branches (i.e., deformation modes) for multistable struc-
tures considering material non-linearities. Instead, some knowl-
edge of the potential stable solutions should be known a priori
such that the proper perturbations may be applied to inuence
the solution towards specic stable branches (Pirrera et al., 2010;
Schlecht et al., 1995).
The RayleighRitz method has been utilized for the study of the
dynamic response (Carrera et al., 2011; Al-Obeid and Cooper, 1995;
Lam et al., 1989; Liew and Lam, 1991), stability (Gigliotti et al.,
2004), and other behaviors in passive plates. Similar approaches
have been considered before with the goal of approximating the
deformation in ber composite plates morphed via thermal expan-
sion (Eckstein et al., 2014; Dano and Hyer, 1998; Eckstein et al.,
2013). The contribution of the current work is that it approximates
the deformation of plates with active SMA layers whose transfor-
mation strains, unlike conventional thermal expansion, are cou-
pled with stress in a non-linear and even hysteretic manner
(Lagoudas, 2008), although the latter is not addressed in this work.
The RayleighRitz method has also been used to approximate the
deformation of composite plates morphed via actuation of SMA
Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
3
wires connected at different locations of the plates (Dano and
Hyer, 2003); however, in such cases the SMA wires were not inte-
grated into the plate itself, but were rather external agents that
provided forces and displacements at discrete points on the plate.
The vibration and thermal buckling behavior of plates with SMA
bers were studied numerically by multiple authors (Li et al., 2010;
Park et al., 2004; Lee and Lee, 2000; Kumar and Singh, 2009), but
their numerical efforts were focused on reinforcement of plates
using the SMA components rather than embedding SMA wires into
compliant medium for morphing purposes. They typically consid-
ered boundary conditions that largely prevented the shapes that
can be attained by the freely actuating SMA-based plates. Most
of those efforts were focused on unidirectional SMA ber layers;
here meshes of SMA wires that may be aligned in multiple arbi-
trary directions are considered, as will be shown in the subsequent
sections. Also, by using an analytical approach, closed-form expres-
sions for the deformation of the SMA-based plates are provided
herein.
The remainder of the article is organized as follows: Section 2
provides the geometrical assumptions, constitutive relations, and
methodology used to obtain closed-form expressions for the defor-
mation of SMA-based plates, Section 3 considers different bound-
ary value problems in which the plate deformation is obtained
using the present method, Section 4 presents a problem in which
the temperature eld required to morph the plate towards a goal
conguration is directly computed using an inversion of the pre-
sent method, and Section 5 provides a summarizing discussion
and concludes the article.
2. Kinematics, solution method, and constitutive relations
2.1. Plate kinematics
Consider a plate assumed to be initially in a planar reference
conguration, having a uniform thickness and its mid-plane coin-
cident with the xy plane. Based on von Krmn plate kinematics
(Berger, 1967; Fernandes et al., 2010), the deformation of the plate
can be completely described by the plate mid-plane displacements
(u ux ; uy ; uz T in Cartesian coordinates). The total innitesimal
strain everywhere in the plate is assumed to be additively decom-
posed as follows:
2 3 2 3 2 3
exx ex0 jx
6 7 6 7 6 7
4 eyy 5 4 ey0 5 z4 jy 5; e e0 zj; 1
exy exy0 jxy
where e0 are the extensional/shear strains of the mid-plane and j
are the out-of-plane curvatures of the plate. The straindisplace-
ment von Krmn equations are given by:
2 3 2 3
ex0 ux;x 12 uz;x 2
6 7 6 7
e0 4 ey0 5 6
4 uy;y 12 uz;y 2 7;
5 2
exy0 1
2
1
ux;y uy;x 2 uz;x uz;y
4 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx
2 3 2 3
jx uz;xx
j6 7 6 7
4 jy 5 4 uz;yy 5:
jxy uz;xy
Note that the in-plane distortion includes second-order geo-
metric contributions due to out-of-plane displacement. Thus, this
can be thought of as a small strains/moderate rotation plate theory
that considers non-linear behavior due to the plate kinematics. It
can be veried from the relations in Eqs. (2) and (3) that the change
in Gaussian curvature Dg of an initially planar plate provides a con-
nection between the mid-plane strains and the curvatures as fol-
lows (Calladine, 1989):
Dg ex0 ;yy  ey0 ;xx 2exy0 ;xy jx jy  j2xy jI jII :
where jI and jII are the principal curvatures (see Section 1). Since
the initial Gaussian curvature is restricted to zero in the current
work, Dg is not only the change in Gaussian curvature but also
the total Gaussian curvature g at any location in the plate. The
compatibility condition in Eq. (4) is a useful intrinsic statement of
the coupling among mid-plane strains and plate curvature.
Specically, it shows that any change in Gaussian curvature of the
plate implies a non-uniform distribution of in-plane distortion
(Manseld, 1989).
2.2. Method of solution
A RayleighRitz-type method is now employed to nd
approximate solutions for boundary value problems involving
plates with SMA constituents (e.g., wire meshes or dense lms).
Additional information on the underlying variational principles
of this approach is provided elsewhere (Slaughter, 2002;
Reddy, 2002). The method begins by considering an additive
set of kinematically admissible displacement elds (i.e., those
satisfying the kinematic boundary conditions of the problem)
given by:
X
N the present method, where the ansatz function space is a set of
u ci ui x;
i1
where N is the total number of individual displacement eld contri-
butions considered, ci are their unknown coefcients to be solved
for, and x are the coordinates of the plate points at the reference
conguration (x x; y; zT ). From such a displacement eld, the
mid-plane strains and curvatures can be found using Eqs. (2) and
(3). Since the displacement eld is considered as input, its associ-
ated mid-plane strain and curvature elds inherently satisfy the
compatibility condition of Eq. (4). Other approaches with reduced
number of degrees of freedom than those of Eq. (5) can be derived
through the usage of auxiliary solutions of plane elasticity (e.g.,
Seffen and Maurini, 2013; Seffen, 2007; Vidoli, 2013). However,
the inelastic transformation strains in the SMA are function
of both temperature and stress, bringing other coupled
stresstemperature-transformation strain equations besides the
elasticity relations (see Section 2.3). Thus, such elasticity solutions
may not be appropriately introduced or may not be accurate for
the problems considered here. Instead, ansatz functions for the
three displacement components are proposed at the expense of a
larger number of degrees of freedom.
The total potential energy of the system, P, is given as follows:
P U  W e;
where U is the elastic strain energy stored in the plate while W e is
the external work done by body forces and boundary tractions.
Assuming an additive decomposition of the total strains into an
elastic contribution eel and a recoverable actuation contribution
e A , the total energy stored in the plate takes the form:
Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
Z Z
1 el 1
U e : C eel dv e  e A : Ce  e A dv ; 7
3 V 2 V 2
where V is the total volume of the plate and C is the material stiff-
ness tensor. The integration over V is performed symbolically by
integrating over the z coordinate through the thickness of the plate
and over the x and y coordinates through the plane-form area of the
plate. The selection of appropriate ansatz displacement elds allows
for such symbolic integration (e.g., polynomial displacement elds
are employed herein). For laminated plates, the integration over z
is performed in a layerwise manner, taking into account the differ-
ent stiffness and actuation strains for each individual layer. The
work done by (deformation-independent) body forces b and plate
4 boundary tractions t is given by:
Z Z
We b  u dv t  u da; 8
V @V
where @V is the boundary of the plate. The actuation strains in this
work correspond to strains due to martensitic transformation in the
SMA regions of the plate. Their evolution depends on stress and
temperature through the constitutive relations required to deter-
mine these transformation strains. The formulations for the consti-
tutive response of monolithic SMA (Lagoudas et al., 2012) and the
effective continuum response of SMA wire meshes used in this work
are presented in Sections 2.3 and 2.4, respectively.
Once the expressions for the actuation strains and the total
strains are obtained, they can be substituted into Eq. (7) to obtain
the total energy stored in the plate. The ansatz displacement elds
can be substituted into Eq. (8) to obtain the work done by applied
loads. This leads to an equation for P in which the only degrees of
freedom are the unknown coefcients of the ansatz displacement
elds (ci ; i 1; . . . ; N).
By the principle of minimum potential energy, among all possible
kinematically admissible displacement elds, the total potential
energy P is rendered stationary by those that are the actual dis-
placement elds (Slaughter, 2002; Bauchau and Craig, 2009). In
5 admissible displacement elds, the exact solution of the displace-
ment eld may not be contained within such a set, and the
obtained solution is an approximate solution. In the special case of
assuming a sufciently rich ansatz function space, however, this
procedure may also yield the exact closed-form solution. To nd
the displacement eld that minimizes P, the following N partial
derivatives are considered:
@P
0; i 1; . . . ; N: 9
@ci
This process produces N non-linear equations that can be solved for
N unknowns ci . The non-linearity of the obtained equations arises
geometrically from the form of the mid-plane strains in Eq. (2)
and physically from the constitutive relations. Since the obtained
equations are non-linear, their solution may not be unique. In this
case, the stability of each solution is evaluated via second deriva-
tives of P. For a conguration to be stable, the generalized
@2 P
(Hessian) stiffness matrix @ci @cj
; i; j 1; . . . ; N, must be positive de-
nite (i.e., all the eigenvalues associated with such a conguration
must be positive).
As mentioned in Section 1, during reverse transformation the
direction of the evolution of transformation strains (assumed to
6 be the only source of actuation strain e A in this work) is prescribed
by the direction of the pre-strains in the SMA constituents. Thus,
their direction is not affected by the evolution or the state of stress.
However, the martensite volume fraction, which denes the mag-
nitude of the transformation strains, is coupled with stresses and is
sensitive to changes in the stress evolution direction. Such changes
may arise during the thermomechanical loading process or
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx 5

transitions between stable congurations. Thus, the evolution of Table 1

transformation strains in SMAs is a path-dependent process since Procedure to nd the approximate deformation of plates with SMA constituents.

these strains are determined from the thermomechanical loading P

1: Input u N i1 c i ui x
history of the material. However, in presented approach, the path 2: Input Tx, surface tractions and body forces
dependence of the problem is not accounted for since the solution 3: Calculate e A x via constitutive relations
is based on a nal set of loading conditions and not their history. 4: Calculate U via Eq. (7)
5: Calculate W e via Eq. (8)
Path dependence could be considered via incremental formula-
6: Form potential energy of the plate P U  W e
tions, which have been applied to the modeling of a broad spec- 7: Evaluate @@cPi 0; i 1; . . . ; N
trum of dissipative problems associated with inelastic material 8: Solve the system of N equations for ci ; i 1; . . . ; N unknowns
response (Ortiz and Repetto, 1999; Ortiz and Stainier, 1999; 9: IF only one solution for ci ; i 1; . . . ; N found THEN
Miehe and Lambrecht, 2003), including thermomechanical a. Output such solution
(Bargmann and Steinmann, 2008) as well as electro- and magne- ELSE
a. Check the stability range for each solution in the set via the
tomechanical problems (Miehe et al., 2011,). However, these @2 P
assessment of the eigenvalues of @ci @cj ; i; j 1; . . . ; N
approaches are beyond the scope of this work where the focus is
b. Output each solution with its associated stability range
to obtain closed-form expressions for the deformation of
10: Form displacement, strain, curvature, stress elds, etc. using each
SMA-based plates. The current implementation limits the problem solution from step 9 (i.e., deformation eld(s) and corresponding
to a single reverse transformation from martensite to austenite (i.e., stability range(s))
no transformation cycles), and to the assumption of proportional
loading. The aforementioned limitations do not hinder the
approach from its primary goal of providing sets of stable shapes
attainable from SMA-based plates under prescribed loading condi- where S is the fourth-order compliance tensor (S 1 C), a is the
tions during preliminary design stages.
second-order thermoelastic expansion tensor, et0 is the
The procedure used to approximate the deformation of plates
second-order pre-strain tensor, and T 0 is the reference temperature.
with SMA constituents is provided in Table 1. The approach allows
The pre-strain tensor et0 represents the initial transformation strain
for exibility in the geometry of the plates, number of layers, con-
state in the SMA and it allows for an initial stress free oriented
stituents for each layer, and boundary conditions. It should be
martensite state. From Eq. (10), the following expression for the
emphasized that, although only SMA transformation strains are
actuation strain e A is obtained (cf. Eq. (7)):
considered as the source of actuation strains, other types of inelas-
tic strains can be included in the present formulation as long as e A aT  T 0 et  et0 : 11
they are sufciently small to be appropriately modeled using the
The effective compliance tensor is obtained using a rule of mix-
presented kinematics.
Alternative formulations of the presented approach can be tures between the compliance tensors of austenite S A and marten-
obtained by exchanging between given and unknown elds. To site S M as follows:
give an example, in Section 4 the problem of nding the tempera-
S S A nS M  S A S A nDS: 12
ture eld required to obtain a goal out-of-plane displacement eld
is considered. In such an example, the out-of-plane displacement For reverse transformation, the inelastic transformation strain
eld uz x) is considered to be a known goal displacement eld evolves such that the time rate of change of both its magnitude
and the temperature eld Tx is found by proposing additive sets and direction are given as:
of ansatz temperature elds and following this solution procedure.
_ t ; Kt etr
e_ t nK rev rev ; 13
2.3. Constitutive model for monolithic SMA nr
where Ktrev is the transformation direction tensor for reverse trans-
2.3.1. Model formulation formation, etr is the transformation strain tensor at transformation
The detailed description and foundation of the constitutive reversal (i.e., the end of the previous forward transformation), and
model for monolithic SMAs used in this work is provided in nr is the martensitic volume fraction at transformation reversal.
Lagoudas et al., 2012 and has been used to analyze a broad spec- For the cases considered in this work, prior full forward transforma-
trum of smart structures with SMA components (Hartl et al., tion is assumed (as it typically occurs during processing), thus
2010, 2011). The complete model formulation presented in nr 1 and Ktrev etr et0 . Constraints on the evolution of the
Lagoudas et al. (2012) considers both forward and reverse transfor-
martensite volume fraction n must be dened. To this end, the
mation under both mechanical and thermal loading in a thermody-
transformation function Utrev is introduced. The constraints on the
namically consistent manner. All the studies in the present work
evolution of martensite volume fraction are then given as:
consider single transformation from oriented martensite to austen-
ite (i.e., heating in the current case, see phase diagram in Fig. 2) and Utrev 6 0; n_ Utrev 0; 0 6 n  1; 14
and therefore only the formulation for reverse transformation is
presented here for the sake of brevity. where the rst two represent the KarushKuhnTucker constraints
The model is three-dimensional, phenomenological, and it con- and the third bounds the martensitic volume fraction that ranges
siders three external state variables: the second-order (Cauchy) from 0 (pure austenite) to 1 (pure martensite). The transformation
stress tensor r, the second-order total innitesimal strain tensor function for reverse transformation is given as follows:
e, and the scalar absolute temperature T. The internal state vari- 1
ables associated with the inelastic transformation process of Utrev 1 Dr : Ktrev  r : DS r  qDs0 T qDu0 f trev n  Y t0 ;
2
SMAs are the scalar martensite volume fraction n and the 15
second-order transformation strain tensor et . Assuming a general
reference state in which the stress and total strain are zero but where D; qDs0 ; qDu0 , and Y t0 are parameters that are calibrated
the transformation strain is not, the stress is given as follows: from an experimentally determined stresstemperature phase dia-
gram (see Section 4 in Lagoudas et al. (2012) for model calibration
r S 1 e  aT  T 0  et et0 ; 10 t
details). The hardening function f rev n has the following form:

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
6 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx

t 1 the expression of Utrev in Eq. (17), the following explicit expression

f rev n a2 1 nn3  1  nn4  a3 ; 16
2 for martensite volume fraction is obtained:
where a2 and a3 are parameters calibrated from the stresstemper-
8
< 1; T < As
ature phase diagram, and n3 and n4 are the hardening coefcients n 1 Dr : Ktrev qDs0 Af  T : 18
chosen to best t the transition between the elastic and the : ; T P As
qDs0 Af  As
phase-transformation response regimes during reverse
transformation. For the plane stress problems considered in this work, the seven
variables required to dene the state of the material (given total
2.3.2. Simplication for the analytical approach strains and temperature) are: the three in-plane stress components
Certain assumptions are made to simplify the previous consti- (rxx ; ryy , and rxy ), the three in-plane actuation strains (exx
A A
; eyy , and
tutive model for SMAs and allow for direct symbolic computation exyA ), and the martensite volume fraction (n). The seven equations
of the martensite volume fraction and stress in the process to be solved for these variables are: Eq. (18) and the plane stress
described in Section 2. First, the Youngs moduli of austenite and versions of Eqs. (10) and (13). The explicit forms of the aforemen-
martensite are assumed to be equal. Similarly, the Poissons ratios tioned equations are the following:
of austenite and martensite are assumed to coincide. These 2 3 2 302 3 2 A 31
assumptions ensure DS 0, simplifying the calculation of the stiff- rxx 1 m 0 exx exx
6 7 E 6 7B6 7 6 e A 7C
ness tensor used in Eq. (7); the alternative requires the use of Eq. 4 ryy 5 4 m 1 0 5 @4 eyy 5  4 yy 5A; 19
1  m2
(12) and inversion of the obtained compliance tensor. This assump- rxy 0 0 1m e xy exyA
tion is sufciently accurate to be applied to the SMA material typ-

ically used to fabricate self-folding sheets (SmartFlex NiTi wire 2 3 2 3 2 3 2 3
exxA etxx et0
xx et0
xx
from SAES Getters) since the Youngs moduli of austenite and 6 e A 7 6 et 7 6 et0 7 6 et0 7
martensite of such material differ only by 10.6% (see 4 yy 5 4 yy 5  4 yy 5 n  14 yy 5; 20
Peraza-Hernandez et al., 2014b). Thermoelastic expansion strains exyA etxy et0
xy et0
xy
are assumed to be negligible since the recovery of strains due to
martensitic transformation is signicantly larger in magnitude. r : Ktrev rxx et0 t0 t0
xx ryy eyy 2rxy exy ; 21
The magnitude of the transformation strain H, which is gener-
ally a function of stress, has been shown to be nearly constant in cf. Eq. (18).
this particular material (Peraza-Hernandez et al., 2014b). This sim-
plies the calibration of the material parameters from Eq. (15). 2.4. Constitutive model for the effective response of SMA wire meshes
Finally, to be able to obtain a closed-form expression for n as a
function of stress and temperature (i.e., to invert Eq. (15) to form One option to decrease the computational demand and simplify
nr; T), n3 and n4 are set to unity. This corresponds to a linear the analysis of applications including periodic SMA wire meshes is
hardening response. Table 2 contains the values of the SMA mate- to represent these meshes using equivalent effective continuum lay-
rial parameters assumed in this work that are largely based on ers. The advantages of modeling periodic truss-like and beam-like
those of Peraza-Hernandez et al. (2014b). Under these assump- structures using continuum representations have been discussed
tions, the transformation function takes the reduced form: in the literature for decades and include computational efciency,
adequacy for design trade studies, and simplicity of the domain
Utrev 1 Dr : Ktrev qDs0 Af  T  nAf  As ; 17 geometry (Nayfeh and Hefzy, 1978, 1981; Noor, 1988). For this
particular work, it also provides the advantage of adequacy for
where D C M  C A =C M C A and qDs0 2HC M C A =C M C A . symbolic computation. The continuum effective response of SMA
The transformation temperatures As and Af and the stress inuence wire meshes was explored previously (Peraza-Hernandez et al.,
coefcients C M and C A are obtained from a stresstemperature 2013c); however, such a model was limited to two sets of equally
phase diagram such as the one schematically shown in Fig. 2. spaced parallel SMA wires oriented orthogonally to each other.
Combining the KarushKuhnTucker constraints in Eq. (14) with The present model is an extension that contains the model of
Peraza-Hernandez et al. (2013c) as a special case and has these
additional features: meshes can be composed of p sets of parallel
Table 2
SMA wires, it is possible to account for a different volume fraction
Shape memory alloy material parameters assumed in this work. The parameters are
associated with the model presented in Lagoudas et al. (2012) and follow from for each set of parallel SMA wires, and the alignment direction of
Peraza-Hernandez et al. (2014b). each set of parallel SMA wires is arbitrary.
Property Value
2.4.1. Model formulation
(Thermoelastic properties)
The considered meshes are composed of p sets of parallel SMA
A
E E M
E 45.5 GPa
wires. The alignment direction of each set of parallel SMA wires
m A mM m 0.33
0
is dened by the structural vector an , where n 1; . . . ; p. For
a A aM
instance, Fig. 4 illustrates meshes of p 2 and p 3.
(Phase diagram properties)
The second-order (Cauchy) stress tensor r is assumed to be
Ms 332.1 K
Mf 302.1 K decomposed into the contributions from the SMA wires and the
As 337.4 K wire joints of the mesh as follows:
Af 362.4 K !
9.54 MPa/K
X
p X
p
CA r an  an v n rn 1  v n rm ; 22
CM 9.0 MPa/K
n1 n1
(Transformation strain properties)
Hmin Hsat H 0.044
where v n is the volume fraction of the set of parallel wires n; rn is
the scalar uniaxial stress in the set of parallel wires n, and rm is
( Smooth hardening properties)
ni ; i 1; . . . ; 4 1
the second-order stress tensor contribution from the wire joints
and it is assumed to be given by rm C m e where C m is the effective

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx
a2
a2 a1
a1
a3 a2
a1
Fig. 4. Meshes of SMA wires. The vectors an ; n 1; 2 or 1; 2; 3 represent the
alignment directions of each set of parallel SMA wires.
fourth-order stiffness tensor of the wire joints. For instance, in
Peraza-Hernandez et al. (2013c) a constant shear stiffness due to
the wire joints was added to the model of two sets of parallel wires
oriented orthogonally to each other in order to obtain a positive def-
inite stiffness tensor. The uniaxial stress in the SMA wires rn is
obtained from a one-dimensional version of the SMA model pre-
sented in Section 2.3. The SMA material parameters shown in
Table 2 are also used when considering this model. The uniaxial
strain in the SMA wires en is obtained from the projection of the
total strain tensor onto the alignment direction of the SMA wires
(en an  an : e an  ean ).
3. Analysis of the deformation of SMA-based plates
The symbolic computational procedure for the analysis of
SMA-based plates summarized in Table 1 has been implemented
in Maple. This software has robust support for the symbolic com-
putation needed to solve the system of equations obtained in Eq.
(9) and to provide closed-form relations for subsequent analytical
study.
Different boundary value problems are considered to test the
utility of the framework and to analyze the behavior of plates com-
prised of different materials and laminate congurations. The spa-
tial domain of the boundary value problems consists of a circular
plate with thickness h and radius R. The temperature of the circular
plate is assumed to vary only in the plate radial direction in a linear
manner. Specically, it is given as follows:
1=2
x2 y2
T T c T e  T c ; 23
R
where T is the local absolute temperature, T e is the temperature at
the edge of the plate, and T c is the temperature at the center of the
plate. The plate is xed with respect to translations and rotations at
its center (x y 0) to avoid rigid body modes.
The lowest applied temperature (either at the center of the
plate or at the edge) is set to the austenite start temperature As
to ensure that transformation is occurring everywhere in the plate.
The maximum applied temperature is kept at or below the austen-
ite nish temperature Af . This temperature range guarantees that
the martensite volume fraction stays in the range 0; 1 everywhere
in the plate.
For the sake of simplicity and computational efciency, the cho-
sen form of the ansatz displacement elds of the plate mid-plane
contains the following terms:
Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
7
ux c1 x c2 xy c3 x2 c4 y2 c5 x2 y c6 xy2 c7 x2 y2 c8 x3 c9 y3 ;
uy c10 y c11 xy c12 x2 c13 y2 c14 x2 y c15 xy2 c16 x2 y2
c17 x3 c18 y3 ;
uz c19 xy c20 x2 c21 y2 :
24
It can be shown that the displacement eld of Eq. (24) satises the
kinematic boundary conditions regardless of the values of ci ,
(i 1; . . . ; 21). The inclusion of terms for ux and uy of higher-order
than those in Eq. (24) is not found to signicantly affect the
obtained out-of-plane deformation of the plates for the problems
considered in this work. The out-of-plane displacement eld in
Eq. (24) is based on the assumption of uniform Gaussian curvature
throughout the plate. This allows for the simplication of the sys-
tem of N equations (Eq. (9)) that has to be solved symbolically for
each problem and reduces the complexity of the obtained displace-
ment eld expressions.
3.1. Example 1: deformation of an SMA plate
In this example, a plate composed of homogeneous isotropic
SMA is considered (Fig. 3(a)). The SMA is homogeneously
pre-strained in the xy plane, which means that it can provide half
of its maximum uniaxial transformation strain H on any two
orthogonal directions in such a plane (et0 H=2; H=2; 0T ; see
Section 2.3 for details of the constitutive model and values for
the SMA material parameters). The expressions for the displace-
ment eld of this boundary value problem are obtained using the
method summarized in Table 1. The full symbolic expressions are
rather lengthy; to allow for simple closed-form expressions, the
numerical values of the SMA material parameters of Table 2 are
assigned while the dimensions of the plate and the temperature
input parameters are left as symbolic entities. For the sake of com-
pleteness, the full symbolic expressions for this particular example
are presented in this Section. The results here show three different
response regimes that are specied in Table 3. Standard SI units are
considered (meters, kilograms, Kelvin, seconds and those derived
from them). For this and the subsequent examples, the curvature
principal directions v jI and v jII are provided using the following
convention:
   
cosh sinh
vj I
; vj II
; 25
sinh cosh
where the parameter h denes the orientation of the principal direc-
tions (h increases counterclockwise and h 0 corresponds to the
positive x-direction). The orientation of the principal directions
may not be unique for certain deformed congurations, in fact, some
congurations have innitely many possible orientations of the
principal directions. In view of this, all the possible principal direc-
tions that may be obtained for each conguration are provided.
The rst regime corresponds to a deformed but planar congu-
ration (i.e., there are no out-of-plane displacements). This planar
deformed conguration is stable when the temperature difference
between the plate center and the edge does not exceed specic
2 2
upper and lower bounds (39934:1 Rh2 and 3001:0 Rh2 , respectively).
In a symbolic representation, the bounds for this regime are as
follows:
b4 b2
 < Tc  Te < : 26
b3 b1
The parameters b1 ; b2 ; b3 , and b4 depend on the SMA material
parameters and the plate geometry as follows:
2
b1 8Af  As 1  m2 HC A R2 ; 27
8 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx

Table 3
Stable/multistable regimes for the SMA circular plate under radially varying temperature.
 
Temperature range jI Principal direction orientations
Principal curvatures
jII
2 2
3001:0 hR2 < T c  T e < 39934:1 Rh2 Planar conguration 0 6 h < 2p
2   
2 1=2

0 6 h < 2p
T c  T e P 39934:1 Rh2 1
 0:0531
R T c  T e  39934:1 Rh2
1
2    
2 1=2 0 6 h < 2p
3001:0 hR2 P T c  T e 1
 0:0531
R T e  T c  3001:0 Rh2
1

 2

2 2
b2 5 2Af  As m2 4m  5C A 4Af  As 1  mEHC A EH h ;
 2
 The plate Gaussian curvature is explicitly related to the radial
b3 4Af  As 2 1  m2 C A 4Af  As 1  mEHC A EH2 HR2 ;
 2
b4 5 2Af  As 3 m2 4m  5C A 4Af  As 2 1  mEHC A
 
2
Af  As EH2 h :
As the plate thickness to radius ratio (h=R) increases, the mag-
nitudes of these bounds increase, thereby expanding the tempera-
ture range of this planar deformed conguration regime.
Specically, a quadratic dependence of such bounds on h=R is
determined (see Table 3). The constant factor multiplying the
square of h=R is one order of magnitude smaller for the lower
bound than that of the upper bound. This indicates that radial tem-
perature gradients of smaller magnitude are needed to induce
instability of the planar deformed conguration when the temper-
ature increases in the radial direction than when it decreases.
When (T c  T e ) exceeds the aforementioned upper bound
2
(T c  T e P 39934:1 Rh2 ), the plate enters a multistable regime with
a deformed conguration associated with negative Gaussian curva-
ture. The  in the curvatures expression for this regime (see
Table 3) indicates two stable principal curvatures of the plate.
The symbolic representation of the principal curvatures in this
regime is the following:
   
jI 1=2 1
b5 b1 T c  T e  b2 ; 31
jII 1
where
1 A plate of radius R 100 mm and thickness h 1 mm is consid-
b5 : 32
51=2 Af  As 1 m1=2 1  m1=2 C A R2
Since geometry, boundary conditions, and material parameters
are axisymmetric with respect to the normal direction at the center
of the plate, the orthogonal principal directions may be oriented
towards any direction in the xy plane (i.e., this conguration is
multistable since its principal curvatures are different and
may be oriented towards any direction in the xy plane). For exam-
ple, the congurations {jx jI ; jy jII ; jxy 0} and {jx 0;
jy 0; jxy jI } are energetically equivalent and both correspond
to a Gaussian curvature of g jI jII < 0.
If (T c  T e ) exceeds the lower bound of the stability range for
the planar deformed conguration (i.e., T c  T e 6 3001:0 Rh2 ),
plate enters a bi-stable regime associated with positive Gaussian
curvature. In a symbolic representation, the curvatures of this
regime are the following:
   
jI 1=2 1
b6 b3 T c  T e  b4 ; 33
jII 1 zero for all the simulations performed.
where
2101=2
b6 : 34
5Af  As 1=2 2Af  As 1  mC A HER2
2 This conguration is axisymmetric and thus the orthogonal
principal directions are indistinguishable from any two orthogonal
28 directions in the xy plane.
29 temperature gradient. For a positive radial temperature gradient,
the transformation strain recovery increases towards the edge of
the plate. This leaves the plate edge in hoop tension and forces
the plate center into compression, hence the plate center buckles
30 and the plate deforms to a positive Gaussian curvature congura-
tion. Similarly, for a negative radial temperature gradient the
transformation strain recovery increases towards the center of
the plate. This leaves the plate center in tension forcing the plate
edge into compression. Thus, the plate buckles at the edge and it
deforms to a negative Gaussian curvature conguration.
Non-linear FEA is used to numerically validate the results of the
analytical approach. The FEA mesh is shown in Fig. 5.2 Finite ele-
ment analysis is performed using Abaqus and a user material sub-
routine (UMAT) that provides the response of the SMA material
(Lagoudas et al., 2012), see Section 2.3 for details. Abaqus S3 ele-
ments are used (3-node triangular general purpose shell, nite mem-
brane strains) due to the circular shape of the domain. A mesh
convergence study was performed for the selection of the mesh
renement and the nal FEA model has a total of 28,686 degrees
of freedom (in contrast to the 21 degrees of freedom selected for
the analytical approach). Quasi-static implicit dynamic analysis
steps and geometric non-linearity are employed (i.e., the effect of
arbitrarily large local rotations has been satisfactorily accounted
for, even under this small strain assumption using the algorithm
of Hughes and Winget, 1980 (Hartl and Lagoudas, 2009)). No
damping-related material parameters are utilized for the implicit
dynamic analysis steps.3
ered for numerical validation. To induce out-of-plane deformed
congurations (i.e., to alleviate numerical problems related to
structural instability), an articial and relatively small
gravity-like volumetric bias force in the z-direction (1 m/s2) is
applied during the nite element analysis. For the analysis of the
conguration with negative Gaussian curvature, slight point loads
(0.01 N) in the z-direction were applied at y R to bias the solu-
tion for which one of the principal directions is aligned with the
y-direction.4
The response metric used for comparison between FEA and the
analytical solutions is the maximum absolute out-of-plane dis-
2
placement umax . The results for the range T c  T e P 39934:1 Rh2 of
z
2
the
2
The global coordinate system [X,Y,Z]T is shown in Fig. 5 and subsequent gures
for the sake of visualization. It is aligned with the plate local coordinate system
x; y; zT that has its origin at the center of the plate.
3
In addition, the energy dissipated by automatic stabilization (ALLSD in Abaqus) is
4
To obtain the magnitude of the applied bias loads, multiple simulations of the
plates are performed, iteratively increasing the bias load magnitudes for each
simulation until loads that induce out-of-plane deformations in the planar unstable
regimes are found. Increments of 0.05 m/s2 were used for the articial gravity-like
volumetric bias force and 0.001 N for the point loads until the reported values were
found.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx 9

Fixed node 15

10
Series2

5
Series1

0
0 5 10 15 20 25

Fig. 5. Circular plate showing mesh and node xed in translations and rotations. 0 7 14 0 0.5 1

15 Z

10 Y
Closed-form h = 0.1
mm X
5 Series1
Fig. 7. The top plot shows normalized maximum out-of-plane displacement for the
2
0 SMA plate in the temperature range T c  T e 6 3001:0h =R2 . The plate is bi-stable
0 5 10 15 20 25 with deformed congurations of positive Gaussian curvature. FEA contour plots of
martensite volume fraction and normalized out-of-plane displacement are also
shown corresponding to the case circled in the plot.

12 0 12 0 0.5 1 To compare the global deformation response between the ana-

Z T c  T e 25 K) and positive (at T e  T c 25 K) Gaussian curvature
Y
X an observed difference between the analytical quadratic
Fig. 6. The top plot shows normalized maximum out-of-plane displacement for the
2
SMA plate in the temperature range T c  T e P 39934:1h =R2 . The plate is multi-
stable with deformed congurations of negative Gaussian curvature. FEA contour
plots of martensite volume fraction and normalized out-of-plane displacement are
also shown corresponding to the case circled in the plot.
deformed congurations with negative Gaussian curvature are pre-
sented in Fig. 6, where maximum normalized out-of-plane dis-
placement vs. T c  T e are shown. A good agreement between
FEA and the analytical results is found. FEA contour plots of
martensite volume fraction and out-of-plane displacement uz nor-
malized by plate thickness are shown for a specic case circled in
the plot for the sake of visualization. A movie of this simulation is
included in the supplementary material of this article. Similarly,
2
the results for the range T c  T e 6 3001:0 Rh2 of deformed congu-
rations with positive Gaussian curvature are shown in Fig. 7, where
maximum normalized out-of-plane displacement vs. T e  T c are
presented. In this case, excellent agreement is observed between
FEA and the analytical results. A movie of this simulation is also
included in the supplementary material of this article.
The implicit-dynamic steps did not develop global convergence
issues, even at the onset of the change from the planar to the
curved conguration regimes. This type of analysis step provides
better convergence characteristics, as compared to static analysis
steps, since kinetic energy is developed during the switch between
stable curvature regimes and alleviates numerical convergence
issues. The applied temperature is raised over a signicantly long
period of time (1000 s) so that the kinetic energy remains negligi-
ble compared to the strain energy (6 or more orders of magnitude
difference). Alternatively, arc-length solution techniques (Ramm,
1981; Criseld, 1981), such as Riks in Abaqus, are other feasible
options and have been used in the literature for bi-stable shells
(Pirrera et al., 2012).
lytical solutions and FEA, the normalized out-of-plane displace-
ment along the y-direction is plotted for the cases of negative (at
deformed congurations in Figs. 8(a) and (b), respectively. There is
out-of-plane deformation eld and the obtained FEA solution in
both cases. This is due to the fact that a quadratic out-of-plane dis-
placement cannot capture the moment-free edge boundary condi-
tions as also pointed by Seffen and Maurini, 2013. Higher-order
polynomial ansatz functions (Pirrera et al., 2010) for the
out-of-plane displacement eld are needed to capture such a
behavior. Nevertheless, the obtained solutions provide a fair quad-
ratic approximation of the FEA solution.
3.2. Example 2: deformation of a laminated plate with meshes of
orthogonally oriented SMA wires
A plate comprised of multiple SMA mesh-based laminae is con-
sidered herein. The laminate layup is shown in Fig. 9; it consists of
two outer layers of SMA meshes separated by a middle elastomeric
layer. Each layer has equal thickness. This system has been studied
elsewhere but considering temperature varying through the
thickness of the plate to achieve bending deformation
(Peraza-Hernandez et al., 2013a,e). The elastomer layer is assumed
to behave as a linear elastic material with effective Youngs
modulus Eel , Poissons ratio mel , and negligible thermal expansion.5
The effective response of the SMA wire mesh layers is modeled using
continuum representative layers. The effective continuum model
formulation for the SMA wire meshes was presented in
Section 2.4. In this example, the sets of wires that compose the
SMA meshes are aligned with the x and y-directions as shown in
Fig. 9, where the vectors a1 and a2 in this gure are collinear to
the alignment direction of the wires in the mesh. The volume frac-
tions of each set of parallel wires composing the mesh layers are
denoted v 1 and v 2 (cf. Eq. (22)). The tensile pre-strain in the axial
direction of the SMA wires is H.
5
Linear elastic behavior is assumed for the small strain range (< 5%) which the
middle layer material experiences during morphing of the plate.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
10 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx

FEA Analytical If (T c  T e ) exceeds the lower bound of the stability range for
2
the planar deformed conguration (i.e., T c  T e 6 136773:1 Rh2 ),
15 the plate enters a bi-stable regime with deformed congurations
10 of positive Gaussian curvature (see Table 4). This conguration is
uzh 1

axisymmetric and thus the orthogonal principal directions are

5 indistinguishable from any two orthogonal directions in the xy
(a) plane.
-1 0 -0.5 0 0.5 1 The out-of-plane deections and curvatures attained by the
-1 -0.5 0 0.5 1 SMA mesh-based laminate are expected to be larger in magnitude
yR 1 than those attained by the isotropic SMA plate (Section 3.1). It
15 should be noted that the key factor providing the difference
between the deformations of the two plates is not the fact that
10 one is single-layered and the other is multi-layered but rather
uzh 1

the difference in the pre-strain magnitude of their SMA

5 constituents (cf. Fig. 3 and Peraza-Hernandez et al., 2013d). The
(b)
0 in the xy plane simultaneously to a magnitude of H=2 (due to
-1 -0.5 0 0.5
yR 1
Fig. 8. Normalized out-of-plane displacement vs. normalized radial distance in the
y-direction for (a) T c  T e 25 K and (b) T e  T c 25 K (thermal states circled in
the plots of Figs. 6 and 7, respectively).
To obtain simple closed-form expressions for this laminated
plate subject to the constant-through-thickness temperature eld
provided in Eq. (23), the numerical values of the SMA material
parameters of Table 2, the properties for the elastomeric middle
layer (Eel 0:6 MPa, mel 0:45), and the volume fractions for the
wire mesh conguration (v 1 v 2 0:025) are assigned to the
obtained symbolic expressions while leaving the dimensions of
the plate and the temperature input as symbolic entities. Three dif-
ferent response regimes are obtained and summarized in Table 4.
The rst regime corresponds to a deformed but planar congura-
tion. This planar deformed conguration is stable when the tem-
perature difference between the plate center and the edge does
not exceed specic upper and lower bounds (84:2 Rh2 and 
2 between the dense SMA plate and this SMA mesh-based laminate,
136773:1 Rh2 , respectively).
When (T c  T e ) exceeds the upper bound, the plate enters a
multistable regime with deformed congurations associated with
negative Gaussian curvature (see Table 4). The orthogonal principal
directions of this conguration are oriented at 45 from the wire
alignment directions. These principal directions are expected since
the SMA mesh layers have their lowest in-plane stiffness in such
directions. Generally, the minimum potential energy conguration
depends on the material anisotropy, geometry, and the applied or
internal loads. For this particular case, the geometry and the
applied thermomechanical loading are axisymmetric with respect
to the plate central normal axis; thus, the minimum potential
energy conguration depends solely on the material anisotropy.
isotropic SMA plate is pre-strained in two orthogonal directions
1 the isochoric transformation strain response in SMAs); the wires
of the SMA mesh-based laminate are pre-strained with a magni-
tude of H in their axial directions (neglecting intersection effects).
This allows for larger transformation strain recovery in the SMA
mesh during martensitic transformation and thus larger global
deections and curvatures. This behavior is indeed predicted by
1
the analytical approach. Analytical results of umax
z h for the SMA
plate and the SMA mesh-based plate are shown in Fig. 10 for plates
of R 100 mm and h 1 mm. These results correspond to the neg-
ative Gaussian curvature regime. It is observed that the SMA
mesh-based plate provides for signicantly larger out-of-plane
displacements. The same trend occurs for the positive Gaussian
curvature regime.
Non-linear FEA is also used to numerically validate the analyt-
ical results of this example. The discretization of the circular plate,
analysis procedures, and numerical values for the plate dimensions
used for the previous example are also utilized here. An articial
and relatively small gravity-like force of 0.025 m/s2 is applied to
the FEA model in order to induce out-of-plane congurations.6 It
2
should be noted that, due to the signicant difference in stiffness
the bias force magnitude required for this example is two orders
of magnitude smaller than that of the dense SMA plate.
In the Abaqus model, a composite shell section is used to
capture the laminate behavior of this particular plate. The UMAT
associated with this model provides the effective response of the
SMA wire mesh layers using the same constitutive relations as
the analytical model, which were presented in Section 2.4.
The maximum absolute out-of-plane displacement umax is again
z
used as the metric of comparison between FEA and the analytical
2
solutions. The results for the range T c  T e P 84:2 Rh2 of deformed
congurations with negative Gaussian curvature are shown in
Fig. 11. There is again a good agreement between FEA and the
analytical results. As predicted, the directions of the principal
curvatures for this conguration are oriented at 45 from the wire
alignment directions. This is observed in the deformed congura-
tion plots of Fig. 11.
2
The results for the temperature range T c  T e 6 136773:1 Rh2
are shown in Fig. 12. In this case, there is only qualitative agree-
ment between the analytical and FEA results in terms of the stable
regime limit for the planar conguration. As previously stated, the
uniform Gaussian curvature description of Eq. (24) does not

Fig. 9. Laminate layup for the SMA mesh-based laminated plate and unit cell of
the SMA mesh with orthogonally oriented wires.
6
The same procedure used to obtain the magnitude of the applied bias load
presented in Section 3.1 is used for this example. In this case, multiple simulations
with increments of 0.005 m/s2 in bias load for each one were considered until the
aforementioned value of 0.025 m/s2 was found to induce out-of-plane deformations
of the plate in the planar unstable regimes.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx 11

Table 4
Stable/multistable regimes for the circular plate with outer layers of meshes of orthogonally oriented SMA wires under radially varying temperature.
 
Temperature range jI Principal direction orientations
Principal curvatures
jII
2 2
136773:1 Rh2 < T c  T e < 84:2 Rh2 Planar conguration 0 6 h < 2p
   
h p4 ; 34p ; 54p ; 74p
2
2 1=2 1
T c  T e P 84:2 Rh2  0:0735 T c  T e  84:2 Rh2
R 1
2    
2 1=2 0 6 h < 2p
136773:1 Rh2 P T c  T e 1
 0:0735
R T e  T c  136773:1 Rh2
1

20 results show that the magnitude of the radial temperature gradient

15
10
5 of largest deections and thus largest strains are oriented at 90
0 tions, these directions coincide with those of lowest stiffness. This
0 5 10 15 20 25
Fig. 10. Normalized maximum out-of-plane displacement vs. (T c  T e ) for the SMA
plate and SMA mesh-based plate in the negative Gaussian curvature regime.
capture the zero moment edge boundary conditions and rather
corresponds to an interior solution. Also, as pointed by Diaconu
et al. (2009), a uniform Gaussian curvature kinematic assumption
may show larger magnitudes for the limits of the planar stable con-
guration regime. Such differences are observed in both Figs. 12
and 6. These differences are due to the fact that such a model
restricts the plate to deform out-of-plane following only uniform
Gaussian curvature shapes. These considerations are addressed in
the works of Pirrera et al. (2010, 2012) where higher order polyno-
mials for the displacement eld were utilized. The current model is
limited to only provide analytical solutions for a uniform Gaussian
curvature ansatz.
Minimization of the total potential energy prescribes the transi-
tion from planar to curved congurations (i.e., such transition
occurs when the total potential energy is lower if the plate adopts
a curved conguration rather than a planar one). The magnitude of
the thermal gradient required to morph the at plate to a curved
shape is dictated by this condition. Here, both analytical and FEA
required to morph the plate towards a negative Gaussian curvature
conguration is lower than that required to morph the plate
towards a positive Gaussian curvature conguration. For the
obtained negative Gaussian curvature conguration, the directions
from one another. Under the given axisymmetric boundary condi-
leaves lower strains at the directions of largest stiffness (those col-
linear to the SMA wires). This is not the case for the positive
Gaussian curvature conguration, where the plate has deections
of similar magnitude everywhere along its periphery. Thus, the
strains in such a conguration are similar at every direction. This
increases the strain energy stored in this conguration since there
exist signicant strains along the stiffer directions of the plate. As a
result, a larger thermal gradient is required for this conguration
to have lower strain energy than the planar shape.
3.3. Example 3: deformation of a laminated plate with meshes of non-
orthogonally oriented SMA wires
The layup of the SMA mesh-based laminated plate considered
in this third example is shown in Fig. 13; it again consists of two
outer SMA mesh layers separated by a middle elastomeric layer.
Each layer has equal thickness. The elastomer material parameters
from Section 3.2 are again used in this example. In this case the
sets of wires that comprise the SMA meshes are oriented at 30
from the x-direction as shown in Fig. 13 (i.e., they are not orthog-
onal). These two sets of wires have a volume fraction of 0.025 each
and a tensile pre-strain magnitude in their corresponding axial
directions of H.

20
15
Closed-form h = 0.1
mm
10
FEA h = 0.1 mm
5
0
0 5 10 15 20 25

17.5 0 17.5 0 0.5 1

a2
Z
a1
Y
X
2
Fig. 11. The top plot shows normalized maximum out-of-plane displacement for the SMA mesh-based plate in the temperature range T c  T e P 84:2h =R2 . The plate is
multistable with deformed congurations of negative Gaussian curvature. FEA contour plots of martensite volume fraction (wires in the a1 direction) and normalized out-of-
plane displacement are also shown corresponding to the case circled in the plot.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
12 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx

2
Fig. 12. The top plot shows normalized maximum out-of-plane displacement for the SMA mesh-based plate in the temperature range T c  T e 6 136773:1h =R2 . The
plate is bi-stable with deformed congurations of positive Gaussian curvature. FEA contour plots of martensite volume fraction (wires in the a1 direction) and normalized
out-of-plane displacement are also shown corresponding to the case circled in the plot.

loading are axisymmetric and thus the minimum potential energy

Fig. 13. Laminate layup for the SMA mesh-based laminated plate and unit cell of
the SMA wire mesh with non-orthogonally oriented wires.
Once again, the results show three different response regimes
as summarized in Table 5. The rst corresponding to a congura-
tion with in-plane deformation only. This planar deformed cong-
uration is stable when the temperature difference between the
plate center and the edge does not exceed specic upper and lower
2 2
bounds (279:9 Rh2 and 77683:0 Rh2 , respectively). When (T c  T e )
exceeds the upper bound, the plate enters a bi-stable regime with
deformed congurations of negative Gaussian curvature. The two
orthogonal principal directions associated with this conguration
are collinear with the y and x-directions, respectively (i.e.,
jy jI ; jx jII ; jxy 0). These principal directions are expected
since the SMA mesh layers have their lowest in-plane stiffness in
the y-direction. Therefore, the direction of the largest curvature
in the plate is aligned with this axis. As noted in Section 3.2, in
these particular examples the geometry and thermomechanical
conguration depends only on the material anisotropy.
If T c  T e exceeds the lower bound of the stability range for
2
the planar deformed conguration (i.e., T c  T e 6 77683:0 Rh2 ),
the plate enters a bi-stable regime with deformed congurations
associated with positive Gaussian curvature. For this conguration
the orthogonal principal directions jI and jII are also collinear with
the y and x-directions, respectively. Unlike the studied congura-
tions of the isotropic SMA plate and the laminated plate with
SMA meshes of orthogonally oriented wires, the deformed cong-
urations in this problem do not have principal curvatures of equal
magnitude (cf. Tables 3 and 4). This is due to the pronounced ani-
sotropy of the SMA mesh layers, where the directions of largest and
lowest in-plane stiffness are oriented towards orthogonal direc-
tions (the x and y-directions, respectively, see Fig. 13). This charac-
teristic allowed for a natural alignment of the orthogonal principal
directions exactly in the directions of lowest and largest in-plane
stiffness and thus principal curvatures with different magnitudes
arose. By comparing the results of the laminated plate with orthog-
onal and non-orthogonal SMA meshes, it is observed that utilizing
the tailored alignment of the SMA wires in the meshes as an addi-
tional degree of freedom may allow for a broader spectrum of
deformed congurations given equal plate dimensions and bound-
ary conditions.
For numerical validation with FEA, the dimensions and dis-
cretization of the circular plate and analysis procedures used for
the previous example are also utilized here. An articial
gravity-like force of 0.0005 m/s2 (found via the procedures
described in the previous sections) is used for stable conguration
biasing. The out-of-plane displacement uz at x; y R; 0 and

Table 5
Stable/bi-stable regimes for the circular plate with layers of meshes of non-orthogonally oriented SMA wires under radially varying temperature.
 
Temperature range jI Principal direction orientations
Principal curvatures
jII
2 2
77683:0 Rh2 < T c  T e < 279:9 Rh2 Planar conguration 0 6 h < 2p
 1=2 

h p2 ; 32p
2
T c  T e P 279:9 hR2 2 2:97
 0:0333
R 279:9 Rh2
T c  T e 
1
   
h p2 ; 32p
2
2 1=2 2:97
77683:0 Rh2 P T c  T e  0:0333 T e  T c  77683:0 Rh2
R 1

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx 13

0; R are used as the response parameters of comparison between throughout the plate and largely increase the complexity of the
FEA and the analytical solutions. The results for the range equations needed to solve for the displacement eld, thereby
2
T c  T e P 279:9 Rh2 of deformed congurations with negative defeating the purpose of this computationally efcient analysis
tool.
Gaussian curvature are shown in Fig. 14. There is once again very
good agreement between FEA and the analytical results. As pre-
dicted, the directions of the principal curvatures are collinear with 4. Morphing an SMA plate towards a goal conguration
the x and y-directions. This is observed in the deformed congura-
tion plots of Fig. 14. The design of morphing structures with intrinsic material actu-
The results for the temperature range T c  T e 6 77683:0 Rh2 are
2 ation entails the challenge of conceiving effective actuation tech-
niques to morph the structures among different goal shapes. In
shown in Fig. 15. There is a good agreement between FEA and ana-
this work, where the intrinsic material actuation is provided by
lytical results for the out-of-plane displacement at x; y R; 0.
the SMA constituents of the plates, the determination of the ther-
However, signicant discrepancies are observed for the
mal input that would induce the morphing of the plate towards
out-of-plane displacement at x; y 0; R. The FEA results show
specic shapes is pursued. Similar problems have been addressed
a much stiffer response in this direction. The inclusion of
for laser forming of doubly curved shapes (Liu et al., 2004) and
higher-order terms in the ansatz out-of-plane displacement eld
for SMA-based plates (Peraza-Hernandez et al., 2014a) under iter-
may reduce this discrepancy; however, the inclusion of such terms
ative optimization frameworks. With the present novel approach,
would break the assumption of uniform Gaussian curvature
the applied conditions (e.g., the temperature eld) that will morph

Fig. 14. The top plot shows normalized out-of-plane displacement at x; y R; 0 and x; y 0; R for the SMA mesh-based plate in the temperature range
2
T c  T e P 279:9h =R2 . The plate is bi-stable with deformed congurations of negative Gaussian curvature. FEA contour plots of martensite volume fraction (wires in the a1
direction) and normalized out-of-plane displacement are also shown corresponding to the case circled in the plot.

20
Series2

15 Series1

10 Series3

5 Series4

0
0 5 10 15 20 25

Z
Y
0 7.5 15 0 0.5 1
X

60 a1
a2

Fig. 15. The top plot shows normalized out-of-plane displacement at x; y R; 0 and x; y 0; R for the SMA mesh-based plate in the temperature range
2
T c  T e 6 77683:0h =R2 . The plate is bi-stable with deformed congurations of positive Gaussian curvature. FEA contour plots of martensite volume fraction (wires in the a1
direction) and normalized out-of-plane displacement are also shown corresponding to the case circled in the plot.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
14 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx
the plate towards a goal shape and minimize the potential energy
stored in plate are directly computed (i.e., without requiring itera-
tive efforts).
For this problem, the plate geometry and material used in
Section 3.1 are considered (i.e., a circular SMA plate). Unlike the
problems in Section 3, where the objective is to approximate the
deformation of the SMA-based plates under given thermal condi-
tions, this problem is focused on the approximation of the temper-
ature distribution required to morph the plate towards a goal
conguration. The approach to obtain the thermal conditions is
developed by exchanging the given temperature eld in the pro-
cess presented in Section 2 with an unknown temperature eld.
Specically, in this case the out-of-plane displacement eld uz x
is cast as given (i.e., a goal displacement distribution exists) while
the temperature eld Tx becomes an unknown eld with coef-
cients to be determined. Knowledge of the temperature eld that
will yield a goal shape will be useful to determine the type and
installment location of heat sources and insulating elements in a
wide range of thermally actuated morphing structures. To this
end, an ansatz temperature eld with unknown coefcients is pro-
posed and the procedure of minimization of the total potential
analogous to that of Eq. (9) is performed. The set of equations to
be solved for this case is given as follows:
@P @P
0; i 1; . . . ; N; 0; j 1; . . . ; M; 35
@ci @cTj
where ci ; i 1; . . . ; N are the N unknown coefcients of the in-plane
displacements ux and uy (those presented in Eq. (24) are again used
here), and cTj ; j 1; . . . ; M are the M unknown coefcients of the
ansatz temperature eld. Eq. (35) provides N M equations that
are solved for such N M unknowns. The optimality conditions of
Eq. (35) refer to the in-plane displacement and temperature elds
that minimize the potential energy given a goal (imposed)
out-of-plane displacement eld. This indicates that in practice, the
obtained temperature eld would generate the inelastic (transfor-
mation) strain eld most compatible with the goal shape.
Therefore, it would provide a deformed conguration approximate
or close to the goal shape.
Exchanging a displacement eld variable with another eld
variable may lead to terms being dropped during minimization
of the total potential energy. For example, if external loads are pre-
sent and their scalar product with the known displacement eld is
non-zero, they will enter into P via W e (Eq. (8)), but their effect
will not be accounted for when evaluating the derivatives of P in
Eq. (35). This would lead to erroneous results. In the problem con-
sidered here, however, there are no tractions or body forces applied
to the plate; therefore, W e 0.
The considered goal shape is one of uniform positive Gaussian
curvature with the maximum absolute out-of-plane deection up .
Since the single layer SMA plate is xed at its center, up corre-
sponds to the deection at the edge of the plate. Given the objec-
tives of a positive Gaussian curvature and a prescribed maximum
out-of-plane deection, the out-of-plane displacement eld is
written as follows:
x2 y 2
uz up : 36
R2
Two different ansatz temperature elds are considered in this
problem. The rst (T 1 ) is limited to linear variation in the radial
direction, such as the imposed temperature eld in the problems
of Section 3, while the other (T 2 ) considers higher-order variations
of temperature through the radial direction. From Section 3 it is
known that temperature elds with increasing temperature in
the radial direction lead to deformed congurations with positive
Gaussian curvature. To assure that the SMA is transforming
Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
everywhere in the plate, the minimum applied temperature is As .
With this in mind, the considered ansatz temperature elds are
given as follows:
1=2
T 1 As cT1 x2 y2 ; 37
3=2 2
T 2 T 1 cT2 x2 y2 cT3 x2 y2 cT4 x2 y2
5=2
cT5 x2 y2 : 38
The material parameters of Table 2 are assigned to the SMA
plate to obtain simple closed-form expressions for the optimal
temperature eld and the total potential energy P. The dimensions
of the plate and up are left as symbolic entities. For the case of T 1
(linear temperature variation in the radial direction) the obtained
eld is as follows:
1=2
u2p x2 y2
T 1 337:4 1361:6 : 39
R3
Note that 337.4 K corresponds to As . The total potential energy of
the deformed plate associated with this temperature eld is the
following:
1:35107 hup  2 
2
2
PTT 1 2
up 100:2h : 40
R
For the ansatz temperature eld T 2 (higher-order variations of
temperature in the radial direction, see Eq. (38)), the terms
cT1 ; cT3 ; cT4 , and cT5 are found to be zero when minimizing the total
potential energy. Hence, the only non-zero term is cT2 , that corre-
sponds to the quadratic temperature variation in the radial direc-
tion. This temperature eld is explicitly given by:
u2p x2 y2
T 2 337:4 1136:4 : 41
R4
The total potential energy of the deformed plate associated with
this temperature eld is:
3
1:35109 h u2p
PTT 2 : 42
R2
The temperature solutions (relative to As ) vs. normalized radial
location associated with the temperature elds of Eqs. (39) and
1
(41) are plotted in Fig. 16 for a goal displacement of up h 10.
It is observed from the forms of Eqs. (39)(42) that the tempera-
ture eld and total potential energy do not depend on the direction
(sign) of up , revealing once again the bi-stable response of the plate
under the computed temperature elds.
Fig. 17 shows plots of P vs. maximum goal displacement nor-
1
malized by plate thickness (up h ) for a plate with R 100 mm
and h 1 mm. The total potential energy stored in the plate is sig-
nicantly higher for the linear temperature eld than for the quad-
ratic one, differing by more than a factor of two in their magnitude
as up h
1
increases. This suggests that lower stresses exist and
transformation strains obtained in a radially quadratic tempera-
ture eld are closer to being compatible (i.e., the difference
between actuation strain eld and total strain eld is smaller) than
those obtained from the linear eld.
Similar observations were made by Fernandes et al., 2010 in
their analysis of bi-stable composite plates. In such a work it was
discussed that the mismatch between the Gaussian curvature asso-
ciated with the in-plane inelastic strain eld and the actual
Gaussian curvature causes an increase of the stresses in the plate.
In the case considered herein, a quadratic temperature eld gener-
ates an approximate quadratic transformation strain eld, which is
known to provide a uniform Gaussian curvature that matches the
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx 15

15 5. Conclusions

10 In this work, an analytical approach to approximate the defor-

T1
mation of SMA-based plates was developed via the RayleighRitz
T2 method. The approach is summarized as follows: (i) an additive
5
set of kinematically admissible displacement elds with unknown
coefcients is initially used to approximate the actual displace-
0 ment eld in the plate; (ii) the total potential is then calculated
0 0.25 0.5 0.75 1
using the set of displacement elds, constitutive relations for the
plate constituents (dense SMA lms or meshes of periodic arrange-
Fig. 16. Difference between local temperature and As vs. normalized radial location ments of SMA wires aligned along arbitrary directions, and elastic
associated with the temperature elds T 1 and T 2 for a normalized prescribed layers), and loading conditions; (iii) the unknown coefcients are
1
maximum displacement up h of 10. found via minimization of the total potential performed in a direct
symbolic computation (i.e., without substitution of symbolic enti-
ties with numerical values or requiring iterative efforts); and (iv)
50 the stability of each obtained conguration is determined via sec-
ond derivatives of the total potential.
40 The approach provides closed-form expressions for the defor-
mation of the plates including multistable congurations. It is sim-
30
ple to implement and has low computational cost compared to
20 higher delity structural analysis methods such as FEA.
Moreover, commonly available nite element analysis codes have
10
no built-in capabilities to determine the possible stable deforma-
0 tion modes for multistable structures considering material
0 4 8 12 non-linearities. Instead, some knowledge of the potentially stable
solutions must be known beforehand, so that the proper perturba-
1
tions may be applied to prompt the solution towards specic stable
Fig. 17. Total potential energy stored in the plate P vs. up h . The plot show results
branches. This work provides an efcient solution for providing
for the linear and quadratic temperature elds, T 1 and T 2 , respectively.
this information. Furthermore, other types of inelastic strains dif-
ferent from SMA transformation strains can be included in the pre-
sent formulation as long as they are sufciently small to be
Temperature range bound Series1 appropriately modeled using the presented kinematics.
25 The deformation of circular plates composed of homogeneous
20 Series3 isotropic SMA and of laminated plates with layers of SMA wire
15 meshes has been considered. The results show that temperature
10 Closed-form elds with a linear variation in the radial direction induce multi-
quad stable deformed congurations in which the plate Gaussian curva-
5
Series4 ture is determined by the direction of the radial temperature
0
gradient. This behavior is observed for SMA plates as well as lam-
0 5 10 15
inated plates with SMA mesh layers with wires oriented in orthog-
onal and non-orthogonal directions.
1
Fig. 18. Difference between edge and center temperatures vs. up h . Analytical and An alternative and novel inversion of the proposed approach is
FEA results for the linear (T 1 ) and quadratic (T 2 ) temperature elds are shown. used to directly compute the approximate temperature eld
required to morph a plate towards a goal shape. If a constant pos-
itive Gaussian curvature goal shape is considered for a traction-free
circular SMA plate, such a conguration can be achieved by the
considered goal shape curvature (see Seffen and Maurini, 2013).
application of an appropriate temperature eld that increases radi-
Therefore, the plate develops lower stresses, as globally quantied
ally in a linear or quadratic manner. Of these two, the axisymmet-
by a lower total potential energy, under this quadratic temperature
ric temperature eld with a quadratic radial variation shows
eld than for the linear one.
superior characteristics, since the total potential (and thus the
The question of how to technologically realize the preferable
overall stresses) of the deformed plate under such a eld is smaller
quadratic temperature distribution in the plate is not the subject
than that of the linear case. This prior knowledge of the tempera-
of this work; however, with further advances in areas such as feed-
ture eld yielding a particular goal shape will for instance be useful
back heating control of SMA structures, answers to this question
to determine the type and installment location of heat sources and
may be found (Schiedeck and Mojrzisch, 2011; Gdouin et al.,
insulating elements in plate structures.
2011; Halbert et al., 2014; Bertagne et al., 2014).
The discretization and articial gravity load used for non-linear
FEA in Section 3.1 is used here to numerically validate the obtained
Acknowledgments
results. A plate of R 100 mm and h 1 mm is considered. Fig. 18
shows the difference in temperature between the edge and the
1
This work is supported by the National Science Foundation
center of the plate vs. up h for both analytical and FEA results. and the Air Force Ofce of Scientic Research under grant
Good agreement is observed between the two methods for the EFRI-1240483. Any opinions, ndings, conclusions or recommen-
considered cases. In the considered temperature range dations are those of the authors and do not necessarily reect
1
(T c As ; As 6 T e 6 Af ), the maximum attainable up h via T 1 is the views of the U.S. Air Force. The rst author would like to
approximately 13.3, while that via T 2 is approximately 14.8 (see acknowledge the support from the National Science Foundation
Fig. 18). International Institute of Materials for Energy Conversion (IIMEC)

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
16 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx

award No. 0844082, and the Institute of Mechanics at TU Kumar, S., Singh, B., 2009. Thermal buckling analysis of SMA ber-reinforced
composite plates using layerwise model. J. Aerosp. Eng. 22 (4), 342353.
Dortmund during a technical exchange in the period 06/1408/
Kuribayashi, K., Tsuchiya, K., You, Z., Tomus, D., Umemoto, M., Ito, T., Sasaki, M.,
14. Finite element analysis was performed using a research license 2006. Self-deployable origami stent grafts as a biomedical application of Ni-rich
granted by Simulia. TiNi shape memory alloy foil. Mater. Sci. Eng. A 419 (1), 131137.
Lagoudas, D., 2008. Shape Memory Alloys: Modeling and Engineering Applications.
Springer-Verlag, New York.
Appendix A. Supplementary data 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. Plast. 3233, 155183.
Supplementary data associated with this article can be found, in Lam, K., Hung, K., Chow, S., 1989. Vibration analysis of plates with cutouts by the
the online version, at http://dx.doi.org/10.1016/j.ijsolstr.2015.05. modied RayleighRitz method. Appl. Acoust. 28 (1), 4960.
007. Lang, R.J., 2007. The science of origami. Phys. World 20 (2), 3031.
Lee, H.J., Lee, J.J., 2000. A numerical analysis of the buckling and postbuckling
behavior of laminated composite shells with embedded shape memory alloy
References wire actuators. Smart Mater. Struct. 9 (6), 780.
Leo, D.J., 2008. Engineering Analysis of Smart Material Systems. John Wiley & Sons
Inc.
Al-Obeid, A., Cooper, J., 1995. A RayleighRitz approach for the estimation of the
Li, S.-R., Yu, W.-S., Batra, R., 2010. Free vibration of thermally pre/post-buckled
dynamic properties of symmetric composite plates with general boundary
circular thin plates embedded with shape memory alloy bers. J. Therm. Stress.
conditions. Compos. Sci. Technol. 53 (3), 289299.
33 (2), 7996.
Bargmann, S., Steinmann, P., 2008. An incremental variational formulation of
Liew, K., Lam, K., 1991. A RayleighRitz approach to transverse vibration of isotropic
dissipative and non-dissipative coupled thermoelasticity for solids. Heat Mass
and anisotropic trapezoidal plates using orthogonal plate functions. Int. J. Solids
Transfer 45 (1), 107116.
Struct. 27 (2), 189203.
Bassik, N., Stern, G.M., Gracias, D.H., 2009. Microassembly based on hands free
Liu, C., Yao, Y.L., Srinivasan, V., 2004. Optimal process planning for laser forming of
origami with bidirectional curvature. Appl. Phys. Lett. 95 (9), 091901.
doubly curved shapes. J. Manuf. Sci. Eng. 126 (1), 19.
Bauchau, O.A., Craig, J.I., 2009. Structural Analysis With Applications to Aerospace
Manseld, E.H., 1989. The Bending and Stretching of Plates, second ed. Cambridge
Structures. Springer.
University Press.
Berger, M., 1967. On von Karmans equations and the buckling of a thin elastic plate,
Miehe, C., Kiefer, B., Rosato, D., 2011. An incremental variational formulation of
I the clamped plate. Commun. Pure Appl. Math. 20 (4), 687719.
dissipative magnetostriction at the macroscopic continuum level. Int. J. Solids
Bertagne, C., Moghadas, P., Malak, R.J., Hartl, D.J., 2014. Feedback control applied to
Struct. 48 (13), 18461866.
nite element models of morphing structures. In: ASME 2014 Conference on
Miehe, C., Lambrecht, M., 2003. A two-scale nite element relaxation analysis of
Smart Materials, Adaptive Structures and Intelligent Systems. Society of
shear bands in non-convex inelastic solids: small- strain theory for standard
Mechanical Engineers, SMASIS2014-7601.
dissipative materials. Comput. Methods Appl. Mech. Eng. 192, 473508.
Calladine, C., 1989. Theory of Shell Structures. Cambridge University Press.
Miehe, C., Rosato, D., Kiefer, B., 2011. Variational principles in dissipative electro-
Carrera, E., Fazzolari, F.A., Demasi, L., 2011. Vibration analysis of anisotropic simply
magneto-mechanics: a framework for the macro-modeling of functional
supported plates by using variable kinematic and RayleighRitz method. J. Vibr.
materials. Int. J. Numer. Methods Eng. 86 (10), 12251276.
Acoust. 133 (6), 061017.
Modes, C., Bhattacharya, K., Warner, M., 2011. Gaussian curvature from at elastica
Cheng, Q., Song, Z., Ma, T., Smith, B.B., Tang, R., Yu, H., Jiang, H., Chan, C.K., 2013.
sheets. Proc. R. Soc. A Math. Phys. Eng. Sci. 467 (2128), 11211140.
Folding paper-based lithium-ion batteries for higher areal energy densities.
Mohd Jani, J., Leary, M., Subic, A., Gibson, M.A., 2014. A review of shape memory
Nano Lett. 13 (10), 49694974.
alloy research, applications and opportunities. Mater. Des. 56,
Criseld, M.A., 1981. A fast incremental/iterative solution procedure that handles
10781113.
snap-through. Comput. Struct. 13 (1), 5562.
Nayfeh, A.H., Hefzy, M.S., 1978. Continuum modeling of three-dimensional truss-
Dano, M.-L., Hyer, M., 1998. Thermally-induced deformation behavior of
like space structures. AIAA J. 16 (8), 779787.
unsymmetric laminates. Int. J. Solids Struct. 35 (17), 21012120.
Nayfeh, A.H., Hefzy, M.S., 1981. Continuum modeling of the mechanical and thermal
Dano, M.-L., Hyer, M., 2003. SMA-induced snap-through of unsymmetric ber-
behavior of discrete large structures. AIAA J. 19 (6), 766773.
reinforced composite laminates. Int. J. Solids Struct. 40 (22), 59495972.
Noor, A.K., 1988. Continuum modeling for repetitive lattice structures. Appl. Mech.
Diaconu, C., Weaver, P., Arrieta, A., 2009. Dynamic analysis of bi-stable composite
Rev. 41 (7), 285296.
plates. J. Sound Vibr. 322 (4), 9871004.
Ortiz, M., Repetto, E.A., 1999. Nonconvex energy minimization and dislocation
Eckstein, E., Pirrera, A., Weaver, P., 2013. Morphing high-temperature composite
structures in ductile single crystals. J. Mech. Phys. Solids 47, 397462.
plates utilizing thermal gradients. Compos. Struct. 100, 363372.
Ortiz, M., Stainier, L., 1999. The variational formulation of viscoplastic constitutive
Eckstein, E., Pirrera, A., Weaver, P., 2014. Multi-mode morphing using initially
updates. Comput. Methods Appl. Mech. Eng. 171, 419444.
curved composite plates. Compos. Struct. 109, 240245.
Otsuka, K., Wayman, C.M., 1999. Shape Memory Materials. Cambridge University
El Feninat, F., Laroche, G., Fiset, M., Mantovani, D., 2002. Shape memory materials
Press, Cambridge, UK.
for biomedical applications. Adv. Eng. Mater. 4 (3), 91.
Pandey, S., Gultepe, E., Gracias, D.H., 2013. Origami inspired self-assembly of
Fernandes, A., Maurini, C., Vidoli, S., 2010. Multiparameter actuation for shape
patterned and recongurable particles. J. Visual. Exp. 72, e50022.
control of bistable composite plates. Int. J. Solids Struct. 47 (10), 14491458.
Park, J.-S., Kim, J.-H., Moon, S.-H., 2004. Vibration of thermally post-buckled
Fernandes, R., Gracias, D.H., 2012. Self-folding polymeric containers for
composite plates embedded with shape memory alloy bers. Compos. Struct.
encapsulation and delivery of drugs. Adv. Drug Delivery Rev. 64 (14), 1579
63 (2), 179188.
1589.
Peraza-Hernandez, E., Hartl, D., Galvan, E., Malak, R., 2013a. Design and
Gdouin, P.-A., Delaleau, E., Bourgeot, J.-M., Join, C., Arbab Chirani, S., Calloch, S.,
optimization of a shape memory alloy-based self-folding sheet. J. Mech. Des.
2011. Experimental comparison of classical PID and model-free control:
135 (11), 111007.
position control of a shape memory alloy active spring. Control Eng. Pract. 19
Peraza-Hernandez, E., Hartl, D., Malak, R., 2013b. Simulation-based design of a self-
(5), 433441.
folding smart material system. In: ASME 2013 International Design Engineering
Gigliotti, M., Wisnom, M., Potter, K., 2004. Loss of bifurcation and multiple shapes of
Technical Conferences and Computers and Information in Engineering
thin [0/90] unsymmetric composite plates subject to thermal stress. Compos.
Conference. American Society of Mechanical Engineers, V06BT07A045, Paper
Sci. Technol. 64 (1), 109128.
No. DETC2013-13439.
Halbert, T., Moghadas, P., Malak, R.J., Hartl, D.J., 2014. Control of a shape memory
Peraza-Hernandez, E.A., Frei, K.R., Hartl, D.J., Lagoudas, D.C., 2014a. Folding patterns
alloy based self-folding sheet. In: ASME 2014 International Design Engineering
and shape optimization using SMA-based self-folding laminates. Proc. SPIE
Technical Conferences and Computers and Information in Engineering
9057, 90571G90571G-13.
Conference IDETC/CIE. pp. DETC2014-34703.
Peraza-Hernandez, E.A., Hartl, D.J., Kotz, A., Malak Jr., R.J., 2014b. Design and
Hartl, D., Lagoudas, D., 2009. Constitutive modeling and structural analysis
optimization of an SMA-based self-folding structural sheet with sparse
considering simultaneous phase transformation and plastic yield in shape
insulating layers. In: ASME 2014 Conference on Smart Materials, Adaptive
memory alloys. Smart Mater. Struct. 18 (10), 104017.
Structures and Intelligent Systems. American Society of Mechanical Engineers,
Hartl, D., Lagoudas, D.C., 2007. Aerospace applications of shape memory alloys. Proc.
SMASIS2014-7540.
Inst. Mech. Eng. Part G J. Aerosp. Eng. 221 (4), 535552.
Peraza-Hernandez, E.A., Hartl, D.J., Lagoudas, D.C., 2013c. Modeling of shape
Hartl, D., Zimmerman, T., Dilligan, M., Mabe, J., Calkins, F., 2010. Analysis of shape
memory alloy wire meshes using effective lamina properties for improved
memory alloy components using beam, shell, and continuum nite elements.
analysis efciency. In: ASME 2013 Conference on Smart Materials, Adaptive
In: ASME 2010 Conference on Smart Materials, Adaptive Structures and
Structures and Intelligent Systems. American Society of Mechanical Engineers,
Intelligent Systems. American Society of Mechanical Engineers, SMASIS2010-
SMASIS2013-3094.
3833.
Peraza-Hernandez, E.A., Hartl, D.J., Lagoudas, D.C., 2013. Shape memory alloy
Hartl, D.J., Lagoudas, D.C., Calkins, F., 2011. Advanced methods for the analysis,
laminate for design of self-folding recongurable structures. In: The 19th
design, and optimization of SMA-based aerostructures. Smart Mater. Struct. 20
International Conference on Composite Materials.
(9), 094006.
Peraza-Hernandez, E.A., Hartl, D.J., Malak Jr, R.J., 2013e. Design and numerical
Hughes, T.J., Winget, J., 1980. Finite rotation effects in numerical integration of rate
analysis of an SMA mesh-based self-folding sheet. Smart Mater. Struct. 22 (9),
constitutive equations arising in large-deformation analysis. Int. J. Numer.
094008.
Methods Eng. 15 (12), 18621867.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007
 E.A. Peraza Hernandez et al. / International Journal of Solids and Structures xxx (2015) xxxxxx 17

Peraza-Hernandez, E.A., Hartl, D.J., Malak Jr, R.J., Lagoudas, D.C., 2014c. Origami-
inspired active structures: a synthesis and review. Smart Mater. Struct. 23 (9),
094001.
Peraza-Hernandez, E.A., Hu, S., Kung, H., Hartl, D., Akleman, E., 2013f. Towards
building smart self-folding structures. Comput. Graph. 37 (6), 730742.
Pirrera, A., Avitabile, D., Weaver, P., 2010. Bistable plates for morphing structures: a
rened analytical approach with high-order polynomials. Int. J. Solids Struct. 47
(25), 34123425.
Pirrera, A., Avitabile, D., Weaver, P., 2012. On the thermally induced bistability of
composite cylindrical shells for morphing structures. Int. J. Solids Struct. 49 (5),
685700.
Pressley, A., 2010. Elementary Differential Geometry. Springer.
Ramm, E., 1981. Strategies for Tracing the Nonlinear Response Near Limit Points.
Springer.
Reddy, J.N., 2002. Energy Principles and Variational Methods in Applied Mechanics,
second ed. John Wiley & Sons, Hoboken, NJ.
Ryu, C., Shin, J.G., 2006. Optimal approximated unfolding of general curved shell
plates based on deformation theory. J. Manuf. Sci. Eng. 128 (1), 261269.
Schiedeck, F., Mojrzisch, S., 2011. Design of a robust control strategy for the heating
power of shape memory alloy actuators at full contraction based on electric
resistance feedback. Smart Mater. Struct. 20 (4), 045002.
Schlecht, M., Schulte, K., Hyer, M., 1995. Advanced calculation of the room-
temperature shapes of thin unsymmetric composite laminates. Compos. Struct.
32 (1), 627633.
Seffen, K., 2007. Morphing bistable orthotropic elliptical shallow shells. Proc. R.
Soc. A Math. Phys. Eng. Sci. 463 (2077), 6783.
Seffen, K., Maurini, C., 2013. Growth and shape control of disks by bending and
extension. J. Mech. Phys. Solids 61 (1), 190204.
Slaughter, W.S., 2002. The Linearized Theory of Elasticity. Birkhuser, Boston.
Srinivasan, A.V., McFarland, D.M., 2001. Smart Structures: Analysis and Design.
Cambridge University Press.
Tolley, M.T., Felton, S.M., Miyashita, S., Aukes, D., Rus, D., Wood, R.J., 2014. Self-
folding origami: shape memory composites activated by uniform heating.
Smart Mater. Struct. 23 (9), 094006.
Vidoli, S., 2013. Discrete approximations of the Fppl-Von Krmn shell model:
from coarse to more rened models. Int. J. Solids Struct. 50 (9),
12411252.
Yu, G., Patrikalakis, N.M., Maekawa, T., 2000. Optimal development of doubly
curved surfaces. Comput. Aided Geomet. Des. 17 (6), 545577.

Please cite this article in press as: Peraza Hernandez, E.A., et al. Analytical investigation of structurally stable congurations in shape memory alloy-ac-
tuated plates. Int. J. Solids Struct. (2015), http://dx.doi.org/10.1016/j.ijsolstr.2015.05.007