Composite Structures 176 (2017) 640–653

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Analysis of functionally graded piezoelectric Timoshenko smart beams

using a multiquadric radial basis function method
T.R.C. Chuaqui, C.M.C. Roque ⇑
INEGI, Faculdade de Engenharia da Universidade do Porto, Universidade do Porto, Rua Dr. Roberto Frias 404, 4200-465 Porto, Portugal

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

Article history: In the present work, a global interpolation scheme is proposed to describe the electro-mechanical static
Received 8 May 2017 response of piezoelectric smart beams considering applied mechanical loads and electric potential. First,
Accepted 11 May 2017 bimorph and three-layered Timoshenko smart beams are investigated and the model is validated for both
Available online 1 June 2017
actuator and sensor capabilities. Lastly, a functionally graded piezoelectric (FGP) bimorph beam is stud-
ied in its actuator configuration. The governing equations of motion are derived by the application of
Keywords: Hamilton’s principle considering an equivalent-single-layer first-order-shear-deformation theory (ESL-
Piezoeletric material
FSDT) with layerwise electric potential. The equations are subsequently solved using a grid free method
Functionally graded material
Timoshenko smart beam
based on Kansa’s asymmetric collocation using multiquadric radial basis functions (RBFs). The method
Radial basis function relies on the Euclidean distance between nodes and on a shape parameter which must be warily selected
to avoid matrix ill-conditioning. Accurate results are obtained while maintaining a very low computa-
tional cost.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Euler–Bernoulli with layerwise linear electric potential for the core
and piezoelectric material respectively have been formulated using
Piezoelectric smart structures have been widely studied in the a classical sandwich beam theory (SBT) [7]. Closed form solutions
past years for their applications in static shape and vibration con- have also been implemented [8]. Finite element models with
trol, energy harvesting, damage detection and sensing technology. assumed linear through-the-thickness distribution of electric
The fields of applications range from the aerospace industry to potential based on SBT [9], equivalent-single-layer (ESL) [10] and
micro and nano-electromechanical sensors and actuators. The layerwise theories [10–12] have been developed. Finite element
piezoelectric material is generally either embedded inside the models for static and dynamic analysis of extension mode smart
structure or bonded to the surfaces in the form of layers or patches. beams have been proposed [13–15] and several higher-order-
The piezoelectric beam is one of the primary smart structures shear-deformation theory based finite element approaches have
being a focus of attention to the scientific community for its typi- also been formulated [3,16–19]. The assumption of through-the-
cally high reliability and simplicity [1–6]. thickness linearity of the electric potential is a mediocre approxi-
The sensing capabilities of a piezoelectric material are associ- mation since induced potential effects lead to a non-linear distri-
ated with the direct effect, where the application of a mechanical bution [1,20–22]. For a better estimation, either sublayers are
load generates an electric field. On the other hand, electrically included to model the physical piezoelectric layers [3,21] and/or
induced deformations in a piezoelectric material by virtue of the higher order distributions of the potential must be considered
converse effect are employed in actuators. The coupling between [5,6,20,22].
mechanical and electrical fields dictates the input and output of One of the main disadvantages of traditional piezoelectric smart
any given piezoelectric system [4]. structures is that these are often multilayered composites with
Several models have been developed to predict the behaviour of adhesive resins being typically applied to bond the layers. High
smart beams, with the first analytical models for extension actua- stress concentrations and discontinuities in the interfaces due to
tors being based on uniform-strain Euler–Bernoulli theory [2]. mechanical and electrical loadings are a common problem leading
Combinations of first-order-shear-deformation theory (FSDT) and to initiation and propagation of micro-cracks which limit the dura-
bility and reliability of these devices. Functionally-graded-piezo
electric-materials (FGPM) have been developed in order to over-
⇑ Corresponding author.
come this issues by presenting a smooth and continuous variation
E-mail address: [email protected] (C.M.C. Roque).

http://dx.doi.org/10.1016/j.compstruct.2017.05.062
0263-8223/Ó 2017 Elsevier Ltd. All rights reserved.
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 641

of electrical and mechanical properties within the thickness and/or severe ill-conditioning [57,58]. Considering a set of nodes
length of the structure [23–25], permitting a significant reduction x1 ; x2 ; . . . ; xN 2 X  Rn , the multiquadric RBFs centred at xj can be
in internal stress concentrations. The exact solution of a FGP layer defined as:
under uniform stretch, bending and twisting load was studied [26]
1=2
and three-dimensional solutions for smart FGP plates have been g j ðxÞ ¼ kx  xj k þ c2 ð1Þ
investigated [27,28]. Exact solutions for simply supported FGP
where kx  xj k corresponds to the Euclidean norm. Other common
laminates in cylindrical bending were formulated using Stroh-
RBFs, which will not be used in this work, are [44,43,57,58]:
like formalism [29] and exact solutions for FGP plates [30] were

1=2
investigated. Closed form solutions for cantilever FGP beams using Inverse multiquadrics: g j ðxÞ ¼ kx  xj k þ c2
two-dimensional elasticity and Airy stress function were obtained 2 kxx k1=2
Gaussians: g j ðxÞ ¼ ec j
[31,32]. FGP bimorph actuators were investigated using layerwise
finite elements [33]. The static and dynamic responses of FGP Thin plate splines: g j ðxÞ ¼ kx  xj k2 log kx  xj k.
Timoshenko actuators [34] and of FGP plates using FSDT [35] were
Considering a boundary-valued problem with a domain X  Rn
studied and third-order-shear-deformation theory (TSDT) was
restricted by the boundary @ X and a linear elliptic partial differen-
employed to investigate the free and forced vibration of FGP beams
tial operator L acting on the bounded region and LB acting on the
[36]. Lastly, a refined sinus finite element model was developed in
boundary:
order to study the static response and the free vibration of FGP
beams [25]. LuðxÞ ¼ sðxÞ  Rn ð2Þ
A considerable number of meshless methods have been imple-
mented to study the behaviour of traditional composite smart LBj@X uðxÞ ¼ f ðxÞ  Rn ð3Þ
beams. The governing equations of piezoelectric devices have been The boundary of the problem and the domain interior are dis-
solved using a mesh free point collocation method [37]. A point cretized by two sets of points or nodes (xj ; j ¼ 1; . . . ; N B ) and
interpolation method for static and frequency analysis of two- (xj ; j ¼ N B þ 1; . . . ; N) respectively, where N B is the number of
dimensional piezoelectric structures was implemented in [38].
boundary nodes and N is the total number of nodes. The radial
Moreover, approximate methods based on two-dimensional
basis function interpolation of the solution uðxÞ is given by:
element-free Galerkin (2D-EFG) for the numerical simulation of
laminated Timoshenko beams have been used [39] and layerwise X
N


mesh-free models have been employed for the static analysis of
sðx; cÞ ¼ aj g kx  xj k; c ð4Þ
j¼1
smart laminated composite beams [40]. However, very few mesh
free radial basis function models have been developed to study For a finite-dimensional static problem, considering the interior
these structures. The method was first used by Hardy for the inter- nodes governed by partial differential equations and the boundary
polation of geographical scattered data [41,42] and later applied in nodes with enforced conditions, one obtains respectively:
the numerical solution of partial differential equations by Kansa
X
N


[43,44]. RBFs have been used for two-dimensional solids [45,46], sB ðx; cÞ ¼ aj LB g kx  xj k; c ¼ fðxi Þ; i ¼ 1; . . . ; NB ð5Þ
for composite laminated beams and plates [47–50] and for the sta- j¼1
tic and vibration analysis of functionally graded structures [51–
54]. Radial basis function based on partition of unity method X
N


(RBF-PUM) was proposed to analyse the static response of piezo- sI ðx; cÞ ¼ aj Lg kx  xj k; c ¼ wðxi Þ; i ¼ N B þ 1; . . . ; N ð6Þ
j¼1
electric structures [55] and a two-dimensional piezoelectric prob-
lem has been solved by converting the original set of differential where fðxi Þ and wðxi Þ denote the prescribed values at the boundary
equations to Poisson-type equations being subsequently solved nodes and the function values at the interior nodes respectively.
with radial basis functions [56]. To the authors’ knowledge, multi- Eqs. (5) and (6) can be rewritten in matrix form as:
quadric RBFs have not been employed to study the static behaviour    
LB g f
of Timoshenko smart beams considering a quadratic through-the- a¼ ð7Þ
thickness distribution of electric potential. In this work, Lg w
equivalent-single-layer first-order-shear-deformation theory
The system (7) is solved using affine space decomposition,
(ESL-FSDT) for the mechanical displacements and layerwise elec-
which decouples the influence between interior and boundary col-
tric potential are considered. Bimorph and three-layered piezoelec-
locations, improving conditioning [59]. After the determination of
tric smart beams are first investigated in both sensor and actuator
the vector of interpolation constants a, the approximate solution
configurations considering different mechanical and electrical  
s  s_B sI can be determined according to (4).
loadings. The study of these structures permitted the validation
of the model. Lastly, a FGP bimorph beam is subsequently studied
2.2. First-order shear deformation theory with layerwise electric
considering different applied electric potentials.
potential

2. Mathematical formulation
The mathematical formulation developed in this work is based
on the equivalent single layer first-order shear deformation theory
2.1. The finite point multiquadric method
(ESL-FSDT) for the mechanical degrees of freedom with the inclu-
sion of a layerwise electric potential. The displacement field equa-
In this paper it is proposed to use Kansa’s asymmetric colloca-
tions in the longitudinal and transverse directions of the
tion method[44,43] to solve the governing equations of the
Timoshenko smart beams demonstrated in Fig. 1 and Fig. 2 can
Timoshenko smart beam. The multiquadric method does not
be written as [60]:
require a mesh, relying only on the Euclidean distance between
nodes and on a user-defined shape parameter c [44,43,57,58].
The accuracy of the solution is largely determined by the selection uðx; z; tÞ ¼u0 ðx; tÞ þ zhðx; tÞ ð8aÞ
of the shape parameter, which if not chosen warily can lead to wðx; z; tÞ ¼w ðx; tÞ 0
ð8bÞ
642 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653

p, the elastic, piezoelectric and dielectric coefficients can be calcu-

lated as follows [5]:
 2
e k ¼ Qk  Qk
Q =Q k22 ð12aÞ
11 11 12

e k ¼ Qk
Q ð12bÞ
55 55

Q kij C kij
¼  C ki3 C kj3 =C k33 ; i; j ¼ 1; 2; 5; ð12cÞ


~ep31 ¼ ep31  ep32 Q p12 =Q p22 ð12dÞ


ep3i ¼ ep3i  ep33 C pi3 =C p33 ; i ¼ 1; 2; ð12eÞ


Fig. 1. Three-layered cantilever smart beam. g~ pzz ¼ g pzz þ ep32 2 =Q p22 ð12fÞ


g pzz ¼ gpzz þ ep33 2 =C p33 ð12gÞ

The electric field can be determined by applying the gradient of

the electric potential U that penetrates the piezoelectric layers of
the beam [1]:
E ¼  rUðx; z; tÞ ð13aÞ
@ Uðx; z; tÞ
Ez ¼  ð13bÞ
@z
Despite linear through-the-thickness distributions of the elec-
tric potential of the beam being recurrently seen in the published
Fig. 2. FGP cantilever bimorph smart beam. literature as a valid simplification, this consideration violates the
first Maxwell equation and often leads to mediocre results [1,20–
22]. One way of avoiding erroneous results while still using a linear
where u0 and w0 denote the axial and transverse displacements of distribution of the potential is to employ a layerwise approach to
the mid-plane respectively, while h corresponds to the rotation the piezoelectric layers [3,21]. For a given physical piezoelectric
angle along the x axis. layer, two or more sub-layers are considered where the potential
Substituting the equations expressed in (8) into the usual within each of them remains linear in nature. Alternatively, one
strain–displacement relations of a one-dimensional plane stress may consider a quadratic or higher-order potential distribution
beam, the following expressions are obtained: in the physical piezoelectric layer without the consideration of
any sub-layers. In this work, a quadratic potential is employed
@uðx; z; tÞ @u0 @h according to [5] for each pth piezoelectric layer.
exx ðx; z; tÞ ¼ ¼ þz ð9aÞ
@x @x @x According to the first Maxwell equation, the divergence of the
@uðx; z; tÞ @wðx; z; tÞ @w0 electric flux is null in the absence of free volumetric charge density.
cxz ðx; z; tÞ ¼ þ ¼hþ ð9bÞ
@z @x @x Hence, the electrostatic equilibrium of a piezoelectric layer can be
The generic constitutive electro-mechanical relations for any written in the differential form as:
given piezoelectric homogeneous material can be written as [60– @Dpz
62]: r:Dp ¼ ¼0 ð14Þ
@z
rij ¼C ijkl ekl  ekij Ek ð10aÞ Substituting the strain–displacement relations (9), the electro-
Di ¼eikl ekl þ gik Ek ð10bÞ mechanical coupling constitutive relations (11) and Eq. (13) into
the equation above:
where rij ; ekl ; Di ; Ek are the components of stress, strain, electric dis-
placement and electric field while C ijkl ; ekij ; gik correspond to the @ 2 /p ~ep31 @h
elastic, piezoelectric and dielectric material constants respectively. ¼ p ð15Þ
@z2 g~ zz @x
Since the problem is one-dimensional and invariably plane-stress
condition exists, ryy ¼ rzz ¼ sxy ¼ syz ¼ cxy ¼ cyz ¼ 0. For the partic- By solving the differential equation with respect to the electric
ular case of a piezoelectric medium operating in extension mode, potential and applying boundary conditions, one arrives at [5]:
the poling direction is parallel to transverse displacement. There-
 p ðx; tÞ þ z  zp ~ p
fore, only axial stress and transverse electric field are accountable Up ðx; z; tÞ ¼ / p / ðx; tÞ
for electro-mechanical coupling, allowing one to assume h

2 !
Ex ¼ Ey ¼ 0. The constitutive relations for the kth layer of the ~ep31 hp 4ðz  zp Þ2 @h
Timoshenko piezoelectric beam can finally be reduced to the fol-  1 
p 2 ð16Þ
8g~ pzz h @x
lowing form:
8 9 2 38 9


e 11 p ¼ / þ / ~p p
< rxx >
> = Q 0 ~e31 > < exx >
=
where / p pþ1 =2; / ¼ /pþ1  /p ; z ¼ zpþ1 þ zp =2 and
6 e 55 7 /pþ1 ; /p , correspond to the surface and bottom electric potential
sxz ¼ 4 0 KQ 0 5 cxz ð11Þ
>
: >
; >
: >
; p
Dz k ~e31 0 g~ zz k Ez k for any given pth piezoelectric layer. h designates the thickness
of the piezoelectric layer.
where K corresponds to the shear correction factor typically used in Considering Eqs. (16 and (13), the electric field of the pth piezo-
FSDT theories (K ¼ 5=6). Evidently, for a non-piezoelectric layer, the electric layer can be written as:
constitutive relations reduce to typical Hooke’s elasticity law and all
~ p ~ep
/ @h
piezoelectric, dielectric, electric field and electric displacement Epz ¼  p
p  p ðz  z Þ
31
ð17Þ
terms vanish. Denoting the piezoelectric layers by the superscript h g~ zz @x
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 643

The governing equations of motion can be determined by the area on which the charge is applied is identical to the surface area of
application of a variational principle. In this work, Hamilton’s prin- the beam, thus S/ ¼ S.
ciple was employed so that the natural boundary conditions are Finally, introducing the variation in potential energy estab-
implicitly established. Variation in kinetic energy was not consid- lished in (21) and the virtual work (23) into the Hamilton’s princi-
ered since only static analysis was conducted in this paper: ple (18), integrating by parts the potential energy in the x
Z t2 Z t2 coordinate relieving the variation in first-order derivatives of the
 dPP dt þ dPF dt ¼ 0 ð18Þ mechanical displacements, one obtains the following governing
t1 t1 equations of motion and natural boundary conditions for the smart
where PP and PF are the potential energy and the work done by beam:
non-conservative forces respectively.
Considering a non-piezoelectric layer, the variation in potential  Governing equations of motion
energy is equivalent to the variation in mechanical strain energy: @ 2 u0 @2h @/~p s
Z B11 þ ðC 11 þ EÞ 2 þ F p þ fu ¼ 0 ð24Þ

 @x 2 @x @x
dPP ¼ dekxx rkxx þ dckxz skxz dV; k – p ð19Þ
V !
@ 2 w0 @h s
For the piezoelectric layers, the variation in electromechanical KB55 þ þ fw ¼ 0 ð25Þ
@x2 @x
potential energy is given by [3]:
Z 
@ 2 u0 @2h @w0 @/~p
dPP ¼ depxx rpxx þ dcpxz spxz  dEpz DPz dV ð20Þ ðC 11 þ EÞ þ ðD þ GÞ  KB55 þ h þ Hp ¼0
11
V @x2 @x2 @x @x
Considering both (19) and (20), substituting the stress and elec- ð26Þ
tric displacement components by the constitutive relations (11)
and replacing the strain and electric field terms by (9) and (13), @u0 @h ~p  q ¼ 0
(16) respectively, one obtains the variation in potential energy of Fp þ Hp  Ip / 0 ð27Þ
@x @x
a generic layer as a function of mechanical displacements and elec-
 Boundary conditions
tric potential:
Z Z  @u0 @h ~ p ¼ 0 _ du0 ¼ 0; x ¼ 0; L
zkþ1
@u0 @h e k @u þ z @h
0
B11 þ ðC 11 þ EÞ þ F p / ð28Þ
dPP ¼ d þ zd Q @x @x
S zk @x @x 11
@x @x
!#
~
/ p ~e p
@h @w 0 @w0
þ ~ep31 p þ 31 ðz  zp Þ þ d þ dh KB55 þ h ¼ 0 _ dw0 ¼ 0; x ¼ 0; L ð29Þ
h g~ pzz @x @x @x
  !
@w0 d/~p @h ~ep31
 K Q 55e þh þ  @u0 @h
p þd ðz  z Þ ~ p ¼ 0 _ dh ¼ 0; x ¼ 0; L
k p
@x h @x g ~ pzz ðC 11 þ EÞ þ ðD11 þ GÞ þ Hp / ð30Þ
@x @x
" !#)
@u0 @h ~ p ~ep
/ @h where the material constants are defined as follows:
 ~e31 p
þz þg ~ zz  p  p ðz  z Þ
p 31 p
dzdS
@x @x h g~ zz @x X
nl
Z L ( "
p 2 ! Bij ¼ e k Ik ; i; j ¼ 1; 5
Q ð31aÞ
@u0 e k k @u0 ~
p  @h ij 0
¼ d Q 11 I0 þ Q e k Ik þ e31 I1  Ip0zp k¼1
0 @x @x 11 1
~
gzzp
@x X
nl
# "
p 2 ! C ij ¼ e k Ik ; i; j ¼ 1; 5
Q ð31bÞ
p ~p p ~
~e / I @h
p  @u 0 ij 1
þ 31 p 0 þ d e k Ik þ e31
Q I  Ip zp k¼1
h @x 11 1
g~ pzz 1 0 @x X
nl

p 2  ! # Dij ¼ e k Ik ; i; j ¼ 1; 5
Q ð31cÞ
~ @h ~e31 / p ~ I z
p p p ij 2
þ Q e k Ik þ e31 Ip2  Ip0 ðzp Þ
2
þ 0 k¼1
11 2
g~ pzz @x h
p
p 2
X
npl
~e31
p 
" # E¼ I1  Ip0 zp ð31dÞ
~
~p p
p e31 I 0 @u
0
 p @h g~ pzz /~ p Ip0 ~
g p
þd/ p þz 
p 2 p¼1 zz
h @x @x h ~ep31 Ip0
  Fp ¼ ð31eÞ
@w0 e k Ik @w 0
h
p
þ d þ dh K Q þh dx ð21Þ
p 2 
@x 55 0
@x X
npl
~e31
Ip2  Ip0 ðzp Þ
2
G¼ ð31fÞ
where: p¼1
~
g p
zz

iþ1  ~ep31 Ip0zp
Iki ¼ zkþ1  ziþ1
k =ði þ 1Þ; i ¼ 0; 1; 2 ð22Þ Hp ¼ ð31gÞ
p
h
The work done by non-conservative forces for any electrome- ~ p p
gzz I
chanical system can be expressed as the sum of the mechanical Ip ¼
p 02 ð31hÞ
and the electrical work. It should be noted that neither volume h
nor point loads will be considered and therefore only the surface The values nl and npl refer to the total number of layers and to the
mechanical forces and electrical loads contribute to the virtual number of piezoelectric layers of the smart beam, respectively.
work:
Z Z

s s  2.3. Homogenization of material properties: FGP bimorph beam
dPF ¼ du0 f u þ dw0 f w dS  d/q0 dS/ ð23Þ
S S/
For a functionally graded piezoelectric bimorph beam, the
s s
where f u; f w
and q0 denote the axial force, the transverse force and homogenized material properties must be determined considering
the surface electrical charge respectively. For the present work, the the distribution of the volume fraction of each constituent.
644 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653

Through-the-thickness variation in the material composition is X

N
@2 gj XN
@g
generally expressed as a power law [63]. Considering V 1 and V 2 KB55 awj þ KB55 ahj j þ f sw ¼ 0 ð38Þ
j¼1
@x2 j¼1
@x
the volume fraction of the substrate material and the piezoelectric
constituent respectively, according to the rule of mixtures, the
XN
@2 gj XN
@ 2 gj XN
@g j
effective property P at an arbitrary point is given by: ðC 11 þ EÞ auj þ ðD11 þ GÞ ahj  KB55 aw
@x 2 @x 2 j
@x
P ¼ P1 V 1 þ P2 V 2 ð32Þ j¼1 j¼1 j¼1

X
N X
N
~p @g j
where P 1 and P2 are values of P for the two constituents. The vol-  KB55 ahj g j þ Hp aj/ ¼0 ð39Þ
ume fraction of the piezoelectric phase is given by: j¼1 j¼1
@x
!t
ð1Þk 2z X
N
@g j XN
@g j XN
V2 ¼ ð33Þ Fp auj þ Hp ahj
~p
 Ip aj/ g j  q0 ¼ 0 ð40Þ
h @x @x
j¼1 j¼1 j¼1

with k ¼ 1; 2 for the bimorph case. Therefore, V 2 ¼ 1 at the bottom

The vector of unknowns has the dimension Nð3 þ nplÞ and is
and upper surface and V 2 ¼ 0 in the middle-plane, corresponding to ~p
a non-piezoelectric core material. The rule of mixtures was now composed of the interpolation parameters auj ; aw
j ; aj and aj .
h /

employed to determine the effective piezoelectric and dielectric For each boundary node i, the boundary conditions are enforced
coefficients and Mori–Tanaka homogenization method was used according to Eqs. (28)–(30):
to find the effective elasticity constants. Thus, the effective bulk X
N XN XN XN
@g i @g ~p
modulus K e and the effective shear modulus Ge are given by [64]: B11 auj þ ðC 11 þ EÞ ahj i þ F p a/j g i ¼ 0 _ auj g i ¼ 0; x ¼ 0; L
j¼1
@x j¼1
@x j¼1 j¼1
Ke  K1 V2
¼ ð34aÞ ð41Þ
K 2  K 1 1 þ ð1  V 2 Þ K 2 K 1
K þ 4G 1 3 1

Ge  G1 V2 X
N
@g i XN XN
¼ ð34bÞ KB55 awj þ KB55 ahj g i ¼ 0 _ awj g i ¼ 0; x ¼ 0; L ð42Þ
G2  G1 1 þ ð1  V 2 Þ GG2 G 1
@x
1 þf 1 j¼1 j¼1 j¼1

where f 1 ¼ G61 ððK9K1 þ2G

1 þ8G1 Þ
. The effective values of the elasticity constants
1Þ XN
@g i XN
@g
can be calculated as follows: ðC 11 þ EÞ auj þ ðD11 þ GÞ ahj i
j¼1
@x j¼1
@x
e k ¼ 9K e Ge ; m ¼ 3K e  2Ge ; Q
Q e k ¼ Ge ð35Þ X
N X
N
11
3K e þ Ge 2ð3K e þ Ge Þ 55
þ Hp
~p
aj/ g i ¼ 0 _ ahj g i ¼ 0; x ¼ 0; L ð43Þ
j¼1 j¼1
For the FGP beam, the material constants previously defined can
be calculated: To simplify the notation, the system of Eqs. (37)–(40) can be
nl Z zkþ1
X written in matrix form as follows:
Bij ¼ e k dz; i; j ¼ 1; 5
Q ð36aÞ  ( )
zk
ij
K dd K dUp ad f
k¼1
¼ ð44Þ
nl Z aU
p
X zkþ1 K Up d K Up Up q
C ij ¼ e k zdz; i; j ¼ 1; 5
Q ð36bÞ
ij
k¼1 zk where K ij are the stiffness terms (which comprise interpolation
nl Z
X zkþ1 functions and material and geometric properties), the subscripts
Dij ¼ e k z2 dz; i; j ¼ 1; 5
Q ð36cÞ d; U denote the displacements and the electric potential and f ; q
ij
k¼1 zk are the mechanical and surface charge loads respectively.
npl Z

p 2
X zpþ1 ~e31 For the particular case of a smart beam operating as an actuator
E¼ ðz  zp Þdz ð36dÞ (closed-circuit), the electric potential is prescribed on the surfaces
p¼1 zp g~ pzz
of the piezoelectric layers, while the interfaces with the substrate
Z
~ep31
zpþ1
material are grounded. The system of Eqs. (44) can be reduced
Fp ¼ dz ð36eÞ since the potentials are known values:
zp h=2
npl Z zpþ1

p 2 
K dd add ¼ f  K dUp aU
p
X ~e31 ð45Þ
z2  ðzp Þ dz
2
G¼ ð36fÞ
p¼1 zp
g~ pzz For the case of the beam operating as a sensor (open-circuit),
Z zpþ1 p p the surface electrical charge is null and potential is unknown:
~e31z
Hp ¼ dz ð36gÞ
h=2  ( )
zp
Z zpþ1
K dd K dUp ad f
¼ ð46Þ
g~ pzz K Up d K Up Up aU
p
0
Ip ¼ 2
dz ð36hÞ
zp ðh=2Þ

3. Numerical examples
2.4. Multiquadrics for interpolation of governing differential equations
The static analysis is carried out for the case of a beam solely
It is now possible to apply the multiquadric radial basis func- operating in extension mode, as previously stated in the mathe-
tion method to the already established equations of motion and matical formulation of the problem. First, a bimorph cantilever
respective boundary conditions. Considering (24), (25), (26), (27), beam composed of two identical piezoelectric layers is considered.
the governing differential equations become: Subsequently, the numerical results correspondent to a three-
X
N
@2 gj XN
@2gj XN
~ p @g j s
layered cantilever composite beam with two symmetrical piezo-
B11 auj þ ðC 11 þ EÞ ahj þ F p aj/ þ fu ¼ 0 ð37Þ electric layers and an aluminium interior substrate (depicted in
@x 2 @x 2 @x
j¼1 j¼1 j¼1 Fig. 1) are demonstrated. It should be noted that inter-laminar
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 645

layers of adhesive material are not considered in any of the cases. describe the non-linear behaviour that occurs within the thickness
Both bimorph and three-layered smart beams are posteriorly of the beam.
tested in their actuator and sensor functionalities individually.
Lastly, a FGP bimorph beam operating in its actuator configura- 3.1.2. Sensor configuration
tion composed of two symmetrical layers of a two phase material In the sensor configuration, no electric potential is applied and
is studied. The FGP bimorph beam can be visualized in Fig. 2. The instead, a tip load of 1000 N deflects the beam generating a
graded material is fully piezoelectric on one side, corresponding potential which is subsequently measured.
to pure PZT-5H, while fully non-piezoelectric on the other side, In Table 3 the tip deflection of the bimorph beam is analysed for
corresponding to pure aluminium. both linear and quadratic distributions of the electric potential. The
The properties of the PZT-5H piezoelectric material and of the absolute value of the relative error between the results obtained
aluminium are demonstrated in Table 1 according to [20]. All and Ansys 2D solution [5] is also presented.
numerical results are obtained considering a regular distribution The results obtained with N ¼ 30 considering a through-the-
of the nodes throughout the full length of the beam. The total num- thickness quadratic distribution of the electric potential are in very
ber of nodes used in the spatial discretization is N ¼ 30 with the good agreement with the reference used for comparison. For
pffiffiffiffi
shape parameter c ¼ 3= N unless otherwise stated. higher values of N, the solution remains approximately the same
for both linear and quadratic distributions.
3.1. Bimorph beam In Fig. 4, the transverse displacement of the same bimorph
beam is plotted and compared with the results from Ansys 2D
3.1.1. Actuator configuration [5]. The displacements computed with the quadratic distribution
For the actuation of the bimorph beam, an electric potential of are once again in very good agreement with the Ansys 2D
10 V is applied on top and bottom surfaces while the interface simulations.
between layers is grounded. The applied potential is considered In Table 4, the electric potential developed at mid-span
constant along the entire length of the beam. The numerical results (x ¼ L=2) of the bimorph beam considering a 1000 N tip load is
are formulated with the quadratically distributed through-the- analysed for the different through-the-thickness distributions of
thickness electric potential demonstrated in the mathematical for- the potential and compared with Ansys 2D [5]. The absolute value
mulation and with a linearly distributed potential for comparison. of the relative error is displayed.
First, the convergence of the numerical method is evaluated and The electric potential developed in each piezoelectric layer
the tip deflection of the beam when subjected to the aforemen- along the length of the bimorph beam when subjected to a
tioned potential can be visualized in Table 2 for a varying number 1000 N tip load can be observed in Fig. 5. The results obtained
of nodes N. The absolute value of the relative error between the with linear and quadratic distributions of the potential are com-
results obtained with both linear and quadratic distributions of pared with the solutions from [5]. The distribution of the potential
the electric potential and Ansys 2D solution [5] is presented. As along the length is linear with a maximum value developed at the
expected, through-the-thickness quadratic distribution of the elec- root of the beam. The results obtained with the quadratic through-
tric potential produces more accurate results than the linear distri- the-thickness electric potential coincide with the expected solu-
bution. It should be noted that for a number of nodes greater than tion and the linear through-the-thickness distribution leads to a
N ¼ 30, results do not improve and the solution remains approxi- slight over-prediction of the developed potential.
mately the same. The through-the-thickness distribution of the electric potential
In Fig. 3, the transverse displacement of the cantilever bimorph of the beam at its mid-span is plotted in Fig. 6 considering the
beam under the same electric actuation considering both linear same applied tip load. The results are calculated using both linear
and quadratic distributions of the potential is depicted and, once and quadratic distributions, with the latter reaching the expected
again, compared with the results obtained in Ansys 2D [5]. solution proposed in [5].
The results obtained with the quadratic distribution of the elec- In Fig. 7, the though-the-thickness distribution of axial stress by
tric potential are in very good agreement with the Ansys 2D solu- the application of the same tip load is measured at the root of the
tion. As expected, despite the prescribed potential on the surfaces beam (x ¼ 0), where it presents its maximum value. In the case of a
being the same for both types of distribution, the linear potential linear distribution of the electric potential, a discontinuity in axial
formulation presents slightly divergent results for its inability to stress is visible at the interface between piezoelectric layers
whereas for the quadratic distribution, the results are in very good
agreement with the Ansys 2D solution from [6].

Table 1 3.2. Three-layered smart beam

Material properties.

Properties PZT-5H Aluminium 3.2.1. Actuator configuration

e 11 [MPa] – 70.3 For the actuation of the three-layered beam, an electric poten-
Q
m – 0.345 tial of 10 V is applied on top and bottom surfaces of the top
q [kg m3 ] 7500 2710 and bottom piezoelectric layers respectively and the interfaces
C 11 [MPa] 126 – with the substrate aluminium layer are grounded. Similar to the
C 12 [MPa] 79.5 – bimorph actuator, the applied potential is considered constant
C 13 [MPa] 84.1 –
along the length of the beam. Both linear and quadratic through-
C 22 [MPa] 126 –
C 33 [MPa] 117 –
the-thickness distributions of the electric potential are considered
C 44 [MPa] 23 – in the numerical results. In Fig. 8 the tip deflection of the actuator
p
C 55 [MPa] 23 – can be observed for varying thickness ratio r ¼ 2h =h, with r ¼ 0
C 66 [MPa] 23.25 –
corresponding to a aluminium beam with no piezoelectric layers
e31 [C m2 ] 6:5 –
and r ¼ 1 corresponding to a bimorph beam.
e32 [C m2 ] 6:5 –
It can be observed from Fig. 8 that for a thinner piezoelectric
e33 [C m2 ] 23.3 –
layer, the tip deflection increases since the electric field across each
gzz [F m1 ] 1:3  108
layer is greater for a constant applied potential, resulting in an
646 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653

Table 2 pffiffiffiffi
Tip deflection [lm] of the bimorph beam with different potential distributions for 10 V actuation, c ¼ 3= N .

N L. potential [V] jR. error [%]j Q. potential [V] jR. error [%]j Ansys 2D [5]
– – – – – 0:790
5 0:168 78.73 0:132 83.27 –
10 0:825 4.46 0:791 0.11 –
15 0:810 2.58 0:776 1.80 –
20 0:808 2.32 0:780 1.28 –
30 0:825 4.48 0:790 0.01 –

Fig. 3. Transverse displacement with different potential distributions along the length of the bimorph beam for 10 V actuation.

Table 3 pffiffiffiffi
Tip deflection [lm] of the bimorph beam with different potential distributions for a 1000 N tip load, c ¼ 3= N .

N L. potential [V] jR. error [%]j Q. potential [V] jR. error [%]j Ansys 2D [5]
– – – – – 57:2
5 48:35 15.47 46:68 18.39 –
10 59:04 3.22 57:10 0.17 –
15 59:36 3.78 57:10 0.17 –
20 59:42 3.88 57:29 0.16 –
30 59:56 4.13 57:24 0.07 –

Fig. 4. Transverse displacement with different potential distributions along the length of the bimorph beam by the application of a 1000 N tip load.
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 647

Table 4
Electric potential developed at mid-span of the bimorph beam with different potential distributions by the application of a 1000 N tip load.

N L. potential [V] jR. error [%]j Q. potential [V] jR. error [%]j Ansys 2D [5]
– – – – – 67:76
5 57:04 15.82 55:05 18.76 –
10 78:02 15.14 75:29 11.11 –
15 70:39 3.88 72:98 7.70 –
20 74:05 9.28 71:33 5.27 –
30 72:79 7.42 67:97 0.30 –

Fig. 5. Electric potential developed across each piezoelectric layer along the length of the bimorph beam by the application of a 1000 N tip load with different potential
distributions.

Fig. 6. Through-the-thickness distribution of developed electric potential at the mid-span of the bimorph beam by the application of a 1000 N tip load with different
potential distributions.

increase in bending of the beam. The results obtained with a linear In Fig. 10 the trough-the-thickness distribution of axial stress at
through-the-thickness distribution of the electric potential deviate the root of the cantilever beam mentioned above is demonstrated
from the Ansys 2D solution [5] as r increases and approaches a for the different applied electric potentials considered.
purely piezoelectric bimorph beam. As expected, an increase in the counteracting applied potential
In Fig. 9, a constant uniformly distributed load of 1000 N is leads to a reduction in the maximum value of axial stress at the
applied to a three-layered cantilever smart beam with the thick- root of the beam since for higher values of the potential, the max-
ness ratio r ¼ 0:2. The transverse displacement is calculated con- imum transverse displacement experienced by the beam reduces.
sidering a quadratic through-the-thickness distribution of electric However, for greater applied electric potentials, the discontinuity
potential. To analyse the influence of the electric potential on the in axial stress in the interfaces between piezoelectric layers and
mechanically loaded beam, different potentials (5; 10 and 15 the aluminium substrate increases significantly, as seen in
V) are applied on top and bottom surfaces of beam in order to Fig. 10, since the stress induced by the potential on the piezoelec-
counteract the transverse displacement due to the distributed load. tric layers opposes the displacement-induced stress verified at the
648 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653

Fig. 7. Through-the-thickness distribution of axial stress at the root of the bimorph beam by the application of a 1000 N tip load with different potential distributions.

p
Fig. 8. Tip deflection of the three-layered smart beam for varying thickness ratio r ¼ 2h =h with 10 V actuation for different potential distributions.

Fig. 9. Transverse displacement along the length of the three-layered smart beam for a 1000 N uniformly distributed load and varying applied electric potential.
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 649

Fig. 10. Through-the-thickness distribution of axial stress at the root of the three-layered smart beam for a 1000 N uniformly distributed load and varying applied electric
potential.

aluminium core. Inter-laminar stresses can significantly contribute bined layers is non-piezoelectric with aluminium being the only
to delamination and consequently reduce the life of a smart beam, phase. The bottom and upper surfaces of the bottom and upper lay-
being one of the most common modes of failure of composite ers respectively are purely made of PZT-5H. The FGP bimorph
materials. Functionally graded piezoelectric materials solve this beam can be visualized in Fig. 2. The distribution of the volume
issue since multiple-layered structures are no longer necessary, fraction of each constituents was previously defined in Eq. (33)
thus avoiding abrupt material discontinuities. and varies significantly with the parameter t. This distribution is
plotted in Fig. 12 considering different values of t.
3.2.2. Sensor configuration The actuator configuration of the FGP beam is studied consider-
For the sensor configuration of the three-layered beam, a tip ing different distributions of the piezoelectric constituent. The
load of 1000 N is applied and the electric potential developed transverse displacement of the beam is computed considering a
at mid-span for varying thickness ratios r is computed. The results constant applied electric potential of 10 V on the top and bottom
are obtained for both linear and quadratic through-the-thickness surfaces, with the aluminium core grounded, and with varying t.
distributions of electric potential, both considering a shape param- The results presented in Fig. 13 are obtained considering a quadra-
pffiffiffiffi
eter of c ¼ 2= N with N ¼ 31. The solutions are posteriorly com- tic distribution of electric potential, with shape parameter
pffiffiffiffi
pared with Ansys 2D results from [5] and can be observed in c ¼ 5= N and with N ¼ 15, since convergence occurs very rapidly.
Fig. 11. Accurate results are obtained with the quadratic formula- The beam with t ¼ 0 corresponds to a purely piezoelectric bimorph
tion, while the linear distribution leads to a slight overestimation beam and the respective transverse displacement is compared
of the developed potential. with Ansys 2D solution [5]. It can be observed that an increase in
t leads to less deflection, since the PZT-5H volume fraction
3.3. FGP bimorph beam decreases and therefore the electric potential loses its influence.
The through-the-thickness distribution of axial stress at the
The functionally graded bimorph beam is composed of two root of the FGP bimorph beam with 10 V actuation is demon-
symmetrical layers of a two-phase material. The core of the com- strated in Fig. 14, considering different PZT-5H volume fraction

Fig. 11. Developed electric potential at mid-span of the three-layered smart beam for varying thickness ratio with a 1000 N tip load for different potential distributions.
650 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653

Fig. 12. Through-the-thickness volume fraction distribution of PZT-5H of the FGP bimorph beam for different values of t.

Fig. 13. Transverse displacement along the length of the FGP bimorph beam for 10 V actuation with different values of t.

Fig. 14. Through-the-thickness distribution of axial stress at the root of the FGP bimorph beam for a 10 V actuation with different values of t.
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 651

distributions. A reduction in t results in higher values of axial stress smooth transition from the aluminium core to the piezoelectric
since transverse displacement is larger, as previously stated. outer surfaces, at least at the macroscale, stress discontinuities
In Fig. 15 the transverse displacement along the length of typical of multiple-layered composites are avoided, thus signifi-
the FGP bimorph beam is displayed considering t ¼ 5 and cantly minimizing delamination probability.
different applied electric potentials. The selection of the param- In Table 5 the tip deflection of the FGP bimorph beam is pre-
eter is purposely done so as to mimic a typical three-layered sented considering different values of the parameter t and for var-
actuator with thin piezoelectric layers and a thicker aluminium ious electric actuations. In Table 6 the corresponding maximum
core. axial stresses calculated at the root of the beam can be observed,
The corresponding through-the-thickness distributions of axial considering the same values of t and the same electric actuations.
pffiffiffiffi
stress with varying applied electric potentials, calculated at the All results are obtained considering the shape parameter c ¼ 5= N
root of the beam, are demonstrated in Fig. 16. Since there is a with N ¼ 15.

Fig. 15. Transverse displacement along the length of the FGP bimorph beam with different electric actuations and for t ¼ 5.

Fig. 16. Through-the-thickness distribution of axial stress at the root of the FGP bimorph beam for t ¼ 5 and with different electric actuations.

Table 5
Tip deflection [lm] of FGP bimorph beam with different electric actuations and for different values of t.

Actuation [V] t¼0 t ¼ 0:2 t ¼ 0:5 t¼1 t¼2 t¼5

5 0:395 0:324 0:252 0:186 0:122 0:060
10 0:789 0:648 0:503 0:372 0:243 0:120
15 1:184 0:969 0:753 0:558 0:365 0:179
20 1:579 1:296 1:007 0:744 0:487 0:239
30 2:368 1:939 1:507 1:117 0:729 0:358
652 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653

Table 6
Maximum axial stress 104 [Pa] at the root of the FGP bimorph beam with different electric actuations and for different values of t

Actuation [V] t¼0 t ¼ 0:2 t ¼ 0:5 t¼1 t¼2 t¼5

5 4:227 3:755 3:301 2:865 2:445 2:039
10 8:454 7:509 6:603 5:729 4:889 4:079
15 12:681 11:264 9:904 8:593 7:334 6:118
20 16:908 15:019 13:205 11:458 9:779 8:158
30 25:362 22:528 19:808 17:187 14:668 12:236

4. Conclusions [9] Raja S, Prathap G, Sinha PK. Active vibration control of composite sandwich
beams with piezoelectric extension-bending and shear actuators. Smart Mater
Struct 2002;11(1):63–71.
In this work Kansa’s asymmetric collocation method based on [10] Robbins DH, Reddy JN. Analysis of piezoelectrically actuated beams using a
multiquadric radial basis functions is employed to solve the static layer-wise displacement theory. Comput Struct 1991;41(2):265–79.
[11] Saravanos DA, Heyliger PR. Coupled layerwise analysis of composite beams
governing equations of three distinct smart beams. ESL-FSDT is
with embedded piezoelectric sensors and actuators. J Intell Mater Syst Struct
used for the mechanical degrees of freedom with layerwise electric 1995;6:350–62.
potentials. In the mathematical formulation of the model, through- [12] Donthireddy P, Chandrashekhara K. Modeling and shape control of composite
beams with embedded piezoelectric actuators. Compos Struct 1996;35
the-thickness distribution of the electric potential is considered
(2):237–44.
quadratic for improved accuracy. [13] Narayanan S, Balamurugan V. Finite element modelling of piezolaminated
The model is validated considering the first two case studies, smart structures for active vibration control with distributed sensors and
where bimorph and three-layered smart beams are investigated actuators. J Sound Vib 2003;262:529–62.
[14] Neto Maria Augusta, Yu Wenbin, Roy Sitikantha. Two finite elements for
in both sensor and actuator configurations. FGP bimorph smart general composite beams with piezoelectric actuators and sensors. Finite Elem
beams, are subsequently tested considering different applied elec- Anal Des 2009;45(5):295–304.
tric potentials and different volume fraction distributions of the [15] Rathi Vivek, Khan Arshad Hussain. Vibration attenuation and shape control of
surface mounted, embedded smart beam. Latin Ame J Solids Struct 2012;1
constituents. Stress discontinuities are avoided using functionally (3):1–25.
graded piezoelectric material, preventing delamination. [16] Adnan Elshafei M, Alraiess Fuzy. Modeling and analysis of smart piezoelectric
The global interpolation scheme presented in this work pro- beams using simple higher order shear deformation theory. Smart Mater
Struct 2013;22(3):14. 035006.
duces accurate results while maintaining a very low computational [17] Peng XQ, Lam KY, Liu GR. Active vibration control of composite beams with
cost. The ease of implementation of the method is another attrac- piezoelectrics: A finite element model with third order theory. J Sound Vib
tive advantage in using RBFs for the interpolation of the governing 1998;209(4):635–50.
[18] Chandrashekhara K, Varadarajan S. Adaptive shape control of composite
differential equations. Moreover, convergence of the solutions is beams with piezoelectric actuators. J Intell Mater Syst Struct 1997;8.
achieved rather rapidly when compared to more popular methods [19] Mitchell JA, Reddy JN. A refined hybrid plate theory for composite
such as finite elements. In fact, Ansys 2D simulations used for com- laminates with piezoelectric laminae. Int J Solids Struct 1995;32
(16):2345–67.
parison are obtained with a refined mesh of 100x40 elements for
[20] Kapuria S, Hagedorn P. Unified efficient layerwise theory for smart beams with
converged results [5,6] while in the present work 31 or fewer segmented extension/shear mode, piezoelectric actuators and sensors. J Mech
nodes are used for the spatial discretization of the beam, consider- Mater Struct 2007;2(7):1267–98.
ing all the cases investigated. [21] Rachmadini Yanti, Lee Sangki, Park Hoon Cheol, Yoon Kwang Joon, Goo Nam
Seo. The effect of electro-mechanical coupling stiffness on the through-the-
One main disadvantage, however, is choosing an adequate value thickness electric potential distribution of piezoelectric actuators. Smart Mater
for the shape parameter in order to avoid matrix ill-conditioning Struct 2005;14(4):754–8.
while still achieving accurate results, since the solution is very sen- [22] Plagianakos Theofanis S, Saravanos Dimitris A. Coupled high-order shear
layerwise analysis of adaptive sandwich piezoelectric composite beams. AIAA J
sitive to small changes of c. 2005;43(4):885–94.
[23] Zhu Xinhua, Meng Zhongyan. Operational principle, fabrication and
displacement characteristics of a functionally gradient piezoelectric ceramic
Acknowledgements actuator. Sens Actuators A 1995;48:169–76.
[24] Carl Wu CM, Kahn Manfred, Moy Walter. Piezoeletric ceramics with functional
The support of Ministério da Ciência, Tecnologia e do Ensino gradients: a new application in material design. J Am Ceram Soc 1996;79
(3):809–12.
Superior and Fundo Social Europeu (MCTES and FSE) under pro- [25] Lezgy-Nazargah M, Vidal P, Polit O. An efficient finite element model for static
grams POPH-QREN and Investigador FCT are gratefully and dynamic analyses of functionally graded piezoelectric beams. Compos
acknowledged. Struct 2013;104:71–84.
[26] Lim CW, He LH. Exact solution of a compositionally graded piezoelectric layer
under uniform stretch, bending and twisting. Int J Mech Sci 2001;43
References (11):2479–92.
[27] Reddy JN, Cheng Z-Q. Three-dimensional solutions of smart functionally
[1] Benjeddou A, Trindade MA, Ohayon R. A unified beam finite element model for graded plates. J Appl Mech 2001;68(2):234–40.
extension and shear piezoelectric actuation mechanisms. J Intell Mater Syst [28] Zhong Z, Shang ET. Three-dimensional exact analysis of a simply supported
Struct 1997;8(12):1012–25. functionally gradient piezoelectric plate. Int J Solids Struct 2003;40:
[2] Crawley Edward F, De Luis Javier. Use of Piezoelectric Actuators as Elements of 5335–52.
Intelligent Structures. AIAA Journal 1987;25(10):1373–85. [29] Lu Pin, Lee HP, Lu C. An exact solution for functionally graded piezoelectric
[3] Clinton YK Chee, Liyong Tong, Grant P Steven. A mixed model for composite laminates in cylindrical bending. Int J Mech Sci 2005;47(3):437–8.
beams with piezoelectric actuators and sensors. Smart Mater Struct 1999;8 [30] Lu Pin, Lee HP, Lu C. Exact solutions for simply supported functionally graded
(3):417–32. piezoelectric laminates by Stroh-like formalism. Compos Struct 2006;72
[4] Wang F, Tang GJ, Li DK. Accurate modeling of a piezoelectric composite beam. (3):352–63.
Smart Mater Struct 2007;16:1595–602. [31] Shi ZF, Chen Y. Functionally graded piezoelectric cantilever beam under load.
[5] Sulbhewar Litesh N, Raveendranath P. An accurate novel coupled field Arch Appl Mech 2004;74(3–4):237–47.
Timoshenko piezoelectric beam finite element with induced potential [32] Tingting Liu, Zhifei Shi. Bending behavior of functionally gradient piezoelectric
effects. Latin Am J Solids Struct 2014;11:1628–50. cantilever. Ferroelectrics 2004;308:43–51.
[6] Litesh Sulbhewar N, Raveendranath P. An accurate higher-order modelling of [33] Lee H. Layerwise laminate analysis of functionally graded piezoelectric
extension mode smart beams with consistent through-the-thickness electric bimorph beams. J Intell Mater Syst Struct 2005;16(4):365–71.
potential distribution. J Intell Mater Syst Struct 2015;26:1191–206. [34] Yang J, Xiang HJ. Thermo-electro-mechanical characteristics of functionally
[7] Zhang XD, Sun CT. Formulation of an adaptive sandwich beam. Smart Mater graded piezoelectric actuators. Smart Mater Struct 2007;16:784–97.
Struct 1996;5:814–23. [35] Behjat B, Salehi M, Armin A, Sadighi M, Abbasi M. Static and dynamic analysis
[8] Abramovich H, Pletner B. Actuation and sensing of piezolaminated sandwich of functionally graded piezoelectric plates under mechanical and electrical
type structures. Compos Struct 1997;38(1–4):17–27. loading. Scientia Iranica 2011;18(4 B):986–94.
 T.R.C. Chuaqui, C.M.C. Roque / Composite Structures 176 (2017) 640–653 653

[36] Doroushi A, Eslami MR, Komeili A. Vibration analysis and transient [50] Ferreira AJM, Roque CMC, Martins PALS. Radial basis functions and higher-
response of an FGPM beam under thermo-electro-mechanical loads using order shear deformation theories in the analysis of laminated composite
higher-order shear deformation theory. J Intell Mater Syst Struct 2011;22 beams and plates. Compos Struct 2004;66(1–4):287–93.
(3):231–43. [51] Ferreira AJM, Batra RC, Roque CMC, Qian LF, Martins PALS. Static analysis of
[37] Ohs RR, Aluru NR. Meshless analysis of piezoelectric devices. Comput Mech functionally graded plates using third-order shear deformation theory and a
2001;27(1):23–36. meshless method. Compos Struct 2005;69(4):449–57.
[38] Liu GR, Dai KY, Lim KM, Gu YT. A point interpolation mesh free method for [52] Ferreira AJM, Batra RC, Roque CMC, Qian LF, Jorge RMN. Natural frequencies of
static and frequency analysis of two-dimensional piezoelectric structures. functionally graded plates by a meshless method. Compos Struct 2006;75(1–
Comput Mech 2002;29(6):510–9. 4):593–600.
[39] Yu Yin, Yang Cao, and Lin Quan Yao. A new meshless method based on 2D-EFG [53] Roque CMC, Ferreira AJM, Jorge RMN. A radial basis function approach for the
for the analysis of piezoelectric laminated Timoshenko beam. In: Proceedings free vibration analysis of functionally graded plates using a refined theory. J
of the 2011 Symposium on Piezoelectricity, Acoustic Waves and Device Sound Vib 2007;300(3–5):1048–70.
Applications, SPAWDA 2011; 2011. [54] Neves AMA, Ferreira AJM, Carrera E, Roque CMC, Cinefra M, Jorge RMN, Soares
[40] Ray MC. Mesh-free models for static analysis of smart laminated composite CMM. Bending of FGM plates by a sinusoidal plate formulation and collocation
beams. Int J Numer Methods Eng 2016. with radial basis functions. Mech Res Commun 2011;38(5):368–71.
[41] Hardy RL. Multiquadric equations of topography and other irregular surfaces. J [55] Li Sen, Yao Linquan, Yi Shichao, Wang Wei. A meshless radial basis function
Geophys Res 1971;76(8):1905–15. based on partition of unity method for piezoelectric structures. Math Problems
[42] Hardy RL. Theory and applications of the multiquadric-biharmonic method. Eng 2016;1–15:2016.
Comput Math Appl 1990;19(8/9):163–208. [56] Fam George SA, Rashed Youssef F. An efficient meshless technique for the
[43] Kansa EJ. Multiquadrics-A scattered data approximation scheme with solution of transversely isotropic two-dimensional piezoelectricity. Comput
applications to computational fluid-dynamics-I surface approximations and Math Appl 2015;69(5):438–54.
partial derivative estimates. Comput Math Appl 1990;19(8–9):127–45. [57] Fasshauer GE. Solving partial differential equations by collocation with radial
[44] Kansa EJ. Multiquadrics-A scattered data approximation scheme with basis functions. In: Le M’ehaut’e A, Rabut C, Schumaker LL, editors. Surface
applications to computational fluid-dynamics-II solutions to parabolic, Fitting and Multiresolution Methods. Vanderbilt; 1997. p. 131–8.
hyperbolic and elliptic partial differential equations. Comput Math Appl [58] Kansa EJ, Hon YC. Circumventing the Ill-conditioning problem with
1990;19(8–9):147–61. multiquadric radial basis functions: applications to elliptic partial
[45] Liu GR, Gu YT. A Local Radial Point Interpolation Method (LRPIM) for free differential equations. Comput Math Appl 2000;39(7–8):123–37.
vibration analyses of 2-D solids. J Sound Vib 2001;246(1):29–46. [59] Ling L, Hon YC. Improved numerical solver for Kansa’s method based on affine
[46] Wang JG, Liu GR. A point interpolation meshless method based on radial basis space decomposition. Eng Anal Boundary Elem 2005;29:1077–85.
functions. Int J Numer Methods Eng 2002;54:1623–48. [60] Reddy JN. Mechanics of laminated composite plates and shells – Theory and
[47] Ferreira AJM. A formulation of the multiquadric radial basis function method analysis. 2nd edition. CRC Press; 2003.
for the analysis of laminated composite plates. Compos Struct 2003;59 [61] Nye JF. Physical properties of crystals. New York: Oxford University Press, USA;
(3):385–92. 1985.
[48] Ferreira AJM. Thick composite beam analysis using a global meshless [62] Tzou HS, Tseng CI. Distributed piezoelectric sensor/actuator design for
approximation based on radial basis functions. Mech Adv Mater Struct dynamic measurement/control of distributed parameter systems: a
2003;10(3):271–84. piezoelectric finite element approach. J Sound Vib 1990;138(1):17–34.
[49] Ferreira AJM, Roque CMC, Martins PALS. Analysis of composite plates using [63] Reddy JN. Analysis of functionally graded plates. Int J Numer Methods Eng
higher-order shear deformation theory and a finite point formulation based on 2000;47:663–84.
the multiquadric radial basis function method. Compos Part B: Eng 2003;34 [64] Mori T, Tanaka K. Average stress in matrix and average elastic energy of
(7):627–36. materials with misfitting inclusions. Acta Metallurgica 1973;21(5):571–4.