ARTICLE IN PRESS

Journal of Sound and Vibration 327 (2009) 925

Contents lists available at ScienceDirect

Journal of Sound and Vibration

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

An electromechanical nite element model for piezoelectric energy

harvester plates
Carlos De Marqui Junior a,, Alper Erturk b, Daniel J. Inman c
a
Department of Aeronautical Engineering, Engineering School of Sao Carlos, University of Sao Paulo, Brazil
b
Center for Intelligent Material Systems and Structures, Department of Engineering Science and Mechanics, Virginia Tech, USA
c
Center for Intelligent Material Systems and Structures, Department of Mechanical Engineering, Virginia Tech, USA

a r t i c l e in fo abstract

Article history: Vibration-based energy harvesting has been investigated by several researchers over the
Received 13 January 2009 last decade. The goal in this research eld is to power small electronic components by
Received in revised form converting the waste vibration energy available in their environment into electrical
5 May 2009
energy. Recent literature shows that piezoelectric transduction has received the most
Accepted 12 May 2009
attention for vibration-to-electricity conversion. In practice, cantilevered beams and
Handling Editor: C.L. Morfey
Available online 11 June 2009 plates with piezoceramic layers are employed as piezoelectric energy harvesters. The
existing piezoelectric energy harvester models are beam-type lumped parameter,
approximate distributed parameter and analytical distributed parameter solutions.
However, aspect ratios of piezoelectric energy harvesters in several cases are plate-like
and predicting the power output to general (symmetric and asymmetric) excitations
requires a plate-type formulation which has not been covered in the energy harvesting
literature. In this paper, an electromechanically coupled nite element (FE) plate model
is presented for predicting the electrical power output of piezoelectric energy harvester
plates. Generalized Hamiltons principle for electroelastic bodies is reviewed and the FE
model is derived based on the Kirchhoff plate assumptions as typical piezoelectric energy
harvesters are thin structures. Presence of conductive electrodes is taken into account in
the FE model. The predictions of the FE model are veried against the analytical solution
for a unimorph cantilever and then against the experimental and analytical results of a
bimorph cantilever with a tip mass reported in the literature. Finally, an optimization
problem is solved where the aluminum wing spar of an unmanned air vehicle (UAV) is
modied to obtain a generator spar by embedding piezoceramics for the maximum
electrical power without exceeding a prescribed mass addition limit.
& 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The research interest in converting ambient vibration energy to usable electrical energy has increased in the last years
[15]. The concept of energy harvesting is particularly useful for wireless sensors powered by batteries and remotely operated
systems with limited energy source. The goal of the research in vibration-based energy harvesting is to provide electrical
energy for such systems by utilizing the vibrations available in their environment. Unmanned air vehicles (UAVs) and micro air
vehicles (MAVs) constitute unique application systems where the possibility of an additional energy source is very important.
UAVs are designed to maximize the endurance and ight range with the limited energy available in operation. A possible

 Corresponding author. Tel.: +55 16 33739417; fax: +55 16 33739590.

E-mail addresses: [email protected] (C. De Marqui Junior), [email protected] (A. Erturk), [email protected] (D.J. Inman).

0022-460X/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jsv.2009.05.015
 ARTICLE IN PRESS
10 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

source of energy for charging the batteries of UAVs is the mechanical vibration energy due to unsteady aerodynamic loads
during the ight [6] and due to ground excitation in perching [7,8]. Different transduction mechanisms (piezoelectric [912],
electromagnetic [1317] and electrostatic [18,19]) can be used for converting vibrations to electricity. However, the recent
literature shows that piezoelectric transduction has received the most attention for vibration-based energy harvesting and
several review articles directly focusing on piezoelectric energy harvesting can be found in the literature [1,35].
Piezoelectric power generators can harvest electrical energy from mechanical vibrations based on the direct
piezoelectric effect. These generators have been extensively studied as a low-cost and efcient alternative for low-level
energy harvesting. Researchers have proposed various models to represent the electromechanical behavior of piezoelectric
energy harvesters, which range from lumped parameter models [9,20] to RayleighRitz type approximate distributed
parameter models [10,11,20] as well as analytical distributed parameter solution attempts [21,22]. Recently, certain issues
observed in some of these lumped parameter and distributed parameter piezoelectric energy harvester models have been
claried in the literature [23]. More recently, the analytical distributed parameter solutions for unimorph [12,24] and
bimorph [25] piezoelectric energy harvester congurations with closed-form expressions have been presented.
Convergence of the RayleighRitz type electromechanical solution [10,11,20] to the analytical solution given by Erturk
and Inman [24] was observed by Elvin and Elvin [26] when sufcient number of admissible functions were used. The
lumped parameter solution [20] have been found useful for a fundamental understanding of the problem and to investigate
the optimization of system parameters for better electrical outputs [27,28]. However, accurate prediction of the
electromechanical behavior of piezoelectric energy harvesters requires using distributed parameter solutions.
Experimental verications and validations were also reported for the approximate [10,11] and analytical [12,25] (beam-
type) distributed parameter electromechanical solutions.
The literature of piezoelectric sensing and actuation includes nite element (FE) models for plates with piezoceramic
materials [2933]. Although these FE models have not been used to study the energy harvesting problem, they provide the
basis for modeling of a piezoelectric energy harvester. As far as the literature of FE modeling is considered, it can be
observed that some of these FE models do not account for the presence of conductive electrodes bracketing the
piezoceramic layer (e.g., [29]); although, in practice, piezoceramic layers usually come with highly conductive electrode
layers from the manufacturer. If the presence of the conductive electrodes is not taken into consideration, a space-
dependent electric potential distribution is obtained throughout the surface of the piezoceramic, yielding a different
electric potential term (i.e., electrical degree of freedom) for each nite element. Some authors have considered the
presence of the electrodes in the electromechanical problem [30,31] and obtained one voltage output (i.e., potential
difference) between the electrode pair covering the piezoceramic. However, regardless of this electrode-based
consideration, most of these models in the literature have focused on structural actuation and damping and plate-type
formulation has not been considered in the literature of energy harvesting.
This work presents an electromechanical FE plate model for piezoelectric energy harvesting. Since piezoelectric energy
harvesters are designed and manufactured as thin structures, the classical plate theory is employed in the formulation. A resistive
electrical load is considered in the electrical domain, in agreement with the simplied analyses followed by others [912,2026].
Electrical circuitry-based work dealing with AC-to-DC (alternating current-to-direct current) converters can be found in the
literature [3437]. First the FE model is veried against the analytical solution for a unimorph cantilever under base excitation
presented by Erturk and Inman [24]. The electromechanical frequency response functions (FRFs) obtained using the FE model are
compared with the analytically obtained FRFs referred from the aforementioned work. As a second case study, the analytical and
experimental FRFs of a bimorph cantilever with a tip mass reported by Erturk and Inman [25] are predicted by using the FE model.
Finally, the FE model is used to solve an optimization problem for UAV applications. The aluminum wing spar of a UAV is modied
to design a generator wing spar. Since mass densities of typical piezoceramics are considerably large for UAV applications, a
limiting value for mass addition is imposed to the problem as a design constraint. Dimensions of the embedded piezoceramic are
identied for the maximum electrical power output of the generator spar with embedded piezoceramics.

2. Finite element modeling of a piezoelectric energy harvester plate

2.1. Generalized Hamiltons principle for a piezoelectric energy harvester

In the absence of magnetic effects, the generalized Hamiltons principle for an electroelastic body is [38]
Z t
2
dT  U W e dW dt 0 (1)
t1

where the total kinetic energy (T), the total potential energy (U) and the electrical energy (W e ) terms are dened as
Z Z
1 1
T rs u_ t u_ dV s rp u_ t u_ dV p (2a)
Vs 2 Vp 2

Z Z
1 t 1 t
U S T dV s S T dV p (2b)
Vs 2 Vp 2
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 11

Z
1 t
We E D dV p (2c)
Vp 2

Here, u is the vector of mechanical displacements, S is the vector of mechanical strain components, T is the vector of
mechanical stress components, D is the vector of electric displacement components, E is the vector of electric eld
components, r is the mass density, V is the volume, t denotes transpose when it is used as a superscript (otherwise it stands
for the time) and an over-dot represents differentiation with respect to time. Here and hereafter, subscripts s and p stand
for the substructure and piezoceramic layers, respectively.
For a set of discrete mechanical forces f applied at locations xi ; yi and for a set of discrete electric charge outputs q
extracted at locations xj ; yj , one can express the variation of the mechanically applied and electrically extracted work as follows:
nf
X nq
X
dW duxi ; yi ; t  fxi ; yi ; t djxj ; yj ; tqxj ; yj ; t (3)
i1 j1

where nf is the number of discrete mechanical forces, jj is the scalar electrical potential and nq is the number of discrete
electrode pairs.
The generalized Hamiltons principle for electroelastic bodies [38] described by the foregoing equations was previously
employed by Hagood et al. [39] where they combined the RayleighRitz method and EulerBernoulli beam theory for
active structural control. The RayleighRitz formulation given by Hagood et al. [39] was recently implemented by Sodano
et al. [10] and duToit et al. [11,20] for predicting the electrical power output of cantilevered EulerBernoulli beams in
energy harvesting. It should be mentioned that, as oppose to the energy harvesting problem, the electric charge is the input
in the structural actuation problem, and therefore the second term in Eq. (3) has a minus sign in Hagood et al. [39]. Note
that the energy dissipation due to mechanical damping is excluded at this point and it will be introduced later in the form
of proportional damping.
The linear-elastic constitutive relation for the substructure material can be written as
T cs S (4a)
and the linear-electroelastic constitutive relation for the piezoceramic material is [40]
  " E # 
T cp et S
(4b)
D e eS E

where c is the elastic stiffness matrix, e is the matrix of piezoelectric constants, e is the matrix of permittivity components,
superscript E and S denote that the parameters are measured at constant electric eld and constant strain, respectively.
Using the constitutive relations given by Eqs. (4a) and (4b) in Eq. (1), the generalized Hamiltons principle for a
piezoelectric energy harvester becomes
Z "Zt2 Z Z Z Z Z
rs du_ t u_ dV s rp du_ t u_ dV p  dSt cs S dV s  dSt cEp S dV p dSt et E dV p dEt eS dV p
t1 Vs Vp Vs Vp Vp Vp
3
Z nf
X nq
X
t S
dE e E dV p duxi ; yi ; t  fxi ; yi ; t djxj ; yj ; tqxj ; yj ; t5 dt 0 (5)
Vp i1 j1

2.2. Electromechanically coupled nite element model

The derivation given in the following is for a unimorph piezoelectric energy harvester conguration, i.e., the harvester
plate has one substructure layer and one piezoceramic layer as shown in Fig. 1. However, the formulation can easily be
extended to bimorph congurations (plates with two piezoceramic layers bracketing a substructure layer) as summarized
at the end of this section. The substructure and the piezoceramic layers are assumed to be perfectly bonded to each other.
The piezoceramic layer (which is poled in the thickness direction) is covered by continuous electrodes (which are assumed
to be perfectly conductive) with negligible thickness. As a common practice in the literature [912,2026], a resistive load
is considered in the electrical domain and the purpose is to estimate the power generated in the electrical domain due to
mechanical vibrations of the energy harvester plate. In general, as depicted in Fig. 1, piezoelectric energy harvesters are
designed to have clamped-free (cantilevered) boundary conditions and the source of excitation is the vibratory motion of
the clamped end. Since piezoelectric energy harvesters are manufactured as thin structures for improved exibility, the
classical (Kirchhoff) plate theory is employed in the formulation.
A rectangular nite element with three mechanical degrees of freedom per node (namely the displacements u, v and w in
x, y and z directions) shown in Fig. 2 is used to model the substructure and the piezoceramic layers. Based on the assumption
that each nite element of the piezoceramic layer is completely covered with perfectly conductive electrodes (on the top and
the bottom surfaces), one electrical degree of freedom (voltage vp across the electrodes) is sufcient for modeling the
electrical response of these elements. Thus, the rectangular nite element in Fig. 2 has 13 degrees of freedom in total.
 ARTICLE IN PRESS
12 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

Fig. 1. A unimorph piezoelectric energy harvester plate with clamped-free boundary conditions.

z
w
y w4
x4
x y
4 y4
fw
x
w1
x1 1 vp 3 w3
x3
y1
y3

w2 2
x2
y2

Fig. 2. Piezoelectric nite element with 12 mechanical degrees of freedom and 1 electrical degree of freedom.

Based on the Kirchhoff plate theory, transverse shear strains and rotary inertias of the nite elements are neglected and
in-plane displacements (u and v) are assumed to be due to the bending (cross-section rotation) of the plate only. The
displacement eld is then
8 9
( )t
<u>
> = qw qw
v z z w (6)
>
: > ; qx qy
w
where the displacement components u, v and w at a thickness level z from the reference (neutral) surface are given in terms
of the transverse deection (w) of the reference surface.
The mechanical strain components can be written in terms of the displacement components as
8 9
> S ( )t ( )t
< x > = qu qv qu qv q2 w q2 w q2 w
Sy z 2 (7)
>
: 2S > ; qx qy qy qx qx2 qy2 qx qy
xy

The transverse displacement of node k of the rectangular nite element shown in Fig. 2 is assumed to vary in the
polynomial form [41]
wk wjxk ;yk Pjxk ;yk l (8a)

and consequently the bending rotations can be given as

 
qw qP
yxk l (8b)
qy xk ;yk qy xk ;yk
 
qw qP
yyk     l (8c)
qx xk ;yk qx xk ;yk
where the polynomial terms are

P 1 x y x2 xy y2 x3 x2 y xy2 y3 x3 y xy3  (9)

and the vector of generalized coordinates is

l a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 t (10)

 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 13

Considering the 4 nodes per element and the 3 degrees of freedom per node, one can dene the 12  1 vector of nodal
variables as

w fw1 yx1 yy1 w2 yx2 yy2 w3 yx3 yy3 w4 yx4 yy4 gt (11)
which can be expressed in the form of
w Al (12)
where A is a 12  12 transformation matrix [41] whose elements are given by P and its derivatives through the denitions
of wk , yxk and yyk given by Eqs. (8a)(8c).
Nodal approximations for the transverse deection as functions of the nodal variables are then
w wk Cw (13)
where

C PA1 (14)
The vector of transverse displacement and cross-section rotations is related to the vector of nodal variables as follows:
( )t
qw qw
w Bg w (15)
qx qy

where
( )t
qC qC
Bg C (16)
qx qy

Similarly, the vector of curvatures can be expressed as

( )t
q2 w q2 w q2 w
2 Bj w (17)
qx2 qy2 qx qy
where
( )t
q2 C q2 C q2 C
Bj 2 (18)
qx2 qy2 qx qy
Here, both Bg and Bj are 3  12 matrices.
The foregoing relations constitute the kinematics of the problem. The displacement components given by Eq. (6) and
consequently the strain components described by Eq. (7) can be expressed as functions of nodal variables to be used in the
Hamiltons principle given by Eq. (5).
Since the piezoceramic is poled in the thickness direction (z-direction), the non-zero electric eld component (which is
assumed to be uniform in the thickness direction) can be expressed as
qj vp
Ez   (19)
qz hp
where the electric potential is assumed to be varying linearly across the electrode pair (i.e., the electric eld is assumed to
be uniform in the thickness direction). Then, the vector of electric eld components becomes
E BE vp (20)
where
( )t
1
BE 0 0 (21)
hp

Based on the Hamiltons principle given by Eq. (5), the element mass matrix m, stiffness matrix k, electromechanical
coupling vector h, capacitance cp and the mechanical forcing vector f can be expressed as
Z Z
m Btg Zt rs ZBg dV s Btg Zt rp ZBg dV p (22a)
Vs Vp

Z Z
k z2 Btj c s Bj dV s z2 Btj c Ep Bj dV p (22b)
Vs Vp

Z
h zBtj e t BE dV p (22c)
Vp
 ARTICLE IN PRESS
14 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

Z
cp BtE e S BE dV p (22d)
Vp

Z
f Ct f w dS (22e)
S

where
2 3
z 0 0
6 7
Z4 0 z 05 (23)
0 0 1
and an over-bar in Eqs. (22a)(22e) respective term (stiffness, piezoelectric constant or permittivity) is dened in plane-
stress conditions so that the formulation is in agreement with the Kirchhoff assumption. It is worthwhile adding that the
mass and stiffness matrices (m and k are 12  12 matrices, the electromechanical coupling (h) and the mechanical forcing
vector (f) are 12  1 vectors and the capacitance term (cp) is a scalar. Note that the mechanical damping matrix will be
introduced to the system in the global coordinates. It should also be recalled that the z term in the respective integrals is
measured from the reference surface of the plate in the thickness direction [24,42].
As mentioned previously, in general, the source of excitation in energy harvesting from cantilevered plates is the motion of the
clamped end. As discussed in the literature [23], if the base is vibrating in the transverse direction (z-direction), the effective force
on the structure is due to the inertia of the structure in the same direction. Therefore, the forcing term in Eq. (22e) can be given by

f w m aB (24)
where m is the mass per unit area of the nite element (including both the piezoceramic and the substructure layers) and aB is
the base acceleration. Here, it is assumed that both layers (piezoceramic and substructure) have the same mesh and the nodes are
coincident.
The plane-stress form of the constitutive equations for the piezoelectric layer can be expressed as
8 9 2 E 38 9
> T1 > c11 c E12 0 e 31 > S1 >
>
> >
> 6 E 7>> > >
>
< T2 > = 6 c E >
< > =
6 12 c 22 0 e 32 7
7 S2
6 E 7 (25)
> T
> 6> > 6
> 4 0 0 
c66 0 5> 7 > S
> 6> >
>
> > > >
>
:D ; :E ;
3 e 31 e 32 0 S33 3

which is the reduced (2-D) form of Eq. (4b). Note that the elastic, piezoelectric and dielectric components in Eq. (25) are
given in the contracted notation (i.e., Voigts notation: 11-1, 22-2, 33-3, 23-4, 31-5 and 12-6) where 1, 2 and 3
directions are coincident with x, y and z directions.
In general, poled piezoelectric ceramics (e.g., PZT-5A, PZT-5 H) exhibit transversely isotropic material behavior (isotropic
in the 12-plane, i.e., xy-plane) and therefore the plane-stress (2-D) components in Eq. (25) can be obtained in terms of the
3-D components as [43]

cE13 2 cE13 cE23 cE23 2

c E11 cE11  ; c E12 cE12  ; c E22 cE22  ; c E66 cE66
cE33 cE33 cE33

cE13 e33 cE23 e33 e233

e 31 e31  ; e 32 e32  ; S33 S33 (26)
cE33 cE33 cE33

Using these plane-stress forms of the elastic stiffness, piezoelectric and dielectric properties of the piezoceramic layer,
the respective integrands of the piezoceramic layer in Eqs. (22a)(22d) can be expressed as follows:
" #
qCt qC qCt qC
Btg Zt rp ZBg rp z2 z2 Ct C (27a)
qx qx qy qy
" #
q2 Ct q2 C q2 Ct q2 C q2 Ct q2 C q2 Ct q2 C
z2 Btj c Ep Bj z2 c E11 2 2
2c E12 2 2
c E22 2 2
4c E66 (27b)
qx qx qx qy qy qy qx qy qx qy
!
z q2 Ct q2 Ct
zBtj e t BE e 31 2
e 32 (27c)
hp qx qx2

S33
BtE e S BE (27d)
h2p
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 15

For an isotropic substructure layer, the 2-D elastic stiffness matrix can be obtained for the plane-stress conditions as
2 3
1 ns 0
6 7
6n 1 0 7 Ys
c s 6 s 7 (28)
4 1  ns 51  n2s
0 0
2
where ns is the Poissons ratio and Y s is the Youngs modulus of the substructure material.
The global equations of motion are then obtained by assembling the element matrices given by Eqs. (22a)(22e):

MW _ KW  Hv F
CW (29a)

Cp v Q Ht W 0 (29b)
where M is the global mass matrix (nm  nm ), K is the global stiffness matrix (nm  nm ) and H are the global
electromechanical coupling matrix (nm  ne ), Cp is the diagonal global capacitance matrix (ne  ne ), F is the global vector of
mechanical forces (nm  1), Q is the global vector of electric charge outputs (ne  1), W is the global vector of mechanical
coordinates (nm  1) and v is the global vector of voltage outputs (ne  1). Here, nm and ne, respectively, are the number of
mechanical and electrical degrees of freedom of the harvester plate. In Eq. (29a), the global mechanical damping matrix
(nm  nm ) is assumed to be proportional to the mass and stiffness matrices:
C aM bK (30)
where a and b are the constant of proportionality.
Since the number of electrical degrees of freedom is equal to the number of elements, the dimension of the global vector
of voltage outputs in Eqs. (29a) and (29b) is equal to number of nite elements used in the mesh of the piezoceramic layer
(ne). This is a general case which assumes that each electromechanical nite element has its own electrode pair, insulated
from one another. Thus, ne number of different voltage outputs can be extracted from the system as Eq. (29b) represents ne
equations. In practice, however, piezoceramics come from the manufacturer with thin and very conductive electrode layers
on the top and bottom surfaces. It is therefore reasonable to assume that all nite elements (ne) generate the same voltage
output so that the elements of vector v are identical (i.e., v1 v2    vne vp ) such that

v fv1 v2 . . . vne gt f1 1 . . . 1gt vp (31)

This transformation takes into account the presence of full electrodes covering the top and the bottom surfaces of the
piezoceramic of the unimorph (Fig. 1) and the potential difference between these two electrodes is simply vp.
After dening the electromechanical coupling vector based on the transformation given by Eq. (31) as H f1 1 . . . 1gt H
(which is a nm  1 vector), taking the time derivative of Eq. (29b) and pre-multiplying by the 1  ne vector f1 1 . . . 1g, one
can obtain the following scalar equation:
t _
C p v_ p Q_ H W 0 (32)
where C p traceCp .
Since the electrical boundary condition due to a resistive load (Rl ) is Q_ vp =Rl , Eqs. (29a) and (32) become

MW _ KW  Hvp F
CW (33a)

vp t _
C p v_ p H W 0 (33b)
Rl
which are the governing electromechanical equations for a unimorph piezoelectric energy harvester plate (Fig. 1).
If a bimorph piezoelectric energy harvester plate is considered (i.e., the substructure is bracketed by two identical
piezoceramic layers) the mass, stiffness and consequently damping matrices are modied accordingly. It is known from the
literature [44,25] that the electrode pairs covering each piezoceramic layer can be connected in series or in parallel to the
external electrical load (for larger voltage or current, respectively). In general, the piezoceramic layers are poled in the same
direction for parallel connection whereas they are poled in the opposite direction for series connection. For the parallel
connection case, the effective electromechanical coupling vector is the sum of the individual contribution of each layer and
the effective capacitance is the sum of each individual capacitances. For the series connection case, the effective
electromechanical coupling vector is equal to that of one piezoceramic layer and the effective capacitance is one half of the
capacitance of one piezoceramic layer.

3. Case studies

This section presents three case studies using the electromechanical FE model described in Section 2. In the rst case
study, the FE model is veried against the analytical results from the closed-form solution given by Erturk and Inman [24]
for a unimorph harvester under base excitation. The second case study aims to predict the analytical and experimental
 ARTICLE IN PRESS
16 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

results for a bimorph harvester with a tip mass presented by Erturk and Inman [25]. The results for these two case studies
are briey discussed here and the details can be found in the respective papers [24,25]. After validating the
electromechanical FE model, power optimization for a UAV wing spar with embedded piezoceramics is studied. The
goal in this last case study is to nd the geometric parameters (length and thickness) of the piezoceramic layers that give
the maximum electrical power output under an imposed mass addition constraint.
The material and electromechanical properties for PZT-5A (used in all cases) are given in Table 1. It is worthwhile
mentioning that manufacturers typically provide limited number of properties for piezoelectric ceramics. For instance, in
the predictions of their analytical model, Erturk and Inman [25] used the data provided by Piezo Systems Inc. [45] as the
data provided by the manufacturer was sufcient for a beam-type formulation. However, the plate-type formulation given
here requires more than what is provided in the manufacturers data sheet (see, for instance, the properties required for the
calculation of the plane-stress elastic, piezoelectric and dielectric components in Eq. (26)). Therefore, the 3-D properties of
PZT-5A [46] displayed in Table 1 are used here.

3.1. Verication against the analytical results for a unimorph conguration

The numerical input data of the unimorph studied by Erturk and Inman [24] is shown in Table 2. The results of the FE
model are compared against the analytical solution in this section. The piezoceramic layer uniformly covers the
substructure layer and the conductive electrodes are connected to a resistive electrical load as depicted in Fig. 1.
Expressions for electromechanical FRFs, namely the voltage across the resistive load, current passing through the resistive
load, electrical power output and relative tip motion, are obtained from the equations of motion dened in the FE model
(Eqs. (33a) and (33b)). The electrical power FRF (per base acceleration) and the mechanical vibration FRF (relative tip
displacement per base displacement) are presented in the following. The material, electromechanical and geometric
properties given in Tables 1 and 2 are used in the FE simulations.
The excitation is due to the harmonic motion of the base in the transverse direction, wB t Y 0 ejot (where wB t is the
base displacement, Y0 is its amplitude, o is the excitation frequency and j is the unit imaginary number). At steady state,
the voltage output-to-base acceleration FRF can be obtained from Eqs. (33a) and (33b) as
 1  1 !1
vp t vp t 1 t 2 1 t
jo joC p H o M joC K jo joC p HH m (34)
aB t o2 Y 0 ejot Rl Rl

where m* is an (nm  1) mass vector obtained from the global forcing term F for the base excitation problem.
The electric current FRF is obtained by dividing the voltage FRF to the load resistance of the energy harvesting
circuit. The electrical power FRF is the product of voltage and current FRFs and it is dened as the ratio of electrical power
output to square of the base acceleration. Note that the modulus form of Eq. (34) is the peak voltage FRF (not the root mean
square value). Therefore the electrical power obtained from this voltage FRF is the peak power and the average power
amplitude is half of the peak power. The moduli of the power FRFs obtained from the FE model (thinner lines) for ve

Table 1
Material and electromechanical properties of PZT-5A.

Mass density of the PZT (kg/m3) 7800

Permittivity (nF/m) 1800  0
cE11 ; cE22 (GPa) 120.3
cE12 (GPa) 75.2
cE13 ; cE23 (GPa) 75.1
cE33 (GPa) 110.9
cE66 (GPa) 22.7
e31 ; e32 (C/m2) 5.2
e33 (C/m2) 15.9

Table 2
Geometric and material properties of the unimorph harvester.

Length of the beam (mm) 100

Width of the beam (mm) 20
Thickness of the substructure (mm) 0.5
Thickness of the PZT (mm) 0.4
Youngs modulus of the substructure (GPa) 100
Mass density of the substructure (kg/m3) 7165
Proportional constanta (rad/s) 4.886
Proportional constantb (s/rad) 1.2433  105
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 17

Fig. 3. Power FRFs for ve different values of load resistance with the enlarged view of mode 1.

different values of load resistance are plotted in Fig. 3. These FRFs are in good agreement with the analytically obtained [24]
curves (thicker lines).
The voltage FRF has a monotonic behavior with increasing load resistance for every excitation frequency. That is, the
voltage across the resistive load increases with increasing load resistance and the frequency of maximum voltage output
moves from short circuit to open circuit resonance frequency. The electrical current FRF exhibits an opposite behavior since
current passing through the electrical load decreases with increasing load resistance for every excitation frequency [24].
The product of these two FRFs gives the electrical power, which shows intersections between the curves of different
electrical loads (Fig. 3). That is, at a xed frequency, the variation of peak power output with load resistance is not
monotonic as can be seen from the enlarged view of the rst mode plotted in Fig. 3. Consequently the short circuit
resonance frequency (47.8 Hz) and the open circuit resonance frequency (48.8 Hz) have considerably different optimum
load values [24] although the difference between these two frequencies is just 1 Hz. The short circuit and the open circuit
resonance frequencies predicted by the FE model are 47.7 and 48.8 Hz, respectively.
The relative tip motion FRF is dened as the ratio of the amplitude of the displacement at the tip of the beam (relative to
the base) to the amplitude of base displacement input and its is obtained from Eqs. (33a) and (33b) as
 1 !1
wrel 1 t
o2 o2 M joC K jo joC p HH m (35)
Y 0 ejot Rl

which denes a vector for the vibration response of all coordinates. Here, the component of interest in the vector dened by
Eq. (35) is the transverse tip displacement.
The mechanical vibration (motion transmissibility) FRFs of the unimorph obtained from the FE model and the analytical
solution are presented in Fig. 4. As can be observed from the enlarged view of Fig. 4, the vibration amplitude at short circuit
resonance frequency (47.8 Hz) decreases as the load resistance is increased from 102 to 105 O due to the resistive shunt
damping effect associated with power generation. The motion at the open circuit resonance frequency (48.8 Hz) is
amplied with increasing load resistance. Simultaneous investigation of the enlarged views in Figs. 3 and 4 provides a
better understanding of the piezoelectric shunt damping effect [47] with changing load resistance. The uncoupled FRF
presented in Fig. 4 is the typical mechanical FRF (transmissibility) when no electromechanical coupling is considered. The
electromechanically coupled vibration FRF converges to the uncoupled FRF when the system is close to short circuit
conditions (Rl ! 0). Note that the inaccuracy in the FE predictions around the anti-resonance in Fig. 4 is related to the FE
mesh size (25  5 nite elements for each layer in this case).

3.2. Verication against the experimental and analytical results for a bimorph conguration

In the second case study, the results obtained from the electromechanical FE model for a cantilevered bimorph with a
tip mass under base excitation (Fig. 5) are compared with the single-mode analytical predictions of the closed-form
solution and experimental results presented by Erturk and Inman [25]. The bimorph harvester conguration has a brass
substructure bracketed by two PZT-5A layers. The piezoceramic layers are poled in the opposite directions and therefore
the combination of the layers to the electrical load in Fig. 5 is the series connection case.
The expressions for the electromechanical FRFs are obtained from the global equations of motion. The voltage FRF has
the same form of the expression for a unimorph harvester given by Eq. (34). However, the contribution of the additional
piezoceramic layer has to be added to the mass, stiffness and damping matrices as well as to the mechanical forcing term.
The tip mass also affects the mass matrix and the mechanical forcing term since the forcing vector is a function of the
 ARTICLE IN PRESS
18 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

Fig. 4. Relative tip motion FRFs for ve different values of load resistance with the enlarged view of mode 1.

Fig. 5. Bimorph piezoelectric energy harvester (series connection).

Table 3
Geometric and material properties of the harvester.

Length of the beam (mm) 50.8

Width of the beam (mm) 31.8
Thickness of the substructure (mm) 0.14
Thickness of the PZT (mm) 0.26 (each)
Youngs modulus of the substructure (GPa) 105
Mass density of the substructure (kg/m3) 9000
Tip mass (kg) 0.012
Proportional constanta (rad/s) 14.65
Proportional constantb (s/rad) 105

inertia of the structure in the base excitation problem. As stated previously, the effective piezoelectric coupling term in the
series connection case is equal to the contribution of one piezoceramic layer and the effective capacitance is equal to one
half of the contribution of one piezoceramic layer. The parameters used in the simulations are presented in Tables 1 and 3.
The voltage FRF is dened here as the voltage output per base acceleration (in terms of the gravitational acceleration,
g 9.81 m/s2) to be in agreement with the experimental and analytical voltage FRFs given by Erturk and Inman [25].
Eq. (34), adapted to represent the bimorph harvester with a tip mass, is also easily modied to give voltage output per base
acceleration. The voltage FRFs for the rst mode of this harvester obtained from the FE model are plotted in Fig. 6 (12  8
nite elements were used for each layer in this case) along with the analytical solution and experimental results for eight
different values of load resistance (1, 6.7, 11.8, 22, 33, 47, 100, 470 kO). A similar monotonic behavior of voltage output with
increasing load resistance is observed for all excitation frequencies according to the numerical (FE model), analytical and
experimental results. The experimental short circuit and open circuit resonance frequencies for the harvester are 45.6 and
48.4 Hz, respectively. The analytical model predicts these frequencies as 45.7 and 48.2 Hz, respectively. The FE model
predictions of the short circuit and open circuit resonance frequencies are 45.7 and 48.3 Hz, respectively.
The mechanical vibration FRFs of the bimorph piezoelectric energy harvester obtained from the FE model, analytical
model and experimental tests are presented in Fig. 7. The tip velocity FRF is dened as the ratio of the amplitude of
velocity at the tip of the beam (relative to the xed frame) to the gravitational acceleration as it is measured by a laser
vibrometer located on the xed ground. This FRF is easily obtained from the expression of the relative tip motion FRF
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 19

Fig. 6. Analytical, FE and experimental voltage FRFs for eight different values of load resistance.

Fig. 7. Analytical, FE and experimental tip velocity FRFs for eight different values of load resistance.

(Eq. (35)) using jg=o1 wrel L; t=Y 0 . It is observed in Fig. 7 that the mechanical FRFs obtained from the FE model are in
agreement with the analytical and experimental results. The vibration amplitude at the short circuit resonance frequency is
attenuated as the load resistance is increased up to 100 kO. Approximately after this value of load resistance, increasing
load resistance amplies the vibration amplitude at the open circuit resonance frequency and the vibration amplitude at
the short circuit resonance frequency is no longer attenuated.

3.3. Design of an energy harvesting wing spar with embedded piezoceramics

Using the electromechanical FE model, an optimization problem is addressed in this section to design a UAV wing spar
with embedded piezoceramics. This class of aircrafts has limited source of energy, limited size and load capacity. The
generator spar, an additional source of energy with acceptable mass addition, could assist the general goal of maximum
endurance and ight range or power small electronics components of the UAV.
The dimensions of the original wing spar are given as 300  30  12 mm3 and the purpose is to replace some part of the
original spar material (aircraft aluminum alloy Al 2024-T3) by piezoceramic material (PZT-5A). Two identical layers of PZT-
5A are embedded to the top and the bottom of the spar by replacing the original material as shown in Fig. 8. As the mass
density of PZT-5A (7800 kg/m3) is considerably larger than that of the aluminum alloy (2750 kg/m3), the critical parameter
in the optimization problem is the amount of mass added to the wing spar. Therefore, the following constraint is imposed
 ARTICLE IN PRESS
20 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

Fig. 8. Generator wing spar with embedded piezoceramics (series connection).

Table 4
Geometric and material properties of the aluminum wing spar.

Length of the spar (mm) 300

Width of the spar (mm) 30
Thickness of the spar (mm) 12
Youngs modulus of the spar (GPa) 70.0
Mass density of the substructure (kg/m3) 2750
Proportional constanta (rad/s) 21.28
Proportional constantb (s/rad) 105

to the optimization problem: The total mass added by the piezoceramic layers to the nal conguration of the piezoelectric
energy harvester wing spar should not exceed 10% of the mass of the original aluminum spar. The goal is to nd the optimal
geometric conguration (length and thickness) of the embedded piezoceramic layers that will give the maximum electrical
power output and satisfy the imposed mass restriction. The width of the piezoceramic is assumed to be the same as that of
the spar.
It is worthwhile to mention that the following analysis assumes that the excitation amplitude (in terms of base
acceleration of the spar, aB) is not large enough to cause failure in the wing spar both for the original and the modied
congurations. It is well known that typical piezoelectric ceramics are very brittle compared to classical engineering
materials (such as the aircraft aluminum alloy considered here which has an ultimate tensile strength of 483 MPa
according to Ref. [48]). Although the compressive strength values of piezoceramics are fairly large (Morganelectroceramics
[49] reports the compressive strength of PZT-5A to be larger than 517 MPa), both the static and dynamic tensile strengths of
PZT-5A are poor compared to typical engineering materials. The reported [49] static and dynamic tensile strength values
of PZT-5A are 75.8 and 27.6 MPa, respectively. Therefore, clearly the original wing spar (made of Al 2024-T3) can carry
larger base accelerations compared to the generator wing spar (Fig. 8). However, the following analysis provides the results
normalized per base acceleration input, which is assumed to be smaller than the acceleration that would cause failure in
the generator spar.
The geometric and material properties of the original spar (aircraft aluminum alloy Al 2024-T3) are presented in Table 4
(the properties of PZT-5A can be found in Table 1). The length-to-thickness ratio of the spar is slender enough to neglect the
shear deformation and rotary inertia effects for the fundamental vibration mode. The source of excitation is assumed to be
base excitation and the spar is assumed to have clamped-free boundary conditions, where the clamped end is the fuselage
side of the spar. The oppositely poled piezoceramic layers are completely covered by conductive electrodes and combined
in series.
The electromechanical behavior of the generator spar is initially investigated using a set of dimensions for two identical
piezoceramic layers (Fig. 9). The length of the piezoceramic layers is increased with a xed increment (10% of the spar
length). At each length, the electromechanical behavior is investigated for eleven different thickness values of the layers,
ranging from 0.3 to 3.3 mm with an increment of 0.3 mm. For each geometric conguration, the electromechanical system
is assumed to be excited at the short circuit resonance frequency of the rst vibration mode. Thus, the power output, the
parameter to be optimized within the boundaries established by the mass restriction, is obtained for a wide range of load
resistance until the optimal one is observed. Variation of the maximum electrical power output with dimensionless length
and dimensionless thickness of the piezoelectric energy harvester is displayed in Fig. 9a (75  8 nite elements were used
for each layer in this case). The dimensionless length (L ) is dened as the length of the piezoceramic layers (Lp ) divided by
the total length of the spar (L) whereas the dimensionless height (h*) is the height of one piezoceramic layer (hp) divided by
the total height of the spar (i.e., L Lp =L and h hp =h). It should be noted that the optimum electrical load that
corresponds to each data point in Fig. 9a is different (Fig. 9b).
Fig. 10 shows the top view of Fig. 9a along with the limiting solution that satises the mass restriction. It is
straightforward to show that the mass restriction implies L  h  0:027225. Therefore, the hyperbola that denes the
limiting solution in Fig. 9b is the curve L 0:027225=h . The combinations L ; h above this curve exceed the given mass
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 21

Fig. 9. Variation of the (a) maximum power output and (b) optimum load resistance with the dimensionless length and dimensionless thickness of the
piezoceramic layers.

1
0.9
Unacceptable
0.8 region
0.7
0.6
L*

Limiting solution curve

0.5 (10% mass addition)
0.4
0.3
Acceptable
0.2 region
0.1
0.05 0.1 0.15 0.2 0.25
h*

Fig. 10. Variation of the power output with the dimensionless length and dimensionless thickness of the piezoceramic layers: top view of Fig. 9a with the
limiting solution curve.

addition limit (10% of the original spar mass) whereas those below the curve satisfy the given condition. The problem is
then to nd the optimal solution on the limiting curve. Before that, it is useful to discuss the trends in Fig. 9a.
As the plane-stress elastic stiffnesses (Youngs moduli) of the aluminum alloy and PZT-5A are similar, the bending
stiffness of any conguration of the generator spar (any length and thickness) is similar to that of the original aluminum
spar (since the cross section is always the same). However, as the mass density of the piezoceramic is much larger than that
of aluminum, the mass of any conguration of the generator spar is larger than that of the original spar. Since the source of
mechanical forcing in the base excitation problem is the inertia of the body itself, the problem analyzed here is more
related to variation of inertia (in terms of amount of added mass and its distribution) than that of stiffness. Furthermore,
since one is interested in the resonance behavior, variation of mechanical damping is also important. It can be recalled from
Eq. (30) that the mechanical damping is assumed to be proportional to the stiffness and mass matrices. However, the
damping ratio in this case study is affected mainly from mass variation since the stiffness of the structure is not changed
considerably. That is, increased mass results in increased mechanical damping as the stiffness is not decreased. The
interesting trade-off in terms of the power output is between the excitation amplitude and the mechanical damping, since
mass addition increases both. Although increased excitation amplitude implies increased power, expectedly, increased
mechanical damping has the opposite effect.
As can be observed from Fig. 9a, the power output increases with increasing dimensionless thickness for any
dimensionless length. One could expect the opposite behavior of power output with increasing thickness based on the fact
that the electromechanical coupling (or piezoelectric coupling matrix in the FE model) is directly proportional to the
distance from the mid-surface of piezoceramic layer to the reference surface of the composite structure. However,
increased thickness of the piezoceramic layers also means increased inertia of the elastic body and consequently larger
mechanical forcing in the base excitation problem. Moreover the dynamic exibility of the generator spar is increased as
can be observed by analyzing the behavior of the short circuit natural frequency with dimensionless length and thickness
plotted in Fig. 11a (note that the length and thickness axes are reversed for clarity). As the case analyzed here is much more
related to mass variation than stiffness variation, natural frequencies are reduced with increasing dimensions of the
piezoceramic layers. It can be observed from Fig. 11a that the lowest natural frequency (72.8 Hz) is obtained for maximum
length and thickness whereas the highest natural frequency (109.1 Hz) is obtained for the minimum dimensions of the
piezoceramic layer. As a consequence, larger strains and larger power outputs are observed with increasing dimensionless
 ARTICLE IN PRESS
22 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

Fig. 11. Variation of the (a) short circuit resonance frequency and the (b) mechanical damping ratio with dimensionless length and thickness of the
piezoceramic layers (for the fundamental vibration mode).

Fig. 12. (a) Power output versus dimensionless length for a range of dimensionless thickness and (b) power output versus dimensionless thickness for a
range of dimensionless length (with the limiting solution).

thickness. Thus, the favorable effect of increased inertia (which constitutes the mechanical forcing in the base excitation
problem) and exibility overcomes the negative effect of reduced distance from the mid-surface of piezoceramic layer to
the reference surface and the negative effect of increased damping (Fig. 11b). It should be noted that this behavior is
amplied for larger values of dimensionless length.
Fig. 12a and b, respectively, show the side views of Fig. 9a along with the limiting solution (the curve in Fig. 10) that
gives 10% mass addition. One can therefore analyze the behavior of power output with increasing dimensionless length for
xed values of dimensionless thickness in Fig. 12a. It is useful to consider two different thickness sets separately in Fig. 12a:
the rst one is for dimensionless thicknesses from 0.02 to 0.075 and the second one is from 0.075 to 0.275. For the rst set
of data curves (thin piezoceramics) a peak value of power output is observed with increasing length. Considering the strain
distribution for a cantilevered beam, a peak value for the power versus length of piezoceramic is not expected for a xed
thickness. It turns out that, for thin piezoceramics with approximately L 40:5, the effect of increased mass and
consequently increased mechanical forcing and increased dynamic exibility (due to reduced natural frequency) cannot
overcome the negative effect of increased damping which is also due to increased mass and which reduces the amplitude of
vibration and the electrical output. The variation of damping ratio with dimensionless length and thickness can be referred
from Fig. 11b. A simple verication was performed by the authors considering damping to be proportional to stiffness only
(which is similar to assuming constant damping for every conguration since the elastic stiffnesses of piezoceramics and
aluminum are similar). In this case, expectedly, a maximum value for the electrical power is not observed with increasing
length for a xed thickness. For the set of curves with thicker piezoceramics in Fig. 12a, it is clear that the maximum value
of power output gets closer to the free end of the generator spar. For the maximum thickness considered here, the
maximum power is observed for L 1:0. Once again, the general behavior is related to the amount of mass added,
mechanical forcing term, structural dynamic exibility (in the sense of reduced natural frequency) and mechanical
damping. Increased excitation due to thicker piezoceramic overcomes the negative effect of damping on the mechanical
and consequently the electrical response amplitudes. One can conclude that the mass added in the region around the free
end of the spar has an effect similar to that of a tip mass on the general behavior of the generator spar. The limiting solution
is also given in Fig. 12b along with the curves of xed length. The power outputs increases monotonically with increasing
thickness for each length of embedded piezoceramic.
If no mass restriction was imposed to the problem, the conguration with maximum length and thickness would be the
best one in terms of energy harvesting. However, the main goal here is to optimize the power output and satisfy the mass
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 23

Table 5
Limiting solutions for the mass restriction problem.

L* h* P (mW/g2) R (kO) osc (Hz)

0.1 0.27225 154.00 410 107.36

0.15 0.18150 156.98 107 107.05
0.2 0.36166 157.69 104 106.97
0.225 0.12100 157.72 84 106.97
0.25 0.10892 157.65 70 106.98
0.275 0.98999 157.48 60 107.00
0.3 0.09075 157.23 51 107.03
0.4 0.06807 155.38 32 107.10
0.5 0.05446 151.91 23 107.04
0.6 0.04538 146.27 18 106.79
0.7 0.03890 138.10 14 106.29
0.8 0.03403 127.56 12 105.55
0.9 0.03025 115.50 10 104.56
1.0 0.02723 103.22 8.5 103.34

restriction constraint which requires L  h  0:027225. The limiting solution (L  h 0:027225) is seen in Fig. 10
from the top view and in Fig. 12a and b from the side views. Sample data points from the limiting solution are also listed in
Table 5. The zoomed views in Fig. 12 allow reading the values of L* and h* that gives the maximum power for 10% mass
addition as L 0:225 and h 0:121. That is, the optimum solution for the given design constraint is obtained when the
two piezoceramic layers are embedded at the root covering 22.5% of the total length, each having a thickness that is 12.1%
of the total thickness. From Table 4, for L 300 mm and h 12 mm, the length and the thickness of each embedded
piezoceramic layer in Fig. 8 are obtained as Lp 67.5mm and hp 1.45 mm, respectively. The electrical power per base
acceleration for the generator spar is obtained as 157.72 mW/g2.

4. Conclusions

In this paper, an electromechanical FE plate model is derived for piezoelectric energy harvesting from base excitations.
The mechanics of the plate is based on the classical (Kirchhoff) plate theory, which is appropriate for modeling of typical
piezoelectric energy harvesters since they are usually designed and manufactured as thin plates. The electromechanical FE
model is derived based on generalized Hamiltons principle for electroelastic bodies and it accounts for the presence of a
pair of conductive electrodes covering the entire piezoceramic layer. A resistive electrical load is considered in the electrical
domain. Derivations are given for predicting the coupled mechanical vibration and electrical power response of the
harvester plate due to base excitation.
The electromechanical model is rst veried against the analytical results obtained from the closed-form solution for a
unimorph cantilever under base excitation reported in the literature. The electromechanical vibration and power FRFs
obtained from the FE model are in very good agreement with the ones obtained from the analytical solution. The second
verication is presented for a bimorph cantilever with a tip mass under base excitation. It is shown that the
electromechanical FE model can successfully predict the analytical and the experimental results of the bimorph reported in
the literature.
UAVs constitute important application systems where generating electrical energy from waste vibration energy
available during the ight has an important practical value. The electrical power generated from structural vibrations of a
UAV can be used to power small electronic components of the UAV and ultimately to increase its ight time if possible.
With this motivation, an optimization problem is investigated in the last case study to design a generator wing spar with
embedded piezoceramics. Although the purpose is to obtain the maximum power output, mass addition is the greatest
penalty in aerospace applications. Thus, the wing spar is modied by embedding piezoceramic into the spar by removing
some of the original material. However, since piezoelectric ceramics are typically much heavier than aircraft materials
(such as aluminum alloys), a design constraint is imposed for the added mass (not to exceed 10% of the original spar mass).
The generator spar with embedded piezoceramics is considered as a bimorph conguration with two piezoceramic layers
in series connection. It is shown that the maximum power for 10% mass addition to the spar is obtained when the two
piezoceramic layers are embedded at the root covering 22.5% of the total length, each having a thickness that is 12.1% of the
total thickness.
In the analysis given here for the UAV application, the source of mechanical forcing is assumed to be base excitation
with no air ow to simplify the problem. The optimization problem investigated in this work becomes more realistic if an
unsteady aerodynamic model is combined with the governing equations. In such a case, the effect of the aerodynamic
damping alters the dynamic behavior of the wing during the ight. The presence of aeroelastic coupling in addition to
piezoelectric coupling results in further variation of the natural frequencies, yielding different values for the electrical
 ARTICLE IN PRESS
24 C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925

power outputs as well as for the optimum values of load resistance. Combination of the piezoelectrically coupled FE model
given here with an appropriate aeroelastic model is currently investigated by the authors.

Acknowledgments

The authors gratefully acknowledge the support of the Air Force Ofce of Scientic Research MURI under Grant number
F 9550-06-1-0326 Energy Harvesting and Storage Systems for Future Air Force Vehicles monitored by Dr. B.L. Lee. The
authors also gratefully acknowledge the support of CAPES (Brazil).

References

[1] H.A. Sodano, D.J. Inman, G. Park, A review of power harvesting from vibration using piezoelectric materials, The Shock and Vibration Digest 36 (2004)
197205.
[2] S.P. Beeby, M.J. Tudor, N.M. White, Energy harvesting vibration sources for microsystems applications, Measurement Science and Technology 13 (2006)
R175R195.
[3] S. Priya, Advances in energy harvesting using low prole piezoelectric transducers, Journal of Electroceramics 19 (2007) 167184.
[4] S.R. Anton, H.A. Sodano, A review of power harvesting using piezoelectric materials (20032006), Smart Materials and Structures 16 (2007) R1R21.
[5] K.A. Cook-Chennault, N. Thambi, A.M. Sastry, Powering MEMS portable devicesa review of non-regenerative and regenerative power supply
systems with emphasis on piezoelectric energy harvesting systems, Smart Materials and Structures 17 (2008) 043001 (33pp.).
[6] S.R. Anton, D.J. Inman, Vibration energy harvesting for unmanned air vehicles, Smart Structures and Materials 2008: Active and Passive Smart Structures
and Integrated Systems II, Proceedings of the SPIE, Vol. 6928, San Diego, CA, March 1013, 2008.
[7] K.C. Magoteaux, B. Sanders, H.A. Sodano, Investigation of energy harvesting small unmanned air vehicle, Smart Structures and Materials 2008: Active
and Passive Smart Structures and Integrated Systems II, Proceedings of the SPIE, Vol. 6928, San Diego, CA, March 1013, 2008.
[8] A. Erturk, J.M. Renno, D.J. Inman, Modeling of piezoelectric energy harvesting from an L-shaped beam-mass structure with an application to UAVs,
Journal of Intelligent Material Systems and Structures 20 (2009) 529544.
[9] S. Roundy, P.K. Wright, J. Rabaey, A study of low level vibrations as a power source for wireless sensor nodes, Computer Communications 26 (2003)
11311144.
[10] H.A. Sodano, G. Park, D.J. Inman, Estimation of electric charge output for piezoelectric energy harvesting, Strain 40 (2004) 4958.
[11] N.E. duToit, B.L. Wardle, Experimental verication of models for microfabricated piezoelectric vibration energy harvesters, AIAA Journal 45 (2007)
11261137.
[12] A. Erturk, O. Bilgen, D.J. Inman, Power generation and shunt damping performance of a single crystal lead magnesium niobatelead zirconate titanate
unimorph: analysis and experiment, Applied Physics Letters 93 (2008) 224102 (3pp.).
[13] C.B. Williams, R.B. Yates, Analysis of a micro-electric generator for microsystems, Sensors and Actuators A 52 (1996) 811.
[14] D. Arnold, Review of microscale magnetic power generation, IEEE Transactions on Magnetics 43 (2007) 39403951.
[15] P. Glynne-Jones, M.J. Tudor, S.P. Beeby, N.M. White, An electromagnetic, vibration-powered generator for intelligent sensor systems, Sensors and
Actuators A 110 (2004) 344349.
[16] S.P. Beeby, R.N. Torah, M.J. Tudor, P. Glynne-Jones, T. ODonnell, C.R. Saha, S. Roy, A micro electromagnetic generator for vibration energy harvesting,
Journal of Micro-electromechanical Engineering 17 (2007) 12571265.
[17] B.P. Manna, N.D. Sims, Energy harvesting from the nonlinear oscillations of magnetic levitation, Journal of Sound and Vibration 319 (2009) 515530.
[18] S. Roundy, P.K. Wright, J. Rabaey, Micro-electrostatic vibration-to-electricity converters, Proceedings of the ASME 2002 International Mechanical
Engineering Congress and Exposition, 2002.
[19] P. Mitcheson, P. Miao, B. Start, E. Yeatman, A. Holmes, T. Green, MEMS electrostatic micro-power generator for low frequency operation, Sensors and
Actuators A 115 (2004) 523529.
[20] N.E. duToit, B.L. Wardle, S.-G. Kim, Design considerations for MEMS-scale piezoelectric mechanical vibration energy harvesters, Integrated
Ferroelectrics 71 (2005) 121160.
[21] S.-N. Chen, G.-J. Wang, M.-C. Chien, Analytical modeling of piezoelectric vibration-induced micro power generator, Mechatronics 16 (2006) 379387.
[22] J.H. Lin, X.M. Wu, T.L. Ren, L.T. Liu, Modeling and simulation of piezoelectric MEMS energy harvesting device, Integrated Ferroelectrics 95 (2007)
128141.
[23] A. Erturk, D.J. Inman, Issues in mathematical modeling of piezoelectric energy harvesters, Smart Materials and Structures 17 (2008) 065016 (14pp.).
[24] A. Erturk, D.J. Inman, A distributed parameter electromechanical model for cantilevered piezoelectric energy harvesters, ASME Journal of Vibration
and Acoustics 130 (2008) 041002 (15pp.).
[25] A. Erturk, D.J. Inman, An experimentally validated bimorph cantilever model for piezoelectric energy harvesting from base excitations, Smart
Materials and Structures 18 (2009) 025009.
[26] N. Elvin, A. Elvin, A general equivalent circuit model for piezoelectric generators, Journal of Intelligent Material Systems and Structures 20 (2009) 39.
[27] N.G. Stephen, On energy harvesting from ambient vibration, Journal of Sound and Vibration 293 (2006) 409425.
[28] J.M. Renno, M.F. Daqaq, D.J. Inman, On the optimal energy harvesting from a vibration source, Journal of Sound and Vibration 320 (2009) 386405.
[29] H.S. Tzou, C.I. Tseng, Distributed piezoelectric sensor/actuator design for dynamic measurement/control of distributed parameter systems: a
piezoelectric nite element approach, Journal of Sound and Vibration 138 (1990) 1734.
[30] W.S. Hwang, H.C. Park, Finite element modeling of piezoelectric sensors and actuators, AIAA Journal 31 (1993) 930937.
[31] D.T. Detwiler, M.-H.H. Shen, V.B. Venkayya, Finite element analysis of laminated composite structures containing distributed piezoelectric actuators
and sensors, Finite Elements in Analysis and Design 20 (1995) 87100.
[32] S.H. Chen, Z.D. Wang, X.H. Liu, Active vibration control and suppression for intelligent structures, Journal of Sound and Vibration 200 (1997) 167177.
[33] S.Y. Wang, A nite element model for the static and dynamic analysis of a piezoelectric bimorph, International Journal of Solids and Structures 41
(2004) 40754096.
[34] G.K. Ottman, H.F. Hofmann, A.C. Bhatt, G.A. Lesieutre, Adaptive piezoelectric energy harvesting circuit for wireless remote power supply, IEEE
Transactions on Power Electronics 17 (2002) 669676.
[35] D. Guyomar, A. Badel, E. Lefeuvre, Toward energy harvesting using active materials and conversion improvement by nonlinear processing, IEEE
Transactions on Ultrasonics, Ferroelectrics and Frequency Control 52 (2005) 584595.
[36] Y.C. Shu, I.C. Lien, Analysis of power output for piezoelectric energy harvesting systems, Smart Materials and Structures 15 (2006) 14991512.
[37] M.J. Guan, W.H. Liao, On the efciencies of piezoelectric energy harvesting circuits towards storage device voltages, Smart Materials and Structures 16
(2007) 498505.
[38] S.H. Crandall, D.C. Karnopp, E.F. Kurtz Jr., D.C. Pridmore-Brown, Dynamics of Mechanical and Electromechanical Systems, McGraw-Hill, New York, 1968.
[39] N.W. Hagood, W.H. Chung, A. Von Flotow, Modelling of piezoelectric actuator dynamics for active structural control, Journal of Intelligent Material
Systems and Structures 1 (1990) 327354.
 ARTICLE IN PRESS
C. De Marqui Junior et al. / Journal of Sound and Vibration 327 (2009) 925 25

[40] IEEE Group on Sonics and Ultrasonics, IEEE Standard on Piezoelectricity, New York, 1978.
[41] K.-J. Bathe, Finite Element Procedures, Prentice-Hall, Englewood Cliffs, NJ, 1996.
[42] F.P. Beer, E.R. Johnston Jr., Mechanics of Materials, McGraw-Hill, New York, 1992.
[43] A.A. Bent, Piezoelectric Fiber Composites for Structural Actuation, PhD Thesis, Massachusetts Institute of Technology, 1994.
[44] Q.M. Wang, L.E. Cross, Constitutive equations of symmetrical triple layer piezoelectric benders, IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control 46 (1999) 13431351.
[45] /http://www.piezo.com/prodsheet1sq5A.htmlS, Accessed December 2008.
[46] /http://www.efunda.com/materials/piezo/material_dataS, Accessed December 2008.
[47] G.A. Lesieutre, G.K. Ottman, H.F. Hofmann, Damping as a result of piezoelectric energy harvesting, Journal of Sound and Vibration 269 (2004)
9911001.
[48] /http://asm.matweb.com/S, Accessed December 2008.
[49] /http://www.morganelectroceramics.com/S, Accessed December 2008.