Date post: | 02-Jun-2018 |
Category: |
Documents |
Upload: | ikhwan-muhammad |
View: | 216 times |
Download: | 0 times |
of 14
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
1/14
Optimization of a right-angle piezoelectric cantilever using auxiliary beams with different
stiffness levels for vibration energy harvesting
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2012 Smart Mater. Struct. 21 065017
(http://iopscience.iop.org/0964-1726/21/6/065017)
Download details:
IP Address: 152.14.136.96
The article was downloaded on 19/08/2013 at 12:42
Please note that terms and conditions apply.
View the table of contents for this issue, or go to thejournal homepagefor more
ome Search Collections Journals About Contact us My IOPscience
http://iopscience.iop.org/page/termshttp://iopscience.iop.org/0964-1726/21/6http://iopscience.iop.org/0964-1726http://iopscience.iop.org/http://iopscience.iop.org/searchhttp://iopscience.iop.org/collectionshttp://iopscience.iop.org/journalshttp://iopscience.iop.org/page/aboutioppublishinghttp://iopscience.iop.org/contacthttp://iopscience.iop.org/myiopsciencehttp://iopscience.iop.org/myiopsciencehttp://iopscience.iop.org/contacthttp://iopscience.iop.org/page/aboutioppublishinghttp://iopscience.iop.org/journalshttp://iopscience.iop.org/collectionshttp://iopscience.iop.org/searchhttp://iopscience.iop.org/http://iopscience.iop.org/0964-1726http://iopscience.iop.org/0964-1726/21/6http://iopscience.iop.org/page/terms8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
2/14
IOP PUBLISHING SMARTMATERIALS ANDSTRUCTURES
Smart Mater. Struct. 21 (2012) 065017 (13pp) doi:10.1088/0964-1726/21/6/065017
Optimization of a right-angle
piezoelectric cantilever using auxiliarybeams with different stiffness levels forvibration energy harvesting
Jia Wen Xu, Yong Bing Liu, Wei Wei Shao and Zhihua Feng
Department of Precision Machinery and Precision Instruments, University of Science and Technology of
China, 230026, Anhui, Peoples Republic of China
E-mail: [email protected]
Received 7 July 2011, in final form 16 January 2012
Published 22 May 2012
Online atstacks.iop.org/SMS/21/065017
Abstract
This paper presents experiments and models of a piezoelectric cantilever generator with a
right-angle structure. Analysis shows that the extended part provides a large torque to the main
beam, which can dramatically smoothen the strain distribution of the main beam. The
auxiliary beam was fabricated with half the length of the main beam. When the auxiliary beam
has a stiffness which is 0.02 times that of the main beam the piezoelectric element has a highly
uniform strain distribution; in addition, its relative utilization efficiency (RUE) is 93% at the
initial resonant frequency, whereas it is 50% for a conventional rectangular piezoelectriccantilever. The performances of three right-angle generators with auxiliary beams having
different levels of stiffness, but constant-stiffness main beams are studied. The RUE of the
piezoelements increases as the auxiliary beams stiffness decreases. A model based on the
RayleighRitz method is established to demonstrate the principle of the strain-smoothing
effect. The voltage and power outputs of the generators are measured. Finite element method
simulations are also presented, and the result fits the experiments well.
(Some figures may appear in colour only in the online journal)
1. Introduction
Wireless apparatuses and portable electronic devices are
mainly powered by conventional batteries. However, replac-
ing or recharging batteries in some situations can become
tedious and expensive. Moreover, batteries offer a limited
energy supply, which means a limited duration for the duty
cycle of the device. Thus, much effort is necessary to seek
a simple way to solve the problem. Recent developments
in low-power electrics that allow the operation of these
devices with much less power have made the application of
microgenerators possible.There exists usable kinetic energy within industrial
machines, human activities and environmental vibration. Thedevelopment of on-site generators has been emphasized
because they can transform available kinetic energy intoelectrical energy [13]. Lead zirconate titanate (PZT) energy
harvester has emerged as one of the primary methods for
harvesting energy. This is superior to other means of energy
harvesting, such as electromagnetic and electrostatic, due
to its simple structure, high electromechanical coupling
coefficient, non-reliance on extra voltage sources and
easy integration into a system [46]. In PZT-based
energy harvesting, different mechanical structures have
been presented to transform vibration energy into electric
energy [715]. Among these structures, a cantilever beam
with a piezoelectric element is a common configuration
because it has a relatively low working frequency and highaverage strain.
10964-1726/12/065017+13$33.00 c2012 IOP Publishing Ltd Printed in the UK & the USA
http://dx.doi.org/10.1088/0964-1726/21/6/065017mailto:[email protected]://stacks.iop.org/SMS/21/065017http://stacks.iop.org/SMS/21/065017mailto:[email protected]://dx.doi.org/10.1088/0964-1726/21/6/0650178/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
3/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
However, the conventional rectangular cantilever has
many disadvantages due to its inherent attributes. The
piezoelectric element could not be fully utilized. Efficiency
is considerably low because its longitudinal strain distribution
changes linearly from the clamped end to the free end; and
it is zero at the free end [2]. With the aim of obtaining the
highest output power per unit of PZT element, many novelstructures have been proposed, and they are usually optimized
in the shapes of the cantilever. Roundy et al discussed
one optimal structure with uniform strain distribution, in
which a trapezoidal structure was considered. For every
unit volume of PZT, the trapezoidal cantilever can supply
twice the energy output of a rectangular structure [16].
Frank further analyzed and tested this concept. He proved
that, compared with the conventional cantilever, 30% more
power can be obtained from the trapezoidal cantilever at a
resonant frequency of under 2.5 g amplitude by a vibration
source [17]. Wood et al, Benasciutti et al and Dietland et alalso discussed the advantages of a trapezoidal cantilever over
a rectangular one [1820]. A method using a tapered beamwas introduced by Mehraeen [21]. He proved that this type
of device could also achieve a uniform longitudinal strain
distribution in the piezoelectric element. Zhao also proposed
a type of stair-shaped beam that had twice the efficiency
of a conventional cantilever in a quasi-stationary state [22].
Concurrently, Paquin and Friswell analyzed the performance
of the tapered beam model [23,24].
A conventional rectangular cantilever would also have
uniform strain distribution if it was excited by a torque
applied at its free end in the quasi-stationary state. In the
current study, a right-angle cantilever is reported, and this
device has an extended part forming a right angle with themain beam. At the series resonant frequency of the main
beam and the extended part, the main beam works in a
quasi-stationary state. Moreover, the auxiliary beam could
provide a considerably large shaking moment directly to the
main beam. Thus, a highly uniform strain distribution of the
main beam could be obtained and the performance of the
energy harvester could be considerably improved.This paper proceeds as follows. First, a simple model
based on the RayleighRitz method for the piezoelectric
compound structure is proposed and used to demonstrate
the principle of the strain-smoothing effect. Next, finite
element method (FEM) simulation is presented to examine the
strain distributions of three devices at their respective initialresonant frequencies. The three devices are made of auxiliary
beams of different stiffness and main beams with constant
stiffness. Experiments are carried out to study the strain
distributions and relative utilization efficiency (RUE) of the
three devices. Finally, their voltage output and power output
under sinusoidal acceleration of 2.5 m s2 are examined and
compared.
2. Prototype structure
Figure1 is a schematic diagram of the right-angle cantilever.
As can be seen, the device has an auxiliary beam in the
orthogonal direction of the main beam, which is regardedas a cantilever fixed to a basement. The auxiliary beam has
Figure 1. Schematic diagram of right-angle cantilever.
one end fixed to the free end of the main beam and the
other end installed with a proof mass. The main beam and
the piezoelectric layer are geometrically uniform in their
longitudinal directions, and the poling axis of the piezoelectric
element is perpendicular to the main beam.
The sinusoidal base movement of constant amplitude
is orthogonally decomposed into Uxx-direction vibration
and Uzz-direction vibration (figure 1). The x-direction is
perpendicular to the main beam, whereas the z-direction is
parallel to it. In this case, Ux and Uz have the same phases,
but they have different amplitudes. They obey the following
conditions:
Ux(t)= U(t) cos, (1)
Uz(t)= U(t) sin. (2)
3. System modeling
3.1. Mathematical model
Models of piezoelectric cantilevers based on the Euler
Bernoulli beam theory have been studied by many
researchers [2530], and these distributed parameter models
demonstrate great accuracy. One example is the L-shaped
piezoelectric cantilever model presented by Erturk and
Inman[31]. In this section, a mathematical model based on
the EulerBernoulli beam theory is also established to analyze
the right-angle cantilever proposed in section2.
The following study is based on the assumptions: (1) theprototype is considered as a composite EulerBernoulli beam;
therefore, the effects of shear deformation and rotary inertia of
the beams are neglected; (2) the PZT layer is perfectly bonded
to the substructure layer; thus, there is no shear strain between
the layers; and (3) no deformation is presented at the joint of
the main beam and the auxiliary beam.
In figure2(a),l1, bm, l2, ba,Mp, hp, hb, ha, p, b andarepresent the following parameters, respectively: length and
width of the main beam, length and width of the auxiliary
beam, proof mass, height of the PZT and titanium substructure
of the main beam, height of the auxiliary beam, mass density
of the substructure, the PZT plate and the auxiliary beam.
The cross-section of the composite beam is shown infigure 3. The coordinate system XOZ is set up with the
2
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
4/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Figure 2. (a) Schematic drawing of the right-angle cantilever, (b) equivalent model of the main beam and the auxiliary beam.
Figure 3. (a) Cross-section of the main beam for the determinationof the neutral plane position. (b) Cross-section of the auxiliary
beam.
origin at the neutral axis NN, and the distance between
the bonding layer and the neutral axis is a. Here, a is
determined by calculating the stress in the cross-section of the
unimorph [28]. Thus,
a=1
2
Ebh2b Eph
2p
Ebhb+ Ephp. (3)
Considering the composite layer of the PZT and the base beam
as one beam with flexural rigidityEIeq = 13 bEb(h
3b 3ah
2b+
3a2hb)+13 bEp(h3p + 3ah2p + 3a2hp),EIarepresents the rigidityof the auxiliary beam given byEIa =
112 baEah
3a .
w1(zm, t) and w2 = (xa, t) denote the transverse
deflections of the main beam and the auxiliary beam,
respectively, at locations x,z and time t in the coordinate
systemXOZlocated at the base. Equations of motion for the
system are derived by extended Hamiltons principles. In this
model, the main beam is considered as a fixed-free cantilever;
the auxiliary beam is considered as a hinged-free cantilever,
as shown in figure2(b).
The kinetic and potential energy of the system under base
excitations ofUxand Uz are respectively expressed as
T =1
2
l1
0m1
w1(zm, t)
t+
Ux
t
2
dzm
+1
2
l10
m1
Uz
t
2dzm
+1
2
l20
m2
Uz
t+
w2(xa, t)
t
2dxa
+1
2Mp
Uz
t+
w2(l2, t)
t
2
+1
2(m2l2+ Mp)
w1(zm, t)
t+ Ux
2
, (4)
V =1
2
l10
EIeq
2w1(zm, t)
z2m
2dzm
+1
2
l20
EIa
2w2(xa, t)
x2a
2dxa, (5)
wherem1and m2are the mass densities per unit length for the
composite beam, given respectively as
m1 = (php+ bhb)bm,
m2 = abaha.
If the viscous damping coefficient for the main beam isdenoted as c1 and the viscous damping coefficient for the
3
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
5/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
auxiliary beam is denoted as c2, then the virtual work is
Wc =
l10
c1w1(zm, t)
tw dzm
l20
c2w2(xa, t)
tw dxa, (6)
where w is the virtual displacement. Using extendedHamiltons principles, we arrive at t2
t1
(T V+ Wc) dt= 0. (7)
Thus, the motion equation for the vibration of the main beamand the auxiliary beam, including the effect of damping, canbe written respectively as
2w21(zm, t)
z2m+c1
w1(zm, t)
t+EIeq
4w1(zm, t)
z4m=0,
for 0 zm l1, (8)
2w22(xa, t)
x2a+c2
w2(xa, t)
t+EIa
4w2(xa, t)
x4a=0,
for 0 xa l2. (9)
Modes of the undamped free vibration are typically usedto estimate the modes of the devices at resonance frequencies.In the case of undamped free vibration, Uz = Ux = 0, andthere is no displacement or rotation at the clamped end of themain beam relative to the base; thus, the boundary conditionsare as follows:
w1(0, t)=w1(zm, t)
zm
zm=0
=0, (10)
w2(0, t)= 0,w1(zm, t)
zm
zm=l1
=w2(xa,t)
xa
xa=0
, (11)
(m2l2+Mp)2w1(zm, t)
t2
zm=l1
EIeq3w1(zm, t)
z3m
zm=l1
=0,
(12)
EIeq2w1(zm, t)
z2m
zm=l1
= EIa2w2(xa, t)
x2a
xa=0
, (13)
EIa3w2(xa, t)
x3a
xa=l2
Mp2w2(xa, t)
t2
xa=l2
=0, (14)
EIa2w2(xa, t)
x2a
xa=l2
=0. (15)
The vibration response relative to the base of the levercan be represented as an absolutely convergent series of theeigenfunctionsmr(zm)andar(xz). Thus,
w1(zm, t)=
r=1
mr(zm)rx(t), (16)
w2(xa, t)=
r=1
ar(xa)rz(t), (17)
where mr(zm) and ar(xz) denote the mass normalizedeigenfunction of the rth vibration mode of the main beam
and the auxiliary beam, respectively. Here, the influence of
gravity on the shapes of the beams is ignored, because gravity
contributes little to the sinusoidal voltage output; r(t) is the
modal mechanical response expression of the device of the
rth vibration mode, and it is the same for both sections of the
device. Eigenfunctionsmr(zm)andar(xz)for the main beam
and the auxiliary beam are denoted respectively asmr(zm)= Amrsin rzm+ Bmrcos rzm
+ Cmrsinhrzm+ Dmrcoshrzm, (18)
ar(xa)= Aarsin rxa+ Barcos rxa
+ Carsinhrxa+Darcoshrxa, (19)
where 4r = 2r
m1EIeq
, 4r = 2r
m2EIa
and r is the undamped
resonance frequency of the rth vibration mode in open-circuit
conditions. In addition, Amr, Bmr, Cmr, Dmr, Aar, Bar, CarandDarshould be evaluated by the aforementioned boundary
conditions. Thus,
Amr= Cmr, Bmr= Dmr, Bar= Dar.
Eigenfunctionsmr(zm)andar(xa)can be simplified as
mr(zm)= Bmr(cosrzmcosh rzm)
+Amr(sinrzmsinhrzm), for 0 zm l1, (20)
ar(xa)= Aarsin rxa+ Barcos rxa+ Carsinhrxa
Barcosh rxa, for 0 xa l2. (21)
The eigenfunctions can be further simplified. The Maclaurin
series expansions of the two expressions are O(z3) of the
second term. The derivation of the first two terms of the
expansion from cosrzm, cosh rzm, sin rzm and sinhrzm
is the general solution of the undamped system, which fits thegeneral boundary condition. Thus, the eigenfunctionsmr(zm)
at the first resonance frequency can be written as
m1(zm)= A1z3m+B1z
2m+ C1zm+ D1,
for 0 zm l1. (22)
In addition,ar(xa)also has a similar simplified form
a1(xa)= A2x3a + B2x
2a + C2xa+ D2,
for 0 xa l2. (23)
Hence,C1 = D1 = D2 =0, and
B2 = 3A2l2,(m2l2+ Mp)(A1l
31+ B1l
21)EIaA2
=Mp(A2l32+B2l
22+ C2l2)EIeqA1,
C2 =3A1l21+2B1l1,
EIaB2 =EIeq(3A1l1+ B1).
The eigenfunctions have one degree freedom, which can
be evaluated by normalizing them according to the following
orthogonality conditions: l10
m1mi(zm)mj(zm) dzm+
l20
m1ai(ya)aj(ya) dya
+ (Mp+ m2l2)mi(l1)mj(l1)+Mpai(l2)aj(l2)=i,j, (24)
4
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
6/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Figure 4. Theoretical diagram ofk= EIaEIeq
versusar.
and l10
EIeq
mi(zm)
mj(zm) dzm
+
l20
EIa
ai(ya)
aj(ya) dyaEIami(l1)
aj(0)
+EIeqmi(0)
aj(0)= i,j
2r. (25)
Here, i,j is the Kronecker delta defined as unity when i = jand zero wheni =j.
The PZT element is bonded to the main beam and works
under the initial resonance frequency; thus, the eigenfunction
m1(zm) should be further studied. This expression can be
described as:
m1(zm)= B1(z2m+ arz
3m), ar=
A1
B1,
for 0 zm l1. (26)
The mathematical expression of the stress distribution of themain beam is:
Sm1(zm)= h2m1(zm)
z2m=2hB1(1+3arzm),
for 0 zm l1. (27)
From the expression, the coefficient arevidently determines
how the strain distribution has been smoothed at the firstresonance frequency. Notably, the expression of the strain
distribution for a conventional cantilever (with the same
parameters as the main beam in this experiment) is also S(x)=
2hB1(1 + 3arx), where a
r =
13l1
. Using the parameters
presented in section5.1, the a rfor a conventional cantilever
is 11.1. The strain distribution is a function of the ratio of
EIa andEIeq, a smaller |ar|indicates a more smoothed stress
distribution. The relation betweenk= EIaEIeq
andaris presented
in figure4.Figure 4shows that a smaller kdenotes a smaller |ar|,
thereby indicating a more smoothed strain distribution. The
physical meaning of this strain-smoothing effect will beexplained in section3.2.Remarkably, this expression of strain
distribution only applies at the first resonant frequency of
the device, with the assumption that it is in a low force
environment. In addition, the strain distribution may change
at other frequencies, as Erturk has described in a previous
study [29].
3.2. Force and torque applied to the main beam
The study of the transient dynamic characteristic of a PZT
cantilever utilizing the impedance method has been performed
in previous studies, and the proposed models have shown
fairly good accuracy [19, 32, 33]. The impedance method
has been studied and implemented in the present research
to get the corresponding displacement of the device under
a generalized force, as well as the interaction between the
main beam and the auxiliary beam. The forcedisplacement
impedance is used to study the models because the electric
charge generated is proportional to the amplitude of the
strain of the device. The forcedisplacement impedance and
admittance are defined as follows [34]:
Z(j)=F(j)
X(j), (28)
H(j)=1
Z(j). (29)
The model in section 2.1 analyzes the dynamic response
of the device in the following discussions. When the device
is solely excited by a vibration in the z-direction, the base
movement first excites the auxiliary beam, which then drives
the main beam. When the device is solely excited by
x-direction vibration, the excitation of the auxiliary beamdepends on its own inertial force. However, the auxiliary
beam would also provide the main beam with a large torque.
Considering the energy dissipation in the beammass system,
a viscous damping factor is added. The modal damping ratio
is identified directly from the frequency response or time
domain measurement.
Figure 5(a) shows the model of impedance of the
right-angle cantilever; figure5(b) is the equivalent circuit of
the impedance of the device [34]. Here, [M1] is a matrix of
the generalized mass of the main beam, [K1] is a matrix of
the generalized stiffness coefficient of the main beam, [c1]is
a matrix of the generalized damping coefficient of the mainbeam,[Mp] is a matrix of the generalized mass of the proof
mass,[M2]is a matrix of the generalized mass of the auxiliary
beam,[K2] is a matrix of generalized stiffness coefficient of
the auxiliary beam and [c2] is a matrix of the generalized
damping coefficient of the auxiliary beam in this device.
[c2] can be ignored in comparison to [c1]. The mass of the
auxiliary beam is much smaller than the proof mass, thus, it
is also ignored in the analysis. {F} is the generalized force
applied at the device. There exist both force and torque at
the joint point of the main and auxiliary beams; hence, the
generalized force is defined as follows:
{F} =
FxFz
= Mp
Ux(t)Uz(t)
. (30)
5
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
7/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Figure 5. (a) Schematic model of the impedance of the right-angle cantilever, (b) equivalent circuit of impedance of the device.
The response of the device {} =xz
obeys the
following equation:[Z0] {} = {F}, where[Z0] = 2[Mp]+
[Z1][Z2][Z1]+[Z2]
[34].
The displacement impedance of the main beam and
auxiliary beam can be calculated using the following
principles[34]:
[Z1] = [K1] +j[c1] 2[M1], (31)
[Z2] = [K2] +j[c2] 2[M2]. (32)
To calculate the generalized stiffness coefficients and
generalized mass of the device, the following model is used:
[K1] =
l10
EIeq(m(z))
2 dz
l10
EIeq(m(z))
2
m(z)dz
l10
EIeq
(m(z))2
m(z)d l1
0EI
eqm(z)
m(z)
2
dz
,
(33)
[M1] =
l1
0m1
2m(z) dz
l10
m12m(z)
m(z)dz
l10
m12m(z)
m(z)dz
l10
m1
m(z)
m(z)
2dz
,
(34)
[K2] =
l2
0EI2
a (x)
a(x)2
dx
l2
0EI2(
a (x))
2a(x) dx
l2
0EI2(
a (x))
2a(x) dx
l2
0EI2(
a (x))
2 dx ,(35)
[M2] =
m2l2 0
0
l20
m22a(x) dx
. (36)
Regarding the device as a series-connected massspring
damper system, the transformation of force and torque in this
system can be calculated using the equivalent circuit of the
impedance in figure5(b). The system works at the resonant
frequency and the mode is defined, thus, regardless of theangle,xcan be expressed byz. The system has one degree
6
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
8/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
of freedom, but has a different displacement impedance in
different directions.
The proof mass, the main beam and the auxiliary beam
have the same generalized displacement at the joint point at
the free end of the auxiliary beam (point a in figure 5). As
such, the impedance of the proof mass and the impedance of
the main and auxiliary beams have the following condition:
[F]
2[Mp]=
[F][Z1][Z2]
[Z1]+[Z2]
, (37)
where [F
] is the force applied to the proof mass, 2[Mp]
is the impedance of the proof mass, [F]is the force applied
to the main beam and the auxiliary beam, and [Z1][Z2][Z1]+[Z2]
is the
impedance of the main beam and the auxiliary beam. Drawn
from figure 5(b), the {F} has infinite impedance; thus, the
force in the x-direction applied at the free end of the main
beam is:
Fx = (m2l2+ Mp)m1(l1)(2)
{F}
[Z0]
T
1
0
+
{F}
[Z0]
[Z1] [Z2]
[Z1] + [Z2]
T 10
. (38)
The torque Mthat is applied to the free end of the main
beam also excites the main beam; here,Mis:
M= Mpl2a1(l2)(2)
{F}
[Z0]
T01
+ {F}T
0
1
Mpl2.
(39)
The expressions for the force and torque show that they
result from two sources: the induced acceleration of the
motion of the proof mass and the inertial acceleration of the
base movement.
At low frequency, the impedance of the device is high;
thus, the induced acceleration of the motion of the device
is extremely low. In this case, the inertial force {F} applied
to the proof mass transfers the force ( {F}[Z0]
[Z1][Z2][Z1]+[Z2]
)T
10
,
as well as generates the torque {F}T
01
Mpl2 directly at
the main beam. However, at the initial resonant frequency,
the impedance of the device is extremely low, and {F}has a phase /2, which is different from the motion of
the device. Thus, the induced inertial force of the proof
mass generates the force (m2l2+Mp)m1(l1)(2)(
{F}[Z0]
)T
10
and torque Mpl2a1(l2)(
2)({F}[Z0]
)T
01
applied to the main
beam. Furthermore, the applied force and torque determine
the modes of the main beam. At the initial resonant frequency,
the force applied to the free end of the main beam Fxis
Fx = (m2l2+ Mp)m1(l1)(2)
{F}
[Z0]
T 10
. (40)
Here,Fx excites the main beam, and the torque Mapplied tothe free end of the main beam also plays the role of activating
the main beam and is given by
M= Mpl2a1(l2)(2)
{F}
[Z0]
T 01
. (41)
As the low rigidity auxiliary beam lowers the series
resonant frequency of the device and makes the main beam
work in a quasi-stationary state, the gradient of the strain
distribution of the main beam is mainly caused by the force
in the x-direction applied at the free end of the main beam.
On the other hand, the torque applied at the free end of the
main beam smooths out the strain distribution. Thus, the strain
distribution of the main beam is determined by the ratio of the
maximum torque generated by the force in thex-direction and
the torque directly applied at the free end of the main beam.
The ratio can be expressed as
Fxl1
M =
(m2l2+ Mp)m1(l1)(2)(
{F}[Z0]
)T
1
0
l1
Mpl2a1(l2)(2)({F}[Z0]
)T
0
1
m1(l1)l1
a1(l2)l2. (42)
Here,m1(l1)anda1(l2)can be simply treated as the relative
amplitude of the proof mass in the x-direction andz-direction.
As the rigidity of the auxiliary beam is reduced, the free end of
the auxiliary beam would inevitably have much larger motion
in the z-direction than in the x-direction. Hence, the induced
acceleration of the proof mass would be much larger in the
z-direction than in the x-direction. In addition, the torque
directly applied at the main beam would be much larger.Thus, the strain distribution of the main beam would also
be smoothed. The auxiliary beam with proof mass acts as an
absorber which transforms the x-direction movement into the
z-direction movement and provides a large torque at the free
end of the main beam. Thus, the strain distribution of main
beam is smoothed.
3.3. Voltage output
In this section, the voltage output across a resistance load is
studied. The constitutive equations of the strips are given by:
S1 = sE11T1+d31E3, (43)
D3 = d31T1+ T33E3, (44)
whereS1 andT1 are the strain and stress along the direction
of the piezoelectric strips, respectively; D3 and E3 are the
electrical displacement and the electric field, respectively; and
d31 and 33 are the compliance values at constant electric
field, that is, the piezoelectric strain constant and the dielectric
constant under constant stress, respectively.
Assuming the leakage resistance of the PZT is much
higher than the load resistance, which is usually insignificant
in an electrical circuit, the electrical part of the model issimplified as a resistive load R between two electrodes of the
7
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
9/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Table 1. Amplitude of the proof mass in the z-direction andx-direction. (Note: unit: mm.)
Device
Amplitude (excited byUz) Amplitude (excited byUx)
|w2(l2, t)| |w1(l1, t)| |w2(l2, t)| |w1(l1, t)|
Model I 10.21 0.54 0.91 0.05Model II 3.16 0.43 1.52 0.19Model III 1.9 0.57 0.91 0.27
Figure 6. Simplified circuit representation.
piezoelectric disc. The electric current output can be written
as follows:
d
dt
A
D n dA
=
v(t)
R, (45)
where
D is the vector of the electric displacement
components in the piezoelectric layer, and n is the unit
outward normal. As shown in figure 6, the relation betweenthe charge on the electrodes q and the voltage across the
piezoelectric discV(t)is obtained as follows:
1
Rx+
s33bl1
hp
dV
dt= e31b
hp
2 +a
l10
3w
z2t. (46)
Assuming thatV,D3and S1vary as ejt, then
V=e31b(
hp2 +a)
l10
3wz2t
1Rx
+jCpejt
= e31b(
hp2 +a)x(t)
l10 Sm(z) dz
1Rx
+jCpejt. (47)
This equation indicates that the piezoelectric elements
voltage output is proportional to the strain integral of the
piezoelectric element when the other parameters are kept
constant. Assuming that the strain distribution in the length
direction is equal to the strain distribution on the surface of
the beam, it can be concluded that the electrical energy stored
in the piezoelectric element is proportional to the square of
the strain integral. A smoothed-strain cantilever can produce
much more energy than an unsmoothed-strain cantilever undera strain limitation at the same resonant frequency.
4. Simulation and discussion
In this section, the strain distribution of three prototypes is
examined by simulation and experiment at the first resonant
frequencies of the devices, where they are most effective. The
three prototypes have auxiliary beams with different stiffness
levels. In this experiment, the length of the auxiliary beam is
half that of the main beam: thus, ki = EIaEIeq
is used to evaluate
the relative stiffness of the auxiliary beams in comparison with
the main beam.
Model I has ki = 0.02, which means that the stiffness ofthe auxiliary beam is much lower than that of the main beam.
Model II haski = 0.25.
Model III has ki = 3, which means that the stiffness of
the auxiliary beam is higher than that of the main beam.
The section5presents the parameter settings in detail.
As discussed in section 3.3, the ratio of the maximum
torque generated by the force in the x-direction and the
torque directly applied at the free end of the main beam is
mainly determined by the amplitude of the proof mass in the
x-direction and z-direction. The amplitude of the proof mass
in the x-direction andz-direction for the three devices, when
they are separately excited by thez-direction vibration and thex-direction vibration, are shown in table1.
Table 1 was obtained at damped free vibration at the
respective initial frequencies of each device using the FEM
method under a sinusoidal acceleration of 2.5 m s2. As the
rigidity of the auxiliary beam decreases, the amplitude of the
motion in the z-direction becomes much larger than that in
thex-direction. Thus, the torque applied at the free end of the
main beam is larger than the maximum torque generated by
the force in the x-direction, and the strain distribution of the
main beam is smoothed. For the low-stiffness Model I, the
stimulated resonance frequency is 22.1 Hz. The stimulated
resonance frequencies for Models II and III are 36 and
47.5 Hz, respectively.
As the three systems have different resonance frequen-
cies, the strain distribution at each resonance frequency
is studied by the FEM method; for better comparison,
normalized differences of strain on the surface of the PZT
element are shown with experimental data in section5.3.
5. Experiments and discussion
5.1. Experimental setup
In the experiment, right-angle cantilevers were fabricated
with the following dimensions: l1 = 30 mm, l2 = 15 mm,b = 20 mm, Mp = 53.4 g, hp = 0.4 mm, hb = 0.8 mm,
8
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
10/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Table 2. Vibration amplitude of the proof mass in the z-direction andx-direction. (Note: unit: mm.)
Device
Amplitude (excited byUz) Amplitude (excited byUx)
|w2(l2, t)| |w1(l1, t)| |w2(l2, t)| |w1(l1, t)|
Model I 11.5 0.62 0.95 0.05Model II 2.95 0.35 1.61 0.17Model III 1.6 0.52 1.01 0.31
Figure 7. Photograph of the experimental system.
p = 7.5 g c m3, b = 4.51 g cm
3, a = 7.8 g c m3,
Eb =109 GPa andEp =106 GPa. The experimental setup is
shown in figure7.First, the piezoelectric element was bonded
to the titanium (TA1) substructure layer. The ceramic material,
PZT-5, was purchased from Jiaye Company, China. It had a
charge coefficient of d31 = 171 pC N1. The brass proofmass, of size 20 mm15 mm20 mm, weighs 53.4 g. The
PZT and polished sheets were bonded using DP460 (epoxy
glue) and cured for 2 h at 70 C. This procedure was repeated
when the unimorph was bonded to the orthogonal steel lever.
The three auxiliary beams of different stiffness for
Models I, II and III were made of steel sheet. These auxiliary
beams were also of different thicknesses. The thickness
for Model I was 0.0003 m, with an estimated stiffness of
0.0049 N m2, less than that of the main beam, which was
0.243 N m2. The thickness for Model II was 0.0007 m, with
an estimated stiffness of 0.062 N m2, and that for Model III
was 0.0016 m, with an estimated stiffness of 0.74 N m2.
5.2. Vibration amplitude and strain distribution
In this experiment, vibration amplitudes of the proof mass in
the z-direction and x-direction were measured using a laser
sensor; strain distributions on the surfaces of the PZT sheets
during vibration were obtained by strain gages. The resistance
of the strain gages was 120 1 , with an active area
of 1.0 mm 1.0 mm. We used models with seven gages
along the mid-lines of the PZT element to measure the strain
distribution and models, with just one gage at the end of the
PZT element to measure the voltage output.
Table2 shows the measured vibration amplitudes of theproof mass in the z-direction and x-direction using a laser
Figure 8. Stimulated and measured strain distributions of the threemodels.
sensor under a sinusoidal acceleration of 2.5 m s2 at each
resonant frequency, which agree well with the stimulation
results.
Figure 8 shows the simulated and measured straindistributions on the surfaces of the PZT elements of three
models at each resonant frequency.
The simulated data are obtained for undamped free
vibration at the initial frequencies of each device, respectively.
It is apparent that a highly uniform strain distribution for the
low-stiffness model exists at its resonance frequency. A peak
strain exists at the clamped end of the main beam, due to
the boundary condition at the clamped end. As the stiffness
of the extended beam increases, the resonance frequency of
the model increases. The gradient of the strain distributions
of Models II and II show a increasing trend, compared with
the low-stiffness model. However, adding a dampener wouldslightly smoothen the strain distribution.
Models with seven gages along the mid-line of the PZT
element are used to measure the strain distribution at resonant
frequencies; the data are also shown in figure 8 and fit
the stimulated result well. The strain distribution is mainly
determined by the ratio of the maximum torque generated by
the force in the x-direction and the torque directly applied at
the free end of the main beam. When the auxiliary beam has
much lower stiffness, the auxiliary beam provides extremely
large torque, and the main beam has a highly uniform strain
distribution. As the stiffness of the extended beam increases,
the gradient of the strain distribution also increases. This isconfirmed by the strain distribution of Models II and III.
9
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
11/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Figure 9. RUE of the PZT element versus frequency.
5.3. RUE of the PZT element
Normally, a piezoelectric energy harvester (unimorph or
biomorph) works at its initial resonant frequency, and the
highest strain in the longitude direction occurs at the clamped
end. From the discussion in section3.3,the charges resulting
from mechanical strain promote the integration of the strain
distribution in longitude, when other parameters are kept
constant. The highest utilization efficiency of a PZT element
can be achieved only when the piezoelectric material has
a uniform strain distribution. In following discussion, a
standard RUE of a PZT element is set for evaluation; thestandard is based on the assumption that the longitudinal strain
distribution of a PZT element is absolutely uniform. Here,
the rectangular PZT element with constant width and height
is studied. In addition, the PZT element used in the current
study is covered with continuum electrodes without any
segmentation, and the influence of the electrode is ignored.
The RUE of the right-angle piezoelectric cantilevers can
be defined as follows:
=
l1
0 (zm) dzm
l10 |(zm)|maxdzm
. (48)
Based on FEM stimulation results, the frequency range is
much lower than the second resonant frequency, and the RUE
of a conventional rectangular piezoelectric cantilever with a
proof mass is 50%. Erturks study also supported the RUE of
a conventional cantilever [29].
Figure9shows the RUE at different frequencies for the
three models, excited by either the z-direction vibration or
the x-direction vibration, respectively. The simulations were
carried out in the open-circuit condition. Figure 9also shows
the value of the RUE of a conventional composite cantilever.
Model I. When the device is excited by the z-direction
vibration, the RUE of the PZT element approximates 97% in a
quasi-static activation. This is due to a torque directly appliedat the free end of the main beam, as discussed in section3.3.
The RUE of the PZT element decreases as the frequency
increases. When the device is excited by the x-direction
vibration, the RUE of the PZT element approximates that of a
rectangular conventional composite cantilever in a quasi-static
activation. This is because the auxiliary beam provides an
inertial force to the main beam, mainly because the device
works in a quasi-static state. The RUE increases as thefrequency increases. The two curves excited by the z-direction
vibration and x-direction vibration intersect at the initial
resonant frequency of the device, which is 22.1 Hz, where
the efficiency approximates 93%. The intersection of the two
curves agrees well with the EulerBernoulli beam theory.
Figure 9 also indicates that, when the device is excited by
the x-direction vibration, the frequency is higher than the
resonance frequency. The RUE of the PZT element has a
peak, due to the transition of shapes between different modes.
However, the device could not be used at these frequencies
because they are not the resonant frequencies, and the device
cannot be driven efficiently.
Model II. When the device is excited by the z-directionvibration, the RUE of the PZT element is equal to that of
Model I in a quasi-static state. The RUE of the PZT element
decreases as frequency increases at a faster rate. When the
device is excited by the x-direction vibration, the RUE of the
PZT element also equates to that of Model I in a quasi-static
condition. The RUE increases as the frequency increases. The
two curves excited by thez-direction vibration andx-direction
vibration intersect at the initial resonant frequency of the
device, at 36 Hz, where the efficiency approximates 87%. The
RUE of the PZT element also has a peak that is higher than the
initial resonance frequency when excited by the x-direction
vibration.Model III. The RUE of the PZT element approximates
the same, although a little lower than that of Model II when
the device is excited by the z-direction vibration. The RUE
of the PZT element shows a slower increase trend, compared
with that of Model II when the device is excited by the
x-direction vibration. The two curves of the device excited by
thez-direction vibration and x-direction vibration intersect at
the initial resonant frequency of the device, at 47.5 Hz, where
the efficiency approximates 80.5%.
Figure 9 indicates that the RUE of the PZT element
can be very high in a model that has a low stiffness of
the auxiliary beam. The increasing stiffness of the auxiliary
beam may be lower than the RUE at the resonant frequency
of the device. The experiments agree with the stimulation
well. Moreover, figure9 indicates the relative capacity of the
power output of the PZT element. The capacity of the power
output of a PZT element is proportional to the square of the
integration of the mechanical strain distribution in longitude at
a certain frequency. A high RUE of the PZT element indicates
a high capacity for power output at a certain frequency with a
determined strain limitation.
5.4. Voltage output
Given that a PZT plate with more uniform strain distributioncan undoubtedly have a greater potential voltage output,
10
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
12/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Figure 10. Strain limitation versus voltage output.
relationships between the output voltage and the maximumstrain of each system were tested and compared in this work.
The mode of the device at the initial resonant frequency and
the voltage output under the strain limit were also defined,
regardless of the directions of the vibration. Thus, the voltage
outputs under a strain limit were measured at each initial
resonance frequency, respectively. The devices were excited
by thez-direction vibration, and the strain gages were bonded
to the surfaces at the clamped ends of the main beams.
This comparison can sufficiently demonstrate the difference
between them.
Figure 10 shows the respective peak-to-peak voltages
generated by the PZT plates at the initial frequencies ofthe devices, with the maximum strain varying from 10 to
26 . Model I shows a maximum voltage output which
is 1.1 times more than that of Model II under a certain
strain limitation, and 1.2 times more than that of Model III.
Notably, the power output of the devices is much larger than
that of a conventional device with the same strain limitation.
In an earlier work, Baker described the advantages of a
strain-smoothed cantilever[16].
Furthermore, the performances of the devices in two
special cases were tested; they were excited by the x-direction
vibration andz-direction vibration.
The peak-to-peak voltage outputs of the three models
shown in figure 11 were obtained under a sinusoidal
acceleration of 2.5 m s2 exerted on the base. Model I reached
its open-circuit peak-to-peak voltage output of 602.7 V at
22.4 Hz, which was excited by the z-direction vibration. When
the cantilever was excited by the x-direction vibration, the
open-circuit voltage output reached its peak of 49.7 V at
22.6 Hz. The voltage output under the z-direction activation
was 12 times that under the x-direction activation. This was
due to the forcedisplacement impedance of the device in
the z-direction, which was lower than that in the x-direction.
The device exercised a larger distance in the direction of the
base movement as well as absorbed and transformed more
energy. The results agree well with the analysis presented insection3.3.
Figure 11. Open-circle voltage versus vibration frequency.
Additionally, Model II reached its open-circuit peak-to-
peak voltage output of 468.6 V at 36.5 Hz, when excited
by the z-direction vibration; when excited by the x-direction
vibration, the voltage reached a peak of 203.2 V at 36.7 Hz.
The device generated 2.3 times more voltage output when
excited by the z-direction vibration than when excited by the
x-direction vibration.
Model III was excited by the z-direction vibration. The
open-circuit peak-to-peak voltage output reached its peak of
326.4 V at 48 Hz. When excited by the x-direction vibration,
the voltage reached a peak of 272.4 V at 48.3 Hz.
Moreover, the frequencies at the peak of the voltage
output were slightly different when the device was excited
by the x-direction vibration and z-direction vibration. The
frequency at the peak of the voltage output was slightly lower
when the device was excited byz-direction vibration than that
by thex-direction vibration. This is due to the effect of gravity
and rotation of the base.
5.5. Power output
This section studies the power output of the three models. To
investigate the power output of the right-angle cantilever, aresistive load is added to the output of the piezoelectric plate.
Only the frequency response around the first natural frequency
was studied. The generator was excited with a sinusoidal
vibration of 2.5 m s2.
Figure12shows the output voltage versus load resistance
and output power versus load resistance, in which the best
match load is a function of frequency. Hence, although the
piezoelectric elements of the three devices have the same size,
their best match loads are different. Best fit resistance loads
can be chosen from figure 12for the devices, namely, 670,
340 and 290 k for Models I, II and III, respectively. The
resulting peak-to-peak voltage Upp across the load resistancewas measured by an oscilloscope. The peak energy was
11
8/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
13/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
Figure 12. (a) Output voltage versus load resistance, (b) output power versus load resistance.
Figure 13. Power output versus frequency.
calculated using the following equation:
Woutput =V2pp
8Rload.
Figure13shows the power outputs of three models with
the same vibration excitation of a sinusoidal acceleration of
2.5 m s2. Model I reached its maximum power output at
22.3 Hz, which was excited by the z-direction vibration. The
output power was 20.12 mW. When excited by the x-direction
vibration, the power reached a peak of 22.3 Hz, with a
power output of 0.17 mW. Model II reached its maximum
power output of 9.3 mW at 36.3 Hz, which was excited bythe z-direction vibration. When excited by the x-direction
vibration, the maximum peak-to-peak output power reached
2.39 mW at 36.4 Hz. When Model III was excited by the
z-direction vibration, the power output reached a peak of
5.7 mW at 47.9 Hz. Finally, when excited by the x-direction
vibration, the power reached a peak of 4.82 mW at 48 Hz.
6. Conclusion
Piezoelectric elements can be attached to mechanical
structures and used to transform vibration energy to
electrical energy. The present paper proposes and analyzes
a right-angle cantilever that can dramatically smoothen thestrain distribution of the piezoelectric plate. The right-angle
cantilever has an auxiliary beam that can provide a large
torque to the main beam; the strain distribution on the
main beam is highly uniform. Hence, the RUE of the PZT
element is considerably heightened and the performance of
the cantilever is greatly improved.The RUE of a PZT element in a conventional rectangular
piezoelectric cantilever with a proof mass is 50%. However,
the right-angle structure can dramatically improve it at the
initial resonant frequencies of the devices, and the RUE
increases as the stiffness of the auxiliary beam decreases. The
RUE of a PZT element is 93% when the auxiliary beam has
a stiffness which is 0.02 times that of the main beam; it is
87% when the auxiliary beam has a stiffness which is 0.25
times that of the main beam; and 80.5% when the auxiliary
beam has a stiffness which is three times that of the main
beam. The voltage outputs and power outputs of three models
under two directions of activation were also studied. Undera strain limitation, the device with higher RUE can generate
higher voltage. Furthermore, the device with higher RUE can
also generate more voltage and power under certain base
movements in thez-direction.
In summary, the right-angle structure can greatly
smoothen the strain distribution of the main beam and
considerably enhance the RUE of the PZT plate, thus
dramatically improving the voltage and power outputs. The
analysis and stimulations are consistent with the experiments.
Acknowledgment
This work was supported by the Fundamental Research Funds
for the Central Universities.
References
[1] Mitcheson P D, Yeatman E M, Kondala Rao G,Holmes A S and Green T C 2008 Energy harvesting fromhuman and machine motion for wireless electronic devicesProc. IEEE96145786
[2] Roundy Set al2005 Improving power output forvibration-based energy scavengersIEEE Pervasive Comput.42836
[3] Paradiso J A and Starner T 2005 Energy scavenging for mobileand wireless electronicsIEEE Pervasive Comput.4 1827
[4] Priya S 2007 Advances in energy harvesting using low profilepiezoelectric transducersJ. Electroceram.19 16784
12
http://dx.doi.org/10.1109/JPROC.2008.927494http://dx.doi.org/10.1109/JPROC.2008.927494http://dx.doi.org/10.1109/MPRV.2005.14http://dx.doi.org/10.1109/MPRV.2005.14http://dx.doi.org/10.1109/MPRV.2005.9http://dx.doi.org/10.1109/MPRV.2005.9http://dx.doi.org/10.1007/s10832-007-9043-4http://dx.doi.org/10.1007/s10832-007-9043-4http://dx.doi.org/10.1007/s10832-007-9043-4http://dx.doi.org/10.1007/s10832-007-9043-4http://dx.doi.org/10.1109/MPRV.2005.9http://dx.doi.org/10.1109/MPRV.2005.9http://dx.doi.org/10.1109/MPRV.2005.14http://dx.doi.org/10.1109/MPRV.2005.14http://dx.doi.org/10.1109/JPROC.2008.927494http://dx.doi.org/10.1109/JPROC.2008.9274948/10/2019 Smart Materials and Structures Volume 21 Issue 6 2012 [Doi 10.1088_0964-1726_21!6!065017] Xu, Jia Wen; Liu, Y
14/14
Smart Mater. Struct. 21 (2012) 065017 J W Xuet al
[5] Khaligh A, Zeng P and Zheng C 2010 Kinetic energyharvesting using piezoelectric and electromagnetictechnologiesstate of the art IEEE Trans. Indust. Electron.5785060
[6] Anton S R and Sodano H A 2007 A review of powerharvesting using piezoelectric materials (20032006) SmartMater. Struct. 16R1
[7] Jiang S N and Hu Y T 2007 Analysis of a piezoelectricbimorph plate with a central-attached mass as an energyharvester IEEE Trans. Ultrason. Ferroelectr. Freq. Control5414639
[8] Priya S 2005 Modeling of electric energy harvesting usingpiezoelectric windmillAppl. Phys. Lett. 87184101
[9] Shahruz S M 2006 Design of mechanical band-pass filters forenergy scavengingJ. Sound Vib.29298798
[10] Kim H Wet al2004 Energy harvesting using a piezoelectriccymbal transducer in dynamic environmentJapan. J.Appl. Phys. 143 617883
[11] Wang Set al2007 Energy harvesting with piezoelectric drumtransducerAppl. Phys. Lett. 90 113506
[12] Shenck N S and Paradiso J A 2001 Energy scavenging withshoe-mounted piezoelectricsIEEE Micro 213042
[13] Jung S M and Yun K S 2010 Energy-harvesting device withmechanical frequency-up conversion mechanism forincreased power efficiency and wideband operationAppl.Phys. Lett.96111906
[14] Renaud Met al2009 Harvesting energy from the motion ofhuman limbs: the design and analysis of an impact-basedpiezoelectric generatorSmart Mater. Struct. 18035001
[15] Giannopoulos G, Monreal J and Vantomme J 2007Snap-through buckling behavior of piezoelectric bimorphbeams: I. Analytical and numerical modeling Smart Mater.Struct.16 114857
[16] Baker J, Roundy S and Wright P 2005 Alternatives geometriesfor increasing power density in vibration energy scavengingfor wireless sensor networksProc. 3rd Int. EnergyConversion Engineering Conf. (San Francisco, CA, 1518
Aug. 2005)pp 95970[17] Goldschmidtboeing F and Woias P 2008 Characterization ofdifferent beam shapes for piezoelectric energy harvestingJ. Micromech. Microeng.18104013
[18] Wood R J, Steltz E and Fearing R S 2005 Optimal energydensity piezoelectric bending actuatorsSensors ActuatorsA11947688
[19] Benasciutti D, Moro L, Zelenika S and Brusa E 2010Vibration energy scavenging via piezoelectric bimorphs ofoptimized shapesMicrosyst. Technol.16 65768
[20] Dietland J M and Garcia E 2010 Beam shape optimization forpower harvestingJ. Intell. Mater. Syst. Struct.2163346
[21] Mehraeen S, Jagannathan S and Corzine K A 2010 Energyharvesting from vibration with alternate scavengingcircuitry and tapered cantilever beamIEEE Trans. Indust.Electron.57 82030
[22] Juan Z, Weiqun L, Yongbin L and Zhihua F 2010 Research on
uniform-strain piezoelectric energy harvesting mechanismPiezoelectr. Acoustoopt.32 21 (in Chinese)[23] Paquin S and St-Amant Y 2010 Improving the performance of
a piezoelectric energy harvester using a variable thicknessbeamSmart Mater. Struct. 19105020
[24] Friswell M I and Adhikari S 2010 Sensor shape design forpiezoelectric cantilever beams to harvest vibration energyJ. Appl. Phys.108 014901
[25] Zhang Y, Ren Q and Zhao Y P 2004 Modeling analysis ofsurface stress on a rectangular cantilever beam J. Phys. D:Appl. Phys. 3721405
[26] Kim Met al2010 Modeling and experimental verification ofproof mass effects on vibration energy harvesterperformanceSmart Mater. Struct. 19045023
[27] Jiang S Net al2005 Performance of a piezoelectric bimorph
for scavenging vibration energySmart Mater. Struct.1476974[28] Brissaud M, Ledren S and Gonnard P 2003 Modelling of a
cantilever non-symmetric piezoelectric bimorphJ. Micromech. Microeng.13 83244
[29] Erturk A and Inman D J 2009 An experimentally validatedbimorph cantilever model for piezoelectric energyharvesting from base excitationsSmart Mater. Struct.18025009
[30] Rafique S and Bonello P 2010 Experimental validation of adistributed parameter piezoelectric bimorph cantileverenergy harvesterSmart Mater. Struct.19 094008
[31] Erturk A, Renno J M and Inman D J 2008 Piezoelectric energyharvesting from an L-shaped beam-mass structureProc.SPIE692869280I
[32] Kim J, Ryu Y-H and Choi S-B 2000 New shunting parametertuning method for piezoelectric damping based onmeasured electrical impedanceSmart Mater. Struct986877
[33] Roundy S and Wright P K 2004 A piezoelectric vibrationbased generator for wireless electronicsSmart Mater.Struct.13113142
[34] Hesheng Z 1987Mechanical Impedance Method andApplication(Beijing: China Machine Press) p 74(in Chinese)
13
http://dx.doi.org/10.1109/TIE.2009.2024652http://dx.doi.org/10.1109/TIE.2009.2024652http://dx.doi.org/10.1088/0964-1726/16/3/R01http://dx.doi.org/10.1088/0964-1726/16/3/R01http://dx.doi.org/10.1109/TUFFC.2007.407http://dx.doi.org/10.1109/TUFFC.2007.407http://dx.doi.org/10.1063/1.2119410http://dx.doi.org/10.1063/1.2119410http://dx.doi.org/10.1016/j.jsv.2005.08.018http://dx.doi.org/10.1016/j.jsv.2005.08.018http://dx.doi.org/10.1143/JJAP.43.6178http://dx.doi.org/10.1143/JJAP.43.6178http://dx.doi.org/10.1063/1.2713357http://dx.doi.org/10.1063/1.2713357http://dx.doi.org/10.1109/40.928763http://dx.doi.org/10.1109/40.928763http://dx.doi.org/10.1063/1.3360219http://dx.doi.org/10.1063/1.3360219http://dx.doi.org/10.1088/0964-1726/18/3/035001http://dx.doi.org/10.1088/0964-1726/18/3/035001http://dx.doi.org/10.1088/0964-1726/16/4/024http://dx.doi.org/10.1088/0964-1726/16/4/024http://dx.doi.org/10.1088/0960-1317/18/10/104013http://dx.doi.org/10.1088/0960-1317/18/10/104013http://dx.doi.org/10.1016/j.sna.2004.10.024http://dx.doi.org/10.1016/j.sna.2004.10.024http://dx.doi.org/10.1007/s00542-009-1000-5http://dx.doi.org/10.1007/s00542-009-1000-5http://dx.doi.org/10.1177/1045389X10365094http://dx.doi.org/10.1177/1045389X10365094http://dx.doi.org/10.1109/TIE.2009.2037652http://dx.doi.org/10.1109/TIE.2009.2037652http://dx.doi.org/10.1088/0964-1726/19/10/105020http://dx.doi.org/10.1088/0964-1726/19/10/105020http://dx.doi.org/10.1063/1.3457330http://dx.doi.org/10.1063/1.3457330http://dx.doi.org/10.1088/0022-3727/37/15/014http://dx.doi.org/10.1088/0022-3727/37/15/014http://dx.doi.org/10.1088/0964-1726/19/4/045023http://dx.doi.org/10.1088/0964-1726/19/4/045023http://dx.doi.org/10.1088/0964-1726/14/4/036http://dx.doi.org/10.1088/0964-1726/14/4/036http://dx.doi.org/10.1088/0960-1317/13/6/306http://dx.doi.org/10.1088/0960-1317/13/6/306http://dx.doi.org/10.1088/0964-1726/18/2/025009http://dx.doi.org/10.1088/0964-1726/18/2/025009http://dx.doi.org/10.1088/0964-1726/19/9/094008http://dx.doi.org/10.1088/0964-1726/19/9/094008http://dx.doi.org/10.1117/12.776211http://dx.doi.org/10.1117/12.776211http://dx.doi.org/10.1088/0964-1726/9/6/318http://dx.doi.org/10.1088/0964-1726/9/6/318http://dx.doi.org/10.1088/0964-1726/13/5/018http://dx.doi.org/10.1088/0964-1726/13/5/018http://dx.doi.org/10.1088/0964-1726/13/5/018http://dx.doi.org/10.1088/0964-1726/13/5/018http://dx.doi.org/10.1088/0964-1726/9/6/318http://dx.doi.org/10.1088/0964-1726/9/6/318http://dx.doi.org/10.1117/12.776211http://dx.doi.org/10.1117/12.776211http://dx.doi.org/10.1088/0964-1726/19/9/094008http://dx.doi.org/10.1088/0964-1726/19/9/094008http://dx.doi.org/10.1088/0964-1726/18/2/025009http://dx.doi.org/10.1088/0964-1726/18/2/025009http://dx.doi.org/10.1088/0960-1317/13/6/306http://dx.doi.org/10.1088/0960-1317/13/6/306http://dx.doi.org/10.1088/0964-1726/14/4/036http://dx.doi.org/10.1088/0964-1726/14/4/036http://dx.doi.org/10.1088/0964-1726/19/4/045023http://dx.doi.org/10.1088/0964-1726/19/4/045023http://dx.doi.org/10.1088/0022-3727/37/15/014http://dx.doi.org/10.1088/0022-3727/37/15/014http://dx.doi.org/10.1063/1.3457330http://dx.doi.org/10.1063/1.3457330http://dx.doi.org/10.1088/0964-1726/19/10/105020http://dx.doi.org/10.1088/0964-1726/19/10/105020http://dx.doi.org/10.1109/TIE.2009.2037652http://dx.doi.org/10.1109/TIE.2009.2037652http://dx.doi.org/10.1177/1045389X10365094http://dx.doi.org/10.1177/1045389X10365094http://dx.doi.org/10.1007/s00542-009-1000-5http://dx.doi.org/10.1007/s00542-009-1000-5http://dx.doi.org/10.1016/j.sna.2004.10.024http://dx.doi.org/10.1016/j.sna.2004.10.024http://dx.doi.org/10.1088/0960-1317/18/10/104013http://dx.doi.org/10.1088/0960-1317/18/10/104013http://dx.doi.org/10.1088/0964-1726/16/4/024http://dx.doi.org/10.1088/0964-1726/16/4/024http://dx.doi.org/10.1088/0964-1726/18/3/035001http://dx.doi.org/10.1088/0964-1726/18/3/035001http://dx.doi.org/10.1063/1.3360219http://dx.doi.org/10.1063/1.3360219http://dx.doi.org/10.1109/40.928763http://dx.doi.org/10.1109/40.928763http://dx.doi.org/10.1063/1.2713357http://dx.doi.org/10.1063/1.2713357http://dx.doi.org/10.1143/JJAP.43.6178http://dx.doi.org/10.1143/JJAP.43.6178http://dx.doi.org/10.1016/j.jsv.2005.08.018http://dx.doi.org/10.1016/j.jsv.2005.08.018http://dx.doi.org/10.1063/1.2119410http://dx.doi.org/10.1063/1.2119410http://dx.doi.org/10.1109/TUFFC.2007.407http://dx.doi.org/10.1109/TUFFC.2007.407http://dx.doi.org/10.1088/0964-1726/16/3/R01http://dx.doi.org/10.1088/0964-1726/16/3/R01http://dx.doi.org/10.1109/TIE.2009.2024652http://dx.doi.org/10.1109/TIE.2009.2024652