Journal of ELECTRONIC MATERIALS

https://doi.org/10.1007/s11664-019-07453-7
Ó 2019 The Minerals, Metals & Materials Society

Design and Optimization of Piezoelectric Transducer

(PZT-5H Stack)

MUHAMMAD ABDULLAH SHEERAZ,1 ZUBAIR BUTT,1,2,3

ABDUL MANAN KHAN,1 SHAHID MEHMOOD,1 AHSAN ALI,2
MUHAMMAD AZEEM,1 ABDUL NASIR,1 and TALHA IMTIAZ1

1.—Department of Mechatronics Engineering, University of Engineering and Technology Taxila,

Chakwal Campus, Chakwal 48800, Pakistan. 2.—Department of Mechanical Engineering,
University of Engineering and Technology Taxila, Taxila 47080, Pakistan. 3.—e-mail:
zubair.butt@uettaxila.edu.pk

Piezoelectric materials are widely used as an input transducer to harvest

electrical energy. In this research, an optimized piezoelectric transducer (PZT-
5H) has been designed for maximum efficiency. The transition system consists
of an ideal mass, spring and damper. It has been modeled using MATLAB
Simscape Toolboxä. Routh–Hurwitz criterion is applied to assess the stability
of the transition system. The root-locus and step-response plots are generated
for the stability analysis. Effects of varying stiffness, damping coefficient and
thickness have been studied in designing of stack PZT-5H. The statistical
analysis based on a Taguchi design of experiment and analysis of variance
approaches is implemented to obtain the optimum output response from the
energy harvester. Consequently, interval plots and a regression model are
designed to establish an adequate relationship between the controllable fac-
tors and optimum output response of the energy harvester. The results are
compared with the literature and they correlate well with each other.

Key words: Piezoelectric material (PZT-5H), transition system, stability

analysis, Taguchi method, ANOVA approach

Energy harvesting systems based on the mechan-

INTRODUCTION
ical vibration can be implemented by three pro-
The process utilized to extract energy from the cesses which are electromagnetic, electro capacitive
sources available in the ambient environment and and piezo-transductions.12 However, from the last
convert it into useful electrical energy is known as few years, researchers are taking a keen interest in
energy harvesting. Energy harvesting is becoming the piezo-based energy harvesters. Piezoelectric
more and more serious for an energy deficient world materials have the property to show two types of
even for the developed economies. One of the effect; (1) direct piezoelectric effect is considered as
methods to harvest energy, while some are already an input transducer as it converts mechanical
in operation, is being performed through piezoelec- signal into an electrical output, (2) Inverse Piezo-
tric materials.1–5 A number of sources are available electric effect is an output transducer as it converts
in the environment mainly including solar power, an electric field into the mechanical vibration.13
ocean waves, mechanical vibration, thermoelectric- Additionally, the material’s ability to conduct volt-
ity, and physical motions. However, for low power age plays a significant role to determine the output
electronics devices, ambient mechanical vibration response of an energy harvester.14 Polarization and
could be a great choice6–11 as observed from Fig. 1. dielectric constant of the piezoelectric materials
decrease with the variation in the thickness
whereas energy storage density improves by
(Received March 7, 2019; accepted July 11, 2019) increasing the relative thickness.15
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

Fig. 1. Availability of natural resources in the environment.

Ottoman and Gerizechick designed an energy Normally, piezoelectric materials are characterized
harvester using Lead Zirconate Titanate (PZT). at room temperature. But the properties of these
They successfully implement the experimental materials could be disturbed when they are heated
setup in the laboratory. They stated that, the to their Curie temperature. One such study was
vibrating Piezo element can be modeled as a sinu- presented in Ref. 21 to investigate the electrical
soidal current source and this model could be used properties of PZT-5A. Results revealed that the
to predict the output response of a certain Piezo output voltage of an energy harvester decreases
element. The goal of this experiment was to find out with the increase of temperature. Mechanical
the optimum values of voltage and power from a parameters such as Young’s modulus, tensile and
Piezo-sensor.16,17 Elvin proposed an energy scav- compressive strengths of an energy harvesting
enging system based on the Polyvinylidene Flouride system play a vital role to obtain the maximum
(PVDF) film attached with a simply supported output response. Several techniques are introduced
beam. Electrical energy was created due to the to make improvements in these parameters.
bending of a beam. In order to make the system Recently, composite materials based on PZT and
more stable, he designed a signal conditioning zinc oxide nanowhiskers (ZnOw) are prepared with
circuit based on the limit switch. The basic function the aid of solid-state processing. The mechanical
of a switch was to allow the capacitor to charge up to quality factor of these composites is significantly
a predefined limit of 1 V. After this, the switch larger than monolithic lead zirconate Titanate.22
should be opened, forcing the capacitor to discharge Sodano proposed an energy harvesting system
through the transmitter. The whole operation of the which was based on the variational principles. This
system was found to generate an appropriate level model was actually used to predict the amount of
of electrical energy depending upon the strain at a electrical energy that could be attained due to
bending level of 2 m.18 Elahi, H proposed and alternative distributions on the piezo-patches. A
analyzed a suspension system for characterizing composite beam with four piezo-patches was tested
PZT-5A. Stability analysis of the system was per- experimentally to validate the results of a simula-
formed by the Nyquist theorem and Routh–Hurwitz tion. After this authentication, the model was used
(RH) criterion.19 In the past few years, researchers to anticipate the output response of an energy
are taking a huge interest to improve the efficiency harvester under the damping of a structure.23
of a PZT based energy harvester. Piezoelectric Double scale Lead Zirconate Titanate ceramic mate-
single crystal nanowires (NWs) are popular because rials are also studied with mechanical characteris-
of their inherent one-dimensional structure which tics that are significantly increased. The authors
helps to possess good electromechanical conver- justified the flexural strength and fracture tough-
sions. A flexible and translucent energy harvester ness of double-scale PZT is about 30% higher than
based on the polydimethylsiloxane (PDMS) films the normal PZT.24 The mechanical quality factor is
having Pt/Ti electrodes was designed and fabri- conspicuously used to describe the energy losses of
cated. The analytical and numerical model also piezoelectric materials. It is observed that the
confirmed that this harvester is capable of produc- mechanical quality factor and the efficiency of piezo
ing the highest open circuit voltage of 10 V.20 materials increase with the applied force. This
Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

argument has also been justified with the finite S3 ¼ d33 T3 þ g3 D3 ; ð1Þ
element analysis on the ABAQUSä.25 In the past,
dense and cracked-free PZT composite films having
the perovskite structure were designed and fabri- E3 ¼ g33 T3 þ b3 D3 : ð2Þ
cated. These films contain micro and nanocrys-
talline particles and their electrical properties are Equation 2 is utilized when a piezoelectric mate-
highly dependent on the annealing temperature and rial is used as an input transducer. The transition
relative thickness.26 During the fabrication of a new system plays a typical role in transforming the
product, engineers take many scientific and direc- mechanical vibration from an input excitation to the
torial decisions at various stages. The main objec- piezo-stack. This section will make use of Newton’s
tive of these decisions was to either increase the laws of motion, ordinary differential equations and
productivity or minimize the time required for that the concept of transfer function to represent the
production. Many researchers are taking more linear time-invariant model of a transition system.
interest in the optimization strategy for piezo-based The ideal mass spring damper system is shown in
energy harvesters. A review of such optimization the Fig. 2a. The ultimate goal of the present anal-
has been observed in Ref. 27. They presented the ysis is to find out the transfer function which will
voltage optimization of PZT-circular discs by con- represent the linear relationship between input and
sidering force, thickness and voltage constant as output of the transition model. For a second-order
controllable factors. They have used the standard system, the standard form of a transfer function is
American Piezo-ceramics manual for output param- observed from Eq. 3.32,33
eters, voltage, and power.
Different piezo-materials such as Quartz, PZT- X ðsÞ x2n
T ðs Þ ¼ ¼ 2 ð3Þ
5A, PZT-4A, PVDF and PZT-5H are used as input F ðsÞ s þ 2fxn s þ x2n
and output transducers. In this research work, an
energy scavenging model on the basis of PZT-5H
In Eq. 3, X(s) and F(S) are output and input,
has been presented. We have examined the perfor-
respectively. Output response can be obtained by
mance of PZT-5H-stack under the influence of a
considering four forces as shown in Fig. 2b. In this
Transition Model. A Transition Model consists of a
analysis, mass is traveling downward due to the
mass which is analogous to inertia, a spring with an
application of applied force, while all other forces
appropriate stiffness and a damping element creat-
impede the motion and act opposite to it.
ing friction effects. Theoretical analysis for a tran-
Hence a differential equation can be written by
sition system was performed to analyze
using Newton’s law:
stable ambient mechanical vibration. MATLAB
X
Simscape Toolboxä is used to (a) create the linear F ¼ m€x:
response of a transition system from which transfer
function is generated and (b) design, simulate and
analyze the energy scavenging system based on the Taking downward direction as a positive, cumu-
Piezo-Stack(PZT-5H). From this energy harvester, lative effect is given in Eq. 5.
voltage and power are analyzed under several
conditions of the stiffness, applied force and damp- F  FC  Fk ¼ m€
x ð4Þ
ing coefficient. Optimization was performed to
obtain desired results under the given experimental x þ C€
m€ x þ Kx ¼ F ð5Þ
conditions. In an energy scavenging system, tradi-
tional methods for the selection of input parameters
do not provide an adequate or desired result. The Eq. 5 with respect to time can be observed
Therefore, statistical analysis has been performed from Eq. 6
on the output values obtained from an energy xðtÞ þ C€
m€ xðtÞ þ KxðtÞ ¼ F ðt; Þ ð6Þ
scavenging system. The analysis was based on the
Taguchi design of experiment (DOE) technique,
ANOVA approach and regression model to obtain C K 1
x€ðtÞ þ x€ðtÞ þ xðtÞ ¼ F ðtÞ : ð7Þ
optimum voltage and power as well as to reduce the m m m
experimental runs.
If we consider the Laplace transform, certain
THEORETICAL ANALYSIS parameters are given in Eq. 8
In the mid-18th century, Carl Linnaeus and x€ðtÞ ¼ S2 X ðsÞ; x€ðtÞ ¼ SX ðsÞ; xðtÞ ¼ X ðsÞ and F ðtÞ
Franz Azepinos proposed the concept of the pyro-
¼ F ðsÞ: ð8Þ
electric effect which is considered as a first source
for the piezo-electricity.28–31 In order to observe the
linear polarization behavior, constitutive laws are Therefore, Eq. 7 in Laplace’s form is observed
given in Eqs. 1 and 2; from Eq. 9
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

Fig. 2. Transition model as (a) equivalent system (b) free body diagram.

Table I. Parametric values for transition system Table II. RH-table for transition model
Sr.# Parameter Value Unit
S2 1 125
1 Mass 6 Kg S1 4.167 0
2 Damper 25 Ns m1 S0 125 0
3 Stiffness 750 Nm1
0:167
TðsÞ ¼ ð12Þ
ðS2 þ 4:167s þ 125Þ

C K 1 For the present analysis, the value of damping

S2 X ðsÞ þ SX ðsÞ þ X ðsÞ ¼ F ðsÞ ; ð9Þ
m m m ratio is 0.18 which is less than 1 so the response is
underdamped. In other words, the system will
  deliver vibration to the piezo-stack with a lesser
2 C K 1
S þ sþ X ðsÞ ¼ F ðsÞ : ð10Þ value of rise time. Moreover, Routh–Hurwitz (R–H)
m m m
criterion is used to perform the stability analysis of
this second order system. The power of this method
And solving for the transfer function yields, actually lies in its design rather than analysis.
1
Completed R–H analysis for this system is observed
X ðsÞ from Table II.
¼  2 Cm K
;
F ðsÞ S þ ms þ m From the second column of Table II, we have
observed that no sign changes occur. Therefore, all
1
the poles exist in the left-half plane. And the system
X ðsÞ will be stable under predefined conditions of mass,
TðsÞ ¼ ¼  2 Cm K
: ð11Þ
F ðsÞ S þ ms þ m spring and damping element. Moreover, the above-
mentioned analysis is verified through MATLAB
simulation. So, by using the present strategy one
By comparing the Eqs. 3 and 11, we have can adjust the standards in order to obtain the
K C stable ambient mechanical vibration which will be
x2n ¼ and 2fxn ¼ : good for a piezo-stack.
m m
SIMULATION AND ANALYSIS
So natural frequency
qffiffiffi and damping ratio of the
K C
MATLAB Simscape Toolboxä provides us a
system are xn¼ m and f ¼ 2mx n
. Values used for graphical programming environment (GPE) to
mathematical modeling and simulation are given in model and analyze the multi-domain physical sys-
Table I. tems. Piezo-stack library is introduced in a Sim-
So the final expression for transfer function is scape environment to model and analyze the
given in Eq. 12. proposed energy scavenging system. Additionally,
Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

Simulink-Ps, Solver-configuration and PS-Simulink Therefore, a derivative block is introduced to obtain

are the blocks which are used to configure the the accelerated part of the vibration which is
Simscape environment. further connected to the PZT-Stack through an
ideal force source as shown in Fig. 8. Block connec-
Stability Analysis of the Transition System tions R and C are mechanical translational conserv-
ing ports, so a signal at port S generates force acting
Figure 3 shows the design of a transition model
from R to C.
on Simscape which is used to transfer the mechan-
PZT-Stack is used as an input transducer by
ical vibration. In the past research, the model under
adjusting the parametric values as given in Table
consideration has constraints of not involving some
III. Due to an input excitation, AC voltage is
parameters like damping element, stiffness, inertia
produced by the piezo-stack which can be sensed
which in turn affect the overall system performance.
through the voltage sensor. The scope block with
In this research work, mechanical vibration for the
PS-Simulink converter is used to observe this
piezo-stack is obtained from this transition model.
voltage. This system is further connected to the
Therefore, certain graphical representations like
electrical interfacing circuit consisting of the bridge
step response, pole-zero (PZ) plot and root locus are
rectifier, filter capacitor and a resistive load to
introduced to analyze the physical performance and
obtain pulsating DC voltage.
stability of the transition system. Figure 4 shows
The equivalent energy harvesting system can be
how to obtain the transfer function through this
observed from Fig. 8. The proposed model has been
simulation.
used to observe the effect of controllable parameters
Solving the generated transfer function with
such as force, damping element, stiffness, resistive
MATLAB, the time response of the transition
load on the rectified output voltage. Moreover,
system is shown in Fig. 5. It shows that the system
parameters like piezo-electric charge constant, elas-
has a rise time up to 0.107 s and settling time is less
tic compliance, dielectric constant, length, size and
than 2 s. So, only a minute amount of time is
width of the piezo stack are also varied during
required to transfer the mechanical vibration to the
simulation to contemplate the output response of an
PZT-stack.
energy harvester. Therefore, the constraints of a
Root-locus of the system is also generated to
transition model are eliminated up to an appropri-
support the results of the R-H criterion. Figure 6
ate extent. It is also observed that thicker piezo-
reveals that all the loop poles are in the left-half
electric stacks are progressively better for higher
plane, so the system is reasonably stable.
voltage generation which is due to the smaller
electrical capacitance. The output response of an
Modeling and Analysis of PZT-5H
energy harvester is decreased with the variation in
In order to design the piezo-scavenging system, the size of the stacked PZT-5H.
certain steps are involved which are as shown in Several readings are obtained by changing the
Fig. 7. The main components of the work-flow values of the input parameters. It is found that
diagram are the PZT-stack itself, configuration output voltage and power are directly proportional
parameters, bridge rectifier with filter capacitor to the applied force and stiffness. While the inverse
and a resistive load. Mechanical vibration from a relationship is observed for the case of damping
transition system is sensed in the form of velocity by coefficient. In order to support these facts, certain
using an ideal translational motion sensor. parametric equations are also involved. As power is

Fig. 3. Design of the transition system on MATLAB.

 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

Fig. 4. Generation of a transfer function.

Fig. 5. Time response of the transition system.

Fig. 6. Stability analysis of transition system via root-locus.

Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

Fig. 7. Components of the energy harvesting system.

Fig. 8. Electromechanical model of energy harvester based on a PZT-stack.

Table III. Material properties and dimension for stacked PZT-5H

Sr.# Parameter Value Unit

1 Length of piezoelectric Stack 40 mm

2 Width of piezoelectric stack 1.5–12 mm
3 Thickness of piezoelectric stack 3–20 mm
4 Piezoelectric charge constant (d31) 110 9 1012 mv1
5 Elastic compliance (S) 13.5 9 1012 m2 N1
6 Relative dielectric constant 1200 F/m
7 Young’s modulus 74 9 109 N/m2

the product of force and velocity, so with the Moreover, at a constant force of 60 N, stiffness of
constant velocity, a relationship is observed from 750 Nm1 and damping coefficient of 25 Ns m1,
Eq. 13. the PZT-Stack is excited by an acceleration of
1.81 9 10+2 m2 s1. Due to this input excitation,
Power / Force ð13Þ an alternating voltage of 5.197 V is obtained which
is further connected to the bridge rectifier to gain
And by keeping the current constant, the rela- DC-values for voltage (3.985 V) and power
tionship between power and voltage can be obtained (8.169 W) as shown in Figs. 9 and 10.
from the power dissipation formula as in Eq. 14.
RESULTS AND DISCUSSION
Power / Voltage ð14Þ
In this section, theoretical and numerical results
are interpreted and discussed.
Equating Eqs. 13 and 14, we have a direct
relationship between the applied force and voltage Phase-1: Voltage and Power at Constant Force
as in Eq. 15. (20 N) with the Variable Stiffness
Force / Voltage ð15Þ Figure 11 shows measurement of the output
voltage at a constant force of 20 N under variable
And in order to observe the effect of stiffness, we stiffness ranging from 150 Nm1 to 1200 Nm1. It is
have Hook’s law as given in Eq. 16: predicted that piezoelectric material shows positive
linear behavior towards different values of stiffness.
Force / Stiffness: ð16Þ This analysis is performed under resistance values
of 1 X, 10 X and 100 X. It is also anticipated that
By equating Eqs. 15 and 16, a direct relationship output voltage is increased with the resistance
between stiffness and output voltage is obtained as value and, hence, satisfying the Ohm’s law. Simi-
given in Eq. 17 larly, Fig. 12 shows the output power under vari-
able stiffness with resistance values of 1 X, 10 X
Stiffness / Voltage: ð17Þ and 100 X. It reveals that both the power and
stiffness are directly proportional to each other.
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

Fig. 9. Outcomes of energy harvester (a) acceleration (b) AC-voltage.

Moreover, it is also observed that power values are plays a typical role in observing the working of the
decreased with a slight variation in the resistance, energy harvester. Therefore, in this phase, the
so satisfying the power dissipation principle. width of a piezo stack is varied ranging from
1.5 mm to 12 mm. The applied force, stiffness and
Phase-2: Voltage and Power at Constant Force damping coefficient are set to be 60 N, 750 Nm1
(20 N) with the Variable Damping Coefficient and 25 Ns m1 respectively. Results revealed that
the output response of an energy harvester
In this phase, both voltage and power are mea-
decreases non-linearly with the width of the PZT-
sured under the various conditions of damping
5H stack. Figure 15 shows that both voltage and
coefficient. Figures 13 and 14 show the output
power are decreased with the width of a PZT-stack.
voltage and power under variable damping coeffi-
The graphical representations of response
cient ranging from 5 Ns m1 to 40 Ns m1. A neg-
obtained from PZT-5H are designed on statistical
ative linear relationship is observed for both voltage
software NCSS 12 to observe the cumulative effect of
and power. It means that by increasing the damping
all the controllable factors such as force (F), stiffness
coefficient, both output voltage and power will
(K) and damping coefficient (C) on the output
decrease. This analysis is performed under the
voltage and power as shown in Fig. 16.
same conditions of resistance values of 1 X, 10 X
and 100 X.
STATISTICAL ANALYSIS
Phase-3: Output of Energy Harvester This section presents the optimization strategy
with Variable Width of a PZT-5H Stack using the Taguchi design of experiment (DOE) and
ANOVA approaches to obtain optimum values of
Although, the obtained output response of energy
voltage and power with the least number of exper-
harvester depends on the length, width and thick-
iments. Experiments are performed on the response
ness of the piezo-stack. But the width of a stack
of an energy scavenging system.
Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

Fig. 10. Output response of PZT-Stack (a) DC-voltage (b) DC-power.

Fig. 11. Measurement of output voltage under variable stiffness. Fig. 12. Measurement of output power under variable stiffness.

Taguchi’s Approach industries. This method has significant importance

The Taguchi method is a standardized statistical in the diverse field of research and engineering.34–36
approach to obtain the optimum output response The present analysis has been used to study the
with the best combination of inputs. This is accom- effect of controllable parameters; force, stiffness,
plished through the partial factorial design of and damping coefficient on two important output
experiments. Although, DOE was originally intro- parameters, voltage and power. Moreover, Fig. 17
duced for robust quality control in manufacturing shows the steps followed for this strategy.
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

tries to minimize the effect of noise factors. For a

PZT-Stack, key controllable factors are force, stiff-
ness and damping coefficient. In the current study,
three levels of each of these factors are considered
and their values are given in Table IV.

Selection of Orthogonal Array

Orthogonal array identifies the manner in which
input parameters change their values during the
experiment. Commonly used arrays are L4, L8, L9,
L12, L16 and L25. Controllable factors with their
levels play a vital role in the selection of such an
array. This research work is presented by using L9
array. It must be noted that traditional experimen-
tal design needs to have 33 runs to perform this
experiment. But the Taguchi method utilizes only 9
Fig. 13. Output voltage of energy harvester under variable damping experimental runs to accomplish this task. The
coefficient. present Taguchi analysis has been presented with
the help of MINITABä 17.0. This analysis is
performed with three trials based on the resistance
values of 1 X, 10 X and 100 X to obtain mean values
of output voltage and power. It is observed that
maximum values of output voltage and power are
3.985 V and 8.169 W, respectively, and occur at the
maximum values of stiffness and applied force
(750 Nm1 and 60 N) with a minimum value of
damping coefficient which is 25 Ns m1.

Quality Characteristics
The signal to noise ratio (S/N) reveals the varia-
tion of the output response relative to nominal or
target value under several conditions of the input
parameters. S/N characteristics are categorized as
(a) nominal and smaller are the best characteristic
Fig. 14. Output power of energy harvester under variable damping
coefficient.
and (b) larger is better. This research work has been
presented by using nominal’s characteristics. Fig-
ures 18 and 19 represent the effect of controllable
factors on the voltage and power by S/N and mean
analysis.
The response chart has been created by consider-
ing the mean-values. This chart has helped us to
determine the parameters that have the highest
and lowest effect on the output response. Table V
shows that in addition to the applied force, stiffness
may have the greatest effect on the output voltage
and power. The effect of each factor is calculated by
determining the range (D).

ANOVA’s Approach
In order to support the results of Taguchi anal-
ysis, we have performed an analysis of variance
Fig. 15. Output response with a varying width of a PZT-stack. (ANOVA) approach which comes under the Taguchi
Grey. ANOVA test is used to determine the indi-
Control Factors and Their Levels vidual contribution of input parameters on the
output response. The present ANOVA is designed
The overall objective of this research work is to
with the confidence interval of 95% having the
maximize the harvested energy. Mainly two types of
significance level of 5%. The ANOVA table contains
factors exist that affect the output performance;
degree of freedom (DOF), sum of squares (Adj SS),
controllable and noise factors. Taguchi method uses
and variance with the desired P value. The
controllable factors with their optimized values and
Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

Fig. 16. Response plot by stiffness and damping coefficient for (a) voltage (b) power.

cumulative effect of all factors has been performed and power are generated, whereas harvested energy
and given in Table VI. The third column of Table VI decreases with the damping coefficient. These mod-
shows that stiffness has the highest effect on the els also show the minor deviation between actual
output response of an energy harvester which can data and the predicted response which can be
be observed from its largest value of the sum of observed from R2 values.
squares (Adj SS). At the end of this research work, the importance
Table VI also shows that P value of all these of stack PZT-5H is evaluated by comparing the
factors is less than 0.05 (significance level), so, proposed results with the literature. Figure 24
accuracy of this model is achieved up to 99% also validates the simulation and optimization results
depicted from R2 value in Table VII. in accordance with the published results of Lin Wei
Interval plots are used to support the results of and Chen Wen.37 This is obtained by varying the
ANOVA and compare the output values with the thickness of the PZT-Stack ranging from 3 mm to
input parameters. These plots are similar to a box or 20 mm. The applied force, length and width of a
a dot plot. But desired confidence interval for the piezoelectric stack are set to be 60 N, 40 mm and
mean of data can be implemented with interval 12 mm, respectively. The trend lines show that
plots so eradicating the limitations of a box or dot current simulation results are better than the
plot. Figures 20 and 21 describe the variation of published results, especially for higher vibrational
voltage and power under various conditions of the modes.
stiffness and damping coefficient by considering the
confidence interval of 95%. Both the output voltage CONCLUSIONS
and power are maximum at stiffness of 750 Nm1
Simulation is an important tool for data manip-
with a damping coefficient of 25 Ns m1. The linear
ulation and analysis of complex engineering prob-
behavior is observed in the study of stiffness while
lems. A transition model plays a vital role in
non-linear performance is noticed in the case of the
transforming the mechanical vibration from input
damping coefficient.
excitation to a piezoelectric material. Therefore, in
The regression model is designed to consider the
this research paper, a transition system has been
graphical relationship between predefined inputs
presented in the phase variable form to interpolate
and output of an energy harvester. In linear regres-
its stable response. Root locus and time response for
sion, variation of the response variable is checked by
a transition system have been generated using
means of the coefficient of determination (R-
MATLAB to validate the results of theoretical
Squared value). The value of R2 is always between
analysis. Results reveal that the system is
0% and 100%. Moreover, the slope of the regression
stable at a 60 N force, 750 Nm1 stiffness and
line predicts the change in the dependent variable
having a damping coefficient of 25 Ns m1.
when an independent variable increases by one
MATLAB Simscape Toolboxä is used to design
unit. Figures 22 and 23 show the regression models
and simulate the energy harvesting system based
for voltage and power with the variation in the
on the stack PZT-5H, as several researchers have
stiffness and damping coefficient. Results displayed
shown adequate relationships between the applied
that by increasing stiffness, more output voltage
force and piezo-voltage. Therefore, present research
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

Fig. 17. Process for Taguchi design of experiment (DOE).

Table IV. Input parameters with levels

Sr.# Input parameters/factors Symbol Unit Level-1 Level-2 Level-3

1 Force A N 20 40 60
2 Damping co-efficient B Ns/m 25 50 75
3 Stiffness C N/m 250 500 750

work is presented to analyze the performance of an piezoelectric stack. Conversely, thicker piezoelectric
energy harvester under various conditions of the stacks are more promising for higher voltage gen-
stiffness and damping coefficients. It is interesting eration. The statistical analysis based on Taguchi
to note that when the stiffness of a transition and ANOVA approaches is implemented for the
system increases, the output voltage and power optimization of a piezoelectric transducer to gain
increase, whereas, the harvested energy decreases maximum voltage and power. It is found that
with the damping coefficient. Additionally, it is maximum values for voltage and power are
found that the output of an energy harvester 3.985 V and 8.169 W obtained at those standards
decreases non-linearly with the width of the of force, stiffness and damping values which are
Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

Fig. 18. Effect of controllable factors on voltage via (a) S/N ratio (b) mean-plot.

Fig. 19. Effect of controllable factors on power via (a) S/N ratio (b) mean-plot.

Table V. Response table by considering nominal

characteristics
Table VII. Model summary by considering
Level Force Damping coefficient Stiffness ANOVA’s approach

1 2.4659 2.8860 0.9264 S R2 R2 (adj) R2 (pred)

2 2.1370 2.0707 2.2623
3 1.7587 1.4049 3.17 0.038 99.08% 99.9% 99.50%
D 0.7072 1.4811 2.24
Rank 3 2 1

Table VI. Analysis of variance (ANOVA) for all factors collectively

Source DOF Adj SS Adj MS F value P value

Force 2 0.7515 0.37574 259.66 0.004

Damping coefficient 2 3.3016 1.65079 1140.82 0.001
Stiffness 2 7.6608 3.83039 2647.08 0.0001
Error 2 0.0029 0.00145
Total 8 11.7167
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

Fig. 20. Measurement of voltage under variable (a) stiffness (b) damping coefficient.

Fig. 21. Measurement of power under variable (a) stiffness (b) damping coefficient.

Fig. 22. Regression model for output voltage via (a) stiffness (b) damping coefficient.

used to obtain stable mechanical vibration from the ACKNOWLEDGMENTS

transition system. The results are in good agree-
The authors gratefully acknowledge the support
ment with the literature. Therefore, it is recom-
of the Mechatronics Engineering Department of the
mended to use stack PZT-5H in the energy
University of Engineering and Technology (UET)
harvesting system.
Design and Optimization of Piezoelectric Transducer (PZT-5H Stack)

Fig. 23. Regression model for output power via (a) stiffness (b) damping coefficient.

Fig. 24. Comparison of proposed results with literature via (a) output voltage (b) output power.

Taxila, Sub-Campus Chakwal, Pakistan for provid- 13. M.U. Khan, Z. Butt, H. Elahi, W. Asghar, Z. Abbas, M.
ing technical assistance to carry out this research. Shoaib, and M.A. Bashir, Microsyst. Technol. (2018). http
s://doi.org/10.1007/s00542-018-4182-x.
REFERENCES 14. M. Asif, F. Rasool, Z. Butt, M. Shahzad, N. Ahmad, and A.R.
Chishti, Trans. Electr. Electron. Mater. 20, 141 (2019).
1. S. Chalasani and J. M. Conrad, in IEEE Southeast Con 2008 15. Q. Zhao, H. Lei, G. He, J. Di, D. Wang, P. Tan, H. Jin, and
(IEEE, 2008), pp. 442–447. M. Cao, Ceram. Int. 42, 1314 (2016).
2. H.A. Sodano, D.J. Inman, and G. Park, J. Intell. Mater. Syst. 16. G.K. Ottman, H.F. Hofmann, A.C. Bhatt, G.A. Lesieutre,
Struct. 16, 799 (2005). IEEE Trans. Power Electron. 17, 669 (2002).
3. J.A. Paradiso and T. Starner, IEEE Pervasive Comput. 1, 10 17. G.A. Lesieutre, G.K. Ottman, and H.F. Hofmann, J. Sound
(2005). https://doi.org/10.1109/mprv.2005.9. Vib. 269, 991 (2004).
4. Z.L. Wang and W. Wu, Angew. Chem. Int. Ed. 51, 11700 18. N.G. Elvin, A.A. Elvin, and M. Spector, Smart Mater. Struct.
(2012). 10, 293 (2001).
5. F. Yildiz, J. Technol. Stud. 35, 40 (2009). 19. H. Elahi, A. Israr, R. F. Swati, H. M. Khan, and A. Tamoor,
6. E.K. Reilly, F. Burghardt, R. Fain, and P. Wright, Smart in 2017 5th International Conference of Aerospace and
Mater. Struct. 20, 125006 (2011). Mechanical Engineering ICASE 2017, vol. 1 (2018), pp. 1–5.
7. G. Altena, M. Renaud, R. Elfrink, M.H. Goedbloed, C. De 20. Q.L. Zhao, G.P. He, J.J. Di, W.L. Song, Z.L. Hou, P.P. Tan,
Nooijer, and R. Van Schaijk, J. Phys. Conf. Ser. 476, 012078 D.W. Wang, M.S. Cao, ACS Appl. Mater. Interfaces 9, 24696
(2013). (2017).
8. T.H. Ng and W.H. Liao, J. Intell. Mater. Syst. Struct. 16, 785 21. Z. Butt and R. A. Pasha, IOP Conf. Ser. Mater. Sci. Eng. 146,
(2005). 012016 (2016).
9. Z. Butt, R.A. Pasha, F. Qayyum, Z. Anjum, N. Ahmad, and 22. L. Hai-Bo, C. Mao-Sheng, Y. Jie, W. Da-Wei, Z. Quan-Liang,
H. Elahi, J. Mech. Sci. Technol. 30, 3553 (2016). and W. Fu-Chi, Chin. Phys. B 17, 4323 (2008).
10. S. Saadon and O. Sidek, Energy Convers. Manag. 52, 500 23. H.A. Sodano, D.J. Inman, and G. Park, J. Intell. Mater. Syst.
(2011). Struct. 16, 67 (2004).
11. Z. Abbas, S. Abbas, Z. Butt, and R.A. Pasha, J. Shanghai 24. W. Da-Wei, J. Hai-Bo, Y. Jie, W. Bao-Li, Z. Quan-Liang, Z.
Jiaotong Univ. 23, 345 (2018). De-Qing, and C. Mao-Sheng, Chin. Phys. Lett. 27, 047701
12. C.B. Williams and R.B. Yates, Sensors Actuators. A Phys. (2010).
52, 8 (1996).
 Sheeraz, Butt, Khan, Mehmood, Ali, Azeem, Nasir, and Imtiaz

25. Z. Butt, Z. Anjum, A. Sultan, F. Qayyum, H.M. Khurram 33. E.K. Orhorhoro, D.S. Polytechnic, M.E. Onogbotsere, D.S.
Ali, and S. Mehmood, J. Electr. Eng. Technol. 12, 846 (2017). Polytechnic, A.E. Ikpe, U. Kingdom, and U. Kingdom, ELK
26. R. Lu, G.S. Jiang, B. Li, Q.L. Zhao, D.Q. Zhang, J. Yuan, Asia Pacific, J. Mech. Eng. Res. 2, 2 (2016).
and M.S. Cao, Chin. Phys. Lett. 29, 5 (2012). 34. D.H. Wu, W.T. Chien, and Y.J. Tsai, IEEE Trans. Ultrason.
27. K. Viswanath Allamraju, Mater. Today Proc. 5, 5322 (2018). Ferroelectr. Freq. Control 52, 480 (2005).
28. A. Qabur and K. Alshammari, Innov. Energy Res. 07, 1 35. M.H.S. Alrashdan, A.A. Hamzah, and B.Y. Majlis, Mi-
(2018). crosyst. Technol. 21, 1607 (2015).
29. J.A. Paradiso and T. Starner, IEEE Pervasive Comput. 4, 18 36. S. Mukhtar, W. Asghar, Z. Butt, Z. Abbas, M. Ullah, and R.
(2005). Atta-Ur-Rehman, J. Cent. South Univ. 25, 2578 (2018).
30. P.D. Mitcheson, E.M. Yeatman, G.K. Rao, A.S. Holmes, and 37. W. Lin, Z. Li, W. Chen, and J. Zhou, Adv. Mater. Res. 267,
T.C. Green, Proc. IEEE 96, 1457 (2008). 1005 (2011).
31. M. Lazarek, P. Brzeski, and P. Perlikowski, Mech. Mach.
Theory 119, 161 (2018). Publisher’s Note Springer Nature remains neutral with
32. C. Li and F. Zhang, Adv. Differ. Equ. (2011). https://doi.org/ regard to jurisdictional claims in published maps and institu-
10.1155/2011/213485. tional affiliations.