Stresa, Italy, 26-28 April 2006
?TIMA Editions/DTIP 2006 -page- ISBN: 2-916187-03-0
OPTIMIZATION OF PIEZOELECTRIC ELECTRICAL GENERATORS
POWERED BY RANDOM VIBRATIONS
Elie LEFEUVRE, Adrien BADEL, Claude RICHARD, Lionel PETIT, Daniel GUYOMAR
Laboratoire de G?nie Electrique et Ferro?lectricit?, INSA de Lyon,
bat. G. Ferri?, 8 rue de la physique, 69621 Villeurbanne Cedex, France
ABSTRACT
This paper compares the performances of a vibration-
powered electrical generators using PZT piezoelectric
ceramic associated to two different power conditioning
circuits. A new approach of the piezoelectric power
conversion based on a nonlinear voltage processing is
presented and implemented with a particular power
conditioning circuit topology. Theoretical predictions and
experimental results show that the nonlinear processing
technique may increase the power harvested by a factor
up to 4 compared to the Standard optimization technique.
Properties of this new technique are analyzed in particular
in the case of broadband, random vibrations, and
compared to those of the Standard interface.
1. INTRODUCTION
The proliferation of transducers and sensors integrated in
many systems raises the problem of of the installation of
wires for power supplies and data transmission. In many
applications, the limited lifespan of batteries may induce
a costly maintenance. Moreover, batteries dying without
warning poses serious problems in safety monitoring
applications. This explains the growing interest in
miniature electrical generators enabling to power
transducers and their associated wireless data
transmission systems. Sustainable power generation may
be achieved in converting ambient energy into electrical
energy. Some possible ambient energy sources are, for
instance, thermal energy, light energy or mechanical
energy. The focus here is on the transformation of
mechanical vibrations, existing almost everywhere. In
this field, electromagnetic and electrostatic generators
have been developed. However, piezoelectric generators
are of major interest due to their solid state nature
facilitating their integration [1]. That?s why different
approaches of energy harvesting using piezoelectric
materials have been developed over the past few years.
Optimization of Piezoelectric Electrical Generators
(PEG) is an essential stage to consider their
miniaturization while preserving a significant output
power level. Previous works proposed a quasi-linear
approach to maximize the PEG output power [2,3].
Although the power electronic interface used for
optimization induces a distortion on the piezoelectric
voltage, the technique is similar to an impedance
adaptation between the load and the piezoelectric element
[4]. More recently, techniques allowing to significantly
enhancing the electromechanical energy conversions
capability of piezoelectric materials were developed
[5,6,7]. These techniques are based on nonlinear
processing of the piezoelectric voltage synchronized with
the mechanical vibrations. Compared to the quasi-linear
impedance adaptation, theoretical and experimental
results showed that these new techniques may increase
the PEG power by a factor above 10 [8].
In addition to increasing the PEG power, some of
these new nonlinear techniques intrinsically solve the
problem of the matching load impedance [7,8]. This
property is particularly interesting to optimize the energy
harvested from broadband vibrations. Indeed, the
matching load impedance required by the other
optimisation techniques remains difficult to tune
effectively in the case of broadband or random vibrations.
This case is the most difficult to solve, but it is also the
most frequent operating condition for PEGs: ambient
vibrations are usually neither of constant amplitude nor
on a single frequency. This paper analyses the broadband
effectiveness of a PEG using a nonlinear self-tuned
optimization technique with mechanical vibrations in the
50 Hz to 1 kHz region.
The theoretical results are experimentally verified for
a PEG prototype composed of a piezoelectric ceramic
bonded on a cantilever steel beam, subjected to harmonic
and random, broadband forces.
2. ARCHITECTURES
A PEG is usually composed of a mechanical part, whose
role is to collect ambient mechanical vibration and subject
the active material to stress and strain variations. Through
its electromechanical coupling property, the active
material (for instance a PZT ceramic) converts the
mechanical energy resulting from stress and strain
variations into electrical energy. In open circuit
Authors
Title. Cut at approx. 70 characters if it is too long....
?TIMA Editions/DTIP 2006 -page- ISBN: 2-916187-03-0
configuration, an alternating voltage thus develops across
the terminal electrodes of the active material.
Most of the electric loads are powered with a DC
voltage, that?s why an electronic power conversion
interface is usually placed between the terminal
electrodes of the active material and the power supply
inputs of the electric load (Figure 1). This power
conversion interface can be as simple as a diode rectifier
associated to a voltage smoothing capacitor. However,
from the electrical side, a piezoelectric element can be
basically modeled by a current source in parallel with a
capacitor (or a voltage source in series with the same
capacitor) and the search for a maximization of the power
leads naturally to a linear impedance adaptation approach
[2,3,4,8]. From its power supply inputs, the electric load
may be equivalent to a resistor, or a voltage source if a
battery is used as energy storage. So, the maximum of
power transmitted by the active material can be tracked
by adapting the load current respectively with the
variations of ambient vibrations. The power conversion
interface may thus play a key role in PEGs power
optimization.
2.1. Mechanical part
Many mechanical structures have been envisioned to
transmit environmental vibrations to active material.
These structures can be classified into two main
categories. The first category includes the structures
sensitive to environmental mechanical acceleration. In
this case, the structures are mechanical resonators, having
only one or several resonance frequencies. Effective
transmission of surrounding vibrations to piezoelectric
material thus depends on the matching between the
resonance frequencies and the environmental frequencies.
And only a few specific frequencies must be considered
for power optimization. Also, mechanical behavior of the
resonator may be significantly influenced by the energy
conversion achieved by the piezoelectric material.
In the second category, PEGs mechanical structures
directly use strain variations of the vibrating host
structure and adapt it to active material. Instead of being
limited to specific resonance frequencies, the frequency
spectrum of mechanical strain/stress transmitted to the
active material is a function of the strain variations
existing in the host structure. The frequency spectrum
may be truly broadband for this PEG category. Thus,
random environmental vibrations will not have the same
effect in both cases.
2.2. Piezoelectric material
Choice among many available piezoelectric materials is
mainly guided by the need of an important intrinsic
electromechanical coupling, leading to a high power
density and facilitating PEG miniaturization. Material
stiffness and maximum stress characteristics help in the
design of the mechanical part. The material mechanical
quality factor may also be also a parameter to consider,
more particularly for power conversion in high
frequencies. The output voltage range needed and the
available stress levels help to determine the inter-
electrodes distance. Most common geometries are plates
or disks, with different possibilities for electrodes
disposition. Electromechanical conversion is mainly
governed by the longitudinal or transverse physical
properties of piezoelectric material, respectively indexed
33 and 31, depending on the relative directions of the
poling axis and the mechanical strain/stress axis.
Energy conversion analysis can be derived from
piezoelectric equations (1) written according to IEEE
standards, where {T}, {S}, {E} and {D} are respectively
the stress vector, the strain vector, the electric field vector
and the electrical induction vector. The piezoelectric
material physical properties are the elastic stiffness matrix
determined at constant electric field [c
E
], the piezoelectric
stress matrix [e] and the permittivity matrix at constant
strain [?
S
]. ?t? refers to matrix transpose. Depending on
the piezoelectric insert geometry and coupling used,
either axial or lateral strain/stress as well as the proper
elastic and piezoelectric coefficients may be considered,
leading to the simplified scalar expressions (2).
?
???? ???
=
? ?????
?? ????
E
tS
TSce
DEe
(1)
?
? =?
?
?
=+
?
?
E
S
TcSeE
DeS E
(2)
For frequencies lower than the proper resonances of
the piezoelectric insert, equations (2) can be expressed as
a function of the average displacement u, the piezoelectric
voltage V, the equivalent mechanical force F and the
outgoing electric charge Q, leading to equation (3).
Equations (4) define the clamped capacitance C
0
, the
force factor ?, and the short-circuit stiffness K
E
of the
piezoelectric insert. t
p
is the distance between electrodes
and w
p
is the insert width in the direction of the
considered strain, A is the surface of an electrode.
0
?
?
? = ?+?
?
?
= ?? ?
?
?
E
P
F Ku V
QuCV
(3)
0
?
=
S
p
A
C
t
,
p
A
e
t
? = ? , .
EE
p
A
Kc
w
=
(4)
Thus, from the electrical point of view, the second
equation (3) shows that the piezoelectric insert used in a
PEG can be basically modeled by a current source
Authors
Title. Cut at approx. 70 characters if it is too long....
?TIMA Editions/DTIP 2006 -page- ISBN: 2-916187-03-0
controlled by the mechanical velocity
0
?=?&I u , in
parallel with a capacitor C
0
.
2.3. Power optimization interfaces
As stated in the beginning of Section 2, the AC-DC
power converter represented on Figure 1 can be used as
power optimization. A first power optimization approach,
so called ?Standard? approach, consists to considering the
structure of Figure 2, where the AC-DC power converter
in composed of a diode rectifier and a smoothing
capacitor C
R
. The electric load is modeled by an
equivalent resistor R
L
. The corresponding displacement,
voltage and current waveforms are represented on Figure
3. For this configuration, expression (5) gives of the
average electrical power produced by the PEG in the case
of an harmonic mechanical displacement u, whose
amplitude and angular frequency are U
M
and ? [2,3,8].
()
2
2
0
2
??
?
?
??
=
??
+
??
??
M
L
L
U
PR
RC
(5)
A thorough analysis would require a model of the
mechanical part of the considered PEG. Indeed, the
mechanical energy converted by the active material may
induce more or less important variations on the
mechanical displacement u and thus on the generated
current I
0
. But considering a weak mechanical damping
effect, the PEG power reaches a maximum P
Max
for a
particular value R
Lopt
of the load equivalent resistance (6).
2
2
max
0
2
M
PU
C
??
?
= ,
0
2
Lopt
R
C
?
?
=
(6)
Considering the above equations, in the case of
harmonic vibrations the PEG power optimisation can be
summarized as a tuning between the displacement
amplitude U
M
and the load resistance R
L
. In an actual
PEG, the load may include an energy storage cell such as
a large capacitor or an electrochemical battery to
overcome variations of ambient vibrations. In this case, a
constant load voltage V
R
is considered instead of the load
equivalent resistance R
L
. The PEG power expression is
then given by (7). The PEG maximum power P
Max
is the
same as in (6), but it is expressed in (8) as a function of
the optimal load voltage V
Ropt
related to the displacement
amplitude.
Maximization of the PEG power whatever the
mechanical displacement amplitude requires an additional
interface shown on Figure 4, allowing a permanent tuning
of the rectified voltage V
R
to the optimal value. This
interface, a DC-to-DC power converter, must induce as
few power losses as possible.
I
V
I
0
C
0
I
Load
V
R
Piezoelectric element
AC-DC
Power
Converter
Electric
Load
Figure 1: PEG including an AC-DC power conversion interface
I
V
I
0
C
0
I
L
V
R
R
L
Piezoelectric element
C
R
Figure 2: PEG using a diode bridge as an AC-DC power
conversion interface
t
u
V
V
R
I
Figure 3: Displacement, piezoelectric voltage and current
0
0
2
R MR
C
PVUV
C
? ?
?
??
=?
??
??
(7)
20
2
Max
Ropt
C
PV
?
?
=?
0
2
Ropt M
VU
C
?
=
(8)
In this domain, switching-mode power converters are
known for their high efficiency. The significant
development of portable electronic devices in the past
years led to important technological progress of low-
power switching-mode power converters: their efficiency
reaches now currently the region of 90%-95%.This first
power optimization approach may be qualified of ?quasi-
linear?, although the diode rectifier bridge and the
smoothing capacitor induce a distortion on the
piezoelectric voltage V. But previous works showed that
some DC-DC power converters may also be used to
increase in a particular way the piezoelectric voltage
nonlinearities [7,8]. As a result, the energy conversion
capability of the active material is greatly enhanced.
Another advantage of this nonlinear technique, so called
?Synchronous Electric Charge Extraction? (SECE), is its
self-optimization property: the PEG power remains at
Authors
Title. Cut at approx. 70 characters if it is too long....
?TIMA Editions/DTIP 2006 -page- ISBN: 2-916187-03-0
maximum whatever the load voltage or equivalent
resistance.
The SECE technique is implemented with the buck-
boost power converter structure represented on Figure 5,
with a particular control: the transistor T is turned ON
each time the rectified voltage V
R
reaches a maximum,
and it is turned OFF each time V
R
reaches zero. So, at
each triggering time, the power converter removes
completely the electric charge stored on the piezoelectric
element electrodes and transmits the corresponding
energy to the load, through the inductor L. The inductor
value is chosen in considering the piezoelement
capacitance, so that the time needed to extract the electric
charge is much shorter than the vibration period. The
corresponding mechanical displacement, piezoelectric
voltage and current waveforms are presented on Figure 6.
The average power converted by the active material is
given in (9). According to equations (6) and (9), with the
same mechanical displacement amplitude the power
converted is four times greater using this technique than
with the firstly presented power optimization approach.
2
2
0
2
M
PU
C
??
?
=
(9)
3. MULTIMODAL VIBRATIONS
Starting from the power optimization analysis in the case
of harmonic mechanical vibrations presented in Section
2.3, it is possible to consider the PEG behavior in the case
of steady state, multimodal mechanical vibrations. The
mechanical part of the considered multimodal PEG is a
cantilever steel beam, whose three first bending modes
frequencies are respectively 56 Hz, 334 Hz and 915 Hz.
The beam vibrations are driven by an electromagnet
controlled by a function generator through a power
amplifier. The active material is P1-89 PZT ceramic
(Saint-Gobain). This piezoelectric insert, composed of six
9x28x0.3 mm
3
plates, is bonded near the clamped edge.
The capacitance value of C
0
is 41.8 nF. A second
identical piezoelectric insert, bonded on the opposite side
of the cantilever beam, is used as mechanical
displacement transducer. Both power optimization
circuits are used: the ?Standard? one is composed of a
diode bridge, and a 2.2 ?F smoothing capacitor in parallel
with a variable resistor; the SECE is implemented as the
circuit represented on Figure 5, but the electrochemical
battery is replaced by a 2.2 ?F capacitor. The control
circuitry of this second optimization interface is powered
by an external power supply. First, an experimental
analysis of the PEG behavior is made for each of the
mechanical modes separately. For each frequency, the
displacement amplitude is tuned in order to reach a 1 mW
maximum power for the Standard interface.
I
V
I
0
C
0
I
R
V
R
R
L
V
L
Piezoelectric element
C
R
Electrochemical battery
DC-DC
Power
Converter
Figure 4: PEG including a DC-DC power converter for power
optimization
I
V
I
0
C
0
R
L
V
L
L
T
V
R
Piezoelectric element
Electrochemical battery
control
circuit
Figure 5: Synchronous electric charge extraction circuit
t
u
V
I
Figure 6: Displacement, piezoelectric voltage and current in the
case of synchronous electric charge extraction
0,0E+00
5,0E-04
1,0E-03
1,5E-03
2,0E-03
2,5E-03
3,0E-03
1,E+03 1,E+04 1,E+05 1,E+06
Load resistance (Ohm)
P (
W
)
Standard
SECE out
SECE in
Figure 7: First mode only (56 Hz)
Experimental results presented on Figure 7, Figure 8
and Figure 9 confirm that maximum power of the
Standard technique is reached with the optimal load
resistance values defined in equation (6), which are
respectively 106 k?, 18 k? and 6.5 k?. Plots referred as
?standard? and ?SECE in? represent the average power
outing from the rectifier bridge respectively for the
Standard interface and the SECE interface.
Authors
Title. Cut at approx. 70 characters if it is too long....
?TIMA Editions/DTIP 2006 -page- ISBN: 2-916187-03-0
0,0E+00
5,0E-04
1,0E-03
1,5E-03
2,0E-03
2,5E-03
3,0E-03
3,5E-03
1,E+03 1,E+04 1,E+05 1,E+06
Load resistance (Ohm)
P (W)
Standard
SECE out
SECE in
Figure 8: Second mode only (334 Hz)
0,0E+00
5,0E-04
1,0E-03
1,5E-03
2,0E-03
2,5E-03
3,0E-03
3,5E-03
1,E+03 1,E+04 1,E+05 1,E+06
Load resistance (Ohm)
P (W)
Standard
SECE out
SECE in
Figure 9: Third mode only (915 Hz)
0,0E+00
5,0E-04
1,0E-03
1,5E-03
2,0E-03
2,5E-03
3,0E-03
1,E+03 1,E+04 1,E+05 1,E+06
Load resistance (Ohm)
P (W)
Standard
SECE out
SECE in
Figure 10: Three modes mixed
Plots referred as ?SECE out? represent the average
power of the terminal load resistance with the SECE
interface. According to (6) and (8), the SECE technique
should theoretically harvest four times more power than
the Standard technique, but measurements show that the
power is only increased by a factor three. Difference
between theoretical and experimental power gains may be
in particular due to the piezoelectric material leakage
resistance which is not taken into account in the model.
-80
-60
-40
-20
0
20
40
60
80
-2-1012
time (s)
Displacement (ar
b
itrary units)
Figure 11: Random displacement as a function of time
-60
-50
-40
-30
-20
-10
0
10
20
30
40
0 200 400 600 800 1000 1200
Frequency (Hz)
log(displacement
)
(a.
u.)
mode 1
mode 2
mode 3
Figure 12: Random displacement spectrum
0,0E+00
5,0E-04
1,0E-03
1,5E-03
2,0E-03
2,5E-03
1,0E+03 1,0E+04 1,0E+05 1,0E+06
Load resistance (Ohm)
P (W)
Standard
SECE out
SECE in
Figure 13: Powers as a function of the load resistance,
PEG powered by random vibrations
Input power of the SECE circuit (?SECE in?)
confirms to be constant whatever the load resistance,
whereas output power (?SECE out?) decreases for high
values of the load resistance because of the circuit
imperfections. The efficiency of the circuit is practically
80% in the range 1 k? to 100 k?, the consumption of the
control circuitry not being taken into account.
When several modes are mixed, experimental results
show that the power harvested with the Standard circuit
remains smaller than with the SECE circuit, but the
power ratio is weaker than in the case of harmonic
Authors
Title. Cut at approx. 70 characters if it is too long....
?TIMA Editions/DTIP 2006 -page- ISBN: 2-916187-03-0
vibrations. Figure 10 shows for instance a case in which
the maximum power with the Standard interface is
1.63 mW whereas the input power of the SECE interface
is 2.76 mW: the power gain is here 1.7 only. In some
cases, the SECE technique may lead to harvested powers
lower than the standard technique, typically in the case of
PEG vibrations composed of a low-frequency mode with
large amplitude, mixed with a high-frequency mode with
very weak amplitude. This drawback may be solved by
selecting the right triggering times for extracting the
electrical charge stored on the piezoelectric element, that
is to say making the good choice among the local extrema
of the piezoelectric voltage.
4. RANDOM VIBRATIONS
The random force driving the PEG cantilever beam is
generated by a HP 35665A random noise generator,
powering the electromagnet through a power amplifier.
An image of the mechanical displacement measured with
a piezoelectric insert bonded on the PEG cantilever beam
near the clamped end is shown on Figure 11. The
displacement spectrum plotted on Figure 12 reveals the
mechanical filtering of the cantilever beam: the
mechanical vibration is mainly composed of the three
bending modes, respectively at 56 Hz, 334 Hz and
915 Hz. The average harvested powers presented on
Figure 13 are the average values of measurement
performed during 100 seconds. The power harvested with
the SECE technique (?SECE in?) is here roughly twice
the maximum power harvested with the Standard
technique.
5. CONCLUSION
The results presented in this paper show that it is possible
to harvest effectively broadband vibrational energy,
having random variations. Comparison between the
Standard power optimization technique and the so called
?Synchronized Electric Charge Extraction? technique
exhibit several advantages of this last one. First, there is
no need for load impedance adaptation. This property
represents a great advantage, because tracking the optimal
load resistance value needed to optimize the Standard
interface power may be relatively complicated in the case
of broadband, random vibrations. Then, for a given
mechanical vibration the power harvested by the new
interface may be increased by a factor up to 400%.
Moreover, practical implementation of the power
optimization technique is much simpler than in the case
of the Standard technique.
Ongoing works aim at improving the efficiency of
electronic circuits for both techniques, in order to make
comparisons of completely self-powered devices.
Another part of these research works will be focused on
the development of smart strategies for selecting the right
mechanical displacement extrema as triggering instants
for electric charge extraction, in order to improve the
broadband performance of the new technique.
6. REFERENCES
[1] C.-N. Xu, M. Akiyama, K. Nonaka, and T. Watanabe,
"Electrical power generation characteristics of PZT
piezoelectric ceramics," IEEE Trans. on Ultrasonics,
Ferroelectrics and Frequency Control, vol. 45, pp. 1065-1070,
1998.
[2] G. K. Ottman, H. F. Hofmann, and G. A. Lesieutre
?Optimized Piezoelectric Energy Harvesting Circuit Using Step-
down Converter in Discontinuous Conduction Mode?, IEEE
Trans. on Power Electronics, vol. 18, pp. 696-703, 2003.
[3] G. K. Ottman, H. F. Hofmann, A. C. Bhatt, and G. A.
Lesieutre, ?Adaptive Piezoelectric Energy Harvesting Circuit
for Wireless Remote Power Supply?, IEEE Trans. on Power
Electronics, vol. 17, pp. 669-676, 2002.
[4] S. Roundy, E. S. Leland, J. Baker, E. Carleton, E. Reilly, E.
Lai, B. Otis, J. M. Rabaey, V. Sundararajan, ?Improving Power
Output for Vibration-Based Energy Scavengers?, IEEE
Pervasive computing, vol. 4, N?1, pp. 28-36, January-March
2005.
[5] D. Guyomar, A. Badel, E. Lefeuvre, and C. Richard
?Towards Energy harvesting using active materials and
conversion improvement by nonlinear processing?, IEEE Trans.
on Ultrasonics Ferroelectrics and Frequency Control, vol. 52,
pp. 584-595, 2005.
[6] A. Badel, E. Lefeuvre, C. Richard, D. Guyomar, ?Efficiency
enhancement of a piezoelectric energy harvesting device in
pulsed operation by synchronous charge inversion?, Journal of
Intelligent Materials Systems and Structures, vol. 16, N?10, pp.
889-901, October 2005.
[7] E. Lefeuvre, A. Badel, C. Richard, D. Guyomar,
?Piezoelectric energy harvesting device optimization by
synchronous electric charge extraction?, Journal of Intelligent
Materials Systems and Structures, vol. 16, N?10, pp. 865-876,
October 2005.
[8] E. Lefeuvre, A. Badel, C. Richard, D. Guyomar. A
comparison between several vibration-powered piezoelectric
generators for standalone systems. Sensors and Actuators A,
vol. 126, n?2, pp. 405-416, 2006.