FT 133
FT 133
FT 133
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
014901-2 M. I. Friswell and S. Adhikari J. Appl. Phys. 108, 014901 共2010兲
冦 冧冦 冧
Using the finite element displacement functions, the
u̇ u̇
charge on the piezoelectric material may be estimated for
˙ ˙ each element and the total charge obtained by summing the
charge over the elements. This is exactly the same as the
q̇1 + u̇ + x2˙ q̇1 + u̇ + x2˙
T= Mf f development for shaped sensors for control or structural
q̇2 + ˙ q̇2 + ˙ health monitoring.20,21 For a beam element with cubic shape
functions,20 the current generated by the piezoelectric is the
q̇3 + u̇ + x3˙ q̇3 + u̇ + x3˙
time derivative of the charge produced, and is
] ]
i pe共t兲 = − f 共6兲
再冎再冎 再冎
e C peq̇e ,
u̇ u̇ u̇ where
= q̇Mq̇ + 2q̇B M + EM 共2兲
冤 冥
˙ ˙ ˙ 36 33ᐉe − 36 3ᐉe
and e31h 3ᐉe − 3ᐉe − ᐉ2e
4ᐉ2e
C pe = , 共7兲
冦冧 冦冧
30ᐉe − 36 − 3ᐉe 36 − 33ᐉe
0 0
U= 0 K f f 0 = qKq, 共3兲 3ᐉe − ᐉ2e − 3ᐉe 4ᐉ2e
q q ᐉe is the element length, qe are generalized displacements for
the element, and fe is the piezoelectric shape defined at the
where q = 关q1q2. . .兴 are the nodal generalized displacements element nodes. Adding the contributions for all nodes gives
of the beam model, not including the node at the support, and
再冎
the total current produced as
are defined relative to the rigid body motion of the beam.
The dot denotes the derivative with respect to time and xi is u̇
the distance of the ith node from the support. M and K are i p = − fC pqq̇ − fC ps , 共8兲
˙
the standard mass and stiffness matrices for the beam with
cantilever boundary conditions. The matrix B M is obtained where C pq and C ps are matrices that are easily assembled
from the columns of the free-free mass matrix, and is given from element coupling matrices, Eq. 共7兲, and f is a vector
by that defines the sensor shape.20,21 The last term in Eq. 共8兲
冤冥
arises because of the imposed motion at the beam support. f
1 0
actually gives the transducer width and slope at the nodes of
0 1 the finite element model, and the shape within the elements
1 x2 is approximated using the beam shape functions.
B M = 关0 I 兴M f f , 共4兲 The force on the beam from the piezoelectric material is
0 1
given by
1 x3
] ] Q p = C
pqfv p , 共9兲
where v p共t兲 is the voltage across the piezoceramic. Thus
再冎
where 0 is the n ⫻ 2 zero matrix and I is the n ⫻ n identity
matrix, where n is the number of degrees of freedom 共length ü
of the vector q兲. Mq̈ + Dq̇ + Kq − C
pqfv p = − B M . 共10兲
¨
再冎
We then obtain the equations of motion as
ü
Mq̈ + Dq̇ + Kq = Q p − B M , 共5兲
¨ III. SIMPLE ELECTRICAL CIRCUITS FOR THE
ENERGY HARVESTER
where Q p is the force arising from charge on the piezoelec-
tric material. Note that a damping matrix, D, has been in- Figure 2 shows two simple electrical circuits that will be
cluded which represents internal damping in the beam; if modeled to investigate the performance of the energy har-
other damping mechanisms were included, for example, air vester. The governing equations are obtained using Kir-
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
014901-3 M. I. Friswell and S. Adhikari J. Appl. Phys. 108, 014901 共2010兲
f C pqq̇ + f C ps 再冎u̇ 1
+ C pv̇ p + v p = 0, 共11兲
B Mr = rB M /mr . 共18兲
再冎
˙ Rl The equations governing the electrical circuit become,
再冎
without the inductor, or
再冎
equation is
ü 1 1 1 1 ü
fC pqq̈ + fC ps + C pv̈ p + v̇ p + v p = 0, 共12兲 r p̈r + C pv̈ p + v̇ p + v p = − fC ps , 共20兲
¨ Rl L Rl L ¨
Often the energy harvester is designed to be resonant, as where C psu is the first column of C ps.
this increases the power output of the harvester for a fixed For the case with the inductor, Eq. 共25兲, becomes
冉 冊
frequency excitation. Thus one of the natural frequencies of
the harvester corresponds to the excitation frequency, and the i 1
− r 2 P r + − C p 2 + + V p = fC psu2u0 . 共26兲
deformation shape of the beam is dominated by a single Rl L
mode. Suppose the rth mode is excited, then we may ap-
The voltage may be estimated by eliminating the dis-
proximate the response using a single degree of freedom,
placement in Eq. 共24兲 and one of Eqs. 共25兲 and 共26兲, and the
namely the rth principal coordinate pr, so that
amplitude of the harvested power is obtained as
q = r pr , 共15兲
兩V p兩2
where r is the rth mode shape. Substituting this displace- Pout = . 共27兲
Rl
ment into Eq. 共5兲, using Eq. 共9兲 and premultiplying by r,
再冎
gives the single degree of freedom mechanical model as
B. Simple approximation for the harvested
r ü
p̈r + 2rr ṗr + r2 pr − v p = B Mr , 共16兲 power
mr ¨
Designing a harvester for a particular situation is diffi-
where r and r are the rth damping ratio and natural fre- cult because all of the parameters interact. To gain some
quency, mr = rMr is the rth modal mass, and insight a number of simplifying assumptions will be made
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
014901-4 M. I. Friswell and S. Adhikari J. Appl. Phys. 108, 014901 共2010兲
冉 冊
Vp = , 共28兲 sues. As the numerical approximation to the transducer shape
1 is improved by using more finite elements, the number of
c r C pi r + + r2
Rl variables to be optimised 共i.e., the length of f兲 increases lead-
ing to poor conditioning and the presence of many local
where cr = 2rrmr is the rth modal damping parameter. optima. The problem of local optima is compounded by con-
Since the beam parameters are assumed to be fixed, the straints on the width of the transducer, which cannot extend
output voltage, and hence power harvested, for a given out- outside the width of the beam; often the global optimum is
put load and excitation frequency, is a function of the elec- located on one of these constraint boundaries. The condition-
tromechanical coupling parameter, r, and the capacitance of ing of the problem can be improved by also minimising the
the piezoelectric material, C p. Both of these parameters de- curvature of the piezoelectric material width,20 given by
pend on the shape of the piezoelectric transducer; as the area
of coverage of the electrode increases, both C p and r tend to Jc = fEf, 共31兲
increase. For a typical case, where r = 100 Hz, Rl where E is the stiffness matrix of a uniform free-free beam
= 100 k⍀, B Mr = 1, mr = 1 kg, r = 0.01, Fig. 3 shows the with a flexural rigidity of unity. The problem then becomes a
power generated for a range of values of C p and r. It is clear multi-objective optimization that can be solved to obtain a
that the electromechanical coupling, r has an optimum Pareto front that highlights the trade-off between maximising
value, and that the piezoelectric capacitance, C p, should be power and minimising curvature.
minimised. For rectangular transducers, these two require- The difficulty in the optimization problem does suggest
ments are often conflicting, and hence a shaped transducer an alternative, where the transducer shape is decoupled from
may be able to help. Increasing the thickness of the piezo- the approximation inherent in f. The shape may be specified
electric material is another way to reduce the capacitance, using a small number of variables, and for a given set of
but this may not be possible. values of these variables the corresponding f may be ob-
If a mode has a vibration node, then it is important that tained and the power estimated. In this way the number of
the transducer only covers parts of the beam where the cur- variables used in the optimization is not related to the quality
vature has the same sign 共or the transducer could be seg- of the approximation 共given by the number of finite elements
mented兲. used兲. Furthermore promising candidate shapes, such as a
Consider now the system with an inductor. Suppose that rectangular transducer of variable length, may be imple-
the natural frequency of the beam has been tuned to the mented and easily optimised.
excitation frequency, so that = r, as above. Furthermore
assume that the electrical circuit has been tuned so that it
VI. NUMERICAL EXAMPLE
also resonates at the excitation frequency 共neglecting the
electrical resistance and the electromechanical coupling兲, so This example is designed to demonstrate that different
that = e where 2e = 1 / 共LC p兲. The voltage generated is then shaped sensors can influence significantly the power gener-
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
014901-5 M. I. Friswell and S. Adhikari J. Appl. Phys. 108, 014901 共2010兲
15
Sensor 1 (mm)
5
Power (W)
10
0
5
−5 0
Capacitance (pF)
60
Sensor 2 (mm)
5
40
0
20
−5 0
0
Sensor 3 (mm)
Coupling
−0.005
0
−0.01
−5 0 50 100 150
Distance along the beam (mm)
Sensor 4 (mm)
5
FIG. 5. The effect of varying the length of a rectangular transducer.
0
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
014901-6 M. I. Friswell and S. Adhikari J. Appl. Phys. 108, 014901 共2010兲
8
strated that for a rectangular sensor there is a transducer J. M. Renno, M. F. Daqaq, and D. J. Inman, J. Sound Vib. 320, 386 共2009兲.
9
L. Mateu and F. Moll, J. Intell. Mater. Syst. Struct. 16, 835 共2005兲.
length that gives the maximum power. The manner in which 10
N. E. duToit and B. L. Wardle, AIAA J. 45, 1126 共2007兲.
the governing equations are developed, where the transducer 11
A. Erturk and D. J. Inman, Smart Mater. Struct. 17, 065016 共2008兲.
12
shape is determined by a small number of variables, gives A. Erturk and D. J. Inman, J. Vibr. Acoust. 130, 041002 共2008兲.
13
the opportunity to optimise the sensor shape to maximise the A. Erturk and D. J. Inman, J. Intell. Mater. Syst. Struct. 19, 1311 共2008兲.
14
A. Erturk and D. J. Inman, Smart Mater. Struct. 18, 025009 共2009兲.
power output. This is the subject of ongoing research. 15
F. Goldschmidtboeing and P. Woias, J. Micromech. Microeng. 18, 104013
共2008兲.
1
N. S. Hudak and G. G. Amatucci, J. Appl. Phys. 103, 101301 共2008兲. 16
J. M. Dietl and E. Garcia, J. Intell. Mater. Syst. Struct. 21, 633 共2010兲.
2
H. A. Sodano, D. J. Inman, and G. Park, Shock Vib. Dig. 36, 197 共2004兲. 17
J. Schoeftner and H. Irschik, Smart Mater. Struct. 18, 125008 共2009兲.
3 18
S. P. Beeby, M. J. Tudor, and N. M. White, Meas. Sci. Technol. 17, R175 W. Kaal, S. Herold, and M. Kurch, Proceedings of IV ECCOMAS The-
共2006兲. matic Conference on Smart Structures and Materials, Porto, Portugal, July
4
S. R. Anton and H. A. Sodano, Smart Mater. Struct. 16, R1 共2007兲. 2009.
5 19
S. Adhikari, M. I. Friswell, and D. J. Inman, Smart Mater. Struct. 18, C. J. Rupp, A. Evgrafov, K. Maute, and M. L. Dunn, J. Intell. Mater. Syst.
115005 共2009兲. Struct. 20, 1923 共2009兲.
6 20
G. Litak, M. I. Friswell, and S. Adhikari, Appl. Phys. Lett. 96, 214103 M. I. Friswell, J. Sound Vib. 241, 361 共2001兲.
共2010兲. 21
M. I. Friswell and S. Adhikari, Mech. Syst. Signal Process. 24, 623
7
N. G. Stephen, J. Sound Vib. 293, 409 共2006兲. 共2010兲.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp