Cansmart 2009

International Workshop SMART MATERIALS AND STRUCTURES

ELECTROMECHANICAL PERFORMANCES OF DIFFERENT SHAPES OF PIEZOELECTRIC ENERGY HARVESTERS

ABSTRACT In recent years, researchers have shown an interest in the possibility to harvest mechanical energy from vibrating structures. A common way to proceed consists of using the direct piezoelectric effect of a bimorph cantilever beam with integrated piezoceramic elements. Several studies focused on the development of analytical models describing the electromechanical coupling. However, these models were limited to simple structures such as constant cross-section cantilever beam harvester. This paper studies the effect of the harvester geometry on its electromechanical performance. A specific geometry will be of interest in this paper: a tapered beam. A semianalytical model is developed using Rayleigh-Ritz approximations and a trigonometric functions set. Numerical simulations are then performed for three different cases: a standard rectangular harvester, an equivalent mass/stiffness tapered beam harvester and an equivalent maximal strain tapered beam harvester. It will be shown that tapered beams lead to a more uniform strain distribution across the piezoelectric material and could increase the harvesting performance by 69%. Tapered beam harvesters are very interesting since they are low-cost and easy to manufacture. Keywords: Piezoelectric energy harvesting, strain distribution, harvester shape, electromechanical modeling

187

2009 Cansmart Workshop

INTRODUCTION Energy harvesting from vibrating mechanical structures has been studied by several researchers in the last decade [1][2]. The energy harvesting performances was predicted for simple geometries such as constant thickness beams and plates. Williams and Yates [3] established a single degree-of-freedom mechanical model having a damper as a dissipative element to represent the energy harvester. Then, Roundy et al. [4] analytically represented piezoelectric bimorph harvester by an electrical circuit were the mechanical elements where represented by their electrical equivalent and solved using Kirchoff law. Later, Sodano et al. [5] developed the constitutive equations for a piezoelectric bimorph harvester using a variational approach. With their model, the electromechanical performance of the harvester (which is a continuous system) is described by a set of two coupled matrix equations. All these work are based on a constant cross-section cantilever beam. Baker et al. [6] proposed to vary the width of the beam (trapezoidal shape) in order to increase the efficiency. By doing so, they reported a 30% increase in energy harvesting. There is no reason for which the geometry should be limited to rectangular and trapezoidal configurations. In this paper, the thickness of the beam is varied (triangular shape) in order to obtain a more uniform strain distribution along the beam, hence to maximize the harvested energy. In the following, a semi-analytic model is briefly described. The model takes both, the mechanical and electromechanical effects into account. A short numerical study is then presented for three different harvester shapes. MECHANICAL MODELING PROCEDURE Fig. 1 shows the harvester under study. It consists of a cantilever tapered beam, two bonded piezoelectric ceramics and two masses at its end. to and tf are the thicknesses of the beam respectively at the fixed-end and at the free-end. tp is the thickness of each piezoceramic plate and tm is the thickness of the masses. L is the length of the beam, L1 is the tapered length and L2 is the length along the beam for which the thickness is constant. Finally, is the taper angle.

Fig. 1: Geometrical model of the tapered beam energy harvester. 2009 Cansmart Workshop 188

Mechanical dynamic model Flexural vibrations of the beam are modeled using Bernouillis assumptions. The displacement field {ux uy uz} can thus be written as: w( x, t ) ux = y x u y = w( x, t ) (1) uz = 0 The vector {ux, uy, uz} represents the displacement of a point either on the beam, the piezoceramic elements or the tip masses. Perfect bonding is considered between the beam and the piezoelectric elements leading to a continuity of displacements at these interfaces. Rayleigh-Ritz approximations are used to solve the system with the general following series expansion:

w( x, t ) = Fi ( x)Gi (t )
i =1

(2)

where Fi(x) are the spatial functions and Gi(t) are the time functions used to describe the displacement. The analytical formulation is based on the variational approach, in which the energy of the whole system is minimized using the Lagrangian. Using the coefficient Gi(t) as the generalized coordinates, Lagrange equations can be written in the general form:
d T & dt Gi T U + =0 Gi Gi

(3)
( i = 1,2,..., N)

where T represents the total kinetic energy of the system and U the total potential energy of the system. Kinetic energy of each element (beam, piezoceramics and tip masses) is described by:
2 w( x, t ) 1 L 1 L w( x, t ) (4) Ti = i Ai ( x) dx + i Ii ( x) dx 2 0 2 0 t xt where i represent the density, Ai(x) and Ii(x) are respectively the cross-section area and the area moment of inertia of each element.
2 2

The first term of (4) represents the kinetic translational energy of each component while the last term represents the kinetic rotational energy of each component. Resolution by Rayleigh-Ritz method allows the introduction of the kinetic rotational energy functions which increases the accuracy of the model. For a tapered beam, rotational energy is greater at the fixed-end than at the free-end of the beam. Thus, rotational energy does have a significant impact on the strain distribution and must be taken into account. These rotational energy functions assume that piezoceramic cross-sections remain perpendicular to the neutral axis. Since the slope angle will remain small, this assumption does not influence significantly the accuracy of the model.

189

2009 Cansmart Workshop

For the whole system, the total kinetic energy is:

(5)

where the subscripts b, p and m indicate respectively the beam, the piezoceramics and the tip masses. Potential energy of each element can be defined as: 1 L (6) Ui = xx xx dAdx i 2 0 Ai with 2 w( x, t ) xx = y (7) and xx = Ei xx x 2 where Ei is the modulus of elasticity of each element. Using (7) and (6) leads to a simplified form of the potential energy of each element:
2 w( x, t ) 1 L Ui = Ei Ii ( x) dx 2 2 0 x
2

(8)

For the whole system, the total potential energy is therefore given by :
U = Ub + U p + U m

(9)

Inserting (5) and (9) into Lagrange equations (3) gives the following differential equations system: && (10) MG + KG = 0 where M and K are respectively the mass and the stiffness matrix. M and K depend on the spatial functions Fi(x) which are defined in the next section. By solving the corresponding eigenvalue problem, it is possible to find the natural frequencies [1, 2, , N] and the mode shapes [p1 p2 pN] of the system. The first mode shape p1 will be later used to simplify the model to one degree-of-freedom. Trigonometric functions set The spatial functions set used in this paper is trigonometric and defined as follow: (2i 1) Fi ( ) = 1 cos 2 where the adimensional axial position =x/L is used.

(11)

Fig. 2 shows the first four functions of this set. This functions set has many advantages including: its simplicity decreases the required computing time, it is bounded between 0 and 2 which avoid truncation errors, it respects both geometrical (Fi(=0) = Fi(=0) = 0) and one natural (Fi(=1) =0) boundary conditions. Moreover, the shape of the first function F1(x) is very close to the first mode of a cantilever beam which helps in reducing the number of functions required to well approximate the first mode.

2009 Cansmart Workshop

190

Fig. 2: Trigonometric functions set: a) i=1 b) i=2 c) i=3 and d) i=4.

ENERGY HARVESTING THEORY The basic principle of vibration energy harvesting is to tune the first natural frequency of the harvester to the frequency of the vibration source. As described by Liao et al. [7], the dynamic model presented in the previous section can be reduced to one degree-of-freedom by considering the first mode only. The direct piezoelectric effect must be taken into account as illustrated in Fig. 3.

Fig. 3: Single degree of freedom electromechanical model. The spatial function (x) corresponding to the first mode can be determined with the spatial functions set Fi(x) and the first mode shape p1, i.e.: ( x) = [ F1 ( x) F2 ( x) L FN ( x)] p1 (11) The transverse displacement can then be written as: w( x, t ) = ( x )h(t ) where h(t) is the time response of the first mode. (12)

By considering the direct piezoelectric effect, one can show that the electromechanical behaviour of the harvester can be modeled with the two following equations [7]: && & M eq h (t ) + Ceq h(t ) + K eq h(t ) v(t ) = Da(t ) (13)
h(t ) + C p v(t ) = q (t )

(14)

with M eq = M b + M p + M m = b Ab ( x) ( x) ( x) dx + b I b ( x) '( x) '( x) dx + p Ap ( x) ( x) ( x) dx +

L1

p I p ( x) '( x) '( x)dx + m Am ( x) ( x) ( x)dx + m I m ( x) '( x) '( x) dx

191

2009 Cansmart Workshop

K eq = K b + K p + K m = Eb I b ( x) ''( x ) ''( x)dx + E p I p ( x) ''( x) ''( x)dx + Em I m ( x) ''( x) ''( x) dx

Ceq = 2 K eq M eq

(15)
L1 L 0 L1

( t f to ) to + t p L1 x+ = 2d31 E p b ''( x) 0 2 L1 2 2 K 3S 0 L1 Cp = Ap ( x)dx 2 t p 0

D = b Ab ( x) ( x )dx + p Ap ( x ) ( x ) dx + m Am ( x ) ( x ) dx
0

dx

In these expressions, a(t), v(t) and q(t) are respectively the base acceleration, the output voltage and the electrical charge. Meq, Ceq and Keq are the equivalent mass, damping and stiffness of the system while D is the inertial loading applied on the system. and Cp are respectively the electromechanical coupling coefficient and the piezoceramics equivalent capacitance. Finally, is the mechanical damping ratio, K3S0 is the permittivity of the piezoceramics and d31 is the piezoelectric charge constant. The harvested energy is dissipated through a resistive load R. Using Ohms law, one obtains the following additional equations: & (16) v(t ) = Rq (t ) By combining (13), (14) and (16), it is possible to obtain a third order linear differential equation and then to determine the power dissipated across the resistive load.

PERFORMANCE STUDY OF DIFFERENT SHAPES OF HARVESTER

In this section, numerical simulations are presented to demonstrate the increase in performances when using tapered beam harvester instead of constant cross section harvester. The general harvester under study consists of a tapered brass beam, two steel masses and two piezoceramics (PIC 151 manufactured by PhysikInstruments) as shown previously in Fig. 1. Material properties of each element are summarized in table 1. For all case under study, the width b and length L of the beam are respectively 25 mm and 90 mm, while the piezoceramics dimensions are (70x25x0.5) mm3. A typical value of 2% is used for the mechanical damping ratio. The input excitation is a base acceleration a(t) having an amplitude of 9.81 m/s2. Table 2 summarizes the three cases under study: a standard rectangular harvester, an equivalent mass/stiffness tapered beam harvester and an equivalent maximal strain tapered beam harvester. They all share a common property: the first short circuit natural frequency is 100Hz which is an arbitrary design frequency. In order to evaluate model quality and the results accuracy, the harvester structure was modeled using finite element with NxNastran 5. The system being symmetrical with respect to the x axis, a 2D grid composed of CQUAD8 was used to perform a modal analysis. Finite

2009 Cansmart Workshop

192

element results agreed very well with the prediction of our model. Due to length restriction, those results are unfortunately not presented in this paper.
Table 1: Materials properties.
Properties Brass modulus of elasticity PIC 151 elasticity Steel elasticity Brass density PIC 151 density Steel density Relative permittivity of ceramics Piezoelectric charge constant Symbol Eb Ep Em b p m Value 100 GPa 66.7 GPa 200 GPa 8740 kg.m-3 7800 kg.m-3 7870 kg.m-3 2068 -210 pC.N-1

K 3S
d31

Table 2: Energy harvesters geometrical parameters.

Case I: Constant thickness beam The first case under study consists of a constant thickness beam energy harvester (=0o). This case represents the one generally proposed in the literature. Fig. 4a) shows the dissipated power as a function of the excitation frequency and the electrical resistance. A maximal dissipated power of 11.5 mW is observed for a 100.74 Hz excitation and a 6.08 k resistive load. Fig. 4b) shows the strain distribution across the piezoceramics for these values. A maximal strain of 1.3710-4 mm/mm is observed at the fixed-end of the beam (x=0) and the strain decreases in a quasi linear manner from the fixed-end to the free-end.

Fig. 4: Case I: a) Harvested power as a function of the frequency and the resistive load and b) piezoceramic strain distribution for optimal resistance and frequency. 193 2009 Cansmart Workshop

Case II : Tapered beam with = 0.3o (same Meq and Keq) The second case under study is a tapered beam energy harvester having a taper angle of 0.3o. Dimensions of the beam were chosen such that the equivalent mass Meq, the equivalent stiffness Keq and the first natural frequency 1 are identical to Case I. Fig. 5a) shows the dissipated power as a function of both the excitation frequency and the resistive load. A maximal dissipated power of 10.2 mW is achieved for a 100.68 Hz excitation and a 4.96 k resistive load. Amazingly, the dissipated power is less than case I. Fig. 5b) compares strain distribution for case I (dashed line) and case II (continuous line), both for optimal resistive load and optimal excitation frequency. It can be observed that the maximal strain is 23% less for case II (1.0510-4 mm/mm instead of 1.3710-4 mm/mm) and that the strain distribution is more uniform.

Fig. 5: Case II: a) Harvested power as a function of the frequency and the resistive load and b) piezoceramic strain distribution for optimal resistance and frequency.

Case III: Tapered beam with = 0.3o (same maximal strain) In order to rigorously compare the performance of two harvesters, they must have the same maximal deformation. Therefore, we modified case II by increasing both the equivalent mass and stiffness in order to keep the same natural frequency but increase the maximal strain to the one of case I. As shown in Fig. 6a), a maximal dissipated power of 19.4 mW (69% energy harvesting increases) can be obtain for a 101.06 Hz excitation frequency and a 6.12 k resistive load. Fig. 6b) compares the strains distributions for case I (dashed line) and case III (continuous line), both for optimal resistive load and optimal excitation frequency. For case III, the strain distribution is more uniform which increases the performance of the harvester. From these results, one can observe that the strain distribution can be optimized by modifying the geometry of the harvester.

2009 Cansmart Workshop

194

Fig. 6: Case III: a) Harvested power as a function of the frequency and the resistive load and b) piezoceramic strain distribution for optimal resistance and frequency.

CONCLUSION

The main objective of this paper was to demonstrate that the efficiency of a cantilever beam vibration energy harvester can be increased by using a variable thickness. A semianalytical model that takes both the dynamic and the electromechanical behaviour of the harvester into account was first described. Using this model, it has been shown that the energy harvested can be increased by 69% when the thickness of the beam is varied (0.3o slope angle). It also leads to a more uniform strain distribution and thus increases the amount of energy that can be harvested. Moreover, the proposed geometry is very easy to manufacture and standard commercially available rectangular piezoceramic elements can be used. Further work will investigate the optimal slope angle, i.e. the one leading to the most uniform stain distribution and will propose design guidelines for cantilever beam vibration energy harvester.

ACKNOWLEDGEMENTS

This research was supported by Fonds de recherche sur la nature et les technologies (FQRNT) scholarships and by NSERC New Discovery Grant. Our thanks to Oliver StAmand for his useful help.
REFERENCES

1. 2. 3. 4.

Anton S.R., Sodano H.A., A review of power harvesting using piezoelectric materials (2003-2006), Smart Mater. Struct. 16, R1-21, (2007) Priya S., Inman D.J., Energy harvesting technologies, 1st edition, Springer, 2009 Williams C.B., Yates R.B., Analysis of a micro-electro generator for micro systems, Sensors and Actuators A52, 8-11, 1996 Roundy S., Wright P.K, A piezoelectric vibration based generator for wireless electronics, Smart Mater. Struct. 13, 1131-1142, 2004

195

2009 Cansmart Workshop

Sodano H.A., Park G., Inman, D.J., Estimation of electric charge output for piezoelectric energy harvesting, Journal of strain 40, 49-58, 2004 6. Baker J., Roundy S., Wright P.K, Alternatives geometries for increasing power density in vibration energy scavenging for wireless sensor networks, AIAA 2005, 3rd international Energy Conversion Engineering Conference, 15-18 August 2005, 12 pp 7. Liao Y., Sodano H.A., Model of a single mode energy harvester and properties for optimal, power generation, Smart Mater. Struct. 17, 14 pp, 2008

5.

2009 Cansmart Workshop

196