ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no.

7, july 2003 773

Transmitting Electric Energy Through a Metal

Wall by Acoustic Waves Using Piezoelectric
Transducers
Yuantai Hu, Xuesong Zhang, Jiashi Yang, and Qing Jiang

Abstract—The feasibility of transmitting electric energy

through a metal wall by propagating acoustic waves using
piezoelectric transducers is examined by studying the effi-
ciency of power transmission and its dependence upon the
relevant system parameters for a simplified system consist-
ing of an elastic plate sandwiched by two piezoelectric lay-
ers. One of these layers models the driving transducer for
generating acoustic wave, and the other layer models the re-
ceiving transducer for converting the acoustic energy into
electric energy to power a load circuit. The output voltage,
the output power, and the efficiency of this system are ex-
pressed as explicit functions of the system parameters. A
numerical example is included to illustrate the dependence
of the system performance upon the physical and geomet-
rical parameters. Fig. 1. An elastic plate sandwiched by two piezoelectric transducers.

ergy into electric energy to power a load circuit, which is

I. Introduction
characterized by its impedance ZL as indicated in Fig. 1.
In Section II we present a mathematical formulation that
here has been recent interest in periodically charging
T and/or recharging batteries that power electronic de-
vices operating in a sealed armor or other environments,
governs thickness-stretch vibrations for this system under
a harmonic electric voltage source connected to the driving
transducer. Our analysis presented in Section III leads to
such as nuclear storage facilities, into which the physical
explicit representations for the output voltage, the output
access is prohibited. For instance, to ensure the reliability
power, and the efficiency of this system. An illustrative
and performance of the nuclear stockpile in the absence
numerical example is included in Section IV. This study
of nuclear testing, there have been proposals that piezo-
indicates that the output voltage of this system is mag-
electric transducers are used to generate acoustic waves
nified at the first several resonant frequencies of the sys-
propagating through the sealed armors for transmitting a
tem. We note with particular interest that the greatest
small amount of power to the electronic devices inside the
magnification does not occur at the fundamental resonant
sealed armors. In the present paper, we examine the feasi-
frequency and that the vibration mode of greatest mag-
bility of such proposals by studying the efficiency of power
nification varies with the system parameters. The system
transmission and its dependence upon the relevant sys-
efficiency in transmitting the energy peaks at selected driv-
tem parameters. In this preliminary analysis, we study the
ing frequencies, which do not necessarily coincide with the
system illustrated in Fig. 1, in which a metal plate, repre-
system resonant frequencies.
senting the metal wall of a sealed armor, is sandwiched by
two piezoelectric layers. These piezoelectric layers model
two piezoelectric transducers. One of these layers is for II. Thickness-Stretch Vibrations of the
generating acoustic waves driven by a prescribed electric Sandwiched Plate
voltage source; the other is for converting the acoustic en-
To set up a mathematical formulation for the proposed
Manuscript received September 8, 2002; accepted January 17, 2003. problem, we model the wall of a sealed armor by a linear
This work was supported by the Office of Naval Research under con-
tract number ONR N00014-96-1-0884. elastic plate and the transducers by two linear piezoelec-
Y. T. Hu is with the Department of Mechanics, Huazhong Uni- tric layers. To maximize the electromechanical coupling
versity of Science and Technology, Wuhan 430074, China, and the along the thickness direction along which acoustic waves
State Key Laboratory of Structural Analysis of Industrial Equip-
ment, Dalian University of Technology, Dalian 116023, China. are propagated, we let the piezoelectric layers be both po-
X. Zhang and J. S. Yang are with the Department of Engineer- larized along the x3 axis, as indicated in Fig. 1. Consider-
ing Mechanics, University of Nebraska, Lincoln, NE 68588 (e-mail: ing that the in-plane dimensions of this structure are sub-
[email protected]).
Q. Jiang is with the Department of Mechanical Engineering, Uni- stantially larger than the thickness, we neglect the edge
versity of California, Riverside, CA 92521. effect in this preliminary analysis by assuming that all the

0885–3010/$10.00

c 2003 IEEE
774 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 7, july 2003

variables are independent of the in-plane coordinates, x1 The driving transducer is electroded on its outer sur-
and x2 , and that the longitudinal acoustic wave, generated face at x3 = h0 + h1 . The electrode is subjected to a time-
by a harmonic electric voltage source, propagates along the harmonic driving voltage V1 (t), and the corresponding me-
thickness direction of the sandwiched plate structure [1] chanical condition is traction-free. This requires:
and is bounced back and forth between the outer surfaces
of the two piezoelectric transducers. From the viewpoint of T3j = 0, φ = V1 (t), on x3 = h0 + h1 . (4)
the theory of plate vibration, the system is driven, equiva-
lently, into thickness-stretch vibrations by a harmonic volt- Owing to (3), the boundary conditions (4) take the fol-
age excitation. In the following, we present the differential lowing form:
equations and the boundary conditions as well as the in-
terface conditions that govern thickness-stretch vibrations T33 = c33 u3,3 + e33 φ,3 = 0, φ = V1 (t), on x3 = h0 + h1 .
for this system. (5)
Motions of the piezoelectric transducers are governed
by the following equations of the linear theory of piezo- The current density flowing out of the driving electrode
electricity [2]: at x3 = h0 + h1 is given by:

Tji,j + ρfi = ρüi , Di,i = 0, j1 = −Q̇1 , Q1 = −D3 (h0 + h1 , t) ,

Tij = cijkl Skl − ekij Ek , Di = eijk Sjk + εSij Ej , D3 (h0 + h1 , t) = e33 u3,3 (h0 + h1 , t)−ε33 φ,3 (h0 + h1 , t) ,
Sij = (ui,j + uj,i ) /2, Ei = −φ,i , (1) (6)

where ui , Ei , and Di are the components of the mechanical where Q1 is the charge per unit area on the driving elec-
displacement vector, the electric field vector, and the elec- trode at x3 = h0 + h1 .
tric displacement vector; Tij and Sij are the components The receiving transducer also is electroded on its outer
of the stress tensor and the strain tensor, respectively; and surface at x3 = − (h0 + h2 ). This electrode has an output
φ denotes the electric potential; cijkl , ekij and εSij are the voltage V2 (t), and the corresponding mechanical condition
elastic, piezoelectric, and dielectric constants. We denote is also traction-free. Thus, we have the following boundary
by ρ the mass density. The Cartesian tensor notation, the conditions:
summation convention for repeated tensor indices, and the
convention that a comma followed by an index denotes T3j = 0, φ = V2 (t), on x3 = − (h0 + h2 ) . (7)
partial differentiation with respect to the coordinate asso-
ciated with the index are used. A superimposed dot repre- Using the equations in (3), one can rewrite these bound-
sents time derivative. With the compact matrix notation ary conditions as follows:
[2], cijkl and ekij can be written as cpq and ekp , with the
convention that p, q = 1, 2, 3, 4, 5, 6. For ceramics poled T33 = c33 u3,3 + e33 φ,3 = 0,
in the positive x3 directions, we have [3]: φ = V2 (t), on x3 = − (h0 + h2 ) . (8)
   
c11 c12 c13 0 0 0 0 0 e31
c21 c11 c13 0 0 0   0 0 e31  Furthermore, one can show that the electric displace-
   
c31 c31 c33 0 0 0   0 0 e33  ment D3 inside the receiving transducer is spatially con-
 , 
 0 0 0 c44 0 0   0 e15 0  , stant, i.e.:
   
 0 0 0 0 c44 0  e15 0 0 
0 0 0 0 0 c66 0 0 0 D3,3 = e33 u3,33 − ε33 φ,33 = 0. (9)
 S 
ε11 0 0
 0 εS11 0  , (2) Thus, we write:
0 0 εS33 D3 (x3 , t) = f (t), (10)
where c66 = (c11 − c12 ) /2. The superscript S of the di-
electric tensor components εSij indicates that the dielectric where f (t) is an undetermined function of time. The cur-
constants should be measured under the fixed strain con- rent density on the receiving electrode at x3 = − (h0 + h2 )
dition. In the following, we drop this superscript for sim- is given by:
plicity. For thickness-stretch vibrations of a piezoelectric
layer [4]–[6], (1) and (2) reduce to: j2 = −Q̇2 , Q2 = D3 (−h0 − h2 , t) = f (t), (11)

c33 u3,33 + e33 φ,33 = ρü3 where Q2 is the charge per unit area on the receiving elec-
e33 u3,33 − ε33 φ,33 = 0, trode at x3 = − (h0 + h2 ).
The middle layer occupying −h0 ≤ x3 ≤ h0 is metallic
E3 = −φ,3 , (3)
and conducting, hence the electric potential φ is spatially
T33 = c33 u3,3 − e33 E3 , constant. We let the middle layer be grounded, hence the
D3 = e33 u3,3 + ε33 E3 . electric potential φ vanishes identically within the middle
hu et al.: acoustic waves, metal walls, and transmitting energy 775

layer. We thus have the following governing equations for sions of the electric potential Φ in terms of the thickness
this elastic layer in thickness-stretch vibrations: displacement U3 through integrating (3)2 :

c033 U3,33 + ρ0 ω 2 U3 = 0, −h0 ≤ x3 ≤ h0

c033 u3,33 = ρ0 ü3
(12) c33 U3,33 + ρω 2 U3 = 0,
T33 = c033 u3,3 .
− h0 − h2 ≤ x3 ≤ −h0 or h0 ≤ x3 ≤ h0 + h1 ,

x3 − h0 e33
 
On the two interfaces, we require that the displacement Φ= V1+ U3 (x3 ) − U3 h+ 0
h1 ε33
u3 , the electric potential φ, and the mechanical tractions 
x3 − h0
 
T3i are continuous, i.e.: − U3 (h0 + h1 ) − U3 h+ 0 ,
h1 (16)
   −  h0 ≤ x3 ≤ h0 + h1 ,
u3 h+0 , t = u3 h0 , t ,
   −  Φ = 0, −h0 ≤ x3 ≤ h0 ,
u3 −h+ 0 , t = u3 −h0 , t ,
 − 
   −  Φ = e33 ε−1
33 U3 (x3 ) − U3 −h0 + ε−1
33 (x3 + h0 ) F,
φ h+0 , t = φ −h0 , t = 0,
  − h0 − h2 ≤ x3 ≤ −h0 ,
T33 h+
0 , t = [c33 u3,3 + e33 φ,3 ] h+ 

0  
0
 where c33 = c33 + e233 ε33 . It follows from (15) that all

= c33 u3,3 h− = T33 h− (13)
0 ,t , the boundary and continuity conditions on the interfaces
 −  0

T33 −h0 , t = [c33 u3,3 + e33 φ,3 ] −h− at x3 = ±h0 are satisfied provided that:

 
0
  
= c033 u3,3 −h+ = T33 −h+ 0 ,t ,
1 2 −1
 
0 0 −
c33 U3,3 h+ e33 ε33 U3 (h1 + h0 ) − U3 h+
0
h1
V1  
where c033 is the elastic constant of the elastic layer and ρ0 = − e33 + c033 U3,3 h− 0 ,
h1 (17)
is its mass density.  −   −1
c33 U3,3 −h0 = c033 U3,3 −h+0 − ε e
33 33 F.
Considering that the system operates under a time-
harmonic driving voltage, we use the following complex We now introduce the output current I2 = J2 S, where S
notation for convenience: denotes the electroded surface area of the receiving trans-
ducer. The receiving transducer is connected to a load cir-
cuit, which represents the battery to be charged and the
[u3 , φ, V1 (t), V2 (t), j1 (t), j2 (t), f (t)] = circuit for charging the battery. Under a harmonic excita-


Re U3 (x3 ) , Φ (x3 ) , V 1 , V 2 , J1 , J2 , F eiωt . (14) tion, the effect of this load
 circuit can be characterized by
its impedance ZL = V 2 I2 .

Substitution of (14) into (13) yields the continuity con-

ditions in terms of the complex quantities: III. Steady-State Characteristics of the System

   − To evaluate the effectiveness of this system, we turn

U3 h+ 0 = U3 h0 ,
   − now to seek the steady-state solution of this system un-
U3 −h+ 0 = U3 −h0 , der a time-harmonic excitation voltage. We outline in this
   −
0 = Φ −h0 = 0,
Φ h+ section the major steps of the solution procedure, neglect-

 ing the detailed calculations, which are very lengthy but
[c33 U3,3 + e33 Φ,3 ] h+ = c033 U3,3 h− , (15)
straightforward.

0  0 0

c33 U3,3 −h+ = [c33 U3,3 + e33 Φ,3 ] −h− . We first integrate the differential equations for the
0 0
thickness displacement U3 in (16)1,2 . Then, using the inter-
face conditions (17), we obtain the solution for the thick-
Next we integrate (3)2 with respect to x3 twice over the ness displacement U3 as follows:
ranges: h0 ≤ x3 ≤ h0 +h1 and − (h0 + h2 ) ≤ x3 ≤ −h0 , re- U3 (x3 ) =
spectively, to obtain the expressions of the electrical volt- 
age Φ in terms of U3 , within the driving and receiving 
A1 sin ηx3 + B1 cos ηx3 , h0 ≤ x3 ≤ h0 + h1 ,



transducers, respectively. These expressions are then sub- A2 sin η x3 + B2 cos η x3 , −h0 ≤ x3 ≤ h0 ,


(18)
stituted into (3)1 to eliminate the electric potential Φ and A3 sin ηx3 + B3 cos ηx3 , −h0 − h2 ≤ x3 ≤ −h0 ,
to obtain the differential equations in which the thickness
displacement U3 is the sole unknown function. We collect, where Am , Bm , m = 1, 2, 3 are three pairs of undeter-
in the following, the equation of motion for the metal plate mined constants. The constants A1 , B1 , A3 , and B3 can
(12)1 , the equations of motion for the transducers by elim- be expressed in terms of the driving voltage V 1 , the elec-
inating the electric potential Φ from (3)1 , and the expres- tric displacement F within the receiving transducer, and
776 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 7, july 2003

A1 = α12 ∆−1 h−1

1 e33 V 1 + ∆
−1
[γ11 α22 − γ21 α12 ] A2 + ∆−1 [γ12 α22 − γ22 α12 ] B2 ,
B1 = −α11 ∆−1 h−1
1 e33 V 1 + ∆
−1
[−γ11 α21 + γ21 α11 ] A2 + ∆−1 [−γ12 α21 + γ22 α11 ] B2 ,
(19)


A3 = α12 ∆−1 −1 −1



−1




1 ε33 e33 F + ∆1 [γ11 α22 − γ21 α12 ] A2 + ∆1 [γ12 α22 − γ22 α12 ] B2 ,


B3 = −α11 ∆−1 −1 −1



−1




1 ε33 e33 F + ∆1 [−γ11 α21 + γ21 α11 ] A2 + ∆1 [−γ12 α21 + γ22 α11 ] B2 ,

where:
  1   1
η = ρω 2 c33 2 , η
= ρ0 ω 2 c033 2 ,
α11 = sin ηh0 , α21 = c33 η cos ηh0 − h−1 2 −1
1 e33 ε33 [sin η (h0 + h1 ) − sin ηh0 ] ,
α12 = cos ηh0 , α22 = −c33 η sin ηh0 − h−1 2 −1
1 e33 ε33 [cos η (h0 + h1 ) − cos ηh0 ] ,
γ11 = sin η
h0 , γ12 = cos η
h0 , γ21 = c033 η
cos η
h0 , γ22 = −c033 η
sin η
h0 , (20)





α11 = sin ηh0 , α12 = − cos ηh0 , α21 = c33 η cos ηh0 , α22 = c33 η sin ηh0 ,


γ11 = sin η
h0 , γ12


= − cos η
h0 , γ21


= c033 η
cos η
h0 , γ22


= c033 η
sin η
h0 ,





∆ = α11 α22 − α12 α21 , ∆1 = α11 α22 − α12 α21 .

the undetermined constants A2 and B2 as given in (19) where:

and (20) (see above).
Substituting the complex notation (14) and the expres- ∆2 = µ11 µ22 − µ12 µ21 . (26)
sions for the electric potential Φ within the transducers
Substituting (25) into (19) yields the following expres-
and the metal plate given by (16), as well as the expres-
sions for the constants A1 , B1 , A3 , and B3 in terms of the
sions for the thickness displacement U3 given by (18), into
driving voltage V 1 and the thickness component of the
the traction free boundary conditions (5) and (8) yields
electric displacement F :
(21) (see next page).
For convenience, we define the parameters in (22) (see A1 = χ11 V 1 + χ12 F, B1 = χ21 V 1 + χ22 F,
next page).
Substituting the expressions for the constants A1 , B1 , A3 = χ31 V 1 + χ32 F, B3 = χ41 V 1 + χ42 F, (27)
A3 , and B3 given in (19) into the boundary conditions
(21), and noting the definitions of τ11 , τ12 , τ21 , and τ22 where definitions are given in (28) (see next page).
given by (22), we obtain: With the expressions for the electric potential Φ in (16),
the expressions for the thickness displacement U3 in (18)
µ11 A2 + µ12 B2 = ϑ1 V 1 , and (27), we obtain the spatial distribution of the electric
(23)
µ21 A2 + µ22 B2 = ϑ2 F, potential Φ within the receiving transducer as given in (29)
(see next page).
where
Thus, the output voltage is given by:
µ11 = ∆−1 [τ11 (γ11 α22 − γ21 α12 )
+ τ12 (−γ11 α21 + γ21 α11 )], V 2 = Φ (−h2 − h0 ) = Γ1 V 1 − Z2 I2 . (30)

µ12 = ∆−1 [τ11 (γ12 α22 − γ22 α12 ) In (30), the thickness component of the electric displace-
+ τ12 (−γ12 α21 + γ22 α11 )], ment F appearing in (29) has been replaced by the electric
current flowing into the receiving electrode: I2 = J2 S =
µ21 = ∆−1



1 [τ21 (γ11 α22 − γ21 α12 )


−iωF S, and the coefficients Γ1 and Z2 are given in (31)





+ τ22 (−γ11 α21 + γ21 α11 )], (see page 778).
(24)
µ22 = ∆−1




1 [τ21 (γ12 α22 − γ22 α12 )
The driving current I1 = J1 S is related to the driving





+ τ22 (−γ12 α21 + γ22 α11 )], voltage V 1 by:

ϑ1 = −∆−1 h−1
1 e33 (∆ + τ11 α12 − τ12 α11 ), I1 = −V 1 Z1 + Γ2 I2 , (32)
ϑ2 = −∆−1 −1


1 ε33 e33 (∆1 + τ21 α12 − τ22 α11 ). with the coefficients Z1 and Γ2 given in (33) (see page
We then solve (23) and obtain the following expressions 778).
for the constants A2 and B2 in terms of the driving voltage In (30) and (32), the coefficients Γ1 and Γ2 are two di-
V 1 and the electric displacement F : mensionless numbers; the coefficients Z1 and Z2 have the
  dimension of impedance. Physically, from (32) it can be
A2 = ∆−1 µ22 ϑ1 V 1 − µ12 ϑ2 F , seen that Z1 may
 be considered as the input impedance
2
  (25) 
B2 = ∆−12 −µ21 ϑ1 V 1 + µ11 ϑ2 F , of the system −V 1 I1 when the output circuit is open
hu et al.: acoustic waves, metal walls, and transmitting energy 777


c33 η cos η (h1 + h0 ) − h−1 2 −1
1 e33 ε33 [sin η (h1 + h0 ) − sin ηh0 ] A1

− c33 η sin η (h1 + h0 ) + h−1 2 −1 −1
1 e33 ε33 [cos η (h1 + h0 ) − cos ηh0 ] B1 = −h1 V 1 e33 , (21)
c33 η cos η (h2 + h0 ) A3 + c33 η sin η (h2 + h0 ) B3 = −ε−1
33 e33 F.

τ11 = c33 η cos η (h1 + h0 ) − h−1 2 −1

1 e33 ε33 [sin η (h1 + h0 ) − sin ηh0 ] ,
τ12 = −c33 η sin η (h1 + h0 ) − h−1 2 −1
1 e33 ε33 [cos η (h1 + h0 ) − cos ηh0 ] , (22)
τ21 = c33 η cos η (h2 + h0 ) , τ22 = c33 η sin η (h2 + h0 ) .

χ11 = α12 ∆−1 h−1

1 e33 + ∆
−1 −1
∆2 ϑ1 [µ22 (γ11 α22 − γ21 α12 ) − µ21 (γ12 α22 − γ22 α12 )] ,
χ12 = ∆−1 ∆−1
2 ϑ2 [µ11 (γ12 α22 − γ22 α12 ) − µ12 (γ11 α22 − γ21 α12 )] ,
χ21 = −α11 ∆−1 h−1
1 e33 + ∆
−1 −1
∆2 ϑ1 [µ22 (γ21 α11 − γ11 α21 ) − µ21 (γ22 α11 − γ12 α21 )] ,
χ22 = ∆−1 ∆−1
2 ϑ2 [µ11 (γ22 α11 − γ12 α21 ) − µ12 (γ21 α11 − γ11 α21 )] ,
(28)
χ31 = ∆−1 −1








1 ∆2 ϑ1 [µ22 (γ11 α22 − γ21 α12 ) − µ21 (γ12 α22 − γ22 α12 )] ,


χ32 = α12 ∆−1 −1 −1 −1








1 ε33 e33 + ∆1 ∆2 ϑ2 [µ11 (γ12 α22 − γ22 α12 ) − µ12 (γ11 α22 − γ21 α12 )] ,
χ41 = ∆−1 −1








1 ∆2 ϑ1 [µ22 (γ21 α11 − γ11 α21 ) − µ21 (γ22 α11 − γ12 α21 )] ,


χ42 = −α11 ∆−1 −1 −1 −1








1 ε33 e33 + ∆1 ∆2 ϑ2 [µ11 (γ22 α11 − γ12 α21 ) − µ12 (γ21 α11 − γ11 α21 )] .

Φ (x3 ) = e33 ε−1

33 [χ31 (sin ηx3 + sin ηh0 ) + χ41 (cos ηx3 − cos ηh0 )] V 1
 −1 (29)
+ ε33 (x3 + h0 ) + e33 ε−1
33 [χ32 (sin ηx3 + sin ηh0 ) + χ42 (cos ηx3 − cos ηh0 )] F.

(I2 = 0). We note that (30) and (32) are in fact general with the load circuit impedance ZL and the input ad-
input-output relations for a linear system. Eq. (31) and mittance approaches a constant. For large loads, the out-
(33) are specific expressions of the coefficients in the lin- put voltage approaches Γ1 , and the input admittance ap-
ear relations, which are determined by the specific struc- proaches 1/Z1 . The input and output powers are given by:
ture shown in Fig. 1. Discussions in the rest of this section
1 ∗
 1 ∗

depend on the general form of (30) and (32), but not on P1 = I1 V 1 + I1∗ V 1 , P2 = I2 V 2 + I2∗ V 2 ,
the special expressions in (31) and (33). Because the driv- 4 4
(35)
ing voltage V 1 is prescribed for the system, solving (30)
and (32) and using the load circuit impedance relation where an asterisk indicates complex conjugate. Then the
ZL = V 2 I2 , we obtain the normalized output voltage, efficiency of the power transmission, defined as the quo-
the output current, and the input admittance as follows: tient P2 P1 , is given by (36) (see next page). Eq. (36)
shows that the system efficiency ν increases with the load
V2 Γ1 ZL
= , impedance ZL linearly for small loads, and decreases to
V1 ZL + Z2 zero for extremely large loads. In the case that the load
Γ1 is a pure resistor, the load impedance ZL is real and the
I2 = V 1, (34)
ZL + Z2 system efficiency ν has the following simple expression:
I1 1 Γ1 Γ2
− = − . ξ1 ZL
V1 Z1 ZL + Z2 ν= , (37)
1 + ξ2 ZL + ξ3 ZL2
The dependence of the output voltage and that of the
input admittance on the load circuit impedance ZL , shown where ξn , n = 1, 2, 3 are real and are dependent of the
explicitly in (34), are of practical interest. It can be seen coefficients Γ1 , Γ2 , Z1 , and Z2 . In this case, the system ef-
that, for small loads, the output voltage increases linearly ficiency is characterized by these three real parameters ξn .
778 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 7, july 2003

Γ1 = −e33 ε−1
33 {χ31 [sin η (h2 + h0 ) − sin ηh0 ] + χ41 [cos ηh0 − cos η (h2 + h0 )]} ,
h2 
Z2 = 1 + e33 h−1
2 χ32 [sin η (h2 + h0 ) − sin ηh0 ] (31)
iωε33 S

+ e33 h−1
2 χ42 [cos ηh0 − cos η (h2 + h0 )] .

iωε33 S 
1/Z1 = 1 − e33 ε−1
33 χ11 [sin η (h1 + h0 ) − sin ηh0 ]
h1

− e33 ε−1 χ
33 21 [cos η (h1 + h 0 ) − cos ηh0 ] , (33)
e33
Γ2 = {χ12 [sin η (h1 + h0 ) − sin ηh0 ] + χ22 [cos η (h1 + h0 ) − cos ηh0 ]} .
h1

Z1 Z1∗ Γ1 Γ∗1 (ZL + ZL∗ )

ν= . (36)
Z1 Z1∗ [Γ∗1 Γ∗2 (ZL + Z2 ) + Γ1 Γ2 (ZL∗ + Z2∗ )] − (ZL + Z2 ) (ZL∗ + Z2∗ ) (Z1 + Z1∗ )

IV. Numerical Results and Discussions

For the purpose of illustration, we include a numerical

example by considering piezoelectric transducers of polar-
ized ceramics PZT-5H [3] for which:

ρ = 7500kg/m3 , c11 = 12.6,

c33 = 11.7, c44 = 2.30,
c12 = 7.95, c13 = 8.41 × 1010 N/m2 ,
c66 = (c11 − c12 ) /2, e15 = 17.0, (38)
e31 = −6.5, e33 = 23.3 C/m , 2

ε11 = 1700ε0 , ε33 = 1470ε0 ,

ε0 = 8.854 × 10−12 farads/m.

For the elastic plate, we choose steel with ρ0 =

7850kg/m3 and c033 = 2.69 × 1011 N/m2 . We include the
damping effect in the structure by allowing cpq to be
0
complex
 [7]. In  c33 and
 our case,  c33 will be0replaced by Fig. 2. The output voltage magnification versus the driving fre-
−1 0 −1
c33 1 + iQ and c33 1 + iQ , here c33 , c33 and Q are quency.
real numbers. We recognize that the metal wall and the
piezoelectric ceramic transducers generally have different
h2 = 2 mm, and S = 0.01 m2 . For clarity, we have normal-
material damping effects. But for the simplicity of this il-
ized, in the numerical results, the impedance of the load
lustrative example, we use a single damping parameter Q,
circuit using the normalizing impedance defined below:
instead of two, which may be considered as a damping pa-
rameter that represents the overall damping effect of the h2
system. Considering that the damping parameter Q for ce- Z0 = . (39)
iωε33 S
ramics is usually in the order of 102 to 103 [7], we take Q
to be 102 in the following illustrative calculations. We ex- In Fig.
2, we plot the normalized output voltage
pect that a slight variation of the Q value would affect the V 2 V 1 versus the driving frequency ω. The output volt-
sharpness of the resonant peaks in our numerical results, age peaks sharply at the thickness-stretch resonant fre-
but not the characteristic behavior of the system, which quencies, and this indicates that the system may effectively
is our primary interest in this illustrative example. We charge the battery inside the armor at frequencies near the
fix the geometrical parameters: h0 = 3 mm, h1 = 1 mm, resonant frequencies of the system. We note particularly
hu et al.: acoustic waves, metal walls, and transmitting energy 779

Fig. 3. The output voltage magnification versus the normalized load

circuit impedance. Fig. 4. The input admittance versus the driving frequency.

that the highest peak in this illustrative example occurs

at the fourth resonant frequency, instead of the first, i.e.,
the fundamental resonant frequency, which corresponds to
the lowest mode of thickness-stretch vibration. In fact, the
peak value increases monotonically with the first several
modes, then decreases. Our detailed numerical analysis re-
veals the following phenomenon. In the lowest mode, the
output transducer is not severely stressed because of the
moderate strain and the traction-free boundary condition
at the outer surface. Correspondingly, the electric field
within the output transducer generated through the piezo-
electric coupling is not particularly strong, and this leads
to a moderate output voltage, noting that the middle layer
is grounded. As the mode number increases from the lowest
to the next several ones, the stress and the corresponding
Fig. 5. The input admittance versus the normalized load circuit
electric field become increasingly stronger, leading to an
impedance.
increasingly larger output voltage. This trend, however,
reverses when the mode becomes so high that there are
nodal points present within the output transducer layer, terest, we also have plotted in Fig. 5 the input admittance
across which the stress reverses its sign and so does the versus the normalized impedance of the load circuit for two
electric field. Correspondingly, the voltage within the out- driving frequencies near the second resonant frequency.
put transducer no longer increases monotonically from the For a prescribed driving voltage V 1 , the input power
interface, in which the voltage is zero, to the outer surface. given in (35)1 depends upon the driving frequency through
This leads to a reduced output voltage. We note that the the input current I1 , which peaks at the resonant frequen-
highest peak does not always occur at the fourth resonant cies, as shown in Fig. 4. The output power given in (35)2
frequency; instead, it varies with the system parameters.
 depends upon the driving frequency through the output
The dependence of the output voltage V 2 V 1 on the current I2 = V 2 ZL and the output voltage V 2 , which

normalized load ZL Z0 is shown in Fig. 3, in which the also peaks at the resonant frequencies as shown in Fig. 3.
two curves correspond to two driving frequencies near the We have plotted in Fig. 6 the system efficiency ν versus
second resonant frequency. As indicated by (34)1 , these the driving frequency ω; and it is evident from compari-
curves are essentially linear for small loads, and the out- son with Fig. 2 that the system efficiency, defined as the
put voltages approach different (saturation) constants for quotient of the output power to the input power, does
extremely large loads. not peak at the resonant frequencies. This indicates that

We have plotted in Fig. 4 the input admittance I1 V 1 detailed analysis is necessary for the choice of driving fre-
versus the driving frequency ω, and the input admittance quency for maximizing the system efficiency.
peaks at the resonant frequencies, typical for resonant We have plotted in Fig. 7 the system
 efficiency ν ver-
piezoelectric components [8]. Considering the practical in- sus the normalized impedance ZL Z0 of the load circuit.
780 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 50, no. 7, july 2003

occur at the fundamental resonant frequency. The system

efficiency in transmitting the energy peaks at selected driv-
ing frequencies, which do not necessarily coincide with the
system resonant frequencies.

Acknowledgment

We would like to thank R. Main at the Lawrence Liv-

ermore National Laboratory for stimulating discussions.

References

[1] J. T. Stewart and Y.-K. Yong, “Exact analysis of the propaga-

tion of acoustic waves in multilayered anisotropic piezoelectric
Fig. 6. The system efficiency versus the driving frequency. plates,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol.
40, pp. 375–390, 1994.
[2] H. F. Tiersten, Linear Piezoelectric Plate Vibrations. New York:
Plenum, 1969.
[3] B. A. Auld, Acoustic Fields and Waves in Solids. vol. 1, New
York: Wiley, 1973, pp. 357–382.
[4] R. D. Mindlin, “Frequencies of piezoelectrically forced vibra-
tions of electroded, doubly rotated, quartz plates,” Int. J. Solids
Struct., vol. 20, pp. 141–157, 1984.
[5] Y. Huo and Q. Jiang, “Effect of polarization switch upon stress
in thickness vibration of a ferroelectric plate,” J. Amer. Ceram.
Soc., vol. 79, pp. 651–654, 1996.
[6] J. S. Yang, H. Fang, and Q. Jiang, “Thickness vibrations of
rotating piezoelectric plates,” J. Acoust. Soc. Amer., vol. 104,
pp. 1427–1435, 1988.
[7] R. Holland and E. P. EerNisse, Design of Resonant Piezoelectric
Devices. Cambridge, Massachusetts: M.I.T. Press, 1968.
[8] O. B. Wilson, An Introduction to the Theory and Design of
Sonar Transducers. Monterey, CA: Naval Postgraduate School,
1985.

Yuantai Hu received his B.S. degree in 1986 from Wuhan University

of Science and Technology, Wuhan, China, and his M.S. and Ph.D.
Fig. 7. The system efficiency versus the normalized load circuit degrees, respectively, from Harbin University of Technology, Harbin,
impedance. China in 1988 and from Shanghai Institute of Applied Mathematics
and Mechanics, Shanghai University, Shanghai, China, in 1992.
From 1993 to 1995 he worked at Shanghai Institute of Applied
As seen in (36), the system efficiency ν increases with the Mathematics and Mechanics as a postdoctoral researcher. Since then
he has been an associate professor of the Department of Mechanics
load impedance for small loads, with a constant initial slop; of Huazhong University of Science and Technology, Wuhan, China.
and after reaching a peak, it decreases monotonically with From April 2000 to May 2001, he was a visiting scholar at the De-
the load impedance and approaches zero. Although the partment of Engineering Mechanics of the University of Nebraska-
Lincoln. His research area is piezoelectric structures and devices.
impedance of the load circuit often is not at the choice of
designers for the transducers, this suggests that one may
enhance the system efficiency by varying the normalizing
impedance, which is a simple function of the dielectric con- Xuesong Zhang received his B.S. degree in
stant and the dimensions of the output transducer. engineering mechanics from Tsinghua Univer-
sity, Beijing, China, in the People’s Republic
of China in 2000. Currently he is a graduate
student in the Department of Engineering Me-
V. Conclusions chanics of the University of Nebraska-Lincoln.
His main research area is numerical and the-
Transmission of electric energy into a sealed armor can oretical studies of piezoelectric devices.
be achieved through thickness-stretch vibrations of the
metal wall together with two piezoelectric transducers, one
as the driving transducer and the other as the receiving
transducer. The output voltage of this system is magni-
fied at the first several resonant frequencies of the system
while, interestingly, the greatest magnification does not
hu et al.: acoustic waves, metal walls, and transmitting energy 781

Jiashi Yang was born in Beijing, China, Qing Jiang received a Ph.D. degree in en-
on June 10, 1956. He received his B.E. and gineering and applied sciences at the Califor-
M.E. degrees in engineering mechanics in nia Institute of Technology, Pasadena, CA, in
1982 and 1985 from Tsinghua University, Bei- 1990. He served at the University of Nebraska-
jing, China. In 1986–1988 he studied at Syra- Lincoln, as assistant professor, associate pro-
cuse University, Syracuse, NY and obtained fessor, and professor of Engineering Mechan-
his M.S. degree in mechanical engineering. ics during 1991–1997. Since 1998, he is a pro-
From 1988–1993 he was a graduate student fessor of mechanical engineering at the Uni-
at Princeton University, Princeton, where he versity of California, Riverside. His research
received his M.A. and Ph.D. degrees in civil has been focused on the performance and re-
engineering. liability of electromechanical devices made of
Dr. Yang was a Postdoctoral Fellow in piezoelectric or ferroelectric materials.
1993–1994 in the Department of Mechanical and Aerospace Engi-
neering of the University of Missouri-Rolla. From 1994–1995 he was
a Postdoctoral Research Associate at the Department of Mechanical
Engineering, Aeronautical Engineering and Mechanics of Rensselaer
Polytechnic Institute, Troy, NY. He was employed by Motorola, Inc.,
Schaumburg, IL, during 1995–1997. Since 1997 he has been an as-
sistant professor of the Department of Engineering Mechanics of the
University of Nebraska-Lincoln.
He is a member of IEEE and its Society of Ultrasonics, Ferro-
electrics, and Frequency Control.