64
64
Advanced structures with integrated self-monitoring and control capabilities are becom-
ing very important due to the rapid development of “intelligent” mechanical systems and
space structures. Since the structures are distributed and flexible in nature, distributed
dynamic measurement and active vibration suppression are of importance to their perform-
ance. In this paper, a new structure (shell or plate) containing an integrated distributed
piezoelectric sensor and actuator is proposed. The distributed piezoelectric sensing layer
monitors the structural oscillation due to the direct piezoelectric effect and the distributed
actuator layer suppresses the oscillation via the converse piezoelectric effect. For modeling
flexibility and versatility, a new piezoelectric finite element with internal degrees of freedom
is derived. The performance of a plate model with distributed piezoelectric sensor/actuator
is evaluated. Applications to distributed dynamic measurement and control of the advanced
structures are also demonstrated.
1. INTRODUCTION
Advanced structures with integrated self-monitoring and control capabilities are increas-
ingly becoming important due to the rapid development of “intelligent” space structures
and mechanical systems [l-4]. Since these structures are, in general, distributed and
tlexible in nature, distributed dynamic measurement and active vibration suppression are
essential to their performance. Vibration suppression and control of distributed parameter
systems (e.g., plates and shells in this study) always represents a challenge, both in theory
and practice. Theoretical development has been constantly advanced in the past 20 years
[2,3,5-g]. However, due to the limitation of materials and actuator design, practical
application of the theory to general distributed parameter systems still needs to be further
explored. Besides, in order to control or suppress the undesirable structural oscillation
of a distributed parameter system, an accurate measurement of the structural vibration
is required. Conventional transducers (such as accelerometers, strain gages, and pressure
transducers, etc.) are “discrete” in nature: i.e., measuring the responses at spatially
“discrete” locations. Some natural frequencies and mode shapes could be missed if the
transducers are placed at nodal modes or lines. Thus, the development of a “distributed”
sensor can be essential for new-generation lightweight, high-performance structures. This
paper is concerned with thin piezoelectric layers which are coupled with conventional
materials and used as distributed sensors and distributed actuators in an intelligent
advanced structure design.
The direct piezoelectric efict, a charge/voltage generated by an imposed force/pressure
to a piezoelectric, has been widely applied to variety of transducer designs: e.g.,
17
0022460X/90/070017+ 18 $03.00/O @ 1990 Academic Press Limited
18 H. S. TZOU AND C. 1. TSENG
accelerometers, pressure transducers, etc. [9]. However, the comer-se piezoelectric efect,
and induced stress/strain due to an externally applied votage/charge, is not that common
as compared with the direct effect. Some applications of the converse effect have been
made in flexible mirrors [lo] and linear translators [ll]. Also, Tzou and Gadre have
designed a piezoelectric exciter and a vibration isolator [ 12,131. In this paper, the advanced
structure (distributed parameter system) is a shell or plate configuration with one
piezoelectric layer serving as a distributed sensor and the other layer serving as a distributed
actuator. The direct effect is used in distributed sensing and the converse effect in
distributed active vibration suppression and control of the advanced structure. Thus, the
sensing layer detects the oscillation of the distributed systems and the actuator controls
the vibration of the system. The peizoelectric material used in the finite element analysis
of the advanced structures is a P-phase polymeric piezoelectric polyvinylidene fluoride
(PVDF). However, the developed piezoelectric finite element can also be used for other
types of piezoelectric materials, e.g., piezoceramics.
Polyvinylidene fluoride (PVDF) was initially discovered by Kawai in 1969 [14]. Raw
polymeric PVDF (a-phase) is an electrical insulator and it does not have any intrinsic
piezoelectric properties. If the raw material is polarized during the manufacturing process,
PVDF transforms to a P-phase-a tough and flexible semi-crystalline material and it can
be made to strain either in one or two directions in the film plane. Since P-phase PVDF
possesses a strong direct piezoelectric effect, it has been used in many transducer applica-
tions: e.g., sonar, medical ultrasonic equipment, robot tactile sensors, acoustic pick-ups,
force and strains gages, etc. [9]. Due to its distinct characteristics, such as flexibility,
durability, manufacturability, etc., PVDF is an ideal material for the distributed sensing
and vibration suppression/control of distributed parameter systems (e.g., beams, plates,
shells, etc.).
Application of the flexible piezoelectric PVDF to vibration control and active damping
of beam structures have been studied in recent years [15,16]. Crystalline piezoceramic
materials also have been investigated in a longitudinal actuation [17] and a distributed
sensor/actuator for beam structures [4,18]. Tzou applied a PVDF film as an active damper
in a flexible structure [19] and as an active vibration isolator and exciter (with Gadre)
[12, 131. A theory of multi-layered shells coupled with the piezoelectric shell actuators
has been derived and evaluated by Tzou and Gadre [l]. Distributed active vibration
control of a shell coupled with PVDF has been investigated [3, 201, and distributed
sensing theory for a shell with PVDF also has been proposed and evaluated [21, 221. Up
to now, research in this area has been primarily focused on experimental and theoretical
studies. General piezoelectric finite element development is relatively limited [20,22]. In
general, experimental models are limited by size, cost, noise and many other laboratory
unknowns. Theoretical models can be more general, but analytical solutions are restricted
to relatively simple geometries and boundary conditions. When the geometry and/or
boundary conditions become relatively complicated, difficulties occur with both theoretical
and experimental models. Thus, the finite element development becomes very important
in modeling and analysis of advanced flexible structures with integrated distributed
piezoelectric sensors and/or actuators.
Finite element techniques are very popular and important in many modern engineering
designs and analyses. A piezoelectric finite element was developed and applied in
piezoceramic transducer designs [23,24]. However, the derived isoparameteric hexa-
hedron and tetrahedral elements are too thick for thin shell/plate applications. In general,
the proposed advanced (intelligent) structure is composed of a master shell/plate with a
coupled or embedded piezoelectric sensor/actuator. The thickness of the master structure
is about two to three orders thicker than that of the piezoelectric layer. It would be very
DISTRIBUTED SENSOR/ACTUATOR DESIGN 19
inefficient and time consuming if the entire structure were to be modeled by hexahedron
or tetrahedral solid elements. Thus, the development of a new thin piezoelectric solid
finite element would be very important to the modeling and simulation of large flexible
distributed systems-shells and plates with distributed piezoelectric sensor/actuator.
In this paper, the development of a “thin” piezoelectric solid element with internal
degrees of freedom (DOFs) is presented, and its application to distributed dynamic
measurement and active vibration suppression and control of an advanc;-d “intelligent”
plate is studied. The distributed sensing phenomena and effectiveness of the piezoelectric
actuator are also evaluated. The dynamic equations of a piezoelectric element are first
derived by using the piezoelectric constitutive equations [25]. This leads to a formulation
of the entire system matrix equation of motion. In order to improve computation efficiency,
a Guyan reduction scheme [26] is used to condense the DOFs associated with the electrical
potential, which can be recovered if the distributed measurement is desired. The dynamic
response of the distributed system is calculated by a direct integration algorithm-the
Wilson-0 method and the pseudo-force method [27].
In this section, a thin piezoelectric finite solid element with internal DOFs is derived
by using a variational method and Hamilton’s principle. The system matrix equation is
also formulated by assembling all of the element matrices.
It is assumed that the mechanical and the electrical forces in an oscillating piezoelectric
are balanced at any given time instant. Thus, the piezoelectric equations can be decoupled:
i.e., a quasi-static approximation is used in the analysis. It is also assumed that the
temperature variation is negligible in a fast oscillating piezoelectric: i.e., the pyroelectric
effect is not considered in the analysis. The linear piezoelectric constitutive equations
coupling the elastic field and the electric field can be respectively expressed as the direct
and the converse piezoelectric equations [25] (a list of nomenclature is given in the
Appendix) :
‘Sll
Dl -e,, e12 S 22
I[:
e13 e14 e15 e16 El1 El2 El3 El
D2 = e21 e22 e23 e24 e25 e26 S 33 + &21 822 &23 E2 . (1)
LI
S 13
s12
T, 1 Cl1 Cl2 Cl3 Cl4 Cl5 cl6 ‘Sll ell e2, e31
T22 Cl2 c22 c23 c24 c25 c26 S 22 e12 e22 e32 El
T 33 Cl3 c23 c33 c34 c35 C36 S 33 et3 e23 e33 E2
=
T 23 Cl4 c24 c34 c44 c45 c46 S 23 cl4 e24 e34 E3
TI3 Cl5 c25 c35 c4s C55 C56 S 13 cl5 e25 e35
T** cl6 c26 c36 c46 c56 C66 S 12 e16 e26 e36
Equation (1) describes the direct effect and equation (2) the converse effect. A piezoeleccric
element defining all mechanical and electrical variables is illustrated in Figure 1.
Equations (1) and (2) can be written simply as
ID] = [el’{S~+[4{W, ~Tl=[cEI~~~-_[~I~~l, (3-4)
20 H. S. TZOU AND C. 1. TSENG
where {D} is the electric displacement vector, [e] is the dielectric permittivity matrix,
[e]’ is the transpose of [e], {S} is the strain vector, [es] is the dielectric matrix at constant
mechanical strain, {E} is the electric field vector, {T} is the stress vector, and [c”] is the
elasticity matrix for a constant electric field.
(8)
DISTRIBUTED SENSOR/ACTUATOR DESIGN 21
To derive the electroelastic matrix relationship for a piezoelectric finite element, the
displacement {q} and electric potential 4 are defined in terms of i nodal variables via
the shape function matrices [IV,] and [IV,]:
{91= LNql{qiI
+ [xl{aj), (16)
where [X] is the extra mode shape function matrix for the added internal DOFs {a,}.
[X] and {ei} are defined as
1
N, 0 0
[Xl=: 0 N, 0 , lo,I=,i9 iT”, 21j Wilt. (17,18)
j=9
[ 0 0 N,
Note that there are eight nodes for a hexahedron element. Thus, the internal DOFs are
defined as those from nine to eleven. The added DOFs {a,} are not actual physical
displacements and they should be regarded as the generalized co-ordinates or as the
displacements relative to the displacements in the local co-ordinate system. The displace-
ments associated with {a,} vanish at all element boundaries, so that these DOFs are
internal and have no effect on interelement compatibility. The strain-displacement
equation (10) now needs to include the internal DOFs:
tsI = iIBql{qiI+
1Yl{“jl* (19)
22 H. S. TZOU AND C. I. TSENG
[Y]==i : N”, 22 .
j=9 (20)
Ni,, d” N;:
N.y Ni,x 0
Note that Ni,x = aN,/ax and [Z3,] is defined as before.
The potential energy % in equation (5) now must be modified to include the internal
DOFs. Applying the principle of minimum potential energy again and substituting
equation (19) into equation (8) yield
(224
where
There are two reasons for considering the condensation of element and system matrices:
(1) the internal DOFs of an element do not have physical significance and (2) some
global DOFs (slave DOFs) are not as significant as the other (master DOFs). Thus, these
DOFs can be reduced from the elemental or system matrices without losing much accuracy
in the analysis. A Guyan reduction algorithm is used in this analysis [26]. The concept
of the scheme is based on a Gaussian elimination technique which eliminates some minor
(slave) DOFs from the system matrices to improve the computation efficiency. In a static
analysis, this reduction is “exact” because the reduced dependent DOFs are exactly
recovered in the back substitution. In a dynamic analysis, however, the mass and damping
matrices are “approximated” by using a congruent transformation technique. In this
analysis, a two-stage matrix condensation is applied: (1) elemental internal DOFs and
(2) global electrical potential DOFs. These are separately discussed in what follows.
DISTRIBUTED SENSOR/ACTUATOR DESIGN 23
3.1. CONDENSATION OF ELEMENTAL INTERNAL DOFs
The enlarged elemental matrix needs to be condensed before assembled into the global
system matrix. Employing the Guyan reduction method to eliminate the internal DOFs
{uj} yields the modified element static matrix equations
(27b)
{gi}=-
JY*
[N&]‘adY>. (27~)
Additional system matrices are defined in equations (23) and (25). In general, all structures
are lightly damped. Thus, adding an artificial linear viscous damping to equation (26)
gives
[m,,l{~}+[c,,l{4}+[k,*,l{s}+[k~~l{~}={f}, (22a)
[k~,l{q}+[k,,l{~}={g}, (28b)
where the damping matrix is defined as a proportional damping: i.e.,
[c,,l= dm,,l+P[k&l, (29)
where cy and /3 are Rayleigh’s coefficients. Combining the displacement {q} and the
electrical potential (4) yields the matrix expression
Assembling all elemental equations gives the global dynamic system equation
Note that the mechanical equation is coupled with the electrical equation. Here {F} is
the external mechanical excitation and {G} the electrical excitation. In the active vibration
control application, {G} is the feedback voltage determined by the control algorithm.
In the distributed sensing and vibration suppression analysis, the displacement time
history is much more important than the electrical potential vector in most applications.
It is not absolutely necessary to save the electrical potential vector at each time step in
the time domain integration unless the dynamic sensing time history is required. Thus,
in order to save computer memory and to improve computation efficiency, the electrical
24 H. S. TZOU AND C. I. TSENG
potential vector is usually condensed in the time domain integration. However, a recovery
scheme can be set up if the sensing information is required.
Consider the static case in equation (31): i.e.,
~~,,l~ci’~+~~,,l~4~+~~*1~~~=~~~-~~,,I~~,,I-’~~~, (34)
where
[K*l= [KJ - v&m&frK$,l. (35)
In equation (34), there are two excitation forces associated with the piezoelectric struc-
tures: i.e., the mechanical forces and the electrical forces. The electrical potential vector
can be recovered by
(4) = VLJ’W~ - [GJ{q~). (364
The system dynamics is governed by equation (34) and the distributed dynamic measure-
ment (voltage distribution) can be calculated by using equation (36). In the free vibration
analysis, {G} is set to zero so that the voltage distribution associated with each mode
can be estimated. Note that {G} is usually zero in the distributed sensor layer. Thus, the
distributed sensor output is estimated by
(4) = Kd’HbJ~d). Wb)
zoelectrlc
Pie uotor loyer
sen
Figure 2. An integrated distributed piezoelectric sensor/actuator design. Plate dimensions for calculations:
10cmx10cmx0~31 cm.
DISTRIBUTED SENSOR/ACTUATOR DESIGN 25
The distributed sensor generates a voltage output when the structure is oscillating; and
this signal is amplified and fed back into the distributed actuator. The micro-structural
action of the piezoelectric actuator, in which a negative voltage introduces positive strains
in the actuator resulting in a moment counteracting the upward motion of the structure,
is shown in Figure 3. Thus, the control law is established such that the moment generated
will oppose the motion in the transverse direction. Hence, a negative voltage should be
applied when the transverse motion is upward and vice uersa.
(Transverse /----?A
Piezoelectrlc
Plate
As discussed earlier, there are two force terms in the system equations of motion: i.e.,
the mechanical force component and the electrical force component as indicated in
equation (34). In the feedback control, the electrical force component, the second term
in equation (34), can be regarded as a feedback force {F,},
~~,,l~~~+~~,,l~4~+~~*l~s~=~~~+~~~,,I~~,,I-’~@I~~,,I-‘~~,,lI~~~. (38)
Since only velocity feedback is considered in this study, an equivalent damping force can
be defined as
[G1{4) = ~~l~~~~~,,I~~,,l-‘~@I~~,,l~‘~-~~,,I~~~~
= ~-~~,~1~~~~1-‘~@1~~~~1~‘~~~,1~~4~~ (39)
Thus, the system equation of motion becomes [ Mg4]{6) + ([C,,] + [C,]){ 4) + [K*](q) =
{F}, which implies that
~~,,1~4~+~~~,,1-~~,,1~~~~1-‘~@1~~~~1-’~~,,1~0+~~*1~~~=~~~. (40)
26 H. S. TZOU AND C. I. TSENG
The control force induced by the feedback surface voltage/charge can effectively
enhance the system damping and therefore suppress the vibration of a distributed system.
As discussed in the finite element formulation, the feedback control forces are assumed
to be proportional to velocity (see equation (39)). Thus, two different feedback control
algorithms are evaluated; (1) constant-gain feedback control, and (2) constant-amplitude
feedback control. The feedback voltages for these two feedback control algorithms are
shown in Figure 4.
r--
In the first case, the feedback gain is constant while the feedback amplitude varies with
respect to the negative oscillating velocity (negative velocity, consfunt-gain feedback con-
trol): i.e.,
In the finite element analysis, the time history responses of the piezoelectric system are
calculated by using a time-domain direct integration algorithm, the modified Wilson-8
method, and a pseudo-force method [27] to accommodate the control force derived from
the applied surface changes.
DISTRIBUTED SENSOR/ACTUATOR DESIGN 27
In this section, the dynamic characteristics of a plate with surface coupled distributed
piezoelectric sensor and actuator are evaluated. The distributed dynamic measurement
of the first three modes is demonstrated by an eigenvalue analysis, in which the mode
shapes and associated voltage distributions are obtained. Evidence of the effectiveness
of the distributed active vibration suppression and control is also presented.
(0.0) mm
TABLE 1
Dielectric permittivity
e31 0.0460 C/m*
e3* 0.0460 C/m*
e33 0~0000 C/m’
Dielectricity
El1 0.1062 x lo-’ F/m
-522 0.1062 x 1O-9 F/m
&33 0.1062 x 1O-9 F/m
Poisson ratio 0.2900
Mass density 0.1800x lo4 kg/m3
Modulus 0.2000 x 1o’O N/m2
28 H. S. TZOU AND C. 1. TSENG
strains settle very fast without time delay; (3) the velocity information can be obtained
and instantaneously used as a feedback signal.
As discussed earlier, the distributed piezoelectric sensing layer should respond to the
plate oscillation and generate an electric voltage representing the distributed dynamic
response of the plate. This distributed sensing phenomena are demonstrated in an
eigenvalue analysis. The active distributed vibration suppression and control of the plate
is studied and evaluated in a snap-back analysis in which an initial displacement (first
mode) was imposed.
In equation (36), the output signals of each node on the distributed piezoelectric sensor
layer can be calculated as a function of the displacements. (Note that the {G} vector is
zero in an eigenvalue analysis.) After the nodal voltage is calculated, the overall voltage
distribution of the plate can be plotted by connecting all nodal voltage amplitudes. Thus,
for a given mode, the modal voltage distribution (distributed sensing phenomena) can
be observed. The first three plate mode shapes and modal voltage distributions are
illustrated in Figures 6-8.
It is observed that the first mode is a bending mode, the second mode a torsion mode,
and the third mode a warping mode. According to equation (36), the output amplitude
is inversely proportional to the displacement: i.e., the maximum positive voltage occurs
at the maximum negative displacement. Voltage drops at two corners, V(O,O)and
V(0,loo), are observed in all three figures. This is because that the strains are (numerically)
smaller at these two boundary nodes in the finite element calculation. The sensitivity of
each mode could also change because the tension and compression vary in different
modes. The modal voltage distribution could also be refined if more elements and a finer
mesh were used in the finite element modeling.
The distributed piezoelectric actuator on the top surface of the plate contracts or
expands depending on negative or positive feedback voltages (the converse piezoelectric
DISTRIBUTED SENSOR/ACTUATOR DESIGN 29
T
o*23
*
P
” 0.18 -
0
‘0 L
P
._
o.oe ll- I I I I I 4
0 250 500 750 1000 1250 1500
vmo, (volts)
0.05 b I I I I 1
0 250 500 750 1000 I250 1500
Feedback gain, C
Figure 10. Damping calculation at constant-amplitude feedback control.
DISTRIBUTED SENSOR/ACTUATOR DESIGN 31
damped out much faster at higher feedback gain, as shown in Figures 9 and 10. Note
that a single node velocity was used in the feedback controls, even though the distributed
voltage was discussed in the previous section. This single signal was amplified and fed
back to all nodes of the actuator.
in this analysis. The performance of other piezoelectric materials, such as PZTs, and the
effect of piezoelectric orientation need to be explored further.
ACKNOWLEDGMENT
This research was supported, in part, by a grant from the National Science Foundation
(No. RII-8610671) and the Kentucky EPSCoR program, and a seed grant (No. 6-AGl)
from the Center for Robotics and Manufacturing Systems at the University of Kentucky.
The open access on IBM 3084 and 3090-3008 computers at the Computer Center is also
gratefully acknowledged.
REFERENCES
1. H. S. Tzou and M. GADRE 1989 Journal of Sound and Vibration 132, 433-450. Theoretical
analysis of a multi-layered thin shell coupled with piezoelectric shell actuators for distributed
vibration controls.
2. H. S. TZOU 1988 Recent Developments in Control of Nonlinear and Distribred Parameter Systems,
ASME-DSC-Vol. (lo), December 1988, 51-58. Integrated sensing and adaptive vibration sup-
pression of distributed systems.
3. H. S. TZOU and C. I. TSENG 1988 Machine Dynamics and Engineering Applications, Xian
Jiaotong University Press, China 1, Gl-G6, August 1988. Sensing and adaptive vibration control
of flexible distributed mechanical systems.
4. E. F. CRAWLEY and J. LUIS 1987 American Institute of Aeronautics and Astronautics Journal
25( lo), 1373-1385. Use of piezoelectric actuators as elements of intelligent structures.
5. J. L. LIONS 1968 Optimal Control of Systems Governed by Partial Dtyerential Equations. Paris:
Dunod and Gauthier-Villars.
6. S. G. TZAFESTAS 1970 International Journal of Control 12, 593-608. Optimal distributed-
parameter control using classical variational theory.
7. A. G. BUTKOVSKII 1962 Automation and Remote Control 22,1429-1438. The maximum principle
for optimum systems with distributed parameters.
8. M. J. BALAS 1988 Recent Developments in Control of Nonlinear and Distributed Parameter
Systems. ASME-DSC- Vol. (lo), 19-22, December 1988. Nonlinear finite-dimensional control
of a class of nonlinear distributed parameter systems using residual mode filters.
9. H. S. Tzou and S. PANDITA 1987 Journal of Robotic Systems (4-6), 719-741. A multi-purpose
dynamic and tactile sensor for robot manipulators.
10. P. HALEVI 1983 Journal of the Optical Society of America 73, 110-l 13. Biomorph piezoelectric
flexible mirror: graphical solution and comparison with experiments.
11. BURLEIGH INSTRUMENTS 1986 Linear Translators, New York, USA.
12. H. S. TZOU and M. GADRE 1988 American Institute of Aeronautics and Astronautics Journal
26, (8), 1014-1017. Active vibration isolation by piezoelectric polymer with variable feedback
gain.
13. H. S. Tzou and M. GADRE 1988 Journal of Sound and Vibrufion 136(3), 477-490. Active
vibration isolation and excitation by piezoelectric slab with constant feedback gains.
14. H. khWA1 1969 Japan Journal of Applied Physics 8,975. The piezoelectricity of poly(vinylidene
tluoride).
15. S. W. SIRLIN 1987 Proceedings of the 114th Meeting of the Acoustic Society of America, November
1987. Vibration isolation for spacecraft using the piezoelectric polymer PVF,.
16. J. M. PLUMP, J. E. HUBBARD and T. BAILY 1987 Journal of Dynamic Systems, Measurement,
and Control 109, 133-139. Nonlinear control of a distributed system: simulation and experi-
mental results.
17. J. L. FANSON and J. A. GARBA 1988 AIAA Paper 88-2207, Proceedings of the
AIAAIASMEIAHS 29th Strucfures, Structural Dynamics, and Materials Conference, 9-17.
Experimental studies of active members in control of large space structures.
18. S. HANAGUD and M. W. OBAL 1988 AZAA Paper 88-2418. Proceedings of the
AIAAIASMEIAHS 29th Structures, Srructual Dynamics, and Materials Conference, Parr 3,
1611-1620. Identification of dynamic coupling coefficients in a structure with piezoelectric
sensors and actuators.
DISTRIBUTED SENSOR/ACTUATOR DESIGN 33
19. H. S. Tzou 1987 Developments in Mechanics (14-C), 1201-1206. Active vibration control of
flexible structures via converse piezoelectricity.
20. H. S. Tzou and C. I. TSENG 1988 Computers in Engineering 1988,3, 599-604. Active vibration
control of distributed parameter systems by finite element method.
21. H. S. Tzou and J. P. ZHONG 1988 Proceedings of 1988 Symposium on Advanced Manufacturing
(SAM88), Lexington, Kentucky September 1988, 91-94. Advanced dynamic measurements for
distributed mechanical systems, part 1: theoretical development.
22. H. S. Tzou and C. I. TSENG 1988 Proceedings ofthe 1988 Symposium on Advanced Manufacturing
(SAM88), Lexington, Kentucky, September 1988, 95-98. Advanced dynamic measurement for
distributed mechanical systems, part 2: finite element analysis.
23. H. ALLIK and T. J. HUGHES 1979 International Journal of Numerical Methods in Engineering
2, 15 1- 168. Finite element method for piezoelectric vibration.
24. M. NAILON, R. H. COURSANT and F. BESNIER 1983 ACTA Electronica 25 (4), 341-362.
Analysis of piezoelectric structures by a finite element method.
25. H. F. TIERSTEN 1969 Linear Piezoelectric Plate Vibrations. New York: Plenum Press.
26. R. J. GUYAN 1965 American Institute of Aeronatics and Astronautics Journal 3(2), 380. Reduca-
tion of stiffness and mass matrices.
27. H. S. Tzou and A. J. SCHIFF 1987 International Journal of Computers and Structures 26(3),
481-493. Development and evaluation of a pseudo-force approximation applied to nonlinear
contacts and viscoelastic damping.
28. K. J. BATHE and E. L. WILSON 1976 Numerical Mathods in Finite Element Analysis. London:
Prentice-Hall.
29. R. D. COOK 1974 Concepts and Applications of Finite Element Analysis. New York: John Wiley.
30. E. L. WILSON, W. P. TAYLOR, A. DOHERTY and J. GHABOUSSI 1973 in Numericaland
Computer Methods in Structural Mechanics. Incompatible displacement models.
APPENDIX: NOMENCLATURE
matrix
vector
transpose of vector or matrix
the elasticity matrix evaluated at constant electric field
element damping matrix
feedback gain
feedback gain matrix
system damping matrix
equivalent non-linear damping matrix
the electric dispiacement vector
the electric field vector
dielectric permittivity matrix
equivalent non-linear force
element charge vector
external applied charge vector
electric enthalpy
element stiffness matrix
element piezoelectric stiffness matrix
element dielectric stiffness matrix
condensed system stiffness matrix
Lagrangian
elasticity operator
element consistent mass matrix
system mass matrix
shape function matrix related to nodal displacement
shape function matrix related to nodal potential energy
body force vector
surface force vector
concentrated load vector
nodal displacement vector
displacement vector
H. S. TZOU AND C. 1. TSENG
velocity vector
acceleration vector
the strain tensor
surface where surface forces applied
surface where surface charges applied
the stress tensor
transformation matrix
internal energy
piezoelectric volume
virtual work
Rayleigh’s coefficients
dielectric matrix evaluated at constant strain
nodal electrical potential vector
electrical potential vector
surface charge
gradient operator