Thesis PDF
Thesis PDF
Thesis PDF
Electrical Loadings
Waleed F. Faris
Doctor of Philosophy
in
Engineering Mechanics
December, 2003
Blacksburg, Virginia
Nonlinear Dynamics of Annular and Circular Plates under Thermal and Electrical Loadings
Waleed F. Faris
(ABSTRACT)
The nonlinear static and dynamic response of circular and annular plates under electrostatic, thermal, and combined loading is investigated. The main motivation for the study of these phenomena
is providing fundamental insights into the mechanics of micro-electro-mechanical-systems (MEMS).
MEMS devices are usually miniaturization of the corresponding macro-scale devices. The basic mechanics of the components of many MEMS devices can be modeled using conventional structural
theories. Some of the most used and actively researched MEMS devices- namely pressure sensors
and micropumps- use circular or annular diaphragms as principle components. The actuation and
sensing principles of these devices are usually electrostatic in nature. Most MEMS devices are
required to operate under wide environmental conditions, thus, a study of thermal effects on the
performance of these devices is a major design consideration.
There exists a wide arsenal of analytic, semi-analytic, and numerical tools for nonlinear analysis
of continuous systems. The present work uses different tools for the analysis of different types of
problems. The selection of the analysis tools is guided by two principles. The first consideration
is that the analysis should reveal the fundamental mechanics and dynamics of the problem rather
than simply generating numerical data. The second consideration is numerical efficiency. Guided
by the same principles, the basic structural model adopted in this work is the von-Karman plate
model. This model captures the basic nonlinear phenomena in the plate with minimal complexity
in the equations of motion, thus providing a balance between simplicity and accuracy.
We address a wide array of problems for a variety of loading and boundary conditions. We start
by analyzing annular plates under static electrostatic loading including the variation of the plate
natural frequencies with the applied voltage. We also analyze parametric resonances in plates
subjected to sinusoidally varying thermal loads. We investigate the prebuckling and postbuckling
static thermal response and the corresponding variation of the natural frequencies. Finally, we
close by investigating the problem of a circular plate under a combination of thermal and electrostatic loading. The results of this investigation demonstrate the importance of including nonlinear
phenomena in the modeling of MEMS devices both for correct quantitative predictions and for
qualitative description of operations.
Dedication
To my mother who could not wait to see this moment.
To my father who tries to hold.
To Heba, Mariem, and Yomna my only joy in this life.
iii
Acknowledgement
I would like to thank all who helped and supported me during the past four years in my research.
Special thanks goes to my academic advisor Dr. Ali Nayfeh for his help, guidance, and unfailing
support without which the completion of this work would have been impossible. I would also like to
express my thanks to my committee members for their support and valuable suggestions, especially
Professor Ragab.
iv
Contents
Motivation
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2
1.3
Literature Review
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3.1
1.3.2
1.3.3
1.3.4
1.1.1
1.4
Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2 General von K
arm
an Formulation of Circular and Annular Plates under Thermal
and Electrostatic Loadings
25
2.1
Basic Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2
Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
34
3.1
Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2
Clamped-Clamped Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3
Clamped-Free Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
47
4.1
Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2
4.3
4.2.1
Perturbation Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2.2
Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Perturbation Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.3.2
4.3.3
4.3.4
71
5.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2
Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
81
6.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.2
Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
vi
91
7.1
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2
Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
vii
List of Figures
1.1
2.1
3.1
3.2
3.3
3.4
The first normalized mode shape of the clamped-clamped plate as the electrostatic
force increases from zero and approaches pull-in level. . . . . . . . . . . . . . . . . . 41
3.5
The second normalized mode shape of the clamped-clamped plate as the electrostatic
force increases from zero and approaches pull-in level. . . . . . . . . . . . . . . . . . 42
3.6
3.7
3.8
3.9
The first mode shape of the clamped-free plate as the electrostatic force increases
45
viii
3.10 The second mode shape of the clamped-free plate as the electrostatic force increases
from zero and approaches the pull-in level. . . . . . . . . . . . . . . . . . . . . . . . . 46
4.1
4.2
4.3
Variation of the steady-state response amplitude with the forcing amplitude for =
0.8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4
Variation of the steady-state response amplitude with the forcing amplitude for = 0.5. 59
4.5
4.6
Variation of the second nondimensional frequency with the nondimensional parameter (temperature) p. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.7
4.8
Variation of the equilibrium solutions with the detuning parameter when p = 7.0
and q = 0.01. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.9
4.10 Variation of the equilibrium solutions with the forcing amplitude (force-response
curves) when = 0.85 and p = 7.0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.1
5.2
5.3
Comparison of the power-series solution with our solution for a simply supported
plate; solid curve: present solution, stars: power-series solution. . . . . . . . . . . . . 79
ix
5.4
5.5
Variation of the first natural frequency sqaured with the load level . . . . . . . . . . 80
5.6
Variation of the first four natural frequencies with the load level . . . . . . . . . . . 80
6.1
6.2
6.3
6.4
6.5
List of Tables
3.1
4.1
xi
Chapter 1
Motivation
Since the electronic revolution in the sixties, miniaturization has become an important goal of
technology. This is quite clear in the doubling of the number of transistors per chip every two years
since the sixties (Michalicek, 2000). Since the first bulk-etched silicon wafers were used as pressure
sensors in the 1970s, a new discipline arose known as Microelectromechanical Systems (MEMS) in
the USA and integrated systems in Europe (Michalicek, 2000). MEMS are usually defined as highly
miniaturized devices or array of devices combining electrical and mechanical components that are
fabricated using integrated circuit (IC) batch-processing techniques (Michalicek, 2000). MEMS
have received a great deal of attention in recent years. This is due not only to the excitement
naturally associated with a new emerging technology, but also to the great promise of increased
miniaturization and improved performance of these devices over conventional devices. Miniaturization means less energy input, increased reliability, reduced cost, and a room for more complicated
and advanced applications.
Applications of MEMS are very wide. In automotive engineering applications, MEMS-based sensors
are used as coolant pressure sensors, manifold gas pressure sensors, airbag accelerometers, and
speedometers. In medical applications, MEMS sensors are used for measuring blood pressure and
micropumps (MEMS pumps) are used for drug delivery (Sensors Magazine, 1997). In defense
Waleed F. Faris
Chapter 1.
applications, MEMS are used in inertial navigation, as distributed unattended sensors for security
surveillance and process control, as integrated fluidic systems for propellant and combustion control,
as mass data storage devices, and in active and conformal surfaces for distributed aerodynamic
control of aircraft (Michalicek, 2000).
The worldwide MEMS market is growing at a very fast rate. In 1993, it was less than 1 Billion US
Dollars and in 2000 it approached 14 Billion US Dollars. The lions share in 2000 was for pressure
sensors, which constituted about one quarter of the worldwide market, and for microfluidic devices,
which constituted slightly less than one quarter of the worldwide market (Michalicek, 2000).
The main theme of this dissertation is micromachined devices, especially sensors and micropumps.
Many micromachined sensors are miniaturized versions of their macroscopic counterparts. Thus, it
is quite instructive to take a look at the macroscopic versions of these devices.
For pressure sensors it is reported (Sensors Magazine, 1997) that many of these devices were
based on diaphragms. Other devices were also sought to improve the amount of deflection of a
simple diaphragm, such as capsules and bellows. Strain gauges are commonly used to measure the
deflections of diaphragm-based devices. Some diaphragm sensors have elaborate systems of levers,
which are linked to electric switches or potentiometer winding. Others, instead of having strain
gauges mounted directly on the diaphragm itself, have a piston, which is driven into a mounted
strain gauge by the motion of the diaphragm. Most sensors for greater than atmospheric pressure
share the common characteristic of deformable diaphragm. In diaphragm-based pressure sensors,
the applied pressure is determined by the deflection it produces in the diaphragm. The shape of
the diaphragm is arbitrary, but generally takes the form of a square or a circle.
Macroscale pumps are generally classified into dynamic-type pumps and positive-displacement
pumps. Dynamic-type pumps include centrifugal, axial, and turbine pumps. Positive-displacement
pumps include reciprocating and rotary pumps. Reciprocating pumps, which are most common,
include piston and diaphragm pumps (Tay and Phoon, 1997).
Micropump technology is currently one of the most advanced technologies in MEMS. They are
designed to handle small and precise volumes in various medical, biomedical, and chemical applications. The function of a micropump is to raise the pressure of a certain volume of gas or liquid.
Waleed F. Faris
Chapter 1.
Usually, micropumps are integrated with other microflow devices and sensors to allow precise control and sensing of flow of the order of microliters. Several pumping principles are available in the
literature and most of them mimic macroscale pumps. Figure 1.1 shows a diagram for a diaphragm
micropump.
Electrode
Isolator
Deformed Membrane
Chamber
Valves
Input
Output
Generally, micropumps are classified into two categories (Tay and Phoon, 1997): mechanical and
non-mechanical. Mechanical micropumps include reciprocating and peristaltic pumps, and nonmechanical micropumps include electrohydrodynamic pumps. Reciprocating micropumps with different actuating techniques are widely used. Piezoelectric, pneumatic, electrostatic, and thermopneumatic actuation principles are used. Electrostatic actuation is gaining popularity because of its
simplicity and high-flow output pressures (Tay and Phoon, 1997). Different shapes of electrostatic
actuating elements are reported in the literature, such as rectangular, square, circular, and annular
(Saif et al., 1999; Tay and Phoon, 1997). They are also used in other MEMS devices, such as
sensors, valves, deformable micromirrors, and micro-turbo generators.
Several transduction techniques have been used for micromachined devices, in general, and micropumps and sensors, in particular. These include piezoelectric, capacitive, optics, and resonance.
So, it is clear now that diaphragm-based MEMS devices are among the most common devices.
Also, capacitive or electrostatic actuation is commonly used in commercial applications because
they suffer less losses and need less mechanisms for their operation. Temperature affects the
Waleed F. Faris
Chapter 1.
sensitivity and performance of such devices to a considerable extent, especially sensors used in
high-temperature environments, such as exhaust manifolds in cars, propulsion engines, and steel
making industry. Also, temperature affects the measurement accuracy of other MEMS devices,
such as accelerometers and resonance-based calibration instruments.
The main objective of this dissertation is to analyze the mechanical response, static and dynamic,
of circular diaphragms under electrostatic and/or thermal loading.
1.1.1
Problem Statement
Analysis of the structural dynamics of MEMS devices poses new challenges. Classical analysis
in structural dynamics does not lend itself easily to the new types of forcing and nonlinearities
one encounters in dealing with MEMS. New approaches have to be developed to deal with these
challenges, combining the rigor of theory and the practicality of engineering design approaches.
In this dissertation, we study the linear and nonlinear problems of thin circular and annular plates
under electrostatic and/or thermal loading. We use different solution techniques to study the local
and global behavior of the dynamic responses. For the local analysis, we use a combination of
perturbation and numerical techniques. For the global analysis, we use a reduced-order model
based on the Galerkin approximation.
1.2
Plates are flat structures whose thickness h is small compared to the other in-plane dimensions.
For a circular plate, the only in-plane dimension is the radius R.
Plate theories are classified in many ways. One way of classification is based on the thickness
(Bairagi 1986); that is, thin- and thick-plate theories. We focus on thin-plate theories because
most of the applications that we address in this work are thin plates. Since we are concerned with
circular and annular plates, the theoretical development we present here is based on the cylindrical
coordinate system (r, , z). Geometrically, a plate is said to be thin if its thinness ratio h/R remains
less than 1/20, otherwise, the plate is said to be thick. In thin-plate theories, the stresses z , rz ,
Waleed F. Faris
Chapter 1.
and z , which are directed towards the z-axis across the thickness, are considered to be of negligible
magnitude compared to the other stresses. In thick-plate theories, the stress components z , rz ,
and z are comparable to the other stresses and cannot be neglected. The various in-plane stresses
r , , rz do not remain proportional to the distance z measured from the neutral plane though
the variation of the stresses rz and z across the thickness can still be assumed to be parabolic.
Thin plates can withstand external loads and moments and develop resistive loads and moments
which can be classified into two distinct groups of action:
1. Membrane (in-plane) action.
2. Bending (transverse) action.
Thin-plate theories are classified according to the transverse deflection w relative to the thickness
h as follows:
1. Plates with small deflection.
In this case, w is small compared to the plate thickness. In theory w must be 0.2h for
the application of thin-plate theory with small deflections to give accurate results, but in
practice this theory is used for deflections up to the order of the plate thickness with fairly
good results for the static case. Because of the smallness of the deflection, the slope and any
higher-order terms in the slope or deflection are neglected. Accordingly, in small-deflection
theory, the bending effect is dominant whereas the membrane effect is totally disregarded.
Though, there are several trials to derive theories within the small deflection framework, the
most in use is the Kirchoff plate theory, which is sometimes referred to as the classical plate
theory or (CPT).
2. Plates with large deflections.
When w h, the results calculated by the small-deflection theory are usually not in good
agreement with the observed results. Such cases are better treated by using theories developed
on the basis of large deflection.
Large deflections should be accounted for, also, in cases that do not necessarily involve deflections larger than the plate thickness. For example, as the thickness of the plate tends to
Waleed F. Faris
Chapter 1.
6
Eh3
,
12(1 2 )
causing the plate to be more flexible. Another example is plates subjected to the action of
in-plane loads. A third example is plates operated in the elasto-plastic range. In the case of
large deflection, both membrane and bending forces take part in resisting the action of the
loads. The larger the deflection is, the greater are the curvature and the membrane actions
and vice versa.
There are several theories for plates under large deflections, the most commonly used of them
is the von Karman plate theory which is sometimes referred to as the Kirchoff-Foppel plate
theory, we use this theory in this thesis. Berger in 1955 (Chia, 1980) proposed an alternate
formulation for the nonlinear static isotropic theory of plates, in which he disregarded the
elastic energy due to the second invariant of the membrane strain compared to the square of
the first invariant. This hypothesis came to be known in the literature as Bergers approximation or hypothesis and the resulting plate equations are known as Bergers plate equations
or model. Nowinski and Ohnabe (1972) showed that the Berger approximation may lead to
grave inaccuracies and even become meaningless if the edge of the plate is free to move in the
in-plane directions.
3. Plates with excessive large deformations.
Plates having very thin sections offer practically no bending resistance to deformation. Also,
plates subjected to excessive loads particularly towards the stage of collapse, are susceptible
excessive deflections. In such cases, a plate will behave as a perfectly flexible structure known
as membrane.
1.3
Literature Review
There is a large number of publications concerning plate theories and plate dynamics, which can
not be covered here. The key issues in our research are nonlinear dynamics of plates and plates
under thermal and/or electrostatic loading. Hence, in the following sections, we review the main
works on the nonlinear dynamics of plates, plates under thermal loading, plates under electrostatic
loading, plates under both loadings, and finally postbuckling analysis of plates under both loadings.
Waleed F. Faris
Chapter 1.
1.3.1
Here, we review the static and dynamic postbuckling of circular plates. We classify the literature
according to the method of solution to make it easier for the reader.
Waleed F. Faris
Chapter 1.
Hadian and Nayfeh (1990) used the method of multiple scales to study the response of circular plates
to a harmonic external excitation. They found out that the multi-mode response loses stability
through a Hopf bifurcation, resulting in harmonically and chaotically modulated motions.
Oh and Nayfeh (1998) have shown experimentally a complicated dynamic behavior of a cantilever
(90/30/-30/-30/30/90)s graphite-epoxy plate. The plate was base excited using a 2000-lb table
shaker near its seventh (third torsional) mode. They monitored the excitation using a base-mounted
accelerometer and the plate response using a laser vibrometer. For some excitation amplitudes and
frequencies, they observed the activation of a low-frequency (first bending) mode accompanied by
amplitude and phase modulations of the seventh mode. Therefore, they concluded that this type
of modal interactions is possible for all ranges of structural stiffnesses and configurations whenever
there exist modes whose natural frequencies are much lower than the natural frequencies of the
directly excited modes. In addition, they also observed simultaneous activation of a two-to-one
internal resonance along with the high- to low-frequency modal interactions for some excitation
amplitudes and frequencies. They used time-history, power-spectra, and force-response plots to
characterize the dynamics of the plate.
Yu et al. (2001) used the method of multiple scales to analyze the nonlinear vibrations of a simply
supported thin rectangular plate and derived a set of ordinary-differential equations governing the
modulation of the amplitudes and phases of the response. They discussed heteroclinic bifurcation
in details.
Azrar et al. (2002) investigated the nonlinear forced vibrations of thin elastic rectangular plates
using an asymptotic-numerical method. Various types of harmonic excitations were considered.
Using the method of harmonic balance and Hamiltons principle, they converted the equations of
motion into an operational formulation. They used the finite-element method to obtain a nonlinear
solution for a given frequency and amplitude of excitation. Applying perturbation techniques in
the vicinity of this solution, they transformed the nonlinear governing equation into a sequence of
linear problems having the same stiffness matrix, which can be easily computed.
Nath and Alwar (1980) studied the nonlinear dynamic response of orthotropic circular plates for
both clamped and simply supported boundary conditions. They expressed the response in terms of
Chebychev polynomials. They focused on the effect of the orthotropic parameter on the nonlinear
Waleed F. Faris
Chapter 1.
response of the plate under three types of dynamic loadings: step function, sinusoidal, and N-shaped
pulse. Nath (1982) studied the nonlinear dynamic response of circular plates resting on Winkler
and Pasternak type elastic foundations. He used Chebychev polynomials and the implicit Houbolt
technique for solving the problem in space and time, respectively. He investigated the effect of the
foundation parameters on the nonlinear response of the plate with clamped and simply supported
boundary conditions. Nath and Kumar (1995) studied the nonlinear dynamic response of cylindrically orthotropic, symmetrically laminated, cross-ply moderately thick plates using Chebychev
polynomials. They demonstrated the effects of transverse shear, rotary inertia, material properties,
number of layers, and boundary conditions on the static and dynamic responses of plates.
Galerkin Method
Yamaki (1961) used Galerkin single-mode expansion to solve the von Karm
an equations directly
for circular plates. He considered simply supported and clamped cases, both with movable and
immovable edges.
Nowinski (1962) utilized the von Karm
an dynamic equations to investigate of the free nonlinear
axisymmetric vibrations of a circular plate built-in at the boundary. He represented the deflections
using a series of separable terms and used an orthogonalization procedure to eliminate the space
variable. Confining the study to one term of the series, the solution in the time variable was found
in the form of elliptic functions.
Wah (1963) used Bergers approximation of the von Karm
an equations to study large-amplitude
free vibrations of clamped and simply supported circular plates. He used a single-mode Galerkin
expansion and studied the ratio between the linear frequencies and their respective nonlinear ones.
Leung and Mao (1995) discretized the Lagrangian of continuous systems by Galerkins method
to obtain the discrete Hamiltonian and Hamiltons equations. They applied symplectic numerical
integration schemes to the resulting ordinary-differential equations to construct the phase diagram. They showed that, due to modal coupling, bounded quasi-periodic and non-periodic free
vibrations are very common for continuous systems. Comparison is made with the Runge-Kutta
method. They showed that the Runge-Kutta integration reduces the total energy of an undamped
Waleed F. Faris
Chapter 1.
10
system while symplectic integration almost preserves it. The symplectic schemes are extended to
ordinary-differential equations by means of the Lagrangian and Hamiltonian, and they used them
to investigate the nonlinear vibrations of beams and plates. Nonlinear modal coupling is emphasized. They noted that extension of these schemes to shell structures involves the complication of
curvature effects.
Abe et al. (1998) used a single-mode Galerkin procedure to reduce the equations of motion of a
laminated plate to a Duffing-type equation in terms of the transverse displacement. They analyzed solutions of the resulting equation using the method of multiple scales. They investigated
the influence of the lamination sequence, the thickness ratio, the number of layers, and the inplane boundary conditions on the subharmonic-resonance response. Abe et al. (1998) investigated
three-mode responses of simply supported laminated plates to harmonic excitations by using a
combination of the Galerkin procedure and the method of multiple scales. They compared their
analytical results with numerical integration results and found good agreement.
Wei (2001) analyzed the global bifurcations and chaos of the response of a simply supported rectangular thin plate to a parametric excitation by applying the Galerkin procedure to the von Karm
an
equations. The method of multiple scales was used to obtain the variational equations. Based on
the averaged equations, they used the theory of normal forms to derive an explicit expression of
the normal form and used it to carry out a global bifurcation analysis.
Zhang (2001) studied the global bifurcations and chaotic dynamics of a parametrically excited
simply supported rectangular thin plate. The Galerkin approximation was used to reduce the von
Karman equations into a system of nonlinear ordinary-differential equations. Then, they used the
method of multiple scales to obtain the variational equations. They used the method of normal
forms to obtain an explicit normal form and used it to analyze global bifurcations.
Waleed F. Faris
Chapter 1.
11
Waleed F. Faris
Chapter 1.
12
the DR method for prebuckling and postbuckling analyses of structures through various numerical
examples. Also, they studied the effect of these higher-order terms in the membrane and curvature
expressions on the overall accuracy of the solution for geometrically nonlinear problems.
Zhu et al. (1995) presented a technique combining the modal expansion with the finite-element
representation of a finite region containing defects to analyze the transient response of a laminated
composite plate. As an illustration, a model problem of a laminated plate containing a normal
surface breaking crack was studied in detail. They presented numerical results showing the effect
of the crack depth on the transient response.
Exact Solution
Maganty and Bickford (1987) derived an accurate set of geometrically nonlinear equations of motion
of circular rings. They concluded that, for nonresonant motions, the initial out-of-plane amplitude
has a marginal effect on the initial in-plane oscillation, whereas the effect of the initial in-plane
amplitude on the out-of-plane oscillation is significant. For resonance motions, they indicated that
there is an exchange of energy between the in-plane and out-of-plane motions.
1.3.2
Although thermal loading of plates is an old subject, very few surveys do exist on the subject. The
most recent of them is by Tauchert (1991) who covers a wide range of topics, such as composite
plates, buckling, and post-buckling. Thornton (1993) reviewed the thermal buckling of plates and
shells since the fifties till the time of his publication. Thermal induced vibrations of isotropic
plates occupy a small portion of the survey. In the following, we present an extensive survey. The
literature on this subject can be classified into two categories. The first category focuses on the
static behavior and the second focuses on thermally induced vibrations. In the following, we review
each category, starting with the first.
Waleed F. Faris
Chapter 1.
13
Waleed F. Faris
Chapter 1.
14
Rossit and Laura (1997) determined the static deflections for two different thermoelastic problems
using the Ritz method employing a unified approach that consists of using polynomial coordinate
functions that satisfy the governing boundary conditions. The first problem was thermoelastic
stresses and deflections in a thin rectangular plate with edges restrained against rotation. The
second problem was thermoelastic plane stress in a rectangular plate with stress-free edges. Both
problems were solved within the framework of linear plate theory.
Waleed F. Faris
Chapter 1.
15
with simply supported and clamped boundary conditions. Jones et al. (1980) examined the largeamplitude vibrations and buckling behavior of a rectangular plate under elevated temperatures
including viscoelastic effect by using Bergers approximation. Chang and Wan (1986) analyzed analytically large-amplitude thermomechanically coupled vibrations of rectangular elastic thin plates
with different boundary conditions using Bergers approximation. They carried out a one-term
Galerkin approximation and found that the effect of coupling is negligible as long as the damping
effect is not required.
Sunakawa (1964) solved approximately the fundamental equations of nonlinear flexural vibration
for a rectangular elastic plate by the method of successive approximation and demonstrated the
influences of temperature changes and large amplitudes on the period of free vibrations. Some
numerical examples were given for a rectangular plate with hinged and immovable edges. He
emphasized the considerable effect of even small temperature differences.
Buckens (1979) examined small deflections of rectangular plates under thermal loads and analyzed
their buckling. Then, he studied the postbuckling behavior using the von Karm
an equations.
Mansfield (1982) gave an exact analysis of certain large-amplitude free vibrations of unsupported
elliptical plates of lenticular section, whose middle surfaces may be flat or have small constant
curvatures. The temperature in the plates may vary quadratically in the plane and linearly through
the thickness. Biswas and Kapoor (1985) studied the effect of temperature on the time period of
vibration of nonlinear orthotropic circular plates with clamped boundary conditions. They used
the von Karman equations written in terms of tranverse and radial displacements and solved them
using a one-term Galerkin approximation.
Sathyamoorthy (1986) studied the effect of thermal loading on large-amplitude free vibrations of
orthotropic clamped plates. He assumed a temperature distribution and solved the von Karm
an
equations to find the mechanical response using a single-mode Galerkin approximation.
Trajkovski and Cukic (1999) studied thermoelastic vibrations of free supported and clamped thin
circular plates caused by a thermal shock impinging on their surface. The system of partialdifferential equations of motion of the coupled system was reduced to a system of first- and secondtype Volterra integral equations in the time domain. In both cases, the solutions were given in the
Waleed F. Faris
Chapter 1.
16
1.3.3
The electrostatic load applied to a plate has an upper limit beyond which the mechanical restoring
force is overcome by the opposing electrostatic force and the structure collapses. This structural
instability phenomenon is known as pull in, and the critical voltage associated with it is called
the pull in voltage.
The literature on plates under electrostatic loading is increasing rapidly due to their wide applications. Microphones, micropumps, sensors, and microswitches are only examples of such vast range
of applications.
One of the earliest works on the subject is that of Warren et al. (1972) who presented nonlinear
equations for the static deflection of the moving electrode due to the electrical field between the
electrodes in capacitance microphones. The equations were solved numerically for circular and
annular membranes. Because of the nonlinearity of the electrical attraction forces, certain values
Waleed F. Faris
Chapter 1.
17
of the microphone parameters exist for which the deflections can become unstable. Warren et
al. (1973) studied the dynamic response of a membrane to an acoustic excitation using finite
differences. They included the effects of static deflections and thin air film between the electrodes.
Wang and Hadaegh (1996) studied micromachined deformable mirrors modeled by either membrane
or plate equations with nonlinear electrostatic actuation. They used numerical methods to compute
the mirror deformation caused by a given actuator voltage and the actuator voltage required for
producing the desired deformations at the actuator locations. They applied their technique to
circular deformable mirrors whose surfaces are modeled by elastic membranes.
Cozma and Puers (1997) investigated the possibility of using electrostatic forces to actuate the
membrane of a pressure sensor in order to create self-testing features. It was shown that the
electrostatic pressure is small compared with the external pressure, unless large voltages or electrical
charges are used. The sensor diaphragm was modeled as a membrane and the finite-element method
was used to solve for the membrane deflection under electrostatic actuation.
Bourouina and Grandchamp (1996) modeled a square-shape micropump under electrostatic loading
using the electrical analogy and the idea of equivalent network. They assumed the micropump
diaphragm to behave like a nonlinear membrane. However, they did not consider the dynamic
behavior of the membrane, as their main interest was the calculation of the flow rate.
Francais et al. (1997) developed a mechanical law with a simple elementary model of the membrane deflection that approximates the micropump square diaphragm. They demonstrated the
static instability (pull-in) of such a system under electrostatic actuation and provided an analytical
solution for the membrane deflection. Also, they proposed a parameter optimization procedure for
the dimensions of the electrostatically actuated membrane in order to achieve maximum volume
displacement. Francais and Defour (1998) studied a micropump made of a square diaphragm for
medical applications. They modeled the diaphragm as a membrane and included two nonlinear
characteristics: the pull-in effect and the hysteresis phenomenon. They solved the problem using
circuit analogy.
Jiang et al. (2000) studied a micropump actuated by a electrostatic force. They used the boundaryelement method to solve the Laplace equation governing the potential difference. A combination
Waleed F. Faris
Chapter 1.
18
of the assumed-mode method and the boundary-element method was employed to solve for the
diaphragm deflections. They presented numerical results showing the time history of the transverse
deflection.
Rajalingham and Bhat (1998) studied in two steps the vibration of a microphone diaphragm subjected to an electrostatic force field, which induces a deflection-dependent load intensity. The
diaphragm was modeled using linear circular plate theory. First, they calculated the static deflection due to the electrostatic load and found that it increases with the field strength. It was noted
that, as the field strength reaches a critical value, it destabilizes the static deflection and causes
the plate to buckle. Second, they solved for small vibrations around the static deflected position.
Saif et al. (1999) presented a methodology based on energy methods to analyze electrostatically actuated micropumps. The micropump diaphragm was modeled as a linear membrane. The
methodology was used to analyze two different types of micropumps: paraboloid and annular cavity
membranes. The relation between the pressure and deflection was demonstrated, but no dynamic
analysis was carried out.
Francais and Dufour (1999) proposed a complete normalized study the behavior of a diaphragm
under classical excitations encountered in microsystem: pneumatic, electrostatic, piezoelectric, and
magnetic. They presented a traditional approximate analytical solution based on energy minimization through the use of a polynomial solution technique. They used both membrane and linear
plate formulations to study the pull-in phenomenon and the relation between the deflection and
the electrostatic loading for different diaphragm shapes.
Konig and Wachutka (1999) analyzed the instability inherent in electrostatically driven microdevices. Starting from a simplified lumped model of two rigid plates under electrostatic loading,
they formulated the governing differential equations and analyzed the stability of their solutions.
They presented a homotopy method that overcomes the stiffness of the problem. They presented
numerical results for micromirrors and membranes.
Seeger and Boser (2002) characterized parallel-plate actuators for oscillations near the mechanical
resonant frequency and amplitudes comparable to the actuator gap. They showed that, at resonance, the structure can move beyond the static pull-in limit, which they calculated as one-third of
Waleed F. Faris
Chapter 1.
19
the gap, but it is still limited to 56 percent of the gap by resonant pull-in. Above the resonant frequency, the structure is not limited by pull-in and can theoretically oscillate across the entire gap.
They developed a function model, which included an amplitude-dependent model for electrostatic
spring tuning, to predict the steady-state frequency response and they verified it experimentally
1.3.4
Here, we review the static and dynamic postbuckling of circular plates. We classify the literature
according to the method of solution to make it easier for the reader.
Waleed F. Faris
Chapter 1.
20
ear algebraic equations. A perturbation procedure was used to examine the effect of imperfection
and aspect ratio on the static transverse deflection of a plate subjected to an in-plane edge displacement. They expressed each unknown in powers of a perturbation parameter, which reduced the
nonlinear equations to a system of equations that can be solved successively by elementary linear
algebra techniques. The segmented deflection history was developed for a variety of imperfection
compositions and aspect ratios.
Brewster (1991) derived an asymptotic expansion for the solution of the axisymmetric von Karm
an
thin circular plates equations under transverse and compressive in-plane loads, with the edge elastically supported against rotation. The equations were reformulated in a form amenable to solution
by singular perturbation techniques, where the perturbation parameter was proportional to the
plate thickness. Solutions of this nonlinear singular perturbation problem were related to the Foppel membrane solutions and exhibited global breakdown. His formal analysis suggested that the
relationship between compressive membrane and plate solutions does not extend to the case of
clamped plates.
Shen (1997) studied the thermal postbuckling of a simply supported composite laminated rectangular plate subjected to uniform and non-uniform tent-like temperature loadings. The formulation is
based on Reddys higher-order shear-deformation plate theory with inclusion of thermal effects. The
analysis was carried out using a mixed Galerkin-perturbation technique to determine the thermal
buckling loads and postbuckling equilibrium paths. The solution procedure starts by substituting
assumed summations of the plate equations variables; namely, the deflection and stress function.
The assumed summations are multiplied by a small perturbation parameter, which upon substitution yield a system of perturbation equations that are solved step by step using the Galerkin
approximation. The effects of transverse shear deformation, thermal load ratio, plate aspect ratio,
total number of plies, fiber orientation, and initial geometric imperfections were studied.
Wang (2000) obtained an analytic axisymmetric postbuckling equilibrium path for circular plates
under uniform radial compression using a power-series method. The solution procedure is based on
a second-order differential equation in the vicinity of the normal singular point. He also investigated
the problem of asymmetric bifurcation buckling from the axisymmetric postbuckling deformation
state using the adjacent equilibrium criterion. The von Karman nonlinear equations in the incre-
Waleed F. Faris
Chapter 1.
21
mental form had been solved using power-series and Fourier-series expansions in the circuferencial
direction.
Galerkin Approximation
Gossard et al. (1953) studied the static behavior of flat and initially imperfect rectangular plates
subjected to thermal buckling. They used the von Karm
an model for the plates and a one-term
Galerkin expansion. The method was used to determine the deflections of a simply supported
panel subjected to a tent-like temperature distribution over the plate surface. The theoretical
and experimental results were in good agreement for the range of temperatures and deflections
considered in the test.
Shulman (1958) derived equations of motion for thermally stressed plates both before and after
initial buckling. He used a linear plate model for prebuckling and the von Karm
an model for
postbuckling. For prebuckling, he solved the problem exactly. For postbuckling, he separated the
static part from the dynamic part by assuming that the deflection is the sum of the static deflection
and a dynamic part around this deflection. The stress function also was expressed similar to the
deflection. These equations were solved by the assumed mode method, which is to a great extent
similar to the Galerkin approximation. These equations yield the free-vibration characteristics
around the two equilibrium states produced by the temperature rise. They considered uniformly
heated plates with two opposite edges simply supported and generalized support conditions on the
other two edges.
Ahmad (1970) developed a one-term Galerkin approximation to study the postbuckling response
of a thin elastic circular plate under the action of an axisymmetric, time-dependent radial thrust,
assuming that the thrust produces resonance of only the principal mode of vibration. He showed
that this assumption is valid for certain simple cases of loading history and for small penetrations
of the postbuckling domain. He presented the radial variation of the stresses and the time history
of the deflection for the case of pulse loading.
Dumir (1986) developed an approximate solution for the large axisymmetric responses of isotropic
thin circular plates resting on Winkler, Pasternak, and nonlinear Winkler foundations, taking into
Waleed F. Faris
Chapter 1.
22
account the in-plane displacement. He considered plates with edges elastically restrained against
rotation. The von Karman plate equations in terms of the transverse deflection and the stress
function were employed. A one-term mode shape was used to approximate the transverse deflection
and Galerkins approximation was used to obtain an equation for the central deflection in the form
of Duffings equation. He determined the nonlinear frequencies, postbuckling response to a radial
load at the edge, and the maximum transient response to a transverse step load.
Ritz Method
Ng and White (1988) studied the dynamic behavior of postbuckled isotropic rectangular plates
through four main steps. First, they derived a simple formula using a one-term Rayleigh method
to give a simple preliminary solution; this solution does not give accurate results for a plate with
a multi-mode imperfection and for vibration modes different from the critical mode. Second, they
used a multi-mode Rayleigh-Ritz approach. Third, they used the finite-element method. And the
final step was to use experiments to validate the theoretical solutions. Comparison showed that
the one-term simple formula was not adequate and that the multi-mode Rayleigh-Ritz and finiteelement results were quite similar. The experimental results showed a clear discrepancy with all of
the three theoretical methods.
Analytic Approaches
Mansfield (1962) gave an exact solution for the bending, buckling, and curling of a thin circular plate
of lenticular section with a uniform temperature gradient through its thickness. The behavior of
such a plate with an initial spherical curvature, susceptible to a further possibility of snap-through
buckling, was also studied.
Gauss and Antman (1984) studied the global properties of the buckled states of nonlinear thermoelastic beams and plates heated at their ends and edges. For plates, they considered axisymmetric
circular plates only. Their analysis relied much on the combination of classical results of ordinarydifferential equations theory and that of bifurcation theory. They emphasized the crucial role of
the constitutive assumptions.
Waleed F. Faris
1.4
Chapter 1.
23
Contributions
Waleed F. Faris
Chapter 1.
24
method. We conclude that a single-mode approximation might give erroneous results and that
implementation of the multi-mode approximation is easier and faster than the power-series
method.
6. We illustrate the importance of including nonlinearity in modeling sensors and other similar
diaphragm-based devices .
Chapter 2
General von K
arm
an Formulation of
Circular and Annular Plates under
Thermal and Electrostatic Loadings
2.1
Basic Assumptions
Waleed F. Faris
Chapter 2.
General Formulation
26
2
r2 , R
r1
2.2
Problem Formulation
The displacement field is assumed to have the form (Librescu, 2001; Reddy, 1999)
w
r
(2.1)
1 w
r
(2.2)
ur (r, , z, t) = u(r, , t) z
u (r, , z, t) = v(r, , t) z
uz (r, , z, t) = w(r, , t)
(2.3)
where u, v, and w are the radial, angular, and transverse displacements of the plate middle surface,
respectively. The nonlinear strains according to von Karm
an are
ur
1 uz 2
rr =
+
r
2 r
1 u
1
ur
+
+
=
r
r
2
zz =
uz
=0
z
1 uz
r
(2.4)
2
(2.5)
(2.6)
Waleed F. Faris
Chapter 2.
General Formulation
ur
1 uz uz
u
2r
u +
+
r
r r
1 u
1 uz
z =
+
=0
2 z
r
1 ur
uz
1 uz uz
rz =
+
+
=0
2 z
r
2 r z
1
=
r
27
(2.7)
(2.8)
(2.9)
We note that zz , z , and rz given by Equations (2.6), (2.8), and (2.9) are identitically zero in the
thin-plate theory approximation. Substituting the displacement field into Equations (2.4), (2.5),
and (2.7) yields
u 1 w 2
2w
rr =
+
z 2
r
2 r
r
u 1 v
1
w
z w 1 2 w
= +
+
+
r
r 2r2
r r
r 2
1 1 u v v 1 w w
2 w
1 w
=
+
+
z
2 r
r
r r r
r r r
(2.10)
(2.11)
(2.12)
(2.13)
where W is the total work: the strain energy, kinetic energy, and work done by the external loads
(Ugural, 1981). In polar coordinates, this principle becomes
Z
W =
0
0
T
h
2
h
2
(rr rr + + r r )dzrdrddt
h
2
h
2
(u r u r + u u + u z u z )dzrdrddt
Z
q(r, , t)wrdrddt
0
(2.14)
where q(r, , t) is the distributed transverse load, T is the final time, and 0 denotes the domain
of the plate.
We define
Z
Nrr =
h
2
Z
N =
h
2
h
2
h
2
rr dz
(2.15)
dz
(2.16)
Waleed F. Faris
Chapter 2.
General Formulation
Z
Nr =
Z
Mrr =
h
2
h
2
h
2
Z
Mr =
h
2
h
2
Z
M =
h
2
h
2
h
2
28
r dz
(2.17)
rr zdz
(2.18)
zdz
(2.19)
r zdz
(2.20)
r = 2r
(2.21)
where Nrr , N , Nr are the in-plane forces and Mrr , M , Mr are the out-of-plane moments. Substituting Equations (2.10)-(2.12) and (2.15)-(2.20) into Equation (2.14), we obtain
Z TZ
u w w
1 w w 1 v u
Nrr
+
+ N
+
+
r
r r
r2
r
r
0
0
1 u v v 1 w w
2 w
1 w 1 2 w
+Nr
+
+
+
Mrr
+
r
r
r
r r
r2
r
r
r 2
2
1 w 1 w
I0 (u
u + v
v + w
w)
2Mr
r r
r
w w
1 w w
I2
+ 2
qw rdrddt = 0
r r
r
(2.22)
where I0 is the mass moment of inertia and I2 is the rotary inertia. Integrating Equation (2.22)
by parts in space and time to relieve u, v, and w from any differentiations, we find that each
individual term simplifies as follows:
Z
Z
Z
u
u
Nrr
r2
Nrr
rdrd =
(rNrr )
drd = (rNrr ) u|r1
Nrr + r
udrd
r
r
r
0
0
0
Z
w w
w w
Nrr
rdrd =
rNrr
drd
r r
r
r
0
0
Z
w
w
r2
= rNrr
w|r1
rNrr
wdrd
r
r
0 r
Z
Z
Z
v
N
1 v
rdrd =
N
drd = N v|21
vdrd
N
r
0
0
0
(2.23)
Z
w w
1
w w
1
N
rdrd =
N
drd
2
r
0
0 r
Z
1
w
1
w
=
N
w|21
N
wdrd
r
(2.24)
(2.25)
(2.26)
Waleed F. Faris
Chapter 2.
General Formulation
Z
Z
v
v
Nr
r2
Nr
(rNr )
Nr + r
rdrd =
drd = (rNr ) v|r1
vdrd
r
r
r
0
0
0
Z
Z
Z
u
Nr
1 u
2
Nr
Nr
rdrd =
drd = Nr u|1
udrd
r
0
0
0
29
Z
w r2
2 w
rdrd = rMrr
Mrr
|r1
rMrr
drd
r2
r
r
0
0 r
Z
w r2
2
r2
(rMrr ) wdrd
= rMrr
|r1
(rMrr ) w|r1
2
r
r
0 r
Z
Z
Z
1 w
w
M
r2
M
rdrd =
M
drd = M w|r1
wdrd
r r
r
0
0
0 r
(2.27)
(2.28)
(2.29)
(2.30)
Z
Z
1 2 w
1 2 w
1
w 2
1 M w
M 2
rdrd =
M
drd = M
|
drd
r 2
r 2
r
1
0
0
0 r
Z
w 2
M
1
2 M
|1
w|21
= M
wdrd
(2.31)
2
r
0
Z
Z
w 1 w
1 2 w 1 w
rdrd = 2
drd
2
Mr
Mr
r r
r
r
r
0
0
Z
Z
Mr w
1
1 Mr
w 2
2
|
drd + 2
Mr w|1
wrdr
= 2 Mr
r 1
r
r
0
0 r
Z
w 2
Mr
2 Mr
r2
= 2 Mr
|
w|r1
wdrd
r 1
0 r
Z
1
1 Mr
+2
Mr w|21
wrdr
(2.32)
r
0 r
Z
Z TZ
I0 (u
u)
rdrddt =
(rI0 ) u
udrddt
Z TZ
2u
u t1
u|t1
(rI0 ) 2 udrddt
= (rI0 )
t
t
(2.33)
I0 (v
v)
rdrddt =
(rI0 ) v
vdrddt
Z TZ
2v
v
(rI0 ) 2 vdrddt
= (rI0 ) v|tt11
t
t
(2.34)
I0 (w
w)
rdrddt =
(rI0 ) w
wdrddt
Z TZ
w
2w
t1
= (rI0 )
w|t1
(rI0 ) 2 wdrddt
t
t
(2.35)
Waleed F. Faris
Z
Chapter 2.
General Formulation
2 w w t1
2 w 2 w
I2
rdrddt = (rI2 )
|
rt rt
rt r t1
(rI2 )
2 w w
drddt
rt r
30
(2.36)
Substituting Equations (2.23)-(2.36) into Equation (2.22) and setting each of the coefficients of
u, v, w in the area integrand equal to zero, we obtain
1 (rNrr ) Nr
2u
u :
+
N I0 2 = 0
r
r
2v
1 (rNr ) N
+
+ Nr I0 2 = 0
v :
r
r
(rNrr w
1 (rMrr ) M
2 Mr
1 2 M
2 Mr
r )
w :
+
2
+
+
+
2
r
r
r
r
r
r
r
!
w
w
1 (N )
2w
1 (r r )
1 2w
+
+ q I0 2 + I2 2
+ 2 2 =0
r
t
t
r r
r
(2.37)
(2.38)
(2.39)
And setting each of the coefficients of u, v, w in the boundary integral equal to zero, we obtain
the following boundary conditions:specify
for r = r1 , r2
N or u = 0
rNr or v = 0
(2.40)
(2.41)
1
w
w
(rMr ) 1 M
+
rMr + N
+ Nr
= 0 or w = 0
r
r
r
(2.42)
1
w
M = 0 or
=0
r
(2.43)
Nrr or v = 0
(2.44)
For = 1 , 2
Nr or u = 0
(rMrr )
Mr
w
w
+2
M + rNrr
+ Nr
= 0 or w = 0
r
Mrr = 0 or
w
=0
r
(2.45)
(2.46)
(2.47)
(2.48)
Waleed F. Faris
Chapter 2.
General Formulation
31
We consider linear thermoelastic isotropic plates in which the elastic stiffnesses are independent of
the temperature and obtain
rr =
E
E
(rr + )
T
1 2
1
(2.49)
E
E
( + rr )
T
2
1
1
(2.50)
r = 2Gr
where is the coefficient of thermal expansion and G =
(2.51)
E
2(1+) .
h
2
h
2
Z
N =
h
2
h
2
E
E
(rr + )
T )dz
2
1
1
(2.52)
E
E
( + rr )
T )dz
2
1
1
(2.53)
Z
Nr =
Z
Mrr =
h
2
h
2
Z
M =
h
2
h
2
h
2
h
2
(2Gr )dz
E
E
T )zdz
(rr + )
2
1
1
(2.55)
E
E
T )zdz
( + rr )
2
1
1
(2.56)
Z
Mr =
h
2
h
2
(2Gr )zdz
v
1 w 2
NT
Eh
1 w 2 u
+
+
u+
+
Nrr =
2
1
2 r
r
r
2r
1
N
(2.54)
!
Eh
1 w 2 u
1
v
1 w 2
NT
=
+
+
u+
+
2
1
2 r
r
r
2r
1
v v 1 u 1 w w
Nr = G
+
+
r
r
r
r r
Z h
2
NT = E
T (r, , z, t)dz
h
2
(2.57)
(2.58)
(2.59)
(2.60)
(2.61)
Waleed F. Faris
Chapter 2.
General Formulation
2 w w 1 2 w
MT
+
Mrr = D
+ (
)
2
2
r
r r
r
1
2
2
w 1 w 1 w
MT
M = D 2 + (
+
)
2
r
r r
r
1
2
1 w
1 w
Mr = (1 )D
r r r
Z h
2
MT = E
T (r, z)zdz
32
h
2
(2.62)
(2.63)
(2.64)
(2.65)
When the in-plane natural frequencies are large compared with the transverse natural frequencies,
the in-plane inertia term can be neglected. Then, we introduce the stress function F , which
sometimes referred to as the Airy function, defined by
Nrr =
1 2F
1 F
+ 2 2,
r r
r
N =
2F
,
r2
Nr =
1 2F
1 F
r2
r r
(2.66)
which satisfy Equations (2.36) and (2.37) exactly. Substituting the axisymmetric form of Equation
(2.66) into the axisymmetric form of Equation (2.38), we obtain
D4 w + hw
=
1 2 w F
1 w 2 F
1
w
+
2 MT
2c
2
2
r r r
r r r
t
1
(2.67)
In the axisymmetric case, the compatibility equation can be deduced from the following equations:
rr =
1
(Nrr N ) + T
Eh
1
(N Nrr ) + T
Eh
1 F
2F
Nrr =
and N =
r r
r2
2
u 1 w
u
rr =
+
and =
r
2 r
r
=
u
1 w 2
1
1 F
2F
=
+
2 + T
r
2 r
Eh r r
r
1
u
=
r
Eh
2F
F
r2
r r
(2.68)
(2.69)
(2.70)
(2.71)
(2.72)
+ T
Eliminating u from Equations (2.72) and (2.73), we obtain the compatibility equation
3F
2F
1 F
1
w 2
T
r 3 +
= Eh
Ehr
r
r2
r r
2
r
r
(2.73)
(2.74)
Waleed F. Faris
Chapter 2.
General Formulation
33
Now adding the external force due to the electrostatic field to Equation (2.67), we obtain the final
form
1 2 w F
1 w 2 F
w
1
V 2
+
2c
2 MT +
2
2
r r r
r r r
t
1
2 (d w)2
3F
2F
1 F
1
w 2
T
r 3 +
= Eh
Ehr
2
r
r
r r
2
r
r
D4 w +hw
=
(2.75)
(2.76)
where V is the applied voltage, d is the capacitor gap distance, and is the dielectric constant of
the medium.
In the following work, we treat the electrostatic and/or thermal loadings.
Chapter 3
34
Waleed F. Faris
3.1
Chapter 3.
35
Problem Formulation
We assume that the transverse deflection w of the plate is axisymmetric and write the equations
of motion as
1 2 w 1 w 2
V 2
2w
w
+
+h 2 =
+
t
t
r r2 r
r r r2
2 (d w)2
Eh 2 w w
4 =
r r2 r
D4 w + c
(3.1)
(3.2)
where t is time, is the material density, c is the viscous damping, h is the plate thickness, d is
the capacitor gap distance, is the dielectric constant of the medium, V is the DC voltage, and
is the stress function. The modulus of rigidity D is
D=
Eh3
12 (1 2 )
(3.3)
where E is Youngs modulus and is Poissons ratio. In the axisymmetric case, the differential
operator 4 is given by
4 =
4
2 3
1 2
1
+
+ 3
4
3
2
2
r
r r
r r
r r
(3.4)
In the axisymmetric case, the fourth-order compatibility equation (3.2) can be reduced to the
following third-order equation (Chapter 2):
3 2 1
1
r 3 +
= Eh
r
r2
r r
2
w
r
2
(3.5)
r = Rr , w = dw , t = R2
c=
h
D
1/2
t , b = Rb
12(1 2 ) 5 1/2
h D
c , = Ehd2
R4
(3.6)
where R is the outer radius of the annular plate. Substituting Equation (3.6) into Equations (3.1)
Waleed F. Faris
Chapter 3.
36
1 w 2
3 2 1
=
r 3 +
r
r2
r r
2 r
where 1 =
12(1 2 )d2
h2
and 2 =
R4
.
2Dd3
(3.7)
(3.8)
(3.9)
(3.10)
To calculate the static deflection, we set the time derivatives in Equations (3.7) and (3.8) equal to
zero and obtain
1 2 ws s ws 2 s
2 V 2
+
+
r r2 r
r r2
(1 ws )2
1 ws 2
3 s 2 s 1 s
=
r
+
r3
r2
r r
2 r
4 ws =
(3.11)
(3.12)
We generate the equations of motion governing the dynamic behavior of the plate around the
deflected shape by substituting Equations (3.9) and (3.10) into Equations (3.7) and (3.8) and
dropping the terms representing the equilibrium position and higher-order terms. The result is
2u
u
1 2 u s 2 ws u 2 s ws 2 2 u u 2
4
+c
+
u
=
+
+
+
+ 2
+
t2
t
r r2 r
r2 r
r r2
r r2
r r
r r2
2
2
2
22 V
32 V
42 V
2
+
u3
(3.13)
3u +
4u +
(1 ws )
(1 ws )
(1 ws )5
2 i
3 s 2 s 1 s
1 h u ws
u
r 3 +
= 2
+
(3.14)
2
r
r
r r
2 r r
r
Next, we study two cases. First, we consider an annular plate clamped along both of the outer and
inner radii; this design was proposed by Saif et al. (1999). Second, we consider an annular plate
clamped along the outer radius and free along the inner radius. The latter design is still under
investigation.
Waleed F. Faris
3.2
Chapter 3.
37
Clamped-Clamped Case
We consider the case of immovable inner and outer edges. In this case, the radial displacement u
is equal to zero. The nondimensional boundary conditions for this case are
2
= 0 at r = b and 1
r2
r r
w
w = 0 and
= 0 at r = b and 1
r
(3.15)
(3.16)
2
1
2 V 2
1
1 00 0
wsiv = ws000 + 2 ws00 3 ws0 +
ws s + ws0 00s +
r
r
r
r
(1 ws )2
1 0
1 0 2
00
r000
w
s + s s =
r
2 s
00s 0s = 0 at r = b and 1
r
ws = 0 and ws0 = 0 at r = b and 1
(3.17)
(3.18)
(3.19)
(3.20)
We solved Equations (3.17)-(3.20) for a polysilicon annular plate with an outer radius R = 50m,
inner radius R = 5m, and thickness h = 1m, a capacitor gap d = 2m, and a range of electrostatic forces varying from zero to the forcing level where structural instability (pull-in) develops.
Figure 3.1 shows an axisymmetric cross-section of the plate as it deflects under an upward electrostatic force increasing from zero to the forcing level corresponding to 2 V 2 = 140.6. The evolution
of the static deflection is most pronounced in the mid-region of the cross-section. Figure 3.2 shows
variation of the maximum deflection WM ax for the same range of 2 V 2 as that in Figure 3.1. As
the electrostatic force, represented by 2 V 2 , increases, the maximum deflection increases.
To calculate the mode shapes and corresponding natural frequencies, we substitute ws (x) into
Waleed F. Faris
Chapter 3.
38
0.4
0.3
0.2
ws
0.1
0.2
0.4
0.6
0.8
Figure 3.1: Nondimensional deflection ws (r) of the clamped-clamped plate under an electrostatic
force ranging from 2 V 2 = 0 to 140.6.
Equations (3.13)-(3.16), drop the nonlinear and damping terms, and obtain
2u
1 2 u s 2 ws u 2 s ws 2
22 V 2
4
+
+
+
u
=
+
+
u
t2
r r2 r
r2 r
r r2
r r2
(1 ws )3
h u w i
3 s 2 s 1 s
s
=
r 3 +
r
r2
r r
r r
00 0 = 0 at r = b and 1
r
u = 0 and u0 = 0 at r = b and 1
(3.21)
(3.22)
(3.23)
(3.24)
To solve these equations for the undamped mode shapes and natural frequencies of the plate, we
assume a harmonic motion in the nth mode in the form
u(r, t) = n (r)ein t
(3.25)
(r, t) = n (r)ein t
(3.26)
where n (r) is the nth axisymmetric mode shape, n (r) is the associated stress function, and n is
the nth nondimensional natural frequency. Substituting Equations (3.25) and (3.26) into Equations
Waleed F. Faris
Chapter 3.
39
0.4
WMax
0.3
0.2
0.1
0
0
30
60
2V2
90
120
150
Figure 3.2: Variation of the maximum deflection WM ax of the clamped-clamped plate with 2 V 2 .
(3.21)-(3.24) yields
V2
2
1
1
1 00 0
2
2
niv + n000 2 n00 + 3 n0
+
n
n n + n0 00n =
r
r
r
r
(1 ws )2
0 0
1 0
00
r000
n + n n = ws n
r
0
00
n n = 0 at r = b and 1
r
n = 0 and n0 = 0 at r = b and 1
(3.27)
(3.28)
(3.29)
(3.30)
To solve this eigenvalue problem, we release the second boundary condition at r = 1.0 and solve
the boundary-value problem iteratively using a shooting method until n , n , and n converge
to within predefined tolerances. This procedure is repeated for each mode shape and associated
natural frequency and stress function, using various sets of initial guesses for the mode shape,
natural frequency, and stress function.
Figure 3.3 shows variation of the fundamental natural frequency 1 of the clamped-clamped plate
with 2 V 2 . The fundamental frequency of the plate decreases at low and high levels of the electrostatic force. In the mid-range, the fundamental frequency remains nearly constant due to the
counterbalance between the electrostatic force and in-plane stresses. As pull-in develops, the fundamental natural frequency is expected to approach zero. Our results do not reflect this fact because
Waleed F. Faris
Chapter 3.
40
30
20
10
0
0
30
60
2V2
90
120
150
Figure 3.3: Variation of the fundamental natural frequency of the clamped-clamped plate with
2 V 2 .
our numerical routine is sensitive to the stiffness of the system of differential equations (3.17),
(3.18), (3.27), and (3.28). The stiffness of these equations increases as 2 V 2 increases. As a result,
it becomes harder to generate adequate initial guesses as pull-in is approached.
Figures 3.4 and 3.5 show the first and second mode shapes of the plate as the electrostatic force is
increased from zero and approaches the pull-in level. In both figures, the mode shape is normalized
with respect to the maximum deflection point. The natural frequencies and mode shapes of the
free-plate vibrations compare well with those reported by Amabili et al. (1996), which we used as
a bench mark to check our algorithm. We show results in Table 1 for clamped-clamped (C-C) and
clamped-free (C-F) annular paltes with a radii ratio of 0.5.
Table 3.1: Nondimensional natural frequencies of an annular plate.
Material
1stmodeC C
2ndmodeC C
3rdmodeC C
1stmodeC F
2ndmodeC F
3rdmodeC F
9.447
9.498
9.660
4.220
4.691
5.667
Present algorithm
9.451
9.510
9.680
4.250
4.700
5.670
Waleed F. Faris
Chapter 3.
41
2V2= 140.6
2V2= 0
0
0
0.2
0.4
0.6
0.8
Figure 3.4: The first normalized mode shape of the clamped-clamped plate as the electrostatic
force increases from zero and approaches pull-in level.
3.3
Clamped-Free Case
In this case, the plate is clamped on the outer edge and free on the inner edge. On the free edge
the boundary condition of the stress function corresponds to the vanishing of the in-plane radial
force Nr . Using Equation (3.7), we find that the boundary conditions at the free edge r = b are
2 w w
= 0,
+
r2
r r
3w 1 2w
1 w
+
2
= 0,
r3
r r2
r r
=0
and
r
(3.31)
while the boundary conditions at the fixed edge r = 1 are the same as those in Equations (3.15)
and (3.16).
We calculate the static deflection of the plate by solving Equations (3.17) and (3.18) for the bound-
Waleed F. Faris
Chapter 3.
42
2 0
-1
0
0.2
0.4
0.6
0.8
Figure 3.5: The second normalized mode shape of the clamped-clamped plate as the electrostatic
force increases from zero and approaches pull-in level.
ary conditions
1
1
at r = b
(3.32)
(3.33)
We solved the boundary-value problem for a range of electrostatic forces varying from zero to the
forcing level where the structural instability (pull-in) develops. Figure 3.6 shows an axisymmetric
cross-section of the plate as it deflects under an upward electrostatic force increasing from zero to
2 V 2 = 21.5. The evolution of the static deflection is most pronounced at the free edge. Figure
3.7 shows variation of the maximum nondimensional deflection WM ax = ws (r = b) with 2 V 2 .
As the electrostatic force, represented by 2 V 2 , increases, the maximum deflection increases. This
relationship is linear at low levels of electrostatic forcing; however as the electrostatic force increases,
the relationship becomes increasingly nonlinear.
To calculate the mode shapes and corresponding natural frequencies, we solve Equations (3.21) and
Waleed F. Faris
Chapter 3.
43
0.6
0.4
ws
0.2
0.2
0.4
0.6
0.8
Figure 3.6: Nondimensional deflection ws (r) of the clamped-free plate under an electrostatic force
ranging from 2 V 2 = 0 to 21.5.
(3.22) in conjunction with the boundary conditions
1
1
u = 0 , u0 = 0, and 00 0 = 0 at r = 1
r
at r = b
(3.34)
(3.35)
We assume a harmonic motion in the nth mode as in Equations (3.25) and (3.26). Substituting
Equations (3.25) and (3.26) into Equations (3.34) and (3.35) yields
n
n00 + n0 = 0 ,
= 0, and n000 +
r
r
n = 0 , n0 = 0, and 00n 0n = 0
r
1 00
1
n 2 n0 = 0
r
r
at r = 1
at r = b
(3.36)
(3.37)
We solve the eigenvalue problem in Equations (3.27), (3.28), (3.36), and (3.37) by releasing one
of the boundary conditions at r = 1.0 and solving the boundary-value problem iteratively using a
shooting method until n , n , and n converge to within predefined tolerances. This procedure is
repeated for each mode shape and associated natural frequency and stress function, using various
sets of initial guesses.
Figure 3.8 shows variation of the fundamental natural frequency of the clamped-free plate with
Waleed F. Faris
Chapter 3.
44
0.6
WMax
0.4
0.2
0
0
10
2V2
15
20
25
Figure 3.7: Variation of the maximum deflection WM ax of the clamped-free plate with 2 V 2 .
2 V 2 . The fundamental frequency decreases as the electrostatic force increases and approaches zero
as pull-in develops. The contrast between the natural frequency results for this case as opposed
to the previous case arises from the fact that the in-plane stresses here are much less pronounced
than in the previous case because of the free edge.
Figures 3.9 and 3.10 show the first and second mode shapes of the clamped-free plate as the
electrostatic force is increased from zero and approaches the pull-in level. In both figures, the mode
shape is normalized with respect to the absolute maximum of the mode shape for a straight plate.
Significant changes are observed in the first mode shape, Figure 3.9, but not in the second mode
shape, Figure 3.10. The natural frequencies and mode shapes compare well with those reported by
Amabili et al. (1996).
Waleed F. Faris
Chapter 3.
45
12
0
0
10
2V2
15
20
25
Figure 3.8: Variation of the fundamental natural frequency of the clamped-free plate with V 2 .
2V2= 21.5
1
2V2= 0
0
0
0.2
0.4
0.6
0.8
Figure 3.9: The first mode shape of the clamped-free plate as the electrostatic force increases from
zero and approaches the pull-in level.
Waleed F. Faris
Chapter 3.
46
2V2= 21.6
2V2= 0
2 0
-1
0
0.2
0.4
0.6
0.8
Figure 3.10: The second mode shape of the clamped-free plate as the electrostatic force increases
from zero and approaches the pull-in level.
Chapter 4
47
Waleed F. Faris
4.1
Chapter 4.
48
Problem Formulation
First, we consider the coupled heat conduction equation and dynamic version of the von Karm
an
equations for a thermally excited circular plate. The basic geometric features are shown in Figure
4.1. We consider an axisymmetric thermal loading and axisymmetric vibrations. They are governed
Z
r
X
ET0 e
T
+
t
1 2 t
1 2 w F
1 w 2 F
w
1
+
2c
2 MT
2
2
r r r
r r r
t
1
Eh 2 w w
4 F =
2 NT
r r2 r
D4 w +hw
=
(4.1)
(4.2)
(4.3)
where
Z
NT = E
12 h
Z
MT = E
and D =
Eh3
.
12(1 2 )
1
h
2
1
h
2
12 h
[T (r, z, t) T0 ]dz
(4.4)
[T (r, z, t) T0 ]zdz
(4.5)
Here, w(r, t) is the plate transverse displacement, F (r, t) is the stress function,
e is the dilatational strain due to the thermal effect, T is the temperature distribution, T0 is the
stress-free temperature, is the material density, h is the plate thickness, cp is the heat capacity
Waleed F. Faris
Chapter 4.
49
1
(Nr N ) + T
Eh
(4.6)
1
(N Nr ) + T
Eh
2F
1 F
and N =
Nr =
r r
r2
2
u 1 w
u
r =
+
and =
r
2 r
r
=
(4.7)
(4.8)
(4.9)
w
r
1
u
=
r
Eh
1
+
Eh
1 F
2F
2
r r
r
2F
F
2
r
r r
+ T
(4.10)
+ T
(4.11)
Eliminating u from Equations (4.10) and (4.11), we obtain the compatibility equation
3F
1
2F
1 F
r 3 +
= Eh
2
r
r
r r
2
w
r
2
Ehr
T
r
(4.12)
r = Rr , t = R
h
D
1/2
t , T = T0 T ,
h2
Eh5
w , z = hz , F = 2 F ,
R
R
12(1 2 ) 5 1/2
h4 q
c=
h
D
c
,
q
=
R4
R6
w=
(4.13)
Substituting Equations (4.13) into Equations (4.1), (4.2), and (4.12) and dropping the asterisks,
we obtain
2 T +
T
e
4qh4
cos(t) = 1
+ 2
4
T0 R
t
t
(4.14)
Waleed F. Faris
Chapter 4.
50
#
"
1 w 2 F
2w
1 2 w F
w
4
+ w =
+
2c
t2
r r2 r
r r r2
t
Z 1
3F
2F
1 F
1 w 2 12T0 (1 + )R3 2
r 3 +
z2 T dz
r
r2
r r
2 r
h3
1
(4.15)
(4.16)
2F
1 F
3F
T0 R4
r
r 3 +
=
r
r2
r r
h4
where
cp
12(1 2 )h2
, 1 =
=
2
R
k
1
2
12
T
dz
r
(4.17)
s
D
E 1
, 2 =
h
(1 2) k
D
h
The last two terms on the right-hand side of Equation (4.14) represent the diffusion of heat and
thermoelastic coupling (Hetnarski, 1987).
Usually, the materials used in plates for mechanical and MEMS applications are Steel, Aluminum
(Al), and Silicon Nitride (SiNi). Using the available mechanical and thermal properties for these
materials, we calculate the nondimensional coefficients 1 and 2 , and present them in Table 4.1.
For thin plates, it follows from Equation (4.14) and Table 4.1 that the thermal diffusion and
thermoelastic coupling terms are negligible because 1 and 2 are very large. Hence, Equation
(4.14) is reduced to
Table 4.1: Thermal and mechanical properties of used materials.
(Kg/m 3 )
E(GPa)
k(W/m.K)
cp (J/Kg.K)
Al
2702
70
237
903
24.0 106
15.86h 106
27.3h 106
Steel
7800
207
60
430
12.0 106
97.43h 106
161.34h 106
SiNi
2330
100
148
712
2.9 106
22.24h 106
9.72h 106
Material
2 T +
4qh4
cos(t) = 0
T0 R4
(4.18)
h4
1 r2 q cos(t)
4
T0 R
(4.19)
Waleed F. Faris
Chapter 4.
1
w
w
+ 4 w =
+
2c
t2
r r2 r
r r r2
t
2
3F
2F
1 F
1 w
r 3 +
=
+ 2qr2 cos(t)
r
r2
r r
2 r
51
(4.20)
(4.21)
Next, we treat two cases of practical importance in thermally loaded circular plates: simply supported and clamped plates, starting with the first case.
4.2
We consider a simply supported plate with no restrictions on its in-plane displacement; that is,
movable boundary condition. This case is important in the structural mechanics of planar bodies,
such as plates (Vinson, 1989). In this case, the thermal stresses in the plate are self-equilibrating
and the radial force Nr vanishes at the outer edge. Therefore, the boundary conditions are
F < and w < at r = 0
F
= 0 at r = 1
r
2 w w
= 0 at r = 1
w = 0 and
+
r2
r r
(4.22)
(4.23)
(4.24)
4.2.1
Perturbation Solution
We seek a first-order uniform expansion of the solution of Equations (4.20)-(4.24) in the form
(Nayfeh, 1981)
w(r, t; ) = w0 (r, t0 , t1 ) + w1 (r, t0 , t1 ) +
(4.25)
F (r, t; ) = F0 (r, t0 , t1 ) +
(4.26)
(4.27)
Waleed F. Faris
Chapter 4.
52
where Dn = /tn . Substituting Equations (4.25)-(4.27) into Equations (4.20)-(4.24) and equating
coefficients of like powers of , we obtain
Order 0
D02 w0 + 4 w0 = 0
3 F0
r
r3
2 F0
r2
1 F0
1 w0
=
r r
2 r
w0 < and F0 < at r = 0
+
(4.28)
+ 2qr2 cos(t0 )
(4.29)
(4.30)
F00 = 0 at r = 1
(4.31)
(4.32)
Order
D02 w1 + 4 w1 = 2D0 D1 w0 +
1 2 w0 F0 1 w0 2 F0
+
2cD0 w0
r r2 r
r r r2
(4.33)
w1 < at r = 0
(4.34)
(4.35)
(4.36)
n=1
n
I1
n
J1
2 n
+ =
1
J0
n
I0
n
and the kn are chosen so that
Z
0
r2n (r)dr = 1
(4.37)
(4.38)
(4.39)
The lowest five nondimensional natural frequencies for the case = 0.3 are 4.935, 29.720, 74.156,
138.318, and 222.215. Hence, the first mode is not involved in any internal resonance with any other
Waleed F. Faris
Chapter 4.
53
axisymmetric mode. Moreover, we assume that this mode is not involved in an internal resonance
with any asymmetric mode. In this analysis, we consider the case of principal parametric resonance
of the first mode. The analysis is also valid for any mode that is not involved in an internal
resonance with other modes. For asymmetric thermal loads, one-to-one internal resonances might
be activated. This case is not considered here.
Substituting Equation (4.36) into Equation (4.29) yields
r
h
i
3 F0 2 F0 1 F0
1X 0 0
i(n +m )t0
i(n m )t0
A
A
e
+
A
A
e
m
n
n
m
r3
r2
r r
2 n,m n m
(4.40)
The solution of Equation (4.40) subject to the boundary conditions (4.30) and (4.31) can be expressed as
F0 =
"
#
An Am e
i(n +m )t0
+ An Am e
i(n m )t0
nm (r)
m,n
+ cc +
1 4
r 2r2 q cos t0
16
(4.41)
where
2
d
d
1 dnm
1
r 2+
= 0n 0m
dr
dr r
dr
2
nm (0) <
0
nm
= 0 at r = 1
Because
2
d
2 d
r
+ r 1 J1 (k r) = k2 r2 J1 (k r)
dr2
dr
0
as
we express nm
0
nm
(r)
bnmk J1 (k r)
(4.42)
(4.43)
(4.44)
(4.45)
(4.46)
k=1
Using the boundary condition (4.44), we find that the m are the roots of
J1 (k ) = 0
(4.47)
The lowest ten roots of Equation (4.47) are 0, 3.832, 7.016, 13.324, 19.616, 22.760, 25.904, 29.047,
32.190, and 35.332. Substituting Equation (4.46) into Equation (4.42) and using Equation (4.45),
Waleed F. Faris
Chapter 4.
54
we obtain
bnmk
rJ100 (k r)
J10 (k r)
k=1
1
1
J1 (k r) =
bnmk k2 rJ1 (k r) = 0n 0m
r
2
(4.48)
k=1
Multiplying Equation (4.48) with J1 (s r) and integrating the result with respect to r from r = 0
to 1, we have
bnms
R1
J1 (s r) 0n (r)0m (r)dr
i
= 0h
R1
2s2 0 rJ12 (s r) dr
(4.49)
#"
#
21
3r
(r2 1)00n +
0n An ein t0 + An ein t0
r
n
"
X
0 0 0
1
+
As An Am ei(n +m +s )t0 + As An Am ei(s +n m )t0
r n,m,s s nm
#
i(
+
)t
i(
)t
+ An Am As e n m s 0 + An Am As e n m s 0 + cc
1
+ qeit0
8
"
(4.50)
Next, we consider the case of principal parametric resonance of the jth mode and assume that
it is not involved in an internal resonance with any other mode. To describe the nearness of the
resonance, we introduce a detuning parameter defined by
= 2j +
(4.51)
00 3r2 1 0
1 X 2
in t0
+ q
D w1 + w1 = 2i
n An + cAn n e
r 1 n +
n
8 n
r
n
"
1
0
0 0
An ei(n +2j )t0 +it1 + An ei(2j n )t0 +it1 +
As An Am ei(n +m +s )t0
r n,m,s s nm
#
i(
+
)t
i(
+
)t
i(
)t
s
n
m
n
m
s
n
m
s
0
0
0
+ As An Am e
+ An Am As e
+ An Am As e
+ cc (4.52)
2
In the presence of damping, all modes that are not directly excited or indirectly excited by an
internal resonance decay with time, and hence the long-time response consists of the jth mode
Waleed F. Faris
Chapter 4.
55
only. Because the homogeneous problem consisting of Equations (4.52), (4.34), and (4.35) has
nontrivial solutions, the nonhomogeneous problem has a solution only if a solvability condition is
satisfied (Nayfeh, 2000).
Because the homogeneous problem is self-adjoint, the solvability condition demands that the righthand of Equation (4.52) be orthogonal to every solution of the homogeneous problem (Nayfeh,
1981). Multiplying the right-hand side of Equation (4.52) with rj (r)eij t0 , integrating the result
from r = 0 to r = 1, and setting the outcome equal to zero, we obtain
(4.53)
where
Z
R1
0
0 0
j 0j jj
dr =
0
02
j jj dr
0
0
Z 1
Z 1
3
0 0
3
1
1
qe = q
r r j j dr = q
r r 02
j dr
8 0
8 0
e =
where j =
(4.54)
(4.55)
rc2j dr.
(4.56)
(4.57)
(4.58)
The equilibrium solutions of Equations (4.57) and (4.58) correspond to periodic motions of the
plate. The equilibrium solutions are given by
qe
aj sin 2j
2j
1
e 3
qe
aj
aj =
aj cos 2j
2
8j
2j
j aj =
(4.59)
(4.60)
There are two possibilities: either aj = 0 or aj 6= 0. In the latter case, eliminating j from
Equations (4.59) and (4.60) yields the frequency-response equation
e 2 2
q2
1
2
aj
= e2
j +
2
8j
4j
(4.61)
Waleed F. Faris
Chapter 4.
or
a2j
8j
=
e
"
qe2
2j
4j2
56
#
(4.62)
Equation (4.62) represents the relation between the response amplitude aj , the forcing amplitude
qe , and detuning (control) parameter . Next, we present results for the first mode.
4.2.2
Numerical Results
For the first mode, the first nine values of bns from Equation (4.49) are: 1.24252, -0.144995,
-0.0272522, -0.0102221, 0.00702803, -0.00507591, 0.00380701, -0.00294182, and 0.00232898. Consequently, we calculate the effective nonlinearity coefficient to be e = 3.77941. Also, we assume
that = 0.01 and consider nondimensional values of q ranging from 1 to 30. In Figure 4.2, we
show typical frequency-response curves for q = 10. The curves are bent to the right, indicating
a hardening-type nonlinearity. The bending of the frequency-response curves leads to multivalued
amplitudes and hence to jumps. To illustrate the jump phenomenon, let us suppose that an experiment is conducted where the excitation amplitude is kept constant at q = 10 while the excitation
frequency is varied very slowly. We start with a frequency far above the natural frequency (i.e.,
is positive) and decrease it. We choose the initial conditions so that the response amplitude is
trivial. As is decreased, the response amplitude remains trivial untill point D is reached, where
the response amplitude experiences a sudden jump up to point C. Point D is a subcritical or reverse
pitchfork bifurcation. The response amplitude decreases as the frequency is further decreased untill
point B is reached, where the response amplitude becomes trivial again. Point B is a supercritical
pitchfork bifurcation.
Starting at point C and sweeping in the reverse direction, we note that, to the first-order approximation, the response amplitude increases without bounds as the frequency increases. In reality,
the stable and unstable branches merge to produce a saddle-node bifurcation. Increasing above
this bifurcation point results in a jump down of the response amplitude to the trivial solution.
Between point D and the saddle-node bifurcation, there are three possible solutions: the trivial
solution, which is stable, and two nontrivial solutions, the larger of which is stable and the smaller
is unstable. In this interval, the response may be trivial or nontrivial, depending on the initial
conditions.
Waleed F. Faris
Chapter 4.
57
q = 10
C
2.5
Response Amplitude a
1.5
0.5
Supercritical
bifurcation
Subcritical
bifurcation
0
1
0.8
0.6
0.4
0.2
0
0.2
Detuning Parameter
0.4
0.6
0.8
Figure 4.2: Variation of the steady-state response amplitude with the detuning parameter when
q = 10.
In Figure 4.3, we show a force-response curve generated when the principal parametric resonance
is negatively detuned by = -0.8. Starting from a forcing amplitude corresponding to point E and
increasing it results in a monotonic increase in the response amplitude. Point E is a supercritical
pitchfork bifurcation. For any excitation amplitude above point E, the response amplitude settles on
the branch EB, irrespective of the initial conditions. Sweeping q down, one finds that the response
amplitude follows the branch BE. Decreasing q below point E results in a smooth transition of the
response from a nontrivial to a trivial response. Increasing the value of the detuning parameter
results in a shift in the critical point E to the left.
We show in Figure 4.4 the force-response curve for the case of a positive detuning of =0.5. In
contrast with the preceding case, there are multivalued responses. Point F is a subcritical pitchfork
bifurcation and point G is a saddle-node bifurcation. Between these bifurcations, the response may
be trivial or nontrivial depending on the initial conditions. Whereas the linear theory predicts
stable trivial responses for forcing amplitudes below point F and unbounded responses above it,
the nonlinear analysis predicts that the trivial response between points F and G is stable for
small disturbances but unstable for large disturbances, a subcritical instability. Moreover, the
nonlinearity puts a cap on the response amplitude. Sweeping q in the forward direction, one finds
Waleed F. Faris
Chapter 4.
58
2.5
Detuning Parameter =0.8
Response Amplitude a
B
1.5
0.5
Supercritical pitchfork
bifurcation
E
0
0
10
15
Forcing Amplitude q
20
25
30
Figure 4.3: Variation of the steady-state response amplitude with the forcing amplitude for =
0.8.
that beyond point F the response amplitude jumps up. On the other hand, sweeping q in the
reverse direction beyond point G results in a jump to the trivial response.
4.3
We consider a clamped plate with an immovable edge. This case is important in most of MEMS
devices, such as sensors and micropumps, as it is a more realistic representation of the actual
boundary condition. The governing nondimensional Equations (4.20) and (4.21) are repeated here
#
"
2w
1 2 w F
1 w 2 F
w
4
+ w =
+
2c
(4.63)
t2
r r2 r
r r r2
t
3F
2F
1 F
1
r 3 +
=
2
r
r
r r
2
w
r
+ 2qr2 cos(t)
+
(Tb 1) = 0 at r = 1
r2
r r
h4
(4.64)
(4.65)
(4.66)
(4.67)
Waleed F. Faris
Chapter 4.
59
3.5
Response Amplitude a
2.5
2
G
Saddle point bifurcation
1.5
0.5
Subcritical pitchfork bifurcation
F
0
0
10
15
Forcing Amplitude q
20
25
30
Figure 4.4: Variation of the steady-state response amplitude with the forcing amplitude for = 0.5.
1 4
r q cos(t)
16
(4.68)
T0 R4
3 2
r q cos t
2(1 )h4 8(1 )
T0 R4 2
1
2(3 ) 2
4
=
r + C2 (t) +
r
r q cos(t)
2(1 )h4
16
1
(4.69)
where is the nonlinear part of the stress function. Substituting Equation (4.69) into Equations
Waleed F. Faris
Chapter 4.
60
(4.63)-(4.67) we obtain
"
2w
1 2 w 1 w 2
2
4
+p
w
+
w
=
+
t2
r r2 r
r r r2
w 1 2 w 2 3
1 w
3
2c
+
r
q
cos(t)
+
3r
q cos(t)
t
4 r2
1
4 r
(1 )r
3 2 1
1 w 2
r 3 +
=
r
r2
r r
2 r
w
w = 0 and
= 0 at r = 1
r
w < and < at r = 0
2
= 0 at r = 1
r2
r r
(4.70)
(4.71)
(4.72)
(4.73)
(4.74)
where
p=
4.3.1
12T0 (1 + )R2
h2
Perturbation Solution
We seek a first-order uniform expansion of the solution of Equations (4.70)-(4.74) in the form
(Nayfeh, 1981)
w(r, t; ) = w0 (r, t0 , t1 ) + w1 (r, t0 , t1 ) +
(4.75)
(r, t; ) = 0 (r, t0 , t1 ) +
(4.76)
(4.77)
where Dn = /tn . Substituting Equations (4.75)-(4.77) into Equations (4.70)-(4.74) and equating
coefficients of like powers of , we obtain
Waleed F. Faris
Chapter 4.
61
Order 0
D02 w0 + p2 w0 + 4 w0 = 0
3 0 2 0 1 0
1
+
=
r3
r2
r r
2
w0 = 0 and w00 = 0 at r = 1
w
r
000 00 = 0 at r = 1
r
(4.78)
(4.79)
(4.80)
(4.81)
(4.82)
Order
1 2 w0 0 1 w0 2 0
D02 w1 +p2 w1 + 4 w1 = 2D0 D1 w0 +
+
2
r r2 r
r r
2
1 w0 2 3
1 w0
3
2cD0 w0 +
q cos(t0 ) +
3r
q cos(t0 ) (4.83)
r
4 r2
1
4 r
(1 )r
w1 = 0 and w10 = 0 at r = 1
(4.84)
w1 < at r = 0
(4.85)
(4.86)
n=1
where
n (r) = [B1 J0 (1n r) + B2 Y0 (1n r) + B3 I0 (2n r) + B4 K0 (2n r)]
i
i
1 hp 2
1 hp 2
21n =
p + 4n + p , and 22n =
p + 4n p
2
2
(4.87)
(4.88)
and B1 , B2 , B3 , and B4 are constants. Applying the boundary condition in Equation (4.81) yields
B2 = 0 and B4 = 0. Then, applying the boundary conditions in Equation (4.80) yields the following
frequency equation:
2n I1 (2n ) J0 (1n ) + 1n J1 (1n ) I0 (2n ) = 0
(4.89)
Waleed F. Faris
Chapter 4.
62
12
10
10
15
Nondimensional Temperature
Figure 4.5: Variation of the first nondimensional frequency with the nondimensional parameter
(temperature) p.
40
39
38
37
36
35
34
33
10
15
Nondimensional Temperature
Figure 4.6: Variation of the second nondimensional frequency with the nondimensional parameter
(temperature) p.
which can be solved numerically for the n as a function of p and hence T0 . In Figures 4.5-7, we
show variation of the first three nondimensional natural frequencies with p. Buckling occurs when
the first natural frequency is zero, which corresponds to p = 14.68.
Waleed F. Faris
Chapter 4.
63
90
89
88
87
86
85
84
83
82
10
15
Nondimensional Temperature
Figure 4.7: Variation of the third nondimensional frequency with the nondimensional parameter
(temperature) p.
Substituting Equation (4.86) into Equation (4.79) yields
r
i
1X 0 0 h
3 0 2 0 1 0
i(n +m )t0
i(n m )t0
A
A
e
+
A
A
e
+
+ cc
m n
n m
r3
r2
r r
2 n,m n m
(4.90)
The solution of Equation (4.90) subject to the boundary conditions (4.81) and (4.82) can be expressed as
0 =
i
Xh
An Am ei(n +m )t0 + An Am ei(n m )t0 nm (r) + cc
(4.91)
m,n
where
Because
1 dnm
1
d
d
= 0n 0m
r 2+
dr
dr r
dr
2
(4.92)
nm (0) <
(4.93)
00
0
nm
nm
= 0 at r = 1
(4.94)
d
d2
r2 2 + r 1 J1 (k r) = k2 r2 J1 (k r)
dr
dr
(4.95)
0
as
we express nm
0
nm
(r)
X
k=1
bnmk J1 (k r)
(4.96)
Waleed F. Faris
Chapter 4.
64
Using the boundary condition (4.94), we find that the m are the roots of
k J1 (k ) (1 + )J0 (k ) = 0
(4.97)
The lowest ten roots of Equation (4.97) are 1.57883, 5.27284, 8.5006, 11.6802, 14.8433, 17.9988,
21.1502, 24.299, 27.4461, and 30.5921. Substituting Equation (4.96) into Equation (4.92) and using
Equation (4.95), we obtain
"
bnmk rJ100 (k r)
k=1
k=1
J10 (k r)
1
J1 (k r)
r
1
bnmk k2 rJ1 (k r) = 0n 0m
2
(4.98)
Multiplying Equation (4.98) with J1 (s r) and integrating the result with respect to r from r = 0
to 1, we have
R1
bnms =
J1 (s r) 0n (r)0m (r)dr
R1
2s2 0 rJ12 (s r)dr
(4.99)
"
3
r
r 0n
1
"
3
"
#
in t0
An e
+ An e
in t0
1 X 0 0 0
)t
i(
)t
n
m
c
n
m
c
0
0
+An Am Ac e
+ An Am Ac e
+ cc
(4.100)
Next, we consider the cases of combination parametric resonance of the additive and difference
types and principal parametric resonance.
4.3.2
We consider the case of combination parametric resonance of the additive type involving the jth
and kth modes and assume that neither is involved in an internal resonance with any other mode.
To describe the nearness of the resonance, we introduce the detuning parameter defined by
= j + k +
(4.101)
Waleed F. Faris
Chapter 4.
65
Because the homogeneous problem consisting of Equations (4.100), (4.84), and (4.85) has nontrivial
solutions, the nonhomogeneous problem has a solution only if solvability conditions are satisfied
(Nayfeh, 2000). Because the homogeneous problem is self-adjoint, the solvability conditions demand
that the right-hand side of Equation (4.100) be orthogonal to every solution of the homogeneous
problem. Multiplying the right-hand side of Equation (4.100) with rs (r)eis t0 , integrating the
result from r = 0 to r = 1, using Equation (4.101), and setting the result equal to zero, we obtain
X
(4.102)
m6=j
(4.103)
m6=k
X
sm As Am Am = 0 for s 6= j and k
2is A0s + s As + ss A2s As +
(4.104)
m6=s
where
Z
Z 1
0 0 0
0
nn = 3
n n nn dr = 3
nn
02
n dr
0
0
Z 1
Z 1
0
0
0
0 0
0m dr
nm = 2
n mm n 4
n nm
0
0
Z 1
Z 1
0
0
=2
mm
02
nm
0n 0m dr
n dr + 4
0
0
0
Z
1 1
3
3
fnm =
n r
r 0m dr
8 0
1
Z
1 1 3 3
=
r
r 0m 0n dr
8 0
1
Z 1
n =
cr2n dr
1
(4.105)
(4.106)
(4.107)
(4.108)
(4.109)
(4.110)
(4.111)
Waleed F. Faris
Chapter 4.
66
Substituting Equations (4.109)-(4.111) into Equation (4.104) and separating the real and imaginary
parts, we obtain
a0s = s as
1
1 X
s as s0 = ss a3s + as
sm a2m
8
8
(4.112)
(4.113)
m6=s
fkj
qaj sin
2k
fkj
1
kk 3 kj
ak k0 = ak +
ak +
ak a2j +
qaj cos
2
8k
8k
2k
a0k = k ak +
(4.114)
(4.115)
(4.116)
(4.117)
where
= j + k
(4.118)
The equilibrium solutions or fixed points of Equations (4.114)-(4.118) correspond to a0j = 0, a0k = 0,
and 0 = 0. There are two possible equilibrium solutions: (a) aj = 0 and ak = 0 and (b) aj 6= 0 and
ak 6= 0. In the latter case, Equations (4.115), (4.117), and (4.118) can be combined to eliminate j
and k and obtain the following equation for 0 :
kj
jk
jj
kk
0
2
= +
+
aj +
+
a2k
8j
8k
8j
8k
aj
ak
+
+
qfjk cos
2aj j
2ak k
(4.119)
Hence, for nontrivial solutions, the modulation equations are reduced from four to three first-order
differential equations. For equilibrium solutions, we set a0j = 0, a0k = 0, and 0 = 0 and solve
Waleed F. Faris
Chapter 4.
67
Equations (4.114), (4.116), and (4.119) for aj , ak , and . The following closed-form solution is
obtained:
s
2 q2
fjk
j
k
e a2j =
j k
j k 4j k
j j 2
a2k =
a
k k j
j k j k
2j j aj
2k k ak
=
= 2
sin =
fjk ak q
fkj aj q
fjk q
(4.120)
(4.121)
(4.122)
where
kj
jk
1 jj
kk j j
e =
+
+
+
8 j
k
j
k k k
(4.123)
In Figure 4.8, we show the frequency-response curves obtained for q = 0.01 and the nondimensional
temperature p = 7.0. The curves are bent to the right, indicating a hardening-type nonlinearity.
The bending of the frequency-response curves leads to multivalued amplitudes and hence to jumps.
To illustrate the jump phenomenon, let us suppose that an experiment is conducted where the
excitation amplitude is kept constant while the excitation frequency is varied very slowly. We start
from a frequency far above the natural frequency (i.e., a positively detuned thermal load) and
decrease it. We choose the initial conditions so that the response amplitude is trivial. As is
decreased, the response amplitudes remain trivial untill point B is reached, where they experience
a sudden jump up to points C and D. Point B is a subcritical or reverse pitchfork bifurcation. The
response amplitudes decrease as the frequency is further decreased untill point A is reached, where
they become trivial again. Point A is a supercritical pitchfork bifurcation.
Starting at points C or D and sweeping in the reverse direction, we note that, to the first-order
approximation, the response amplitudes increase without bounds as the frequency increases. In
reality, the stable and unstable branches merge to produce a saddle-node bifurcation. Increasing
the frequency above this point results in a jump down of the response amplitudes to the trivial
solution. Between point B and the saddle-node bifurcation, there are three possible solutions: the
trivial solution, which is stable, and two nontrivial solutions, the larger of which is stable and the
smaller is unstable. In this interval, the response may be trivial or nontrivial, depending on the
initial conditions.
Waleed F. Faris
Chapter 4.
x 10
68
4
3.5
C
a1
a1 & a2
2.5
2
1.5
a2
D
1
0.5
A
0
0.5
1
1
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
Figure 4.8: Variation of the equilibrium solutions with the detuning parameter when p = 7.0 and
q = 0.01.
In Figure 4.9, we show the force-response curves generated when the system is negatively detuned
by 0.8. Starting from a forcing amplitude corresponding to point A and increasing it results in a
monotonic increase in the response amplitudes. Point A is a supercritical pitchfork bifurcation. For
any excitation amplitude above point A, the response amplitudes settle on the branches AB and
AC, irrespective of the initial conditions. Sweeping q down, one finds that the response amplitudes
follow the branches BA and CA. Decreasing q below point A results in a smooth transition of the
response from a nontrivial to a trivial response. Increasing the value of the detuning parameter
results in a shift of the critical point A to the right.
We show in Figure 4.10 the force-response curves generated for the case of a positive detuning of
0.85. In contrast with the preceding case, there are multivalued responses. Point A is a subcritical
pitchfork bifurcation and points D and E are saddle-node bifurcation. Between these bifurcations,
the response may be trivial or nontrivial depending on the initial conditions. Whereas the linear
theory predicts stable trivial responses for values of q below point A and unbounded responses
above it, the nonlinear analysis predicts an instability of the trivial response between points A and
D for large disturbances, a subcritical instability. Moreover, the nonlinearity puts a cap on the
response amplitudes. Sweeping q in the forward direction, one finds that the response amplitudes
Waleed F. Faris
Chapter 4.
x 10
69
-3
= 0.8
a1
4
a1 & a2
a2
2
C
0.05
0.1
0.15
0.2
0.25
Figure 4.9: Variation of the equilibrium solutions with the forcing amplitude (force-response curves)
when = 0.8 and p = 7.0.
jump up beyond point A. On the other hand, sweeping q in the reverse direction results in a jump
to the trivial response beyond point D.
4.3.3
(4.124)
can be obtained from the preceding case by changing the sign of j from positive to negative. Then,
it follows from Equation (4.121) that nontrivial solutions do not exist.
4.3.4
For the case of principal parametric resonance of the kth mode, we let
= 2k +
(4.125)
Waleed F. Faris
Chapter 4.
70
-3
x 10
6
= 0.85
a1
5
B
a1 & a2
a2
2
C
1
E
A
-1
0.05
0.1
0.15
0.2
0.25
Figure 4.10: Variation of the equilibrium solutions with the forcing amplitude (force-response
curves) when = 0.85 and p = 7.0.
Then, all modes except the kth mode decay exponentially with time. Then, using Equation (4.125)
along with Equations (4.100), (4.84), and (4.85), we obtain
(4.126)
where kk , fkk , and k are defined in Equations (4.105), (4.107), and (4.108).
Next, we substitute Equation (4.110) into Equation (4.126), separate the real and imaginary parts,
and obtain
fkk
qak sin(2k )
2k
kk 3
fkk
1
a +
qak cos(2k )
ak k0 = ak +
2
8k k 2k
a0k = k ak +
(4.127)
(4.128)
It follows from Equations (4.127) and (4.128) that their equilibrium solutions are given by
s
"
#
2 q2
f
8
1
k
kk
a2k =
2k
(4.129)
kk 2
4k2
The frequency- and force-response curves generated using Equation (4.129) are qualitatively similar
to those shown in Figures 4.2-4.4.
Chapter 5
Introduction
The vibration characteristics of thermally stressed plates have become important with the increasing
flight speeds of air and space vehicles and the resulting thermal environment of the airframe. The
buckling of skin panels, whether caused by air loads or by thermal stresses, causes a marked
reduction in their stiffnesses. the changes in frequencies and mode shapes, which take place due to
the thermal stresses, affect considerably their static and dynamic aeroelastic instabilities. These
changes are the subject of this study.
The literature on the postbuckling behavior of plates, whether static or dynamic, is not as extensive
as the literature on the pre-buckling behavior of plates.
71
Waleed F. Faris
5.2
Chapter 5.
72
Governing Equations
We consider large-amplitue deflections and vibrations of a postbuckled circular plate whose edge
temperature is maintained at a constant value Tb . Neglecting thermal diffusion and thermoelastic
damping, one finds that the plate temperature T (r, t) = Tb . We consider both clamped and simply
supported boundary conditions. Axisymmetric deflections and vibrations are governed by
4w
D
+ h
1 2w
2w
F 1 w
2 F
=
+
r r2
r
r r r2
t2
(5.1)
where w(
r, t) is the plate transverse displacement, F (
r, t) is the stress function, is the mass
density, h is the plate thickness, D = Eh3 /[12(1 2 )] is the plate rigidity, E is the modulus of
elasticity, and is Poissons ratio. The compatibility equation for axisymmetric vibrations is given
by
3 F 2 F 1 F
1
r 3 +
= Eh
2
r
r
r r
2
2
(5.2)
= 0 at r = R
r
F < and w
< at r = 0
2
F
F
+ Eh(Tb T0 ) = 0 at r = R
2
r
r r
w
= 0 and
(5.3)
(5.4)
(5.5)
And for a simply supported plate with immovable edges, the boundary conditions are
2w
w
= 0 at r = R
+
r2
r
r
F < and w
< at r = 0
2
F F
+ Eh(Tb T0 ) = 0 at r = R
r2
r r
w
= 0 and
(5.6)
(5.7)
(5.8)
where T0 is the zero-stress temperature, is the coefficient of thermal expansion, and R is the
radius of the plate.
We introduce nondimensional variables, defined as follows:
r = Rr, t = R
h
D
1/2
t, w
= hw, F = Eh3 F
(5.9)
Waleed F. Faris
Chapter 5.
1
w
+ 4 w = 1
+
t2
r r2 r
r r r2
3F
2F
1 F
1
r 3 +
=
2
r
r
r r
2
w
r
73
(5.10)
w
= 0 at r = 1
r
F < and w < at r = 0
2F
F
(Tb T0 )R2
+
= 0 at r = 1
r2
r r
h2
w = 0 and
(5.11)
(5.12)
(5.13)
(5.14)
where
1 = 12(1 2 ).
We transform Eqs. (5.10-14) first into a problem with homogeneous boundary conditions. To this
end, we solve the linear part of Eq. (5.11) subject to the boundary conditions (5.13) and (5.14). The
general solution of the linear part of Eq. (5.11), which is bounded at the origin, can be expressed
as
F = C1 (t)r2 + C2 (t)
(5.15)
(Tb T0 )R2
2(1 )h2
(Tb T0 )R2 2
r + C2 (t)
2(1 )h2
(5.16)
"
#
2w
1 2 w 1 w 2
2
4
+ p w + w = 1
+
t2
r r2 r
r r r2
3 2 1
1 w 2
r 3 +
=
r
r2
r r
2 r
w
w = 0 and
= 0 at r = 1
r
w < and < at r = 0
2
= 0 at r = 1
r2
r r
(5.17)
(5.18)
(5.19)
(5.20)
(5.21)
Waleed F. Faris
Chapter 5.
74
where
p=
The nondimensional parameter p depends on the temperature and thermal properties of the material.
REDUCED-ORDER MODEL
A general analytical solution of the nondimensional coupled Eqs. (5.17)-(5.21) is not available yet,
hence we seek an approximate analytical solution in the form
w(r, t) =
N
X
m (t)m (r)
(5.22)
m=1
where m (r) is the mth shape function, m (t) is the mth generalized coordinate for the mth
shape function, and N is the number of retained shape functions. As N approaches infinity, the
approximation in Eq. (5.22) becomes exact if the chosen shape functions form a complete set. Once
we solve for all the m (r) and m (t), we calculate an approximation to the deflection w(r, t) from
Eq. (5.22).
We choose the shape functions m (r) to be the first N axisymmetric modes of the linear undamped
plate with the in-plane load due to temperature. For this case, the mode shapes are given by
m (r) =
J0 (r1m ) I0 (r2m )
J0 (1m )
I0 (2m )
(5.23)
where J0 is the Bessel function of the first kind, I0 is the modified Bessel function of the first kind,
1m
i
i
1 hp 2
1 hp 2
=
p + 4m + p , 2m =
p + 4m p
2
2
(5.24)
and m is the known mth nondimensional natural frequency associated with the mth linear undamped axisymmetric mode m (r). We note that the mode shapes are orthonormal; that is,
Z
0
rm (r)n (r)dr = mn
(5.25)
Waleed F. Faris
Chapter 5.
75
Substituting Eq. (5.22) into Eq. (5.18) and integrating, we have (Vogl and Nayfeh, 2003)
(r, t) =
N
X
(5.26)
m,n=1
Rr 0 0
Rr 0 0
1
4r 0 mn d + 4r
0 m n d
R1 0 0
R
1+ 1
0
0
+ 4r 0 mn d + 4r 1
0 m n d
(5.27)
Here, the prime denotes differentiation with respect to the space variable r and the overdot denotes
0
differentiation with respect to the time variable t. We note that mn
is symmetric in m and n for
m, n = 1, 2, . . . , N .
We then substitute Eqs. (5.22), (5.23), and (5.26) into Eq. (5.17) and obtain
N
N
X
X
1 0
00
2
m m + m =
m + m m m + p
r
m=1
m=1
#
"
N
1 X
00 0
0
00
1
m n p m np + m np
r
(5.28)
m,n,p=1
We multiply Eq. (5.28) with rq (r), integrate the outcome over r [0, 1], use the orthonormality
condition (5.25), and obtain
N
X
2
m
q + q q p
m=1
Z
0
r0m 0q dr
= 1
N
X
m,n,p=1
Z
m n p
0
0q 0m np
dr
(5.29)
where q = 1, 2, . . . , N . Consequently, the discretizations of w(r, t) and (r, t) in Eqs. (5.22) and
(5.26), respectively, have reduced the general nonlinear distributed-parameter system, Eqs. (5.17)(5.21), into the system of N nonlinearly coupled ordinary-differential Equations (5.29).
0 are calculated for n, p = 1, 2, . . . , N , one evaluates the integrals in Eq. (5.29).
Once the n and np
Then, the system of equations (5.29) is solved for the m (t). Then, once all of the m (t) are
determined, the plate deflection w(r, t) can be calculated using Eq. (5.22).
STATIC BEHAVIOR
Setting all of the time derivatives equal to zero, we reduce the system of coupled differential
equations (5.29) into a system of coupled algebraic equations. Choosing values for the parameters
Waleed F. Faris
Chapter 5.
76
and N , we solve this algebraic system for the m and then the static deflection using Eq. (5.22).
In what follows, we choose = 0.3 (a very common value for metals) and N = 15.
Figure 5.1
1.4
1.2
w*
0.8
0.6
0.4
0.2
0
0
10
12
14
16
shows variation of the maximum postbuckling deflection with the load level =
p
pcr ,
where pcr is
the first buckling load. Convergence was obtained using five modes, including more modes does not
affect the converged results. Figure 5.2 shows variation of the normalized radial stresses with the
number of modes N retained in the approximation. The significance of the shown results lies in
the fact that the value of N needed for the radial stress to converge is higher than the value of N
needed for the deflection to converge. Also, the number of modes needed for convergence depends
on the load level, the higher the load level is the higher N is.
Bodner (1954) solved for the radial stress of a clamped plate using a power-series solution. He
could not carry out the solution for higher load levels because of convergence problems. Figure
5.3 shows a comparison between our solution and Bodners. It is clear that our solution can reach
higher load levels easily with less number of modes compared with the number of terms one needs
for the power-series solution. Also, it indicates that the power-series solution starts to deviate from
our results for > 3.
For the sake of comparison, we solved the same simply supported plate problem of Friedrichs and
Stoker (1942) and reproduce their power-series solution results. They used a maximum of 23 terms
Waleed F. Faris
Chapter 5.
77
0.2
0.4
0.6
0.8
1.2
1.4
0
10
15
Number of Modes
Figure 5.2: Normalized edge load vs. number of modes needed for convergence at = 15
in their solution and in our reproduction we used 50. The number of modes N retained in our
solution is 15. The results compare fairly well for much less computational effort. Figure 5.4 shows
this bench mark comparison.
NATURAL FREQUENCIES
In this section, we determine the natural frequencies of the axisymmetric modes of the deflected
plate. The procedure followed here is similar to that of Vogl and Nayfeh (2003). First, we perturb
eq
each coefficient function m (t) with a harmonic term from its equilibrium value m
; that is,
eq
m (t) = m
+ m eit ,
(5.30)
Then, we substitute Eq. (5.30) into the reduced-order model, Eq.(5.29), and keep only the linear
terms in m . Because the constant terms, which are due to equilibrium, cancel out, the reducedorder model becomes a system of N linear homogeneous algebraic equations in the unknowns m
and ; that is,
[Z()]{m } = {0},
(5.31)
Waleed F. Faris
Chapter 5.
78
0.8
0.6
0.4
0.2
0.2
0.4
0
10
12
14
16
18
(5.32)
Equation (5.32) is an algebraic equation of order 2N with N roots associated with unique modes
of vibration. The positive roots correspond to the natural frequencies of the deflected plate and
the negative roots correspond to unstable equilibrium states.
Figure 5.5 shows variation of the first nondimensional axisymmetric natural frequency with the
load level and Figure 5.6 shows variation of the square value of the first natural frequency with
the load level . The reason for showing both results is that some previous publications mixed
both results and showed the squared value in place of the value of the frequency, both behviors are
clearly different. Figure 5.7 shows variation of the first four nondimensional axisymmetric natural
frequencies with the load level .
Waleed F. Faris
Chapter 5.
79
1
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
0
10
12
14
16
18
Figure 5.3: Comparison of the power-series solution with our solution for a simply supported plate;
solid curve: present solution, stars: power-series solution.
50
45
40
35
30
25
20
15
10
5
0
0
10
12
14
16
Figure 5.4: Variation of the first natural frequency with the load level .
Waleed F. Faris
Chapter 5.
2500
2000
1500
1000
500
0
0
10
12
14
16
Figure 5.5: Variation of the first natural frequency sqaured with the load level .
180
160
nondimensional frequencies
140
120
100
80
60
40
1
20
10
11
Figure 5.6: Variation of the first four natural frequencies with the load level .
80
Chapter 6
Introduction
Sensors are one of the most commonly used devices in industry. They are composed of several
components and the heart of them is the sensing element, which may be either a beam, or a bridge,
or a diaphragm. Sensors are increasingly used as a major safeguard against thermal disasters,
especially after the disaster of the Space Shuttle Columbia (Frauenfelder, 2003). High-performance
sensors capable of operating at temperatures as high as 300 C and in corrosive environments are in
demand in many industries, such as automotive engine control, subterranean heat exploration, and
industrial pressure instruments (Ishida, 2001). Also, pressure and vibration-measurement sensors
in gas-turbine engines of manned aircraft operate in high-temperature environments (up to 600 C);
they are based on measurements of the lower natural frequencies of diphragms. Therefore, there
is a wide interest in analyzing, controlling, and compensating for the influence of temperature on
sensor performance (Smith and Senturia, 1995). The focus of this paper is on the modeling and
analysis of such an influence on a circular diaphragm sensor.
81
Waleed F. Faris
6.2
Chapter 6.
82
Problem Formulation
2w
F 1 w
2 F
V 2
1 2w
+
=
+
r
r2 r
r r r2
2(d w)
2
t2
(6.1)
where w(
r, t) is the plate transverse displacement, F (
r, t) is the stress function, is the mass
density, h is the plate thickness, D = Eh3 /[12(1 2 )] is the plate rigidity, E is the modulus of
elasticity, is Poissons ratio, d is the initial gap between the electrodes, V is the applied voltage,
and is the dielectric constant of the medium.
As shown in Chapter 2, the compatibility relation can be expressed as
3 F 2 F 1 F
1 w 2
r 3 +
=
r2
r r
2 r
(6.2)
= 0 at r = R
r
F < and w
< at r = 0
2
F F
+ Eh(Tb T0 ) = 0 at r = R
r2
r r
w
= 0 and
(6.3)
(6.4)
(6.5)
where T0 is the zero-stress temperature, R is the radius of the plate, and is the coefficient of
thermal expansion.
We introduce nondimensional variables, defined as follows:
1/2
2 h
r = Rr, t = R
t, w
= dw, F = Ehd2 F
D
Substituting Equation (6.6) into Equations (6.1)-(6.5), we obtain
"
#
2 w F
2F
2w
1
1
w
2 V 2
4
+
w
=
+
+
1
t2
r r2 r
r r r2
(1 w)2
(6.6)
(6.7)
Waleed F. Faris
Chapter 6.
3F
2F
1 F
1
r 3 +
=
2
r
r
r r
2
w
r
83
w
= 0 at r = 1
r
F < and w < at r = 0
F
2F
(Tb T0 )R4
= 0 at r = 1
+
r2
r r
h4
w = 0 and
(6.8)
(6.9)
(6.10)
(6.11)
where
1 =
12(1 2 )d2
R4
and
=
2
h2
2Dd3
and the coefficient 1 represents the strength of the nonlinear terms. In general the gap to thickness
ratio can be the order of 1-2. and it can be less than one in case of what is called Touch Mode
Capacitive Sensors.
We transform Equations (6.7)-(6.11) first into a problem with homogeneous boundary conditions.
To this end, we solve the linear part of Equation (6.8) subject to the boundary conditions (6.10)
and (6.11). The general solution of the linear part of Equation (6.8), which is bounded at the
origin, can be expressed as
F = C1 (t)r2 + C2 (t)
(6.12)
(Tb T0 )R4
2(1 )h4
(Tb T0 )R4 2
r + C2 (t)
2(1 )h4
(6.13)
(6.14)
Waleed F. Faris
Chapter 6.
3 2 1
1 w 2
r 3 +
=
r
r2
r r
2 r
w
w = 0 and
= 0 at r = 1
r
w < and < at r = 0
2
= 0 at r = 1
r2
r r
84
(6.15)
(6.16)
(6.17)
(6.18)
where
p=
The coefficient p depends on the temperature and thermal properties of the material. For example,
in case of RF MEMS capacitive switches where either aluminium or gold membranes are used and
the temperature varies between 50 and 70 C, p can take values ranging from 5 to 16. This is a wide
range covering prebuckling, postbuckling, and collapse of the plate.
REDUCED-ORDER MODEL
A general analytical solution of the nondimensional coupled Equations (6.14)-(6.18) is not available
yet, hence we seek an approximate analytical solution in the form
w(r, t) =
N
X
m (t)m (r)
(6.19)
m=1
where m (r) is the mth shape function, m (t) is the mth generalized coordinate for the mth
shape function, and N is the number of retained shape functions. As N approaches infinity, the
approximation in Equation (6.19) becomes exact if the chosen shape functions form a complete
set. Once we solve for all the m (r) and m (t), we calculate an approximate deflection w(r, t) from
Equation (6.19).
We choose the shape functions m (r) to be the first N axisymmetric modes of the linear undamped
case (1 = 0) with the in-plane load due to temperature and no electrostatic forcing (V = 0). For
this case, the mode shapes are given by
m (r) =
J0 (r1m ) I0 (r2m )
J0 (1m )
I0 (2m )
(6.20)
Waleed F. Faris
Chapter 6.
85
where J0 is the Bessel function of the first kind, I0 is the modified Bessel function of the first kind,
and
1m =
i
i
1 hp 2
1 hp 2
p + 4m + p and 2m =
p + 4m p
2
2
and m is the known mth nondimensional natural frequency associated with the mth linear undamped axisymmetric mode m (r). We note that the mode shapes are orthonormal; that is,
Z
0
rm (r)n (r)dr = mn
(6.21)
N
X
(6.22)
m,n=1
Z
r 1 0m 0n
+
d +
4 0
0
mn
(r) =
Z r
1
0m 0n d
4r 0
Z
r1+ 1 0 0
m n d
41 0
(6.23)
Here, the prime denotes differentiation with respect to the space variable r and the overdot denotes
0
differentiation with respect to the time variable t. We note that mn
is symmetric in m and n for
m, n = 1, 2, . . . , N .
We then substitute Equations (6.19), (6.20), and (6.23) into Equation (6.14) and obtain
!2
N
N
N
X
X
X
1 0
2
00
2
m + m m m + p
m m + m = 2 V
1
i i
+
r
m=1
m=1
i=1
"
#
N
1 X
0
00
1
m n p 00m np
+ 0m np
r
(6.24)
m,n,p=1
2
P
We multiply Equation (6.24) with rq (r) 1 N
Waleed F. Faris
Chapter 6.
86
2
q + q2 q 2
m + m
m i
p
"
N
X
Z
m
m=1
r0m 0q dr
N
X
m n p
+ 2p
i j m n p
i,j,m,n,p=1
i m
r0m (i q )0 dr
N
X
i,j,m=1
N
X
Z
i m n p
i,m,n,p=1
1
i,j,m=1
0
0q 0m np
dr + 2
N
X
N
X
i,m=1
m,n,p=1
ri m q dr +
i,m=1
Z
N
X
2
m + m
m i j
0
(i j q )0 0m np
dr + 2 V 2
Z
0
Z
i j m
ri j m q dr
r0m (i j q )0 dr =
0
(i q )0 0m np
dr
rq dr
(6.25)
Equation (6.25). Then, once all m (t) are determined by solving the N coupled equations (6.25),
the plate deflection w(r, t) can be calculated using Equation (6.19).
STATIC BEHAVIOR
Setting all time derivatives equal to zero, we reduce the system of coupled differential equations
(6.25) into a system of coupled algebraic equations. Choosing values for the parameters 1 , , p,
and N , we solve this algebraic system for the m and then the static deflection using Equation
(6.19). In what follows, we choose 1 = 1 corresponding to a gap the order of the plate thickness,
= 0.3 (a very common value for metals), and N = 5. We present results for different values of p
below the critical (buckling) value pcr .
Figure 6.1 shows the interplay between the electrostatic and thermal loadings. For each thermal load
level =
p
pcr ,
we plot variation of the maximum plate deflection wm = w(0) with the electrostatic
load coefficient 2 V 2 . For each 2 V 2 , there are two values of wm , the smaller value is stable and
the larger one is unstable. For each , the stable value of wm increases with increasing 2 V 2 up
Waleed F. Faris
Chapter 6.
87
1
1=1
0.9
0.8
=0.0
0.7
wm
0.6
0.5
=0.9
=0.7
=0.5
=0.3
=0.1
0.4
0.3
0.2
0.1
0
0
2 V2
10
15
Figure 6.1: Variation of the normalized maximum deflection with the electrostatic coefficient.
to a critical value 2 Vp2 beyond which the plante collapses. The voltage Vp is known as the pull-in
voltage. At this value, the stable solution branch (solid line) collides with the unstable branch
(dashed line) and destroy each other in a saddle-node bifurcation.
In Figure 6.2, we plot variation of the maximum plate deflection wm = w(0) with the electrostatic
load coefficient 2 V 2 for 1 = 0.0, 3.0, and 8.0. As mentioned above, the coefficient 1 represents
the strength of the geometric nonlinearity. As 1 increases, the pull-in voltage increases. Also, as
1 increases, the value of wm at which pull-in occurs increases. It follows from Figure 6.2 that 1
does not have a significant effect on the stable maximum deflection below 2 V 2 10.
NATURAL FREQUENCIES
In this section, we determine the natural frequencies of the axisymmetric modes of the deflected
plate. The procedure followed here is similar to that reported by Vogl and Nayfeh (2003). First,
eq
we perturb each coefficient function m (t) with a harmonic term from its equilibrium value m
;
that is,
eq
m (t) = m
+ m eit
(6.26)
Waleed F. Faris
Chapter 6.
88
1
0.9
1=24
0.8
0.7
=1
1
wm
0.6
p=0
=12
1
0.5
0.4
0.3
0.2
0.1
0
0
2 V2
10
15
20
Then, we substitute Equation (6.26) into the reduced-order model (6.25) and keep only the linear
terms in m . Because the constant terms, which are due to equilibrium, cancel out, the reducedorder model becomes a system of N linear homogeneous algebraic equations in the unknowns m
and ; that is,
[Z()]{m } = {0},
(6.27)
(6.28)
Equation (6.28) is an algebraic equation of order 2N with N roots associated with unique modes
of vibration. The positive roots correspond to the natural frequencies of the deflected plate and
the negative roots correspond to unstable equilibrium states.
Figure 6.3 shows variation of the first five nondimensional axisymmetric natural frequencies with
the nondimensional temperature p. Figures 6.4 and 6.5 show variation of the first natural frequency
with the electrostatic coefficient for three thermal load levels, namely, 0.0,0.3, and 0.7. In Figure
6.4, 1 = 1; it indicates a weak nonlinear effect. In Figure 6.5, 1 = 12; it indicates a strong effect
of the nonlinearity. It follows from both figures that increasing the thermal load decreases the first
Waleed F. Faris
Chapter 6.
89
250
5
Frequency
200
4
150
100
50
2
1
10
15
Nondimensional Temperature p
Figure 6.3: Variation of the first five frequencies with the nondimensional temperature p.
natural frequency as well as the pull-in voltage. Moreover, increasing the strength of the mid-plane
stretching nonlinearity increases the natural frequency and the pull-in voltage.
Waleed F. Faris
Chapter 6.
12
1 = 1.0
10
= 0.0
= 0.3
= 0.7
4
10
2V 2
15
12
1 = 12
10
= 0.0
= 0.3
= 0.7
10
12
14
16
2 V2
90
Chapter 7
Summary
In the first two chapters of this work we reviewed the relevant literature and formulated the general
von Karman equations for circular plates under both thermal and electrostatic loadings.
In Chapter 3, we studied the mechanical behavior of an electrostatically actuated micropump. We
modeled the actuating element as an annular plate using the von Karm
an nonlinear model for
plates to account for in-plane stresses and moderately large deformations. The electrostatic forcing
adds another nonlinearity to the model. We solved for the nonlinear static deflection first and
then solved the linearized eigenvalue problem around the calculated deflected shape. We presented
results describing the static characteristics and the linear natural frequencies and the corresponding
mode shapes of the actuating element under two different boundary conditions.
Most of the previous research on thermally induced vibrations in circular plates has been limited
to small deflections. Moreover, the works that deal with large deflections were based on Bergers
approximation. In Chapter 4, we investigated the nonlinear response of a circular plate to a thermal
loading consisting of a steady component and a sinusoidal component. We based the analysis on the
91
Waleed F. Faris
92
heat conduction equation and the dynamic analogue of the von Karm
an equations. We neglected
the dissipation terms in the heat equation and solved it explicitly for the temperature distribution.
Then, we substituted the resulting distribution into the equations describing the displacement and
the stress function. We used the method of multiple scales to determine a first-order approximation
of the plate response with simply supported boundary conditions in the case of principal parametric
resonance of an axisymmetric mode, which is not involved in an internal resonance with any other
mode, and obtained two first-order nonlinear ordinary-differential equations governing the amplitude and phase of the response. We used these modulation equations to generate frequency-and
force-response curves. The numerical results showed that the heat variation can lead to largeamplitude responses, including multivalued responses and jumps. We followed the same procedure
for the second case of a clamped circular plate to investigate combination parametric resonances of
the additive and difference types and also principal parametric resonance.
In Chapter 5, we investigated the static and dynamic behavior of postbuckled circular plates under a thermal load using the Galerkin approximation. We formulated the dynamic postbuckling
problem of a circular plates under a constant temperature using the von Karman plate model. We
used the Galerkin procedure to reduce the distributed-parameter problem into a finite system of
ordinary-differential equations in time. The linear undamped modes with in-plane loading were
used as basis functions in the discretization. For given system parameters, the reduced-order system was used to calculate approximations to the plate static as well as dynamic deflections. First,
we solves for the equilibrium state due to the thermal load. We concluded that five modes are adequate for determining the static deflection, but they were not adequate for calculating the radial
stresses, which needed 15 modes for convergence. Then, we repeated the calculations for the simply
supported case and comapared the results with those obtained using the power-series solution. We
used five modes to determine the natural frequencies of the axisymmetric modes for the deflected
plate under an electrostatic actuation. As expected, the fundamental natural frequency decreases
as the load level increases in the post-buckling regime.
The present analysis shows that the one-term Galerkin approximation might yield misleading results. Using a three-mode approximation might produce acceptable deflections, but not radial
stresses. The present analysis also provides designers with a tool that can be used to analyze the
Waleed F. Faris
93
7.2
Future Work
We recommend the following topics as a continuation of this thesis. They will provide a complete
picture of the modeling and analysis of the cases examined above.
1. We recommend experiments to verify the frequency change due to thermal effects.
Waleed F. Faris
94
References
1. A. Abe, Y. Kobayashi, and G. Yamada, Three-mode response of simply supported rectangular
laminated plates, JSME International Journal 41 (1998), 51-59.
2. A. Abe, Y. Kobayashi, and G. Yamada, Analysis of subharmonic resonance of moderately
thick antisymmetric angle-ply laminated plates by using method of multiple scales, Journal
of Sound and Vibration 217 (1998), 467-484.
3. J. Ahmed, Dynamic post-buckling characteristics of circular plates, Journal of the Franklin
Institute 289 (1970), 57-66.
4. G. A. Altay and M. C. Dokmeci, Karm
an-Mindlin plate equations for thermoelastic vibrations
of temperature dependent materials, ARI 50 (1997), 110-126.
5. M. Amabili, G. Frosali, and M.K. Kwak, Free vibrations of annular plates coupled with fluids,
Journal of Sound and Vibration 191 (1996), 825-846.
6. L. Azrar, E. H. Boutyour, and M. Potier-Ferry, Nonlinear forced vibrations of plates by an
asymptotic-numerical method, Journal of Sound and Vibration 252 (2002), 657-674.
7. K. Bairagi, Textbook of Plates, Khanna, Bombay, 1986.
8. P. Biswas and P. Kapoor, Nonlinear vibrations of circular plates at elevated temperature, 4th
International Conference on Numerical Methods in Thermal Problems Swansea, UK (1985),
1493-1501.
9. F. Bloom and D. Coffin, Handbook of Thin Plate Buckling and Postbuckling, Chapman and
Hall/CRC, Boca Raton, Florida, 2001.
95
Waleed F. Faris
References
96
10. R. Bodner, The post-buckling behavior of a clamped circular plate, Quarterly of Applied
Mathematics 12 (1954), 397-401.
11. B. A. Boley and J. H. Weiner, Theory of Thermal Stresses, Wiley, New York, 1960.
12. T. Bourouina and J-P. Grandchamp, Modeling micropumps with electrical equivalent networks, Journal of Micromechanics and Microengineering 6 (1996), 398-404.
13. M. E. Brewster, The post-buckling behavior of a clamped circular plate, SIAM Journal on
Applied Mathematics 51 (1991), 1255-1283.
14. F. Buckens, Vibrations in a thermally stressed thin plate, Journal of Thermal Stresses 2
(1979), 367-385.
15. W. P. Chang and S. M. Wan, Thermomechanically coupled non-linear vibration of plates,
International Journal of Non-Linear Mechanics 21(5) (1986), 375-389.
16. C-Y. Chia, Nonlinear Analysis of Plates, McGraw-Hill, New York, 1980.
17. H-N. Chu and G. Herrmann, influence of large amplitudes of free flexural vibrations of rectangular elastic plates, Journal of Applied Mechanics 23 (1956), 532-542.
18. A. Cozma and R. Puers, Electrostatic actuation as a self-testing method for silicon pressure
sensors, Sensors and Actuators A 60(5) (1997), 32-36.
19. E. M. Dombourian, C. V. Smith, and R. L. Carlson, A Perturbation solution to a plates
postbuckling problem, International Journal of Non-Linear Mechanics 11 (1976), 49-58.
20. P. C. Dumir, Nonlinear vibration and postbuckling of isotropic thin circular plates on elastic
foundations, Journal of Sound and Vibration 107 (1986), 253-263.
21. P. C. Dumir, R. Kumar, and M. L. Ghandi, Nonlinear axisymmetric vibration of orthotropic
thin circular plates on elastic foundation, Journal of Sound and Vibration 103 (1985), 273283.
22. P. C. Dumir, R. Kumar, and M. L. Ghandi, Nonlinear axisymmetric vibration of orthotropic
thin circular plates with elastically restrained edges, Journal of Computers and Structures 26
(1986), 677-686.
Waleed F. Faris
References
97
23. O. Francais, I. Dufour, and E. Sarraute, Analytical static modeling and optimization of
electrostatic micropump, Journal of Micromechanics and Microengineering 7 (1997), 183185.
24. O. Francais and I. Dufour, Dynamic simulation of an electrostatic micropump with pull-in
and hystersis phenomena, Sensors and Actuators A 70 (1998), 56-60.
25. O. Francais and I. Dufour, Normalized abacus for the global behavior of diaphragms: pneumatic, electrostatic, piezoelectric, or electromagnetic actuation, Journal of Modeling and Simulation of Microsystems 1 (1999), 149-160.
26. K. O. Friedrichs and J. J. Stoker, Buckling of the circular plate beyond the critical thrust,
Journal of Applied Mechanics 9 (1942), A7-A14.
27. B. E. Gatewood, Thermal Stresses, McGraw-Hill, New York, 1957.
28. R. C. Gauss and S. S. Antman, Large thermal buckling of nonuniform beams and plates,
International Journal of Solids and Structures 20 (1982), 979-1000.
29. C. Gehin, C. Barthod, and Y. Teisseyre, Design and characterization of a force resonant
sensor, Sensors and Actuators A: Physical 84 (2000), 65-69.
30. M. L. Gossard, P. Seide, and W. M. Roberts, Thermal Buckling of Plates, NACA Technical
Note 2771, 1952.
31. J. Hadian and A. H. Nayfeh, Modal interaction in circular plates, Journal of Sound and
Vibration 142 (1990), 279-292.
32. R. R. Heldenfels and W. M. Roberts, Experimental and theoretical determination of thermal
stresses in a flat plate, NACA Technical Note 3578, 1956.
33. G. Herrmann, Influence of large amplitudes on flexural motions of elastic plates, NACA
Technical Note 2769, 1952.
34. R. B. Hetnarski, Thermal Stresses II, North-Holland, The Netherlands, 1987.
35. C-L. Huang and B. E. Sandman, Large amplitude vibrations of a rigidly clamped circular
plates, International Journal of Non-Linear Mechanics 6 (1971), 451-468.
Waleed F. Faris
References
98
36. T. Irie and G. Yamada, Thermally induced vibration of circular plate, Bulletin of the Japanese
Society of Mechanical Engineers 21 (1978), 1703-1709.
37. T. Y. Jiang, T. Y. Ng, and K. Y. Lam, Dynamic analysis of an electrostatic micropump, 3rd
International Conference on Modeling and Simulation of Microsystems, MSM2000 California
2000.
38. R. Jones, J. Mazumdar, and Y. K. Cheung, Vibration and buckling of plates at elevated
temperatures, International Journal of Solids and Structures 16 (1980), 61-70.
39. E. R. Konig and G. Wachutka, Analysis of unstable behavior occurring in electro-mechanical
microdevices, 2nd International Conference on Modeling and Simulation of Microsystems,
MSM1999 California 1999.
40. A. Y. T. Leung and S. G. Mao, A symplectic Galerkin method for non-linear vibration of
beams and plates, Journal of Sound and Vibration 183 (1995), 475-491.
41. L. Librescu, Plates and Shells, Unpublished Lecture Notes, 2001.
42. L. Librescu and W. Lin, Non-linear response of laminated plates and shells to thermomechanical loading: Implications of violation of interlaminar shear traction continuity requirement,
International Journal of Solids and Structures 36 (1999), 4111-4147.
43. R. M. Lin, M. K. Lim, and H. Du, Large deflection analysis of plates under thermal loading,
Computer Methods in Applied Mechanics and Engineering 117 (1994), 381-390.
44. S. P. Maganty and W. B. Bickford, Large amplitude oscillation of thin circular rings, Journal
of Applied Mechanics 54 (1987), 315-322.
45. E. H. Mansfield, On the large-deflexion vibrations of heated plates, Proceedings of the Royal
Society of London, Series A 379 (1982), 15-39.
46. K. Marguerre, Thermo-elastische platten-gleich-ungen, Zeitschrift fur Angewandte Mathematik und Mechanik 15 (1935), 369-372.
47. M. A. Michalicek, Introduction to Microelectromechanical Systems: A Short Course, Air Force
Research Laboratory, New Mexico, 2000.
Waleed F. Faris
References
99
48. Y. Nath and R. S. Alwar, Nonlinear dynamic analysis of orthotropic circular plates, International Journal of Solids and Structures 16 (1980), 433-443.
49. Y. Nath, Large amplitude response of circular plates on elastic foundation, International
Journal of Non-Linear Mechanics 17 (1982), 285-296.
50. Y. Nath and S. Kumar, Large amplitude response of layered circular plates, Journal of Engineering Mechanics 121 (1995), 37-49.
51. A. H. Nayfeh, Introduction to Perturbation Techniques, Wiley, New York, 1981.
52. A. H. Nayfeh, Nonlinear Interactions, Wiley, New York, 2000.
53. A. H. Nayfeh and D. T. Mook, Nonlinear Oscillations, Wiley, New York, 1979.
54. M. Newman and M. Forray, Axisymmetric large deflections of circular plates subjected to
thermal and mechanical loads, Journal of Aerospace Science 29 (1962), 1060-1066.
55. C. F. Ng and R. G. White, Dynamic behavior of postbuckled isotropic plates under in-plane
compression, Journal of Sound and Vibration 120 (1988), 1-18.
56. J. L. Nowinski, Nonlinear transverse vibrations of circular elastic plates built-in at the boundary, 4th US National Congress on Applied Mechanics 1 (1962), 325-332.
57. J. L. Nowinski and H. Ohnabe, On certain inconsistencies in Berger equations for large
deflections of elastic plates, International Journal of Mechanical Science 14 (1972), 165-170.
58. K. Oh and A. H. Nayfeh, High- to low frequency modal interactions in a cantilever composite
plate, Journal of Vibration and Acoustics 120 (1998), 579-587.
59. M. C. Pal, Large deflections of heated circular plates, Acta Mechanica 8 (1969), 82-103.
60. M. C. Pal, Large amplitude free vibration of circular plates subjected to aerodynamic heating,
International Journal of Solids and Structures 6 (1970), 301-313.
61. M. C. Pal, Static and dynamic non-linear behaviour of heated orthotropic circular plates,
International Journal of Non-Linear Mechanics 8 (1973), 489-504.
Waleed F. Faris
References
100
Waleed F. Faris
References
101
75. Y. Shulman, On the Vibration of Thermally Stressed Plates in the Pre-Buckling and PostBuckling States, MIT Technical Report 25-25, 1958.
76. I. S. Sokolnikoff and E. S. Sokolnikoff, Thermal stresses in elastic plates, Transactions of the
American Mathematical Society 45 (1939), 235-255.
77. S. Sridhar, D. T. Mook, and A. H. Nayfeh, Nonlinear resonances in the forced responses of
plates, Part I: Symmetric responses of circular plates, Journal of Sound and Vibration 41
(1975), 359-373.
78. S. Sridhar, D. T. Mook, and A. H. Nayfeh, Nonlinear resonances in the forced responses of
plates, Part II: Asymmetric responses of circular plates, Journal of Sound and Vibration 59
(1978), 159-170.
79. A. V. Srinivasan, Nonlinear vibrations of beams and plates, International Journal of NonLinear Mechanics 1 (1966), 179-191.
80. E. Suhir, Structural Analysis in Microelectronic and Fiber-Optic Systems, van Nostrand Reinhold, New York, 1991.
81. M. Sunakawa, Influence of temperature changes and large amplitude on free flexural vibration
of rectangular elastic plates, Transactions of JSME 30 (1964), 558-566.
82. T. R. Tauchert, Thermally induced flexure, buckling, and vibration of plates, Applied Mechanics Review 44 (1991), 347-360.
83. E.H. Tay and K.H. Phoon, Review of the technology of micropumps, Journal of Institution
of Engineers, Singapore 37 (1997), 68-72.
84. E. A. Thornton, Thermal buckling of plates and shells, Applied Mechanics Review 46 (1993),
485-506.
85. D. Trajkovski and R. Cukic, A coupled problem of thermoelastic vibrations of a circular plate
with exact boundary conditions, Mechanics Research Communications 26 (1999), 217-224.
86. A. C. Ugural, Stresses in Plates and Shells, McGraw-Hill, New York, 1981.
Waleed F. Faris
References
102
87. K. L. Verma, Thermoelastic vibration of a transversely isotropic plate with thermal relaxations, International Journal of Solids and Structures 38 (2001), 8529-8546.
88. J. R. Vinson, The Behavior of Thin Walled Structures, Kluwer, Amsterdam, The Netherlands,
1989.
89. Vogl, G. W. and Nayfeh, A. H., A reduced-order model for electrically actuated clamped
circular plates, ASME International 19th Biennial Conference on Mechanical Vibration and
Noise (VIB), September 2-6, 2003, Chicago, Illinois.
90. T. Wah, Vibration of circular plates at large amplitudes, Journal of Engineering Mechanics
Division, Proceedings of American Society of Civil Engineers EM 5 (1963), 1-15.
91. A. Wang, Axisymmetric postbuckling and secondary bifurcation buckling of circular plates,
International Journal of Non-Linear Mechanics 35 (2000), 279-292.
92. P. K. C. Wang and F. Y. Hadaegh, Computation of static shapes and voltages for micromachined deformable mirrors with nonlinear electrostatic actuators, IEEE Journal of Microelectromechanical Systems 5 (1996), 205-220.
93. J. E. Warren, A. M. Brzezinski, and J. F. Hamilton, Capacitance-microphone static membrane
deflections, Journal of the Acoustical Society of America 52 (1972), 711-719.
94. J. E. Warren, A. M. Brzezinski, and J. F. Hamilton, Capacitance-microphone dynamic membrane deflections, Journal of the Acoustical Society of America 54 (1973), 1201-1213.
95. Z. Wei, Analysis of global dynamics in a parametrically excited thin plate, Acta Mechanica
Sinica (English Series) 17 (2001), 71-85.
96. S. Xuefeng, C. Yiping, and Z. Xiaoqing, The study of nonlinear thermoelastic vibration of
circular plate, Acta Mechanica Solida Sinica 20 (1999), 245-250.
97. N. Yamaki, Influence of large amplitudes on flexural vibrations of elastic plates, Zeitschrift
fur Angewandte Mathematik und Mechanik 41 (1961), 501-510.
Waleed F. Faris
References
103
98. Y-L. Yeh, C-K. Chen, and H-Y. Lai, Chaotic and bifurcation dynamics for a simply supported rectangular plate of thermo-mechanical coupling in large deflection, Chaos, Solitons
and Fractals 13 (2002), 1493-1506.
99. P. Yu, W. Zhang, and Q. Bi, Vibration analysis on a thin plate with the aid of computation
normal forms, International Journal of Non-Linear Mechanics 36 (2001), 597-627.
100. W. Zhang, Global and chaotic dynamics for a parametrically excited thin plates, Journal of
Sound and Vibration 239 (2001), 1013-1036.
101. J. Zhu, S. K. Datta, and A. H. Shah, Modal representation of transient dynamics of laminated
plates, Journal of Composite Engineering 5 (1995), 949-963.
Vita
Waleed Fekry Faris was born on December 15, 1966 in Sharkia, Egypt. He joined the College
of Engineering at Zagzig University in 1984 and graduated from the Department of Mechanical
Engineering in 1989 with honors and he was ranked the first in the graduating class at Zagzig
University. After finishing his military service, he joined the Mechanical Engineering Department
as an instructor. He finished his MS in 1996. In 1999, he joined the Department of Engineering
Science and Mechanics at Virginia Tech to pursue a PhD in nonlinear dynamics, which he earned
in December 2003.
104