Handbook of Thin Plate Buckling and Postbuckling
Handbook of Thin Plate Buckling and Postbuckling
Handbook of Thin Plate Buckling and Postbuckling
Thin Plate
Buckling
and
Postbuckling
229x152 HB
Library of Congress Cataloging-in-Publication Data
This book contains information obtained from authentic and highly regarded sources. Reprinted material
is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable
efforts have been made to publish reliable data and information, but the author and the publisher cannot
assume responsibility for the validity of all materials or for the consequences of their use.
Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic
or mechanical, including photocopying, microfilming, and recording, or by any information storage or
retrieval system, without prior permission in writing from the publisher.
The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for
creating new works, or for resale. Specific permission must be obtained in writing from CRC Press LLC
for such copying.
Direct all inquiries to CRC Press LLC, 2000 N.W. Corporate Blvd., Boca Raton, Florida 33431.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation, without intent to infringe.
Preface
References
Preface
In the Spring of 1996, the authors initiated a research program at the In-
stitute of Paper Science and Technology (Atlanta, Ga.) whose goal was
to model cockle, i.e., the hygroscopic buckling of paper. In the course
of researching the aforementioned problem, an extensive examination
was made of the expansive literature on the buckling and postbuckling
behavior of thin plates in an effort to understand how such phenomena
were controlled by the wide variety of factors listed below. A compara-
tive study of plate buckling was then produced and published as a pair of
IPST reports; this book presents the essential content of these reports.
The essential factors which influence critical buckling loads, initial
mode shapes, and postbuckling behavior for thin plates are studied,
in detail, with specific examples, throughout this treatise; among the
factors discussed which affect the occurrence of initial buckling and the
initial mode shapes into which plates buckle are the following:
(i) Aspect Ratios: Included in our discussions will be examples
of buckling for rectangular plates, circular plates, and annular plates;
we will look at how, e.g., critical buckling loads for the various types of
plates are affected by changes in the aspect ratio, associated with the
plate, given that all other factors (support conditions, load conditions,
etc.) are held fixed. For a rectangular plate, the aspect ratio is just the
ratio of its sides; for a circular plate it is just the plate radius while, for
an annular region, it is the ratio of the inner to the outer radius.
(ii) Support Conditions: The edge(s) of a plate may be supported
in many different ways; mathematically, an edge support condition is
reflected in a constraint on the out-of-plane deflection of that plate along
its edge(s). The basic types of support that will be considered are those
which correspond to having the edges(s) of the plate clamped, free, or
simply supported. For a rectangular plate or an annular plate, one edge
(or pair of edges) may be supported in one manner while the other edge
(or pair) of edges can be supported in a different manner.
(iii) Load Conditions: The edge(s) of a thin plate may be subjected
© 2001 by Chapm
to compressive, tensile, or shear loadings or to some combinations of
such loadings along different edges or pairs of edges. The effect of using
different load conditions on critical buckling loads and mode shapes for
rectangular, circular, and annular plates are illustrated throughout the
book.
(iv) Material Symmetry: For thin plates exhibiting a particular
constitutive response, the material symmetry exhibited by the plate may
have a marked effect on, e.g., the magnitude of the lowest buckling load,
when all other factors, such as aspect ratio or edge support conditions
are held constant. The basic types of material symmetry that will be
examined in this work are isotropic symmetry, rectilinearly orthotropic
symmetry, and cylindrically orthotropic symmetry.
(v) Constitutive Behavior: Our examination of the influence of
constitutive response on initial buckling of thin plates was guided by the
behavior observed with respect to the deformation of paper, i.e., that
under various combinations of loading conditions, paper may exhibit
linear elastic, nonlinear elastic, viscoelastic, or elastic-plastic. Examples
of all four kinds of such constitutive response are presented throughout
this treatise.
(vi) Hygroscopic or Thermal Stress Distributions: Thin plates
are often exposed to variations in temperature or moisture or hygroex-
pansive strains which result in local compressive stress distributions that
can cause buckling even in the absence of applied external loading.
Beyond the factors delineated above, which influence both critical
buckling loads and mode shapes for thin plates and also influence the
subsequent postbuckling behavior, two other important factors come
into play with respect to postbuckling behavior.
(vii) Initial Imperfections: Thin plates often possess some small
initial transverse deflection (e.g., a paper sheet with an initial cockle
which is wetted, dried, and experiences further local buckling) or may
be subjected to a small normal loading (transverse) to the initially flat
unbuckled state of the plate; the presence of either or both of these con-
ditions amounts to an “initial imperfection” which may have a profound
effect on the subsequent postbuckling behavior of the plate.
(viii) Secondary Buckling: After a thin plate has buckled into an
initial mode shape, it is possible that, with increasing load, instead of
obtaining a buckled shape which deforms continuously during postbuck-
ling, as the magnitude of the loading increases, a critical load larger than
the initial buckling load is reached at which the buckle pattern changes
suddenly and a new mode shape appears; such phenomena have been
exhibited with respect to plate buckling and will be examined with par-
ticular emphasis on the unsymmetric wrinkling that has been observed
in circular plates.
Throughout this treatise, several results appear which cannot be found
in earlier books on plate buckling, e.g., the discussion of the Johnson-
Urbanik generalization of the von Karman Equations for plates exhibit-
ing nonlinear elastic behavior; other results appear which have not pre-
viously been published elsewhere, most notably the analysis of initial
buckling for annular rectilinearly orthotropic thin plates, which was pre-
sented in the Ph.D. Thesis of the second author.
The authors wish to express their gratitude to the Institute of Paper
Science and Technology and its member companies for the support which
made possible the production of the IPST reports that form the basis
for the present work.
© 2001 by Chapm
THIS BOOK IS DEDICATED TO
the memory of Sidney Bloom
(1912-1999)
and to the memory of
his grandson
Scott Bloom
(1969-1997)
Chapter 1
© 2001 by Chapm
cited above, as well as [18] and [19], which employ singularity theory for
maps, an outgrowth of the catastrophe theory of Thom [20]). Besides
problems in buckling and postbuckling of structures and, in particular,
the specific problems associated with the buckling and postbuckling be-
havior of thin plates, general ideas underlying bifurcation and nonlinear
stability theory have been used to study problems in fluid dynamics re-
lated to the instabilities of viscous flows as well as branching problems
in nonlinear heat transfer, superconductivity, chemical reaction theory,
and many other areas of mathematical physics. Beyond the references
already listed, a study of various fundamental issues in branching theory,
with applications to a wide variety of problems in physics and engineer-
ing, may be found in references [21]-[58], to many of which we will have
occasion to refer throughout this work.
To illustrate the phenomena of bifurcation within the specific context
of buckling and postbuckling of structures, we will use the example of the
buckling of a thin rod under compression, which is due to Euler, op. cit,
1744, and which is probably the simplest and oldest physical example
which illustrates this phenomena. In Fig. 1.1, we show a homogeneous
thin rod, both of whose ends are pinned, the left end being fixed while
the right end is free to move along the x-axis. In its unloaded state, the
rod coincides with that portion of the x-axis between x = 0 and x = 1.
Under a compressive load P , a possible state of equilibrium for the
rod is that of pure compression; however, experience shows that, for
sufficiently large values of P , transverse deflections can occur. Assuming
that the buckling takes place in the x, y plane, we now investigate the
equilibrium of forces on a portion of the rod which includes its left end;
the forces and moments are taken to be positive, as indicated in Fig.
1.2.
Let X be the original x-coordinate of a material point located in a
portion of the rod depicted in Fig. 1.2. This point moves, after buckling,
to the point with coordinates (X + u, v). We let ϕ be the angle between
the tangent to the buckled rod and the x-axis and s the arc length along a
portion of the rod (measured from the left end). Although more general
constitutive laws may be considered, we restrict ourselves here to the
case of an inextensible rod in which the Euler-Bernoulli law relates the
moment M acting on a cross-section with the curvature dϕ/ds. Thus
s = X while
dϕ
M = −EI (1.1)
ds
with EI the (positive) bending stiffness. The constitutive relation (1.1)
is supplemented by the geometric relation
dv
= sin ϕ (1.2)
ds
and the equilibrium condition
M = Pv (1.3)
Combining the above relations, we obtain the pair of first order non-
linear differential equations
λv = − dϕ , λ = P/EI
ds (1.4)
dv
= sin ϕ
ds
with associated boundary conditions
d2 ϕ
+ λ sin ϕ = 0, 0 < s < (1.6a)
ds2
ϕ (0) = ϕ () = 0 (1.6b)
1 dϕ
V = EI ds − P − cos ϕds (1.7)
2 0 ds 0
and that setting the first variation δV = V (ϕ)δϕ = 0 yields the differ-
ential equation (1.6a) with the natural boundary conditions (1.6b).
The linearized version of (1.6a,b) for small deflections v, and small
angles ϕ, is obtained by substituting ϕ for sin ϕ (a precise mathematical
justification for considering the linearized problems so generated will be
considered below); the linearized problem for v then becomes
© 2001 by Chapm
2
d v + λv = 0
ds2 (1.8)
v(0) = v() = 0
n2 π 2
which has eigenvalues λn = , n = 1, 2, · · · with corresponding
2
nπx
eigenfunctions vn = c sin( ). It is desirable to be able to plot v
versus λ but, unfortunately, v is itself a function (an infinite-dimensional
vector) so we content ourselves instead, in Fig. 1.3, with a graph of the
maximum deflection vmax versus λ = P/EI. For loads below the first
π 2 EI
critical load P1 = no buckling is possible. At the load P1 , buckling
2 πx
can take place in the mode c sin but the size of the deflection is
undetermined due to the presence of the arbitrary constant c. Now, if
an analysis of the buckled behavior of the rod is based entirely on the
linearized problem, then we see that, as the load slightly exceeds P1 ,
the rod must return to its unbuckled state until the second critical load
4π 2 EI
P2 = is reached, when buckling can again occur, but now in
2πx
the new mode c sin , which is still of undetermined size. Except
n2 π 2 EI
at the critical loads Pn = there is no buckling. On physical
2
grounds the picture provided only by the linearized problem is clearly
unsatisfactory.
The nonlinear problem (1.4), (1.5), gives a more reasonable prediction
of the buckling phenomena. Again, no buckling is possible until the
compressive load reaches the first critical load of the linearized theory
(the reason for this being discussed below) and as this value is exceeded
the possible buckling deformation is completely determined except for
sign. For values of the load between the first two critical loads there are
therefore three possible solutions as shown in Fig. 1.4. When P exceeds
the second critical load a new, determinate, pair of nontrivial solutions
appears, and so forth. The values of λ corresponding to these critical
loads are said to be branch points (or bifurcation points) of the trivial
solution because new solutions, initially of small size, appear at these
points.
A diagram like Fig. 1.4 is called a branching (or bifurcation diagram)
and we may justify the statements, above, about the buckling behav-
ior for the problem of the compressed thin rod by exhibiting the closed
form solutions of (1.6a,b). The deflection v can then be obtained from
the first relation in (1.4). We first note certain simple properties of the
boundary value problem (1.4), (1.5), namely, if any one of the triplets
(λ, v, ϕ), (λ, v, ϕ + 2nπ), (λ, −v, −ϕ), or (−λ, −v, ϕ + π + 2nπ), is a solu-
tion so are the other three; all four solutions yield congruent deflections.
In fact, the first two are identical, while the third is a reflection of the
first about the x-axis and the fourth is a reflection of the first about the
origin; in this last case, a reversal in the sign of the load is accounted for
by interchanging the roles of the left-hand and right-hand ends of the
rod. Similar remarks apply to the system (1.6a,b).
All nontrivial lateral deflections v(s) of the rod are therefore generated
by those solutions of the initial value problem
2
d ϕ + λ sin ϕ = 0, λ > 0
ds2 (1.9)
ϕ(0) = α, ϕ (0) = 0, 0 < α < π
The initial value problem (1.9) has one and only one solution and
this solution may be interpreted as representing the motion of a simple
pendulum, with x being the time and ϕ the angle between the pendulum
and the (downward) vertical position. In view of the initial conditions,
ϕ(s) will be periodic. If we multiply the equation in (1.9) by dϕ/ds,
then this equation becomes a first order equation which can be solved
(explicitly) by elliptic integrals, i.e.
√
ϕ(s) = 2 arc sin k sn( λs + K) (1.10)
with
α
k = sin( 2 )
π2 (1.11)
dζ
K =
0 1 − k 2 sin2 ζ
and
u3 u5
sn u = u − (1 + k 2 ) + (1 + 14k 2 + k 4 ) (1.12)
3! 5!
7
u
−(1 + 135k 2 + 135k 4 + k 6 ) + · · ·
7!
© 2001 by Chapm
π
The parameter K takes on values between and +∞ and there is a one-
2
to-one correspondence between these values and those of α in 0 < α < π.
The period of oscillation for ϕ is 4K and the condition ϕ () = 0 is
satisfied if and only if
√
K= λ, n = 1, 2, · · · (1.13)
2n
π
As K > , in order for nontrivial solutions to exist, we must have
2
λ > π 2 /2 . Also, given a value of λ > π 2 /2 there are as many distinct
nontrivial solutions√ ϕ(s), with 0 < ϕ(0) < π, as there are integers n
such that (/2n) λ > π/2. Thus, we have for
0 ≤ λ ≤ π 2 /2 , only the trivial solution
π 2 /2 < λ ≤ 4π 2 /2 , one nontrivial solution
2 2 2
n π / < λ ≤ (n + 1)2 π 2 /2 , n nontrivial solutions
If we then also take into account other values of ϕ(0), and the possibility
of negative λ, we obtain Fig. 1.5, which depicts the maximum value of
ϕ(s) versus λ; the only (physically) significant portion of this figure is,
of course, the part drawn with solid (unbroken) curves, i.e., for λ > 0
and −π < ϕmax < π.
The broken curves for λ > 0 are those attributable to angles ϕ that are
determined only by modulo 2π, while the dotted curves for λ < 0 corre-
spond to having the free end of the rod at x = −. The deflection v(s) is
then obtained from the first equation in (1.4) and yields the bifurcation
diagram in Fig. 1.4, where it is known that the various branches do not
extend to infinity because vmax does not grow monotonically (at large
loads the buckled rod may form a knot so that the maximum deflection
can decrease, although the maximum slope must increase.)
We have presented a complete solution of the buckling problem for a
compressed thin rod which is based on classical analysis of an associ-
ated initial value problem for the boundary value problem (1.6a,b) first
treated by Euler in 1744. The equivalent problem for the deflection v(s)
is (1.4), (1.5) and the results of the analysis, based on a closed form
solution using elliptic integrals, seem to indicate a prominent position
for the linearized (eigenvalue) problem (1.8), or, equivalently, for the
linearization of (1.6a,b) namely,
2
d ϕ + λϕ = 0, 0 < s <
ds2 (1.14)
ϕ (0) = ϕ () = 0
Indeed, this system governs the initial buckling of the rod for reasons
that are explained below; our explanation will be phrased in just enough
generality so as to make our remarks directly applicable to the study of
buckling and postbuckling behavior for thin plates. For the case of linear
elastic response, the von Karman system of nonlinear partial differen-
tial equations, with associated boundary conditions, which govern plate
buckling, are derived in § 1.2.
As we will specify below, the nonlinear boundary value problem (1.6a,
b), which governs the behavior of the compressed rod, can be put in the
form
G(λ, u) = 0 (1.15)
where λ is a real number, u is an element of a real Banach space B
with norm
·
, and G is a nonlinear mapping from R × B into B, R
being the real numbers. The restriction to real λ and a real Banach
space B is based on the needs in applications where only real branching
is of interest. Strictly speaking, a solution of (1.15) is an ordered pair
(λ, u) but we often refer to u itself as the solution (either for fixed λ, or
depending parametrically on λ); to study branching (or bifurcation) we
must have a simple, explicitly known solution u(λ) of (1.15). We may
make the assumption that
so that u(λ) = 0 is a solution of (1.15) for all λ and this solution is then
known as the basic solution (for the problem considered above, this solu-
tion corresponds to the compressed, unbuckled rod). The main problem
is to study branching from this basic solution (i.e., the unbuckled state)
although within the context of plate buckling we will also discuss, in §5,
branching from nontrivial solutions of (1.15), i.e., secondary buckling
of thin plates. Thus, the goal is to find solutions of (1.15) which are of
small norm (small “size” in the relevant Banach space B); this motivates
the following:
o
Definition. We say that λ = λ is a branch point of (1.15) (equivalently
a bifurcation point or, for the case in which (1.15) represents the equi-
librium problem for a structure, such as the compressed rod considered
© 2001 by Chapm
o
above, a “buckling load”) if every neighborhood of (λ , 0), in R × B,
contains a solution (λ, u) of (1.15) with
u
= 0.
We note that the above definition is restricted only to small neighbor-
o
hoods of (λ , 0); also, the definition is equivalent to the existence of a
sequence of solutions (λn , un ) of (1.15), with
un
= 0 for each n, such
o
that (λn , un ) → (λ , 0), as n → ∞. Before placing the compressed rod
problem within the context of the general formulation (1.15), and then
explaining the reason for the apparent principal position of the associ-
ated linearized problem, it is worthwhile to list some of the principal
problems of bifurcation theory (equivalently, buckling and postbuckling
theory for structures); these are covered by the following questions:
Answers to questions 1-3, above, are the only ones which clearly fall
within the domain of bifurcation theory; problem 4 is related to stability
theory, while problem 5 requires, for an answer, techniques of global
analysis which fall outside the strict domain of branching theory.
Remarks: Quite often (such as will be the case for the buckling of thin
plates governed by the von Karman equations, or some modification
thereof) the functional equation G(λ, µ) = 0, in a Banach space B,
subject to the condition (1.16), will assume the particular form
Au − λu = 0, with A0 = 0 (1.17)
lim =0 (1.21)
v→0
v
© 2001 by Chapm
In most cases of importance in applications, the concept of Frechét
derivative f (u) is actually equivalent to the definition of the so-called
Gateaux derivative of f at u, namely, for v any other element (i.e., func-
tion) in the space in question, f (u) is that linear mapping which is
defined by
f (u + εv) − f (u)
f (u)v ≡ lim (1.22)
ε→0 ε
or, equivalently,
∂
f (u)v = [f (u + εv)]|ε=0 (1.23)
∂$
For our purposes in this treatise, we may think of the Frechét derivative
of a nonlinear mapping G, such as the one in (1.18), (which depends
on the parameter λ ), as being given by (1.22) or (1.23), e.g., at λ =
λ0 , ϕ = ϕ0 , the Frechét derivative of G(λ, u), as defined by (1.18), is
given by
∂
Gϕ (λ0 , ϕ0 )Ψ = [G(λ0 , ϕ0 + $Ψ)]|ε=0 (1.24)
∂ε
with Ψ ∈ B, as defined in (1.19). With the concept of Frechét derivative
in hand, we may state the following version of the implicit function the-
orem which, in essence, underlies all of bifurcation (or buckling) theory:
Theorem: Let G (parametrized by λ, as in (1.18)) be a Frechét dif-
ferentiable, nonlinear mapping, on some Banach (or Hilbert) space and
suppose that (λ0 , u0 ) is an (equilibrium) solution, i.e, G(λ0 , u0 ) = 0. If
the linear map Gu (λ0 , u0 ) has an inverse which is a bounded (i.e. con-
tinuous) mapping then locally, for |λ − λ0 | sufficiently small, there exists
a differentiable mapping (i.e. function) u(λ) such that G(λ, u(λ)) = 0.
Furthermore, in a sufficiently small neighborhood of (λ0 , u0 ), (λ, u(λ)) is
the only solution of G(λ, u) = 0
Remarks: Under the hypothesis (1.16), (λ, 0) is an equilibrium solu-
tion for all λ (note that the mapping G as given by (1.18) satisfies this
condition). Thus, if Gu (λ, 0) is a (linear) map with an inverse, which is
bounded, then u(λ) = 0, for all λ, is the only solution of G(λ, u(λ)) = 0
and one may conclude that no branching (or bifurcation) of solutions
can occur from the trivial solution u(λ) = 0.
As a consequence of the implicit function theorem, one may conclude
that branching of solutions can only occur if the linear mapping Gu ,
evaluated at a specific (equilibrium) solution (λc , u0 ) is singular, so that
Gu (λc , u0 ) does not have a bounded inverse; such a point (λc , u0 ) is then
a candidate for a branch (or bifurcation) point.
Example (Consider the compressed thin rod, as governed by G(λ, ϕ) =
0, G defined by (1.18), with ϕ in the space B given by (1.19). Using the
definition (1.23) of Frechét derivative, we easily find that for any ϕ0 , in
the space given by (1.19), the derivative Gϕ (λ, ϕ0 ) is the linear mapping
L, which acts on a function Ψ in B as follows:
d2 Ψ
LΨ = Gϕ (λ, ϕ0 )Ψ ≡ + λΨ cos ϕ0 (1.25)
ds2
so that, in particular, for the equilibrium solution (λ, ϕ0 = 0)
d2 Ψ
LΨ = Gϕ (λ, 0)Ψ ≡ + λΨ (1.26)
ds2
where, as Ψ ∈ B, Ψ must satisfy Ψ (0) = Ψ () = 0. But L is invertible
if and only if the only solution of LΨ = 0, for Ψ in B, is given by Ψ ≡ 0.
Thus, L is not invertible if there are values λ = λc such that LΨ = 0,
for Ψ ∈ B, has nontrivial solutions, i.e., if there exists λ = λc such that
2
d Ψ +λ Ψ=0
c
ds2 (1.27)
Ψ (0) = Ψ () = 0
has at least one nontrivial solution Ψ; such solutions (of 1.27) are, of
course, eigenfunctions, corresponding to the eigenvalue λ = λc , which
then becomes a candidate for being a branch (or bifurcation) point for
the boundary value problem (1.6a,b) associated with the compressed
(thin) rod. The eigenvalue problem (1.27) is the linearized problem
associated with (1.6a,b); in the parlance of buckling theory for elastic
structures, λ = λc , such that there exists a nontrivial solution Ψ = Ψc (s)
of (1.27), is a possible buckling “load” – actually the buckling load di-
vided by EI in this case – and the eigenfunction Ψc is the associated
buckling mode. In all cases of interest in classical buckling theory, the
linearization of an equilibrium equation (or set of equations), such as
(1.15), about an equilibrium solution (λ0 , u0 ), leads, in the manner de-
scribed above, to an eigenvalue-eigenfunction problem
for the buckling “loads” λc and the associated buckling modes vc ; this
will, of course, be the case for the von Karman equations which we derive
in §1.2.
© 2001 by Chapm
We remark that it is possible to construct equations (and systems of
equations) of the form G(λ, u) = 0, with G(λ, 0) = 0, for all λ, and
Gu (λ, 0) not invertible, for which branching does not take place from
each eigenvalue of the linear operator L = Gu (λ, 0). However, there are
general theorems of bifurcation theory, which apply, e.g., to branching
of solutions of the von Karman equations, and which guarantee that
eigenvalues of the linear operator L = Gu (λ, 0) are not only, by virtue
of the implicit function theorem, candidates for branch points, but are,
indeed, values of λ where bifurcation occurs; we state only one such
theorem: Let G(λ, u) = Au − λu, with A a nonlinear operator such that
A(0) = 0, and suppose that A is completely continuous (or compact).
o
Then if λ = 0 is an eigenvalue of odd multiplicity for the linearized
o
operator L = A (0), λ is a branch-point of the basic solution u(λ) ≡ 0
of the nonlinear problem.
In this section we will derive the classical von Karman equations which
govern the out-of-plane deflections of thin isotropic and orthotropic lin-
ear elastic plates as well as the linearized equations which mediate the
onset of buckling; the equations will be presented in both rectilinear
coordinates and in polar coordinates. We begin with a derivation, in
rectilinear coordinates, of the von Karman equations for linear elastic,
isotropic, (and then orthotropic), behavior.
1.2.1 Rectilinear Coordinates
We begin with the case of isotropic symmetry. Our derivation follows
that in the text by Troger and Steindl [66] in which the authors begin
with the derivation of the equilibrium equations for shallow shells, un-
dergoing moderately large derivations, and then specialize to the case
of the plate equations (which follow as a limiting case corresponding to
zero initial curvature); such an approach is of particular interest to us
in as much as we will want, later on, to look at imperfection buckling of
thin plates.
In Fig. 1.6, we depict the undeformed middle surface of a shallow
shell; this middle surface is represented in Cartesian coordinates by the
function w = W (x, y) and the displacements of the middle surface corre-
sponding to the x, y, z directions are denoted by u, v, and w respectively.
To derive the nonlinear shell equations we
(ii) Relate the strain tensor to the stress tensor by using the consti-
tutive law representing linear, isotropic elastic behavior.
(iii) Derive the equilibrium equations which relate the stress tensor
to the external loading.
We begin (see Fig. 1.7) by considering an infinitesimal volume element
dV = hdxdy of the shell. Let r be the position vector to a point P , in
the interior of the shell, which is located at a distance ζ from the middle
surface; then with ei , i = 1, 2, 3, the unit vectors along the coordinate
axes
r = x e1 +y e2 +(W + ζ) e3 (1.29)
© 2001 by Chapm
are the displacement components of P when it moves, after the displace-
ment, (which arises, e.g., because of the application of middle surface
forces at the boundary, or variations in temperature or moisture content)
to a point P̂ . The position vector r̂ to the point P̂ is then
r̂ = (x + u − ζw,x ) e1 +(y + v − ζw,y ) e2
(1.31)
+(W + ζ + w) e3
The components of the strain tensor for this deformation of the shell are
then obtained by computing the difference of the squares of the lengths
of the differential line elements d r and d r̂ ; a length computation based
on (1.31) and (see Fig. 1.6) r = x e1 +y e2 +W e3 yields
1 2
|d r̂ | − |d r | = εxx dx2 + 2εxy dxdy + εyy dy 2
2
(1.32)
2
where
1 2 2 2
$xx = u,x + u,x + v,x + w,x + W,x w,x
2
−ζw,xx + · · ·
1
$yy = v,y + u,2y +v,2y +w,2y + W,y w,y (1.33)
2
−ζw,yy + · · ·
γxy ≡ 2$xy = u,y +v,x +u,x v,y +w,x w,y
+W,x w,y +W,y w,x −2ζw,xy + · · ·
and terms of at least third order have been neglected. If we now take
into consideration the fact that, for stability problems connected with
thin-walled structures, the displacement w, which is orthogonal to the
middle surface, is much larger than the displacements u, v in the middle
surface, then the terms quadratic in u and v, in (1.33), may be neglected
in comparison with those in w; we thus obtain the (approximate) kine-
matical relations
1
$xx = u,x + w,2x +W,x w,x −ζw,xx
2
1 2
$yy = v,y + w,y +W,y w,y −ζw,yy
2 (1.34)
γ = 2$xy = u,y +v,x +w,x w,y +W,x w,y
xy
+W,y w,x −2ζw,xy
For the particular case in which the shallow shell is a plate whose middle
surface, prior to buckling, occupies a region in the x, y plane (so that
W ≡ 0) (1.34) reduces further to
1
$xx = u,x + w,2x −ζw,xx
2
1 2 (1.35)
$yy = v,y + w,y −ζw,yy
2
γxy ≡ 2$xy = u,y +v,x +w,x w,y −2ζw,xy
We now use the generalized Hooke’s law to describe the material be-
havior of the isotropic, linear elastic, shallow shell and we make the
usual assumption for thin-walled structures that σzz 0. The stress
components, σxz and σyz , which will appear in the equilibrium equa-
tions, do not contribute to the constitutive relationship as the corre-
sponding strains are zero (due to the Kirchhoff hypothesis). Thus, a
plane strain problem over the thickness h of the shell is obtained for
which the relevant constitutive equations may be written in the form
1
hE(u,x + w,2x +W,x w,x ) = Nx − νNy
2
1 2
hE(v,y + w,y +W,y w,y ) = Ny − νNx
2 (1.36)
Gh(u,y +v,x +w,x w,y +
W,x w,y +W,y w,x ) = Nxy
are the averaged stresses over the shell thickness h, which is assumed to
be small. The bending moment Mx is defined to be
h/2
Mx = σxx ξdξ (1.38)
−h/2
© 2001 by Chapm
and by using (1.34) and (1.36) this may be computed to be
h/2
E
Mx = ($xx + ν$yy )ξdξ
−h/2 1 − ν2 (1.39)
= −K(w,xx +νw,yy )
with the plate stiffness K given by
Eh3
K= (1.40)
12(1 − ν 2 )
With analogous definitions (and computations) for the bending moments
My and Mxy we find that
My = −K(w,yy +νw,xx )
(1.41)
Mxy = −(1 − ν)Kw,xy
The next step consists of introducing the Airy stress function Φ(x, y),
which is defined so as to satisfy
for the two unknowns, the (extra) deflection w(x, y) and the Airy stress
function Φ(x, y), where ∆2 denotes the biharmonic operator in rectilin-
ear Cartesian coordinates, i.e.,
∂4w ∂4w ∂4w
∆2 w = + 2 + (1.48)
∂x4 ∂x2 ∂y 2 ∂y 4
To obtain a second partial differential equation for w(x, y) and Φ(x, y),
we make use of the identity
into which we substitute, from the constitutive relations (1.36), for the
displacement derivatives u,x , v,y , and u,y +v,x . One then makes use
of the definition of the Airy function Φ to replace the stress resultants
Nx , Ny , and Nxy ; there results the following equation
∆2 Φ = Eh (w2 ,xy −w,xx w,yy )
(1.50)
+2W,xy w,xy −W,xx w,yy −W,yy w,xx
© 2001 by Chapm
If we introduce the nonlinear (bracket) differential operator by
so that
then the system consisting of (1.47) and (1.50) can be written in the
more compact form
1 2 1
∆ Φ = − [w, w] − [W, w] (1.52b)
Eh 2
In particular, for the deflection of a thin plate, which in its undeflected
configuration occupies a domain in the x, y plane, so that W ≡ 0, (1.52a),
(1.52b) reduce to the von Karman plate equations for an isotropic linear
elastic material, namely,
Remarks: The curvatures of the plate in planes parallel to the (x, z) and
(y, z) planes, are usually denoted by κx and κy , respectively, while the
twisting curvature is denoted by κxy . Strictly speaking, the curvature
κx , e.g., is given by
∂2w
− 2
κx = ∂x (1.55)
2 32
∂w
1+
∂x
where the minus sign is introduced so that an increase in the bending
moment Mx results in an increase in κx . As w,2x is assumed to be small,
even within the context of the nonlinear theory of shallow shells, the
curvatures are usually approximated by, e.g. κx = −w,xx , in which case
for linear isotropic response
κx = (Mx − νMy )/ (1 − ν 2 )K
κ = (My − νMx )/ (1 − ν 2 )K (1.56)
y
κxy = (1 + ν)Mxy / {(1 − ν)K}
© 2001 by Chapm
∂ 1 2
u,x +v,y + w,x +w,2y
∂x 2 (1.59b)
1−ν
+ (2 u + w,y 2 w) = 0
1+ν
and
∂ 1 2 2
u,x +v,y + w,x +w,y (1.59c)
∂y 2
1−ν
+ (2 v + w,y 2 w) = 0
1+ν
where 2 is the usual Laplacian operator.
We now proceed (still in rectilinear coordinates) to derive the appro-
priate form of the von Karman plate equations for the case of a thin plate
which exhibits linear elastic behavior and has (rectilinear) orthotropic
symmetry. Thus, consider an orthotropic thin plate for which the x
and y axes coincide with the principal directions of elasticity; then the
constitutive equations have the form
σxx c11 c12 0 $xx − β1 ∆H
σyy = c21 c22 0 $yy − β2 ∆H (1.60)
σxy 0 0 c66 γxy
where we have included in the strain components possible hygroexpan-
sive strains βi ∆H, i = 1, 2 where the βi are the hygroexpansive coef-
ficients and ∆H represents a humidity change; alternatively, we could
replace the βi ∆H by thermal strains αi ∆T with the αi thermal expan-
sion coefficients and ∆T a change in temperature. In (1.60)
c11 = E1 /(1 − ν12 ν21 )
c12 = E2 ν21 /(1 − ν12 ν21 )
c21 = E1 ν12 /(1 − ν12 ν21 ) (1.61)
c22 = E 2 /(1 − ν ν
12 21 )
c66 = G12
with E1 ν12 = E2 ν21 so that c12 = c21 . In (1.61), E1 , E2 , ν12 , ν21 , and
G12 are, respectively, the Young’s moduli, Poisson’s ratios, and shear
modulus associated with the principal directions. The constants
h3
Dij = cij (1.62)
12
are the associated rigidities (or stiffness ratios) of the orthotropic plate,
specifically,
E1 h3 E2 h3
D11 = , D22 = (1.63)
12(1 − ν12 ν21 ) 12(1 − ν12 ν21 )
are the bending rigidities about the x and y axes, respectively, while
G12 h3
D66 = (1.64)
12
is the twisting rigidity. The ratios D12 /D22 , D12 /D11 are often called
reduced Poisson’s ratios. For the thin orthotropic plate under consid-
eration, the strains $xx , $yy , and γxy , the averaged stresses (or stress
resultants) Nx , Ny , and Nxy , and the bending moments Mx , My , and
Mxy are still given by (1.35), (1.37), (1.38), and the relevant expressions
for My and Mxy , which are analogous to (1.38). Thus, with
σxx = c11 ($xx − β1 ∆H) + c12 ($yy − β2 ∆H)
σyy = c21 ($xx − β1 ∆H) + c22 ($yy − β2 ∆H)
σxy = c66 γxy
we have
1
σxx = c11 (u,x + w,2x −ζw,xx ) (1.65a)
2
1
+c12 (v,y + w,2y −ζw,yy )
2
−(c11 β1 ∆H + c12 β2 ∆H)
1
σyy = c21 (u,x + w,2x −ζw,xx ) (1.65b)
2
1
+c22 (v,y + w,2y −ζw,yy )
2
−(c21 β1 ∆H + c22 β2 ∆H)
and
1
τxy = c66 (u,y +v,x ) + w,x w,y −2ζw,xy (1.65c)
2
© 2001 by Chapm
moments Mx , My , and Mxy to be
h/2
1 2
Mx = c11 ξ u,x + w,x − ξ w,xx dξ
2
2
−h/2
h/2
1 2
+c12 ξ v,y + w,y − ξ w,yy dξ
2
(1.66a)
−h/2 2
h/2
−
ξ (c12 β1 ∆H + c22 β2 ∆H) dξ
−h/2
h/2
1 2
My = c21 ξ u,x + w,x − ξ w,xx dξ
2
2
−h/2
h/2
1 2
+c22 ξ v,y + w,y − ξ w,yy dξ
2
(1.66b)
−h/2 2
h/2
−
ξ (c11 β1 ∆H + c22 β2 ∆H) dξ
−h/2
and
Mxy =
h/2
1 (1.66c)
c66 ξ [u,y +v,x ] + w,x w,y − 2ξ 2 w,xy dξ
−h/2 2
and
h2
ξ3
Mxy = −2c66 · w,xy
3 −h (1.67c)
2
= −2D66 w,xy
The corresponding modification of the first von Karman equation for the
deflection of a thin, linearly elastic, orthotropic plate, when there exists
an initial deflection z = w0 (x, y), is easily obtained from (1.45) and
(1.67a,b,c) by setting W = w0 and defining, as in (1.57), w̃ = w + w0 .
To obtain the appropriate modification of the second von Karman equa-
tion (1.53b) for the case of a linearly elastic, thin, orthotropic plate, we
© 2001 by Chapm
once again begin with the identity (1.49). We then use the constitutive
relations (1.65 a,b,c) to compute the average stresses (or stress resul-
tants) Nx , Ny , and Nxy , introduce the Airy function through (1.46),
and solve the resulting equations for u,x , v,y , and u,y +v,x ; these, in
turn, are substituted into (1.49) and there results the partial differential
equation
1 1 1 2ν12
Φ,yyyy + − Φ,xxyy (1.69)
E1 h h G12 E2
1 1
+ Φ,xxxx = − [w, w]
E2 h 2
if one makes use of the fact that E2 ν21 = E1 ν12 .
The complete set of von Karman equations in rectilinear Cartesian
coordinates thus consists of (1.68) and (1.69). In an isotropic plate
E1 = E2 = E, G12 = G = E/2(1 + ν), with ν12 = ν21 = ν, and
the system of equations (1.68), (1.69) reduces to the system (1.53a,b),
with t ≡ 0, where K is the common value of the principal rigidities
D1 = D11 , D2 = D22 , and D3 = D12 ν12 + 2D66 .
Remarks: The system of equations (1.68), (1.69), as well as their spe-
cializations to (1.53 a,b), for the case of linear, isotropic, elastic response,
may be obtained from a variational (minimum energy) principle based
on the potential energy
1 h/2
U = c11 ($xx − β1 ∆H)2 (1.70)
2 A −h/2
+2c12 ($xx − β1 ∆H)($yy − β2 ∆H)
+c22 ($yy − β2 ∆H)2 + c66 γxy
2
} dzdA
where the outer integral is computed over the area A occupied by the
plate. In (1.70), or its equivalent for the case where the plate ex-
hibits isotropic response, we must first substitute from (1.35) in order
to express the integrand as a polynomial expression in the displacement
derivatives.
Remarks: In the case of very thin plates, which may have deflec-
tions many times their thickness, the resistance of the plate to bending
can, often, be neglected; this amounts, in the case of a plate exhibiting
isotropic response, to taking the stiffness K = 0, in which case the prob-
lem reduces to one of finding the deflection of a flexible membrane. The
equations which apply in this case were obtained by A. Föppl [69] and
turn out, of course, to be just the von Karman equations (1.53a,b) with
K set equal to zero.
∂ur 1 ∂w 2 ∂2w
err = + ( ) − ζ
∂r 2 ∂r ∂r2
ur 1 ∂uθ 1 ∂w 2 1 ∂w 1 ∂2w
eθθ = + + 2( ) − ζ( + 2 2)
r r ∂θ 2r ∂θ r ∂r r ∂θ (1.71)
e = ∂u u 1 ∂u 1 ∂w ∂w
θ
−
θ
+ (
r
)+ ( )( )
rθ
∂r r r ∂θ r ∂r ∂θ
1 ∂2w
1 ∂w
−2ζ( − )
r ∂r∂θ r2 ∂θ
With the components of the stress tensor σrr , σrθ , σθθ , σzz , σrz and
σθz as shown in Fig. 1.10, (σrθ = σθr ), (σrz = σzr ), and Fr , Fθ the com-
ponents of the applied body force in the radial and tangential directions,
© 2001 by Chapm
the equilibrium equations are
∂σ 1 ∂σrθ σrr − σθθ ∂σrz
rr
+ + + + Fr = 0
∂r r ∂θ r ∂z
(1.72a)
∂σrθ + 1 ∂σθθ + 2 σ + ∂σθz + F = 0
rθ θ
∂r r ∂θ r ∂z
and
1 1
σrr w,r +w,rr + σθθ w, θθ
r r2
2 1 ∂σrr
+σrθ w,rθ + σrz + w,r
r r ∂r
∂σθθ 1 ∂σrθ 1
+ w,θ + w,θ (1.72b)
∂θ r2 ∂r r
∂σrθ 1 ∂σrz ∂σrz
w,r + + w,r
∂θ r ∂r ∂z
1 ∂σθz ∂σθz 1
+ + w,θ + Fz = 0
r ∂θ ∂z r
σrr = σxx cos2 θ + σyy sin2 θ + 2σxy sin θ cos θ
σ = σxx sin2 θ + σyy cos2 θ − 2σxy sin θ cos θ (1.73)
θθ
σrθ = (σyy − σxx ) sin θ cos θ + σxy cos2 θ − sin2 θ
with analogous transformation formulae for σrz and σθz . If we set, for
the deflection w and the Airy stress function Φ, w̄ (r, θ) = w (r cos θ,
r sin θ), Φ̄ (r, θ) = Φ (r cos θ, r sin θ), and then drop the superimposed
bars in the polar coordinate system, it can be shown directly that the
stress resultants (or averaged stresses) Nr , Nθ , and Nrθ are given in
terms of Φ by
1 1
Nr = Φ,r + 2 Φ,θθ
r r
Nθ = Φ,rr (1.74)
Nrθ = 1 Φ,θ − 1 Φ,rθ
r2 r
while the operators ∆2 and [ , ] are given by
2 1
∆2 w = w,rrrr + w,rrr − 2 w,rr
r r
2 1 2
+ w,rrθθ + 3 w,r − 3 w,rθθ (1.75a)
r 2 r r
1 4
+ w,θθθθ + 4 w,θθ
r4 r
and
1 1
[w, Φ] = w,rr Φ,r + 2 Φ,θθ
r r
1 1
+ · w,r + 2 w,θθ Φ,rr
r r
1 1 1 1
−2 w,rθ − 2 w,θ Φ,rθ − 2 Φ,θ
r r r r
1 1
[w, Φ] = Nr w,rr −2Nrθ w,θ − w,rθ
r2 r
1 1
+ Nθ · w,r + 2 w,θθ (1.75b)
r r
2 1 1
∆2 w = w,rrrr + w,rrr − 2 w,rr + 3 w,r (1.76)
r r r
and
1
[w, Φ] = Nr w,rr +Nθ w,r (1.77)
r
Thus, for the von Karman equations for a thin, linearly elastic, isotropic
plate, in polar coordinates, we have (with t ≡ 0):
© 2001 by Chapm
2 1 2
K w,rrrr + w,rrr − 2 w,rr + 2 w,rrθθ
r r r
1 2 1 4
+ 3 w,r − 3 w,rθθ + 4 w,θθθθ + 4 w,θθ
r r r r
1 1
= w,rr Φ,r + 2 Φ,θθ (1.78)
r r
1 1
+ w,r + 2 w,θθ Φ,rr
r r
1 1 1 1
−2 w,rθ − 2 w,θ Φ,rθ − 2 Φ,θ
r r r r
and
1
2 1 2
Φ,rrrr + Φ,rrr − 2 Φ,rr + 2 Φ,rrθθ
Eh r r r
1 2 1 4
+ 3 Φ,r − 3 Φ,rθθ + 4 Φ,θθθθ + 4 Φ,θθ (1.79)
r r r r
2
1 1 1 1
= − w,rr w,r + 2 w,θθ − w,rθ − 2 w,θ
r r r r
Eh 0
Nr = err + νe0θθ
1−ν 2
2
Eh dur 1 dw ur (1.82a)
= + +ν
1 − ν 2 dr 2 dr r
and
Eh 0
Nθ = eθθ + νe0rr
1−ν 2
2
Eh ur dur ν dw (1.82b)
= +ν +
1 − ν2 r dr 2 dr
Remarks: In lieu of (1.78), a useful (equivalent) form for the first von
Karman equation (especially for our later discussion of the buckling of
annular plates) is
1 1
K∆2 w = Nr w,rr −2Nrθ w, θ − w,rθ (1.83)
r2 r
1 1
+Nθ w,r + 2 w,θθ
r r
If σrr , σrθ , and σθθ are independent of the variable z, in the plate, then
Nr , Nrθ , and Nθ in (1.83), as well as in all the other expressions prior
to (1.83), where these stress resultants appear, may be replaced, respec-
tively, by hσrr , hσrθ , and hσθθ .
Remarks: It is easily seen that, for the case of an axially symmetric
deformation of the plate, the relevant equations, i.e., (1.80), (1.81) may
be rewritten in the form
1 d d 1 d dw
K r r =
r dr dr r dr dr
(1.84)
1 dΦ d2 w 1 d2 Φ dw
+
r dr dr2 r dr2 dr
© 2001 by Chapm
and
1 1 d d 1 d dw 1 dw d2 w
r r =− (1.85)
Eh r dr dr r dr dr r dr dr2
Remarks: From the structure of the operators ∆2 and [ , ], in the polar
coordinate system, i.e., (1.75a), (1.75b), it is clear that a troublesome
singularity arises at r = 0; the boundary conditions which must be
imposed to deal with this difficulty at r = 0 will be discussed in the next
subsection.
The quantities
d2 w 1 dw
κr = 2
, κθ = , (1.86)
dr r dr
for the case of radially symmetric deformations of a plate, are the middle-
surface curvatures. If the plate is circular, with radius R, then the
strain-energy of bending for the isotropic, linearly elastic plate is
1
VB = K (κ2r + 2νκr κθ + κ2θ )dA
2
R A 2
(1.87a)
w, r 1
= πK w,2rr +2νw,rr · + w,r rdr
0 r r
W (u, w, P ) ≡ VB + VS + VL (1.89)
We now turn to the equations, in polar coordinates, for a linearly
elastic, orthotropic, body with cylindrical anisotropy; in this case, there
are three planes of elastic symmetry, one of which is normal to the axis
of anisotropy, the second of which passes through that axis, and the
third of which is orthogonal to the first two. For a plate, we choose the
first plane of elastic symmetry to be parallel to the middle plane of the
plate; in this case the constitutive relations assume the form
1 νθ
err = σrr − σθθ
E E
r θ
νr 1
eθθ = − σrr + σθθ (1.90)
Er Eθ
1
γrθ = σrθ
Grθ
with Er , Eθ being the Young’s moduli for tension (or compression) in
the radial and tangential directions, respectively, νr and νθ the corre-
sponding Poisson’s (principal) ratios, and Grθ the shear modulus which
characterizes the change of angle between the directions r and θ. As
Er νθ = Eθ νr , the constitutive equations (1.90) can be recast in the
form
1
rr
e = (σ − ν r θθ
σ )
E
rr
r
1
eθθ = (σθθ − νθ σrr ) (1.91)
Eθ
1
γrθ = σrθ
Grθ
so that
σrr =
Er νr E θ
err + eθθ
1 − ν r ν θ 1 − ν r ν θ
E r νθ Eθ (1.92)
σθθ = err + eθθ
1 − ν r ν θ 1 − ν r ν θ
σrθ = Grθ γrθ
One may compute the strains err , eθθ , and γrθ by using (1.71) and, then,
by employing (1.92), the stresses σrr , σθθ , and σrθ . The equations of
equilibrium which apply in this situation are (1.72 a,b) and these then
produce stress components σrz and σθz . The stresses in the cylindri-
cally orthotropic plate then lead to stress resultants Nr , Nθ , and Nrθ
and bending and twisting moments Mr , Mθ , and Mrθ . By employing
© 2001 by Chapm
straightforward calculations, we are led to the following results for a
cylindrically orthotropic plate:
1 1
M r = −D r w, rr +ν θ w, r + w,θθ
r r2
1 1 (1.93)
Mθ = −Dθ νr w,rr + w,r + 2 w,θθ
r r
Mrθ = Mθr = −2D̃rθ w ,rθ
r
and
Er h ∂ur 1 2
N r = + w
1 − νr νθ ∂r 2 ,r
νr Eθ h ur 1 ∂uθ 1 2
+ + + 2 w,θ
1 − νr νθ r r ∂θ 2r
νθ E r h ∂ur 1 2
Nθ = + w,r (1.94)
1 − ν ν ∂r 2
r θ
Eθ ur 1 ∂uθ 1 2
+ + + 2 w,θ
1 − ν ν r r ∂θ 2r
r θ
∂ur ur 1 ∂ur 1
Nrθ = Grθ h − + + w,r w,θ
∂r r r ∂θ r
E r h3 E θ h3
Dr = , Dθ = (1.95)
12(1 − νr νθ ) 12(1 − νr νθ )
and
Grθ · h3
D̃rθ = (1.96)
12
while
Drθ = Dr νθ + 2D̃rθ (1.97)
Using the expressions for Mr , Mθ , Mrθ in (1.94), those for Nr , Nθ , and
Nrθ in (1.74), and a compatibility equation for the displacements in po-
lar coordinates, we find the following form of the von Karman equations
for a linearly elastic, thin plate exhibiting cylindrically orthotropic sym-
metry, (where we have once again introduced the Airy stress function
through the relations (1.74)):
1 1
Dr w,rrrr +2Drθ 2
w,rrθθ +Dθ 4 w,θθθθ (1.98)
r r
1 1 1
+2Dr w,rrr −2Drθ 3 w,rθθ −Dθ 2 w,rr
r r r
1 1
+2(Dθ + Drθ ) 4 w,θθ +Dθ 3 w,r
r r
1 1 1 1
= Φ,r + 2 Φ,θθ w,rr +Φ,rr ( w,r + 2 w,θθ )
r r r r
1 1 1 1
+2 Φ,θ − Φ,rθ w,rθ − w,θ
r2 r r r2
and
1 1 2νr 1
Φ,rrrr + − Φ,rrθθ (1.99)
Eθ Grθ Er r2
1 1 2 1
+ 4
Φ,θθθθ + Φ,rrr
Er r Eθ r
1 2νr 1 1 1
− − Φ,rθθ − Φ,rr
Grθ Er r 3 Er r 2
1 − νr 1 1 1 1
+ 2 + 4
Φ,θθ + Φ,r
Er Grθ r Er r 3
1 1
= −h w,rr w,r + 2 w,θθ
r r
2
1 1
− w,rθ − 2 w,θ
r r
Equations (1.98), (1.99) may also be obtained directly from many sources
in the literature, e.g., the paper by Uthgenannt and Brand [70].
Equations (1.98), (1.99), which govern the general deflections of a
cylindrically orthotropic, linearly elastic, thin plate reduce to those which
govern the deflections, in polar coordinates, of an isotropic plate, i.e.
(1.78), (1.79) when
Eh3
Dr = Dθ = Drθ = K ≡ (1.100a)
12(1 − ν 2 )
© 2001 by Chapm
and
Er = Eθ = E, νr = νθ = ν (1.100b)
σrr Er 1 νθ err
= (1.104a)
σθθ 1 − νr νθ νθ β eθθ
and
err = exx cos2 θ + eyy sin2 θ + γxy cos θ sin θ
eθθ = exx sin2 θ + eyy cos2 θ − γxy cos θ sin θ (1.105)
γrθ = 2(eyy − exx ) cos θ sin θ + γxy (cos2 θ − sin2 θ)
1 1 1 1
(rMr ),rr + 2 Mθ,θθ − Mθ,r + Mrθ,rθ + Nr w,rr (1.106)
r r r r
w
1 1
+Nθ w,r + 2 w,θθ + 2Nrθ =0
r r r ,rθ
where the stress resultants Nr , Nθ , and Nrθ are, once again, given by
(1.74) in terms of the Airy function Φ(r, θ); we note that the sum of
the last three terms in (1.106) is (again) identical with the right-hand
side of (1.75b), i.e., with [w, Φ]. From the work in [71], we deduce the
following expressions for the bending moments (which may, of course,
be obtained by directly transforming the expressions in (1.67a), (1.67b)
and (1.67c) into polar coordinates):
© 2001 by Chapm
1 1
Mr = −D̃1 w,rr −D̃12 w,θθ + w,r
r2 r (1.107a)
1
−2D̃16 w ,rθ
r
1 1
Mθ = −D̃12 w,rr −D̃2 2 w,θθ + w,r
r r (1.107b)
1
−2D̃26 w ,rθ
r
1 1
Mrθ = −D̃16 w,rr −D̃26 w,θθ + w,r
r2 r
(1.107c)
1
−2D̃6 w ,rθ
r
where,
D̃1 = D1 cos4 θ + D3 cos2 θ sin2 θ + D2 sin4 θ
D̃2 = D1 sin4 θ + D3 cos2 θ sin2 θ + D2 cos4 θ
D̃12 = ν1 D2 + (D1 + D2 − 2D3 ) cos2 θ sin2 θ
(1.108)
D̃6 = D66 + (D1 + D2 − 2D3 ) cos2 θ sin2 θ
$
D̃ = (D − D ) sin2
θ − (D − D ) cos 2
θ cos θ sin θ
16
$
2 3 1 3
D̃26 = (D2 − D3 ) cos2 θ − (D1 − D3 ) sin2 θ cos θ sin θ
© 2001 by Chapm
rectilinearly orthotropic plate appear to have been derived anywhere else
in the literature. The calculation, however, which is required to obtain
the equation which complements (1.109) may be carried out in one of
two ways: first of all, by transforming the three fourth order partial
derivatives Φ,xxxx , Φ,xxyy , and Φ,yyyy which appear on the left-hand
side of (1.69); the right-hand side of (1.69), in polar coordinates, will be
identical with the right-hand side of (1.79).
Alternatively, to obtain the form of the second of the von Karman
equations, we may rewrite the constitutive relations for a linearly elastic,
rectilinearly orthotropic material, i.e.,
1 ν21
exx = σxx − σyy
E E2
1
ν12 1
eyy = − σxx + σyy (1.111)
E1 E2
1
γxy = σxy
G12
(where we have, for now, not considered the presence of possible hy-
groexpansive or hygrothermal strains) in the polar coordinate form
err = a11 σrr + a12 σθθ + a13 τrθ
eθθ = a21 σrr + a22 σθθ + a23 τrθ (1.112)
γrθ = a31 σrr + a32 σθθ + a33 τrθ
where the aij = aij (θ), in contrast to the case of a cylindrically or-
thotropic material, i.e. (1.91), in which the constitutive “coefficients”
are θ-independent. The strain components in (1.112) are given by the
relations (1.71) in terms of the displacements ur , uθ , and w, where
h h
− < ζ < ; if we think in terms of averaging the constitutive re-
2 2
lations (1.112) over the thickness of the plate we may, in essence, ignore
the expressions involving ζ in (1.71). For the in-plane stress distribution
(prebuckling), w = 0, in which case
r2
L(ur , uθ ) = − [w, w] (1.115)
2
where
2
1 1 1 1
[w, w] = 2 w,rr w,r + 2 w,θθ − w,rθ − 2 w,θ (1.116)
r r r r
The essential idea behind the derivation of the second of the von
Karman equations is to compute the polar coordinate form (1.112) of
the constitutive relations (1.111), substitute (1.112) into (1.115), and
then set
1 ∂Φ 1 ∂2Φ
σrr = + 2 2
r ∂r r ∂θ
∂2Φ
σθθ = (1.117)
∂r2
∂2 1
τrθ = − Φ
∂r∂θ r
in the resulting equation. We begin by recalling the transformation
(1.105) of the principal strains exx , eyy , and γxy to the polar coordinate
system and the analogous transformation of the principal stresses, i.e.,
σrr = σxx cos2 θ + σyy sin2 θ + 2σxy sin θ cos θ
σθθ = σxx sin2 θ + σyy cos2 θ − 2σxy sin θ cos θ (1.118)
σrθ = (σyy − σxx ) sin θ cos θ + σxy cos2 θ − sin2 θ
Using the transformation (1.105) of the principal strains, in conjunction
with the constitutive relations (1.111), we find that
1 ν12
err = cos θ −
2 2
sin θ σxx
E E1
1
1 ν21
+ sin θ −
2 2
cos θ σyy (1.119)
E2 E2
sin 2θ
+ σxy
G12
© 2001 by Chapm
1 ν12
eθθ = sin2 θ − cos2 θ σxx
E E1
1
1 ν21
+ cos θ −
2 2
sin θ σyy (1.120)
E2 E2
sin 2θ
− σxy
G12
and
(1 + ν12 ) (1 + ν21 )
erθ = − sin 2θ · σxx + sin 2θ · σyy
2 2
(1.121)
cos 2θ
+ · σxy
G12
By solving the relations in (1.118) for σxx , σyy , and σxy in terms of
σrr , σθθ , and σrθ and then substituting these results for σxx , σyy , and
σxy into (1.119), (1.120), and (1.121), and simplifying, we obtain (1.112)
with
cos2 θ̄ sin2 θ̄
a11 = (cos2 θ̄ − ν12 sin2 θ̄) + (sin2 θ̄ − ν21 cos2 θ̄)
E1 E2 (1.122a)
1
sin2 2θ̄
4G12
sin2 θ̄ cos2 θ̄
a12 = (cos2 θ̄ − ν12 sin2 θ̄) + (sin2 θ̄ − ν21 cos2 θ̄)
E1 E2
(1.122b)
1
− sin2 2θ̄
4G12
1 1
a13 = sin 2θ̄ (ν12 sin2 θ̄ − cos2 θ̄) + (sin2 θ̄ − ν21 cos2 θ̄)
E1 E2
(1.122c)
cos 2θ̄
+
2G12
cos2 θ̄ sin2 θ̄
a21 = (sin2 θ̄ − ν12 cos2 θ̄) + (cos2 θ̄ − ν21 sin2 θ̄)
E1 E2
(1.122d)
sin2 2θ̄
−
4G12
sin2 θ̄ cos2 θ̄
a22 = (sin2 θ̄ − ν12 cos2 θ̄) + (cos2 θ̄ − ν21 sin2 θ̄)
E1 E2 (1.122e)
sin2 2θ̄
+
4G12
1 1
a23 = (ν12 cos2 θ̄ − sin2 θ̄) + (cos2 θ̄ − ν21 sin2 θ̄)
E1 E2
(1.122f)
cos 2θ̄
−
2G12
(1 + ν21 ) (1 + ν12 )
a31 = sin θ̄ cos θ̄ sin2 θ̄ − cos2 θ̄
E2 E1
(1.122g)
cos 2θ̄
+
2G12
(1 + ν21 ) (1 + ν12 )
a32 = sin θ̄ cos θ̄ cos2 θ̄ − sin2 θ̄
E2 E1
(1.122h)
cos 2θ
−
2G12
(1 + ν12 ) (1 + ν21 )
a33 = sin 2θ̄ sin θ̄ cos θ̄ +
E1 E2
(1.122i)
cos2 2θ̄
+
2G12
To summarize, the second of the von Karman equations in polar co-
ordinates, for a linearly elastic, thin plate, possessing rectilinear or-
thotropic symmetry, is obtained by first inserting the aij (θ), as given
by (1.122a)-(1.122i) into the constitutive relations
1 1
err = a11 Φ,r + 2 Φ,θθ + a12 Φ,rr
r r
(1.123a)
1
−a13 Φ
r ,rθ
1 1
eθθ = a21 Φ,r + 2 Φ,rr + a22 Φ,rr
r r
(1.123b)
1
−a23 Φ
r ,rθ
© 2001 by Chapm
1 1 1
γrθ = a31 Φ,r + 2 Φ,rr + a32 Φ,rr − a33 Φ (1.123c)
r r r ,rθ
∂ 2 err ∂2 ∂2
+ r (reθθ ) − (rγrθ )
∂θ2 ∂r2 ∂r∂θ
2
∂err 1 1 1 1
−r = − w,rr w,r + 2 w,θθ − w,rθ − 2 w,θ
∂r r r r r
(1.124)
A comprehensive study of postbuckling for rectilinearly orthotropic
plates with circular geometries will not be attempted in the present
work; therefore, we will forgo carrying out the remainder of the deriva-
tion of the second of the von Karman equations for this situation leaving,
instead, the straightforward calculations as an exercise for the reader.
The initial buckling of rectilinearly orthotropic (circular) annular plates
will be treated in Chapter 4 and the in-plane displacement differential
equations associated with the buckling of rectilinearly orthotropic circu-
lar plates will be obtained in Chapter 3.
Along the edge of the plate, at r = a, we prescribe, in general, the
radial and tangential components pr (θ) and pθ (θ), respectively, of the
applied traction where
$ '
pr (θ) = h σxx cos2 θ + σyy sin θ cos θ + σxy sin 2θ 'r=a
(1.125)
pθ (θ) = h [(σyy − σxx ) sin θ cos θ − σxy cos 2θ]|r=a
i.e.
Ω = (x, y) |x2 + y 2 ≤ R2 (1.127a)
so that
∂Ω = (x, y)|x2 + y 2 = R2 (1.127b)
© 2001 by Chapm
and f = f (r, θ) is defined on Ω, with first partial derivatives continuous
up to the boundary ∂Ω, then
∂f ∂f 1 ∂f
= f,r and = (1.128)
∂n ∂s r ∂θ
the latter result being a consequence of the fact that s = rθ, so that
∂ ∂ ∂s
= . Although there are many different types of (physical)
∂θ ∂s ∂θ
boundary conditions which can be considered for the deflection w of
a thin plate, three types are most prevalent in the literature on plate
buckling: clamped edges, simply supported edges, and free edges. With
respect to the general geometry shown in Fig. 1.11, and irrespective of
whether we are dealing with isotropic or orthotropic symmetry, these
three sets of boundary conditions lead to the following requirements on
∂Ω:
∂w
(i) ∂Ω is clamped: w = 0 and = 0, on ∂Ω (1.129a)
∂n
(iii) ∂Ω is Free: Mn = 0
∂Mtn (1.129c)
and Qn + = 0, on ∂Ω
∂s
where Mn is the bending moment on ∂Ω in the direction normal to
∂Ω, Mtn is the twisting moment on ∂Ω, with respect to the tangential
and normal directions on ∂Ω, and Qn is the shearing force associated
with the direction normal to ∂Ω. We now specify the conditions in
(1.129 a,b,c) for the cases of isotropic and orthotropic symmetry in both
rectilinear and circular geometries.
and
w,x (0, y) = 0, 0≤y≤b
w,x (a, y) = 0, 0≤y≤b
(1.130b)
w,y (x, 0) = 0, 0≤x≤a
w,y (x, b) = 0, 0≤x≤a
∂w(x, y)
where w,x (0, y) ≡ |x=0 , etc.
∂x
(ii) ∂Ω is Simply Supported
In this case, as a consequence of (1.129b), we have, first of all, the
conditions (1.130a), because w = 0 on ∂Ω. The condition Mn = 0
translates, in this case, into Mx = 0, for x = 0, x = a, 0 ≤ y ≤ b, and
My = 0 for y = 0, y = b, 0 ≤ x ≤ a, or, in view of the expressions for
the bending moments Mx , My in (1.39) and (1.41), respectively,
w,xx +νw,yy |x=0 = 0, 0 ≤ y ≤ b
w,xx +νw,yy |x=a = 0, 0 ≤ y ≤ b
(1.131)
w, +νw, | = 0, 0 ≤ x ≤ a
yy xx y=0
w,yy +νw,xx |y=b = 0, 0 ≤ x ≤ a
© 2001 by Chapm
(iii) ∂Ω is Free
If all four edges of the rectangular plate are free then, first of all, as
a consequence of the condition Mn = 0, on ∂Ω, in (1.129c), the four
relations in (1.131) must hold. The second relation in (1.129c) becomes
∂Mxy
Qyz + |y=0 = 0, 0≤x≤a
∂x
∂Mxy
|y=b = 0, 0≤x≤a
Qyz +
∂x
(1.134)
∂Myx
Qxz + |x=0 = 0, 0≤y≤b
∂y
Qxz + ∂Myx |x=a = 0, 0≤y≤b
∂y
or, if we employ the moment equilibrium equations (1.44):
My,y + 2Mxy,x |y=0 = 0, 0 ≤ x ≤ a
My,y + 2Mxy,x |y=b = 0, 0 ≤ x ≤ a
(1.135)
Mx,x + 2Myx,y |x=0 = 0, 0 ≤ y ≤ b
Mx,x + 2Myx,y |x=a = 0, 0 ≤ y ≤ b
However, by virtue of (1.41),
Thus, if all four edges of the plate were free, we would have
w,yyy +(2 − ν)w,xxy |y=0 = 0, 0 ≤ x ≤ a
w,yyy +(2 − ν)w,xxy |y=b = 0, 0 ≤ x ≤ a
(1.137)
w,xxx +(2 − ν)w,xyy |x=0 = 0, 0 ≤ y ≤ b
w,xxx +(2 − ν)w,xyy |x=0 = 0, 0 ≤ y ≤ b
Remarks: In any actual problem that one would want to consider
with respect to buckling or postbuckling of a (rectangular) thin, linearly
elastic, isotropic plate, there would usually be a mixing of the various
boundary conditions delineated above along parallel pairs of edges. For
example, if the edges along x = 0, x = a, 0 ≤ y ≤ b, were clamped while
those along y = 0, y = b, 0 ≤ x ≤ a, were simply supported then, by
(1.130a,b) and (1.131), the full set of boundary conditions would read
as follows:
for 0 ≤ y ≤ b, and
w(x, 0) = w(x, b) = 0, 0 ≤ x ≤ a
w,yy +νw,xx |y=0 = 0, 0 ≤ x ≤ a (1.138b)
w,yy +νw,xx |y=b = 0, 0 ≤ x ≤ a
© 2001 by Chapm
the circle of radius R2 = b. If ∂Ω1 is clamped, then as a consequence of
(1.129a)
w(Ri , θ) = 0, w,r (Ri , θ) = 0, 0 < θ ≤ 2π (1.139)
where
1 1
∆w = w,rr + w,r + 2 w,θθ (1.142)
r r
Remarks: As was the case for a rectangular plate, for a thin, linearly
elastic, isotropic, annular plate, one may mix and match the various
sets of boundary conditions delineated above, e.g., if the outer radius at
r = b is clamped, while the inner radius at r = a is free, the boundary
conditions would read as follows:
1−ν 1
(∆w) ,r + w,θ ,rθ = 0, 0 < θ ≤ 2π (1.143c)
r r r=a
© 2001 by Chapm
(i) ∂Ω is Clamped
In this case, the change from isotropic to orthotropic symmetry is incon-
sequential; the general conditions in (1.129a) once again translate into
(1.130a,b).
(ii) ∂Ω is Simply Supported
Because we still have w = 0 on ∂Ω, the conditions delineated in (1.130a)
still apply in this case. As a consequence of the second condition in
(1.129b), however, we have, in lieu of (1.131), the following statements,
which are, by virtue of (1.67a,b), equivalent to Mx = 0, for x = 0, x =
a, 0 ≤ y ≤ b, and My = 0, for y = 0, y = b, 0 ≤ x ≤ a:
D11 w,xx +D12 w,yy |x=0 = 0, 0 ≤ y ≤ b
D11 w,xx +D12 w,yy |x=a = 0, 0 ≤ y ≤ b
(1.146)
D21 w,xx +D22 w,yy |y=0 = 0, 0 ≤ x ≤ a
D21 w,xx +D22 w,yy |y=b = 0, 0 ≤ x ≤ a
However, by (1.61) - (1.63), and the fact that E1 ν12 = E2 ν21 :
and
so that, for 0 ≤ y ≤ b,
and
D11 w,xxx +(D12 + 4D66 )w,xyy |x=a = 0 (1.152)
© 2001 by Chapm
1.3.1.4 Cylindrical Orthotropic Response: Circular
Geometry
∂w(r, θ)
w(b, θ) = 0, | r=b = 0, (1.154b)
∂r
for 0 < θ ≤ 2π.
(iii) ∂Ωi is Free
With ∂Ωi free, i = 1, 2, (1.153) will still apply but, in lieu of the
vanishing of either w or w,r at r = a, b, we must impose the condition
1
Qr + Mrθ,θ = 0, i = 1, 2 (1.155)
r r=Ri
© 2001 by Chapm
(i) ∂Ωi is Clamped
These conditions are, once again, identical with (1.139), for i = 1, 2.
1 1
D̃1 w,rr +D̃12 w,θθ + w,r
r2 r
(1.157a)
1
−2D̃16 w ,rθ =0
r r=a
1 1
D̃1 w,rr +D̃12 w,θθ + w,r
r2 r
(1.157b)
1
−2D̃16 w ,rθ =0
r r=b
© 2001 by Chapm
is applied normal to the edges at x = 0, x = a, for 0 ≤ y ≤ b. If, e.g.
all four edges are simply supported, and the thin plate is isotropic and
linearly elastic, then, along all four edges of the plate, w = ∆w = 0.
Referring to the von Karman equations, (1.53a,b), which apply in this
case, with t ≡ 0, for a state of generalized plane stress (1.53a) is satisfied
identically while (1.53b) reduces to
y2
Φ̄0 (x, y) = −λh (1.160)
2
inasmuch as we do not care about linear and constant terms in Φ̄0 (be-
cause the expressions involving Φ in (1.53a,b) always appear as second
derivatives in the Airy function.) From (1.160) it is clear that we may
h
take, in accordance with (1.158), Φ0 (x, y) = − y 2 . From (1.160) and
2
(1.46) we see that, N̄x0 = −λh, N̄y0 = 0, N̄xy
0
= 0.
Returning to the general situation we make the following observations:
(i) In every case considered in section 1.2, with respect to the buck-
ling of a thin, linearly elastic plate, either for isotropic or orthotropic
response, and whether it be for the case of rectilinear or circular geom-
etry, the structure of the von Karman equations is as follows:
L1 w = [Φ, w] (1.161a)
1
L2 Φ = − [w, w] (1.161b)
2
1 1
[Φ, w] = Φ,rr r w,r + r2 w,θθ
1 1
+ Φ,r + 2 Φ,θθ w,rr
r r
1 1 1 1
−2 w, rθ − w,θ Φ,rθ − Φ,θ
r r2 r r2
1 1
= w,rr Nr − 2 2
w,θ − 2 w,rθ Nrθ (1.163)
r r
1 1
+ w,r + 2 w,θθ Nθ
r r
−2
1 1
[w, w] = w, rr w, r + w, θθ
r r2
2
1 1
− w,rθ − 2 w,θ
r r
(ii) We set
L2 Φ0 = 0, in Ω, (1.164)
with some appropriate boundary conditions on ∂Ω. Then, (1.161a,b)
may be put in the form
L1 w − λ [Φ0 , w] = [ϕ, w]
(1.165)
L2 ϕ = [w, w]
© 2001 by Chapm
with ϕ(x, y) the “extra” Airy stress (function) generated by the out-of-
plane deflection w.
Remarks: The term [Φ0 , w] may also be written in terms of the stress
resultants (or averaged stresses) Nx0 , Ny0 , Nxy
0
(rectilinear Cartesian
0 0 0
coordinates) or Nr , Nθ , Nrθ (polar coordinates) corresponding to the
state of generalized plane stress in the plate when w ≡ 0.
From our general discussion in section 1.1, we know that branching
from an equilibrium solution (λ0 , u0 ) of a general system of equations
G(λ, u) = 0 can occur only if Gu (λc , u0 ) is not invertible (as a linear
map), i.e., if the eigenvalue-eigenfunction problem
Gu (λ, u0 )v = 0 (1.166)
has a nontrivial solution vc (a “buckling mode”) for some value of λ,
say, λc (a “buckling load”). For the general von Karman system given
by (1.165), which includes all the cases we have considered in section
1.2, the problem equivalent to (1.166) may be formulated as follows:
We take, for the “vector” u the pair (w, ϕ) and we write
u0 = (0, 0) (1.170)
i.e., by w = 0 and Φ = Φ0 = λΦ0 . Computing the Frechét derivative of
the mapping in (1.169), at (λ, u0 ) ≡ (λ, (0, 0)), and setting v = (ŵ, ϕ̂),
we find that
L1 ŵ − λ [Φ0 , ŵ]
Gu (λ, u0 )v = (1.171)
L2 ϕ̂
so that Gu (λ, u0 )v = 0 is equivalent to the system of equations
L1 ŵ − λ [Φ0 , ŵ] = 0
(1.172)
L2 ϕ̂ = 0
and
© 2001 by Chapm
Remarks: For certain types of loadings we will encounter, in Chapter
2, for example, we may not be able to express Φ̄0 directly in the form
λΦ0 , with λ representing the magnitude of a loading along a particular
edge, e.g., compressive loading of the rectangular plate along the edges
parallel to both the x and y axes with different loading magnitudes along
the different pairs of parallel sides. However, the algorithm we have
delineated above is still valid: we simply replace Φ(x, y) in (1.161a) by
Φ̄0 (x, y).
Example: We return to the linearly elastic, isotropic plate depicted
in Fig. 1.12; the plate is simply supported along all four edges and
is subjected to a compressive loading of magnitude λ along the sides
x = 0, x = a, for 0 ≤ y ≤ b. We have already seen, in this case, that
h
Φ̄0 is given by (1.160), so that Φ0 = − y 2 . The relevant von Karman
2
equations are (1.53a,b). Suppressing (1.53b), and replacing Φ in (1.53a)
by Φ̄0 = λΦ0 , we have
K∆2 w = λ%
[Φ0 , w] &
= λ Φ0 ,yy w,xx +Φ0,xx w,yy −2Φ0,xy w,xy
= −λhw,xx
so that the eigenvalue–eigenfunction problem we are interested in is
K∆2 w + λhw,xx = 0, in Ω
(1.174)
w = ∆w = 0, on ∂Ω
with Ω = {(x, y)|0 ≤ x ≤ a, 0 ≤ y ≤ b} . Other examples of linearized
problems governing the initial buckling of thin plates will be considered
as they arise in Chapters 2–5; in certain of these later examples, we
will want to modify the above discussion to account for the presence
of initial imperfections or for constitutive behavior which is other than
linear elastic; however, the same basic logic which took us, in this section,
from the full set of von Karman equations to the (linear) eigenvalue–
eigenfunction problem governing initial plate buckling, will still apply.
1.5 Figures: Plate Buckling and the von Karman
Equations
FIGURE 1.1
Thin rod in compression.
© 2001 by Chapm
FIGURE 1.2
Forces and moments acting on a portion of a buckled thin rod.
FIGURE 1.3
Buckling of the thin compressed rod based only on the lin-
earized equations.
© 2001 by Chapm
FIGURE 1.4
Buckling of the thin compressed rod based on the nonlinear
equations.
FIGURE 1.5
Graph of ϕmax vs. λ for various ranges of the initial angle ϕ(0).
© 2001 by Chapm
FIGURE 1.6
The middle surface of the shallow shell in rectilinear cartesian
coordinates.
FIGURE 1.7
An infinitesimal volume element dV of the shell; the coordinate
ζ measures the distance from the middle surface.
© 2001 by Chapm
FIGURE 1.8
Stress resultants and distributed loading t(x, y) acting on a
differential shell element.
FIGURE 1.9
Moments acting on a differential shell element (Mxy = Myx ).
© 2001 by Chapm
FIGURE 1.10
Components of the stress tensor in polar coordinates.
FIGURE 1.11
A thin elastic plate which occupies the region Ω in the x, y
plane.
© 2001 by Chapm
FIGURE 1.12
A rectangular plate under an in-plane loading.
Chapter 2
© 2001 by Chapm
even within the context of isotropic response. In section 2.3, we present
some results on initial buckling and postbuckling behavior of rectangular
plates exhibiting either elastic-plastic or viscoelastic response. Through-
out our discussions in sections 2.1 through 2.3 of this chapter, we present
the most pertinent graphs that we have found in the literature to de-
scribe the specific initial or postbuckling behavior under consideration;
these graphs then form the basis for our discussion in section 2.4 where
we compare the initial and postbuckling behavior of rectangular plates
with respect to variations in constitutive response, material symmetry,
and boundary conditions.
b2 h
K̃ = λcr (2.2)
π2 K
where λcr will represent an eigenvalue (i.e. buckling load) for (2.1), sub-
ject to the conditions of simple support along the loaded edges, and var-
ious combinations of boundary conditions along the longitudinal edges
at y = 0 and y = b.
(a) The Edges y = 0, y = b, 0 ≤ x ≤ a are Simply Supported.
In this subcase w = 0 along y = 0, y = b, 0 ≤ x ≤ a, and the second
set of boundary conditions in (1.131) also applies along these edges.
We assume, because of the nature of the loading, and the conditions of
simple support along the loaded edges, that the plate buckles into m
sinusoidal half-waves in the direction of loading; this leads, naturally,
to a search for solutions of the boundary-value problem associated with
(2.1) of the form
mπx
w(x, y) = f (y) sin (2.3)
a
Substitution of (2.3) in (2.1) shows that f must have the form
and that the critical values λcr of the compressive stress must satisfy
the characteristic equation
where
mπ 2 m
p/q = K̃ = (2.5)
φ φ
with
m2 π 2 m2 π 2
r 2 = p2 − ν 2
, s2 = q 2 + ν 2 (2.6)
φ φ
a
φ = , and K̃ given by (2.2). For m = 1, the solution of (2.4) is given
b
by
K̃ = φ−2 + φ2 + 2 (2.7)
which is graphed in Fig. 2.1. With m = 1, the value of K̃, and, thus,
a
of λcr , increases monotonically as φ = increases from φ = 1 (square
b
© 2001 by Chapm
plate) and for large φ (a long, narrow plate), K̃ ∼ φ2 . If we consider the
case of two half-waves in the x-direction, then with the same definitions
as in (2.5), (2.6), it can be shown that the value of λcr is unchanged if
Nm
m/φ is replaced by for any integer N . As a consequence, it may,
Nφ
for example, be shown that λcr for a plate with aspect ratio 2φ, buckling
into the two half-wave mode (m = 2), is the same as that for a plate
with aspect ratio φ buckling into the one half-wave mode (m = 1). The
K̃ versus φ curves, in general, are shown in Fig. 2.2 for various values
of m, where
K̃ ≡ K̃m = m2 φ−2 + m−2 φ2 + 2 (2.8)
Note, e.g., that for aspect ratio φ = 2, the minimum value of λcr is
associated with buckling into the two-half-wave mode (m = 2) and, in
fact, in this case (λcr )min corresponds to the value K̃ = 4. Also, a
transition occurs from buckling into m half-waves to buckling into m + 1
half-wave whenever successive curves intersect. From (2.8) it follows
that K̃m = K̃m+1 provided
φ2 = m(m + 1) (2.9)
© 2001 by Chapm
The solution to (2.14), for m = 1, is depicted in Fig. 2.5 and an energy
analysis yields the result
πφ2 π − 4ν
K̃ = φ−2 + + (2.15)
16(3π − 8) 2(3π − 8)
A1 ps sinh p + A2 ps cosh p + A3 qr sin q − A4 qr cos q = 0
2 2 2 2
A1 r2 cosh p + A2 r2 sinh p − A3 s2 cos q − A4 s2 sin q = 0
with
πφ
α + K̃ + K̃ − 4
2a
(2.18b)
β = πφ
K̃ − K̃ − 4
2a
and imposition of the boundary conditions along the four edges produces
a characteristic equation relating K̃ and φ (and hence, λcr and φ) whose
solution is depicted in Fig. 2.7; it may be noted that modes with even
numbers of half-waves in the x-direction are antisymmetric. The solution
to this problem was first obtained by Schleicher [75]. As φ increases, the
value of K̃ approaches that of a plate simply supported on all four sides,
because the half-wave length of the buckles in the y-direction approaches
a value equal to the width of the plate. Finally, the transition from m
to m + 1 half waves, in the x-direction, takes place when φ2 = m(m + 2).
with m having odd values for buckling into an odd number of waves
in the x-direction, and even values for buckling into an even number of
© 2001 by Chapm
waves; the conditions in (I.130b) are now imposed upon this form of the
solution which generates the system of equations
∞
# mnkm + m2 tn
2 = 0 (n = 1, 3, 5, . . .)
m2 m2 K̃
2
−
m = 1, 3, 5, . . . φ2 + n φ2
or 2, 4, 6 . . .
# ∞ (2.19)
n 2
k + nmt m = 1, 3, 5 . . .
2
m
2
n
=0
or 2, 4, 6
m m2 K̃
n = 1, 3, 5 . . . + n 2
−
φ2 φ2
The system of homogeneous, linear, algebraic equations for km and tn ,
(2.19) produces a characteristic equation whose solution is shown in Fig.
2.8, which is taken from Levy [76]; on this graph we also depict the results
for the earlier case, where the loaded edges at x = 0, x = a, 0 ≤ y ≤ b,
were simply supported, and the edges along y = 0, y = b, 0 ≤ x ≤ a,
are clamped. From the graph in Fig. 2.8 it follows that as φ increases,
K̃ approaches the value that it has for a long plate, which is simply
supported along its loaded edges, and clamped along the edges y = 0
and y = b. In the present case (loaded edges clamped), the change
from m = 1 to m = 2 half-waves in the x-direction takes place at the
intersection of the m = 1 and m = 3(K̃ versus φ) curves which apply
in the earlier case where the loaded edges were simply supported and, it
is clear from Fig. 2.8, that this pattern continues as we transition from
m = 2 to m = 3 half-waves in the x-direction, etc.
when −1/2
2ξ ξ 1
φ ≡ φmin = 1− , 0< < (2.26)
λ λ 2
ξ 1
For values of λ and ξ such that > , the corresponding values
λ 2
of φmin are imaginary. The graph of K̃x versus the aspect ratio φ for
the plate is depicted in Fig 2.10; from this figure we may deduce that
the graph of K̃x , as a function of φ, is asymptotic to the horizontal line
K̃x = λ/ξ. Also, for the case in which the plate is square (i.e., φ = 1)
and is subjected to uniform compression on all sides (λ = ξ) we have
K̃x = 2, which is one-half of the value of K̃ for a simply supported
square plate which is compressed in only one direction. From (2.25) and
(2.26), it follows that for ξ < 0 (i.e. if a tensile force acts on the edges
y = 0, y = b, 0 ≤ x ≤ a)φmin is not imaginary and (K̃x )min may be
larger than 4; this implies that the presence of a tensile stress along the
© 2001 by Chapm
edges y = 0, y = b can increase the initial buckling load for a simply
supported plate which is acted upon by compressive stresses applied to
the edges at x = 0 and x = a. An alternative rendering of the results
for this problem may be found in the work of Przemieniecki [78]; these
results are summarized in Fig. 2.11.
(b) The Edges at x = 0, x = a, 0 ≤ y ≤ b, and y = 0, y =
b, 0 ≤ x ≤ a, are Clamped.
In this subcase, we are still dealing with (2.20), but this equation is
now subjected to the boundary conditions in (1.130a), (I.130b). An
approximate solution for this situation was derived by Timoshenko [64]
who assumed that φ ≈ 1(i.e., the plate does not differ much from a
square) and that λ ≈ ξ; the deflection of the buckled plate is sought in
the form
1 2πx 2πy
w(x, y) = A 1 − cos 1 − cos (2.27)
4 a b
Using the hypothesized form of the eigenfunction given by (2.27), and
equating the work done by the applied compressive loads to the strain
energy of bending, Timoshenko [64] shows that
2π2 K −2 2 8
λ+φ ξ = 2 4φ + 4φ + (2.28)
b h 3
For the case φ = 1, ξ/λ = 1 (i.e. a square plate under uniform compres-
sion) K̃x = 5.33.
(c) The Edges at x = 0, x = a, 0 ≤ y ≤ b, are Simply
Supported and the Edges at y = 0, y = b, 0 ≤ x ≤ a are
Clamped.
This subcase has also been treated by Przemieniecki in [78] ; the bound-
ary conditions are the first pair of relations in (1.130a), and the first
pair of relations in (1.131), along x = 0, x = a, for 0 ≤ y ≤ b, as well as
the second set of relations in (1.130a), and the second set of relations in
(I.130b), along y = 0, y = b, for 0 ≤ x ≤ a. The results in [78] are dis-
played graphically in Fig. 2.12 in the form of curves which relate λ/λcr
with φ for different values of ξ/ξcr , where λcr is the (critical) buckling
load for a simply supported plate under unidirectional compression, i.e.
2
4π 2 E h
λcr = (2.30)
12(1 − ν 2 ) b
under the hypothesis that the buckled form of the plate consists of m
half-waves in the x-direction, we compute the strain energy of bending
to be 2
∞
Kabπ 2 # 2 1 n2
U= An + 2 (2.32)
8 n=1
a2 b
while, for the work done by the compressive edge forces, we have
2
1 c a b
∂w
T = λ 1− y h dxdy (2.33)
2 b o 0 ∂x
where only values for q are summed over such that n + q is odd. Equat-
ing U and T then produces an expression for λcr , which contains the
© 2001 by Chapm
coefficients An , and these must now be chosen so as to minimize the
value of λcr ; standard max–min calculations yield the following system
of equations
2 c
An 1 + n2 φ2 − K̃φ2 1 −
2
∞ (2.35)
8c # Aq · nq
− 2
φ K̃ =0
2
q=1
(n2 − q 2 )2
In view of (2.39),
∞ #
# ∞ #
∞ #
∞
mnpq
T = −4ζ · h Amn Apq (2.41)
m=1 n=1 p=1 q=1
(m2 − p2 )(q 2 − n2 )
It is noted in [32], that the system (2.42) may be split into two subsys-
tems of equations each of which can be solved separately; one subsystem
© 2001 by Chapm
contains constants Amn for which (m+n) is odd, and represents antisym-
metric buckling, while the other contains constants for which (m + n) is
even and represents symmetric buckling. Some of the solutions obtained
by Stein and Neff [80] for this case are depicted in Fig. 2.17; according
1
to this figure, the lowest value of K̃ for .5 < < 1.0 is associated with
φ
1
one buckle (symmetric), and for .3 < < .5 with two buckles (anti-
φ
1
symmetric), while for → 0, K̃ approaches its value for an infinitely
φ
long plate. In analyzing the plate subjected to shearing forces along its
edges, with the edges simply supported, we may interchange the roles
of the sides of dimensions a and b, respectively; thus, it makes no sense
1
to consider values of in Fig. 2.17 which exceed one. A reasonable
φ
parabolic approximation to the graph in Fig. 2.17 is given by
© 2001 by Chapm
2.1.2 Rectilinear Orthotropic Symmetry:
Initial Buckling
For several cases analogous to the ones considered in the isotropic
case, we now consider the results which are available for initial buckling
of a rectilinearly orthotropic rectangular plate which, again, occupies
the region in the x, y plane defined by 0 ≤ x ≤ a, 0 ≤ y ≤ b. From
(1.68), the first of the von Karman equations in this case reads
where, by (1.63)
E1 h3
D11 = ·
1 − ν12 ν21 12
D22 = E2 h3
·
1 − ν12 ν21 12
and
E2 ν21 h3 E1 ν12 h3
D = · = · = D21
12
1 − ν12 ν21 12 1 − ν12 ν21 12
D h3
66 = G12 ·
12
The partial differential equation governing the eigenvalue–eigenfunction
problem, for initial buckling of the plate, is then, by virtue of our dis-
cussion in Chapter 1, given by (2.49) with Φ, the total Airy function,
replaced by Φ̄; we recall that Φ̄ is the solution of (1.69), for w ≡ 0,
subject to the appropriate edge loading conditions, i.e., Φ̄ satisfies
1 ∂ 4 Φ̄ 1 1 ∂ν12 ∂ 4 Φ̄ 1 ∂ 4 Φ̄
4
+ − 2 2
+ =0 (2.50)
E1 h ∂y h G12 E1 ∂x ∂y E2 h ∂x4
Equations (2.49), with Φ = Φ̄, and (2.50), apply in the domain 0 < x <
a, 0 < y < b.
(i) Compression in the x-Direction with the Loaded Edges
Simply Supported
Just as in subcase (i), for the isotropic situation, Fig. 1.12 applies with
σxx = −λ, λ > 0, along the edges at x = 0 and x = a; also, along these
edges, for 0 ≤ y ≤ b, the first pair of boundary conditions in (1.146)
hold together with w(0, y) = w(a, y) = 0, 0 ≤ y ≤ b. The relevant
eigenvalue problem is governed by
* (2.55)
2(D12 + 2D66 ) D22 φ
+ √ +
D11 D22 D11 m
© 2001 by Chapm
in order to determine λcr . The various cases which present themselves
are as follows:
*
D11
1) If the aspect ratio satisfies φ = m 4 , for m an integer, then
D22
m = m provides the minimal λm1 in (2.55), in which case λcr is given
by
√
2π 2 D11 D22 [D12 + 2D66 ]
λcr = 1+ √ (2.56)
b2 h D11 D22
Following the point of view taken in [51], another way to look at this is
as follows: for m = 1 in (2.55) we have
π 2 D11 2
λ11 = + 2[D12 + 2D66 ] + D22 φ (2.57)
b2 h φ2
in which case, the smallest value of λ11 occurs for the aspect ratio φ =
*
D11
and this smallest value is just given by (2.56). The relation (2.56)
D22
can be expressed in the form
√
K̃min π 2 D11 D22
λcr = (2.58)
b2 h
where
[D12 + 2D66 ]
K̃min = 2 1 + √ (2.59)
D11 D22
2) If
*
D11
φ = m(m + 1) 4
≡ φm (2.60)
D22
for m any integer, then at the same critical load there are two possible
buckling modes, namely,
mπx πy
w = Am1 sin sin (2.61)
a b
and
(m + 1)πx πy
w = Am+1,1 sin sin (2.62)
a b
and the φm represent, therefore, the aspect ratios where a transition
from buckling in the mode (2.61) to buckling in the mode (2.62) occurs.
From (2.60) we deduce directly that, at λ = λcr , if
*
D11
0 < φ < 1.41 4 , then m = 1
D22
* *
D11 D11
1.41 4
< φ < 2.45 4 , then m = 2, (2.63)
D D22
22
* *
2.45 4 D11 < φ < 3.46 4 D11 , then m = 3
D22 D22
and so forth.
3) For a given aspect ratio φ, therefore, (2.60) can be used to determine
the value of m which holds at buckling; the use of this integer m in
(2.55) then produces the corresponding critical load λcr for the given φ.
In general, by virtue of (2.59), we may write that
√
π 2 D11 · D22
λcr = k
b2 h
where
* 2 * 2
D11 m 2(D12 + 2D66 ) D22 φ
k ≡ km = + √ + (2.64)
D22 φ D11 D22 D11 m
In Fig. 2.21 (and Table 2.1) we show a typical graph of k against φ
for the case of compression along the external fibers while in Fig. 2.22
(and Table 2.2) we show the analogous results for compression across
the external fibers; in the first case, D11 > D22 , while in the latter case,
D11 < D22 . The figures referenced above clearly show that the number
of half-waves in the x-direction is much larger for a given aspect ratio
φ when compression is across the external fibers than when it is along
them. In Fig. 2.23 we show the graph of k against β̄ = φ · 4 D22 /D11
(which [43] defines to be an “effective” aspect ratio); the parameter α,
which varies along each family of curves, is
D12 + 2D66
α= √ (2.65)
D11 D22
© 2001 by Chapm
satisfies the conditions of simple support along the loaded edges at x = 0
and x = a, for 0 ≤ y ≤ b. If we substitute (2.66) into (2.51), we find
that f (y) must satisfy
mπ 2
f (y) − 2 [D12 + 2D66 ]f (y)
a
(2.67)
mπ 4 mπ 2
+ D11 − λh f (y) = 0
a a
and equating the work of the compressive edge forces with the potential
energy acquired during bending; it is determined that λ must satisfy
√ * 2
π 2 D11 D22 D11 m
λ ≡ λmn =
b2 h D22 φ
2
* (2.72)
8 2 [D12 + 2D66 ] 16 4 D22 φ
+ n √ + n
3 D11 D22 3 D11 m
The results for (2.73) are now quite analogous to those for the case of
compression in the x-direction with four simply supported edges:
*
4 D11
1) If the aspect ratio φ satisfies φ = .658m , for some integer m ,
D22
then at criticality, m = m and
√
8π 2 D11 D22 √ [D12 + 2D66 ]
λcr = 3 + √ (2.74)
3b2 h D11 D22
© 2001 by Chapm
3) For a value of the aspect ratio φ which does not satisfy either the
conditions in 1 or 2, above, the value of m may be determined by looking
at the transitions governed by the relation in 2 (above) i.e., if
*
D11
0 < φ < .931 4
, then m = 1
D22
* *
D11 D11
.931 4 < φ < 1.61 4 , then m = 2 (2.76)
D 22 D22
* *
1.61 4 D11 < φ < 2.28 4 D11 , then m = 3
D22 D22
and so forth; once m has been determined, the corresponding value of
λcr is then given by (2.73). In Fig. 2.23 we also show the graph of
* 2
D11 m 8 [D12 + 2D66 ]
k ≡ km = + √
D22 φ 3 D11 D22
* 2 (2.77)
16 D22 φ
+ ,
3 D11 m
against β̄ = φ 4 D22 /D11 , for various values of α, as defined by (2.65),
when the loaded ends of the plate at x = 0, x = a are simply supported
and the ends at y = 0, y = b are clamped.
(ii) Compression in the x-Direction with the Loaded Edges
Clamped.
The relevant differential equation is still (2.51); the boundary conditions
along x = 0 and x = a, 0 ≤ y ≤ b, are given by the first pair of conditions
in each of (1.130a,b).
(a) The Edges y = 0, y = b, 0 ≤ x ≤ a are Simply Supported.
In this subcase w = 0 along y = 0, y = b, 0 ≤ x ≤ a, and the second
pair of conditions in (1.148) hold as well. Analysis, similar to that
presented in the case where the loaded edges were simply supported,
again produces a relation of the form
π2
λcr = k D11 D22 (2.78)
b2 h
for an appropriately defined buckling coefficient k = km . The graph in
Fig. 2.24 depicts k vs. β, again for varying values of the parameter α.
(b) The Edges y = 0, y = b, 0 ≤ x ≤ a, Are Clamped
In this subcase, we have the second pair of conditions, in each of (1.130a)
and (1.130b), holding along y = 0 and y = b, for 0 ≤ x ≤ a. A relation
of the type (2.78) again applies with the results displayed on the graph
in Fig. 2.24.
(iii) Compression in both the x and y Directions The situation
here corresponds to that depicted in Fig. 2.9; instead of equation (2.20),
which governs initial buckling for this loading condition, in the isotropic
case, we have (λ = −σxx |x=0,a , ξ = −σyy |y=0,b ):
We consider only the case where the edges of the plate are simply sup-
ported on all four sides so that w = 0 for x = 0, x = a, 0 ≤ y ≤ b, w = 0,
for y = 0, y = b, 0 ≤ x ≤ a, and all four conditions in (1.148) apply
as well. Because of the condition of simply supported edges, we again
look for eigenfunctions of the form (2.52), in which case, substitution
into (2.79) yields
m 2 n 2 m 4
π2
λ +ξ = D11 +
a b h a
mn 2 n 4 (2.80)
2[D12 + 2D66 ] + D22
ab b
2[D12 + 2D66 ] 2
+ √ ·n (2.81)
D11 D22
* +
( 2
2
D22 φ 4 φ
+ n 1+η n2
D11 m m
© 2001 by Chapm
and the problem again consists of seeking values of the integers m and
n which give the smallest value of λ and, thus, λcr ; numerical analysis
yields the following [51] :
when D11 > D22 (x-axis directed along the external fibers). The plate,
upon initial buckling, exhibits one-half sine waves in the direction of the
y-axis.
In this case, for compression in the x-direction (along the external fibers)
and tension in the y-direction, we have, m = 2, n = 1, and, (approxi-
mately),
√
π 2 D11 D22
λcr = 19.67 (2.83)
b2 h
while for compression in the y-direction (across the fibers), and tension
in the x-direction, m = 2, n = 1, and, (approximately),
√
π 2 D11 D22
λcr = 3.72 (2.84)
b2 h
Thus, the plate is more stable in the first case, represented by (2.83),
as opposed to the second case, represented by (2.84) since the critical
λ in (2.83) is more than five times the size of the critical λ in (2.84).
Other subcases of this particular problem which have been treated in
Lekhnitskii [51] include the following:
(c) The Plate is Simply Supported along all Four Edges, with
a Compressive Stress σxx = −λ Uniformly Distributed along
the Edges x = 0, x = a, 0 ≤ y ≤ b, and a Constant Tension
given by σyy = +ξ̄ Uniformly Distributed along the Edges
y = 0, y = b, 0 ≤ x ≤ a.
Setting ξ = −ξ¯ in (2.80), we find that
√ * 2
π 2 D11 D22 D11 m [D12 + 2D66 ] 2
λ= +2 √ n
b2 h D22 φ D11 D22
* 2 ¯ 2 2
(2.85)
D22 φ 4 ξhb φ
n + 2√ n2
D11 m π D11 D22 m
From (2.86) it follows that an additional tension (load) along the edges
y = 0, y = b raises the critical load and, thus, improves the stability of
the plate. We also note that, by virtue of (2.86), for a given aspect ratio
φ, m may be large if either D11 is small or D22 is large.
2) The aspect ratio at which a transition from m half sine waves to
m + 1 half sine waves (in the x-direction) takes place is
m(m + 1)
φ ≡ φm = (2.88)
4 D ¯ 2h
ξb
22
+ 2
D11 π D11
© 2001 by Chapm
y
have, along the edges x = 0, x = a, 0 ≤ y ≤ b, σxx = λ 1 − c .
b
Using an energy method, Lekhnitskii [51] shows that
2 2
a b
1 ∂ w
λ= a b 2 D11
c ∂w 0 0 ∂x
1− y dxdy
0 0 b ∂x
2 2 2 2
∂2w ∂2w ∂ w ∂ w
+2D11 ν21 2 + D22 + 2[D12 + 2D33 ] dxdy
∂x ∂y 2 ∂y 2 ∂x∂y
(2.89)
The most general form for an initial buckling mode satisfying the con-
ditions of simple support on all four edges is
∞ #
# ∞
mπx nπy
w= Amn sin sin
m=1 n=1
a b
a first approximation to which may be taken to be
mπx πy
w ≡ wm = Am1 sin sin (2.90)
a b
Using (2.90), we obtain from (2.89),
√ * 2
π 2 D11 D22 D11 m
λcr = 2
b h(1 − c/2) D22 φ
* 2
(2.91)
2(D12 + 2D66 ) D22 φ
+ √ +
D11 D22 D11 m
We note that (2.91) is valid only for small c and, in fact, loses its meaning
for c = 2, which corresponds to the case of pure bending. The calcu-
lations may be found in Lekhnitskii [51] for the case in which a second
approximation to w of the form
πy 2πy mπx
w = Am1 sin + Am2 sin sin (2.92)
b b a
is employed in (2.89), but we will not pursue them in detail here; some
of the results, however, are well worth noting, as follows: If we set
* 2 * 2
D11 m 2(D12 + 2D66 ) D22 φ
am1 =
+ √ +
D22 φ D11 D22 D11 m
* (2.93)
* 2 2
D m 8(D + 2D ) D φ
am2 = 11
+ √
12 66
+ 16
22
D22 φ D11 D22 D11 m
then the energy method yields
1 1
√ −
π 2 D11 D22
1 c (am1 + am2 )
2 2
λ= 2 2
b2 h 1 16c
1− c −
2 9π 2
2 (2.94)
1 1 16c
1 − c (am1 − am2 ) +
2
am1 am2
4 2 9π 2
± 2 2
1 16c
1− c −
2 9π 2
© 2001 by Chapm
results on the number of half-sine waves in the x-direction, which occur
upon initial buckling, hold:
*
D11
0<φ< 4 , then m = 1
D22
* *
D11 D11
4
< φ < 1.73 4 , then m = 2 (2.98)
D D
22 22
* *
1.73 4 D11 < φ < 2.45 4 D11 , then m = 3
D22 D22
and so forth. Using these results, for a given value of the aspect ratio
φ we may determine the number m of half-sine waves in the x-direction
into which the plate buckles and then, from (2.93), (2.96), and (2.97),
the corresponding critical load. It may again be demonstrated that, as
was the case with compression of the simply supported, rectilinearly or-
thotropic plate, (along the edges x = 0, x = a, 0 ≤ y ≤ b) if D11 < D22
a large number of half-sine waves occur upon initial buckling as com-
pared with the case in which D11 > D22 (loading parallel to the external
fibers of the plate as opposed to loading perpendicular to the external
fibers). For plywood plates deflected by a pure bending moment, we in-
dicate in Table 2.3 the values of k and m (relative to (2.96), (2.97)) for
various values of the aspect ratio φ, both in the case where D11 > D22
and in the case where D11 < D22 .
(v) Rectangular Orthotropic Plates Subjected to
Shearing Loads
The situation is again depicted, as in the isotropic case, in Fig. 2.16.
Taking, once again, ζ = σxy (along the four edges of the plate), the
partial differential equation which mediates the initial buckling of the
plate is
∂4w ∂4w
D11 4 + 2(D12 + 2D66 ) 2 2
∂x ∂x ∂y
(2.99)
∂4w ∂2w
+D22 2 + 2ζh 2 2 = 0
∂y ∂x ∂y
The principal directions of the plate are, of course, taken to be parallel
to the edges of the plate, all of which are simply supported. Lekhnit-
skii [51] employs the energy method to obtain a solution of the initial
buckling problem in this case. The work done by the edge forces is, once
again, given by (2.40b), as in the isotropic case, and equating this to the
potential energy of bending yields
(
a b a b
∂w ∂w
ζ= Ψdxdy 2h dxdy (2.100)
0 0 0 0 ∂x ∂y
with
2
∂2w ∂2w ∂2w
Ψ(x, y) ≡ D11 + 2D11 ν21
∂x2 ∂x2 ∂y 2
2 2 (2.101)
∂2w ∂2w
+D22 + 4(D12 + 2D12 )
∂y 2 ∂x∂y
The initial buckling mode is again sought in the form (2.38), which
automatically satisfies the simple support conditions along the edges;
the rest of the solution now proceeds as in the isotropic case. It is noted
in [51] that, with the external fibers of the plate parallel to the long sides
of the plate, good accuracy is achieved with five terms in the expression
(2.38) for the initial buckling mode, i.e., with
πx 3πx πy
w = A11 sin +A31 sin sin
a a b
2πx 2πy
+A22 sin sin (2.102)
a b
πx 3πx 3πy
+ A13 sin + A33 sin sin ,
a a b
if the plate is square (φ = 1); other approximations (for the aspect ratios
φ = 2 and φ = 3) are also reported in Lekhnitskii [51]. In each of the
cases referred to above, the critical (shear) buckling load assumes the
form 2
4 h
ζcr = 10 k (2.103)
b
where the buckling coefficient k, which depends on φ, is tabulated in
Table 2.4.
Remarks : Shearing of a Very Narrow or Infinite
Orthotropic Strip
A special case of the problem treated above occurs when the ratio of the
sides is large, i.e., φ > 4; in this case, as Lekhnitskii [51] points out, the
effect of the short sides is negligible and the plate may be considered
© 2001 by Chapm
to be an infinite strip, as depicted below in Fig. 2.25. The rigorous
solution of this problem is due to Seydel [84]–[86] who considered the
cases where the principal directions are both parallel and perpendicular
to the edges of the plate for the case of both simply supported as well
as clamped edges; he found approximate solutions for both infinite and
finite rectangular plates for arbitrary values of the Dij . The results of
Seydel’s investigations for infinite plates were presented by Bergmann
[87] in the following form, for the case of simply supported edges:
2D22 (D12 + 2D66 ) D11 D22
ζcr = 2
8.3 + 1.525
(b/2) h (D12 + 2D66 )2
(2.104)
2 2
D11 D22
−.493
(D12 + 2D33 )4
when
D11 D22 < (D12 + 2D66 )2 (2.105)
and
4 3 2
D11 D22 8.125 + 5.64 (D12 + 2D66 )
ζcr =
(b/2)2 h D11 D22
(2.106)
2
(D12 + 2D33 )
−.6
D11 D22
when
D11 D22 > (D12 + 2D66 )2 (2.107)
For the finite, long, rectangular, orthotropic plate under shear along its
edges, Seydel, op. cit., seeks a solution of (2.99) of the form
2η
w = f (y) exp ix (2.108)
b
in which case f (y) must satisfy
2
2η
D22 f − 2(D12 + 2D66 ) f
b
4 (2.109)
2η 2η
+iζh · f + D11 f =0
b B
with associated characteristic equation
2 4
2η 2η 2η
D22 r − 2(D12 + 2D66 )
4
r + iζh · r + D11
2
= 0 (2.110)
b b b
An analysis of (2.10) produces an expression for the initial buckling
mode of the form
2η
w = exp ix c1 e2β1 iy/b + c2 e2β2 iy/b
b
+c3 e2β3 iy/b + c4 e2β4 iy/b (2.111)
In Fig. 2.26, taken from Seydel op. cit., K̃, the buckling coefficient,
is plotted against 1/β̄ where β̄ is the effective aspect ratio, β̄ = φ
4
D22 /D11 , for various values of the plate parameter θ(≥ 1), as given by
(2.112); for the plot shown in Fig. 2.26, all the edges of the plate are
simply supported. If the plate parameter θ satisfies θ < 1, instead of
θ ≥ 1, we have, in lieu of (2.113), an expression of the form
ζcr = K̃π 2 D22 (D12 + 2D66 )/b2 h (2.114)
for the critical shear stress at which the plate buckles; another (similar)
description of the results for this case is depicted in Fig. 2.27.
© 2001 by Chapm
which governs initial buckling is
∂2w ∂2w
+λh 2
+ 2ζh =0 (2.115)
∂x ∂w∂y
and the plate is assumed to be simply supported along all four edges.
According to Lekhnitskii [51], a solution of the initial buckling problem
is known only for the case of an infinite strip, i.e. aspect ratio φ = ∞;
The method consists of seeking an initial buckling mode in the form
πy π
w = A sin sin (x − y tan ψ) (2.116)
b s
where
γ = (b/s)2 , α = tan ψ (2.118)
2
ζ = 2π D11 D22 · α (D√12 + 2D66 ) γ + D22 (3 + α2 γ)
b2 h D11 D22 D11
(2.119)
*
D22
γ =
D11 + 2(D12 + 2D66 )α2 + D22 α4
Substituting the value of γ from (2.119) into (2.117), as well as into the
expression for ζ in (2.119), yields the following results:
√ + *
2π 2 D11 D22 2[D12 + 2D66 ] D22
λ= √ + 3α 2
b2 h D11 D22 D11
(2.120)
*
2(D12 + 2D33 ) 2 D22 4
+ 1+ α + α − 2αζh
D11 D11
√ + *
2π 2 D11 D22 D22
ζ= 2
·α 3
b h D11
(2.121)
[D12 + 2D66 ] + α2 D22
+*
2(D12 + 2D66 ) 2 D22 4
1+ α + α
D11 D11
Equations (2.120), (2.121) yield a relation between λ and ζ in para-
metric form; as the parameter α is varied from 0 to ∞ [ψ varies from 0
radians to π/2 radians ] we may plot this relation with the values of λ
as the abscissa and the values of ζ as the ordinate (see Fig. 2.28); the
graph which results is a parabolic curve of the form
λ ζ2
+ 2 =1 (2.122)
λ̄cr ζ̄cr
© 2001 by Chapm
and the following result for the case of a clamped strip:
2
ζcr
λcr + + * 6
12π 2 D22
3D22 + {D12 + 2D66 }
b2 h D11 (2.124)
15π 2
= D11 D22 + {D12 + 2D 66 }
b2 h
© 2001 by Chapm
Next, equating all terms of order δ reproduces the small-deflection or
initial buckling equation, namely,
h2 4 ∂ 2 w1 ∂u0 ∂v0 ∂ 2 w1 ∂v0 ∂u0
w1 = +ν + + ν
12 ∂x2 ∂x ∂y ∂y 2 ∂y ∂x
2
(2.127)
∂ w1 ∂u0 ∂v0
+(1 − ν) +
∂x∂y ∂y ∂x
and have been chosen to illustrate the technique because they yield par-
ticularly simple results. The parameter
a a δ is taken to be the deflection
at the center of the plate, i.e., at , ; while we have already seen
2 2
that, for high enough values of the compression, two or more buckles can
occur in the direction of the compression, we will confine our attention
here to plates exhibiting a single initial buckle.
The solution of (2.126) which satisfies (2.128) is
u0 = −k0 x, v0 = 0 (2.129)
© 2001 by Chapm
the solution of which, subject to (2.128), yields
πx πy πx 3πy
w3 = (A + B) sin sin + A sin sin
a a a a
(2.135)
3πx πy
+B sin sin
a a
with
A = 3(1 − ν)(1 + ν)2 /16(24 + 25ν − 9ν 2 )h2
(2.136)
B = 3(1 − ν)(1 + ν)2 /16(16 + 25ν − ν 2 )h2
and
π2
(3 − ν)
k2 = (2.137)
8a2
The cycle of operations described above now continues so as to first
determine u4 , v4 and, then, w5 , etc. If δucr represents the amount
by which the edges at x = 0, x = a approach each other upon initial
buckling, then it is possible to relate the amount by which the loaded
edges approach each other, beyond that point, to the magnitude of the
center deflection δ, namely,
π 2 (3 − ν) 2
δu − δucr = δ + O(δ 2 ) (2.138)
δa
Of more practical importance
a is the relation between δu − δucr and
P − Pcr , where P = h σxx dy denotes the total applied compressive
0
load. However,
2 2
a
Eh ∂u ∂v 1 ∂w 1 ∂w
P =− +ν + + ν dy (2.139)
1 − ν2 0 ∂x ∂y 2 ∂x 2 ∂y
π 2 Eh 2
P − Pcr = δ + O(δ 4 ) (2.140)
4(1 + ν)a
Combining (2.140) with (2.138) then yields
P − Pcr 2Eh
= + O(δ 2 ) (2.141)
δu − δucr (1 + ν)(3 − ν)
As (P −Pcr )/(δu−δucr ) is the stiffness of the plate, immediately after the
onset of buckling, and Eh/(1−ν 2 ) is the plate stiffness prior to buckling,
the ratio of the stiffnesses is (approximately) dependent only on the
Poisson ratio ν and may be computed to be 2(1 − ν)/(3 − ν). A graph of
the postbuckling behavior for the square isotropic plate discussed above
is shown in Fig. 2.29, in which P/Pcr is plotted against δu /δucr ; another
postbuckling graph of P/Pcr , against the deflection δ of the center of the
plate, when the plate buckles (initially, i.e., (2.131)) into one-half sine
wave in the x-direction may be obtained by employing (2.140), i.e.,
π2 E
P/Pcr = 1 + · δ 2 + O(δ 4 ) (2.142)
4Pcr (1 + ν)a
ĒE11 h
Nxx = 2 ($xx + ν12 $yy ) (2.143a)
E − ν12
© 2001 by Chapm
ĒE11 h
Nyy = (Ē$yy + ν12 $xx ) (2.143b)
Ē − ν12
2
and
with
ĒE11 h
Nxn = (un,x + ν12 vn,y ), (2.146)
Ē − ν12
2
Nxyn = Gh(un,y + vn,x ) (2.147)
and analogous expressions for the other coefficients Nxnm , Nyn , Nynm
and Nxynm . Substitution of the expansions in (2.145) into (1.42) and
(1.68) then yields a recursive collection of sets of three equations each
for the u2k , v2k , w2k+1 , the first two of which are
Nx0,x + Nxy0,y = 0
Ny0,y + Nxy0,x = 0 (2.148)
L(D)w1 − (Nx0 w1,xx + Ny0 w1,yy + 2Nxy0 w1,xy ) = 0
∂4 ∂4 ∂4
where L(D) = D11 + 2(D 12 + 2D66 ) + D22 , and
∂x4 ∂x2 ∂y 2 ∂y 4
Nx2,x + Nxy2,y = −(Nx11,x + Nxy11,y )
Ny2,y + Nxy2,x = −(Ny11,y + Nxy11,x )
L(D)w3 − (Nx0 w3,xx + Ny0 w3,yy + 2Nxy0 w3,xy ) (2.149)
= (Nx2 + Nx11 )w1,xx + (Ny2 + Ny11 )w1,yy
+2(Nxy2 + Nxy11 )w1,xy
© 2001 by Chapm
' '
a ' a '
non-loaded edges, Nyy ''y=0 dx = Nyy '' dx = 0; if into these
0 0 y=b
load conditions on the edges we substitute the perturbation expansions
for Nxx and Nyy from (2.45) and, also, replace P by Pcr + η 2 Pcr ,
then we obtain a sequence of boundary conditions for the coefficients
Nxn , Nxnm , . . . , Nxynm of the form
b '
'
'
Nx0 ' dy + Pcr = 0
'
0
x=0,a
'
b
'
'
(Nx2 + Nx11 )' dy + Pcr = 0 (2.152)
'
0
'
x=0,a
b '
'
(Nxy + 2Nx13 )' dy = 0
'
0
x=0,a
We note that solutions of the first two equations in (2.148) provide the
prebuckling stresses in the rectangular, orthotropic plate; these equa-
tions may be put in the form
Au0,xx + G12 hu0,yy + Bv0,xy = 0
(2.154)
ĒAv0,yy + G12 hv0,xx + Bu0,xy = 0
where
ĒE11 h
A= , B = ν12 A + G12 h (2.155)
Ē − ν12
2
Pcr
Nx0 = − , Ny0 = Nxy0 = 0 (2.156b)
b
The use of (2.156b) then permits us to write the last equation in (2.148)
in the form
Pcr
L(D)w1 + w1,xx = 0 (2.157)
b
and imposition of the simple support condition along the four edges of
the plate again leads to an initial buckling mode of the form
mπx nπy
w1 (x, y) = A1 sin sin (2.158)
a b
As in our previous discussion of the initial buckling problem, for a simply
supported, orthotropic, rectangular plate which is compressed in the x-
direction, we substitute (2.158) into (2.157) so as to determine that
mπ 4 mπ 2 nπ 2
b
Pcr = D 11 + 2(D 12 + 2D 66
(mπ/a)2 a a b
nπ 4 (2.159)
+D22
b
ĒE h
Nxnm = 11 2 (wm,x wn,x + ν12 wm,y wn,y )
2 Ē − ν12
© 2001 by Chapm
ĒE h
Nynm = 11 2 (Ēwn,y wm,y + ν12 wn,x wm,x )
2 Ē − ν12
and
Nxymn = G12 h(wm,x wn,y )
Thus, if we rewrite the first pair of equations in (2.149) in terms of u2
and v2 , and use (2.158), we find that Au2,xx + G12 hu2,yy + Bv2,xy
mπ AA2 mπ 2 nπ 2 2mπx
=− 1
− + ν12 sin
a 4 a b a
mπ A2 mπ 2 nπ 2
− 1
A + (B + G12 h) (2.160)
a 4 a b
2mπx 2nπy
× sin cos
a b
and
ĒAv2,yy + G12 hv2,xx + Bu2,xy
nπ AA2 nπ 2 mπ 2 2nπy
=− 1
−Ē + ν12 sin (2.161)
b 4 b a b
nπ A2 nπ 2 mπ 2
− 1
ĒA + (B + G12 h)
b 4 b a
2nπy 2mπx
× sin cos
b a
where A, B are given by (2.155). The solutions of (2.160), (2.161), sub-
ject to (2.150), (2.151), and the second condition in (2.152), are
Pcr A21 mπ 2 a
u2 = − + x− (2.162)
E11 hb 8 a 2
A2 (mπ/a)2 − ν12 (nπ/b)2 2mπx
− 1 sin
16 (mπ/a) a
mπ 2mπx 2nπy
− sin cos
a a b
and
Pcr ν12 A21 nπ 2
v2 = · − (y − b/2) (2.163)
E11 hb Ē 8 b
2 2
A2 Ē (nπ/b) − ν12 (mπ/a) 2nπy
− 1 sin
16 Ē (nπ/b) b
nπ
2mπx 2nπy
− cos sin
b a b
Pcr
L(D)w3 + w3,xx (2.164)
b
Pcr mπ 2 E11 hA21 mπ 4
= A1 −
b a 16 a
nπ 4 mπx nπy
+Ē sin sin
b a b
nπ 4 −1
16 Pcr mπ 2 mπ 4
A21 = · + Ē (2.165)
E11 h b a a b
and
mπx nπy
w3 (x, y) = A3 sin sin
a b
(2.166)
mπx 3nπy 3mπx nπy
+ A31 sin sin + A32 sin sin
a b a b
© 2001 by Chapm
with
E11 hA31 mπ 4 mπ 4
A31 = D11
16 a a
mπ 2 3nπ 2
+2(D12 + 2D66 ) (2.167a)
a b
4
−1
3nπ
Pcr mπ 2
+ D22 −
b b a
and
4
E11 hA31 nπ 4 3mπ
A32 = D11
16 b a
2
3mπ nπ 2
+2(D12 + 2D66 ) (2.167b)
a b
nπ 4 P 3mπ 2 −1
cr
+D22 −
b b a
3 η 2 A21
where Λ = m/φ, µ2 = Ē − ν12
2
, and
4 h2
$ 4
3 Ā31 Λ4 + 8Ē 2 Ē − 1 / Ē − ν12 2
n + Ē Ā32 n4
Ā3 = $
2 Λ4 + Ēn4 + 2Ē Ē − 1 / Ē − ν12 2 n2 Ēh2 − ν12 Λ2
(2.169)
with
Λ4 Ē G12 −1
Ā31 = ν12 Ē + 2 Ē − ν12 Λ2 n2 +5Ē 2 n4 (2.170a)
2
16 E11
and
Ēn4
Ā32 = (2.170b)
8 9Λ4 − Ēn4
−1
Ē − 1 2
+ 6ν 12 n Ā31
Ē − ν12
2
© 2001 by Chapm
and plates A–F , but for plates L, K, J, I, H, and G the buckling param-
eter first decreases, and then increases again, as Λ is increased. Finally,
in Figs. 2.39 through 2.45, we show some of the (postbuckling) load-
shortening curves from [96]; from the graphs it may be deduced that
postbuckling strength in the case of rectangular orthotropic plates de-
pends on the ratios Ē, G12 /E11 , ν12 , and the parameter Λ = m/φ,
where m is the number of half-sine waves in the x-direction which are
present in the buckling mode. In particular, plates G–E (for which
Ē > 1) have a higher postbuckling strength as compared with plates
A–F (for which Ē < 1). For Λ = 1, plates E, F, I, J, K, and L pos-
sess distinct postbuckling strengths, whereas the postbuckling behavior
of plates A–D is the same as that of the isotropic plate; however, as Λ
increases, the postbuckling strengths (load-shortening curves) of plates
A–F are very different from that of the isotropic case (i.e., Fig. 2.41).
The graph in Fig. 2.43 is a plot of non-dimensional effective width
against non-dimensional load P/Pcr for several of the cases in Table 2.5.
The last of this series of graphs, i.e., Fig. 2.45, displays some results
from [96] for the isotropic plate that show the change in the nature
of the postbuckling curves (plotted as non-dimensionalized load against
non-dimensionalized end shortening) as Λ is varied; the authors [96] in-
dicate that the results displayed here are very close to those obtained
earlier by Stein [93] for postbuckling of an isotropic plate.
∂H(E)
σij = (2.174)
∂Eij
with the strain-displacement relations
1 ∂ui ∂uj ∂uk ∂uk
Eij = + + (2.175)
2 ∂xj ∂xi ∂xi ∂xj
In (2.172) - (2.175), x1 = x, x2 = y, x3 = z, the ui are the displace-
ment components and H is the strain-energy function. In general, the
ui in (2.175) and, thus, the Eij may depend on z. In [73] the authors
take the middle surface of the undeformed plate to be given by x3 = 0,
the thickness to be h, and introduce a lateral length scale L, so that
for a thin plate $ = h/L is a small parameter. Scaling the coordinates,
displacement components, and strain components as in the von Kar-
man theory (i.e., x1 , x2 = O(L), x3 = O(h), u1 , u2 = O($h), u3 =
O(h), E11 , E22 , E12 , E33 = O($2 ), and E13 , E23 = O($3 )) it turns out
that the lowest order terms in the strain-displacement relations (2.175)
can be written in the form
E11 = $1 − K1 x3
E22 = $2 − K2 x3 (2.176)
E12 = $12 − K12 x3
© 2001 by Chapm
while
1 2
$1 = u,x + w,x
2
1 2
$2 = v,y + w,y (2.177)
2
$ = 1 (u + v + w w )
12 ,y ,x ,x ,y
2
∂H(E)
σ11 (2.179)
∂E11
'
'
If we set H̃($1 , $2 , $3 ) = H(E)' , then
x3 =0
' '
∂H '' 1 ∂ 2 H '' 2
H(E) H̃($1 , $2 , $3 ) + x3 + x (2.180)
∂x3 '0 2 ∂x23 '0 3
where the zero subscripts denote evaluation at x3 = 0. Then,
' ' '
∂H '' ∂H '' ∂H ''
= · (−K1 ) + · (−K2 )
∂x3 '0 ∂E11 'x3 =0 ∂E22 'x3 =0
'
∂H ''
+ · (−K12 ) (2.181)
∂E12 'x3 =0
∂ H̃ ∂ H̃ ∂ H̃
− K1 − K2 − K12
∂$1 ∂$2 ∂$12
On the other hand, for |x3 | small we have, by virtue of (2.179)
' 2 '
∂ H̃ ∂ ∂H '' 1 ∂ ∂ H ''
σ11 + ' x + x23 (2.182)
2 ∂$1 ∂x23 '0
3
∂$1 ∂$1 ∂x3 0
in which case
h/2
Mx ≡ σxx ζdζ
−h/2
'
h/2
∂ H̃ h/2
∂ ∂H ''
ζdζ + ζ 2 dζ
−h/2 ∂$1 −h/2 ∂$1 ∂x3 '0
' (2.183)
1 h/2
∂ ∂ 2 H ''
+ ζ 3 dζ
2 −h/2 ∂$1 ∂x23 '0
'
h/2
∂ ∂H ''
≡ ζ 2 dζ
−h/2 ∂$1 ∂x3 '0
Finally, if we use (2.181) in (2.183), we have
+ 6
h/2
∂ 2 H̃ ∂ 2 H̃ ∂ 2 H̃
Mx = −κ1 2 − κ2 − κ12 ζ 2 dζ
∂$1 ∂$1 ∂$2 ∂$1 ∂$12 −h/2
or
h3
Mx = − κ1 H̃11 + κ2 H̃12 + κ12 H̃13 (2.184)
12
where the H̃ij are the bending stiffness moduli, H̃11 = ∂ 2 H̃/∂$21 ,
H̃12 = ∂ 2 H̃/∂$1 ∂$2 , H̃13 = ∂ 2 H̃/∂$1 ∂$12 , etc. The other two moment-
curvature relations are
h3
My = − κ1 H̃12 + κ2 H̃22 + κ3 H̃23 (2.185)
12
© 2001 by Chapm
h3
Mxy = − κ1 H̃13 + κ2 H̃23 + κ12 H̃33 (2.186)
24
As noted in [73] , linear isotropic plate theory corresponds to the
(approximate) strain-energy
E E
H̃ = ($1 + $2 )2 − ($1 $2 − $212 ) (2.187)
2(1 − ν 2 ) 1+ν
E1
H̃ = ($2 + ν21 $1 $2 )
2(1 − ν12 ν21 ) 1
E2 (2.188)
+ ($2 + ν12 $1 $2 )
2(1 − ν12 ν21 ) 2
+2G12 $212
We also note that, for the constitutive theory at hand, it follows directly
that the averaged stresses are
∂ H̃ ∂ H̃
Nx = h ∂$1 , Ny = h ∂$2
(2.190)
1 ∂ H̃
Nxy = Nyx = h
2 ∂$12
and explicit expressions for the H̃ij in (2.185) - (2.187), in terms of H̃ (e)
and H̃ (e), may be computed as well. For unaxial extension in the y di-
rection, Nx = 0, which implies that $1 = −ν21 $2 ; also, in that particular
−1
situation, it follows from (2.192), (2.193), that e = (ν12 − ν21 )$22 and,
−1
as $12 = 0, Nxy = Nyx = 0, while Ny = 2h(ν12 − ν21 )H̃ (e)$2 . Explicit
forms for the H̃ij , for this case, are also computed in [73], i.e.
−1
H̃11 = 2ν21 H̃ (e)
−1
H̃22 = 4(ν12 −1
− ν21 )2 H̃ (e)$22 + 2ν12 H̃ (e)
(2.194)
H̃12 = 2H̃ (e), H̃33 = 2cH̃ (e)
H̃ = H̃ = 0
13 23
© 2001 by Chapm
In (2.195) the parameters c1 and c2 must be determined through an ex-
perimental data fit. For uniaxial deformation in the x-direction, (2.195)
leads to * *
ν12 c2 ν12
σxx ($1 ) = c1 tanh $1 (2.196)
ν21 c1 ν21
while for uniaxial deformation in the y-direction one obtains
Finally, if we employ the form (2.195) of the strain energy function H̃(e),
in (2.194), then for compression in the y-direction of the plate one finds
that the bending stiffness moduli are given by
c1 ν12 /ν21 tanh(c2 $2 /c1 )
H̃11 = ·
1 − ν12 ν21 $2
c2 ν12 ν21 c1 tanh(c2 $2 K1 )
H̃22 = + ·
cosh 2
(c $ /c ) 1 − ν12 ν21 $2
2 2 1
Using (1.45), and (2.184) - (2.186), and assuming that the middle surface
strain field is homogeneous, so that the H̃ij in (2.198) are independent of
x and y, we obtain as the equation governing the initial buckling mode
It is noted in [73] that the even solutions lead to lower buckling loads
than the odd solutions; the imposition of support conditions along the
edges that are parallel to the x-axis, in an (assumed) long plate, then
yields transcendental characteristic equations for the buckling strain as a
function of the wave number λ and the critical buckling strain is obtained
by minimizing the buckling strain with respect to λ. If we introduce
the dimensionless (buckling) strain $̂ = c2 $2 /c1 , and the dimensionless
2
stiffness coefficient S = (c2 /c1 ) (h/2a) ν12 /ν21 , where a is the half-
width of the plate in the x direction, then Fig. 2.46 depicts a plot
of $̂ as
√
a function of S for three mean Poisson’s ratios ν = ν12 ν21 and two
different edge support conditions along the non-loaded edges at x = ±a;
similarly, Fig. 2.47 is a plot of the corresponding normalized buckling
© 2001 by Chapm
stress σ̂ against S for the same three mean Poisson’s ratios and the same
two-edge support conditions along x = ±a. The ratio of the clamped
edge buckling stress to the simply supported buckling stress is graphed
in Fig. 2.48 as a function of S for the same three values of the mean
Poisson ration ν. Finally, in Fig. 2.49 we depict, for the same type of
edge supports along x = ±a and the same mean Poisson’s ratios, the
predicted variation with S of the ratio of the buckling stress given by
the nonlinear elastic plate theory of Johnson-Urbanik to the buckling
stress for a linear elastic, rectilinearly orthotropic plate, with the same
E1 , E2 , ν12 , ν21 , and G12 .
© 2001 by Chapm
As in the case of elastic buckling, (2.210) must now be solved subject
to known support conditions along the edges. For a long plate whose
unloaded edges are simply supported Stowell [102] reports the critical
plastic stress for one transverse half-wave at buckling as
2
π 2 Es h 1 3 Et 1
λP
c = + + φ 2
+ 2 (2.211)
9 b 4 4 Es φ2
where φ = a/b is, once again, the aspect ratio of the plate. It is noted
in [102] that the formula in (2.211) is similar to the one obtained for
elastic buckling of a compressed plate which is simply supported along its
unloaded edges, except that ν is replaced by 1/2, E by Es , and a factor
multiplying the term 1/φ2 has been introduced. In fact, as indicated
in
[102], the ratio of the minimum plastic buckling stress λP c min to the
minimum elastic buckling stress (λec )min , in this case, is given by
P + * 6
λc min 4Es 1 1 1 3Et
η= = + + (1 − ν 2 ) (2.212)
(λec )min 3E 2 2 4 4Es
Much of the more recent work in plastic buckling, i.e., Hutchinson [17],
Tvergaard [103], Hutchinson [45], [46], and Needleman and Tvergaard
[104] is based on a small strain theory of plasticity in which the relation
between the stress rates and strain rates is assumed to be of the form
1 2
J2 = Sij Sij 1/3 σ11 2
+ σ22 2
+ 3σ12 − σ11 σ22 (2.218)
2
1
where Sij = σij − σkk δij is the stress deviator. The instantaneous
3
moduli for J2 flow theory with isotropic elastic properties, are given,
e.g., in Hutchinson [17] as
E 1
Lijk = (δik δj + δi δjk )
1+ν 2
(2.219)
ν Fδij Skl
+ δij δkl −
1 − 2ν 1 + ν + 2FJ2
where
and
© 2001 by Chapm
3 E
F= − 1 , for J2 = (J2 )max and J˙2 ≥ 0,
4J2 Et
i.e., for loading in the plastic range
dG
where G = ; for unloading G = G = 0, while for loading, G =
dJ 2
E
3 − 1 . The relations (2.219), (2.220), however, were proposed in
Es
conjunction with the problem of buckling of a complete spherical shell
under external pressure. For the specific problem of buckling of square
elastic-plastic plates under axial compression, Needleman and Tvergaard
[104] assume (2.217), with the elastic tensor of moduli Leijk isotropic,
i.e.,
e E 1 ν
Lijk = (δik δj + δi δjk ) + δij δk (2.221)
1+ν 2 1 − 2ν
Sij Sk
Lijk = Leijk − H (2.222)
σe2
where *
3
σe = Sij Sij (2.223)
2
and
0, if σe < Y or σ̇e < 0
H=
3 E 1 − Et /E
· , if σe = Y and σ̇e ≥ 0
2 1+ν
(1 + ν) Et + 1 − (Et /E)
2
3 E
(2.224)
In (2.224), Y is the flow (yield) stress which is taken to be the greater of
the maximum value of σe (over the stress history) and the (initial) yield
stress σy associated with uniaxial stress-strain behavior; the uniaxial
stress-strain behavior chosen is that of a piecewise power law with yield
stress σy and a continuous tangent modulus, i.e.
σ/σy , for σ < σy
$
= n (2.225)
$y
1 σ 1
+ (1 − ), for σ ≥ σy
n σy n
© 2001 by Chapm
and the initial slope λ1 of the λ(ξ) curve is uniquely determined by the
condition that plastic loading takes place through the plate except at
one point where neutral loading occurs; in accordance with the origi-
nal concept of Shanley [101], the slope λ1 > 0, so that buckling takes
place under increasing load. After buckling, a region of elastic unload-
ing spreads out from the point of neutral loading and the next term in
the postbuckling expansion for λ takes the penetration of this region of
elastic unloading into account and turns out to be proportional to ξ 4/3
(e.g., Hutchinson [17]):
λ = λc + λ1 ξ + λ2 ξ 4/3 + . . . (2.229)
In all cases considered, the authors of [104] determine that λ2 < 0 and,
thus, the first three terms of (2.229) can be used to estimate the value of
the maximum stress λmax that can be supported by the plate. In [104]
one also finds an analysis of the imperfection sensitivity of compressed,
simply supported, elastic-plastic square plates which ignore elastic un-
loading; we will not go into any details of the imperfection-sensitivity
analysis here but do note the following: attention in [104] is confined to
imperfections which have the shape of the buckling mode (2.227) with
the amplitude of the imperfection denoted by ξ. ¯ For λ < λc , a regular
perturbation analysis shows that the lowest order effect of the imperfec-
tion on the magnitude ξ, of the out-of-plane deflection of the plate, is
given by
λmax
= 1 − µ(ξ)2/(2Ψ+1) (2.232)
λc
with
−2Ψ/(2Ψ+1)
λ2 2Ψλ2
µ = − (2Ψ + 1) − · p(λc )2/(2Ψ+1) . (2.233)
λc λc
The results of the analysis presented in [104] for three different cases are
depicted in Figs. 2.51, 2.52, and 2.53.
Finally, we turn our attention to the analysis of the initial and post-
buckling behavior of rectangular viscoelastic plates; a formulation of the
buckling equations which are suitable for dealing with this problem may
be found in the work of Brilla [106] although only the stability problems,
both linearized and nonlinear are dealt with. The buckling of circular
viscoelastic plates has been treated in [140] and will be discussed in the
next chapter. The governing equations for the large deflection theory of
viscoelastic plates considered in [106] are of two types, the first of which
is of the form
h3
Kijk (D)w,ijk
12 (2.234)
= K(D)(q + h$ik $j w,ij Φ,k )
with i, j, k, = 1, 2, and
∂
are polynomial operators in D = , the Lijk (D), and K(D), have
∂t
analogous representations, and
−1
K(D) [Kijk (D)] = L(D)−1 Lijk (D) (2.237)
The operators K(D), L(D), Kijk (D), and Lijk (D) are the differential
operators of linear viscoelasticity, w and Φ are, of course, the transverse
© 2001 by Chapm
plate deflection and Airy function, respectively, q is the transverse plate
loading, h the plate thickness, and $ij the standard alternating tensor.
∂
By w,i we mean w, etc., and we employ the usual convention of
∂xi
summing over repeated indices. The coefficients Kijk are assumed to
possess the symmetries given by
(δ) (δ) (δ) (δ)
Kijk = Kjik = Kijk = Kkij (2.238)
and satisfy
(δ)
Kijk (D)$ij $k ≥ 0 (2.239)
with equality holding only for =0. The polynomials Kijk and L are
assumed to have only negative real roots. The boundary conditions
considered in [106] are either of the form
∂w
w= = 0, on ∂Ω, (2.240)
∂n
∂Ω being the boundary of the domain Ω occupied by the thin plate (in
which (2.234), (2.235) are to hold), or
where the νkn = cos(xk , n) are the direction cosines of the (unit) out-
ward normal n to ∂Ω. Also, with respect to the Airy function Φ, we
have either
∂Φ ∂3Φ
= = 0, on ∂Ω (2.242)
∂n ∂n3
or
∂2Φ ∂2Φ
= = 0, on ∂Ω (2.243)
∂n2 ∂s∂n
∂ ∂
where , denote, respectively, the normal and tangential derivatives
∂n ∂s
on ∂Ω. Finally, if the differential operator Lijk (D) is of order k, then
the initial conditions for the problem assume, in general, the form
γ '
∂ w ''
= wγ (γ = 0, 1, . . . , k − 1)
∂tγ '
t=0
' (2.244)
∂ γ Φ ''
= 0 (γ = 0, 1, . . . , k − 1)
∂tγ '
t=0
A second set of governing equations considered in [106] are those ap-
propriate for the large deflection theory of viscoelastic plates of Boltz-
mann type, namely,
h3 t ∂
Gijk (t − τ ) w,ijk (τ )dτ
12 0 ∂τ (2.245)
= q + h$ik $j w,ij Φ,k
and t
∂
$im $jn $kr $s Jmnrs (t − τ ) Φ,ijk (τ )dτ
0 ∂t (2.246)
1
= − $ik $j w,ij w,k
2
The boundary and initial conditions associated with the system (2.245),
(2.246) are of the same type as those delineated, above, for the system
(2.234), (2.235).
We begin our discussion of the analysis in [106] by detailing the de-
scription of the linearized stability problem, i.e., the problem of initial
buckling. Since we are dealing now with a stability problem for a time-
dependent process, we must consider perturbations from an equilibrium
state; in the usual manner, we neglect the nonlinear terms in (2.234)
and (2.235) when we address the linearized stability problem. We as-
sume that, as a consequence of the loading and boundary conditions,
◦
the plate is subjected to an in-plane stress distribution of the form λσij .
◦ ◦
With Nij = −hσij , we have as a consequence of (2.234), with q ≡ 0, the
following equation governing the initial buckling of the plate within the
context of the (linear) viscoelasticity model of differential type:
h3 ◦
Kijk (D)w,ijk +λNij w,ij = 0 (2.247)
12
Equation (2.247) is subject to boundary conditions of the type (2.240)
or (2.241) and initial conditions
'
∂ γ w ''
w|t=0 = w0 , = 0, γ = 1, 2, · · · , r − 1 (2.248)
∂tγ 't=0
If we seek a solution of (2.247) in the form
w(x, y, t) = eµt u(x, y) (2.249)
then it is easily seen that u(x, y) must satisfy the equation
h3 # (γ) γ #
r s
◦
Kijk µ u,ijk +λ Kγ µγ Nij u,ij = 0 (2.250)
12 γ=0 γ=0
© 2001 by Chapm
A nontrivial solution u(x, y) of (2.250) exists only for values µ = µn ,
the generalized eigenvalues of the linearized problem. Specifically, if we
assume that r = s, then (2.250) becomes
# r 3
γ h (γ) ◦
µ K u,ijk +λKγ Nij u,ij = 0 (2.251)
γ=0
12 ijk
it follows that, for sufficiently small λ, there exists κ > 0 such that
# r 3
γ h (γ) ◦
µ Kijk (φ,ij , φ,k ) − λKγ Nij (φ,i , φ,k ) ≥ κ
φ
2 (2.254)
γ=0
12
where
φ
2 = (φ, φ).
The polynomial on the left-hand side of (2.254) has positive coeffi-
cients for λ sufficiently small and is a monotonically increasing function
for µ > 0; this, in turn, implies that the roots of this polynomial are
either negative or have negative real parts; for a real viscoelastic mate-
rial the roots must, in fact, all be negative. If, in addition, the roots are
simple, then the solution of (2.247) can be expressed in the form
∞ #
# r
w(x, y, t) = Ank won φn (x, y)e−µnk t (2.255)
n=1 k=1
where the (real) roots of (2.252) have been denoted by −µnk and won =
(w0 , φn ). It can be shown that the coefficients Ank in the expansion
(2.255) are given by [106]:
µn1 µn2 · · · µn(k−1) µn(k+1) · · · µnr
Ank =
(µnk − µn1 ) · · · (µnk − µn(k−1) )(µnk − µn(k+1) ) · · · (µnk − µnr )
(2.256)
With all the µnk > 0 the solution is stable, while if at least one
µnk < 0 then the solution is unstable. As the left-hand side of (2.252) is
a continuous, increasing function of λ, when λ increases critical values
of λ are reached at which the roots of (2.252) successively change their
signs; to determine these critical roots the polynomial equation (2.252),
which can be written in the form
#
r
µγn Anγ (λ) = 0 (2.257)
γ=0
h3 (0)
K (φn,ij , φn,k ) − λK0 Ny0 (φn,i , φn,j ) = 0 (2.259)
12 ijk
and
h3 (r)
K (φn,ij , φn,k ) − λKr Nij0
(φn,i , φn,j ) = 0 (2.260)
12 ijk
If the Laplace transform is applied to (2.247), we find that the values of
λ, in question, are eigenvalues of the system
h3 (0) 0
K w,ijk (∞) + λNij K0 w,ij (∞) = 0 (2.261)
12 ijk
and
h3 (r) 0
K w,ijk (0) + λNij Kr w,ij (0) = 0 (2.262)
12 ijk
Whenever λ is an eigenvalue of (2.261), one of the roots of (2.252) is
equal to zero and, as λ increases above this value, one of the µnk becomes
negative; when λ is an eigenvalue of (2.262) one of the roots of (2.252)
must be infinite. In [106], eigenvalues of (2.261) are called critical values
for infinite critical time and are denoted by λcr . For λ < min λcr each of
the µnk > 0 and the basic solution of (2.47) is stable. For λ = min λcr
we have neutral stability, while for λ > min λcr at least one of the
µnk < 0 and the solution is said to be unstable (with infinite critical
time). On the other hand, eigenvalues of (2.262) are said to be critical
© 2001 by Chapm
values of instant instability, or critical values for finite critical time, and
are denoted as λ0cr . When λ achieves the value min λ0cr the viscoelastic
plate becomes (instantly) unstable.
The results delineated above hold for viscoelastic plates of “differential
type,” i.e., those which are governed by the large deflection equations
(2.234), (2.235); for the case of viscoelastic plates of the “Boltzmann
type,” namely, those for which large deflection relations of the form
(2.245), (2.246) are valid, the critical values are shown in [106] to corre-
spond to the eigenvalues of the system
0
Gijk (∞)w,ijk (∞) + λNij w,ij (∞) = 0
(2.263)
0
Gijk (0)w,ijk (0) + λNij w,ij (0) = 0
and
1
(1 + βD) 2 Φ̃ = − E(1 + αD)[w, w] (2.265)
2
where Φ0 (x, y) satisfies 2 Φ0 = 0, with boundary conditions that cor-
respond to the given loading condition(s) on the edge(s), K is the plate
bending stiffness, E the Young modulus, and α, β constants which serve
to characterize the viscoelastic behavior of the plate. For α = β = 0,
(2.264), (2.265) reduce to the usual set of von Karman equations for a lin-
early elastic, isotropic plate, in Cartesian coordinates, with Φ = λΦ0 + Φ̃
the (total) Airy function. Using the boundary conditions (2.240), or
(2.241), and the initial conditions
and
1
2 Φ̃ = − E[w, w] (2.268)
2
The result referenced above may be deduced as follows: we rewrite equa-
tions (2.264), (2.265) in the form
α 2 α 1 t 2 − β1 (t−τ )
K w+ 1− w e dτ
β β β 0
α
− 2 w0 · e− β t
1
(2.269)
β
= h λ [Φ0 , w] + Φ̃, w − λ [Φ0 , w] e− β t
1
and
β 2 β 1 t 2 − 1 (t−τ )
Φ̃ + 1 − Φ̃ e α dτ
α α α 0
(2.270)
1
= − E [w, w] − [w0 , w0 ] e− α t
1
2
Taking the scalar product of (2.269) with w, and then using [w, w] as
given by (2.270), it is possible to show that
t
α α 1 −β1
K (w, w) + 1 − w, (w) e (t−τ )
dτ
β β β
0
2h β
−hλ (Φ0 , [w, w]) + Φ̃, Φ̃
E α
t 1
β 1
+ 1− Φ̃, Φ̃ e− α (t−τ ) dτ
α α
α
0 1
= K (w0 , w0 ) e− β t − hλ (Φ0 , [w0 , w]) e− β t − h Φ̃, [w0 , w0 ] e− α t
1 1
β
(2.271)
from which, with the assumption of finite total energy, we obtain, in the
limit, as t → ∞
2
K (w, w) − hλ (Φ0 , [w, w]) + h Φ̃, Φ̃ = 0 (2.272)
E
Because (Φ̃, Φ̃) ≥ 0, a nonzero solution of (2.272) exists if and only
if λ is larger than the critical values (with infinite critical time) of the
linearized problem, which are also critical values of the non-linear prob-
lem (2.267), (2.268). If w0 = 0, the right-hand side of (2.271) is equal
to zero and, setting t = 0, we then find that
2β 2
Kα (w, w) − hλβ (Φ0 , [w, w]) + h Φ̃, Φ̃ = 0 (2.273)
Eα
© 2001 by Chapm
Also, the corresponding reduced (generalized) von Karman equations
(which are obtained from (2.269), (2.270)), are
Kα w = βh λ [Φ0 , w] + Φ̃, w
2
(2.274)
β 2 Φ̃ = − 1 Eα [w, w]
2
It is not difficult to show that (2.273), and, thus, also the system (2.274)
can have a non-zero solution if and only if
which leads to the following conclusion: the critical values for zero criti-
cal time, which are the eigenvalues of (2.274), are critical values for zero
critical time of the linearized problem. No numerical computations for
the critical values or the corresponding eigenfunctions (buckling modes)
have been carried out in [106]; there exists no postbuckling analysis ei-
ther, so that these are likely subjects for future research work in this
area.
© 2001 by Chapm
along the edges y = 0, y = 0, 0 ≤ x ≤ a, affects the variation with
φ of the normalized loading λ/λcr , in the x-direction, for a rectangular
isotropic plate compressed in both directions; the parameter along the
curves is the compressive stress ξ normalized by λcr . Here λcr denotes
the critical value of the compressive stress in the x-direction for the
given support conditions in the absence of the compressive stress ξ;
among the results that may be gleaned from Figs. 2.11 and 2.12 are
the following: fix the value of ξ/λcr , say, ξ/λcr = 1.0; then for φ = 1
(square plate), λ/λcr < 1 for the plate which is simply supported along
the edges at x = 0, x = a, 0 ≤ y ≤ b and clamped along the edges at
y = 0, y = b, 0 ≤ x ≤ a, while λ/λcr > 1 (in fact, approximately 1.8)
for the plate which is simply supported along all four edges. Also, in the
first case, the plate buckles into one-half wave in the x-direction, while in
the second case the plate buckles into two half-waves in the x-direction.
Figure 2.19 clearly displays the comparative results for shear buckling
of a linearly elastic isotropic rectangular plate for four different kinds of
edge support conditions; all the graphs of K̃ against φ are monotonically
decreasing and as the graphs do not intersect for φ > 1 it is sufficient
to display, as we have done in Table 2.9, the results for just one aspect
ratio, say φ = 2. Once again, we see an increase in the buckling load as
we move from edges which are all simply supported to edges which are
pairwise simply supported and clamped to the case where all edges are
clamped.
For linearly elastic rectangular plates which are orthotropic Figs. 2.23
and 2.24 display the variation of the buckling constant k (for a plate
which is
3 compressed in the x-direction) with the effective aspect ratio
β̄ = φ 4 DD11 . The parameter α along the curves is defined in (2.65),
22
© 2001 by Chapm
process of going from isotropic to (rectilinearly) orthotropic but also the
process of going from one degree of orthotropy, as measured, e.g., by the
ratio Ē = E22 /E11 , to another.
The most revealing of the graphs presented in [96] that are related to
initial buckling is depicted in Fig. 2.38; this figure depicts the variation
in buckling load, as a function of Λ = m/φ, as one varies the orthotropic
behavior of the plate among cases A-L in Table 2.5. The curves in
Fig. 2.38 turn out to be ordered from bottom to top (F, E, D, C, B, A),
isotropic, (G, H, I, J, K, L) according to increasing Ē = E22 /E11 ; for
fixed m and φ the buckling load increases as the ratio of the Young’s
moduli E22 to E11 increases. In Figs. 2.33 through 2.37, we show the
initial buckling graphs of critical stress versus aspect ratio a/b as Ē
increases through the values Ē = .072, .51, 1.0 (isotropic), 1.97, and
13.73, respectively; these graphs again illustrate the increase in buckling
load with aspect ratio for a wide range of buckling wavelengths in the
direction of compression; perhaps, more importantly, they depict the
aspect ratios at which a transition from m = 1 to m = 2 half-waves
occurs. From Table 2.11 and the graphs in Figs. 2.33 through 2.37, it
should be clear that, as Ē increases, the transition from m = 1 to m = 2
half-waves in the direction of compression of the plate comes at smaller
and smaller values of the aspect ratio.
Variations in postbuckling behavior for linearly elastic rectangular
orthotropic plates may be gauged from the graphs in Figs. 2.39 through
2.42, which depict normalized load versus normalized shortening (or
contraction) curves for a simply supported plate in compression; in Figs.
2.39–2.41, Λ = m/φ varies over the values 1.0, 1.33, and 2.0, respectively,
while, in Fig. 2.42, the results are for an infinitely long plate. The
values of Ē for the plates depicted are L (13.73), K (7.6), J (3.22), I
(1.97), A (.83), B (.70), C (.506), D (.303), E (.131), and F (.072).
For the case Λ = 1 one cannot distinguish the postbuckling behavior
for plates A, B, C, D, and an isotropic plate (Ē = 1.0); otherwise, it
is clear from the graphs in the first three figures in this group that
the steepness of the postbuckling curves decrease as the value of Ē =
E22 /E11 decreases, i.e., for a fixed load, beyond the critical load, the
normalized shortening increases as the value of Ē decreases. For any
given plate, the postbuckling curves, for the corresponding values of
Ē, exhibit a decrease in steepness as the value of Λ increases. Note
that, for the infinitely long plate, it is clear from Fig. 2.42 that the
steepness of the load-shortening curves decreases as the value of Ē, for
the plate, increases; thus for the compressed, simply supported, infinitely
long plate, and a given load beyond the critical load, the shortening
(contraction) of the plate increases as the ratio Ē of E22 to E11 increases.
In Fig. 2.43 we depict the monotonic decreases in the effective widths of
simply supported, compressed rectangular plates as the load increases
beyond the critical load; we note that the steepness of the curves (or the
rate of decrease of effective width with increasing load) increases as the
value of the ratio Ē(= E22 /E11 ) decreases. For a given load, beyond
the critical load, a smaller effective width is obtained for smaller values
of Ē.
2
π 2 Es h 1 3 Et 1
λP
c = + 2
+φ +2
9 b 4 4 Es φ2
© 2001 by Chapm
where φ = a/b is, again, the aspect ratio of the plate while Et and
Es are, respectively the tangent and secant moduli associated with the
elastic-plastic behavior of the plate. To compare the critical stress levels
required to initiate plastic and (linear) elastic buckling, we have the rela-
tion (2.212) for the plasticity reduction factor η, which holds in the case
where the plate is simply supported along its unloaded edges, namely,
+ * 6
(λP
c )min 4Es 1 1 1 3Et
η≡ e = + + (1 − ν 2 ).
(λc )min 3E 2 2 4 4Es
© 2001 by Chapm
Table 2.3 Values of Coefficient k for Plywood Plates De-
flected by Bending Moment [51] (From Lekhnitskii, S.G.,
Anisotropic Plates, Gordon and Breach Pub. Co., N.Y., 1968
and OPA, N.V. With permission)
Table 2.5 Elastic Constants for the Various Cases of Composite Plates A-N [96] (Reprinted from the
Int. J. Mech. Sci., 15, Chandra, R. and Raju, B., Postbuckling Analysis of Rectangular Orthotropic Plates,
81-97, 1973, with permission from Elsevier Science)
Designation of
G12
different cases E11 E22 G∗12 Ē = E22 /E11 ν21 ν12
E11
A 1.887 1.5772 0.6123 0.8358 0.3245 0.3 0.25074
B 3.761 2.648 1.061 0.7041 0.2823 0.3 0.21123
C 7.068 3.578 1.409 0.5064 0.1994 0.3 0.15192
D 14.130 4.282 1.714 0.3031 0.1213 0.3 0.09093
E 36.390 4.788 1.959 0.1315 0.0538 0.3 0.03945
F 69.660 5.071 2.062 0.0728 0.0296 0.3 0.02184
G 1.5772 1.87 0.6123 1.1964 0.3882 0.25074 0.3
H 1.648 3.761 1.061 1.4202 0.4009 0.21123 0.3
I 3.578 7.068 1.409 1.9747 0.3937 0.15192 0.3
J 4.282 14.13 1.714 3.2292 0.4002 0.09093 0.3
K 4.788 36.39 1.959 7.6045 0.4091 0.03945 0.3
L 5.071 69.66 2.062 13.7362 0.4065 0.02184 0.3
M 2.6 7.5 2.1 2.89 0.424 0.08675 0.25
N 7.5 2.6 1.1 0.347 0.147 0.25 0.08673
Table 2.6 Buckling Parameter K̃(∼ λcr ) for a Linearly Elas-
tic Isotropic Rectangular Plate Compressed in the x-Direction;
the Aspect Ratio is φ = 1. Based on the Data Presented in
[32].
© 2001 by Chapm
Table 2.8 Transition from m = 1 to m = 2 Half-Sine Waves
in the x-Direction for Initial Buckling of a Compressed, Lin-
early Elastic Isotropic Rectangular Plate. Based on the Data
Presented in [32].
Loaded Longitudinal
K̃ φ(m = 1 → m = 2)
Edges Edges
.072 2.72 .8
© 2001 by Chapm
FIGURE 2.1
Buckling of simply supported plates [32]. (From Bulson, P.S.,
The Stability of Flat Plates, American Elsevier, N.Y., 1969)
FIGURE 2.2
K̄ ∼ φ curves for long plates [32]. (From Bulson, P.S., The
Stability of Flat Plates, American Elsevier, N.Y., 1969)
© 2001 by Chapm
FIGURE 2.3
Buckling of plates with built-in longitudinal edges [32]. (From
Bulson, P.S., The Stability of Flat Plates, American Elsevier,
N.Y., 1969)
FIGURE 2.4
One longitudinal edge built-in, one simply supported [32].
(From Bulson, P.S., The Stability of Flat Plates, American
Elsevier, N.Y., 1969)
© 2001 by Chapm
FIGURE 2.5
One longitudinal edge built-in, one free [32]. (From Bulson,
P.S., The Stability of Flat Plates, American Elsevier, N.Y.,
1969)
FIGURE 2.6
One longitudinal edge simply supported, one free [32]. (From
Bulson, P.S., The Stability of Flat Plates, American Elsevier,
N.Y., 1969)
© 2001 by Chapm
FIGURE 2.7
Built-in loaded edges. (Adopted, in modified form, from [75].)
FIGURE 2.8
All edges built in. (Adopted, in modified form, from [76].)
© 2001 by Chapm
FIGURE 2.9
Rectangular plate compressed in both directions.
FIGURE 2.10
Kx for simply supported edges [32]. (From Bulson, P.S., The
Stability of Flat Plates, American Elsevier, N.Y., 1969)
© 2001 by Chapm
FIGURE 2.11
Relationship between λ and ξ [78].
FIGURE 2.12
Relationship between λ and ξ [78].
© 2001 by Chapm
FIGURE 2.13
Linearly varying edge forces [32]. (From Bulson, P.S., The
Stability of Flat Plates, American Elsevier, N.Y., 1969)
FIGURE 2.14
Simply supported plates with linearly varying edge forces [32].
(From Bulson, P.S., The Stability of Flat Plates, American
Elsevier, N.Y., 1969)
© 2001 by Chapm
FIGURE 2.15
Longitudinal edges built-in. (Adopted, in modified form, from
[79].)
FIGURE 2.16
Plate subjected to a combined loading.
© 2001 by Chapm
FIGURE 2.17
Simply supported plates in shear. (Adopted, in modified form,
from [80].)
FIGURE 2.18
Built-in plates in shear. (Adopted, in modified form, from
[81].)
© 2001 by Chapm
FIGURE 2.19
Shear buckling coefficient for various edge conditions [32].
(From Bulson, P.S., The Stability of Flat Plates, American
Elsevier, N.Y., 1969)
FIGURE 2.20
Shear and compression, restrained edges, φ < 1. (Adopted, in
modified form, from [83].)
© 2001 by Chapm
FIGURE 2.21
Compression along the external fibers [51]. (From Lekhnitskii,
S.G., Anisotropic Plates, Gordon and Breach Pub. Co., N.Y.,
1968 and OPA, N.V. With permission)
FIGURE 2.22
Compression across the external fibers [51]. (From Lekhnitskii,
S.G., Anisotropic Plates, Gordon and Breach Pub. Co., N.Y.,
1968 and OPA, N.V. With permission)
© 2001 by Chapm
FIGURE 2.23
Coefficients of plate buckling in compression, I [43]. (From
Hoff, N.J. and Stavsky, Y., “Mechanics of Composite Struc-
tures”, in Composite Engineering Laminates, A.G.H. Dietz,
ed., MIT Press, Cambridge, MA., 1969, 5-59, with permis-
sion.)
FIGURE 2.24
Coefficients of plate buckling in compression, II [43]. (From
Hoff, N.J. and Stavsky, Y., “Mechanics of Composite Struc-
tures”, in Composite Engineering Laminates, A.G.H. Dietz,
ed., MIT Press, Cambridge, MA., 1969, 5-59, with permis-
sion.)
© 2001 by Chapm
FIGURE 2.25
Shearing of a very narrow or infinite orthotropic strip.
FIGURE 2.26
Coefficients of plate buckling in shear. (Adopted, in modified
form, from [84].)
© 2001 by Chapm
FIGURE 2.27
Orthotropic plates in shear. (Adopted, in modified form, from
[84].)
FIGURE 2.28
Orthotropic plates subjected to a combination of shear and
compressive loading [51]. (From Lekhnitskii, S.G., Anisotropic
Plates, Gordon and Breach Pub. Co., N.Y., 1968 and OPA,
N.V. With permission)
© 2001 by Chapm
FIGURE 2.29
Postbuckling behavior for a square isotropic plate. (Adopted,
in modified form, from [61].)
FIGURE 2.30
Postbuckling deflections of rectangular plate in end compres-
sion [32]. (From Bulson, P.S., The Stability of Flat Plates,
American Elsevier, N.Y., 1969)
© 2001 by Chapm
FIGURE 2.31
Postbuckling deflections. (Adopted, in modified form, from
[94].)
FIGURE 2.32
Postbuckling deflections of plates under a linearly varying end load.
(Reprinted with permission of The Random House Group Limited c
A.H. Chilver, Editor.)
© 2001 by Chapm
FIGURE 2.33
Critical stress versus a/b ratio for plate C(Ē = 0.51) [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
FIGURE 2.34
Critical stress versus a/b ratio for plate F (Ē = 0.072) [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 2.35
Critical stress versus a/b ratio for plate I(Ē = 1.97) [96] .
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
FIGURE 2.36
Critical stress versus a/b ratio for plate L(Ē = 13.73) [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 2.37
Critical stress versus a/b ratio for the isotropic plate (Ē =
1.00) [96]. (Reprinted from the Int. J. Mech. Sci., 15, Chan-
dra, R. and Raju, B., Postbuckling Analysis of Rectangular
Orthotropic Plates, 81-97, 1973, with permission from Else-
vier Science)
FIGURE 2.38
Variation of the buckling load parameter with Λ for various
cases, Λ = m/φ [96]. (Reprinted from the Int. J. Mech.
Sci., 15, Chandra, R. and Raju, B., Postbuckling Analysis of
Rectangular Orthotropic Plates, 81-97, 1973, with permission
from Elsevier Science)
© 2001 by Chapm
FIGURE 2.39
Load-shortening curves for the case Λ = 1.0, Λ = m/φ [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
FIGURE 2.40
Load-shortening curves for the case Λ = 1.33, Λ = m/φ [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 2.41
Load-shortening curves for the case Λ = 2.0, Λ = m/φ [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
FIGURE 2.42
Load-shortening curves for the infinitely long plate [96].
(Reprinted from the Int. J. Mech. Sci., 15, Chandra, R. and
Raju, B., Postbuckling Analysis of Rectangular Orthotropic
Plates, 81-97, 1973, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 2.43
Effective width versus load curves for the case Λ = 1.0, Λ =
m/φ [96]. (Reprinted from the Int. J. Mech. Sci., 15, Chan-
dra, R. and Raju, B., Postbuckling Analysis of Rectangular
Orthotropic Plates, 81-97, 1973, with permission from Else-
vier Science)
FIGURE 2.44
Comparison of load-shortening curves [96]. (Reprinted from
the Int. J. Mech. Sci., 15, Chandra, R. and Raju, B., Post-
buckling Analysis of Rectangular Orthotropic Plates, 81-97,
1973, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 2.45
Load-shortening curves for different values of λ for the isotropic
case [96]. (Reprinted from the Int. J. Mech. Sci., 15, Chan-
dra, R. and Raju, B., Postbuckling Analysis of Rectangular
Orthotropic Plates, 81-97, 1973, with permission from Else-
vier Science)
FIGURE 2.46
Plots of ˆ as a function of S for three mean Poisson’s ratios
and two edge-support conditions [74]. (From Johnson, M. and
Urbanik, T.J., Wood and Fiber Science, 19, 135, 1987, with
permission)
© 2001 by Chapm
FIGURE 2.47
Plots of σ̂ as a function of S for three mean Poisson’s ratios
and two edge-support conditions [74]. (From Johnson, M. and
Urbanik, T.J., Wood and Fiber Science, 19, 135, 1987, with
permission)
FIGURE 2.48
Ratio of fixed-edge buckling stress to simply-supported buck-
ling stress as a function for S for three mean Poisson’s ratios
[74]. (From Johnson, M. and Urbanik, T.J., Wood and Fiber
Science, 19, 135, 1987, with permission)
© 2001 by Chapm
FIGURE 2.49
Ratio of nonlinear elastic buckling stress to linear elastic buck-
ling stress as a function of S for three mean Poisson’s ratios
and two edge-support conditions [74]. (From Johnson, M. and
Urbanik, T.J., Wood and Fiber Science, 19, 135, 1987, with
permission)
FIGURE 2.50
Young’s modulus and tangent modulus for an elastic-plastic
material.
© 2001 by Chapm
FIGURE 2.51
Load versus buckling mode displacement for square plate
(σy /E = 0.00337, h/a = 0.035, n = 10, ν = 0.3) [104].
(Reprinted from the Int. J. Solids and Structures, 12,
Needleman, A. and Tvergaard, V., An Analysis of the Imper-
fection Sensitivity of Square Elastic-Plastic Plates under Axial
Compression, 185-201, 1976, with permission from Elsevier
Science)
FIGURE 2.52
Load versus buckling mode displacement for a Square
plate (σy /E = 0.00337, h/a = 0.035, n = 3, ν = 0.3) [104].
(Reprinted from the Int. J. Solids and Structures, 12,
Needleman, A. and Tvergaard, V., An Analysis of the Imper-
fection Sensitivity of Square Elastic-Plastic Plates under Axial
Compression, 185-201, 1976, with permission from Elsevier
Science)
© 2001 by Chapm
FIGURE 2.53
Load versus buckling mode displacement for a Square
plate (σy /E = 0.00337, h/a − 0.031, n = 10, ν = 0.3) [104].
(Reprinted from the Int. J. Solids and Structures, 12, Needle-
man, A. and Tvergaard, V., An Analysis of the Imperfection
Sensitivity of Square Elastic-Plastic Plates under Axial Com-
pression, 185-201, 1976, with permission from Elsevier Science)
Chapter 3
© 2001 by Chapm
3.1 Plates with Linear Elastic Behavior
Circular plates exhibiting linear elastic behavior may be classified ac-
cording to whether the plate, which is assumed to be thin and with-
out initial imperfections, possesses isotropic, cylindrically orthotropic,
or rectilinearly orthotropic material symmetry.
2 1 1
K[w,rrrr + w,rrr − 2 w,rr + 3 w,r ]
r r r
(3.1)
0 1 0
= Nr w,rr + Nθ w,r
r
1
where Nr0 = (Φ0 ),r , and Nθ0 = (Φ0 ),rr . An alternative form for the
r
full set of von Karman equations, in the case of axially symmetric de-
formations, may be obtained by multiplying both (1.80) and (1.81) by
r and integrating; this yields the system
d 1 d dw dΦ dw
Kr r = + c1 (3.2)
dr r dr dr dr dr
and 2
1 d 1 d dΦ 1 dw
r r = + c2 (3.3)
Eh dr r dr dr 2 dr
At r = 0, the left-hand sides of (3.2), (3.3) vanish and, as a consequence
of the radial symmetry
' of the solutions and the physical requirement of
dw ''
regularity, = 0. Thus, c1 = c2 = 0 in (3.2), (3.3). If we expand
dr 'r=0
the left-hand sides of (3.2), (3.3) we then have, in lieu of (1.80), (1.81),
the equations
1 1 1
K(w,rrr + w,rr − 2 w,r ) = Φ,r w,r (3.4)
r r r
and
1 1 1 1
(Φ,rrr + Φ,rr − 2
Φ,r ) = (w,r )2 (3.5)
Eh r r 2r
With Φ = Φ0 (r) in (3.4), the initial buckling equation for an isotropic
circular plate is now obtained in the form
1 1
K(w,rrr + w,rr − 2 w,r ) = Nr0 w,r (3.6)
r r
If we let ϕ be the angle between the central axis of the plate and the
normal to the deflected surface w = w(r) at any point, then elementary
geometry establishes that ϕ = w,r so that, e.g., (3.6) may be rewritten
in terms of ϕ as
+ o 6
1 Nr 1
ϕ,rr + ϕ,r − + 2 ϕ=0 (3.7)
r K r
One other form of the von Karman equations for axisymmetric buck-
ling of an isotropic, thin, linearly elastic circular plate which is prevalent
in the literature, especially in the work of Friedrichs and Stoker [72],
[143], is obtained by introducing the functions
p = 1 Φ,r (≡ Nr )
r (3.8)
q = − a w,r
r
√
the constant η = K/a, and the linear operator
2
a2 d 1 d 2 d 3 d
G≡ r = a2 + (3.9)
r dr r dr dr2 r dr
η 2 Gq − pq = 0 (3.10a)
© 2001 by Chapm
1 1
· Gp = q 2 (3.10b)
Eh 2
Remarks: In the work of Friedrichs and Stoker, op.cit., the first equa-
tion in the set (3.10) (with the definitions (3.8), (3.9)) has the form
η 2 Gq + pq = 0; the difference in sign appears as a result of defining
1
the Airy function so that, in the radially symmetric case, Nr = Φ,r
r
1
instead of Nr = − Φ,r . If we had defined the Airy function so that
r
1
Nr = − Φ,r , for axially symmetric deformations, then in lieu of (3.4)
r
we would have
1 1 1
K(w,rrr + w,rr − 2 w,r ) = − Φ,r w,r (3.4)
r r r
and, with the definitions of p, q, and G in (3.8), (3.9), the first equation
in (3.10) would read
η 2 Gq + pq = 0 (3.10a)
while the equilibrium equation in (3.7) would become
0
1 Nr 1
ϕ,rr + ϕ,r + − 2 ϕ=0 (3.7)
r K r
dw
with ϕ = .
dr
Remarks: If Φ(r) and w(r) are to have continuous fourth derivatives,
so that (1.80), (1.81) may be interpreted in the classical sense, then the
boundary conditions which must be satisfied at the center of the plate
are shown in [72], [143] to be
' '
dp '' dq ''
= 0, =0 (3.11)
dr 'r=0 dr 'r=0
Remarks: For circular, isotropic, linearly elastic plates the two basic
support conditions that we will consider along the edge at r = a are of
the clamped and simply supported kinds; in the first case, we have, as
a consequence of (1.139)
for 0 < θ ≤ 2π, with the obvious simplifications for the axially symmetric
case.
Remarks: For the case where one considers only rotationally symmetric
buckled states of an isotropic, linearly elastic, circular plate, which is
simply supported along its edge at r = a, and is subject to a uniform
compressive thrust Nr = λh, per unit length, along r = a, which lies in
the plane of the plate, several fairly general results have appeared, most
notably in [72], [144], and [145]; in the first paper it was shown there
is one pair of buckled states with no internal node for all values of the
thrust parameter λ greater than the first eigenvalue (predicted by the
linear buckling theory) and that no other buckled states exist when λ is
less than or equal to the second eigenvalue. In the second of the papers
referenced above, i.e., [140], it was proven that for every positive integer
n, a pair of axially symmetric buckled states with n–1 internal nodes
exists when the thrust λ is slightly larger than λn , the nth eigenvalue
of the linear buckling problem. In [145], Wolkowisky uses the Schauder
fixed point theorem to prove that the pairs of buckled states established
in [140] continue to exist for all λ > λn ; in fact, Wolkowisky [145] proves
that for all λ > λn there exist n pairs of nontrivial solutions of (3.10a,b);
the members of each pair differ only in the sign of q. The q in the first
pair has no internal nodes, the q in the second has one internal pair, and
so on up to the q in the nth pair which has n–1 internal nodes.
We begin our delineation of the results on initial buckling of a lin-
early elastic, isotropic, circular plate subject to the action of a radial
compressive force Nr = −λh along its edge at r = a, by looking at
rotationally symmetric buckling into the first mode for a plate with a
clamped edge. In this case, the equation governing initial buckling is
(3.7), with ϕ = w (r) and Nr0 = −λh, and the boundary conditions are
(see (3.11)) w(a) = w (a) = 0. If (3.7) is multiplied through by r2 we
obtain
2
0 2
2d ϕ dϕ Nr r
r +r + −1 ϕ=0 (3.14)
dr2 dr K
which is Bessel’s differential equation. It can be shown (see, e.g., Timo-
shenko and Gere [64]) that the solution of (3.14) subject to the clamped
boundary conditions w(a) = w (a) = 0 leads to a critical value of the
© 2001 by Chapm
compressive stress λ, which is given by
14.68K
λcr = (3.15)
a2 h
The result in (3.15) will also follow, later on, from a perturbation ex-
pansion approach to the postbuckling analysis for this case; the clamped
plate in edge compression is depicted in Fig. 3.1.
The next case is that of the isotropic circular plate which is subject to
a uniform radially compressive force and is simply supported along its
edge at r = a; we consider axially symmetric buckling into the first mode
ν
so that w (0) = 0 and w(a) = 0, as well as w (a)+ w (a) = 0. The case
a
of a simply supported circular plate is more amenable to postbuckling
analysis than that of a clamped plate and has received far more attention
in the literature. Using Bessel functions (see, once again, Timoshenko
and Gere [64]), it may be shown that
4.20K
λcr = (3.16)
a2 h
a result which will also follow from the postbuckling analysis of this
problem.
A more general problem than that considered in either of the two
preceeding cases concerns the case of elastically restrained edges, from
which one may deduce (i.e., Reismann [146]) a critical stress for any
degree of restraint along the edge at r = a between clamped and simply
supported. If one restricts consideration to buckling only into the first
axially symmetric mode, then a coefficient µ of restraint may be intro-
duced into the support condition along r = a in such a way that µ = 0
corresponds to the simply supported case, while µ = ∞ corresponds to
a clamped edge; specifically, we have, in all situations, w(a) = 0, and
µφ(a) dφ ν
=− + φ |r=a (3.17)
a dr r
Physically, (3.17) says that the inclination of the middle surface of the
plate with respect to the plane of the undeflected plate is proportional
to the radial edge moment. We still maintain the condition w (0) =
0 at the center of the plate. By imposing the conditions w(a) = 0,
w (0) = 0, and (3.17), on the general solution of (3.14)—which will be
discussed, in detail during the course of the postbuckling analysis—it
may be determined that
kK
λcr = 2 (3.18)
a h
where the relationship between the buckling parameter k and µ is indi-
cated in Fig. 3.2, which is taken from Reismann’s work [146]; it is clear
from this figure that as µ → 0, k → 14.68 and as µ → ∞, k → 4.20,
in agreement with (3.15) and (3.16), respectively, for the clamped and
simply supported conditions along the edge at r = a.
Higher order (non-axially symmetric) buckling modes, which are as-
sociated with the uniform compressive buckling of linearly elastic, iso-
tropic, circular plates of radius a, are governed by the linear partial dif-
1
ferential equation (1.78) with Nr = Nr0 ≡ − Φ0,r , Φ0,θ = Φ0,rθ = Φ0,θθ = 0,
r
where Φ0 is, once again, the Airy function corresponding to the pre-
buckled in-plane stress distribution; under this condition, of a uniform,
radially compressive, force Nr = −λh per unit length, applied to the
1
edge of the plate at r = a, (1.78), with Nr = Nr0 ≡ − Φ0r , may be
r
rewritten in the form
2 2
∂ 1 ∂ 1 ∂2 ∂ 1 ∂ 1 ∂2 Nr0
+ + + + + w(r, θ) = 0
∂r2 r ∂r r2 ∂θ2 ∂r2 r ∂r r2 ∂θ2 K
(3.19)
The usual procedure at this point (which we will substantially elaborate
upon when we consider the problem of initial buckling of isotropic annual
plates) is to look for solutions of (3.19) of the form
∞
#
w(r, θ) = An (r) sin nθ (3.20)
n=1
© 2001 by Chapm
the circumferential direction with a nodal line along the diameter as in
Fig. 3.3. For such a case we obtain, therefore, n = 1, and a numerical
solution giving the buckling parameter
a2 h
k= λcr (3.22)
K
as a function of µa/K is shown as the upper curve in Fig. 3.4, where the
lower curve again depicts the same function for the axially symmetric
first mode.
We now turn our attention to the postbuckling analysis for a uniformly
(radially) compressed linearly elastic, isotropic circular plate which is ei-
ther clamped or simply supported along the edge at r = a; the earliest
analysis of this problem which yielded substantive results appears to be
in the work of Friedrichs and Stoker [72], [143] which treats the sim-
ply supported case; further work on the case where the boundary at
r = a is simply supported has been carried out by Sherbourne [147],
while the work of Friedrichs and Stoker, op. cit, has been summarized
in the monograph by Stoker [148]. The methods employed by Friedrichs
and Stoker, op. cit, for the simply supported case have been extended
by Bodner [149] to treat the case where the edge at r = a is clamped.
All the work referenced above deals with the von Karman equations for
this problem written in terms of Φ and w; attention is also confined,
with respect to the problem of buckling beyond the critical load, to that
situation where the plate buckles, initially, into an axisymmetric mode.
Beyond reviewing the results of Friedrichs and Stoker, [72], [143], and
Sherbourne [147] (for a simply supported edge), and those of Bodner
[149] (for a clamped edge), we will also discuss, for both types of sup-
port conditions along r = a, results which follow, for axially symmetric
initial buckling, from carrying out the postbuckling analysis for the von
Karman equations in terms of the in-plane (u = u(r)) and out-of-plane
(w = w(r)) displacements instead of in terms of Φ and w.
We begin with the work of Friedrichs and Stoker [72], [143], [148] for
a simply supported edge; to bring our results completely into line with
1
their work we will take Nr = − Φr , replace G in (3.9) by G̃ = G/Eh,
r
and η 2 = K/a2 by η̃ 2 = Ehη 2 , in which case we have, in place of
(3.10a,b), the equations (we drop the ˜):
η 2 Gq + pq = 0 (3.23a)
1 2
Gp = q (3.23b)
2
1
Here p = Φ,r is the compressive radial membrane stress, and q =
r
a a dw qr
− φ≡− , so that is the slope of the deflected plate. At the
r r dr a
center of the plate at r = 0, p(r) and q(r) must satisfy the conditions
in (3.11). At the edge of the plate, at r = a, we have the boundary
conditions '
1 '
p = Φ,r '' = − Nr |r=a = p̄(= −λh) (3.24)
r r=a
which implies the vanishing of the bending moment at the edge of the
plate.
a a a
Remarks: As q = − w,r , q = − w,rr + 2 w,r
r r r
so
dq a aw,r
r + (1 + ν)q = (−aw,rr + w,r ) − (1 + ν)
dr r q
and (3.25) is, thus, a direct consequence of the condition of simple sup-
port, (1.140), with w,θ = 0 and R2 = a.
Remarks: If one has determined p and q, then all other physical quan-
tities of interest may be easily computed, e.g., the circumferential mem-
brane stress is
pc = B0 p ≡ rp (r) + p(r), (3.26)
while the normal deflection w is
a
1
w(r) = q(r)rdr (3.27)
a r
© 2001 by Chapm
p and q with respect to some convenient parameter $ which vanishes at
the onset of buckling, i.e.,
p̄ = p̄0 + $ p̄2 + $ p̄4 + · · ·
2 4
p = p0 + $2 p2 + $4 p4 + · · · (3.28)
q = $q1 + $ q3 + $ q5 + · · ·
3 5
dQ1 dQ1
u2 +u + (u2 − 1)Q1 = 0 (3.30)
du du
Equation (3.30) is the Bessel equation of the first order and has the
general solution
Q1 (u) = c1 J1 (u) + c2 Y1 (u) (3.31)
where J1 (u) and Y1 (u) are, respectively, the Bessel functions of the first
and second kinds of the first order. Because Y1 is not bounded in a
neighborhood of u = 0, the conditions in the physical problem require
that c2 = 0, and the remaining boundary conditions then imply that
'
dJ1 ν '
c1 λ0 + J1 '' =0 (3.32)
du u u=λ0
and '
1 dJ1 1 '
c1 a−1 λ20 − 2 J1 '' =0 (3.33)
u du u u=0
1
As J1 (u) = J0 (u)− J1 (u), where J0 is the Bessel function of zero order,
u
except for the trivial solution in which c1 = 0, the boundary condition
(3.32) is satisfied only if λ0 is a root of the transcendental equation
© 2001 by Chapm
√
p̄/p̄0 ≡ λ2 /λ20 as the basic buckling parameter and to choose $ = Λ − 1
so that the onset of buckling is characterized by Λ = 1 and $ = 0. In
this section, we will present the results of a calculation based on the
Lyapunov-Schmidt reduction ([10], [12]), which shows that $ in (3.28)
should have the form
a 12
q12 dr √
$=
0
a
Λ−1 (3.36)
q1 a3 dr
0
with λ̄ = p̄ − p̄0 . For various values of Λ = p̄/p̄0 , the results for radial
and circumferential membrane stresses and radial bending stress which
follow from the work of Friedrichs and Stoker [72], [143] are displayed,
respectively, in Figs. 3.5, 3.6, and 3.7; the corresponding postbuckling
graphs for the slope and deflection of the (uniformly) radially compressed
circular plate are depicted in Figs. 3.8 and 3.9. It must be emphasized
that the results shown are for first-order (axisymmetric) buckling corre-
sponding to the lowest eigenvalue of the linear (initial) buckling theory;
there are many indications that this first buckled state becomes unstable
for sufficiently large Λ and that a second buckling occurs which, subse-
quently, also becomes unstable, e.g., the second-order buckling, which
corresponds to the second eigenvalue of the linear problem L1 , appears
in the work of Friedrichs and Stoker [143] for Λ in the range Λ > 5.
The issue of secondary buckling into non-axisymmetric modes (“wrin-
kling”) for compressed, linearly elastic, isotropic, circular plates will be
discussed in detail in § 5.2.
A slightly different analysis of the postbuckling behavior of a simply
supported uniformly (radially) compressed, elastic, isotropic plate has
been given by Sherbourne [147] who again recovers the result (3.16) for
the critical stress at which the plate will first buckle into an axisymmetric
mode; for large elastic deformations, the compressive stress λ = σrr |r=a
then increases as the central deflection δ = w(0) grows. Sherbourne
[147] shows that the relationship between λ/λcr and δ/h takes the form
2
λ δ
= 1 + 0.241 (3.38)
λcr h
p̄(1 − ν)
w(r) = 0, u(r) ≡ uo (r) = − r (3.39)
Eh
and, if we set v(r) = u(r) − u0 (r), then in terms of w, v, the von Karman
© 2001 by Chapm
equations assume the form
1 1 12 1 2 ν
w,rrr + w,rr − 2 w,r − 2 v,r + w,r +v w,r
r r h 2 r
p̄ (3.40a)
+ w,r = 0
K
and
1 1 (1 − ν) 2
v,rr + v,r − 2 v + w,r w,rr + w,r = 0 (3.40b)
r r 2r
For a simply supported plate we have, furthermore, the boundary
conditions
w,r ''
w(a) = w + ν ' =0
,rr
r r=a
'
1 2 v'
v,r + w,r + ν '' =0 (3.41a)
2 r
r=a
ẇ(0) = v(0) = 0,
while, for a clamped plate, these conditions read
w(a) = w,r |r=a = 0
'
1 2 v '' (3.41b)
v,r + w,r + ν ' =0
2 r r=a
ẇ(0) = v(0) = 0,
In the spirit of the perturbation analysis employed by Friedrichs and
Stoker [143], we seek solutions of (3.40a,b) in the form
w(r) = $w1 (r) + $2 w2 (r) + $3 w3 (r) + · · ·
v(r) = $v1 (r) + $2 v2 (r) + $3 v3 (r) + · · · (3.42)
2
p̄ = p̄ + $p̄ + $ p̄ + · · ·
0 1 2
2
with $ once again the independent expansion parameter which regu-
lates progress along the post-buckling path. Substituting the expansions
(3.42) into the equilibrium equations (3.40a,b) we generate a sequence
of equations (all linear) for the coefficients wi and vi , the first three of
which in each case are (out-of plane equations):
1 1 p̄
0
w1,rrr + w1,rr − 2 w1,r + w1,r = 0 (3.43a)
r r K
1 1 12 w1,r v1
w2,rrr + w2,rr − 2 w2,r − w 1,r v1,r + ν
r r h2 r (3.43b)
p̄ p̄
0 1
+ w2,r + (1 − ν )w1,r = 0
2
K K
1 1 12
w3,rrr + w3,rr − 2 w3,r − {w1,r v2,r
r r h2
w1,r v2 1 3 w2,r v1
+ν + w1,r + w2,r v1,r + ν (3.43c)
r 2 r
1 1
+ p̄0 w3,r + p̄1 w2,r + p̄2 w1,r = 0
K 2
and (in-plane equations):
1 1
v1,rr + v1,r − 2 v1 = 0 (3.44a)
r r
1 1 1
v2,rr + v2,r − 2 v2 + w1,r w1,rr + (1 − v)w1,r
2
=0 (3.44b)
r r 2r
1 1
v3,rr + v3,r − 2 v3 + w1,r w2,rr
r r (3.44c)
w1,r w2,r
+w2,r w1,rr + (1 − ν) =0
r
For the simply supported circular plate, the conditions (3.41a) imply
that
ẇi (0) = vi (0) = 0, all i
' (3.45a)
wi (a) = 0, wi,rr + ν wi,r '' = 0, all i
r r=a
ν ''
v + v1 ' =0
1,r
r r=a
'
1 2 ν ''
v2,r + w1,r + v2 ' =0 (3.45b)
2 r r=a
'
v3,r + w1,r w2,r + ν ν3 '' =0
r r=a
The first out-of-plane equation has the solution
© 2001 by Chapm
with A1 , B1 constants of integration, γ = p̄0 /K, p̄0 the critical load
for buckling of the plate, and J1 and Y1 the Bessel functions of the
first order. Regularity at r = 0 requires that B1 = 0. If we set the
parameter, $ = δ ≡ w(0), the central deflection of the plate, we find that
w1 (0) = 1, wj (0) = 0, j = 1; using this result, together with w1 (a) = 0,
we obtain
w1,r = γJ1 (γr)/(J0 (γa) − 1) (3.47)
and
γ2 J1 (γr)
w1,rr = J0 (γr) − (3.48)
J0 (γa) − 1 γr
with J0 , once again, the Bessel function of the first kind of zero order.
Using the second condition
' of simple support along the edge at r = a,
ν '
i.e., w1,rr + w1,r ' = 0, we then obtain the relation
r r=a
−γrJ12 (γa)
We may now write the third-order equation for the lateral deflection
in the form
1 1 p̄
0
w3,rrr + w3,rr − 2 w3,r + w3,r
r r K
p̄ (3.53)
12 1 2 ν2 2
= 2 v2,r + w1,r +ν w1,r − w1,r
h 2 r K
so that the right-hand side of (3.53) is known except for p̄2 . If (3.53)
is multiplied by rw1,r and integrated with respect to r, from r = 0 to
r = a, and the boundary conditions at r = 0 and r = a are used, we
find that
p̄2 = 2.269K/a2 h2 (3.54)
so that the leading terms in the postbuckling expansion for the load
yield
2
δ
p̄/p̄0 = 1 + 0.27 , δ = w(0) (3.55)
h
which is in line with Sherbourne’s [147] result (3.38). The corresponding
analysis for the clamped (plate) boundary conditions yields the equation
γJ1 (γr)
= 0, γ = p̄0 /K (3.56)
J0 (γa) − 1
whose solution again produces (3.15), or p̄0 = 14.68K/a2 , and the buck-
ling mode (3.50) with γ now satisfying γa = 3.832. We also find that,
for the case of clamped boundary conditions, in lieu of (3.54) we have
© 2001 by Chapm
Setting φ = (p, q), we define the mapping G : φ → Φ by
1
Φ1 = G1 (p, q) ≡ Gp − q 2
2 (3.60)
Φ2 = G2 (p, q) ≡ η Gq + pq
2
and, then,
Gφ (p̄, 0) [(p, q)] ≡ A(p, q) = Gp, η 2 Gq + p̄q (3.62)
thus, if N (A) denotes the null space of A, we have (p, q) ∈ N (A) if and
only if Gp = 0 and η 2 Gq + p̄q = 0. Therefore,
We write
p = p̄ + p̃
(3.64)
q = $q1 + q̃
where
p̃ = $2 p2 + · · ·
(3.65)
q̃ = $3 q3 + · · ·
and, as q̃ belongs (via the Lyapunov-Schmidt decomposition) to N (A)⊥ ,
the orthogonal complement of N (A),
a
q̃(r)q1 (r)dr = 0 (3.66)
0
The von Karman equations for this problem have the form Gφ = 0 and,
thus, employing the projection P ( onto N (A)) we have P Gφ = 0 or
1 2 2
# Gp − q , η Gq + pq , (p̄, q1 )$ = 0 (3.68)
2
a
where < (p, q), (p , q ) >= (pp + qq )dr. Thus,
0
a
1 a
p̄ (Gp − q 2 )dr + η 2 Gq + pq q1 dr = 0 (3.69)
0 2 0
However,
a
1
(Gp − q 2 )dr
0 a2
1 a
= G(p̄ + p̃)dr − ($q1 + q̃)2 dr
2 0
a
0
(3.70)
1 a 2 2
= Gp̃dr − ($ q1 + 2$q1 q̃ + q̃ 2 )dr
2 0
0
a
1 2 a 2 1 a 2
=$ 2
Gp2 dr − $ q1 dr − q̃ dr
0 2 0 2 0
a
1 2
as q1 q̃dr = 0. Using the fact that Gp2 = q we have, from (3.70),
0 2 1
a
1 2 1 a 2
(Gp − q )dr = − q̃ dr
2 2 0
0
a
1 (3.71)
=− ($3 q3 + · · ·)2 dr
2 0
= O($6 )
a
Next, we need to compute (η 2 Gq + pq)q1 dr. We begin with
0
η 2 Gq + pq = η 2 Gq + p̄q + p̃q
But
η 2 Gq1 = −p̄0 q1
so
η 2 Gq + pq = −$p̄0 q1 + η 2 Gq̃ + p̄q + p̃q
© 2001 by Chapm
where, once again, we have set λ = p̄ − p̄0 . Thus,
a
0= (η 2 Gq + pq)q1 dr
a 0
a a
= $λ q12 dr + η 2 q1 Gq̃dr + p̄ q̃q1 dr
0
a 0 0
+ p̃qq1 dr
a 0 a
= $λ q12 dr + η 2 q1 G($3 q3 )dr
0 0
k
+ ($ p2 )($q1 + $3 q3 + ··)q1 dr + O($7 )
2
0
a
where we have used the fact that q̃q1 dr = 0. Therefore,
0
a a
0 = $λ q12 dr + $3 η 2 (Gq3 )q1 dr
0
a
0
a (3.72)
+$3 p2 q12 dr + $5 p2 (q3 + $2 q5 + · · ·)dr + O($7 )
0 0
or
a a
0 = $λ q12 dr +$ 3
(−p̄0 q3 − p2 q1 )q1 dr
0
a
0
(3.73)
+$3 p2 q12 dr + O($5 )
0
as
η 2 Gq3 + p̄0 q3 + p2 q1 = 0, (3.74)
Therefore,
a a a
(η 2 Gq + pq)q1 dr = $λ q12 dr − p̄0 $3 q3 q1 dr + O($5 )
0 0 0 (3.75)
=0
Hence,
λA − p̄0 $ B = 0
2
a a (3.76)
A = q12 dr, B = q3 a1 dr
0 0
or
a 12
q12 dr *
p̄
$=
0
a
−1 (3.77)
p̄0
q1 q3 dr
0
Therefore,
q = $q1 + ($3 q3 + · · ·) (3.78)
© 2001 by Chapm
where Dr , Dθ , and Drθ are given by (1.95), (1.96), and (1.97) and
σ o · h ≡ N o = 1 Φ0,r + 1 Φ0,θθ
rr r
r r2 (3.81)
o
σθθ · h ≡ Nθ = Φ0,rr
o o
so that σrr , σθθ are the non-zero components of the in-plane (prebuck-
ling) stress tensor in polar coordinates.
The initial buckling problem governed by (3.80) was analyzed by
Woinowsky-Krieger [150] and a more general solution which includes the
results in [148] was then presented by Mossakowski [151]. We assume
that the deflection w(r, θ) can be expressed in the form
∞
#
w(r, θ) = w0 (r) + wr (r) cos nθ (3.82)
n=1
© 2001 by Chapm
Iwinski and Nowinski; other notable contributions include the work of
Pandalai and Patel [153] and C.L. Huang [154]. In the work which we
present below, we follow the analysis in the survey paper of Sherbourne
and Pandey [155]; this comprehensive work contains a comparison of the
postbuckling behavior of circular and annular plates exhibiting cylindri-
cally orthotropic symmetry. The authors [155] also make note of the fact
that, with the exception of a handful of studies, most of the postbuckling
analyses that have been carried out for plates with a circular geometry
and cylindrically orthotropic symmetry, have been effected for annular
as opposed to circular plates; they note that “this somewhat surprising
negligence of postbuckling analysis for circular orthotropic plates may
not be coincidental as the problem is characterized by singularities due
to orthotropy and axisymmetry of deformations thus contributing to
mathematical complexity.” One study of postbuckling analysis which
the authors [155] do point to specifically is the finite-element analysis of
Raju and Rao [156], which highlights one of the difficulties inherent in
the subject: it is shown in [156] that for the cylindrically orthotropic,
circular plate, an increase in the orthotropy ratio increases the linear
buckling load but decreases the postbuckling stiffness, i.e., the load-
carrying capacity of the plate.
The plane stress solution for a circular orthotropic plate appears, e.g.,
in Lekhnitskii [51] and Pandalai and Patel [153]; the plate is assumed
to be subjected to a uniform compressive stress λ = σrr |r=a . For the
isotropic circular plate, the prebuckling, in-plane stress distribution is
characterized by a uniform compressive stress distribution throughout
the entire plate; however, for the case of a cylindrically orthotropic,
circular plate, the in-plane prebuckling stress state varies along the di-
rection of the radial coordinate according to
r √β−1
σrr = − (σrr |r=a )
a
√
√ r β−1 (3.88)
σθθ = − (σrr |r=a ) · β
a
σrθ = 0
so that the qualitative nature of the prebuckling stress distribution is
governed by the orthotropy ratio. Thus, for β > 1, stresses decrease as
we approach the center of the plate, with the opposite situation holding
for β < 1 ( and the stresses becoming infinite at the center of the plate);
stress singularities, therefore, appear at r = 0, whenever β < 1, and,
as the authors [155] explicitly point out, “this peculiar feature of the
plane stress solution is tacitly recognized in the literature since most
solutions for circular and annular plates are limited to cases of k > 1.
The only exception appears to be the study by Turvey and Drinali [157]
where numerical results are reported for cases including β < 1 using the
dynamic relaxation method.
Remarks: As has been pointed out in [155], G.F. Carrier [158] argued
that it is pointless to speak of material orthotropy at the center of a
circular plate and that such a concept is simply a mathematical ab-
straction. In practical terms, cylindrical orthotropy does not exist at
r = 0 because of the coincidence there of the radial and circumferential
directions.
The scope of the nonlinear analysis which is presented in [155] is lim-
ited to axisymmetric deformations and proceeds as follows: we first de-
fine, in the usual way, the in-plane strains by
1 1
err = u (r) + w2 (r), eθθ = u(r) (3.89)
2 4
where u(r), w(r) have the same interpretation as for the case of the
isotropic circular plate undergoing axisymmetric deformations. As the
plate is assumed to be subjected to a uniform radial compression applied
along its edge at r = a, σrθ = 0 in (1.91) and, thus, the inverted version
of these constitutive relations assumes the form
σrr Er 1 νθ err
= (3.90)
σθθ (1 − νθ νr ) νθ β eθθ
Eθ νθ
with β = = , while the moment-curvature relation is
Er νr
Mr Er h3 1 νθ κr
= (3.91)
Mθ (1 − νθ νr ) νθ β κ θ
© 2001 by Chapm
and
2
d3 w 1 d2 w β dw 12 dw du 1 dw u
= − + 2 + 2 + + νθ (3.93)
dr3 r dr2 r dr h dr dr 2 dr r
d3 ω d2 ω β dω dω dv 1 dω v
ρ 3 =− 2 + + 12ρ + + νθ (3.96)
dρ dρ ρ dρ dρ dρ 2 dρ ρ
which implies that the deformed shape of the circular plate remains
smooth at the center of the plate (at x = 0); in order that (3.101)
hold, it follows, from (3.99), (3.100), that the coefficients of those terms
involving 1/x in these equations must vanish, which, in turn, implies
that √ √
(νθ − 2 β − 1)( β + 1)
U (0) = √ · W 2 (0) (3.102)
2(3 β + 1)
and √
12( β + νθ )
W (0) = √ √ U (0)W (0) (3.103)
(3 β + 1)( β + 1)
Expanding U (x) and W (x) in Maclaurin series, in a neighborhood of
x = 0, substituting the results in (3.99), (3.100), and then setting x = 0,
one obtains for U (0) and W (0) the relations
√ √
2 6(4 β + 3 − νθ )( β + νθ )
U (0) = −U (0)W (0) √ √ (3.104)
(4 β + 2)(3 β + 1)
√
√ 12(2 β + νθ + 1)
(4 β + 2)W (0) = √ U (0)W (0)
( β + 1)
√ (3.105)
12( β + νθ )
2
+ √ Ω(0)W (0) + 6( β + 1)W (0)W (0)
( β + 1)
If we introduce the variables Yi , i = 1, · · · , 5,
Y1 = W, Y2 = U, Y3 = W , Y4 = U , Y5 = W (3.106)
© 2001 by Chapm
boundary condition along the edge of the plate at r = a; if this boundary
condition is not satisfied, then W (0) is systematically varied, keeping
U (0) constant, and the system of ODEs is numerically integrated until
the prescribed boundary condition is met.
We want to elaborate upon some of the results which follow from
numerical solutions of the system (3.107); however, it is first worthwhile
to append some remarks to the discussion of initial buckling for the
cylindrically orthotropic circular plate that was presented earlier.
Remarks: In [160] Swamidas and Kunukkasseril have shown that the
buckling equation for a clamped, circular, cylindrically orthotropic plate
has the form
√ √
2 β 2 λcr
Jη (σ) = 0; η = √ , σ= √ , (3.108)
( β + 1) ( β + 1)
with Jη the Bessel function of the first kind of order η, while for sim-
ply supported plates, the critical buckling loads are obtained from the
solutions of the equation
√
β+1
( β + νθ )Jη (σ) − σJη+1 (σ) = 0 (3.109)
2
σrr σθθ Mr Mθ
βr = , βθ = , γr = , γθ = (3.110)
fr fr mr mr
where
fr = Er /(1 − νθ νr ) (3.111)
and
mr = Er h2 /12(1 − νθ νr ) (3.112)
For simply supported plates, the qualitative nature of the load versus
deflection and load versus radial shortening curves are similar to those
for clamped cylindrically orthotropic circular plates. Increases in the
orthotropy ratio result in increases in the postbuckling stiffness and a
reduced transverse deflection. Also, the zone of the normalized tensile
radial stresses which surrounds the origin r = 0, i.e. βr = σrr /fr , fr ,
as given by (3.111), is much larger than that for clamped plates; this
is depicted in Fig. 3.23a. In Fig. 3.23b we depict the corresponding
normalized circumferential stresses for the simply supported case. Both
Figs. 3.23a and 3.23b correspond to a load ratio of λ/λcr = 3. Finally,
contrary to the case of an isotropic plate, a peak in the tensile stress βr
© 2001 by Chapm
is observed at some distance away from the center of the plate; this fact
is quite evident from Fig. 3.23b.
Remarks: Although we have not yet considered the initial or postbuck-
ling behavior of cylindrically orthotropic annular plates, this is a good
place to make note of some of the differences of a qualitative nature
which have been observed in the literature. As noted in [155], Huang
[154] has reported an increase in the maximum transverse deflection of
annular, cylindrically orthotropic plates (with a free inner edge and an
outer edge that is either clamped or simply supported) as the orthotropy
ratio increases, while the opposite trend is observed in the solutions for
cylindrically orthotropic circular plates, which are given in [155]. It is
natural to ask whether one can recover the circular plate solutions from
the solutions obtained for annular plates by letting the hole size in the
annular case shrink to zero; this does not appear to be mathematically
feasible. At the free inner edge of an annular plate the boundary con-
ditions are βr = γr = 0, and these lead to solutions which yield finite
values for the circumferential stress and moment at the free edge. How-
ever, in formulating the problem for the cylindrically orthotropic circular
plate, we impose the boundary conditions
βr = βθ = γr = γθ = 0, at r = 0 (3.113)
© 2001 by Chapm
∂2u ∂u ∂2u ∂u ∂2u
I2,1 2
+ I2,2 + I2,3 u + I2,4 2 + I2,5 + I2,6 +
∂r ∂r ∂θ ∂θ ∂r∂θ
∂2v ∂v ∂2v ∂v ∂2v
J2,1 2 + J2,2 + J2,3 v + J2,4 2 + J2,5 + J2,6 =0
∂r ∂r ∂θ ∂θ ∂r∂θ
(3.115)
In the “in-plane” equations (3.114), (3.115) for u = ur (r, θ) and v =
uθ (r, θ) the coefficients, modulo a factor of h (the plate thickness) are
given by
1 a b
I1,1 = cos 2θ + cos 4θ + c (3.116a)
r 2 8
3
I1,2 = − b cos 4θ + c (3.116b)
8
3
−rI1,3 = − b cos 4θ + c (3.116c)
8
1 1
rI1,4 = − b cos 4θ + b (3.116d)
8 8
1
rI1,5 = b sin 4θ (3.116e)
2
1 1
I1,6 = − a sin 2θ − b sin 4θ (3.116f)
2 4
1 1 1
I2,1 = − a sin 2θ − b sin 4θ (3.116g)
r 4 8
3 3
I2,2 = − a sin 2θ + b sin 4θ (3.116h)
4 8
3 3
rI2,3 = a sin 2θ − b sin 4θ (3.116i)
4 8
1 1
rI2,4 = − a sin 2θ + b sin 4θ (3.116j)
4 8
1 1
rI2,5 = −a cos 2θ + b cos 4θ + b + (D3 + G12 ) (3.116k)
2 2
1 1
I2,6 = − b cos 4θ + b + (D3 − G12 ) (3.116l)
4 4
and
1 1 1
J1,1 = − a sin 2θ − b sin 4θ (3.117a)
r 4 8
1 3
J1,2 = a sin 2θ + b sin 4θ (3.117b)
4 8
1 3
rJ1,3 = − a sin 2θ − b sin 4θ (3.117c)
4 8
1 1
rJ1,4 = − a sin 2θ + b sin 4θ (3.117d)
4 8
1 1
rJ1,5 = b cos 4θ − b − (D3 + G12 ) (3.117e)
2 2
1 1
J1,6 = − b cos 4θ + b + (D3 − G12 ) (3.117f)
4 4
1 1 1
J2,1 = b cos 4θ − b + G12 (3.117g)
r 8 8
1 3 1
J2,2 = − a cos 2θ + b cos 4θ + b + G12 (3.117h)
2 8 8
1 3 1
rJ2,3 = a cos 2θ − b cos 4θ − b − G12 (3.117i)
2 8 8
1 1
rJ2,4 = − a cos 2θ + b cos 4θ + c (3.117j)
2 8
r 1 1
J2,5 = − b sin 4θ + a sin 2θ (3.117k)
4 8 4
© 2001 by Chapm
1 1
J2,6 = − a sin 2θ + b sin 4θ (3.117l)
2 4
where the parameters a, b, c are given by
a = D1 − D2
b = (D1 + D2 ) − 2D3 (3.118)
c = 3 (D1 + D2 ) + 1 D3
8 4
with
E1 E2 E1 ν12
D1 = , D2 = , D3 = + 2G12 (3.119)
1 − ν12 ν21 1 − ν12 ν21 1 − ν12 ν21
Along the edge of the plate, at r = a, we prescribe, in general, the radial
and tangential components pr (θ) and pθ (θ), respectively, of the applied
traction where, as in Chapter 1
$ '
pr (θ) = h σxx cos2 θ + σyy sin θ cos θ + σxy sin 2θ 'r=a
(3.120)
pθ (θ) = h [(σyy − σxx ) sin θ cos θ − σxy cos 2θ]|r=a
i.e.
pr (θ) = hσrr |r=a , pθ (θ) = hσrθ |r=a (3.121)
and regularity conditions must, in addition, be prescribed at the center
of the plate, i.e., at r = 0, with respect to both the Airy function Φ
and the transverse deflection w. To the best of the author’s knowledge,
the system of equations (3.114), (3.115) and the associated structure
(3.116a)–(3.120) for the coefficients, have never appeared explicitly in
the literature on plate buckling.
© 2001 by Chapm
easier than the nonlinear elastic case because of the assumed linearity of
the constitutive response although, of course, memory effects must now
be taken into consideration, as is clear from the discussion in Section
2.2. The general system of governing equations for viscoelastic buckling
presented in [106] has been specialized in [140] to treat the problems of
initial buckling and postbuckling for thin, circular, viscoelastic plates;
∂
these equations assume the form (recall that D ≡ ) :
∂t
∂ 3 ∂w (r, t)
(1 + αD) r
∂r ∂r (3.122)
= (1 + βD) r3 w (r, t) [Φ (r, t) − λ]
∂ 3 ∂Φ (r, t)
(1 + βD) r
∂r ∂r (3.123)
= −(1 + αD)r3 w2 (r, t)
for 0 < r < 1, 0 ≤ t ≤ T, T < ∞, where w is the spatial derivative of
the transverse displacement of the plate, Φ is the spatial derivative of
the Airy stress function, λ is, in the usual way, the positive parameter
of proportionality for the given uniform radial loading along the edge of
the circular plate (located at r = 1) and α, β are positive “viscous” pa-
rameters, where we assume that α < β. Associated with the generalized
von Karman equations (3.122), (3.123) are the boundary conditions
© 2001 by Chapm
It may be shown that the eigenvalues of (3.132) which correspond to
the time t = 0 are simple and constitute a sequence of discrete numbers
λon → ∞, as n → ∞, while the eigenvalues of (3.132) which correspond
to time t = ∞ are also simple and form a sequence λ∞n → ∞, as n → ∞.
Furthermore, for each n = 1, 2, · · ·
α ∞
λon = λ (3.133)
β n
For λ = λon the linearized problem (3.132) may be solved in the form
∞
#
1 λ∞
n −λ
w (r, t) = cn φn (r)exp − t (3.134)
n=1
β λon − λ
where the φn (r) are the eigenfunctions which correspond to the eigen-
values λon , and the cn are the (generalized) Fourier coefficients in the ex-
pansion of the initial transverse perturbation w0 (r) with respect to the
φn (r). For λ < λ∞ n , it is clear from (3.134) that all transverse perturba-
tions of the viscoelastic plate tend to zero as t → ∞; for λ∞ n < λ < λn
o
i = 1, 2, · · ·, with
w0 (r, t) = Zλ−1 {S[wd (r, t)]} (3.138)
for 0 < r < 1, 0 < t < T , where wd (r, t) = w (r, 0), 0 < t < T, w+ (r, 0)
+
being the only positive solution of (3.135), then for λ1 < λ < λc the
operator Zλ−1 is monotone. As the operator S is also monotone, in
β
(3.137), for 0 < r < 1, and 0 < t < T , we have, with λ1 < λ < λ∗ ,
α
with the sequence, defined by (3.137), bounded from above by the func-
tion w̄ (r, t) = w+
(r, ∞), 0 < t < T, where w+ (r, ∞) is the only positive
solution of (3.136). By taking the limit as i → ∞, in (3.137), one ob-
tains a function w(r, t), which satisfies the first equation in (3.129), for
© 2001 by Chapm
0 < r < 1, 0 < t < T , so that λ1 = λ01 is a bifurcation point (i.e.
buckling load). The postbuckling behavior relative to branching from
λ1 = λ01 is shown in Fig. 3.24 and our results may be summarized as
follows: For λ < λ1 , only the trivial solution exists; however, in con-
trast to the situation for the stationary von Karman equations, where
for λ = λ1 only the trivial solution exists, for the generalized von Kar-
man equations associated with buckling of a circular viscoelastic plate,
of differential type, exactly two nontrivial solutions exist for λ = λ1 .
One of these two solutions, for λ = λ1 , is increasing while the other is
decreasing from the value w = 0, at time t = 0, to the values of w,
which are given by the solutions of the associated limiting stationary
equations at t = ∞. For λ > λ1 , there exist solutions starting from the
values given by the solutions of the limiting stationary equations at time
t = 0 which assume, as t → ∞, values determined by the solutions of the
limiting stationary equations as t → ∞. A more detailed (numerical)
analysis than the one carried out in [140] remains to be done, inasmuch
as no numerical approximations are provided for the eigenvalues λ0n , or
for the associated eigenfunctions φn (r), and the same remarks apply, of
course, to the eigenvalues and associated eigenfunctions of (3.132) which
are associated with time t = ∞.
∂
where (˙) ≡ . The stress rates are related to the strain rates in (3.140),
∂t
by σ̇ = Lė where the tensor of moduli L possesses two branches, one
of which corresponds to loading and the other corresponds to elastic
unloading. The elastic part of L, which we denote by L, is taken to be
isotropic in [141] so that
E 1 ν
Lijkl = (δik δjl + δil δjk ) + δij δkl (3.141)
1+ν 2 1 − 2ν
© 2001 by Chapm
where
12
3
σe = Sij Sij
2 (3.143)
Sij = σij − 1 δij σkk
3
and
0, if σe < (σe )max or
c= σe = (σe )max and σ̇e ≤ 0 (3.144)
1, if σe = (σe )max and σ̇e ≥ 0
The tangent modulus, which we have depicted in Fig. 2.50, in connection
with our discussion in Chapter 2.3, is the slope of the uniaxial stress-
strain curve and is a function of σe . The uniaxial stress-strain behavior
employed in [141] has a representation in terms of a modified piecewise
power law of the form
σ
$ σy , σ ≤ σy
= n (3.145)
$y
1 σ 1
+ 1 − , σ ≥ σy
n σy n
Lαβ33 Lγδ33
Kαβγδ = Lαβγδ − (3.146)
L3333
with the stress increments related to the strain increments by the plane
stress moduli (3.146); the variational principle described above serves as
the basis in [141] for implementing a finite element solution of the buck-
ling problem. We now delineate the initial buckling and postbuckling
results which follow from the elastic-plastic model of the circular plate,
beginning with the onset of buckling.
In both the elastic and plastic ranges, the prebuckling solution is that
of uniform radial compression, so that in the plastic range, the plane
stress moduli (3.146) at buckling correspond to an isotropic material.
The critical stress for buckling in the plastic range is (see Chapter 2.3)
given by the same expression as that in the elastic range except that,
instead of E and ν, an effective Young’s modulus Ef and an effective
Poisson ratio νf are employed. The critical compressive stress is given
by + 6
2
β̃ 2 Ef h
σc ≡ λcr = (3.150)
12 1 − νf2 a
with
E 1 E
=1+ −1 (3.151)
Ef 4 Et
and
Ef 1 E
νf = ν− −1 (3.152)
E 4 Et
where, for a clamped plate, β̃ is the smallest non-zero root of J1 (β̃) = 0,
while for a simply supported plate β̃ is the smallest non-zero root of
The functions J0 , J1 are, once again, the Bessel functions of the first
kind of orders zero and one, respectively, while the subscript c, above,
© 2001 by Chapm
denotes evaluation of the effective Poisson ratio νf at buckling. For both
the simply supported and clamped boundary conditions, the buckling
mode wc (r), associated with σc , as given by (3.150), has the form
J0 (β̃r/a) − J0 (β̃)
wc (r) = wc (0) (3.153)
1 − J0 (β̃)
For the case of a clamped plate it may be computed that β̃ 3.832.
In Fig. 3.25, we depict the buckling stress for both the clamped and
simply supported cases, with ν = 1/3, n = 12, as a function of the
parameter η̃ where
2
2
β̃E h
η̃ = (3.154)
12(1 − ν )$y a
2
with β̃E the value of β̃ when buckling occurs in the elastic range. For the
1
simply supported plate β̃E = 2.069, when ν = , while for the clamped
3
plate β̃E = 3.832. We also show, in Fig. 3.25, the uniaxial stress-strain
curve for the modified piecewise power law (3.145) with strain-hardening
exponent n = 12.
1
Remarks: It has been noted in [141] that, for a circular plate with ν =
3
and n = 12, the discrepancy between the buckling stresses predicted by
J2 flow theory and those predicted by a J2 deformation theory is within
one percent, provided η̃, as given by (3.154), satisfies η̃ ≤ 2; for larger
values of η̃, e.g., for thicker plates, the discrepancy between the values
of the predicted buckling stresses can be significant.
We now present those results of Needleman [141] which pertain to the
initial postbuckling behavior of circular plates in the plastic range. The
results presented are for both the clamped and simply supported situa-
tions and follow as a consequence of the asymptotic theory of Hutchin-
son [17]; in this analysis, the amplitude ξ of the relevant eigenmode
is employed as the expansion variable in the postbuckling perturbation
analysis.
The average radial stress applied at the edge of the circular plate at
bifurcation is given by
ξ
σav = σc + σ1 + ··· (3.155)
h
where σc , the bifurcation stress, is given by (3.156) and
'
1 h/2 '
'
σav = − σrr (a, η)dη ≡ Nr ' (3.156)
h −h/2 '
r=a
so that σav > 0 for compressive loading. In (3.155), the initial slope σ1
is uniquely determined by the condition that plastic loading takes place
throughout the plate except at one point where neutral loading occurs;
this leads to
σ1 3(1 + νf )c
= (3.157)
σc 1 − J0 (β̃)
and σ1 > 0 so that buckling takes place under increasing load.
The next term in the postbuckling expansion for the average applied
4
compressive stress σav turns out to be proportional to ξ 3 , i.e.,
43
ξ ξ
σav = σc + σ1 + σ2 + ··· (3.158)
h h
and 13
3 ∆
σ2 = − σ1 (3.159)
4 1 − J0 (β̃)
where ∆ is given by
2 '
h dLf ''
β̃ 2
2
∆ = 24β̃ Nc I1 + I0 − '
a dσc '
24
c
3 1−ν −1
+ (1 − νf )c Nc (3.160)
4 1+ν
+ 62
3 [1 − J0 (β̃)]J1 (β̃) 1−ν
− (1 − νf )c Nc−1
2 β̃ 1+ν
with
1 1
I0 = xJ02 (β̃x)dx, I1 = xJ12 (β̃x)dx (3.161a)
2 0
and
' '
Ef '' Lf '
'
Lf = ' , N = (3.161b)
(E/(1 − ν)) − Lf 'c
c
1 − νf c
The first three terms in the postbuckling expansion (3.158) are used,
in [141], to estimate σmax , the maximum applied stress that can be
supported by the plate, as well as the value of the amplitude ξ of the
buckling mode at σmax ; the results obtained have the form
σmax 3 (1 + νf )c
=1+ · (3.162a)
σc 4 ∆
© 2001 by Chapm
and
ξ 1 − J0 (β̃)
= (3.162b)
h max ∆
Also, for both the clamped and simply supported circular plates, the
point at which neutral loading occurs at buckling turns out to be r =
1
0, z = h.
2
When ξ increases from zero, the neutral loading surface (i.e., the sur-
face which separates the regions of plastic loading and elastic unloading),
1
spreads out from the point r = 0, z = h. The equation of the neutral
2
loading surface is
1/3
1 ξ z 1 β̃ 2 r 2
+ − − =0 (3.163)
2 ξmax h 2 8 a
It is noted in [141] that, at maximum load, the greatest penetration
of the neutral loading surface, through the thickness of the circular
plate, occurs at r = 0, where the neutral loading surface has pene-
trated halfway through the thickness of the plate. Away from the axis
of symmetry, at r = 0, the greatest penetration occurs on the surface
1
z = h. In Table 3.4, we display the results of the postbuckling analysis
2
obtained in [141] for three different cases; among other things, Table 3.4
shows that the maximum support load, which is given by (3.162a), is
only slightly above the buckling load for all three cases indicated and
that, for the simply supported case, the analysis predicts that, at the
maximum load, the plastic zone extends all the way out to the edge of
the circular plate, while for the clamped plate rmax /a is 0.52 at the max-
imum load irrespective of the amount of plastic strain prior to buckling.
In [141], Needleman also presents numerical results based on a finite-
element analysis of the variational equation δI = 0, where the functional
I is given by (3.149); in this analysis an imperfection w0 (r) is considered
which is proportional to the buckling mode as given by (3.153). The im-
perfection mode in [141] is specified by ξ, ˆ where ξˆ = w0 (0). The results
given in [141], for a “perfect” plate, are actually those which are com-
puted using the very small imperfection ξˆ = 10−4 h. We will put off a
complete discussion of the results obtained in [141], relative to the imper-
fection sensitivity of elastic-plastic circular plates, until Chapter 5. For
now, however, we reproduce from [141], Figs. 3.26–3.29, which indicate,
respectively, the curves of average applied stress vs. buckling deflection
for two different cases of a simply supported circular plate, the curves
governing buckling deflection versus applied displacement for a simply
supported circular plate, and the curves governing average applied stress
versus buckling deflection for a clamped circular plate; besides indicat-
ing the imperfection sensitivity for the various cases considered, all these
graphs clearly depict the nature of the postbuckling behavior for the per-
fect elastic-plastic circular plate.
In Fig. 3.27, buckling of the perfect plate takes place just outside the
elastic region; this plate is simply supported with σc /σy = 1.02, η̃ =
1
1.28, n = 12, and ν = ; for this case the initial slope σ1 , as computed
3
from (3.157), is also shown on the graph (which is valid only for very
small values of the buckling deflection because the maximum load, which
is given by σmax /σc = 1.006, is reached at ξ = 0.008h). The unloading
region for the perfect plate of Fig. 3.27, at maximum load, is the region
h
1 < z < ; beyond the maximum load point, the load decreases rapidly
2
and then levels off at larger buckling deflections.
The graph in Fig. 3.28 depicts the curves of buckling deflection ξ
vs. applied edge displacement U for the same values of the parameters,
geometrical and physical, as those considered in the case depicted in Fig.
3.27; the applied displacement has been normalized by its value Uc at
buckling. For the perfect plate, the buckling deflection ξ is zero until U
reaches the value Uc then, at U = Uc , the deflection begins to grow (but
very slowly). Near the maximum load point, ξ grows at nearly constant
edge displacement U .
In Fig. 3.29, we depict the average applied stress versus buckling
deflection curves for a case in which the buckling occurs further into the
plastic range. The circular plate for the case depicted in Fig. 3.29 is
1
clamped with σc /σy = 1.12, η̃ = 2.0, ν = , and n = 12; the initial
3
slope σ1 , as computed from (3.157), is once again shown for the perfect
plate but, as in the earlier case, this slope is only valid for very small
values of the buckling deflection ξ. Also, the maximum load, which is
given by σmax /σc = 1.018, occurs at a value of ξ = 0.065h, which is
larger than the corresponding value of ξ in the previous example. The
graph associated with the perfect plate in Fig. 3.29 shows that, when
buckling occurs well into the plastic range, the load falls off much more
gradually after the maximum load point is achieved than when σc ≈ σy .
Finally, Fig. 3.30 depicts the development of the plastic zone and the
unloading zone for the same case as that shown in Fig. 3.29. For the
© 2001 by Chapm
1
perfect plate, unloading starts at r = 0, z = h. As the primary zone of
2
unloading grows, a region near the clamped edge of the circular plate is
subjected to unloading; however, this unloading region near the clamped
edge does not have a significant effect on the overall deformation of the
plate because this region is highly constrained by the condition that the
edge is clamped. Once the maximum load point is reached, this zone near
the clamped edge undergoes loading again. A fundamental conclusion,
based on the analysis of perfect plates in [160], is that whenever buckling
occurs in the plastic range of the circular plate, the maximum support
stress is only slightly higher than the buckling stress; the asymptotic
analysis of the postbuckling behavior for the perfect plate in the plastic
range, as based on the theory first formulated by Hutchinson [17], [46],
yields a good estimate of the maximum support stress and of the growth
of the unloading zone in the plate after buckling.
and
r(s) sin θ(s)
τ (s) = ∗
, σ(s) = ∗ , µ(s) = θ (s) (3.168)
r (s) r (s)
defining strains ν(s), η(s), µ(s), τ (s), and σ(s). We note that |r ,s | =
thus
ν 2 (s) + η 2 (s). The strain ν(s) = r ,s · a is the local ratio of deformed
area to reference area for the parallelogram with sides r ,s and b; it is
convenient to think of ν as measuring “stretching of the base curve r ∗ ,”
while η is a measure of “shear,” i.e., of the reduction in the angle between
r ∗,s and b∗ . The strain measure µ is related to changes in curvature
which are associated with pure bending as opposed to stretching, while
τ (s) measures stretching in the azimuthal direction and σ(s) measures
the amount of bending about a. The strain “vector” associated with
axisymmetric deformations of the shell is, therefore,
S = (ν, η, τ, σ, µ) (3.169)
Now, let n1 (s0 , φ) and m1 (s0 , φ) denote the resultant contact force
and contact couple, per unit reference length of the circle r ∗ (s0 , φ), of
radius r∗ (s0 ), that are exerted across this section at r ∗ (s0 , φ), by the
material with s ≥ s0 on the material with s < s0 . Let n2 (s, φ) and
m2 (s, φ) denote the resultant contact force and contact couple per unit
reference length of the curve r ∗ (s, φ) that are exerted across this section
at r ∗ (s, φ) by the material with φ0 ≤ φ0 ≤ φ0 + $ on the material with
φ0 − $ ≤ φ < φ0 , where $ > 0 is a small number. It may be shown that,
in view of our limitation to axisymmetric deformations, the force and
© 2001 by Chapm
couple resultants have the form:
n1 (s, φ) = N (s)a(s, φ) + H(s)b(s, φ)
n2 (s, φ) = T (s)e2 (φ)
(3.170)
m1 (s, φ) = −M (s)e2 (φ)
m2 (s, φ) = Σ(s)a(s, φ)
If we assume that the shell is subjected to an axisymmetric body force
of intensity,
f (s, φ) = f1 (s)e1 (φ) + f3 (s)k (3.171)
and a body couple of intensity −(s)e2 (s, φ), per unit reference area of
r ∗ at (s, φ), then by summing all the forces and moments on a segment
which consists of all those material points with coordinates (ξ, ψ), s1 ≤
ξ ≤ s, 0 ≤ ψ ≤ φ, and subsequently differentiating the resulting integral
balance laws with respect to s and φ, we obtain the equilibrium equations
for the shell in the form:
∂ ∗
[r (s)N (s)a(s, φ) + H(s)b(s, φ)]
∂s (3.172a)
−T (s)e1 (φ) + r∗ (s)f (s, φ) = 0
and
d ∗
[r (s)M (s)] − Σ(s) cos θ(s)
ds
+r∗ (s) (ν(s)H(s) − η(s)N (s)) (3.172b)
+r∗ (s)(s) = 0
The shell is defined to be nonlinearly elastic if there are functions
T̂ , N̂ , Ĥ, Σ̂, and M̂ , of S and s, such that
When the edge pressure has intensity λ units of force per deformed
length, g(r) = r, while, if it has intensity λ units of force per reference
length, g(r) = 1; in [22] the latter situation is assumed. The integral
form of the balance of forces equation for the plate produces the relations
1
−sN (s) = λ + T (ξ)dξ cos θ(s) (3.179a)
s
1
sH(s) = λ + T (ξ)dξ sin θ(s) (3.179b)
s
© 2001 by Chapm
and, upon differentiation, these yield the system
d
[sN (s)] = sH(s)θ (s) + T (s) cos θ(s) (3.180a)
ds
d
− [sH(s)] = sN (s)θ (s) − T (s) sin θ(s) (3.180b)
ds
which is equivalent to (3.172a) for the present situation.
By substituting from (3.179a,b) into (3.172b), we obtain
d
[sM (s)] − Σ(s) cos θ(s)
ds
1 (3.180c)
+ λ+ T (ξ)dξ (ν(s) sin θ(s) + η(s) cos θ(s)) = 0
s
T̂ (k, k, 0, 0, 0) = −λ (3.185)
with
2 λ λ
γ(λ) = 0 (λ)
k(λ) + 0
M,µ H,η (λ)
(3.188)
N 0 (λ) N 0 (λ)
=− 0 k(λ) − 0
M,µ (λ) H,η (λ)
The eigenvalues of (3.187), subject to θ(0) = θ(1) = 0, namely, λ0 , are
the solutions of J1 (γ(λ)) = 0, where J1 is the Bessel function of the first
kind of order one and the corresponding eigenfunction is J1 (γ(λ0 )s).
© 2001 by Chapm
Remarks: Global bifurcation results for the above problem are obtained
in [22] by transforming this problem as follows: we substitute the con-
stitutive relations (3.173) into (3.180a,b,c) and carry out the indicated
differentiations; the positive-definiteness of the first Jacobian matrix in
(3.174) allows us to solve the resulting equations for sν , sη , and sµ
and an application of (3.165) then produces
L1ρ ≡
d
(sρ ) − ρ/s = f1
ds
d
L2 z ≡ (sz ) = f2 (3.189)
ds
d
L3 θ ≡ (sθ ) − (θ/s) = f3
ds
and
0
N,ν ρ (1) + N,τ0 ρ(1) = −N̂ (S(1))
(3.190b)
0
−λ + N,ν ρ (1) + N,τ0 ρ(1)
If we define u = (u1 , u2 , u3 ) by
√
(L1 ρ)(s) ≡ √su1 (s)
(L2 z)(s) ≡ su2 (s) (3.191)
√
(L3 θ)(s) ≡ su3 (s)
© 2001 by Chapm
Now, let ν # (τ, ν) denote the unique solution of the equation N̂ (τ, ν) = n
and set T # (τ, n) ≡ T̂ (τ, ν # (τ, n)). Then the boundary-value problem
(3.202), (3.203) is equivalent to
d
(sτ ) = ν # (τ, n) (3.199a)
ds
d
(sn) = T # (τ, n) (3.199b)
ds
sτ (s) → 0, as s → 0 (3.199c)
It has been proven in [22] (see also [161]) that, for λ ≥ 0, (3.196), (3.197)
has a unique solution r0 (·, λ) and, equivalently, (3.199a-d) has a unique
solution (τ0 (·, λ), N 0 (·, λ)); to analyze the buckling problem, we set
T 0 (s, λ) − N,τ0 (s, λ)
A1 (s, λ) ≡ ,ν
o
2N,ν (s, λ)
T,τ0 (s, λ)
B1 (s, λ) ≡ A21 (s, λ) + 0 (3.200)
N,ν (s, λ)
α1 (λ) ≡ A1 (0, λ)
β1 (λ) ≡ B1 (0, λ)
Then, the following result has been established in [22]: if −∞ < N 0 (0, λ)
< 0, the limits of τ0 (s, λ) and ν0 (s, λ), as s → 0, exist and are equal and
positive. Moreover, α1 (λ) + β1 (λ) ≥ 1 or, equivalently,
and there are numbers A(λ), B(λ), C(λ), and D(λ) such that
r0 (s, λ) sτ0 (0, λ) A(λ)
= + sα1 (λ)+β1 (λ)
N 0 (s, λ) N 0 (0, λ) B(λ)
(3.202)
C(λ)
+ ρ2[α1 (λ)+β1 (λ)] + O(s2[α1 (λ)+β1 (λ)] ,
D(λ)
as s → 0
Remarks: For the aelotropic, nonlinear elastic circular plate, the lin-
earized problem is of the form
$
0
sM,µ (s, λ)θ − 2M,µ
(s, λ)A3 (s, λ)θ
$ θ (3.203)
(s, λ) θ + M,µ (s, λ) A23 (s, λ) − B32 (s, λ)
0 )
M,σ
s
$ 0
+sN (s, λ) N (s, λ)/H,η (s, λ) − r0 (s, λ) θ = 0
0 0
where
Σ0,µ (s, λ) − M,σ
0
(s, λ)
A (s, λ) ≡
3 0
2M,µ (s, λ)
(3.204)
Σ0,σ (s, λ)
B 3 (s, λ) ≡ A 2 (s, λ) +
3 0 (s, λ)
M,µ
with
0
M,µ (s, λ) = M,µ (r0 (s, λ)/s, r0 (s, λ), 0, 0, 0), (3.205)
the value of M,µ at the trivial state (τ0 (0, λ), N (0, λ)). As has been
noted in [22], the analysis of (3.203) is a far more difficult task than the
analysis of the linearized problem in the transversely isotropic case, i.e.,
(3.187). The expansions (3.202) may be used as the basis of an analysis
to determine the nature of the eigenvalues and eigenfunctions of (3.203)
subject to the boundary conditions θ(0) = θ(1) = 0.
© 2001 by Chapm
nonlinear elastic equivalent of a Johnson-Urbanik type model has yet
to be formulated and studied, while serious numerical studies of the
geometrically exact model for nonlinear elastic buckling of a circular
plate do not appear to have been carried out. The results we have
presented for viscoelastic buckling refer, as they must, to a dynamic as
opposed to a static problem, and the limiting cases for times t = 0 and
t → ∞ have not been subjected to serious numerical analysis. Finally,
the models of plastic buckling we have presented are highly dependent
on the type of flow rule assumed, on the nature of the hardening law, and
even on whether the plate, assumed to be subject to a uniform applied
radial stress at its edge, buckles in the elastic domain (below yield) or in
the plastic domain (above yield). Even within the domain of the linear
elastic circular plate, the case of the rectilinearly orthotropic circular
plate appears to have not been studied at all. Thus, our discussion
here is focused, primarily, on the variations in initial and postbuckling
behavior for linearly elastic circular plates with cylindrically orthotropic
symmetry (including isotropic response as a special case) as one varies
the degree of orthotropy and the support conditions along the edge of
the plate.
© 2001 by Chapm
dially) compressive load along its edge at r = a are depicted in Fig.
3.20; these graphs show the postbuckling variations of transverse de-
flection and radial shortening, respectively, with respect to the applied
compressive radial loading where the applied load has been normalized
by the buckling load of a similar isotropic circular plate. The graphs in
Fig. 3.20 correspond to orthotropy ratios of β = 1, 4, 25, and 64 and it
follows, e.g., that the load vs. deflection curves become steeper as the
orthotropy ratios increase, i.e., the plates with higher orthotropy ratios
are stiffer with respect to their postbuckling response (in addition to
having higher buckling loads). Figure 3.21 depicts the distributions of
normalized radial and circumferential stresses for the clamped, cylin-
drically orthotropic circular plate while Fig. 3.22 depicts, for the same
situation, the graphs of the normalized radial and circumferential bend-
ing moments; both membrane and bending stresses tend to increase as
the degree β of orthotropy increases. As we have indicated earlier, for
the simply supported case, the qualitative nature of the load vs. de-
flection and load vs. radial shortening curves are similar to those for
cylindrically orthotropic, clamped circular plates, e.g., increases in the
orthotropy ratio result in an increase in the postbuckling stiffness and
a reduction in the transverse deflection. One difference between the
clamped and simply supported cases (see Fig. 3.23 is that the zone of
normalized tensile radial stresses which surrounds the origin of the plate
at r = 0, is much larger, for each orthotropy ratio β, for the simply
supported cylindrically orthotropic circular plate than it is for the cor-
responding clamped plate. In Table 3.3, we indicate how the maximum
displacement, i.e., w(0), and the radial shortening of the circular plate,
vary with increasing orthotropy ratio β in both the clamped and sim-
ply supported cases; for both types of support the maximum deflection
decreases and the radial shortening increases as β increases.
Remarks: Although we have presented concrete analytical and graphi-
cal results for elastic-plastic plates, we will defer a discussion of compar-
ative results for the initial and postbuckling behavior of such plates to
Chapter 5, i.e., until we discuss the critical issue of imperfection sensitiv-
ity for such plates; such comparative results are already clearly present,
however, in Figs. 3.26–3.29.
3.4 Initial Buckling/Postbuckling Figures, Graphs,
and Tables: Circular Plates
DE BC, at r = 0 BC, at r = R
dp0
L0 : Gp0 = 0 =0 p0 = p̄0
dr
dq1
L1 : η 2 Gq1 + p0 q1 = 0 =0 Bν q1 = 0
dr
1 2 dp2
L2 : Gp2 = q =0 p2 = p̄2
2 1 dr
dq3
L3 : η 2 Gq3 + p0 q3 + p2 q1 = 0 =0 Bν q3 = 0
dr
etc. ··············· ········· ·········
© 2001 by Chapm
Table 3.2 Linear Buckling Loads [160]
β = Eθ /Er Clamped Simple
(1) (2) (3)
1 14.6819 4.1977
2 24.0349 7.1979
5 49.2942 15.5208
8 73.0251 23.4736
Table 3.3 Maximum Displacement at Load (λ/λcr ) = 3
[155] (From Sherbourne, A.N. and Panday, M.D., Postbuck-
ling of Polar Orthotropic Circular Plates—Retrospective, J.
Eng. Mech. Div., ASCE, 118, 2087, 1992. Reproduced by
permission of the publisher, ASCE.)
β Clamped Simple
w0 uR w0 uR
(1) (2) (3) (4) (5)
1 3.4738 0.1733 2.84 0.0713
2 3.29 0.1723 2.7252 0.0954
5 3.1705 0.202 2.657 0.1327
8 3.14 0.23 2.6426 0.1583
© 2001 by Chapm
Table 3.4 Buckling Data for the Elastic-Plastic Circular Plate [141] (Reprinted from Int. J. Mech.
Sci., 17, Needleman, A., Postbifurcation behavior and imperfection sensitivity of elastic-plastic cir-
cular plates, 1–13, 1975, with permission from Elsevier Science.)
Boundary
∗
n v ρ σc /σy condition β ζmax /t σmax /σc rmax /Pc
1
12 1.28 1.02 S.S. 1.999 0.002 1.002 1.0
3
1
12 2.0 1.12 cl. 3.832 0.016 1.007 0.52
3
1
4 2.0 1.33 cl. 3.832 0.019 1.010 0.52
3
∗
S.S., Simply supported; cl., clamped.
FIGURE 3.1
Circular plate in edge compression [32]. (From Bulson, P.S.,
The Stability of Flat Plates, American Elsevier, N.Y., 1969)
© 2001 by Chapm
FIGURE 3.2
Buckling of circular plates with restrained edges. (Adopted, in
modified form, from [146].)
FIGURE 3.3
Second buckling mode of a circular plate [32]. (From Bulson,
P.S., The Stability of Flat Plates, American Elsevier, N.Y.,
1969)
© 2001 by Chapm
FIGURE 3.4
Buckling coefficient for circular plates [32]. (From Bulson, P.S.,
The Stability of Flat Plates, American Elsevier, N.Y., 1969)
FIGURE 3.5
Radial membrane stresses. (Adopted, in modified form, from
[143].)
© 2001 by Chapm
FIGURE 3.6
Circumferential membrane stress. (Adopted, in modified form,
from [143].)
FIGURE 3.7
Radial bending stress. (Adopted, in modified form, from
[143].)
© 2001 by Chapm
FIGURE 3.8
Slope and deflection. (Adopted, in modified form, from [143].)
FIGURE 3.9
Slope and deflection. (Adopted, in modified form, from [143].)
© 2001 by Chapm
FIGURE 3.10
Large deflections of buckled circular plates. (Adopted, in mod-
ified form, from [147].)
FIGURE 3.11
Radial membrane stresses for clamped and hinged circular
plates. (Adopted, in modified form, from [149].)
© 2001 by Chapm
FIGURE 3.12
Notation for the circular plate [262] (Reprinted from Int. J.
Mech. Sci., 23, Das, S., Note on Thermal Deflection of Regular
Polygonal Viscoelastic Plates, 323-329, 1981, with permission
from Elsevier Science.)
FIGURE 3.13
Postbuckling solutions: A, based on (3.55) for simple supports;
B, Friedrichs and Stoker [143] with simple supports; C, based
on (3.58) with clamped boundary conditions.
© 2001 by Chapm
FIGURE 3.14
Form of the buckling mode w1 (r). (Adopted, in modified form,
from [143].)
FIGURE 3.15
Buckling of circular plate with cylindrical orthotropy.
(Adopted, in modified form, from [151].)
© 2001 by Chapm
FIGURE 3.16
Clamped circular plates with anti-symmetric buckling modes
[32]. (From Bulson, P.S., The Stability of Flat Plates, Ameri-
can Elsevier, N.Y., 1969)
FIGURE 3.17
Simply supported circular plate buckling in first mode.
(Adopted, in modified form, from [151].)
© 2001 by Chapm
FIGURE 3.18
Simply supported circular plate buckling in anti-symmetric
mode. (Adopted, in modified form, from [151].)
FIGURE 3.19
Comparative study: clamped plate (β = 2) [155]. (From Sher-
bourne, A.N. and Panday, M.D., Postbuckling of Polar Or-
thotropic Circular Plates - Retrospective, J. Eng. Mech. Div.,
ASCE, 118, 2087, 1992. Reproduced by permission of the pub-
lisher, ASCE.)
© 2001 by Chapm
FIGURE 3.20
Postbuckling behavior of clamped plates: (a) load vs. trans-
verse deflection (b) load vs. radial displacement [155]. (From
Sherbourne, A.N. and Panday, M.D., Postbuckling of Polar
Orthotropic Circular Plates - Retrospective, J. Eng. Mech.
Div., ASCE, 118, 2087, 1992. Reproduced by permission of
the publisher, ASCE.)
FIGURE 3.21
Membrane stresses in clamped plates (Λ = 3): (a) normal-
ized radial stress and (b) normalized circumferential stress
[155]. (From Sherbourne, A.N. and Panday, M.D., Postbuck-
ling of Polar Orthotropic Circular Plates - Retrospective, J.
Eng. Mech. Div., ASCE, 118, 2087, 1992. Reproduced by
permission of the publisher, ASCE.)
© 2001 by Chapm
FIGURE 3.22
Bending moments in clamped plates (Λ = 3): (a) normal-
ized radial moment and (b) normalized circumferential mo-
ment [155]. (From Sherbourne, A.N. and Panday, M.D., Post-
buckling of Polar Orthotropic Circular Plates - Retrospective,
J. Eng. Mech. Div., ASCE, 118, 2087, 1992. Reproduced by
permission of the publisher, ASCE.)
FIGURE 3.23
Membrane stresses in simply supported plates (Λ = 3): (a)
normalized radial stress and (b) normalized circumferential
stress [155]. (From Sherbourne, A.N. and Panday, M.D., Post-
buckling of Polar Orthotropic Circular Plates - Retrospective,
J. Eng. Mech. Div., ASCE, 118, 2087, 1992. Reproduced by
permission of the publisher, ASCE.)
© 2001 by Chapm
FIGURE 3.24
Postbuckling behavior of a circular viscoelastic plate [140].
(Reprinted from Mech. Res. Comm., 17, Brilla, J., Postbuck-
ling Analysis of Circular Viscoelastic Plates, 263-269, 1990,
with permission from Elsevier Science.)
FIGURE 3.25
Uniaxial stress-strain curve and bifurcation stresses for
clamped and simply supported circular plates (n = 12) [141].
(Reprinted from Int. J. Mech. Sci., 17, Needleman, A., Postb-
ifurcation Behavior and Imperfection Sensitivity of Elastic-
Plastic Circular Plates, 1-13, 1975, with permission from Else-
vier Science.)
© 2001 by Chapm
FIGURE 3.26
Curves of average applied stress vs. buckling deflection for
a simply supported circular plate ν = 13 , σc /σy = 0.72 [141].
(Reprinted from Int. J. Mech. Sci., 17, Needleman, A., Postb-
ifurcation Behavior and Imperfection Sensitivity of Elastic-
Plastic Circular Plates, 1-13, 1975, with permission from Else-
vier Science.)
FIGURE 3.27
Curves of average applied stress vs. buckling deflec-
tion for a simply supported circular plate (ν = 13 , η̄ = 1.28,
σc /σy = 1.02) [141]. (Reprinted from Int. J. Mech. Sci.,
17, Needleman, A., Postbifurcation Behavior and Imperfec-
tion Sensitivity of Elastic-Plastic Circular Plates, 1-13, 1975,
with permission from Elsevier Science.)
© 2001 by Chapm
FIGURE 3.28
Curves of buckling deflection vs. applied displacement
for a simply supported circular plate (ν = 13 , n = 12,
η̄ = 1.28, σc /σy = 1.02) [141]. (Reprinted from Int. J. Mech.
Sci., 17, Needleman, A., Postbifurcation Behavior and Imper-
fection Sensitivity of Elastic-Plastic Circular Plates, 1-13, 1975,
with permission from Elsevier Science.)
FIGURE 3.29
Curves of average applied stress vs. buckling deflection for a
clamped circular plate (ν = 13 , n = 12, η̄ = 2.0, σc /σy = 1.12)
[141]. (Reprinted from Int. J. Mech. Sci., 17, Needleman,
A., Postbifurcation Behavior and Imperfection Sensitivity of
Elastic-Plastic Circular Plates, 1-13, 1975, with permission
from Elsevier Science.)
© 2001 by Chapm
FIGURE 3.30
Development of plastic zone (shaded area) for a clamped
circular plate ν = 13 , n = 12, η̃ = 2.0, σc /σy = 1.12 [141].
(Reprinted from Int. J. Mech. Sci., 17, Needleman, A., Postb-
ifurcation Behavior and Imperfection Sensitivity of Elastic-
Plastic Circular Plates, 1-13, 1975, with permission from Else-
vier Science)
Chapter 4
In this chapter, we present some of the basic results which pertain to the
buckling and postbuckling behavior of annular plates. We begin by look-
ing at annular plates which exhibit linear elastic behavior, treating first
the problems of initial buckling for plates which possess either isotropic
or cylindrically orthotropic symmetry. Next, we present an analysis of
the postbuckling behavior of isotropic and cylindrically orthotropic lin-
ear elastic annular plates; this is followed by a discussion of the nature
of the prebuckling stress distributions and a study of the initial buckling
problem for a linearly elastic annular plate which exhibits rectilinearly
orthotropic material symmetry. As was the case for buckling and post-
buckling of circular plates possessing rectilinear orthotropic symmetry,
the corresponding problem for annular plates is once again complicated
by the mismatch between the geometry of the plate and its material
symmetry behavior; indeed, for the linearly elastic annular plate, the
problem of delineating the postbuckling behavior for the case where the
plate exhibits rectilinear orthotropic symmetry appears to have not been
treated at all. For plates which are annular and exhibit either nonlinear
elastic, viscoelastic, or elastic-plastic behavior, very few results appear
to exist in the literature; those results which do exist for annular plates
which do not exhibit linear elastic response are presented in §4.2. In
§4.3 we offer a comparison of the initial and postbuckling behavior of
annular plates which results from varying the support conditions along
the edges of the plate (inner and outer), the type of material symmetry,
and the nature of the constitutive response.
© 2001 by Chapm
4.1 Plates with Linear Elastic Behavior
Annular plates exhibiting linear elastic behavior may be classified ac-
cording to whether the plate possesses isotropic symmetry, cylindrically
orthotropic symmetry, or rectilinear orthotropic symmetry.
λha2 2
∇4n un + ∇n u n = 0 (4.5)
K
where
d2 1 d n2
∇2n = + − (4.6)
dρ2 ρ dρ ρ2
The solutions to (4.5) may be expressed in terms of Bessel functions and
then studied for a variety of support conditions along the edges at r = b
and r = a; the loading condition of uniform compression along both
edges has been studied by a number of authors, but the most general
treatment appears to belong to Yamaki [162], whose results we present
below.
(i) The edges at r = a and r = b are clamped.
∂w
In this case, the boundary conditions are that w = 0 and = 0, at
∂ρ
both r = a and r = b. The general solution of (4.5) has the form
12
12
λhr2 λhr2
u0 = A0 J0 + B0 Y0 (4.7)
K K
r
+C0 + D0 lnρ ( for n = 0), ρ =
a
and
12
12
λhr2 λhr2
un = An Jn + Bn Yn
K K
(4.8)
r
+Cn ρn + Dn ρ−n ( for n = 1, 2, · · ·), ρ =
a
with the A0 , B0 , An , Bn , etc., arbitrary constants of integration and
Jn , Yn , respectively, the Bessel functions of the first and second kinds of
order n.
© 2001 by Chapm
By imposing the clamped edge boundary conditions on the general
solution (4.7), (4.8) we obtain, for each fixed n, four homogeneous linear
equations in the An , Bn , Cn , and Dn ; the vanishing of the determinant
of the matrix of coefficients of this system then provides the condition
for determining the eigenvalues of the boundary value problem associ-
ated with (4.5) and, hence, the critical edge stress λcr . The solution of
the initial buckling problem for this case may be displayed in graphical
form as in Fig. 4.1, which is taken from Yamaki [161]; the situation
in which both edges are clamped is represented by the curve (i) where
the buckling parameter k is defined by λcr = kK/a2 h. In earlier work
on the annular plate problem in which both edges were subject to uni-
form compression (i.e., Olsson [163] , Shubert [164]) only axisymmetric
buckling was studied but, as we will see later in this subsection, the
axisymmetric buckled state is not always necessarily associated with the
lowest critical load. It should be noted from Fig. 4.1 that the number
of diametric nodes increases as the hole diameter increases. As has been
b
noted in [162], it can be shown that as → 1, the values of the buck-
a
ling parameter k approach those associated with a clamped infinite strip
subject to uniform compression.
(ii) The edge at r = a is clamped, while the edge at r = b is
simply supported.
In this case, as we have demonstrated in Chapter 1, the boundary con-
∂w
ditions are w = 0, = 0, at r = a, while w = 0 and
∂ρ
∂ 2 w ν ∂w 1 ∂ 2 w
+ + =0
∂ρ2 ρ ∂ρ ρ ∂θ2
at r = b. The initial buckling behavior is shown as curve (ii) in Fig.
4.1, which indicates that the annular plate will first buckle with one
diametric note but that, as the hole size increases, this buckling mode
changes to the one (n = 0) which reflects axisymmetric buckling of the
plate.
(iii) The edge at r = a is clamped, while the edge at r = b is
free.
∂w
Once again, we have w = 0, = 0, at r = a while, as shown in
∂ρ
Chapter 1, at the free edge at r = b we require that
∂ 2 w ν ∂w 1 ∂ 2 w
+ + =0
∂ρ2 ρ ∂ρ ρ ∂θ2
and
∂ ∂ 2 w 1 ∂w 1 ∂2w
+ +
∂ρ ∂ρ2 ρ ∂ρ ρ2 ∂θ2
(1 − ν) ∂ ∂ 1 ∂w ∂w
+ +K =0
ρ ∂θ ∂ρ ρ ∂θ ∂ρ
the second condition indicating that the edge shearing force at r = b is
equal to the normal component of the applied load at r = b. The initial
buckling behavior is shown as curve (iii) in Fig. 4.1 and the annular
isotropic plate, in this case, always buckles into the axisymmetric form
(n = 0) regardless of the hole size.
(iv) The edge at r = a is simply supported, while the edge at
r = b is clamped.
This case is the same as case (ii) but the roles of the edges at r = a, r = b
have been exchanged with respect to the boundary support conditions.
The initial buckling behavior is described by curve (iv) in Fig. 4.2 and
the annular plate first buckles into a mode with one diametric node but
then changes to the axisymmetric mode as the hole size increases. If
we compare the graphs for cases (ii) and (iv) we see that they are very
similar but that the values of the buckling parameter k for a fixed value
of b/a are slightly lower in case (iv).
(v) The edges at r = a and r = b are both simply supported.
In this situation, the boundary conditions are just
w=0
at r = a, r = b
∂ 2 w ν ∂w 1 ∂ 2 w
2 + + 2
=0
∂ρ ρ ∂ρ ρ ∂θ
The graph depicting the initial buckling behavior for this case is shown
b √
as curve (v) in Fig. 4.2; as → 0, k → 3.625.
a
(vi) The edge at r = a is simply supported, while the edge at
r = b is free.
The conditions at r = a are delineated in case (v) above, while the
conditions that the edge at r = b be free are delineated in case (iii)
above. The initial buckling behavior in this case is shown as curve (vi)
in Fig. 4.2; the plate always buckles into the axisymmetric form and
the buckling stress decreases as the hole size increases. This decrease
of buckling stress with an increasing hole size does not happen for any
© 2001 by Chapm
other case considered here except case (viii), which deals with the same
boundary support conditions but with the roles of the edges at r = a
and r = b interchanged.
(vii) The edge at r = a is free and the edge at r = b is
clamped.
The boundary support conditions here are the same as for case (iii) but
with the roles of the edges at r = a and r = b interchanged. The initial
buckling curve is shown as curve (vii) in Fig. 4.1 and the results indicate
a behavior which differentiates this case from the others we have con-
sidered: as the hole size is increased, first one diametric node appears
in the buckling mode, then two nodes, and again, one. Finally, at yet
larger hole sizes, the annular isotropic plate buckles into the axisymmet-
ric mode with n = 0.
(viii) The edge at r = a is free, while the edge at r = b is
simply supported.
At r = b, the boundary support conditions are those delineated in case
(v), while at r = a they are the conditions delineated in case (iii). The
initial buckling curve is depicted as curve (viii) in Fig. 4.1 so that the
initial buckling behavior in this situation is similar to that for case (vi).
Cases (i)–(viii), considered above, correspond to the situation in which
the isotropic annular plate is subject to uniform compression along both
its inner and outer edges; the basic reference, as noted earlier, is the work
of Yamaki [161]. The next basic situation concerns the initial buckling
behavior of an isotropic annular (linear elastic) plate which is subjected
to compression or tension along its inner edge at r = b; the basic ref-
erence for this situation is the paper of Mansfield [165]. Because the
solution in [165] is drawn from the results for an infinite plate supported
along two concentric circles and subjected to radial stresses along the
inner circle, it is applicable only if the outer edge is supported by a mem-
ber possessing the required tensile stiffness. Mansfield [162] has noted
that, when a uniform radial compressive stress is applied to the circum-
ference of a circular hole in an infinite plate, the resulting radial stresses
in the plate decay inversely as the square of the distance from the center
of the circle; there are also tensile hoop stresses of the same magnitude
and varying in the same way. An identical stress distribution exists in
a similarly loaded annular plate of thickness h, say, which is bounded
by circles of radii b and a, if the outer circle is supported by a member
of sectional area ah/(1 + ν). Mansfield, [165], also notes that such an
annular plate will buckle as soon as the applied radial stress achieves a
critical value which depends on the boundary support conditions; if the
radial stress is compressive, the plate will buckle into an axially symmet-
ric mode, while, if the radial stress is tensile, the plate will buckle into
a number of circumferential waves because the hoop stresses are now
compressive. The situation considered in [165] is depicted in Fig. 4.3.
If we denote the hoop stress by σθθ , then the equation of equilibrium
for the infinite plate follows from (4.2), where it is now convenient to
replace σrr → −σrr and σθθ → −σθθ so that
1 1
K∇4 w + hσrr w,rr + hσθθ w,r + 2 w,θθ = 0 (4.9)
r r
If we denote the radial stress (at r = b), which causes buckling, by λcr ,
then the radial and hoop stresses in the plate are of the same magnitude
and are given by
2
b
σrr = −σθθ = λcr (4.10)
r
Thus, as already noted above, these stresses decay inversely as the square
of the distance r from the center of the circle of radius b and the same
stress condition obtains if an outer circle of radius a > b in the infinite
plate is bounded by a member of sectional area equal to ah/(1 + ν).
Using the stress distribution (4.10), we rewrite (4.9) in the form
2 2
∂ 1 ∂ 1 ∂2 ∂ w 1 ∂w 1 ∂2w
+ + + + 2 2 (4.11)
∂r2 r ∂r r2 ∂θ2 ∂r2 r ∂r r ∂θ
k ∂ 2 w 1 ∂w 1 ∂2w
+ − − = 0,
r2 ∂r2 r ∂r r2 ∂θ2
λcr b2 h
with k = , and again seek solutions of the form
K
w(r, θ) = f (r) cos nθ (4.12)
with f (r) the transverse deflection function and n the number of dia-
metric nodes; this problem has been solved by Mansfield, [165], for a
number of boundary support conditions. When λcr is compressive, the
buckling mode, as we have already noted, is axially symmetric; the re-
sults obtained in [165] for this subcase are depicted in Fig. 4.4 with
1
k(1 − γ)2 /γ graphed against for various combinations of boundary
γ
support conditions at r = a and r = b. Solutions of the initial buckling
problem are also presented in [165] for the case where buckling occurs
© 2001 by Chapm
as a result of tensile stresses along the edge at r = b; such solutions are
derived for those situations in which the edges at r = a and r = b are
either both clamped or both simply supported. When buckling occurs
because of tensile stresses applied along the edge at r = b, a number of
diametric nodes are formed which decrease in number as the hole size
is decreased; this behavior is depicted in Fig. 4.5. It is shown in [165]
that, as a → ∞, and γ → 0, the compressive buckling stress for axially
symmetric buckling converges to K/b2 h, i.e., k → 1, provided we have
zero radial slope at the inner edge at r = b; furthermore, the tensile
buckling stress λcr → −3K/b2 h. Therefore, the tensile buckling stress
associated with radial pressure applied along the edge of a hole in an
infinite plate is three times the compressive buckling stress.
Remarks: Using (4.11), (4.12), it may be shown that the equation
governing the transverse deflection f (r) has the form
d4 f 2 d3 f 1 + 2n2 − γ d2 f
+ −
dr4 r dr3 r2 dr2
(4.13)
1 + 2n − γ df
2
n 4−n −γ
2 2
+ − f =0
r3 dr r4
and the particular case of buckling with axial symmetry is then obtained
by setting n = 0 in (4.13). Solutions of (4.13), which is homogeneous in
r, may be expressed in the form
#
r
f (r) = Ai ρ1+βi (4.14)
i=1
w = w,r = 0, at R = a (1.139 )
and, at r = b,
∂2w 1 ∂w 1 ∂2w
+ν + 2 2 =0 (1.140 )
∂r2 4 ∂r r ∂θ
2 2
∂ ∂ w 1 ∂w 1 ∂ w
2
+ + 2 2
∂r ∂r r ∂r r ∂θ
(1.141 )
1 − ν ∂ 1 ∂2w 1 ∂w
+ − =0
r ∂θ r ∂r∂θ r2 ∂θ
If we set
Nr |r=a = σrr |r=a h ≡ −N (= −λh) (4.18)
then the prebuckling stress distribution in the annular circular plate is
given by
a2 b2
Nr = −N 2
0
1− 2 (4.19a)
a − b2 r
a2 b2
Nθ0 = −N 2 1+ 2 (4.19b)
a − b2 r
© 2001 by Chapm
where
N a2 b2
µ= (4.21)
K (a2 − b2 )
dA0
where Ψ = . In terms of the variables Ψ and ρ, equation (4.26) is
dρ
subject to the boundary conditions
a
Ψ = 0, at ρ = µ 2
1
b
dΨ ν 1
+ Ψ = 0, at ρ = µ 2 (4.27)
dρ ρ
d dΨ 1 1
+ Ψ = 0, at ρ = µ 2
dρ dρ ρ
dφ
By using the transformations A1 = ρφ and Ψ = , (4.29) may be
dρ
reduced to
d2 Ψ 3 dΨ 1 − p2 C
+ + 1 + Ψ= 3 (4.30)
dρ2 ρ Cρ ρ2 ρ
where p2 = µ + 4, and C is an arbitrary constant. For the general
solution of (4.30) we set
π J−p (ρ) Jp (ρ)
S−1,p (ρ) = Jp (ρ) dρ − J−p (ρ) dρ (4.31)
2 sin(pπ) ρ ρ
Then,
1
Ψ(ρ) = (AJp (ρ) + BJ−p (ρ) + CS−1,p (ρ)) (4.32)
ρ
© 2001 by Chapm
with A and B arbitrary constants. If we rewrite Ψ(ρ) in terms of A1 (ρ),
in (4.32), we find that
dA1
ρ − A1 = ρ [AJp (ρ) + BJ−p (ρ) + CS−1,p (ρ)] (4.33)
dρ
dA1 a
1
A1 =
dρ
= 0, at ρ = µ 2
b
d2 A
1 ν dA1 1
+ 2 ρ − A1 = 0, at ρ = µ 2
dρ 2 ρ dρ
d3 A
1 d 2 A1 (3 − ν) dA1 (3 − ν)
1 1
+ − + A1 = 0, at ρ = µ 2
dρ3 ρ2 dρ2 ρ2 dρ ρ3
(4.34)
If we impose the boundary conditions (4.34) on the general solution
(4.33), we obtain the system of equations
1 a 1 a
AJ µ + BJ−p µ2 =0
2
p
b b
1 1
1
A (1 + ν − p) J−p µ 2 + µ 2 Jp−1 µ 2 (4.35)
1 1
+B (1 + ν + p) J µ2 + µ2 J
1
µ2 =0
−p −p−1
1 a 1 1
1
Jp µ 2 + (1 + ν − p) Jp µ 2 + µ 2 Jp−1 µ 2 = 0 (4.36)
b
and this leads to the initial buckling curve for n = 1, which is shown in
Fig. 4.6.
Approximate solutions of the general problem in Majumdar [166] are
generated by minimizing the potential energy using a Rayleigh-Ritz pro-
cedure where
2 2
a 2π
1 ∂w 1 ∂w
V = Nr + 2 Nθ
2 b 0 ∂r r ∂θ
2
∂ 2 w 1 ∂w 1 ∂2w
K + +
∂r2 r ∂r r2 ∂θ2
(4.37)
∂ 2 w 1 ∂w 1 ∂2w
−2 (1 − ν) 2 + 2 2
∂r r ∂r r ∂θ
2
1 ∂2w 1 ∂w
+2 (1 − ν) − rdθdr
r ∂r∂θ r2 ∂θ
© 2001 by Chapm
(1.141 ), Troger and Steindl [66], [167] proceed as follows: first of all,
the linearized equation is written in the form
Lw = N · M w (4.39)
and
b2 1 1
Mn g = − − 2 g,rr
1 − b2 b2 r
(4.43b)
1 1 1 n2
+ 2
+ 2 g,r − 2 g
b r r r
© 2001 by Chapm
double eigenvalues for b & 0.51, while for 0 ≤ b . 0.51 the axisymmet-
ric buckling mode corresponds to a simple eigenvalue; the nongeneric
points of intersection correspond to three-and four-fold eigenvalues. As
noted in [66], the large number of multiple eigenvalues is due to the cir-
cular symmetry of the problem. We will return to a further discussion
of the results in [66] and [167] in the course of our examination of the
postbuckling behavior of annular plates.
© 2001 by Chapm
Eθ
values of the orthotropy ratio β = , with β = 1 corresponding to an
Er
isotropic elastic (circular) annular plate. In fact, β ranges from β = 1
(isotropic) to β = 30, while b/a ranges from zero (a circular plate) to
b/a = 0.8, which approaches a ring. For all the calculations displayed,
Poisson’s ratios νr = νθ = 0.3 were used. As noted by the authors in
[70], each figure referenced above shows the pronounced effect of the or-
thotropic nature of the material on the critical buckling loads, and shows
comparative results obtained by Shubert [164], Mossakowski [151], Ya-
maki [162], Woinowsky-Krieger [170], Roza [171], Meissner [172], and
Pandalai and Patel [153] for those special subcases in which the cylin-
drically orthotropic annular plate is either a circular plate (i.e., b = 0)
or an isotropic plate (or both).
with
d2 3 1−β
L= + + (4.53)
dρ2 ρ ρ2
which, together with the boundary support conditions, must be solved
in order to determine the axisymmetric behavior of annular plates sub-
jected to loads greater than the critical buckling load. In dimensionless
form, the boundary conditions are
S = 0
b
at ρ = : dv , (4.54)
a ρ + (1 + νr ) V = 0
dρ
© 2001 by Chapm
i.e., the inner edge is free and,
S = −P
at ρ = 1 : , (4.55)
V =0
i.e., the outer edge is clamped and loaded, with P the magnitude of
the external compressive load. A finite-difference scheme, similar to
that employed by Keller and Reiss [177], is used to solve the boundary
value problem consisting of (4.52)–(4.55) and to obtain results for the
variation, with ρ, of the following quantities in nondimensional form:
the deflection w, the membrane tangential stress SθM ,
dS
SθM = ρ + S, (4.56a)
dρ
B
the bending radial stress SR ,
B dV
SR =ρ + (1 + νr )V, (4.56b)
dρ
dV
SθB = νθ ρ + (1 + νr ) V (4.56c)
dρ
νr Eθ
= =β (4.57)
νθ Er
© 2001 by Chapm
variables r w
ρ= , W = , φ = Φ/haEθ
a a (4.58)
p = P/haEθ
and rewrites the governing system of von Karman equations in the form
d4 W 2 d3 W β d2 W β dW
+ − + 3 (4.59)
dρ4 ρ dρ3 ρ2 dρ2 ρ dρ
a 2
1 d dW
= 12 β− νθ2 φ
h ρ dρ dρ
2
d2 φ 1 dφ β 1 dW
+ − φ = − (4.60)
dρ2 ρ dρ ρ 2ρ dρ
with boundary support conditions
2
d W νθ dW
+ =0
dρ 2 ρ dρ
d3 W 1 d2 W β dW
+ − (4.61a)
dρ3
ρ dρ2 ρ2 dρ
a 2 1 dW
−12 β − νθ2
φ =0
h ρ dρ
at ρ = b/a and either
dW
W = = 0, at ρ = 1, for a clamped edge (4.61b)
dρ
or
d2 W dW
W = + νθ = 0, at ρ = 1,
dρ2 dρ (4.61c)
for a simply supported edge
Along the edge at ρ = 1 we also have the uniform, radial, compressive
load of magnitude P. In the calculations presented in Huang [174], the
aspect ratio γ = b/a is taken to be 0.4, while the orthotropy ratio varies
over the range β = 1.0, 3.0, 5.0, and 10.0, for the case where the outer
edge at r = a is simply supported, while β = 5.0 for the case in which
the outer edge at r = a is clamped. With the buckling parameter λ
given by
a 2
λ = 12 β − νθ2 p (4.62)
h
the postbuckling (and initial buckling) behavior of the cylindrically or-
thotropic annular plate is indicated in Table 4.1 and displayed graph-
ically in Figs. 4.26 and 4.27 where ν = νθ ; the corresponding distri-
butions of (normalized) radial bending stresses, circumferential bend-
ing stresses, radial membrane stresses, and circumferential membrane
stresses are graphed in Figs. 4.28, 4.29, 4.30, and 4.31, respectively,
where δ = w(γ)/h, while the rigidity D0 is given by
D0 = Eθ h3 /12 1 − νθ2 (4.63)
Finally, based on the analysis in [174], we show in Fig. 4.32 the (normal-
ized) transverse displacement shapes for the annular plate as functions
of the nondimensionalized plate radius ρ, thus indicating the influence
of postbuckling load magnitudes on the plate deflection shapes.
Two other noteworthy analyses of the axisymmetric buckling behav-
ior of cylindrically orthotropic elastic annular plates are the papers of
Alwar and Reddy [176] and Dumir, Nath, and Gandhi [173]; the earlier
paper [176] presents a large deflection analysis of orthotropic annular
plates subjected to uniformly distributed loads using Chebyshev series
expansions, with a substantial part of the work geared toward dynamic
analysis. The work of Dumir, Nath and Gandhi [173] subsumes, for the
case of a static analysis, much of the work in [176]; in [173] both w and Φ
are expanded in finite power series and an orthogonal point collocation
method is used to obtain discretized algebraic equations from the gov-
erning set of von Karman differential equations. Results are presented
in [173] for annular plates with and without a rigid plug (in the hole of
radius b) and some of these results are displayed in Figs. 4.33 and 4.34.
In Fig. 4.33 we show the deflection versus load graphs for both the cases
of a clamped and a simply supported edge at r = a, and for orthotropy
1 1
ratios of 1, , and , as well as aspect ratios of .5 and .25; for com-
3 10
parison purposes, the results obtained in these situations by Alwar and
Reddy [176] are shown on the same graphs. Figure 4.34 displays the
same kind of results as Fig. 4.33, but this time, for the case in which
a rigid plug occupies the region 0 ≤ r ≤ b; for all the cases referenced
here the Poisson ratio νr = .25. It can be easily observed from Fig. 4.33
that the effect of varying β on the deflection response is very small in
the simply supported case as compared with the clamped case; this is
not the situation, of course, when the hole is occupied by a rigid plug,
as is made clear in Fig. 4.34.
Remarks: As we have noted earlier in this subsection, a somewhat
deeper analysis of the postbuckling behavior of elastic annular plates
© 2001 by Chapm
subjected to a uniform compressive load along the edge at r = a, with the
edge at r = a clamped and the edge at r = b free, has been carried out
in Machinck and Troger [167] as well as in Troger and Steindl [66]; both
analyses assume the plate to be isotropic and focus on the nongeneric
situation in which buckling occurs at eigenvalues of multiplicity greater
than one. In both [66] and [167] the operator ∆2 in (4.38) is inverted so
as to yield
1
Φ = − ∆−2 [w, w] (4.64)
2
and (4.64) is then inserted in (4.38) so as to provide the functional
equation
b2 1 1 2
2
F (w, N ) = ∆ w + N + 2 ∆w − 2 w,rr
1 − b2 b2 r r (4.65)
1$ −2
+ w, ∆ [w, w] = 0
2
The buckling problem is then dealt with as a problem in bifurcation
theory, after the linearized problem which governs initial buckling is
analyzed: the functional equation is reduced locally in the neighborhood
of a bifurcation point N = Ncr to a system of algebraic equations
Ai (x1 , · · · , xk , λ) = 0, i = 1, · · · , k, (4.66)
is made where x0 , x1c , x1s are the amplitudes of the buckling modes (or
critical variables) and Ψ0 = g01 (r), Ψ1c = g11 (r) cos θ, Ψ1s = g11 (r) sin θ
are the buckling modes (eigenfunctions) corresponding to the triple ei-
genvalue. The function W includes non-critical variables which corre-
spond to modes that are orthogonal to the buckling modes and can be
expressed in terms of the critical variables by means of the Liapunov-
Schmidt reduction. It may be shown [66], [167] that
2 2 2
W (x0 , x1c , x1s ) = O |x0 | + |x1c | + |x1s | (4.68)
In (4.69) the terms γ0 , γ1c , γ1s are derived from a transverse loading on
the annular plate (and thus, vanish if such a loading (imperfection) is
not present). With the inner product defined by
2π 1
(Ψi , Ψj ) ≡ Ψi Ψj rdrdθ (4.70)
0 b
1 $
A= Ψ0 , ∆−2 [Ψ0 , Ψ0 ] , Ψ0 = 0.00361
2
1 $
B = Ψ1c , ∆−2 [Ψ1c , Ψ1c ] , Ψ1c = 0.00584
2 (4.71)
1 $
Ψ0 , ∆−2 [Ψ1c , Ψ1c ] , Ψ0
C=
2
$
+ Ψ1c , ∆−2 [Ψ1c , Ψ0 ] , Ψ0 = 0.00919
© 2001 by Chapm
and
b2 1 1 2
α0 = − + 2 ∆Ψ0 − 2 Ψ0,rr, Ψ0 = 1
1 − b2 b2 r r
2
b 1 1 2
α1 = − + 2 ∆Ψ1c − 2 Ψ1c,rr, Ψ1c = 1 (4.72)
1 − b2 b2 r r
1
γ = (q, Ψ ) , γ = (q, Ψ )2 + (q, Ψ )2 2
0 0 1 1c 1s
© 2001 by Chapm
may occur in the region b ≤ r ≤ a, 0 ≤ θ < 2π, when the applied load
along the inner edge at r = b reaches a critical value; in [71] the applied
tractions are not restricted to a uniform radial loading, as in Mansfield
[165], but are assumed to be self-equilibrating and symmetric about the
principal material axes.
'
pr (θ) = Nx cos2 θ + Ny sin θ + 2Nxy cos θ sin θ 'r=b
2
' (4.76)
pθ (θ) = (Ny − Nx ) cos θ sin θ − Nxy cos2 θ − sin2 θ 'r=b
with pr (θ) the radial component, and pθ (θ) the tangential component,
of the applied traction at r = b. All stresses are required to be bounded
as r → ∞ and to assume the same values at θ = 0 and θ = 2π. In [71]
the boundary conditions (4.76) are rewritten in integral form and the
components of the applied tractions in the negative x and y directions
are then expanded as function of θ in Fourier series; the results are
subsequently substituted back into the integral form of the boundary
conditions (4.76) so as to yield equivalent conditions with respect to the
Fourier coefficients.
The boundary value problem for the prebuckling stress distribution
which we have delineated above was first solved by Lekhnitskii [51]. We
will content ourselves here with presenting some of the basic results
which address both uniform as well as nonuniform radial loading along
the edge at r = b. The basic material parameters which enter into the
analysis in [71] are the ratios E1 /E2 , E1 /G12 , and the Poisson ration ν21 ;
furthermore, in all the calculations reported in [71], it is assumed that
E2 /E1 ≥ 1 so that the y-direction is the direction of maximum stiffness.
The first problem considered in [71] vis à vis prebuckling stress distri-
butions involves the effect of varying the ratio E2 /E1 of Young’s moduli;
in that problem, the shear modulus ratio E1 /G12 is fixed at 2.6, while
ν21 = 0.3. The applied loading at r = b is taken to be a uniform radial
loading with pr (θ) = pi , pθ (θ) = 0. For E2 /E1 = 1, it can be shown that
the prebuckling stress distribution is axisymmetric, and the magnitudes
of both radial and tangential stress components are proportional to r−2 ,
while the polar stress component is identically zero. For E2 /E1 = 1,
the prebuckling stress distribution is not axisymmetric and the polar
shear stress is non-zero; sample results are displayed in Figs. 4.37, 4.38,
and 4.39 for a ratio E2 /E1 = 10, with the stress contour lines indicated
for the first quadrant, and b ≤ r ≤ 26. The most significant effect of
varying E2 /E1 (from the isotropic case) is on the distribution of the
tangential normal stresses that now exhibit a large compressive stress
concentration at r = b, θ = 0 given by
9 +
* E : * 6
: E2 E1 E1
−;
2
Nθ (b, 0) = pi − 2ν12 + 2 (4.77)
E1 E1 G12 E2
In Fig. 4.40 we show the effect of varying the ratio of Young’s mod-
uli on the tangential normal stress distribution at θ = 0; note that at
large radial distances from the edge, at r = b, the (prebuckling) tangen-
tial stress distributions in the isotropic plate and the inextensible plate
(E2 /E1 = ∞) are nearly the same. Figure 4.41 depicts a comparison
π
of the tangential normal stresses along the line θ = for the cases
2
E2 /E1 = 1 and E2 /E1 = ∞, and shows that these compressive stresses
are influenced by changes in E2 /E1 , only at small radial distances from
the edge at r = b. In Fig. 4.42, we illustrate the effect of changing
E2 /E1 on the tangential stress distribution along the edge at r = b. As
π
E2 /E1 increases, stress concentrations develop at θ = 0 and θ = . Fig-
2
ure 4.43 shows the effect of varying the ratio E2 /E1 of Young’s moduli
© 2001 by Chapm
on Ny (along the line x = b). The anisotropy ratios vary in the range
1 ≤ E2 /E1 ≤ 105 and, as this ratio increases, the stresses become com-
pressive for all y, with increasing magnitudes all along x = b, leading
to a compressive stress concentration along the line x = b. As noted in
[71], the line x = b is quite significant in a highly anisotropic annular
plate, of the type under consideration here; it tends to separate the plate
into two regions: |x| < b, in which Ny is tensile, and |x| ≥ b in which
Ny is compressive, with an abrupt change taking place at x = b. The
above discussion, when coupled with Figs. 4.37 through 4.43, highlights
the considerable influence on the prebuckling stress distribution in an
elastic (infinite) annular plate, which is exerted by variations in the ratio
E2 /E1 of Young’s moduli.
The second problem treated in [71], in relation to the nature of the
prebuckling stress distributions, involves the effect of varying the shear
modulus ratio E1 /G12 . Taking the load along the edge at r = b to
again be a uniform radial load, we show in Figs. 4.44, 4.45, and 4.46,
respectively, the effects of varying the shear modulus ratio on the radial
normal, tangential normal, and polar shear stress distributions. Figure
π
4.44 shows that, along the line θ = , the radial stress decreases rapidly,
4
even becoming slightly compressive, while Fig. 4.45 indicates the forma-
tion of two compressive stress concentrations at the edge r = b, for θ = 0
π
and θ = ; for E2 /E1 = 1 these stress concentrations are of magnitude
2
*
π E1
Nθ (b, 0) = Nθ b, = pi 1 − + 2 (1 − ν12 ) (4.78)
4 G12
π
In Fig. 4.47 we depict the radial stress distribution along the line θ =
4
for various values of E1 /G12 which range from 2.6 to 108 ; the graph
indicates
√ that a compressive stress concentration begins to form near
r = 2b as the ratio E1 /G12 increases. Figure 4.48 depicts, for b ≤
r ≤ 2b, the relative displaced shape of the elastic annular plate for a low
shear modulus (E1 /G12 = 100) when it is subjected to a uniform radial
load along the edge at r = b; the plate, as in [71], is in a state of bi-axial
compression for r > 2b.
Also considered in [71] are problems which correspond to nonuniform
distributions of applied tractions along the edge at r = b, i.e., applied
loads which are given in (Cartesian) vector form by (p1 cos θ, p2 sin θ)
where p1 and p2 both act in the negative x and y directions, respectively,
in the first quadrant of the annular a plate. In polar coordinates, the
applied load thus has the form
pr (θ) = p1 cos2 θ + p2 sin2 θ
(4.79)
pθ (θ) = [p2 − p1 ] sin θ cos θ
© 2001 by Chapm
stress distribution. In (1.109), [w, Φ] is given by (1.74) and (1.75b)
while the D̃j , j = 1, 2, 3, 6, and the coefficients D̃12 , D̃16 , and D̃26 , are
given by (1.108) and (1.110) with D1 = D11 , D2 = D22 , and D3 =
D2 ν12 + 2D66 , the principal rigidities, defined as in (1.63), (1.64). In
Cartesian coordinates (1.109), with Φ= Φ0 , assumes the simpler form
(1.68), with Φ = Φ0 (x, y), where b ≤ x2 + y 2 ≤ a.
Remarks:
(i) If V denotes the potential energy function associated with the
rectilinearly orthotropic annular plate, then for positive material coeffi-
cients E1 , E2 , ν21 , and G12 , δ 4 V > 0, which implies that at the smallest
eigenvalue associated with the initial buckling equation the associated
equilibrium state is stable with respect to small perturbations.
(ii) If w(x, y) is decomposed as
we (x, y) = we (x, −y), w0 (x, y) = −w0 (x, −y), then we (x, y) and w0 (w, y)
separately satisfy (1.68), with Φ = Φ0 (x, y), if the applied load is even
about the x-axis (the averaged stress components Nx0 and Ny0 being even
0
functions, while Nxy is an odd function about the x-axis). In addition,
the 2π-periodicity of w, as a function of r and θ, implies that w0 and its
even-order derivatives with respect to y vanish at y = 0, while odd-order
derivatives of we with respect to y also vanish at y = 0; this implies that
the symmetric and antisymmetric components of the buckling modes
are uncoupled from each other so that the buckled mode shape is either
symmetric or antisymmetric about the x-axis. In a similar fashion, for
applied loads which are symmetric about the y-axis, the buckled modes
are either symmetric or antisymmetric with respect to the y-axis. When
the loads are not symmetric with respect to the principal material di-
rections in the rectilinearly orthotropic annular plate, the even and odd
components of the buckling modes will be coupled together; such general
conditions were not considered in [71].
(iii) Boundary support conditions must be associated with (1.109),
with Φ = Φ0 , at r = b and r = a in order to have a well-defined eigen-
value problem. The analysis presented in [71] is valid only for the case
of clamped supports at the inner and outer edges of the unsupported
region of the plate; therefore, it is noted in [71] that the buckling behav-
ior which we delineate, below, will change if different boundary support
conditions are used or if we employ the prebuckling stress distributions
that are associated with a finite annulus.
(iv) For the analysis reported in [71], a finite-difference scheme was
deemed to be the most appropriate method of solution to solve the
eigenvalue problem and determine the buckling mode shapes; because
of the symmetry assumptions with respect to the loads, considerations
could be limited to the first quadrant of the annular region b ≤ r ≤ a.
As noted in [71], using a more sophisticated scheme such as a finite-
element approach could, conceivably, allow for consideration of different
loading and boundary support conditions along the edges of the plate,
but any such work would be complicated by the expected appearance of
high stress concentrations and singular behavior if the annular plate is
highly anisotropic.
© 2001 by Chapm
of Young’s moduli and variations in the shear modulus, as well as in the
effect of using nonuniform applied loadings.
In Figs. 4.55 and 4.56, we displace contour plots of the lateral dis-
placement associated with the buckled mode shapes for Young’s moduli
ratios of E2 /E1 = 1 and E2 /E1 = 100, respectively, an aspect ratio of
a/b = 2.0, and (assumed) symmetry about the x and y axes; the mesh
size used was 10 × 45. Of course, Fig. 4.55 displays the results for an
isotropic annular plate and the wave number obtained, i.e., n = 10, cor-
responds to what is expected from the exact solution; wrinkling, in the
tangential direction, occurs at 18-degree intervals and we note that these
E2
wrinkles become wider as n increases. For = 1, the isotropic annular
E1
plate, 20 wrinkles occur over the entire plate; for E2 /E1 = 10 this drops
to 10 wrinkles in the buckled mode shape over the entire plate, and the
wrinkles are no longer oriented in the radial direction as in the isotropic
case. As the plate tends towards anisotropy, the wrinkles occurring in
the buckled mode shapes become more prominent as E2 increases in
relation to E1 . For E2 /E1 = 100 (Figure 4.56) the number of fully de-
veloped wrinkles in the buckled mode shape drops down to six. With
E2 /E1 = 1, 10, and 100 the buckled mode shapes were again generated,
but this time for an aspect ratio of a/b = 10.0 in place of a/b = 2.0;
the results are displayed in Figs. 4.57, 4.58, and 4.59, respectively. For
E2 /E1 = 1 (isotropic annular plate) the wave number, in this situation,
is n = 4. As the ratio of Young’s moduli E2 /E1 is increased, first to 10
and then to 100, the wrinkle which was oriented in the isotropic plate
π
along the line θ = becomes narrower and its orientation moves closer
4
to the y axis. On the other hand, the wrinkle centered on the x-axis
in the buckled mode shape, for the isotropic plate, broadens out as the
ratio E2 /E1 is increased. The cases illustrated in Figs. 4.55 through
4.59 illustrate buckling modes which are symmetric with respect to the
x-axis. In Fig. 4.60 we depict a mode shape which is anti-symmetric
with respect to the y-axis; in this case, the calculations produce wrin-
kles which are somewhat narrower and more closely oriented toward the
y-axis. One may summarize the results reported above by noting that,
for a rectilinearly orthotropic elastic annular plate, as the ratio E2 /E1
increases, wrinkles in the buckled mode shapes located near the y-axis
tend to narrow, while those located near the x-axis tend to broaden and
the orientation of the wrinkles shifts from the radial direction to the
π
direction of the ray θ = .
2
The second problem considered in [71], with respect to the initial
buckling behavior of an elastic rectilinearly orthotropic plate, concerns
the effect on the buckling mode shapes of varying the shear modulus;
some of the results obtained in [71] are depicted in Figs. 4.61, 4.62,
and 4.63; in all these figures, the shear modulus ratio is 100, while the
aspect ratio a/b varies over 2, 5, and 10. For each of those cases in
which the shear modulus G12 is much lower than Young’s moduli, the
π
plate buckles along the rays θ = ± because, along these directions, the
4
stiffness of the plate is a minimum and the compressive tangential and
radial stresses combine so as to buckle the annular plate.
The effect of nonuniform applied tractions on the buckled mode shape
of the rectilinearly orthotropic annular plate is depicted in Fig. 4.64
for the case in which a/b = 2.0, E2 /E1 = 10.0, E1 /G12 = 2.6, and
ν21 = 0.3. The applied loads in this case assume the form (4.79) and
the ratio p1 /p2 goes from 0.0 to 0.8; it is easy to see from Fig. 4.64
that changes in the ratio p1 /p2 significantly influence the buckled mode
shapes, with increases in p1 /p2 leading to wrinkles in the buckled mode
shape which shift from being centered around the x-axis, and oriented
in the y-direction, to being centered around the y-axis and oriented in
the y-direction. Note that at the ratio of p1 /p2 = 0.712 the wrinkles
are distributed around the entire circumference of the annular plate.
Because of the anisotropy of the plate, those wrinkles which are close
to the y-axis are relatively narrow, while those which are close to the
x-axis are relatively broad.
Figures 4.65 through 4.67 depict the results of calculations carried
out in [71] to determine the buckled mode shapes for a rectilinearly
orthotropic elastic annular plate with orthotropic in-plane properties and
isotropic bending properties; in all three cases shown, a/b = 2.0, ν21 =
0.3, and the assumption of isotropic bending means that D2 /D1 = 1.0.
In Fig. 4.65, the orthotropy is due to the large ratio E2 /E1 = 103 of
Young’s moduli, and the plate is subjected to a uniform radial traction,
while the same conditions apply in the case depicted in Fig. 4.66 except
that the plate is subjected to a uniform radial displacement. In the case
shown in Fig. 4.67 we again employ a uniform radial traction but now
the orthotropic in-plane behavior is a consequence of the large shear
modulus ratio, i.e., E1 /G12 = 100.0. As can easily be seen in Fig. 4.65
π
the largest buckle, in this case, occurs along the ray θ = ; this may be
4
compared with the buckling that can be expected to occur around the
y-axis if the ratio of the bending properties were the same as the ratio of
the in-plane material properties. When a uniform radial displacement
is applied at the inner edge of the plate, at r = b, (Fig. 4.66) the largest
© 2001 by Chapm
π
buckle again appears near the ray θ = , with the wrinkle oriented
4
uniformly in the x-and y-directions because of the equality of the two
bending stiffnesses. Finally, in Fig. 4.67, we see that when the in-plane
orthotropy results from the large ratio of E1 /G12 = 100, and the plate
is subjected to a uniform radial traction at r = b, the most pronounced
π
buckle still develops along the ray at θ = because the combination
4
of compressive radial and tangential stresses easily buckles the annular
plate in the vicinity of this direction.
Finally, the issue of the way in which the critical buckling loads vary
for the various cases studied has also been addressed in [71]. In Table 4.4,
we show the changes which occur in the predicted values of the buckling
loads as the ratio of Young’s moduli varies; the results clearly indicate
that the critical buckling load increases with increasing stiffness of the
annular plate in the y-direction, i.e., with E1 (and, thus, D1 fixed ),
that pi increases as E2 increases, and that the critical buckling load also
increases with increasing stiffness of the annular plate in the x-direction,
i.e., with E2 (and, thus, D2 fixed), pi increases as E1 increases. In Table
4.5, we record the effect on the critical buckling load of the annular
plate as the shear stiffness G12 decreases: when the overall stiffness (as
measured by E2 /G12 ) increases, the critical buckling loads increase very
rapidly, while for E2 (and hence, D2 ) are fixed, with the shear stiffness
G12 decreasing, the critical buckling load exhibits a slow decrease.
2
2π a
∂ 2 w 1 ∂w
1 1 ∂2w
VB
e
= +K + 2 2
∂r2 2 r ∂r r ∂θ
0 b
∂ 2 w 1 ∂w 1 ∂2w
−K(1 − ν) 2 + 2 2
∂r r ∂r r ∂θ
2
1 ∂2w 1 ∂w
+K(1 − ν) − rdrdθ
r ∂r∂θ r2 ∂θ (4.83)
π 2 1
= Kc2 2 + n2 − 1 ln
2 γ
2
n
+n (1 − γ) (1 + γ)
2
+ 1 − ν − 2(n − 1)
2
2
π
≡ Kc2 F (n, γ)
2
where γ = b/a and F (n, γ) has the obvious definition which is implied
by (4.83). If the annular plate is subjected to a radial tensile stress
σrr |r=b = λ along its inner edge, then the prebuckling elastic stress
distribution in the plate is given by
2
λb2 a
σrr = a2 − b2 r2 − 1
0
2 (4.84)
λb2 a
σθθ = − 2
0
+1
a − b2 r 2
which is analogous to (but not identical with) the prebuckling stress
distribution in an isotropic elastic annular plate subjected to a uniform
© 2001 by Chapm
radial compressive force applied at the outer edge at r = a, i.e., (4.19
a,b). By virtue of (4.82) and (4.84) the work done by the membrane
forces in the plate is given by
2 2
1 2π a ∂w 1 ∂w
W =−
e 0
σrr h 0
+ σθθ h rdrdθ
2 0 b ∂r r ∂θ
π 2 λb2 1 2 1
= hc 2 3 a − b − a ln
2 2
2 a − b2 2 γ
1
+n2 b2 − a2 + b2 + a2 ln
γ (4.85)
π 2 2
λa γ 1 1
= hc2 3 1 − γ 2 − ln
2 1 − γ2 2 γ
$
+n2 γ 2 − 1 + γ 2 + 1
π 2 2
≡ hc λa G(n, γ)
2
where G(n, γ) has the obvious definition which is implied by (4.85). By
equating VB e
, as given by (4.83), with W e , as given by (4.85), we obtain
the critical condition for buckling in the elastic domain of the annular
plate in the form
a2 h F (n, γ)
λcr = (4.86)
K G(n, γ)
or, as K − Eh3 /12(1 − ν 2 ),
2
1 h F (n, γ)
λcr = · (4.87)
12(1 − ν 2 ) a G(n, γ)
For ν = 0.3, some numerical results for the problem delineated above are
b
depicted in Fig. 4.68, as well as in Fig. 4.69, where ξ = 1 − ≡ 1 − γ.
a
When the annular plate has a large outer radius a, but is relatively
a−b
thin, i.e. is small, Yu and Johnson [179] note that initial buckling
a
will usually be governed by the elastic model described above but, for
a−b
≈ 1, the stress distribution in the plate which is associated with
a
λcr , as given by (4.87), will usually cause yielding before buckling; in
this case
2a2 2
(σrr − σθθ )|r=b = λcr = λcr =Y (4.88)
a −b
2 2 1 − γ2
or
1 − γ2 1
λcr = Y ≡ ξ(2 − ξ)Y (4.89)
2 2
where Y is the yield-stress associated with the elastic-plastic material
constituting the plate; plastic yield will first occur at the inner edge of
the plate at r = b. Equation (4.89) imposes a limitation with respect to
the applicability of (4.87) for determining the critical buckling load of
the elastic-plastic annular plate; using (4.87), in conjunction with (4.89),
this limitation can be expressed in the form
a F (n, γ) E 1
≥ · (4.90)
h G(n, γ) Y G(1 − ν 2 )(1 − γ 2 )
© 2001 by Chapm
where the buckling modulus E0 is given by
4EEt
E0 = √ √ (4.96)
( E + Et )2
© 2001 by Chapm
when the edge
√ at r = a is clamped but the edge at r = b is simply
supported, k 7.2 when the √ edge at r = a is simply supported while
the edge at r = b is clamped, k √ 5.4 when the edges at both r = a
and r = b are simply supported, k 3.4 when the edge at r = a is
clamped while the edge at r = b is free, and so forth.
The second basic situation associated with the initial buckling of an
isotropic annular plate is the one considered first by Mansfield [165],
i.e., the buckling of an infinite annular plate of hole size b, which is
supported by a member of sectional area ah/(1 + ν) along the circle of
radius r = a, and which is subjected to either tension or compression
along the edge at r = b. If we refer to Figs. 4.4 and 4.5, we see that,
with compressive loading along the edge at r = b, the plate will always
buckle into the axially symmetric mode, while with tensile loading along
the edge at r = b the plate buckles initially into modes with several
diametric nodes. From Fig. 4.4, it is clear that (for all the combinations
of boundary support conditions indicated) increases in γ = b/a (i.e.,
decreases in 1/γ = a/b) lead to decreases in the normalized buckling
parameter k(1 − γ)2 /γ; however, as γ = b/a increases, f (γ) = (1 − γ)2 /γ
decreases; and, thus, it is difficult to gauge from Fig. 4.4 the effect on
a2 h
the buckling parameter k = λcr of increasing the ratio b/a. What is
K
clear from Fig. 4.4 is that the critical buckling edge stress λcr increases,
for any fixed γ, 0 ≤ γ < 1, as one further restricts the joint lateral
movement of the plate along the edge at r = b and along the circle of
radius edge at r = a. In Fig. 4.5 we show some results associated with
applying an (inward) tensile stress along the edge at r = b ; for both the
case in which the plate is clamped at r = a and r = b, and the case in
which the plate is simply supported at r = a and r = b, the parameter
(1 − γ)2
k again decreases as γ = b/a increases. Now, however, the
γ
annular plate buckles into modes possessing several diametrical nodes.
At any fixed value of γ = a/b more diametrical nodes appear in the
initial buckling modes when the plate is clamped at both r = a and
r = b as opposed to the case where the plate is simply supported at
r = a and r = b ; in both cases, the number of diametrical nodes in an
initial buckling mode increases as the aspect ratio γ = a/b of the plate
increases. By comparing Figs. 4.4 and 4.5 it is easy to see that, for any
given aspect ratio γ, the critical buckling loads for the case when the
annular plate is subjected to a tensile load along the edge at r = b are
much higher than the critical buckling loads which are indicated for the
case in which the plate is subjected to a compressive loading along the
edge at r = b; this is true both for the clamped boundary conditions at
r = a, r = b and for the simply supported boundary conditions at r = a
and r = b. For both types of loading along the edge at r = b, and any
fixed γ, the clamped boundary conditions at r = a and r = b lead to
higher values of λcr than the simply supported boundary conditions at
r = a and r = b.
The last major subcase associated with the initial buckling of an
isotropic linearly elastic annular plate is the one in which the plate is
subjected to a uniform radial compressive loading only along the edge
at r = a, with the outer edge clamped and the edge of the hole at
r = b free; this is the case first considered by Majumdar [166] and, later,
by Troger and Steindl [66] and Machinek and Troger [167]. The initial
buckling curves are depicted in Fig. 4.7, which is (roughly) equivalent
to Fig. 4.10. Because results appear to exist in the literature for only
one set of boundary support conditions, we note only that as the aspect
ratio γ = b/a is increased, the critical buckling edge stress λcr increases
and the number of diametrical nodes in the initial buckling mode also
increases. Up to an aspect ratio of approximate 0.5, the buckling mode
is axially symmetric.
Figures 4.12 through 4.17 provide the basis for making a comparative
study of the initial buckling behavior of cylindrically orthotropic annular
elastic plates as one varies the material symmetry assumptions (i.e., the
degree of orthotropy) as well as the boundary support conditions. If we
recall that β = Eθ /Er , then the case of an isotropic elastic annular plate
corresponds to β = 1.0 so that the magnitude of β gauges the deviation
from isotropy. In Figs. 4.12 through 4.17 the buckling parameter
µ = −λcr ha2 /Dr ,
Dr = Er h3 /12(1 − νr νθ )
is graphed against the plate aspect ratio b/a for various combinations of
boundary support conditions along the edges of the plate at r = a and
r = b ; the plate is assumed to be subjected to an in-plane compressive
loading along one or both edges. For all the combinations of boundary
support conditions considered, it is easy to see that for any fixed value of
γ = b/a, the critical buckling edge stress λcr increases as the degree, β,
of orthotropy increases. For the first three cases considered (both edges
clamped and loaded, both edges simply supported and loaded, outer edge
fixed and loaded and inner edge free), λcr increases monotonically with
γ = b/a for any degree of orthotropy β. For the fourth and sixth case
(outer edge simply supported and loaded, inner edge free and outer edge
© 2001 by Chapm
simply supported, inner edge free, with both edges loaded) λcr decreases
monotonically with increasing γ = b/a for any degree of orthotropy.
Finally, in the fifth case depicted, where the outer edge is fixed, the
inner edge is free, and both edges are loaded, it is only in the isotropic
case that λcr increases monotonically with γ = b/a; in all the cases
(β > 1.0), λcr first decreases and then increases as γ = b/a increases
with the initial rate of decrease of λcr higher the larger the degree, β, of
orthotropy. If one fixes any particular value of β in Figs. 4.12 through
4.17 and looks at the values of µ (i.e. λcr ) for a given value of the aspect
ratio, then it is easy to isolate the effect that varying the boundary
support conditions along the edges at r = a, r = b has on the initial
buckling of the elastic annular plate; for example, with β = 5.0 and
γ = 0.6 the (approximate) values of µ are
so that we once again see the trend toward higher (initial) buckling loads
as the (joint) degree of lateral movement allowed at the edges of the plate
is increasingly restricted.
Because of the noted mismatch between the circular geometry of an-
nular elastic plates and the other deviation from isotropic material sym-
metry we have discussed, namely, rectilinear orthotropy, it is not pos-
sible to directly mesh this latter variation from isotropic response into
the comparative study delineated above; this is due, analytically, to our
inability to compare, in a simple way, the measure β̄ = E2 /E1 of devia-
tion from isotropy in the rectilinearly orthotropic case with the measure
β = Er /Eθ of deviation from isotropy in the cylindrically orthotropic
case. Also, for an elastic (circular) annular plate exhibiting rectilinear
orthotropic material symmetry, the available data base, even for the
problem of initial buckling, appears to be limited to the results in [71]
; these results, in turn, are limited to the type of loading and support
conditions first studied for isotropic material response by Mansfield [165]
and are, in essence, delineated in Tables 4.4 and 4.5 (critical loads versus
the ratios of Young’s moduli and shear moduli) as well as in Figures 4.55
through 4.67 (variations in buckled mode shapes).
We now want to look at the effect of varying the nature of the mate-
rial symmetry assumption (i.e. degree of orthotropy β) and the type of
boundary support conditions imposed along the edges, on the postbuck-
ling behavior of elastic annular plates. Because there appear to be no
results available for (circular) elastic annular plates with rectilinear or-
thotropic material symmetry, our discussion will be confined to those
postbuckling results which are available for cylindrically orthotropic
plates; all such results appear to be restricted to axially symmetric de-
formations of the plate. In [70] the outer edge at r = a is clamped and
subjected to a uniform, in-plane, compressive loading, while the inner
edge at r = b is free; from Fig. 4.23 it is easily seen that with Λ = λ/λcr ,
(the normalized postbuckling load), the maximum deflection not only in-
creases with Λ for each fixed β = Er /Eθ and fixed value of γ = b/a,
but for each fixed value of the aspect ratio γ = b/a and each fixed
postbuckling load λ > λcr , the maximum deflection Wmax increases as
the variance from isotropy increases. The very best comparative results
which are available in connection with the postbuckling behavior of elas-
tic annular plates may be those which follow from the work of Huang
[175]; in this case, the inner edge at r = b is free, the outer edge at r = a
is subjected to a uniform radial compressive loading, and that edge may
be either clamped or simply supported. Thus, in this case, it is possible
to gauge the effect on the postbuckling behavior of the plate of varying
both the boundary support conditions and the degree of deviation from
isotropic material response. Taking the measure of plate deflection to be
|w(γ)/h|, γ = b/a ≡ 0.4, Table 4.1 shows the effect on the postbuckling
deflection of varying the postbuckling load λ > λcr for the case of a
simply supported edge at r = a, for β = 1, 3, 5, and 10, and for the
case of a clamped edge at r = a for β = 5. For β = Er /Eθ = 5, we
see that achieving particular values of |w(γ)/h| in the case where the
edge at r = a is clamped requires a postbuckling load about three times
larger than that which is needed to produce the same deflection when
the edge at r = a is simply supported. Also, in addition to the obvious
conclusion that for both types of boundary support conditions at r = a,
and any fixed β, most buckling deflections increase monotonically with
increasing λ > λcr , we also note (at least for the case in which the outer
edge of the plate at r = a is simply supported) that achieving any par-
ticular value of |w(γ)/h| requires an increasing postbuckling edge stress
© 2001 by Chapm
λ > λcr as the degree, β, of orthotropy increases. For β = 5.0, Fig.
4.26 depicts the effect of varying the boundary support conditions at
r = a on the postbuckling behavior; a considerably higher postbuckling
load is needed to obtain a given postbuckling deflection for the case of a
clamped edge as opposed to a simply supported edge. In Fig. 4.27, the
normalized postbuckling loads λ/λcr are graphed against |w(γ)/h| for
both types of boundary support condition at r = a and the same varia-
tions in β as depicted in Table 4.1; here we note that for a fixed value
of |w(γ)/h|, in the simply supported case, higher values of β correspond
to lower values of the normalized load λ/λcr . Finally, for β = 5.0, Fig.
4.32 displays the postbuckled (axially symmetric) displacement shapes
for the two types of boundary support conditions along the outer edge
at r = a and variations in the postbuckling loads λ/λcr (which are indi-
cated on these graphs as variations in the postbuckling deflections w(γ));
other comparative results concerning the postbuckling behavior of cylin-
drically orthotropic (circular) annular plates may be read directly off the
graphs in Figs. 4.28 through 4.31, which depict, for β = 5.0 and γ = 0.4,
in both the clamped and simply supported cases at r = a, the distribu-
tions of radial bending stresses, circumferential bending stresses, radial
membrane stresses, and circumferential membrane stresses and their re-
spective variations with the postbuckling edge stress λ > λcr (as gauged
by variations in δ = w(γ)/h).
As we have already noted, very little (if any) serious work appears
to have been done on the initial or postbuckling behavior of nonlinear
elastic, viscoelastic, or elastic-plastic annular plates; a good treatment
within the domain of elastic-plastic response is that of Yu and John-
son [179], but this work treats only initial buckling and does not allow
for a comparative study based on either variations in loading condi-
tions/boundary support conditions at r = a and r = b, or on variations
in the degree of deviation from isotropic material symmetry response
within the elastic domain of the plate. Variations in both initial and
postbuckling response for (circular) annular plates, as a consequence of
variations in constitutive response, may constitute a good area for future
research efforts. For additional work on the buckling and postbuckling
behavior of (circular) annular plates, one may consult [182]–[190]; none
of these references concern buckling in the presence of imperfections,
which will be dealt with in the next chapter.
4.4 Initial Buckling/Postbuckling Figures, Graphs,
and
Tables: Annular Plates
†
When β = 1, the material of the annulus is isotropic.
© 2001 by Chapm
Table 4.2 Convergence of the Eigenvalue for Increasing Mesh
Size. (Based on [71])
© 2001 by Chapm
Table 4.4 Comparison of Critical Loads Versus Ratio of
Young’s Moduli. (Based on [71])
© 2001 by Chapm
FIGURE 4.1
Buckling of annular plates. (Adopted, in modified form, from
[162].)
FIGURE 4.2
Buckling of annular plates. (Adopted, in modified form, from
[162].)
© 2001 by Chapm
FIGURE 4.3
Annular plate stiffened along outer edge [165]. (From Mans-
field, E.H., “On the Buckling of an Annular Plate,” Quart. J.
Mech. Appl. Math. XIII, 16, 1960. Reprinted by permission
of Oxford University Press.)
FIGURE 4.4
Buckling of annular plate with stiffener along outer edge [165].
(From Mansfield, E.H., “On the Buckling of an Annular Plate,”
Quart. J. Mech. Appl. Math. XIII, 16, 1960. Reprinted by
permission of Oxford University Press.)
© 2001 by Chapm
FIGURE 4.5
Buckling of annular plates under inward tensile stress [165].
(From Mansfield, E.H., “On the Buckling of an Annular Plate”,
Quart. J. Mech. Appl. Math. XIII, 16, 1960. Reprinted by
permission of Oxford University Press.)
FIGURE 4.6
Comparison between exact and approximate buckling loads.
(Adopted, in modified form, from [166].)
© 2001 by Chapm
FIGURE 4.7
Buckling load for various wave numbers. (Adopted, in modified
form, from [166].)
FIGURE 4.8
Eigenvalue curves showing for n = 0 the influence of m on the critical
load N . The physically relevant value is m = 1. (Machinek, A. and
Troger, H., “Postbuckling of Elastic Annular Plates,” Dynamics and
Stability of Systems, 3, 1988, 79–88. With permission of the Taylor
& Francis Group plc.)
© 2001 by Chapm
FIGURE 4.9
Eigenfunctions for b = 0.3; the physically relevant eigenfunction is g01 .
(Machinek, A. and Troger, H., “Postbuckling of Elastic Annular Plates,”
Dynamics and Stability of Systems, 3, 1988, 79–88. With permission
of the Taylor & Francis Group plc.)
FIGURE 4.10
Stability boundary in parameter space (N, b) for m = 1 and n =
0, 1, . . . , 10. (Machinek, A. and Troger, H., “Postbuckling of Elastic An-
nular Plates,” Dynamics and Stability of Systems, 3, 1988, 79–88.
With permission of the Taylor & Francis Group plc.)
© 2001 by Chapm
FIGURE 4.11
Axonometrical representation of the deflection surface of the annular
plate for m = 1 and n = 0, 1, 2. (Machinek, A. and Troger, H., “Post-
buckling of Elastic Annular Plates,” Dynamics and Stability of Sys-
tems, 3, 1988, 79–88. With permission of the Taylor & Francis Group
plc.)
FIGURE 4.12
Critical buckling loads: buckling parameter µ = −λcr ha2 /Dr
versus b/a, both edges fixed and loaded. (Adopted, in modified
form, from [169].)
© 2001 by Chapm
FIGURE 4.13
Critical buckling loads: buckling parameter µ = −λha2 /Dr
versus b/a with edges simply supported and loaded. (Adopted,
in modified form, from [169].)
FIGURE 4.14
Critical buckling loads: buckling parameter µ = −λha2 /Dr
versus b/a, outer edge fixed and loaded, inner edge free.
(Adopted, in modified form, from [169].)
© 2001 by Chapm
FIGURE 4.15
Critical buckling loads: buckling parameter µ = −λha2 /Dr
versus b/a, outer edge simply supported and loaded, inner edge
free. (Adopted, in modified form, from [169].)
FIGURE 4.16
Critical buckling loads: buckling parameter µ = −λha2 /Dr
versus b/a, outer edge fixed, inner edge free; both edges loaded.
(Adopted, in modified form, from [169].)
© 2001 by Chapm
FIGURE 4.17
Critical buckling loads: buckling parameter µ = −λha2 /Dr
versus b/a, outer edge simply supported, inner edge free, both
edges loaded. (Adopted, in modified form, from [169].)
FIGURE 4.18
Nondimensional deflection (w) versus nondimensional radius
(ρ) [70]. (From Uthgenannt, E. and Brand, R., J. Appl. Mech.,
40, 559, 1973. With permission.)
© 2001 by Chapm
FIGURE 4.19
Nondimensional membrane radial stress (S) versus nondimen-
sional radius (ρ) [70]. (From Uthgenannt, E. and Brand, R.,
J. Appl. Mech., 40, 559, 1973. With permission.)
FIGURE 4.20
Nondimensional membrane tangential stress (SθM ) versus
nondimensional radius (ρ) [70]. (From Uthgenannt, E. and
Brand, R., J. Appl. Mech., 40, 559, 1973. With permission.)
© 2001 by Chapm
FIGURE 4.21
B
Nondimensional bending radial stress (SR ) versus nondimen-
sional radius (ρ)[70]. (From Uthgenannt, E. and Brand, R., J.
Appl. Mech., 40, 559, 1973. With permission.)
FIGURE 4.22
Nondimensional bending tangential stress (SθB ) versus nondi-
mensional radius (ρ) [70]. (From Uthgenannt, E. and Brand,
R., J. Appl. Mech., 40, 559, 1973. With permission.)
© 2001 by Chapm
FIGURE 4.23
Nondimensional maximum deflection (w) versus Annular plate
parameter (b/a) [70]. (From Uthgenannt, E. and Brand, R.,
J. Appl. Mech., 40, 559, 1973. With permission.)
FIGURE 4.24
Nondimensional maximum membrane tangential stress (SθM )
versus annular plate parameter (b/a) [70]. (From Uthgenannt,
E. and Brand, R., J. Appl. Mech., 40, 559, 1973. With per-
mission.)
© 2001 by Chapm
FIGURE 4.25
B
Nondimensional maximum bending radial stress (SR ) versus
annular plate parameter (b/a) [70]. (From Uthgenannt, E. and
Brand, R., J. Appl. Mech., 40, 559, 1973. With permission.)
FIGURE 4.26
Postbuckling loads and transverse displacement parameter for
an orthotropic annulus [174]. (Reprinted from the Int. J.
Nonlinear Mech., 10, Huang, C.L., On Postbuckling of Or-
thotropic Annular Plates, 63-74, 1974, with permission from
Elsevier Science)
© 2001 by Chapm
FIGURE 4.27
Ratio of postbuckling load to linear buckling load and trans-
verse displacement parameter [174]. (Reprinted from the Int.
J. Nonlinear Mech., 10, Huang, C.L., On Postbuckling of Or-
thotropic Annular Plates, 63-74, 1974, with permission from
Elsevier Science)
FIGURE 4.28
The distribution of radial bending stresses [174]. (Reprinted
from the Int. J. Nonlinear Mech., 10, Huang, C.L., On Post-
buckling of Orthotropic Annular Plates, 63-74, 1974, with per-
mission from Elsevier Science)
© 2001 by Chapm
FIGURE 4.29
The distribution of circumferential bending stresses [174].
(Reprinted from the Int. J. Nonlinear Mech., 10, Huang, C.L.,
On Postbuckling of Orthotropic Annular Plates, 63-74, 1974,
with permission from Elsevier Science)
FIGURE 4.30
The distribution of radial membrane stresses [174]. (Reprinted
from the Int. J. Nonlinear Mech., 10, Huang, C.L., On Post-
buckling of Orthotropic Annular Plates, 63-74, 1974, with per-
mission from Elsevier Science)
© 2001 by Chapm
FIGURE 4.31
The distribution of circumferential membrane stresses [174].
(Reprinted from the Int. J. Nonlinear Mech., 10, Huang, C.L.,
On Postbuckling of Orthotropic Annular Plates, 63-74, 1974,
with permission from Elsevier Science)
FIGURE 4.32
Transverse displacement shape of the buckled annular plate
[174]. (Reprinted from the Int. J. Nonlinear Mech., 10,
Huang, C.L., On Postbuckling of Orthotropic Annular Plates,
63-74, 1974, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 4.33
Deflection response for a uniformly distributed load [173].
(Reprinted from the Int. J. Nonlinear Mech., 19, Dumir, P.C.,
Nath, Y., and M. Gandhi, Non-Linear Axisymmetric Static
Analysis of Orthotropic Thin Annular Plates, 255-272, 1984,
with permission from Elsevier Science)
FIGURE 4.34
Deflection response for a plugged hole under a uniformly dis-
tributed load [173]. (Reprinted from the Int. J. Nonlinear
Mech., 19, Dumir, P.C., Nath, Y., and M. Gandhi, Non-Linear
Axisymmetric Static Analysis of Orthotropic Thin Annular
Plates, 255-272, 1984, with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 4.35
Branches of the bifurcated solutions for an annular plate b < bc
(a) no imperfections present (γ0 = γ1 = 0); (b) 0 < γ1 <
0.36, γ0 = 0, smooth transition from x1 mode to x0 mode; (c)
0.36 < γ1 < 0.51, γ0 = 0, jump transition from x1 Mode to x0
mode: (d) transition point γ1 = 0.51 and (e) γ1 > 0.51, γ0 = 0,
plate stays in x1 mode [66].
FIGURE 4.36
Sectional view of one quadrant of a partially restrained infinite
annular plate. (Original figure from the second author’s Ph.d.
Thesis [71].)
© 2001 by Chapm
FIGURE 4.37
Radial normal stress distribution in an annular plate with or-
thotropy due to variation in Young’s Moduli. (Original figure
from the second author’s Ph.d. Thesis [71].)
FIGURE 4.38
Tangential normal stress distribution in an annular plate with
orthotropy due to variation in Young’s moduli. (Original figure
from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.39
Polar shear stress distribution in an annular plate with or-
thotropy due to variation in Young’s moduli. (Original figure
from the second author’s Ph.d. Thesis [71].)
FIGURE 4.40
Effect of the ratio of Young’s moduli on the tangential normal
stress distribution at θ = 0. (Original figure from the second
author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.41
Effect of the ratio of Young’s moduli on the tangential normal
stress distribution at θ = π/2. (Original figure from the second
author’s Ph.d. Thesis [71].)
FIGURE 4.42
Effect of the ratio of Young’s moduli on the tangential stress
distribution along the inner edge of the annular plate. (Origi-
nal figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.43
Effect of the ratio of Young’s moduli on the y component of the
stress distribution at x = b. (Original figure from the second
author’s Ph.d. Thesis [71].)
FIGURE 4.44
Radial normal stress distribution in an annular plate with or-
thotropy due to variation in the shear modulus ratio. (Original
figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.45
Tangential normal stress distribution in an annular plate with
orthotropy due to variation in the shear modulus ratio. (Orig-
inal figure from the second author’s Ph.d. Thesis [71].)
FIGURE 4.46
Polar shear stress ddstribution in an annular plate with or-
thotropy due to variation in the shear modulus ratio. (Original
figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.47
Effect of shear modulus ratio on the radial normal stress distri-
bution at θ = π/4. (Original figure from the second author’s
Ph.d. Thesis [71].)
FIGURE 4.48
Relative displacement shape of an annular plate with low shear
modulus subjected to uniform load (b ≤ r ≤ 2b). (Original
figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.49
Ratio of Young’s moduli versus ratio of traction parameters
yielding uniform radial displacement. (Original figure from
the second author’s Ph.d. Thesis [71].)
FIGURE 4.50
Tangential normal Stress distribution in an orthotropic annular
plate with the applied traction acting in direction of maximum
stiffness. (Original figure from the second author’s Ph.d. The-
sis [71].)
© 2001 by Chapm
FIGURE 4.51
Radial normal stress distribution in an orthotropic annular
plate with the applied traction acting in the direction of maxi-
mum stiffness. (Original figure from the second author’s Ph.d.
Thesis [71].)
FIGURE 4.52
Polar shear stress distribution in an orthotropic annular plate
with the applied traction acting in the direction of maximum
stiffness. (Original figure from the second author’s Ph.d. The-
sis [71].)
© 2001 by Chapm
FIGURE 4.53
Effect of variation in the ratio of traction parameters on the
tangential normal stress distribution along the inner edge of
the annular plate. (Original figure from the second author’s
Ph.d. Thesis [71].)
FIGURE 4.54
Convergence of the finite-difference solution of buckled mode
shape in the radial direction for a plate aspect ratio of 2. (Orig-
inal figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.55
Buckled mode shape for an isotropic plate with plate aspect
ratio of 2. (Original figure from the second author’s Ph.d.
Thesis [71].)
FIGURE 4.56
Buckled mode shape for an orthotropic plate with ratio of
Young’s moduli equal to 100 and a plate aspect ratio of 2.
(Original figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.57
Buckled mode shape for an isotropic plate with a plate aspect
ratio of 10. (Original figure from the second author’s Ph.d.
Thesis [71].)
FIGURE 4.58
Buckled mode shape for an orthotropic plate with ratio of
Young’s moduli equal to 10 and a plate aspect ratio of 10.
(Original figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.59
Buckled mode shape for an orthotropic plate with ratio of
Young’s moduli equal to 100 and a plate aspect ratio of 101.
(Original figure from the second author’s Ph.d. Thesis [71].)
FIGURE 4.60
Buckled mode shape for an orthotropic plate with ratio of
Young’s moduli equal to 10 and a plate sspect ratio of 5 (sym-
metric about x axis and antisymmetric about y axis ). (Origi-
nal figure from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.61
Buckled mode shape for an orthotropic plate with a shear mod-
ulus ratio of 100 and a plate aspect ratio of 2. (Original figure
from the second author’s Ph.d. Thesis [71].)
FIGURE 4.62
Buckled mode shape for an orthotropic plate with a shear mod-
ulus ratio of 100 and a plate aspect ratio of 10. (Original figure
from the second author’s Ph.d. Thesis [71].)
© 2001 by Chapm
FIGURE 4.63
Buckled mode shape for an orthotropic plate with a shear mod-
ulus ratio of 100 and a plate aspect ratio of five (antisymmetric
about the x and y axes). (Original figure from the second au-
thor’s Ph.d. Thesis [71].)
FIGURE 4.64
Effect of nonuniform applied tractions on the buckled mode
shape. (Original figure from the second author’s Ph.d. Thesis
[71].)
© 2001 by Chapm
FIGURE 4.65
Buckled mode shape for a plate with orthotropic in-plane prop-
erties and isotropic bending properties and subjected to a uni-
form radial traction. Orthotropy due to ratio of Young’s mod-
uli. (Original figure from the second author’s Ph.d. Thesis
[71].‘)
FIGURE 4.66
Buckled mode shape for a plate with orthotropic in-plane prop-
erties and isotropic bending properties and subected to a uni-
form radial displacement. Orthotropy due to ratio of Young’s
moduli. (Original figure from the second author’s Ph.d. Thesis
[71].)
© 2001 by Chapm
FIGURE 4.67
Buckled mode shape for a plate with orthotropic in-plane prop-
erties and isotropic bending properties and subjected to a uni-
form radial traction. Orthotropy due to the shear modulus
ratio. (Original figure from the second author’s Ph.d. Thesis
[71].)
FIGURE 4.68
The Critical load for the elastic buckling of a flange [179].
(Reprinted from Int. J. Mech. Sci., 24, Yu, T.X. and John-
son, W., The Buckling of Annular Plates in Relation to the
Deep-Drawing Process, 175-188, 1972, with permission from
Elsevier Science)
© 2001 by Chapm
FIGURE 4.69
The number of waves for the elastic buckling of a flange [179].
(Reprinted from Int. J. Mech. Sci., 24, Yu, T.X. and Johnson,
W., The Buckling of Annular Plates in Relation to the Deep-
Drawing Process, 175-188, 1972, with permission from Elsevier
Science)
FIGURE 4.70
The critical condition for the plastic buckling of a flange when
n ≤ 11 [179]. (Reprinted from Int. J. Mech. Sci., 24, Yu, T.X.
and Johnson, W., The Buckling of Annular Plates in Relation
to the Deep-Drawing Process, 175-188, 1972, with permission
from Elsevier Science)
© 2001 by Chapm
FIGURE 4.71
The critical condition for the plastic buckling of a flange when
n ≥ 10 [179]. (Reprinted from Int. J. Mech. Sci., 24, Yu, T.X.
and Johnson, W., The Buckling of Annular Plates in Relation
to the Deep-Drawing Process, 175-188, 1972, with permission
from Elsevier Science)
FIGURE 4.72
The number of waves for the plastic buckling of a flange [179].
(Reprinted from Int. J. Mech. Sci., 24, Yu, T.X. and Johnson,
W., The Buckling of Annular Plates in Relation to the Deep-
Drawing Process, 175-188, 1972, with permission from Elsevier
Science)
© 2001 by Chapm
Chapter 5
with
K1 K2 K3
λc = , λ1 = , λ2 = (5.5)
L L L
© 2001 by Chapm
so that
λ1 /λc = K2 /K1 (5.6)
and these solution paths in the λ − ξ plane are depicted in Figs. 5.4a
and 5.4b for both positive and negative values of the initial deviation
¯ the intersecting equilibrium paths for the perfect structure ξ¯ = 0 ,
ξ:
which are shown as dotted curves, have been deformed into disjointed
branches by the presence of the initial imperfection ξ¯ = 0 and, with
respect to a loading that increases from λ = 0, those branches that do
not pass through the origin have become irrelevant. For various values
¯ the significant branches which do contain the origin make up the
of ξ,
two' families
' of curves depicted in Figs. 5.5a and 5.5b; in these graphs,
as 'ξ¯' converges to zero, the members of each family approach different
pairs of branches of the equilibrium paths which are associated with the
¯ paths from
perfect structure. We note that, irrespective of the sign of ξ,
the origin in the λ − ξ plane never approach the fundamental path above
λ = λc .
With a sufficiently small positive value of ξ¯ the corresponding equi-
librium path (Fig. 5.5a) possesses a local maximum with respect to
the load at the point (ξs , λs ), with λs < λc , and with the rotation ξs
depending on ξ. ¯ Now, suppose that the actual loading is increased be-
yond the load λs ; for λ > λs static equilibrium becomes impossible at
values of ξ in a neighborhood of ξs : a dynamic process must ensue,
and the final static configuration (if one exists) cannot be arbitrarily
close to ξs no matter how slightly λs is exceeded. In accord with this
observation, λs is designated as the buckling load of the imperfect struc-
ture; this kind of buckling, which is associated with a local maximum
on a static equilibrium path in the λ − ξ plane, and an expected jump
to a non-neighboring equilibrium configuration when the local maxi-
mum is exceeded, is known as snap-buckling (or, sometimes, limit-point
buckling). On the other hand, the graph depicted in Fig. 5.5b for the
case ξ¯ < 0, suggests a somewhat less drastic behavior: for loads in a
neighborhood of λc , significant increases occur in the rate of the growth
of rotation with load; however, λc still warrants being designated the
buckling load of the structure and the buckling cannot, in any case, be
compared with the shape and possible catastrophic snap buckling which
occurs at λ = λs , when ξ¯ > 0.
While Figs. 5.5a and 5.5b depict the situation at hand when λ1 /λc ≡
K2 /K1 < 0, other combinations of the parameters in (5.2) are also of
interest. For λ1 > 0, small values of ξ¯ < 0 may lead to snap-buckling
(i.e., Fig. 5.6a), while for λ1 = 0, where we have a symmetric bifurca-
tion for the perfect structure, snap-buckling may be induced by initial
rotations with either ξ¯ > 0 or ξ¯ < 0 if λ2 < 0 (i.e. Fig. 5.6b); the
case λ2 > 0 yields only mild buckling (i.e. Fig. 5.6c). It is, therefore,
the postbuckling behavior of the perfect structure, with ξ¯ = 0, and,
in particular, the issue of whether the load increases or decreases after
initial buckling, which determines what kind of buckling behavior one
may expect for the imperfect structure with ξ¯ = 0. We now turn to a
discussion of imperfection sensitivity for the simple structure pictured
in Fig. 5.3.
When snap-buckling occurs, the magnitude of λs (see Fig. 5.5a) may
be well below that of λc . If we maximize λ(ξ) in (5.7), with respect to
ξ, then we find that (λ (ξs ) = 0 and λ(ξs ) = λs )
¯ = f (ξs )
λs L cos(ξs + ξ) (5.8)
© 2001 by Chapm
perfect structure with ξ¯ = 0; structures in which this phenomena can
occur are said to be imperfection sensitive. For the model depicted in
Fig. 5.3, imperfection sensitivity is implied by the conditions λ1 = 0 or
λ1 = 0, with λ2 < 0, and the extent of the imperfection sensitivity in
these cases then depends on the magnitudes of λ1 /λc or λ2 /λc .
Remarks: In lieu of prescribing the load λ which acts on the imper-
fect structure in Fig. 5.3, one could impose a vertical displacement (or
shortening) ∆ at the point of the application of the load (see Fig. 5.1)
of the form
∆/L = cos(ξ + ξ)¯ − cos ξ¯ (5.11)
1
For the limiting case of the perfect model, i.e. ξ¯ = 0, ∆/L ≈ ξ 2 and
2
then, by virtue of (5.4),
1
λ/λc ≈ 1 + (λ1 /λc ) (2∆/L) 2 , (ξ > 0)
1
λ/λc ≈ 1 − (λ1 /λc ) (2∆/L) 2 , (ξ < 0) (5.12)
λ1 = 0
λ/λc ≈ 1 + 2 (λ2 /λc ) (∆/L) ,
λ2 = 0
where w(x, y) = w̃(x, y) + αw0 (x, y) yields the final shape of the plate;
in rectilinear coordinates, the [ , ] is given by (1.51a) and the form of
(5.13) results from assuming that the thin plate is both linearly elastic
and isotropic.
The influence of small deviations w0 (x, y), from initial flatness, on
the out-of-plane deflections of simply supported isotropic elastic square
plates which are subjected to unidirectional edge compression, has been
investigated by Hu, Lundquist, and Batdorf [109], while other types
of boundary support conditions along the edges have been studied by
Yamaki [94] and Coan [192]; only in a neighborhood of the critical elastic
buckling stress do the resulting lateral deflections differ substantially
from those of an ideal flat plate during the loading process. Thus, very
large postbuckling deflections appear not to be greatly affected by small
initial irregularities. For the square plate which is simply supported
along all four edges, the postbuckling curves of Hu, Lundquist, and
Batdorf [109] are depicted in Fig. 5.8, where δ0 is the initial deflection
at the center of the plate while δ̃ is the additional deflection at the
center of the plate; the type of imperfection buckling displayed in this
figure is similar to that pictured in Fig. 5.7d for the case ξ¯ > 0. In [94]
Yamaki computed the values of δ̃/h for other types of boundary support
conditions along the edges of a square, linearly elastic isotropic thin
plate with an initial deviation (imperfection) given by δ0 /h = 0.1; these
results are depicted in Figs. 5.9 and 5.10 along with the corresponding
dotted curves for the perfect plate in which there is no initial deflection.
It is also worthwhile to comment, briefly, on the type of influence on the
buckling of a rectangular, isotropic linearly elastic plate, which is exerted
by the presence of a distributed load t(x, y) normal to the (initially) flat
configuration for the plate. As noted in [32], “the critical stress of a
rectangular plate increases when normal loads of sufficient magnitude
are applied ... this increase is due to the effect of the additional tensile
membrane stresses caused by large deflections of the plate under normal
loads and is related to the magnitude of deflection that can be allowed.”
© 2001 by Chapm
An analysis of the initial buckling of isotropic rectangular elastic plates
subjected not only to edge compression, but also to a distributed normal
load (with associated pressure function p), and having an aspect ratio
a/b = 4, was carried out by Levy, Goldenburg, and Zibritosky [193]; the
plate was assumed to be simply supported and the lateral deflection w
was approximated by
πx πy 3πx πy
w(x, y) = A11 sin sin + A31 sin sin
a b a b
(5.14)
5πx πy 7πx πy
+A51 sin sin + A71 sin sin
a b a b
pb4
The analysis in [193] leads to four simultaneous equations relating ,
Eh2
2
h , and the coefficients A11 , A31 , A51 , and A71 , in (5.14), which are
shown to have more than one real solution; thus, the plate may be in
equilibrium for a given combination of lateral and end loading in more
than one buckle pattern, and this phenomenon is depicted in Fig. 5.11,
which shows the relationship between the parameter λb2 /Eh2 , (which
is based on the edge stress λ = σxx |x=0,a ) and the parameter $xx b2 /h2 ,
which is based on the axial strain for the constant normal pressure given
Eh4
by p = 24.03 4 . We note that the shift in the buckle pattern is as-
b
sociated with a drop in the axial load and that the plate can be in
equilibrium with either one, three, or seven buckles in the initial buck-
ling mode. In Fig. 5.12 we show the development of longitudinal buckles
in the initial buckling mode, thus indicating that, for low values of λ,
the resulting deflection is a long shallow bulge. When high values of λ
prevail, one obtains the regular initial buckling pattern again and any
initial deflection of the plate has all but disappeared. For all the con-
ditions investigated by Levy, Goldenburg, and Zibritosky [193], it was
noted that the presence of a distributed normal load leads to an increase
in the critical buckling load of the plate. When the edges of the rect-
angular plate are clamped instead of simply supported, the existence of
a distributed normal load leads to results similar to those obtained in
[193]; this follows from the analysis in Woolley, Corrick, and Levy [120]
in which the same gradual transition from a long bulge to a regular ini-
tial buckling pattern was observed as in the case of a simply supported
plate. One notable difference between the cases of a simply supported
and a clamped rectangular plate subject to distributed normal loading
as well as edge loads, is that the increase in the critical buckling load
is much less in the latter case; this result is illustrated in Fig. 5.13,
where the critical buckling strain parameter $cr b2 /h2 is compared with
the parameter pb4 /Eh4 for the two different types of boundary support
conditions.
We now turn to a description of the postbuckling behavior of general
linear elastic anisotropic rectangular plates possessing imperfections in
the form of an initial prebuckling displacement w0 (x, y). The analysis,
which is based on the work of Romeo and Frulla [114], includes the case
of rectilinear orthotropic symmetry as a particular subcase of the general
anisotropic response considered, and applies to several types of boundary
support conditions for plates subjected to combined biaxial compression
and shear loads. As indicated in [114], plates possessing initial imperfec-
tions w0 have been studied on the basis of the Marguerre approximate
nonlinear theory by Chia [33]; the resultant strain-displacement relations
are, essentially, (1.34) with W replaced by w0 , i.e.,
1 2
$xx = u,x + w,x + w0,x w,x − zw,xx
2
1 2
$yy = v,y + w,y + w0,y w,y − zw,yy (5.15)
2
γxy = u,y + v,x + w,x w,y + w0,y w,x
+w0,x w,y − 2zw,xy
while the equilibrium equations consist of (1.42) and the analog of (1.45)
with W replaced by w0 , i.e.
+2Nxy w0,xy = 0
Introducing the Airy function as in (1.46), the first two equilibrium equa-
tions (1.42) are, once again, satisfied identically, while (5.16) becomes
© 2001 by Chapm
$0xx,yy + $0yy,xx − $0xy,xy = w,xy
2
− w,xx w,yy
(5.18)
+2w0,xy w,xy − w0,yy w,xx − w0,xx w,yy
in which [A] and [D] are, respectively, the extensional and bending
stiffness matrices. Employing (5.19) and (5.20) in (5.17) and (5.18) and
introducing the normalizations
x y z
ξ= , η = , ζ = , φ = a/b (5.21a)
a b h
−1 ∗
[A∗ ] = [A] , Ā = A22 [A∗ ] (5.21c)
1
∗
[D ∗ ] = [D] , D̄ = [D ∗ ] (5.21d)
h2 A22
∗ ∗
∗ ∗
2
D̄11 W,ξξξξ + 4D̄16 φW,ξξξη + 2 D̄12 + 2D̄66 φ W,ξξηη
∗ 3 ∗ 4
+4D̄26 φ W,ξηηη + D̄22 φ W,ηηηη
(5.22)
−φ2 (F,ηη W,ξξ + F,ξξ W,ηη − 2Fξη W,ξη
N x b2 N y b2 Nxy b2
λξ = , λ η = , λ ξη = (5.24)
h2 A22 h2 A22 h2 A22
ξ = 0, 1 : Φ,ηη = λξ , Φ,ξη = −φλξη , W = 0, Mξ = 0
(5.25)
η = 0, 1 : Φ,ξξ = φ2 λη , Φ,ξη = −φλξη , W = 0, Mη = 0
ξ = 0, 1 : Φ,ηη = λξ, Φ,ξη = −φλξη, W = 0, W,ξ = 0
(5.26)
η = 0, 1 : Φ,ξξ = φ2 λη, Φ,ξη = −φλξη, W = 0, Mη = 0
© 2001 by Chapm
To satisfy the boundary conditions, the displacement and Airy func-
tions are chosen in the respective forms
#
i #
j
W = Cpq ω(ξ, η) (5.29)
p=1 q=1
and
η2 ξ 2 φ2
Φ= λξ + λη − φξηλξη
2 2
#m # n (5.30)
+ Fnk Xh (ξ)Yk (η)
h=1 k=1
and
Yk (η) = cosh(µk η) − cos(µk η)
(5.31b)
−αk (sinh(µk η) − sin(µk η))
In Fig. 5.14, the results obtained in [114] are displayed and compared
with the results obtained by Yamaki [94] for an isotropic square plate
under uniaxial compression, both with an initial imperfection of the
form (5.34) with T = 0.1, and without the imperfection under various
constraint conditions: I is a plate with all edges simply supported, II
is the plate with the sides clamped and the ends simply supported, for
III the sides are simply supported and the ends are clamped, while IV
represents a plate clamped along all edges. The results obtained in [114]
are very close to those of Yamaki [94]. In Fig. 5.15, results are shown for
the clamped/simply supported isotropic plate under axial compression,
for plate aspect ratios of φ = 1 and φ = 3, and these are compared with
those of Sheinman, et.al. [194] for T = 0.1 with m = n = i = j = 4 or 6 in
(5.29), (5.30); it may be noted that, as the plate aspect ratio increases,
the initial imperfection causes a postbuckling behavior which is very
different from the theoretical behavior without the imperfections. In Fig.
5.16, results are shown for an anisotropic square plate under uniaxial
compression which has its edges clamped and its sides simply supported;
these results are compared with those of Minguet, et. al. [195]. Finally,
in Fig. 5.17, we show a comparison of analytical and experimental results
(based on the work in [114]) for a rectangular anisotropic plate under
uniaxial compression. The plate has clamped edges, simply supported
sides, and an initial imperfection of the form (5.34) with T = 0.05; the
results are compared with those of Engelstad, et. al. [196].
(b) Plates Exhibiting Elastic-Plastic Response
In §3 of Chapter 2 we discussed the initial and postbuckling be-
havior of square elastic-plastic plates subjected to axial compression;
that discussion was based, to a large extent, on the work presented by
Needleman and Tvergaard [104] and concerned a plate material which
is strain-hardening and characterized by a flow theory of plasticity with
a smooth yield surface. The constitutive theory employed in [104] is
given by (2.221)–(2.225). Two types of in-plane boundary conditions
were considered in 2.3: one which requires all four edges of the plate
to remain straight throughout the loading history and one which leaves
the edges unconstrained; in the elastic range, these in-plane boundary
© 2001 by Chapm
conditions play a major role in determining the initial postbuckling be-
havior of the plate but the work in [104] also investigates the effects
of the in-plane boundary conditions on the postbuckling behavior and
imperfection sensitivity in the plastic range. For buckling of the per-
fect plate in the plastic range, the authors [104] employ Hutchinson’s
postbuckling theory [17] to obtain an asymptotically exact description
of the initial postbuckling behavior, i.e., (2.229); they also note that for
a plate with a small initial curvature compressed into the plastic range,
there is no available general theory which relates the behavior of the im-
perfect plate to the postbuckling behavior of the corresponding perfect
plate as in, e.g., Koiter’s theory of elastic stability [5]. Thus, in [104],
an approximate analysis based on the approach used by Hutchinson and
Budiansky [197], which neglects elastic unloading, is employed in order
to assess the imperfection sensitivity of simply supported square plates
when buckling of the perfect plate occurs in the plastic range.
Following the discussion in 2.3, we confine our attention to imperfec-
tions in the shape of the initial buckling mode (2.227) with the amplitude
of the imperfection denoted by ξ. ¯ For λ < λc we have noted, in Chapter
2, that the lowest order effect of the imperfection on the magnitude of
the lateral deflection ξ is given by (2.230) where Ψ ≥ 1 and p(λ) is finite
at λ = λc . For λ close to λc , a singular perturbation analysis produces
(2.231) and matching (2.230) and (2.231) produces
1
c = −λc (p(λc )) Ψ , γ = 2Ψ + 1 (5.35)
When λ2 < 0, (2.231) produces a maximum support stress λmax which
satisfies (2.232) with µ given by (2.233). For λc < σy , λ2 reduces, as
noted in [104], to the appropriate value for a linear elastic plate.
Figures 2.51, 2.52 and 2.53, which we have already referenced in Chap-
ter 2, for the buckling behavior of the perfect (i.e., ξ¯ = 0) elastic-plastic
square plate, are based on numerical results obtained in [104]. Figures
2.51 and 2.52 display results which correspond to buckling of the square
plate occurring in the plastic range; in both cases, h/a = 0.035 and
σy /E = 0.0037. In Fig. 2.51, the hardening parameter n in (2.225) is
10 while λc /σy = 1.196. In Fig. 2.52, we have n = 3 and λc /σy = 1.259.
Although initial buckling occurs well into the plastic range of the mate-
rial, it turns out that the predictions of J2 -flow theory and those of the
simplest deformation theory, i.e. J2 -deformation theory, do not differ
by a large amount; for example, in Fig. 2.51, with n = 10, the dis-
crepancy noted in [104] is 5.8%, while in Fig. 2.52, with n = 3, the
discrepancy is just 2.3%. In Fig. 2.53 we have h/a = 0.031, ν = 0.3,
and σy /E = 0.00337 and initial buckling of the perfect plate occurs just
after the onset of plastic yield; as noted in [104], in this case there exists
virtually no discrepancy between the bifurcation predictions of J2 -flow
theory and those of J2 -deformation theory. In Fig. 2.53, n = 10 and
λc /σy = 1.015. As we have already noted in Chapter 2, for the perfect
plates the behavior in the immediate neighborhood of initial buckling is
described by (2.229) in that buckling takes place under increasing load
and the curvature of the load deflection curve is large and negative; also,
in the immediate neighborhood of the initial buckling point, the initial
postbuckling behavior is seen to be independent of the boundary condi-
tions. We now turn to the imperfection buckling behavior described in
figures 2.51, 2.52, and 2.53.
In Fig. 2.51, where the hardening parameter n = 10, the load falls
off rapidly after the maximum load point is reached. The postbuckling
behavior is nearly identical for both sets of in-plane boundary condi-
tions considered, as is the behavior of a plate possessing a small initial
imperfection given by ξ¯ = 0.01. However, as the amplitude of the initial
imperfection is increased, the effect of the in-plane boundary conditions
is more pronounced. For the largest imperfection considered in Fig. 2.51,
i.e. ξ¯ = 1.0, the plate subjected to the constrained boundary conditions
has a maximum support load which is about 25% higher than the one
with the unconstrained in-plane boundary condition.
In Fig. 2.52, the postbuckling behavior of the perfect plate depends
very strongly on the in-plane boundary conditions. Under the uncon-
strained in-plane boundary conditions, the load reaches a maximum and
then decreases monotonically, whereas with the constrained boundary
condition, the load continues to increase; in this case the mode ampli-
tude ξmax , which is obtained from (2.229), is much smaller than the
numerical value obtained for the plate subjected to the unconstrained
boundary condition. The load-deflection behavior for a plate with an
initial imperfection also depends, in this case, on the in-plane boundary
conditions. An imperfect plate subjected to the constrained in-plane
boundary condition is capable of supporting loads in excess of the buck-
ling load, while plates with the unconstrained boundary condition clearly
exhibit a mild imperfection sensitivity.
In Fig. 2.53, in which n = 10, the load on the perfect plate is seen
to reach a maximum shortly after buckling and then decrease monoton-
ically. When compared with the example considered in Fig. 2.51, in
which initial buckling of the perfect plate occurs further into the plastic
range, the effect of the in-plane boundary conditions shows up much ear-
lier in Fig. 2.53. Even with an initial imperfection amplitude of ξ¯ = 0.01,
the effect of the in-plane boundary conditions on the load-deflection be-
havior of the plate is apparent. We also note, that in Fig. 2.53, for each
initial imperfection amplitude ξ¯ considered, the imperfection sensitivity
is less than that for the corresponding example in Fig. 2.51 and that the
effect of the in-plane boundary conditions on the load-deflection behav-
ior of the plate increases with the amplitude of the initial imperfection.
Overall, the results described above for square elastic-plastic plates ex-
hibit a far smaller degree of imperfection sensitivity than the comparable
results for circular elastic-plastic plates, which are presented in the next
section.
¯ − 0.1r̄ − 0.9r̄2 ), r̄ = r
w0 (r) = ξ(1 (5.36)
a
while for the case of a clamped plate
1
w0 (r) = ξ¯ cos πr̄ (5.37)
2
In the analysis presented in [157] the initial deflection parameter ξ¯
was varied over the values 0, 0.1, 0.5, and 1.0. For the analyses of the
imperfection buckling of an annular plate, the inner edge of the plate
was assumed to be free; at the center of a circular plate one imposes, at
r = 0, the usual conditions, i.e., u(0) = w (0) = 0 where u is the plate
displacement in the radial direction. The axially symmetric large defor-
mation states for cylindrically orthotropic circular and annular plates
possessing an initial deflection w0 (r) are governed by that straightfor-
ward modification of (1.101), (1.102), (1.103) which is analogous, e.g.,
in the case of the first von Karman equation, in rectilinear Cartesian
coordinates, to retaining W = w0 in (1.45); the relevant constitutive
relations are given by the obvious modifications of (1.104 a,b) where,
e.g., σrr , err , etc., must be replaced by σrr − σrr
0
, err − e0rr , with σrr , err
the stress and strain components in the present state of the plate, while
0
σrr , e0rr correspond to the state given by the initial deflection w0 . The
plates in the examples considered in [157] are buckled by controlling
the radial shortening applied uniformly at the outer circumference at
r = a; the flexural conditions assumed to exist along this boundary are
those which correspond to either the simply supported or clamped edge
conditions.
In the work presented in [157] the authors employ a finite difference
discretization of the governing equations and first consider the computa-
tion of the initial buckling loads and subsequent postbuckling behavior
of flat isotropic circular and annular plates; the results are summarized
in Table 5.1 and are compared with earlier results of Nadai [186] and
Meissner [172], where the first column in the table refers to the number
of mesh divisions in the finite difference scheme. A comparison of the
accuracy of two different mesh divisions is shown in Fig. 5.19, the results
being those in the postbuckling regime (i.e. load-radial shortening re-
sponse) for initially flat, clamped, and simply supported circular plates.
To check that the initial deflections had been correctly incorporated into
the program, the computations were checked against Massonnet’s [198]
© 2001 by Chapm
results for imperfect clamped isotropic circular plates; in this case, the
edge compression versus additional plate center deflection comparison
is shown in Fig. 5.20 and there appears to be good agreement between
the two sets of results over the range of initial deflection amplitudes
considered. To check that the cylindrical orthotropy of the plates had
been appropriately incorporated into the program, yet another set of
computations were performed for the postbuckling response of flat or-
thotropic annular plates (with aspect ratio φ = 0.4) having both simply
supported and clamped outer edges; the benchmark in this case was
the work of Huang [174] and the edge compression versus free edge de-
flection comparison is shown in Fig. 5.21. In Figs. 5.22 and 5.23 we
reproduce the results obtained by Turvey and Drinali [157] relative to
the postbuckling response of imperfect polar orthotropic annular plates
having an aspect ratio of φ = 0.4. The postbuckling response was com-
puted for three orthotropy ratios and four imperfection amplitudes, i.e.,
1
ξ¯ = 0, 0.1, 0.5, and 1.0 and β = 3, 1, and and the edge compression
3
versus additional deflection (at r = 0.4a) responses are shown in Figs.
5.22 and 5.23 for clamped and simply supported edges, respectively. The
general postbuckled response for the clamped plates, as depicted in Figs.
5.22a through 5.22c does not appear to depend on the orthotropy pa-
rameter and, as the edge compression becomes large, the magnitude of
the additional deflection depends less and less on the initial deflection
amplitude, i.e., the postbuckling response of the imperfect plate tends
to that of the flat plate with the trend being more pronounced as the
orthotropy parameter increases. However, the edge compression versus
edge deflection response of the simply supported plates, as noted on Figs.
5.23a through 5.23c is quite different from those of clamped plates: in
the simply supported case the additional imperfect plate deflection does
not converge to that of the flat plate deflection; indeed, the additional
imperfect plate deflection becomes less than the flat plate value once
5 3
the edge compression exceeds a value between and times the criti-
4 2
cal buckling load value for the flat plate. Additionally, if one compares
the clamped and simply supported postbuckling plate responses shown
in Figs. 5.22 and 5.23, it is observed that (for edge compression ratios
greater than unity) the additional deflection is generally smaller for the
simply supported plate than it is for the clamped plate. It may, there-
fore, be deduced that the presence of the initial deflection has a greater
transverse stiffening effect, in the postbuckling regime, on simply sup-
ported plates as opposed to clamped plates. Finally, it has been noted
in [157] that none of the computed flat or imperfect plate responses ex-
hibited any buckling mode change tendencies as either the imperfection
amplitude and/or the edge compression increased, i.e., the initial ax-
isymmetric single half-wave buckling mode was maintained throughout.
(b) Plates Exhibiting Elastic-Plastic Response
In Chapter 3 we considered the initial buckling and postbuckling be-
havior of (perfect) thin elastic-plastic circular plates, basing our analysis
on the work of Needleman [141] which is, a priori , restricted to axisym-
metric deformations. As noted in [141], however, the postbuckling be-
havior of elastic-plastic circular plates is characterized by a high degree
of imperfection sensitivity, which is the issue that we will be concerned
with in this subsection. We begin by reviewing, briefly, the mathemat-
ical formulation of the buckling problem for an elastic-plastic circular
plate as presented in §2 of Chapter 3.
As in Chapter 3, the plate is assumed to have radius a and thickness
h and is subjected to an in-plane radial compressive loading. The strain
rates ėrr , ėθθ are given by (3.140) with u(r) and w(r), respectively, the
radial in-plane and lateral displacements. Stress rates σ̇ are related to
the strain rates ė by σ̇ = Lė where L, the elastic part of the tensor of
moduli L, is given by (3.141). A J2 flow theory with isotropic hardening
is employed so that the uniaxial stress-strain behavior is given by (3.145)
with n the strain-hardening exponent and σy the yield stress. The full
tensor L is then represented by (3.142) with σe , sij , and c given, respec-
tively, by (3.143) and (3.144). The tangent modulus is a function of σe
and represents the slope of the uniaxial stress-strain curve as depicted in
Fig. 3.25. As noted in Chapter 3, each surface in the plate which is par-
allel to the middle surface is in an (approximate) state of plane stress,
implying that only the in-plane stresses need to be admitted into the
constitutive theory; this observation leads, in turn, to the introduction
of the tensor K of in-plane moduli which is defined as in (3.146). For
a simply supported plate, the relevant conditions at r = a are given by
(3.147) while for a clamped plate (3.148) holds. The equilibrium equa-
tions are given by the variational principle δI = 0, where the functional
I is given by (3.149) and all those incremental fields u̇(r) and ẇ(r) which
satisfy the kinematic boundary condition at r = a are admissible.
The critical buckling stress in the plastic range is given by (3.150)
where E/Ef and νf are defined, respectively, by (3.151), (3.152); for
a clamped plate β̃ is the smallest nonzero root of J1 (β̃) = 0 while, for
a simply supported plate, it is the smallest non-zero root of β̃J0 (β̃) −
(1 − νf )c J1 (β̃) = 0, the c subscript denoting evaluation of the effective
Poisson ratio νf at buckling. For both types of boundary support con-
ditions, the associated buckling mode wc (r) has the form (3.153). The
critical buckling stress σc for both the clamped and simply supported
cases is depicted in Fig. 3.25 as a function of the parameter η̃, which is
defined by (3.154), where β̃E is the value of β̃ when buckling occurs in
the elastic range.
Employing the amplitude ξ of the relevant eigenmode as the expansion
variable in a postbuckling perturbation analysis, the average radial stress
applied at the edge of the plate at buckling is given by (3.155), where
σc is the critical buckling stress as given by (3.150) and σav is defined
by (3.156). The initial slope σ1 in the expansion (3.155) is uniquely
determined by the condition that plastic loading takes place throughout
the plate except at one point where neutral loading occurs; this leads
to the relation (3.157) between σ1 and σc . As a consequence of (3.157),
σ1 > 0. Carrying the expansion for the average applied compressive
stress one term further leads to (3.158), where σ2 is given by (3.159),
(3.160) and (3.161a,b). The first three terms that are depicted in the
expansion (3.158) for σav may be used to estimate the maximum applied
stress σmax that can be supported by the plate as well as the amplitude
ξ of the buckling mode at σmax ; these results are given by (3.162a,b).
As ξ increases from zero, the neutral loading surface expands out from
1
the point r = 0, z = h and has the form (3.163).
2
In Chapter 3 we have referenced Figs. 3.26 thorough 3.29 and pre-
sented a fairly thorough discussion of the postbuckling behavior for the
perfect elastic-plastic circular plate for each of the respective cases con-
sidered in these graphs; the graphs in question depict the (postbuckling)
curves of average applied stress vs. buckling deflection for two different
cases: the (postbuckling) curves governing buckling deflection vs. ap-
plied displacement for a simply supported plate, and the (postbuckling)
curves depicting average applied stress vs. buckling deflection for a
clamped plate. The results displayed in Figs. 3.26 through 3.29 are
based on a finite-element analysis of the variational equation δI = 0
where I is given by (3.149); in this analysis the author [141] allows for
an initial imperfection w0 (r) which is proportional to the (initial) buck-
ling mode (3.153) and the amplitude is then specified by ξˆ = w0 (0). The
results for perfect plates, which have been discussed in detail in Chap-
ter 3, actually result from a calculation in which the small imperfection
amplitude ξˆ = 10−4 h has been employed.
For σc << σy the elastic postbuckling analysis shows that the initial
postbuckling behavior is stable in the sense that applied stresses higher
than the buckling stress σc can be supported by the plate; in this range,
the plate is imperfection insensitive and these observations have already
been made in Chapter 3. However, for σc > σy the maximum support
load of the circular plate is not much larger than σc and the analysis
by Hutchinson [45], [46] of imperfection sensitivity for buckling in the
plastic range indicates that, in general, one may expect imperfection
sensitivity to be prevalent when σc > σy .
(i) For initial buckling occurring in the elastic range of the perfect plate,
imperfections can lead to a substantial reduction in the load-carrying
capacity of the plate if σc σy .
(ii) For initial buckling occurring in the plastic range of the perfect
plate, σmax is only slightly higher than σc ; furthermore, an asymptotic
analysis of the postbuckling behavior of the perfect plate in the plastic
range, which is based on the work of Hutchinson [45], [46], yields a good
estimate of σmax as well as of the development of the unloading region
after buckling.
© 2001 by Chapm
Reiss [201] note, however, Morozov [207] did not prove that secondary
buckling actually occurs. Yanowitch [209], on the other hand, did show
that buckling of a circular elastic plate can take place from an initially
deformed state of the plate by proving that the radially symmetric pri-
mary buckled state is unstable for λ sufficiently large.
In [201] the following (dimensionless) form of the von Karman theory
for clamped circular plates is used:
2
∆ w = [Φ, w]
(0 ≤ r < 1, 0 ≤ θ < 2π) (5.38)
∆2 Φ = − 1 [w, w]
2
w(1, θ) = w,r (1, θ) = 0
(0 ≤ θ < 2π) (5.39)
Φ(1, θ) = 0, Φ,r (1, θ) = −λ
where r is the dimensionless radial coordinate, Φ(r, θ) the dimensionless
Airy function, w(r, θ) the dimensionless lateral deflection of the plate, ∆
the Laplacian in polar coordinates, and the nonlinear (bracket) operator
[Φ, w] is given by (1.163). A dimensionless edge thrust λ may be defined
by
1
λ = c2 (λ/E)(R/h)2 , c2 = 12(1 − ν 2 ) 2 (5.40)
with R and h, respectively, the radius and thickness of the plate, and
E and ν Young’s modulus and Poisson ratio. Dropping the ' on λ, the
unbuckled state is given by
1
w ≡ 0, Φ ≡ Φ̄ = λ(1 − r2 ) (5.41)
2
and is a solution of(5.38), (5.39), for all values of λ, which corresponds
to a state of uniform compressive stress. Substituting
Φ = Φ̄ + Φ̃, w = w̃ (5.42)
into (5.38), (5.39), and subsequently dropping the ˜ on Φ and w, we
obtain the reduced forms, i.e.,
2
∆ w + λ∆w = [Φ, w]
(5.43)
1
∆2 Φ = − [w, w]
2
and (5.39), where Φ now denotes the extra Airy stress function. The
classical linearized buckling problem is, of course, obtained from (5.43),
(5.39) by deleting the nonlinear terms in (5.43). We have already noted,
in our work in Chapter 3, that the smallest eigenvalue of the linearized
problem for the clamped isotropic elastic circular plate, which corre-
sponds to the primary buckling load, is given by λ1 ≈ 14.7 and that λ1
is the square of the smallest root of the Bessel function J1 (z); also, the
eigenfunction corresponding to λ1 is radially symmetric, i.e., the primary
buckling mode is axisymmetric. In fact, two axisymmetric solutions of
(5.43), (5.39), which differ only in the sign of the deflection w, branch
at λ = λ1 , and either of them may be referred to as the primary buckled
state. The primary buckled state will be denoted by
{Φ0 (r; λ), w0 (r; λ)}
and is known to exist for all λ > λ1 .
In order to study the secondary buckling of an isotropic elastic circular
plate near a secondary buckling load λ(0) , the authors in [201] seek
unsymmetric solutions of (5.43), (5.39) of the form
∞
#
w(r, θ; λ) = w0 (r; λ) + $ w(m) (r, θ)$m (5.44)
m=0
∞
#
Φ(r, θ; λ) = Φ0 (r; λ) + $ ϕ(m) (r, θ)$m (5.45)
m=0
∞
#
λ($) = λ(0) + $ λ(m) $m (5.46)
m=0
where $ is a small parameter which is defined by
2
2
$ = (w − w0 )2 + (Φ − Φ0 ) rdrdθ (5.47)
Ω
and Ω denotes the unit circle. The choice of the perturbation parameter
is not unique, of course. To determine the coefficients w(m) , ϕ(m) , and
λ(m) in the expansions (5.44)–(5.46), the primary state is first expanded
about λ = λ(0) , i.e., about $ = 0. Thus, it is assumed in [201] that
∞
#
w0 (r; λ($)) = wom (r)$m
m=0 (5.48)
∞
#
Φ0 (r; λ($)) = Φom (r)$m
m=0
© 2001 by Chapm
The coefficients wom and Φom , m = 0, 1 · · · are determined using a
numerical evaluation of the primary state with {woo , Φoo } the primary
state evaluated at λ = λ(0) . The expansion coefficients w(m) , ϕ(m) , and
λ(m) are computed by substituting (5.44)–(5.46), and (5.48), into (5.43),
(5.39), and (5.47); this yields a sequence of linear boundary value prob-
lems with associated normalizing conditions, e.g., w(0) , ϕ(0) , and λ(0)
must satisfy the eigenvalue problem:
(0) (0)
L1 w , ϕ
≡ ∆2 w(0) + λ(0) ∆w(0)
$ $
+ w00 , ϕ(0) + Φ00 , w(0) = 0
(0) (0) $
L2 w , ϕ ≡ ∆2 ϕ(0) + w00 , w(0) = 0
(5.49)
(0) (0) (0) (0)
w = w,r = ϕ = ϕ,r = 0, at r = 1
w (0)2
+ ϕ (0)2
rdrdθ = 1
Ω
© 2001 by Chapm
5.2.2 Secondary Buckling for Rectangular Plates.
In [203] the Movchan-Liapunov theorem, which we describe in the
analysis to follow, is used to establish a sufficiency condition for the
stability of equilibrium configurations of a rectangular plate under edge
loads and displacements; this sufficiency criterion may, as we will see,
also be used to characterize the possible occurrence of secondary bifur-
cation from the first buckled state of the perfect plate. A postbuckling
analysis of the secondary solution can then be used to formulate an
explanation of the mode-jumping phenomena which is experimentally
observed for rectangular plates. The Movchan-Liapunov criterion is dy-
namical in nature and the relevant dynamical equations are those for
small motion near equilibrium; they are derived from the full set of
equations by an appropriate perturbation method.
Experimental studies indicate not only the presence, for this prob-
lem, of the primary buckling solutions but also that other secondary
solutions may exist as well. As the authors [203] point out, work by
Stein [210], et al., with rectangular plates under longitudinal compres-
sion, have shown that, as the load is slowly increased, in the postbuckled
domain, the plate appears to suddenly snap from the equilibrium con-
figuration which corresponds to the original lowest mode to one which
corresponds to a higher mode; this apparent mode-jumping has led sev-
eral investigators to speculate that transition paths, which are, perhaps,
unstable, connect the primary buckled solutions. The phenomena is il-
lustrated in Fig. 5.26 in which equilibrium configurations corresponding
to combinations of a load parameter λ and some characteristic measure
w∗ of the deflection are depicted. The λ axis represents, of course, the
zero deflection solution which is possible under all loads for the perfect
system; the primary buckled states Ci (λ) of the perfect system branch
from the trivial solution as the load levels exceed λi , which are the
eigenvalues of the classical linear buckling problem for the plate. The
states Ci (λ) exist for all loads λ ≥ λi but may (as we saw for the cir-
cular plate, i.e., Fig. 5.25) change their local stability characteristics by
branching again, thus yielding the secondary buckled states Ci,j (λ); or
snapping may occur, i.e., local extrema may appear, such as at point
A, in Fig. 5.26, without branching. The local behavior, at branching,
for these buckled states, can also be related to the imperfection con-
figurations of the plate which are represented in Fig. 5.26 by curves
such as C ∗ (λ). Several numerical and analytical studies have supported
the hypothesis that secondary instability on the first primary branch
corresponds to branching; in particular, the work of Bauer and Reiss
[205], in which the postbuckling behavior of rectangular plates was nu-
merically analyzed, yields an asymmetric solution for the problem under
consideration, which appears to branch from a primary solution.
To formulate the dynamical theory to which the Movchan-Liapunov
stability theorem may be applied, we let w(xα ) and Φ(xα ), α = 1, 2, de-
note, as usual, the transverse displacement (deflection) and Airy func-
tion corresponding to an equilibrium configuration of the plate; for a
“motion” configuration, close to this equilibrium configuration, the dis-
placement and stress functions will be given by
w̃(xα , t) = w(xα ) + $ŵ(xα , t)
(5.52)
Φ̃(xα , t) = Φ(xα ) + $Φ̂(xα , t)
The boundary conditions relative to w (or ŵ) are the ones associated
with either clamped edges, i.e., (1.130a,b) for a plate with aspect ra-
tio φ = a/b, or simply supported edges, i.e., (1.130a), (1.131), or some
combination of clamped and simply supported edges. The boundary
conditions for the in-plane equilibrium problem, which are connected
with the edge loading of the (rectangular) plate, are those associated
with either prescribed edge tractions or prescribed edge displacements.
We now denote by Nαβ , Eαβ the stress resultants and strain components
associated with the equilibrium configurations determined by (w, Φ) and
by nαβ , eαβ the same quantities associated with the perturbed configu-
rations determined by (ŵ, Φ̂). Also, we let Uα denote the displacement
components associated with an equilibrium configuration while the uα
are the corresponding quantities for a perturbed configuration. Then
© 2001 by Chapm
1
Eαβ = 2 (Uα,β + Uβ,α + w,α w,β )
(5.55)
1
eαβ = (uα,β + uβ,α + ŵ,α w,β + ŵ,β w,α )
2
and the constitutive relations for the linear isotropic plate may be ex-
pressed in the form
Nαβ = (1 − σ 2 )−1 [(1 − σ)Eαβ + σEγγ δαβ ]
(5.56)
nαβ = (1 − σ 2 )−1 [(1 − σ)eαβ + σeγγ δαβ ]
To derive a quasi-static principle which is sufficient for establishing
the stability of solutions to the equilibrium problem (5.53) with given
loads, we study the boundedness of solutions to the dynamical equation
(5.54) and introduce the two energy functionals E and E:
1 $
E= (∆w)2 − (1 − σ)[w, w]
2 R
+(1 − σ 2 )−1 {(1 − σ)Eαβ Eαβ + σEαα Eββ } dxdy (5.57)
=
− T̄α Uα ds
with R denoting the domain occupied by the rectangular plate, and the
T̄α being the components of the applied traction on ∂R. Also,
2
1 ∂ ŵ $
E= + (∆ŵ)2 − (1 − σ)[ŵ, ŵ]
2 R ∂t
+(1 − σ 2 )−1 [(1 − σ)eαβ eαβ + σeαα eββ ]
(5.58)
+ Nαβ ŵ,α ŵ,β dxdy
˙ ŵ)
ρ = (ŵ, ˙ + (ŵ, ∆2 ŵ) + (ŵ, ŵ) (5.59)
where
(f, g) = f (xα )g(xα )dxdy (5.60)
R
1 2
V (θ) = (∆ θ) − (1 − σ) [θ, θ] dxdy
2 R
+ (1 − σ 2 )−1 {(1 − σ)eαβ eαβ + σeαα eββ } dxdy (5.61)
R
+ Nαβ θ,α θ,β dxdy
R
© 2001 by Chapm
of the plate, it may be used to study either the stability of the trivial
solution w = 0 or of any buckled state. To pursue the study of secondary
bifurcation in [203], the authors now restrict the form of the boundary
conditions to those that give the classical initial buckling loads. Thus,
we set
T̄α = λTα0 , Ūα = λUα0 (5.64)
where the Ūα are the components of the applied displacements on ∂Ru
o o
and λ is the proportional loading parameter. In (5.64), Tα , Uα are fixed
boundary tractions and displacements corresponding to the prebuckled
o o
in-plane equilibrium state of the plate. If we set Φ̄ = λΦ , where Φ
is the Airy function associated with the prebuckled in-plane state, the
linearized equilibrium eigenvalue problem which follows from (5.53) is
governed by
o
∆2 w − λ[Φ , w] = 0 (5.65)
which, (together with the associated clamped/simply supported edge
conditions) has only the trivial solution w ≡ 0, except for the eigen-
values λ = λi at which w = cψi the ψ(xα ) being the associated eigen-
functions; the same result may be obtained from the energy criterion
≥
V (θ) 0, for with w = 0 one has (θ[ζ̂(θ), w]) = 0 and (θ, [Φ, θ]) =
o
−λ Nαβ θ,α θ,β dxdy. Thus, the functional V (θ) assumes the form
R
of the classical potential energy functional for the plate problem; we
note that V (θ) ceases to be positive definite at λ = λ1 , θ = cψ1 .
To investigate secondary branching from a primary buckled state, one
must compute the critical value λc of the load parameter λ, the form of
the primary buckling state, and the branching behavior of the solution
for loads close to the critical load. Thus, for λ ≥ λ1 we let (w, Φ)
represent the first buckled solution, set
o
Φ = λ 1 Φ + Φ∗ (5.66)
and introduce the linear operator
o
Lw = ∆2 w − λ1 [Φ , w] (5.67)
and set
w(x, y; λ) = µW (x, y; µ2 )
(5.71)
Φ(x, y; λ) = µ2 φ(w, y; µ2 )
with W (x, y; 0) = 0, φ(x, y; 0) = 0. The functions W and φ satisfy
LW = µ2 [φ, W ]
(5.72)
∆2 φ = − 1 [W, W ]
2
and o
να Nαβ (φ) = λ1 Tβ , on ∂RT
o (5.73)
Uα (φ, W ) = λ1 Uα , on ∂Ru
The in-plane boundary conditions involve fixed traction and displace-
o
ment vectors which are the same as those that hold for Φ = λ1 Φ at the
first branching. Problem (5.72), (5.73) for (W, φ) clearly admits a zero
eigenvalue and the energy functional takes the form
1 1
V (θ, µ2 ) = (θ, Lθ) − µ2 {(θ, [φ, θ]) + (θ, [ζ, W ])} (5.74)
2 2
where ζ = µ−1 ζ̂ satisfies
∆2 ζ + [W, θ] = 0, in R (5.75)
A criterion is now sought that determines the first positive value µ2c of
µ2 such that V (θ, µ2c ) = 0. Using the work of Rogers [211] it may be
established that the real eigenvalues of the Euler-Lagrange equations for
V have a variational characterization; to establish that characterization,
an orthogonality condition is imposed to eliminate the effects of the zero
eigenvalue, i.e., θ(xα ) is split as
θ = α1 ψ1 + α2 θ̄ (5.76)
© 2001 by Chapm
where
ψ1 , [φ, θ̄] + [ζ̄, W ] = 0 (5.77)
and the auxiliary function ζ̄ is defined below. In fact (5.77) can also be
written in the form
θ̄, [φ, ψ1 ] + [F1 , W ] = 0 (5.78)
where
∆2 F1 + [ψ1 , W ] = 0
(5.79)
∆2 ζ̄ + [θ̄, W ] = 0
but (ψ1 , [F1 , W ]) ≤ 0, while (ψ1 , [φ, ψ1 ]) < 0 for µ2 sufficiently small.
Therefore, the first term in V is positive, as the first bifurcating branch
of solutions is traversed; this means that the first critical value of µ2
comes from the second term in V , i.e., at the first positive µ2 = µ2c , V
becomes zero for θ = α2 θ̄ with θ̄ the eigenfunction corresponding to the
eigenvalue µ2c . The variational equation for the restricted function
λ − λc λ1 (µ2 − µ2c )
η= =
λc λc
and denote the first eigensolution for (5.82), (5.79) by (θ̄, ζ̄). Following
[203] we set
w̄(x, y; γ, ξ) = w(x, y; η) + ξθ(x, y; η, ξ) (5.83)
so that w̄(x, y; η, 0) = w(x, y; η), write θ as a perturbation power series
in the new scalar bifurcation parameter ξ, i.e.,
© 2001 by Chapm
to be observed. In [205], Bauer and Reiss proposed an energy mechanism
based on total potential energy for the phenomenon of mode-jumping; in
their scenario, there exist energy-crossing parameter values λc at which
one mode shape is preferred over another, i.e., small disturbances in
a mode shape at such values may cause the given state to jump to a
state of lower energy through an intermediate state. The authors [205]
speculated that there may exist secondary branches which connect the
bifurcation points on two primary branches with the bifurcation point
on the second branch at a lower load level than that on the first.
To illustrate the method employed in [203], the example of a rectan-
gular simply supported plate in longitudinal compression (with edges
constrained so that the normal, in-plane edge displacement is uniform,
and the edge shear force is zero) is treated; the plate occupies the region
0 ≤ x1 ≤ , 0 ≤ y ≤ 1. The vanishing of the second normal derivative of
the deflection on all four edges of the plate may be shown to be equiv-
alent to the physical requirement that the edge moment be zero. The
load parameter λ is just the negative of the resultant load on the edges
at x = 0, x = , i.e.
1 1
λ=− N11 (0, y)dy = − N11 (, y)dy (5.87)
0 0
while the mean load on the edges at y = 0 and y = 1 is zero. The
requirements that the normal edge displacements be constant while the
edge stresses vanish may be expressed by the conditions
u1,2 (0, y) = u1,2 (, y) = u2,1 (x, 0) = u2,1 (x, 1) = 0
(5.88)
u2,1 (0, y) = u2,1 (, y) = u1,2 (x, 0) = u1,2 (x, 1) = 0
© 2001 by Chapm
perturbations and pointing to the likelihood of snapping occurring at an
unstable bifurcation. The authors [203] also compute the snapping loads
for the imperfect systems and the energy-crossing loads for the first few
primary branches. In Table 5.2 we have summarized, following the work
in [203], the numerical values of the secondary buckling loads and the
values of the parameters ā, b̄, governing the (approximate) postbuckling
analysis, for the four types of θ̄ series; three different aspect ratios are
considered in this table.
5.3 Imperfection Buckling and Secondary Buckling
Figures, Graphs, and Tables
© 2001 by Chapm
Table 5.2 Numerical Values of the Secondary Buckling Loads
for a Rectangular Plate [203] (From Stroebel, G.J. and Warner,
W.H., Stability and Secondary Bifurcation for von Karman
Plates, J. Elast., 3, 185, 1973. Reprinted with kind permission
from Kluwer Academic Publishers.)
© 2001 by Chapm
FIGURE 5.2
Equilibrium paths. (a) λ1 < 0; (b) λ1 = 0, λ2 > 0; (c)
λ1 = 0, λ2 < 0 [16]. (From “Theory of Buckling and
Post·Buckling Behavior of Elastic Structures” by B. Budian-
sky in ADVANCES IN APPLIED MECHANICS, Volume 14,
copyright 1974
c by Academic Press, reproduced by permis-
sion of the publisher.)
FIGURE 5.3
Initially imperfect simple model [16]. (From “Theory of Buck-
ling and Post·Buckling Behavior of Elastic Structures” by B.
Budiansky in ADVANCES IN APPLIED MECHANICS, Vol-
ume 14, copyright 1974
c by Academic Press, reproduced by
permission of the publisher.)
© 2001 by Chapm
FIGURE 5.4
Equilibrium paths for imperfect model. (a) ξ̄ > 0; (b) ξ̄ < 0
[16]. (From “Theory of Buckling and Post·Buckling Behavior
of Elastic Structures” by B. Budiansky in ADVANCES IN AP-
PLIED MECHANICS, Volume 14, copyright 1974 c by Aca-
demic Press, reproduced by permission of the publisher.)
FIGURE 5.5
Equilibrium paths for various values of initial imperfection
ξ̄(λ1 < 0) (a) ξ̄ > 0; (b) ξ̄ < 0 [16]. (From “Theory of Buck-
ling and Post·Buckling Behavior of Elastic Structures” by B.
Budiansky in ADVANCES IN APPLIED MECHANICS, Vol-
ume 14, copyright 1974
c by Academic Press, reproduced by
permission of the publisher.)
© 2001 by Chapm
FIGURE 5.6
Equilibrium paths for perfect and imperfect models. (a) λ1 >
0; (b) λ1 = 0, λ2 < 0; (c) λ1 = 0, λ2 > 0 [16]. (From “Theory
of Buckling and Post·Buckling Behavior of Elastic Structures”
by B. Budiansky in ADVANCES IN APPLIED MECHANICS,
Volume 14, copyright 1974
c by Academic Press, reproduced
by permission of the publisher.)
FIGURE 5.7
Load-shortening relations for the simple model (a)
λ1 < 0, ξ̄ ≥ 0; (b) λ1 < 0, ξ̄ ≤ 0; (c) λ1 = 0, λ2 < 0, ξ̄ ≥ 0;
(d) λ1 = 0, λ2 > 0, ξ̄ ≥ 0 [16]. (From “Theory of Buckling
and Post·Buckling Behavior of Elastic Structures” by B.
Budiansky in ADVANCES IN APPLIED MECHANICS,
Volume 14, copyright 1974
c by Academic Press, reproduced
by permission of the publisher.)
© 2001 by Chapm
FIGURE 5.8
Deflection of plates with initial deviation from flatness.
(Adopted, in modified form, from [109].)
FIGURE 5.9
Deflections of plates with δ0 /t = 1, for various support condi-
tions. (Adopted, in modified form, from [94].)
© 2001 by Chapm
FIGURE 5.10
Deflections of plates with δ0 /t = 1, and sides free to wave.
(Adopted, in modified form, from [94].)
FIGURE 5.11
Shifting of the buckle pattern in plates under lateral and end
loads. (Adopted, in modified form, from [193].)
© 2001 by Chapm
FIGURE 5.12
Development of longitudinal buckles. (Adopted, in modified
form, from [193].)
FIGURE 5.13
Critical buckling strain of plates under lateral and end loading.
(Adopted, in modified form, from [120].)
© 2001 by Chapm
FIGURE 5.14
Load-deflection curves for various boundary conditions
isotropic plate under uniaxial compression. (Adopted, in mod-
ified form, from [94].)
FIGURE 5.15
Load-deflection curves for isotropic plate with various aspect
ratios [114]. (Reprinted from the Int. J. Solids and Struc-
tures, 31, Romeo, G. and Frulla, G., Nonlinear Analysis
of Anisotropic Plates with Initial Imperfections and Various
Boundary Conditions Subjected to Combined Biaxial Com-
pression and Shear Loads, 763-783, 1994, with permission from
Elsevier Science)
© 2001 by Chapm
FIGURE 5.16
Load-deflection behavior for an anisotropic plate under uniax-
ial compression [114]. (Reprinted from the Int. J. Solids and
Structures, 31, Romeo, G. and Frulla, G., Nonlinear Analysis
of Anisotropic Plates with Initial Imperfections and Various
Boundary Conditions Subjected to Combined Biaxial Com-
pression and Shear Loads, 763-783, 1994, with permission from
Elsevier Science)
FIGURE 5.17
Load-deflection curves of an anisotropic plate under uniaxial
compression. (a) Out-of-plane deflection at a quarter length.
(b) Overall deflection at P/Pcr = 2 [114]. (Reprinted from the
Int. J. Solids and Structures, 31, Romeo, G. and Frulla, G.,
Nonlinear Analysis of Anisotropic Plates with Initial Imper-
fections and Various Boundary Conditions Subjected to Com-
bined Biaxial Compression and Shear Loads, 763-783, 1994,
with permission from Elsevier Science)
© 2001 by Chapm
FIGURE 5.18
Plate geometry and coordinate system. (a) Annular sector showing posi-
tive polar coordinate and principal orthotropy directions. (b) Radial sec-
tion through annular plate with an initial deflection. (Turvey, G.J. and
Drinali, H., “Elastic Postbuckling of Circular and Annular Plates with
Imperfections,” Proc. 3rd. Int. Conf. Composite Struct. Appl.
Sci. Pub. 1985, 315–335. With permission of Elsevier Science.)
FIGURE 5.19
Edge compression-radial shortening response of an isotropic flat circular
plate: (ν = 0.3) (a) clamped edge (b) simply supported edge. (Turvey,
G.J. and Drinali, H., “Elastic Postbuckling of Circular and Annular Plates
with Imperfections,” Proc. 3rd. Int. Conf. Composite Struct.
Appl. Sci. Pub. 1985, 315–335. With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 5.20
Edge compression-plate center additional deflection response of isotropic
clamped imperfect circular plates (ν = 0.3). (Turvey, G.J. and Drinali,
H., “Elastic Postbuckling of Circular and Annular Plates with Imperfec-
tions,” Proc. 3rd. Int. Conf. Composite Struct. Appl. Sci. Pub.
1985, 315–335. With permission of Elsevier Science.)
FIGURE 5.21
Edge compression-hole-edge deflection response of orthotropic flat annu-
lar plates (φ = 0.4) and (νrθ = 0.3333). (Turvey, G.J. and Drinali,
H., “Elastic Postbuckling of Circular and Annular Plates with Imperfec-
tions,” Proc. 3rd. Int. Conf. Composite Struct. Appl. Sci. Pub.
1985, 315–335. With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 5.22
Edge compression-hole-edge additional deflection response for clamped
orthotropic imperfect annular plates (φ = 0.4) (a) β = 1.0, νrθ = 0.1
(b) β = 1/3, νrθ = 0.3. (Turvey, G.J. and Drinali, H., “Elastic Post-
buckling of Circular and Annular Plates with Imperfections,” Proc. 3rd.
Int. Conf. Composite Struct. Appl. Sci. Pub. 1985, 315–335. With
permission of Elsevier Science.)
FIGURE 5.23
Edge compression-hole-edge additional deflection response for clamped
orthotropic imperfect annular plates (φ = 0.4) (a) β = 3 νrθ = 0.1
(b) β = 1, νrθ = 0.3 (c) β = 1/3, νrθ = 0.3. (Turvey, G.J. and Dri-
nali, H., “Elastic Postbuckling of Circular and annular Plates with Imper-
fections,” Proc. 3rd. Int. Conf. Composite Struct. Appl. Sci.
Pub. 1985, 315–335. With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 5.24
Maximum average applied stress of an elastic-plastic simply
supported circular plate when buckling takes place in the elas-
tic range (ν = 13 , n = 12) [141]. (Reprinted from Int. J. Mech.
Sci., 17, Needleman, A., Postbifurcation Behavior and Imper-
fection Sensitivity of Elastic-Plastic Circular Plates, 1-13, 1975,
with permission from Elsevier Science)
FIGURE 5.25
A sketch of the response of the radially compressed and
clamped circular plate [201]. (From Cheo, L.S. and Reiss, E.L.,
Secondary Buckling of Circular Plates, SIAM J. Appl. Math.,
490, 26, 1974. With permission)
© 2001 by Chapm
FIGURE 5.26
Schematic bifurcation diagram [203]. (From Stroebel, G.J. and
Warner, W.H., “Stability and Secondary Bifurcation for von
Karman Plates”, J. Elast., 3, 185, 1973. Reprinted with kind
permission from Kluwer Academic Publishers.)
Chapter 6
and
h/2
x
M = σxx zdz
−h/2
h/2
My = σyy zdz (6.6b)
−h/2
h/2
xy
M = σxy zdz
−h/2
© 2001 by Chapm
In view of the definitions of $xx , $xy , and $yy , if we define, in the usual
way, the middle surface strains by (see (1.34), (1.35))
1 2
$oxx = u,x + w,x
2
o 1 2
$yy = v,y + w,y (6.7)
2
1
$o = (u,y + v,x + w,x w,y )
xy
2
so that
$ = $oxx − ζw,xx
xx
$xy = $oxy − ζw,xy (6.8)
$yy = $oyy − ζw,yy
then, by virtue of (6.5), (6.6a), and (6.8)
Eh
Nx = 1 − ν 2 ($xx + ν$yy ) − NHT
o o
Eh (6.9)
Ny = ($o + ν$oxx ) − NHT
1 − ν 2 yy
Nxy = 2Gh$oxy
where h/2
E
NHT = · $HT dz (6.10)
1−ν −h/2
The relations (6.12), with MHT = M T may also be found, e.g., in §9.4
of [217].
In certain situations it may be the case that the coefficients α and/or
β are field-dependent, i.e., α = α(δT ), β = β(δH). In such case, one
would have, e.g.,
h/2
E
N = 1 − ν −h/2 α(δT ) · δT dz
T
h/2 (6.16)
MT = E
α(δT ) · δT zdz
1 − ν −h/2
∂2 ∂2
where = 2
+ 2 is the two-dimensional Laplacian while 2 is the
∂x ∂y
biharmonic operator. Modifications (which will be discussed later) must
be made to (6.19) if imperfection buckling is considered, i.e., if the plate
possesses an initial prebuckling deflection w0 = w0 (x, y) or is subject to
a transverse normal loading. As in Chapter 1, we may introduce the
Airy stress function Φ(x, y) by
where $0xx , $0yy , $0xy are the middle surface strains, as given by (6.7) and
the constitutive relations (6.9), we easily obtain the second of the two
generalized von Karman equations which apply in the case of hygrother-
mal buckling, i.e.,
© 2001 by Chapm
is the twisting rigidity. Often, the ratios D12 /D22 and D26 /D11 are
termed reduced Poisson’s ratios.
We write (6.25) out in the form
σxx = c11 ($0xx − ζw,xx ) + c12 ($0yy − ζw,yy )
−c11 $1HT − c12 $2HT
σyy = c21 ($0xx − ζw,xx ) + c22 ($0yy − ζw,yy ) (6.27)
−c21 $1HT − c22 $2HT
σxy = 2c66 ($0xy − ζw,xy )
as α1 = α2 = α, β1 = β2 = β. Then
h/2
E
NHT + NHT =
11 12
$HT dz
1 − ν −h/2
h/2 (6.30)
E
21
NHT 22
+ NHT = $HT dz
1 − ν −h/2
11 12
and it is clear that (6.28) reduces to (6.9) with NHT = NHT + NHT =
21 22
NHT + NHT given by (6.10).
From (6.28) we have, immediately, that
1
$0xy = Nxy (6.31a)
2G12 h
yields
1 E2 1
$0xx = (Nx + ÑHT )
|cij |h 1 − ν12 ν21
(6.31b)
E2 ν21
− 2
(Ny + ÑHT )
1 − ν12 ν21
1 E1 2
$0yy = (Ny + ÑHT )
|cij |h 1 − ν12 ν21
(6.31c)
E1 ν12
− 1
(Nx + ÑHT )
1 − ν12 ν21
© 2001 by Chapm
with ' '
' E1 E2 ν21 '
' '
' 1 − ν12 ν21 1 − ν12 ν21 '
|cij | = '' ',
'
' E1 ν12 E2 '
'1−ν ν 1−ν ν '
12 21 12 21
i.e.,
|cij | = E1 E2 /(1 − ν12 ν21 ) (6.32)
© 2001 by Chapm
and then substitute into (6.22) so as to obtain
1 1 1 2ν12
Φ,yyyy + ( − )Φ,xxyy
E1 h h G12 E2
1 1
+ Φ,xxxx = − [w, w] (6.39)
E2 h 2
1 1
− 1
(ÑHT − ν21 ÑHT
2
),yy − (Ñ 2 − ν12 ÑHT
1
),xx
E1 h E2 h HT
where err , eθθ , and γrθ are given in terms of the displacement com-
ponents ur , uθ in the middle surface of the plate and the out-of-plane
displacement w = w(r, θ) by the relations in (1.71), i.e.,
∂ur 1 ∂w 2 ∂2w
err = + ( ) −ζ 2
∂r 2 ∂r ∂r
ur 1 ∂uθ 1 ∂w 2 1 ∂w 1 ∂2w
eθθ = + + ( ) − ζ( + )
r r ∂θ ∂r2 ∂θ r ∂r r2 ∂θ2
2
γrθ = ∂uθ − uθ + 1 ( ∂ur ) + 1 ( ∂w )( ∂w ) − 2ζ( 1 ∂ w − 1 ∂w )
∂r r r ∂θ r ∂r ∂θ r ∂r∂θ r2 ∂θ
Middle surface strains eorr , eoθθ , γrθ
o
are obtained from these relations by
simply setting ζ = 0.
The stress components σrr , σθθ , and σrθ must satisfy the equilibrium
equations delineated, e.g., in (1.72a, b); they are related to the stress
components in rectangular Cartesian coordinates by the equations in
(1.73), i.e.,
Assuming that the thin plate exhibits linearly elastic response, we have,
in lieu of (6.4), the constitutive relations
1
err − e∗HT = (σrr − νσθθ )
E
∗ 1
eθθ − eHT = (σθθ − νσrr ) (6.42)
E
1+ν
erθ = ( )σrθ
E
© 2001 by Chapm
whose inverted form is
E E ∗
σrr = (err + νeθθ ) − e
1−ν 2 1 − ν HT
E E ∗
σθθ = (e + νerr ) −
2 θθ
e (6.43)
1 − ν 1 − ν HT
E
σrθ = ( )erθ ≡ Gγrθ
1+ν
The averaged stresses and bending moments in polar coordinates are
defined, in the obvious way, as the natural counterparts of (6.6a, b), i.e.
h/2
Nr = −h/2 σrr dz
h/2 (etc.)
Mr =
σrr zdz
−h/2
(6.47)
1
2 Φ(r, θ) = − Eh[w(r, θ), w(r, θ)] − (1 − ν)NHT ∗
2
where h/2
E
N ∗
= e∗ dz
HT 1 − ν −h/2 HT
h/2 (6.48)
E
∗
e∗ zdz
MHT =
1 − ν −h/2 HT
while 2 w is given by (1.75a), i.e.,
2 1
2 w = w,rrrr + w,rrr − 2 w,rr
r r
2 1 2
+ 2 w,rrθθ + 3 w,r − 3 w,rθθ
r r r
1 4
+ 4 w,θθθθ + 4 w,θθ ,
r r
with an analogous expression for 2 Φ, while
1 1
[Φ, w] = w,rr ( Φ,r + 2 Φ,θθ )
r r
1 1
+( w,r + 2 w,θθ )Φ,rr
r r
1 1 1 1
−2( w,rθ − 2 w,θ )( Φ,rθ − 2 Φ,θ ),
r r r r
1 1
= Nr w,rr − 2Nrθ ( 2 w,θ − w,rθ )
r r
1 1
+Nθ ( w,r + 2 w,θθ )
r r
1 1
[w, w] = 2{w,rr ( w,r + 2 w,θθ )
r r
1 1
−( w,rθ − 2 w,θ )2 }
r r
© 2001 by Chapm
and
∗ ∗ 1 ∗
MHT (r, θ) = (MHT ),rr + (MHT ),r
r (6.49)
1 ∗
+ 2 (MHT ),θθ
r
∗
with an analogous expression for NHT (r, θ). If δH,θ = δT,θ = 0 so that
2 1
K[w + w − w ]
r r (6.51a)
Nθ ∗ 1 ∗
= Nr w + w − {MHT + MHT },
r r
h/2
d 1 E
where = , Nr = Φ , Nθ = Φ, MHT
∗
(r) = e∗HT (r, z)zdz,
dr r 1−ν −h/2
and
1 2 1 1
[Φ + Φ − 2 Φ + 3 Φ ]
Eh r r r
(6.51b)
1 (1 − ν) ∗ 1 ∗
=− ww − {NHT + NHT }
r Eh r
Both the systems (6.47) and (6.51a, b) neglect the effects of initial (pre-
buckling) deflections and an applied transverse loading; special cases of
these systems, which have appeared in the literature in connection with
problems associated with the thermal bucking of isotropic linearly elastic
circular plates, will be analyzed later in this report.
For a linearly elastic orthotropic body exhibiting cylindrical orthotropy
there exist three planes of elastic symmetry; one of these is normal to
the plane of anisotropy, the second passes through that axis, and the
third is orthogonal to the first two. For a thin plate, the first plane
of elastic symmetry, for a cylindrically orthotropic material, is chosen
parallel to the middle plane of the plate; with this convention, the con-
stitutive equations generalize those in (6.42) for an isotropic plate (in
polar coordinates) and assume the form
1 νθ
err − erHT = σrr − σθθ
E E
r θ
νr 1
eθθ − eθHT = − σrr + σθθ (6.52)
Er Eθ
1
γrθ = σrθ
Grθ
where the radial and angular hygrothermal strains erHT and eθHT are
defined, respectively, by
erHT = βr δH(r, θ, z) + αr δT (r, θ, z)
(6.53)
eθHT = βθ δH(r, θ, z) + αθ δT (r, θ, z)
© 2001 by Chapm
surface strains, by
err = eorr − ζw,rr
1 1
eθθ = eoθθ − ζ( w,r + 2 w,θθ ) (6.55)
r r
γ = γ o − 2ζ( 1 w − 1 w )
rθ rθ ,rθ ,θ
r r2
where
1
eorr = ur,r + (w,r )2
2
1 1 1
eθθ = ur + uθ,θ + 2 (w,θ )2
o (6.56)
r r 2r
γo = u − 1 u + 1 u + 1 w w
rθ θ,r θ r,θ ,r ,θ
r r r
Inverting the relations (6.54) we obtain
Er νr E θ
σrr = err + eθθ
1 − νr νθ 1 − νr νθ
(6.57a)
1
− {Er erHT + νr Eθ eθHT }
1 − νr νθ
νθ E r Eθ
σθθ = err + eθθ
1 − νr νθ 1 − νr νθ
(6.57b)
1
− {νθ Er erHT + Eθ eθHT }
1 − νr νθ
For the cylindrically orthotropic case, the averaged stresses and bending
moments are still defined by the relations following (6.43); thus, by
(6.57a, b, c)
Er h νr E θ h o
Nr = eorr + e − NHT r
1 − ν ν
r θ 1 − νr νθ θθ
νθ Er h o Eθ h (6.58)
Nθ = err + eoθθ − NHT
θ
1 − νr νθ 1 − ν r νθ
Nrθ = 2Grθ heorθ
where
h/2 h/2
Er νr E θ
N r
= er
dz + eθ dz
HT 1 − νr νθ −h/2 HT 1 − νr νθ −h/2 HT
h/2 h/2 (6.59)
νθ E r Eθ
θ r
eθ dz
HT
N = e dz +
1 − νr νθ −h/2 HT 1 − νr νθ −h/2 HT
then
1
Mr = −Dr [w,rr + νθ ( w,r +
1
w,θθ) − MHT
r
r r2
1 1
Mθ = −Dθ [νr w,rr + ( w,r + w,θθ )] − MHT
θ (6.62)
r r2
Mrθ = −2D̃rθ ( w ),rθ
r
with
h/2 h/2
Er νr E θ
r r θ
MHT = 1 − νr νθ −h/2 eHT zdz + 1 − νr νθ −h/2 eHT zdz
h/2 h/2 (6.63)
νθ E r Eθ
MHT
θ
erHT zdz + eθ zdz
=
1 − νr νθ −h/2 1 − νr νθ −h/2 HT
The expressions in (6.58) for the averaged stresses and (6.62), for
the bending moments generalize, for the case of cylindrical (polar) or-
thotropic behavior, the corresponding relations (1.93) and (1.94) for the
non-hygrothermal case.
The first of the generalized von Karman equations for a plate possess-
ing cylindrically orthotropic symmetry is obtained by substituting the
© 2001 by Chapm
expressions in (6.62), (6.63), for the bending moments, into the polar
coordinate equivalent form of (6.18), namely,
1 1 1
(rMr ),rr + 2 Mθ,θθ − Mθ,r
r r r
1
+ Mrθ,rθ + Nr w,rr (6.64)
r
1 1 w
+Nθ ( w,r + 2 w,θθ ) + 2Nrθ ( ),rθ = 0
r r r
where the stress resultants Nr , Nθ , and Nrθ are, again, given by (6.46)
in terms of the Airy stress function Φ(r, θ). Equation (6.64) is identical
with (1.106) of Chapter 1. Carrying out the process indicated above, we
obtain
1 1
Dr w,rrrr + 2Drθ 2 w,rrθθ + Dθ 4 w,θθθθ
r r
1 1 1
+2Dr w,rrr − 2Drθ 3 w,rθθ − Dθ 2 w,rr (6.65)
r r r
1 1
+2(Dθ + Drθ ) 4 w,θθ + Dθ 3 w,r
r r
1 1
= ( Φ,r + 2 Φ,θθ )w,rr
r r
1 1
+Φ,rr ( w,r + 2 w,θθ )
r r
1 1 1 1
+2( 2 Φ,θ − Φ,rθ )( w,rθ − 2 w,θ )
r r r r
1 1 1
+ (rMHTr θ
),rr + 2 (MHT ),θθ − (MHT
θ
),r
r r r
To obtain the polar coordinate equivalent form of the strain compati-
bility relation without transforming this relation directly, we proceed as
follows: with respect to the linearized strains
err = ur,r
1 1
eθθ = ur + uθ,θ (6.66)
r r
γ = u + 1 u − 1 u
rθ θ,r r,θ θ
r r
it is easy to check that
(rγrθ,θ ),r − err,θθ − (r2 eθθ,r ),r + rerr,r = 0 (6.67)
Moreover, in view of (6.56),
1
eorr = err + (w,r )2
2
1
eθθ = eθθ + 2 (w,θ )2
o (6.68)
2r
γo = γ + 1 w w
rθ rθ ,r ,θ
r
Thus, strain compatibility, when written in terms of the middle surface
strains, requires that
o
(rγrθ,θ ),r − eorr,θθ − (r2 eoθθ,r ),r + reorr,r
1 1
= [r( w,r w,θ ),θ ],r − [ (w,r )2 ],r (6.69)
r 2
1 1
−[r2 ( 2 (w,θ ),r ],r + r( (w,r )2 ),r
2
2r 2
Expanding the right-hand side of (6.69), and simplifying, we obtain
o
(rγrθ,θ ),r − eorr,θθ − (r2 eoθθ,r ),r + reorr,r
1 (6.70)
= w,rr (rw,r + w,θθ ) − (w,rθ − w,θ )2
r
If we invert the relations in (6.44) we find that
1
err = E h {(Nr + NHT ) − νr (Nθ + NHT )}
o r θ
r
1
eoθθ = {(Nθ + NHTθ
) − νθ (Nr + NHT
r
)} (6.71)
E
θh
1
γrθ
o
= Nrθ
Grθ h
Computing, in succession, therefore, the expressions on the left-hand
side of the compatibility relation (6.70), we now obtain, through the use
of (6.71)
1
o
(rγrθ,θ ),r = {Nrθ,θ + rNrθ,rθ } (6.72a)
Grθ h
1
eorr,θθ = {(Nr,θθ + NHT,θθ
r
) − νr (Nθ,θθ + NHT,θθ
θ
)} (6.72b)
Er h
© 2001 by Chapm
r2
(r2 eoθθ,r ),r = {Nθ + NHT
θ
),rr − νθ (Nr + NHT
r
),rr }
Eθ h
(6.72c)
2r
+ {(Nθ + NHT
θ
),r − νθ (nr + NHT
r
),r }
Eθ h
and
r
reorr,r = {(Nr + NHT
r
),r − νr (Nθ + NHT
θ
),r } (6.72d)
Er h
1
[(Nr − νr Nθ ),θθ + r(Nr,r − νr Nθ,r )]
Er h
1
+ (Nrθ,θ − rNrθ,rθ )
2Grθ h
1
+ [(r2 (Nθ,rr − νθ Nr,rr ) + 2r(Nθ,r − νθ Nr,r )]
Eθ h
1 (6.73)
= w,rr (rw,r + w,θθ ) − (w,rθ − w,θ )2
r
1
− [(NHT − νr NHT ),θθ + r(NHT,,r − νr NHT,r
r θ r θ
)]
Er h
1
− [r2 (NHT,rr
θ
− νθ NHT,rr
r
)
Eθ h
θ
+2r(NHT,rθ − νθ NHT,r
r
)]
To rewrite (6.73) in terms of the Airy function Φ(r, θ), we substitute for
Nr , Nθ , and Nrθ in (6.73) from (6.46), multiply the resulting equation
through by −h/r2 , and simplify; there results the following equation:
1 1 2νr 1
Φ,rrrr + ( − ) Φ,rrθθ
Eθ Grθ Er r 2
1 1 2 1
+ Φ,θθθθ + Φ,rrr
Er r 4 Eθ r
1 2νr 1 1 1
−( − ) 3 Φ,rθθ − Φ,rr
Grθ Er r Er r 2
1 − νr 1 1 1 1
+(2 + ) 4 Φ,θθ + Φ,r
Er Grθ r Er r 3
1 1 (6.74)
= −h[w,rr ( w,r + 2 w,θθ )
r r
1 1
−( w,rθ − 2 w,θ )2 ]
r r
1 1 1 r
+ [ 2 (NHT − νr NHT ),θθ + (NHT,r − νr NHT,r
r θ θ
)]
Er r r
1 2 θ
θ
+ [(NHT,rr − νθ NHT,rr
r
) + (NHT,rθ − νθ NHT,r
θ
)]
Eθ r
© 2001 by Chapm
point on the boundary, while t denotes the unit tangent vector to the
boundary at that point. The normal derivative of a function f on ∂Ω is
∂f ∂f
denoted by equivf,n while the tangential derivative is given by ,
∂n ∂s
s being a measure of arc length along the boundary. Thus, e.g., if Ω
is a disk centered at (0, 0) of radius R > 0 and f = f (rθ) is defined
on Ω and is of class C 1 (Ω) with first derivative continuous up to ∂Ω,
∂f ∂f 1
then = f,r while = f,θ . As noted in Chapter 1, the three
∂n ∂s r
most prevalent types of boundary conditions in the buckling literature
are those which correspond to clamped edges, simply supported edges,
and free edges; regardless of whether we are considering plates with
isotropic or orthotropic symmetry (either cylindrical or rectilinear), the
basic forms assumed by these various sets of boundary conditions are
still the same as those delineated in (1.129a, b, c), i.e.,
∂w
(i) ∂Ω is clamped: w = 0 and = 0, on ∂Ω
∂n
∂Mtn
(iii) ∂Ω is free: Mn = 0 and Qn + = 0, on ∂Ω
∂s
where Mn is the bending moment on ∂Ω in the direction normal to
∂Ω, Mtn is the twisting moment on ∂Ω, with respect to the tangential
and normal directions on ∂Ω, and Qn is the shearing force associated
with the direction normal to ∂Ω.
For the work to be considered in this report only rectangular and circu-
lar (or annular) domains Ω will be covered. If ∂Ω is clamped, therefore,
or if, as in the case of a rectangular plate, one or more edges are clamped,
the pertinent boundary conditions will be exactly the same as for the
non-hygrothermal case, i.e., for a rectangle of width a and length b,
exhibiting isotropic response, the clamped boundary conditions are ex-
pressed by (1.130a, b). For a circular plate (or annular plate) exhibiting
isotropic response the relevant conditions are those in (1.139). Condi-
tions (1.130a, b) apply equally well to a rectangular plate exhibiting
rectilinear orthotropic response when all the edges are clamped, while
(1.139) still applies for circular (or annular) plates with clamped edge(s)
when the plate exhibits cylindrically orthotropic behavior.
We now consider thin elastic plates subject to hygrothermal expansion
or contraction which have one or more edges simply supported. For a
rectangular plate of width a and length b the condition Mn = 0 on ∂Ω
becomes
Mx = 0, x = 0, x = a; 0 ≤ y ≤ b
(6.75)
My = 0, y = 0, y = b; 0 ≤ x ≤ a
Thus, by virtue of (6.12),
K(w,xx + νw,yy ) + MHT = 0,
(6.76a)
x = 0, x = a; 0 ≤ y ≤ b
and
K(w,yy + νw,xx ) + MHT = 0
(6.76b)
y = 0, y = b; 0 ≤ x ≤ a,
where h/2
MHT = E
$HT zdz
1−ν −h/2 (6.77)
$HT = βδH(x, y, z) + αδT (x, y, z)
For the same rectangular elastic plate, now assumed to exhibit recti-
linear orthotropic symmetry, the conditions in (6.75) take, as a direct
consequence of (6.35) and (6.37), the following form:
1
(D11 w,xx + D12 w,yy ) + M̃HT = 0,
(6.78a)
x = 0, x = a; 0 ≤ y ≤ b
2
(D21 w,xx + D22 w,yy ) + M̃HT = 0,
(6.78b)
y = 0, y = b; 0 ≤ x ≤ a
where Dij = cij h3 /12, the constitutive constants cij are given by (1.61)
and
h/2
1
M̃HT = c11 $1HT (x, y, z)zdz
−h/2
(6.79a)
h/2
+c12 $2HT (x, y, z)zdz
−h/2
h/2
2
M̃HT = c21 $1HT (x, y, z)zdz
−h/2
(6.79b)
h/2
+c22 $2HT (x, y, z)zdz
−h/2
© 2001 by Chapm
with, as per (6.26),
$1HT = β1 δH(x, y, z) + α1 δT (x, y, z)
where h/2
M∗ E
= e∗HT zdz
HT
1−ν −h/2 (6.81)
e∗HT = βδH(r, θ, z) + αδT (r, θ, z)
On the other hand, for a circular (or annular) plate with edge(s) at
r = Ri , i = 1, 2, which exhibits cylindrically orthotropic symmetry, we
have as the expression of Mn = 0 on ∂Ω the condition
1 1 r
Dr [w,rr + νθ ( w,r + 2 w,θθ )] + MHT = 0,
r r (6.82)
r = Ri , i = 1, 2
and Dr = Er h3 /12(1 − νr νθ ).
Along any edge of a thin plate which is free, we must have Mn = 0
∂Mtn
as well as Qn + = 0. Conditions equivalent to Mn = 0 along
∂s
a portion of ∂Ω (or all of ∂Ω) for various cases of interest have been
elucidated above. For rectangular plates of width a and length b, it has
∂Mtn
been shown in Chapter 1 that the condition Qn + = 0 on ∂Ω is
∂s
equivalent to the following relations
My,y + 2Mxy,x = 0,
(6.83a)
y = 0, y = b; 0 ≤ x ≤ a
Mx,x + 2Myx,y = 0,
(6.83b)
x = 0, x = a; 0 ≤ y ≤ b
and
K[w,xxx + (2 − ν)w,xyy ] + MHT,x = 0
(6.84b)
x = 0, x = a; 0 ≤ y ≤ b
h/2
E
with MHT = $HT zdz and $HT = βδH + αδT.
1 − ν −h/2
For the case in which the rectangular plate exhibits rectilinear or-
thotropic symmetry, (6.83a,b) still represent the conditions equivalent
∂M=n
to Qn + = 0 along all four edges, but now, the bending moments
∂s
Mx , My , and Mxy are given by (6.35). An easy computation then shows
that in lieu of (6.84a, b) for the isotropic case we have
2
D21 w,xxy + D22 w,yyy + 4D66 w,xxy + M̃HT,y = 0,
(6.85a)
y = 0, b; 0 ≤ x ≤ a
and
1
D11 w,xxx + D12 w,xyy + 4D66 w,xyy + M̃HT,x = 0,
(6.85b)
x = 0, a; 0 ≤ y ≤ b
1 2
where M̃HT , M̃HT are defined by (6.34) and (6.37) with $1HT , $2HT as
given by (6.26).
To elucidate the free edge boundary conditions which apply with re-
spect to thin elastic plates with a circular geometry, we note that (see
© 2001 by Chapm
[217], §2.2 ) in the orthogonal (s, n) coordinate system introduced along
the boundary ∂Ω of a domain Ω in the x, y plane
∂Mn ∂Mtn
Qn = +2 (6.86a)
∂n ∂s
∂Mn ∂Mtn
+2 =0 (6.86b)
∂n ∂s
1 1
K{(w,rr + w,r + 2 w,θθ ),r
r r
(6.88)
1 1 ∗
+(1 − ν)( 2 w,θθr − 3 w,θθ )} + MHT,r = 0, r = Ri , i = 1, 2
r r
∗
where MHT (r, θ) is given by (6.81). For the same annular plate, this time
exhibiting cylindrically orthotropic behavior, we have as a consequence
of (6.62)
1 1
Dr [w,rrr + νθ ( w,r + 2 w,θθ ),r ]
r r
4 w r (6.89)
+ D̃rθ ( ),rθθ + MHT,r = 0,
r r
r = Ri , i = 1, 2
where $oxx , $oxy , $oyy are the middle surface strains as defined by (6.7) and
N T is given by (6.11a). The bending moments are given as
Mx = −K(w,xx + νw,yy ) − M
T
My = −K(w,yy + νw,xx ) − M T (6.92)
Mxy = −(1 − ν)Kw,xy
© 2001 by Chapm
with the thermal moment M T given by (6.14).
The generalized von Karman equations for the hygrothermal case now
reduce to (see (6.21), (6.24)):
and
1
2 Φ = − Eh[w, w] − (1 − ν)N T (6.93b)
2
where t = t(x, y) is the applied transverse force. In considering small
deflection theory one ignores the bracket operator on the right-hand side
of (6.93b), in which case, the Airy function, as given by (6.20), satisfies
has been considered; for this specific type of distribution it is easy to see
that, as a direct consequence of (6.11a) and (6.14),
αEH
NT = To (x, y)
1−ν (6.95)
3
M T = αE · h T1 (x, y)
1 − ν 12
For such a temperature distribution within the context of the small
deflection equations, it will generally be the case that one is dealing
with a thermal bending problem when N T = To = 0 and a thermal
buckling problem when M T = T1 = 0. The boundary conditions
associated with the thermal bending and buckling of, say, a rectangular
plate of width a and length b exhibiting isotropic material symmetry are
as follows:
(i) If all the edges are clamped, then the conditions coincide with
(1.130a, b).
(ii) If all the edges are simply supported, then w = 0 along each edge
and, in addition,
K(w,xx + νw,yy ) + M T = 0
x = 0, a; 0 ≤ y ≤ b
(6.96)
K(w,yy + νw,xx ) + M T = 0
y = 0, b; 0 ≤ x ≤ a
(iii) If the edges are all free, then the conditions in (6.96) hold as well
as
K[w,yyy + (2 − ν)w,xxy ] + M,y T
=0
y = 0, b; 0 ≤ x ≤ a
(6.97)
K[w,xxx + (2 − ν)w,xyy ] + M,x T
=0
x = 0, a; 0 ≤ y ≤ b
with the coefficients of linear thermal expansion along the x and y axes,
α1 and α2 , respectively, taken to be constants. The constitutive relations
in this situation are given as follows (where $oxx , $oxy , $oyy are, once again,
the middle surface strains):
E1 h E2 ν21 H
Nx = { }$oxx +} }$oyy − ÑT1
1 − ν ν 1 − ν ν
12 21 12 21
E1 ν12 h E 2h (6.99)
Ny = { }$oxx + { }$oyy − ÑT2
1 − ν 12 ν21 1 − ν 12 ν21
Nxy = (2G12 h)$oxy
© 2001 by Chapm
where (see (6.29)):
h/2
E α + E2 α2 ν21
ÑT1 = ( 1 1
) δT (x, y, z)dz
1 − ν12 ν21 −h/2
h/2 (6.100)
E1 α1 ν12 + E2 α2
2
TÑ = ( ) δT (x, y, z)dz
1 − ν12 ν21 −h/2
and the cij are given by (1.61). The relevant von Karman equations for
thermal bending/buckling become (see (6.36), (6.39)):
D11 w,xxxx + {D12 + 4D66 + D21 }w,xxyy
(6.103a)
+D22 w,yyyy = [Φ, w] − M̃T,xx
1
− M̃T,yy
1
+t
and
1 1 1 2ν12
Φ,yyyy + ( − )Φ,xxyy
E1 h h G12 E2
1 1
+ Φ,xxxx = − [w, w]
E2 h 2
(6.103b)
1
− (Ñ 1 − ν21 ÑT2 ),yy
E1 h T
1
− (Ñ 2 − ν12 ÑT1 ),xx
E2 h T
As for the boundary data in this case, with respect, e.g., to a rectangular
plate of width a and length b, we have the following:
(i) If all four edges are clamped, then conditions (1.130a, b) still apply.
(ii) If the four edges are simply supported, then w = 0 along each edge
and, in addition,
(D11 w,xx + D12 w,yy ) + M̃T1 = 0,
x = 0, a; 0 ≤ y ≤ b
(6.104)
2
(D 21 w,xx + D 22 w,yy ) + M̃ T = 0,
y = 0, b; 0 ≤ x ≤ a
and
(c21 α1 + c22 α2 )h3
M̃T1 = T1 (x, y) (6.106b)
12
© 2001 by Chapm
For a circular (or annular) plate exhibiting isotropic material symmetry,
the thermal strain (see (6.40)) is given by
e∗T = αδT (r, θ, z) (6.107)
For the isotropic case, the constitutive relations (6.44) reduce to
Eh
(eo + νeoθθ ) − NT∗
Nr =
1 − ν 2 rr
Eh (6.108)
Nθ = (eo + νeorr ) − NT∗
1 − ν 2 θθ
Nrθ = 2Gheorθ
where the eorr , eorθ , eoθθ are the middle surface strains as given by (1.71)
with ζ = 0, while
h/2
∗ Eα
NT = δT (r, θ, z)dz (6.109)
1 − ν −h/2
The bending moments for the isotropic case in polar coordinates are
given by (see [217], §4):
1 1
Mr = −K[w,rr + ν( w,r + 2 w,θθ )] − MT∗
r r
1 1
Mθ = −K[νw,rr + ( w,r + 2 w,θθ )] − MT∗ (6.110)
r r
Mrθ = −(1 − ν)K(w,rθ − 1 w,θ )
r2
where h/2
Eα
MT∗ = δT (r, θ, z)zdz (6.111)
1−ν −h/2
The von Karman system for this case now assumes the form
∗
K w = [Φ, w] − MT + t(r, θ)
2
(6.112)
2 Φ
1
= − Eh[w, w] − (1 − ν)NT∗
2
where [Φ, w] and [w, w], as well as 2 w, are given by the expressions
∂2 1 ∂ 1 ∂2
directly following (6.48) and = 2 + + 2 2 ; for the small de-
∂r r ∂r r ∂θ
flection case one again deletes the “bracket” [w, w] in the second equation
in (6.112). Thermal bending alone for the circular (or annular) isotropic
plate is governed by the first of the two equations in (6.112) with Φ ≡ 0.
For a temperature distribution varying linearly through the thickness
of the plate we have, in lieu of (6.94),
δT (r, θ, z) = To (r, θ) + zT1 (r, θ) (6.113)
In this special case (6.109) and (6.111) become, respectively,
∗ Eαh
NT = 1 − ν To (r, θ)
(6.114)
∗
Eαh3
MT = T1 (r, θ)
12(1 − ν)
We now delineate the boundary conditions that are associated with the
system (6.112), or a specialization thereof, for the case of an annular
plate with edges at r = Ri , i = 1, 2, R1 = a, R2 = b > a.
∂w
(i) If both edges are clamped, then w = 0 and = 0, along r =
∂r
Ri , i = 1, 2.
(ii) If the plate edges at r = Ri , i = 1, 2 are simply supported, then
w = 0, for r = Ri , i = 1, 2, and, in addition,
1 1
K[w,rr + ν( w,θθ + w,r )] + MT∗ = 0
r2 r (6.115)
for r = Ri , i = 1, 2
Our last case in this sequence concerns the thermal bending and/or
buckling of thin elastic circular (or annular) plates exhibiting cylindri-
cally orthotropic behavior. The thermal strains in the radial and angular
directions are given by (see (6.53))
erT = αr δT (r, θ, Z), eθT = αθ δT (r, θ, z) (6.117)
© 2001 by Chapm
In lieu of (6.108)–(6.111) for the isotropic case, we now have the following
sets of expressions for the resultant forces and bending moments:
Er h νr E θ h o
Nr = eorr + eθθ − NTr
1 − ν ν
r θ 1 − ν ν
r θ
νθ E r h o Eθ h (6.118)
Nθ = err + eoθθ NTθ
1 − ν r ν θ 1 − ν r νθ
Nrθ = 2Grθ heorθ
with
h/2
Er αr + νr Eθ αθ
N r
= ( ) δT (r, θ, z)dz
T 1 − νr νθ −h/2
h/2 (6.119)
νθ Er αr + Eθ αθ
θ
T
N = ( ) δT (r, θ, z)dz
1 − ν r νθ −h/2
and
1
Mr = −Dr [w,rr + νθ ( w,r +
1
w,θθ )] − MTr
r r2
1 1
Mθ = −Dθ [νr w,rr + ( w,r + w,θθ )] − MTθ (6.120)
r r2
w
Mrθ = −2D̃rθ ( ),rθ
r
with
h/2
E α + νr Eθ αθ
MTr = ( r r
) δT (r, θ, z)zdz
1 − νr νθ −h/2
h/2 (6.121)
νθ Er αr + Eθ αθ
θ
T
M = ( ) δT (r, θ, z)zdz
1 − νr νθ −h/2
Employing (6.120) in (6.64), and using (6.46), we obtain the first of the
von Karman equations for thermal bending and/or buckling of a thin,
elastic, cylindrically orthotropic plate, i.e.,
1 1
Dr w,rrrr + 2Drθ · w,rrθθ + Dθ · 4 w,θθθθ
r2 r
1 1 1
+2Dr · w,rrr − 2Drθ 3 w,rθθ − Dθ 2 w,rr
r r r
1 1
+2(Dθ + Drθ ) 4 w,θθ + Dθ 3 w,r
r r
1 1
= ( Φ,r + 2 Φ,θθ )w,rr
r r
(6.122)
1 1
+Φ,rr ( w,r + 2 w,θθ )
r r
1 1 1 1
+2( 2 Φ,θ − Φ,rθ )( w,rθ − 2 w,θ )
r r r r
1 r 1 θ
+ (rMT ),rr + 2 (MT ),θθ
r r
1
− (MTθ ),r + t(r, θ)
r
while the second of the relevant von Karman equations for this case
becomes
1 1 2νr 1
Φ,rrrr + ( − ) Φ,rrθθ
Eθ Grθ Er r2
1 1 2 1
+ · Φ,θθθθ + · Φ,rrr
Er r 4 Eθ r
1 2νr 1 1 1
−( − ) 3 Φ,rθθ − · 2 Φ,rr
Grθ Er r Er r
1 − νr 1 1 1 1
+(2 + ) 4 Φ,θθ + · 3 Φ,r (6.123)
Er Grθ r Er r
1 1 1 1
= −h[w,rr ( w,r + 2 w,θθ ) − ( w,rθ − 2 w,θ )2 ]
r r r r
1 1 1 r
+ { 2 (NTr − νr NTθ ),θθ + (NT,r − νr NT,r
θ
)}
Er r r
1 2 θ
+ {(NT,r θ
− νθ NT,rr
r
) + (NT,rθ − νθ NT,r
θ
)}
Eθ r
© 2001 by Chapm
(6.123) may be simplified somewhat, e.g.,
h/2
NTr − νr NTθ = Er αr δT (r, θ, z)dz
−h/2
(ii) For small deflections, the first term on the right-hand side of
(6.123), in square brackets, is deleted.
(iii) The (purely) thermal bending problem is governed by (6.122) with
Φ = 0.
(iv) For an axially symmetric problem, both (6.122) and (6.123) reduce
to (variable coefficient) ordinary differential equations in the radial
variable r; we have w = w(r), Φ = Φ(r) and T = T (r, z) so that
the resultants NTr , NTθ and moments MTr , MTθ are functions only of
r. In (6.122) and (6.123), therefore, derivatives of all quantities,
of any order, with respect to θ vanish.
The boundary data for an annular elastic plate of inner radius R1 =
a and outer radius R2 = b > a, exhibiting cylindrically orthotropic
symmetry, may be specified as follows:
∂w
(i) If both edges are clamped, then w = 0 and = 0, along r =
∂r
Ri , i = 1, 2.
(ii) If the plate edges are simply supported, then w = 0, for r = Ri , i =
1, 2 and, additionally,
1 1
Dr [w,rr + νθ ( w,r + 2 w,θθ )] + MTr = 0
r r (6.126)
for r = Ri , i = 1, 2
The usual considerations apply if one edge is, e.g., clamped while the
other is simply supported, or if one edge is simply supported while the
other is free, etc.; for a circular plate of radius R = b, R1 = a = 0.
Remarks: All of the problems considered above may be posed in terms
of the middle surface displacements u, v and the out-of-plane deflection
w in lieu of w and the Airy function Φ; the idea is most feasible within
the context of small deflection theory. For an isotropic rectangular plate,
small deflection theory corresponds to the substitution of the expressions
for the middle surface strains $oxx , $oxy , and $oyy from (6.7) into the con-
stitutive relations (6.91), suppression of all those terms involving the
out-of-plane displacement w, and then substitution of the resultant ex-
pressions for Nx , Ny , and Nxy into the in-plane equilibrium equations
(6.17a); this process leads to the pair of equations
Eh Eh T
1 − ν 2 (u,xx + νv,xy ) + 2(1 + ν) (u,yy + v,xy ) = N,x
(6.128)
Eh (u,xy + v,xx ) + Eh (νu,xy + v,yy ) = N,y
T
2(1 + ν) 1 − ν2
which must be solved in conjunction with (6.93a), which we repeat here
as
K2 w = Nx w,xx + Nxy w,xy + Ny w,yy − M T + t (6.129)
For the small deflection case considered above, (6.128) and (6.129) are
decoupled, just as (6.93a), (6.93b) are if, in (6.93b), we delete the bracket
© 2001 by Chapm
[w, w]. Thus (6.128) must be solved subject to appropriate boundary
conditions for the middle surface displacements u, v which, in turn,
are used to compute Nx , Nxy , and Ny for subsequent substitution in
(6.129). Little use of the displacement formulation of the thermal bend-
ing/buckling problem will be made in the work presented below and
we will, therefore, not pursue the issue further with respect to other
geometries or other classes of material symmetry.
where A is the area of the middle surface of the plate, while the potential
of the transverse loading t(x, y) is
V =− twdxdy
A
b a
Eh Eh
Π= { (u,x + v,y )2 + [(u,y + v,x )2
o o 2(1 − ν )
2 4(1 + ν)
K
−4u,x v,y ] + (w,xx + w,yy )2
2 (6.130)
+(1 − ν)K[w,xy
2
− w,xx w,yy ]
b a
Eh
δΠ = {−[ (u,xx + νv,xy )
o o 1 − ν2
Eh
+ (u,yy + v,xy ) − N,x
T
]δu
2(1 + ν)
Eh Eh (6.131)
−[ (u,xy + v,xx ) + (νu,xy + v,yy ) − N,y
t
]δv
2(1 + ν) 1 − ν2
+[K(w,xxxx + 2w,xxyy + w,yyyy )
T
+M,xx T
+ M,yy − t]δw}dxdy
© 2001 by Chapm
b
+ {[Nx δu]ao + [Nxy δv]ao
o
(a,b) (a,o)
−[2Mxy δw](o,b) + [2Mxy δw](o,o)
In order that (6.131) be satisfied for arbitrary variations δu, δv, δw,
δ(w,x ), and δ(w,y ), all the expressions within the square brackets in
(6.131) must vanish identically, which leads to (6.128) and (6.129), with
Nx = Nxy = Ny = 0; the resultant system governs the thermal bending,
under an arbitrary temperature distribution δT (x, y, z), of a linearly
elastic isotropic rectangular plate. From (6.131) it is readily deduced
that the following (natural) boundary conditions must be satisfied:
(i) On the edges x = 0, a, for 0 ≤ y ≤ b
u is prescribed or Nx = 0
v is prescribed or Nxy = 0
(6.132a)
w is prescribed or Mx,x + 2Mxy,y = 0
w,x is prescribed or Mx = 0
u is prescribed or Nxy = 0
v is prescribed or Ny = 0 (6.132b)
w is prescribed or My,y + 2Mxy,x = 0
(iii) At the corners (0, 0), (a, 0), (0, b), and (a, b)
MT
w=− (x2 + y 2 ) + d + ex + f y (6.135)
2(1 + ν)K
1 T 12 EαT (z)
σxx = σyy = N + 3 MT z − , σxy = 0 (6.136)
h h 1−ν
For the case where the plate has clamped edges, instead of free edges,
it again follows that a simple solution exists. With constant M T , and
© 2001 by Chapm
t ≡ 0, (6.93a) and the boundary conditions are satisfied by taking w = 0.
Then, by virtue of (6.92),
Mx = My = −M T , Mxy = 0 (6.137)
EαT (z)
σxx = σyy = − , σxy = 0 (6.139)
1−ν
If, on the other hand, the middle surface of the plate is free of in-plane
tractions, then Nx = Ny = Nxy = 0 and
1 T EαT (z)
σxx = σyy = N − , σxy = 0 (6.140)
h 1−ν
Chapter 7
© 2001 by Chapm
edges, in which case w = 0 for x = 0, a, 0 ≤ y ≤ b, w = 0 for y = 0, b, 0 ≤
x ≤ a, and the conditions (6.96) apply as well. The analysis proceeds by
expressing the thermal moment and transverse load as double Fourier
sine series of the form
#∞ # ∞
M T
= Mmn sin αm sin βn y
m=1 n=1
#∞ # ∞ (7.2)
t = tmn sin αm sin βn y
m=1 n=1
mπ nπ
with αm = , βn = and
a b
b a
4
(Mmn , tmn ) = (M T , t) sin αm sin βn ydxdy
ab o o
To satisfy the boundary conditions we take the deflection to have the
form
#∞ #∞
w= γmn sin αm sin βn y (7.3)
m=1 n=1
∞ #
# ∞
NT = Nmn sin αm sin βn y (7.7)
m=1 n=1
with
b a
4
Nmn = N T sin αm sin βn ydxdy,
ab o o
The expressions in (7.8) are now substituted into (6.128) with the result
that
−αm Nmn (1 − ν 2 )
a mn = 2 + β2 )
Eh(αm n
(7.9)
−βn Nmn (1 − ν 2 )
bmn =
Eh(αm 2 + β2 )
n
© 2001 by Chapm
Employing (7.7)–(7.8) in (7.6) we compute, for the stress resultants
∞ # ∞
# Eh
− {
N x =
− ν2
(αm amn + νβn bmn )
1
m=1 n=1
+Nmn } sin αm sin βn y
#∞ # ∞
Eh
Ny = − { (ναm amn + βn bmn ) (7.10)
1 − ν2
m=1 n=1
+Nmn } sin αm sin βn y
∞ # ∞
# 1 Eh
N = (βn amn + αnm bmn ) cos αm cos βn y
xy 21+ν
m=1 n=1
If, in lieu of the boundary data (7.6), one assumes that tangential dis-
placements are allowed along each edge, but that normal displacements
are prevented, the relevant boundary conditions are
1 Eh
Nxy = (u,y + v,x ) = 0
21+ν (7.11a)
for x = 0, a; 0 ≤ y ≤ b, and y = 0, b; 0 ≤ x ≤ a
and
u = 0, for x = 0, a; 0 ≤ y ≤ b
(7.11b)
v = 0, for y = 0, b; 0 ≤ x ≤ a
with
ζmn b a T
N̄mn = N cos αm cos βn ydxdy
ab o o
4, m > 0, n > 0
ζmn = 2, m > 0, n = 0 or m = 0, n > 0
1, m = n = 0
while the displacements, chosen so as to identically satisfy the boundary
conditions in (7.11a, b), have the form
∞ # ∞
#
u = āmn sin αm cos βn y
m=1 n=0
∞ # ∞ (7.13)
#
v = b̄mn cos αm sin βn y
m=0 n=1
The same procedure described above for the first set of boundary con-
ditions now leads to
αm N̄mn (1 − ν 2 ) βn N̄mn (1 − ν 2 )
āmn = 2 + β2 )
, b̄ mn = 2 + β2 )
(7.14)
Eh(αm n Eh(αm n
and
∞ # ∞
# Eh
{
x
N =
− ν2
(αm āmn + νβn b̄mn )
1
m=0 n=0
−N̄mn } cos αm cos βn y
#∞ # ∞
Eh
Ny = { (ναm āmn + βn b̄mn ) (7.15)
1 − ν2
m=0 n=0
−N̄mn } cos αm cos βn y
∞ # ∞
# 1 Eh
N = − (βn āmn + αm b̄mn ) sin αm sin βn y
xy
21+ν
m=1 n=1
© 2001 by Chapm
x = 0 and x = a are simply supported, that the plate is symmetric
b b
with respect to the x axis, so that − ≤ y ≤ , and that, for now,
2 2
1
the boundary conditions along y = ± b are arbitrary. As w = 0 along
2
x = 0 and x = a it follows that w,yy = 0 along these edges as well. The
conditions of simple support of the plate along x = 0 and x = a may,
therefore, be expressed as
1 T
w = 0, w,xx = − M
K (7.17)
for x = 0, a; 0 ≤ y ≤ b
with
a2 x x x
Ho (x) = { − 3( )2 + ( )3 }
6K a a a
(7.19)
a2
x x
Ha (x) = { −( ) }3
6K a a
Using (7.19) in (7.18) and substituting the resultant expression for
w(x, y) into (7.1), it follows that
where
t(x, y) 2 M T
F (x, y) = −
K K (7.21)
− 4 {M T (0, y)Ho (x) + M T (a, y)Ha (x)}
and
∂2w
w= = 0, for x = 0, a; 0 ≤ y ≤ b (7.22)
∂x2
We express F in terms of the Fourier series
∞
# mπ
F (x, y) = fm (y) sin αm x; αm = (7.23)
m=1
α
with a
2
fm (y) = F (x, y) sin αm xdx
a o
and take w in the form
∞
#
w(x, y) = Ym (y) sin αm x (7.24)
m=1
where
0, m even
km = 4M T
3
, m odd
aKαm
© 2001 by Chapm
in which case,
∞
#
w= (Am cosh αm y + Dm y sinh αm y + km ) sin αm x (7.29)
m=1,3,···
1 T
w = 0, w,yy = − M
K
(7.30)
1
for y = ± b; 0 ≤ x ≤ a
2
We write
∞
#
MT = 2
Mm sin αm x; mm = Kαm km (7.31)
m=1
and substitute (7.29) into (7.30); after solving for Am and Dm we obtain
∞
#
4M T 1 cosh αm y
w(x, y) = (1 − ) sin αm x (7.32)
aK α 3 1
m=1,3,··· m cosh αm b
2
1
For the case in which the plate is clamped along the edges at y = ± b,
2
and subject to a constant thermal moment M T , it has been noted in
[217] that the deflection, once again, assumes the form in (7.29), but
now with
1 1
Am = −km ( 2 αm b cosh 2 αm b
1
+ sinh αm b)/m
2
(7.34)
1 1 1
Bm = 2 αm b + sinh 2 αm b cosh 2 αm b
m = 1 αm b + sinh 1 αm b cosh 1 αm b
2 2 2
References for the thermal bending of an isotropic, elastic rectangular
plate, under other combinations of edge conditions, may be found in
[217].
Next, we consider the problem of thermal bending of an isotropic
annular plate; we assume that the plate is subjected to a transverse
loading t = t(r, θ) and a general temperature variation δT (r, θ, z). For
this situation, ignoring for now the effect on deflections of loads in the
plane of the plate, the relevant equation is the first partial differential
equation in (6.112) with Φ ≡ 0, i.e.,
2 1
K(w,rrrr + w,rrr − 2 w,rr
r r
2 1 2
+ 2 w,rrθθ + 3 w,r − 3 w,rθθ
r r r
(7.35)
1 4
+ 4 w,θθθθ + 4 w,θθ )
r r
∗ 1 ∗ 1 ∗
= t − (MT,rr + MT,r + 2 MT,θθ )
r r
h/2
Eα
with MT∗ = δT (r, θ, z)zdz being the thermal moment.
1 − ν −h/2
Equation (7.35) holds for a ≤ r < b, 0 ≤ θ < 2π. The associated
clamped, simply supported, and free edge boundary conditions are given,
respectively, by
∂w
(i) w = 0 and = 0, at r = a, b if the edges are clamped
∂r
1 1
(ii) w = 0 and K[w,rr + ν( 2 w,θθ + w,r )] + MT∗ = 0, at r = a, b if the
r r
edges are simply supported
© 2001 by Chapm
1 1
(iii) K[w,rr + ν( w,θθ + w,r )] + MT∗ = 0 and
r2 r
1 1 1 1 ∗
K[(w,rr + w,r + 2 w,θθ ),r +(1−ν)( 2 w,θθr − 3 w,θθ )]+MT,r = 0,
r r r r
at r = a, b if both edges are free.
Also, for isotropic response, the bending moments in polar coordinates
Eα
are given by (6.110), the resultant forces by (6.108), with NT∗ =
1 −ν
h/2
δT (r, θ, z)dz and the stresses may be expressed by
−h/2
1 12z Eα
σrr = (Nr + NT∗ ) + 3 (Mr + MT∗ ) − δT
h h 1−ν
1 12z Eα
σθθ = (Nθ + NT∗ ) + 3 (Mθ + MT∗ ) − δT (7.36)
h h 1−ν
σrθ = 1 Nrθ + 12z Mrθ
h h3
The simplest case of thermal bending with respect to an annular plate
is that of axisymmetric bending in which it is assumed that the loading
and boundary conditions are independent of the angular coordinate θ.
If, in addition, t ≡ 0, then (7.35) reduces to
1 2 ∗
4 w = − MT , a < r < b (7.37)
K
h/2
∗ Eα
where w = w(r), MT = δT (r, z)zdz,
1 − ν −h/2
d4 2 d3 1 d2 1 d
4 = 4 +
− + 3
dr r dr3 r2 dr2 r dr
2 d2
= +
1 d
≡
1 d d
(r )
dr2 r dr r dr dr
The general solution of (7.37) is easily computed to be
r r
w = C1 + C2 r2 + C3 1n + C4 r2 1n
a a
b r (7.38)
1 1 ∗
+ ( MT (r)rdr)dr
r r a K
with the Ci , i = 1, · · · , 4, arbitrary constants of integration. For the
problem at hand, a straightforward computation based on (7.38) yields
the following expressions for the relevant moments and shear force re-
sultant:
C3
Mr = −K{2(1 + ν)C2 − (1 − ν) 2 + (3 + ν)C4
r
r 1−ν r ∗
+2(1 + ν)C4 1n } − 2 MT (r)rdr
a r a (7.39)
Mrθ = 0
Q ≡ M + Mr − Mθ = −4K C4
r r,r
r 4
For the case of a solid plate, in which a = 0, the constants C3 and C4
in (7.38) must vanish so that Mr and Qr remain finite at r = 0; if the
solid plate is clamped along its edge at r = b, then it follows from (7.38)
and the fact that w = w,r = 0 at r = b that
b
1
C1 = −b C2 = −
2
MT∗ (r)rdr (7.40)
2K 0
while, if the edge at r = b is simply supported,
b
1−ν
C1 = −b C2 =
2
MT∗ (r)rdr (7.41)
2(1 + ν)K 0
When (δT ),θ = 0 a solution w = w(r, θ) must be obtained for (7.35);
for simplicity we again set t ≡ 0; such problems have, e.g., been con-
sidered by Forray and Newman [225] for the special case in which the
thermal gradient is assumed to vary linearly through the thickness of
the plate. Specifically, it is assumed in [225] that the thermal moment
MT∗ may be expressed in the form
∞ #
# ∞ ∞ #
# ∞
MT∗ = Akm rk cos mθ + Bkm rk sin mθ (7.42)
m=0 k=0 m=1 k=0
© 2001 by Chapm
and Td the temperature difference between the upper and lower faces of
the plate. Using the definition of MT∗ we then easily compute that
1+ν
MT∗ = αTd (r, θ) (7.44)
hK
while (7.35) becomes (with t ≡ 0)
1+ν
4 w = α 2 Td (7.45)
hK
The solution to (7.45) consists of the sum of the general solution of
1+ν
4 wg = 0 and a particular solution of 2 wp = αTd .
hK
In fact, the general solution to (7.45) can be shown, as in [225], to
have the form
© 2001 by Chapm
h h b b
so that T (x, y, ) = T1 , T (x, y, − ) = T2 , for 0 ≤ x ≤ a, − ≤ y ≤ ,
2 2 d 2
where T1 and T2 are, respectively, the constant temperatures at the top
and bottom of the plate; thus the temperature is assumed to remain
constant in any plane which is parallel to the x, y plane. The edge
conditions are given by
b b
w = 0, Mx = 0; on x = 0, a, for − ≤y≤ (7.51a)
2 2
and
b
w = 0, w,y = 0; on y = ± , for 0 ≤ x ≤ a (7.51b)
2
where Mx is given by (6.101). Actually, Misra [226] writes the term M̃T1
h/2
1
in (6.102) in the form M̃T = β̄1 MT , with MT = δT zdz, so that
−h/2
The superposed bar over the β1 in (7.52) does not appear in [226] and has
been placed there so as not to confuse this parameter with a hygroscopic
coefficient. With the definition of MT , as given above, and (7.50) it is
easily seen that
1 2
MT = h (T1 − T2 ) ≡ k (7.53)
12
which has the Fourier representation
∞
#
4k 1 mπ
MT = sin x, (7.54)
π m=1,3,5,...
m a
where
4kπ β̄1
P = (7.56)
a2
From (7.54) it follows that MT = 0 along the edges at x = 0 and x = a.
Then, by virtue of (7.51a), it follows that both w and w,yy must vanish
along x = 0 and x = a. However, as Mx = 0 along x = 0 and x = a, it
would follow from (6.101) that w,xx = 0 along x = 0 and x = a only if
MT = 0 along these edges, which it does not—the Fourier representation
not withstanding! Thus, the edge conditions in (7.51a), which in [226]
are now written in the form
w = 0, w,xx = 0, on x = 0, a
b b (7.57)
for − ≤ y ≤
2 2
are open to suspicion, as is the remainder of the solution presented below.
A solution of the homogeneous equation associated with (7.55) which is
compatible with the edge conditions (7.57) is sought in [226] in the form
∞
# mπ
w= Ym (y) sin x (7.58)
m=1,3,..
a
© 2001 by Chapm
in which case, the complete solution of (7.55) assumes the form
∞
# mP
w= { 4 D
+ Am cos hpm y cos qm y
m=1,3,5...
αm 11 (7.64)
+Bm sinh pm y sin qm y} sin αm x
The edge conditions (7.57) are automatically satisfied by (7.64), while
those in (7.51b) are satisfied if and only if Am , Bm are connected by the
relations
mp bpm bqm
4 D
+ Am cosh cos
αm 11 2 2
(7.65a)
bpm bqm
+Bm sinh sin =0
2 2
bpm bqm
Am (pm tan h − qm tan )
2 2
(7.65b)
bqm bpm
+Bm (pm tan + qm tanh )=0
2 2
These relations may be solved for Am , Bm (see [226] for the details),
which are then substituted back into (7.64) so as to yield the deflection
at any point of the plate. The expression obtained for w(x, y) can also
be employed in (6.101) so as to compute the thermally induced moments
at any point in the plate, e.g.,
∞
#
Mx = D11 2
αm {Am cosh pm y cos qm y
m=1,3,5,..
+Bm sinh pm y sin qm y} sin αm x
∞
#
−D12 [Am {(p2m − qm
2
) cosh pm y cos qm y
m=1,3,5,...
−2pm qm sinh pm y sin qm y} (7.66)
+Bm {(p2m − qm
2
) sinh pm y sin qm y
∞
#
4β̄1 k 1
− sin αm x
a α
m=1,3,5,... m
with analogous expressions for My and Mxy . Finally, the deflection at
a
the center of the plate, i.e., at x = , y = 0, is computed in [226] to be
2
∞
# mP mπ
w= ( 4 D
+ Am ) sin (7.67)
m=1,3,5...
αm 11 2
The expressions for the moments, e.g., (7.66) and the final result (7.67)
for the deflection at the center of the plate are subject to the criticism
(because of the relevance of the boundary conditions (7.57)) which has
been levied above.
An alternative approach to solving problems of thermal deflection for
plates that has been used extensively in the literature is based on the
concept of an influence function and usually goes under the title of May-
sel’s method; this approach is actually an extension of Betti’s reciprocal
theorem to thermoelastic problems, and excellent treatments have ap-
peared in several places in the literature, e.g., in Nowacki [222] and in
Tauchert [217]. In what follows, we will adhere closely to the presenta-
tion in [217] and will assume that the plate exhibits isotropic response;
we will also take t ≡ 0, so that, in either rectangular or polar coordinates,
the relevant partial differential equation is given by
If the plate occupies the domain A in the x, y plane when in its unde-
flected state, and w∗ (ξ, y; x, y) is the Green’s function for the operator
K2 then it is easily shown (i.e. [217] or [222]) that
w(x, y) = − M T (ξ, η) 2 w∗ (ξ, η; x, y)dA(ξ, η) (7.69)
A
∂2 ∂2
2 = 2
+ 2
∂ξ ∂η
© 2001 by Chapm
shape and assigned support conditions. As an alternative to (7.69) one
may use the form obtained by employing Green’s formula, i.e.,
w(x, y) = − w∗ (ξ, η; x, y) 2 M T (ξ, η)dA(ξ, η)
A (7.70)
∂w∗ ∂M T
− MT
− w∗ ds
∂A ∂n ∂n
where n, s denote, respectively, the directions that are normal and tan-
gential to the plate boundary ∂A. If 2 M T = 0, such as for the case in
which M T is constant, then (7.70) reduces to
∂w∗ ∂M T
w(x, y) = − MT − w∗ ds (7.71)
∂A ∂n ∂n
and if the plate is simply supported, so that w∗ = 0 along ∂A, then
∂w∗
w(x, y) = − M T (ξ, η) (ξ, η; x, y)ds (7.72)
∂A ∂n
Of course, if 2 M T = 0 in A, and the plate is clamped along ∂A, then
∂w∗
w∗ = = 0 along ∂A in which case w ≡ 0 throughout the plate.
∂n
As a first example of the influence function method we consider the
simply supported rectangular plate which is depicted in Fig. 7.2. We
assume that the thermal moment is nonzero within an arbitrary region
AT of the plate, while M T = 0 is the complement of this region. It
is easily shown that the deflection w∗ at an arbitrary point (ξ, η) of
the same simply supported plate subject to a concentrated unit force at
(x, y) is
∞ ∞
4 # # sin αm ξ sin βn η sin αm x sin βn y
w∗ (ξ, η; x, y) = 2 + β 2 )2
(7.73)
abK m=1 n=1 (αm n
mπ nπ
where αm = , βn = , provided M T vanishes along the edges of the
a b
plate. Substituting (7.73) into (7.69), and carrying out an elementary
computation, we are led to the following expression for the deflection:
∞ ∞
4 # # sin αm x sin βn y
w(x, y) = 2 + β 2 )2
abK m=1 n=1 (αm n
(7.74)
× M T (ξ, η) sin αm ξ sin βn ydξdη
AT
If the thermal moment is constant over the entire plate, say, M T = M ,
then (7.74) formally reduces to
∞
# ∞
#
16M sin αm x sin βn y
w(x, y) = 2 + β2 )
(7.75)
abK α
m=1,3,... n=1,3,... m n
β (αm n
but, once again, such a solution is subject to the criticisms raised earlier
as now M T does not vanish along the edges of the plate.
As a second example, we consider the application of Maysel’s relation
(7.69) to the thermal deflection of a solid circular plate of radius b; in
this case (7.69) assumes the following form in terms of polar coordinates:
2π b
w(r, θ) = − M T (ρ, ψ) 2 w∗ (ρ, ψ; N, θ)ρdρdψ (7.76)
0 0
where
∂2 1 ∂ 1 ∂2
2 = 2
+ + 2
∂ρ ρ ∂ρ ρ ∂ψ 2
In lieu of (7.76) we may write, in analogy with (7.70), that
2π b
w(r, θ) = − w∗ (ρ, ψ; r, θ) 2 M T (ρ, ψ)ρdρdψ
0 0
2π
∂w∗ (b, ψ; r, θ)
− M T (b, ψ) (7.77)
∂ρ
0
T
∂M (b, ψ)
−w∗ (b, ψ; r, θ) bdψ
∂ρ
∂w∗
If the circular plate is clamped along its edge at r = b then w∗ = =
∂ρ
0, for r = b, 0 ≤ θ < 2π, and (7.77) reduces to
2π b
w(r, θ) = − w∗ (ρ, ψ; r, θ) 2 M T (ρ, ψ)ρdρdψ (7.78)
0 0
© 2001 by Chapm
with ρ = ρ/b and r = r/b. Various forms of the Green’s function are
available for the case of the simply supported solid circular isotropic
plate but, as noted in [217] the expressions tend to be quite complex.
Before concluding this brief description of thermal bending of plates
(and moving on to describe some problems of thermal buckling within
the context of small deflection theory) we want to note some approxi-
mate methods that have been employed to deal with problems of thermal
flexure of plates; two of the better-known techniques are the Rayleigh-
Ritz and Galerkin procedures. In the Rayleigh-Ritz method, the dis-
placement field w is approximated by functions which contain a finite
number of independent coefficients. The functions employed are chosen
so as to satisfy the kinematic boundary conditions, but they do not have
to satisfy the static boundary conditions. The unknown coefficients in
the assumed solution are then determined by employing the principle of
minimum potential energy. For the problem of thermal flexure we may,
in particular, represent the transverse displacement w(x, y) in the form
#
M #
N
w(x, y) = cmn φmn (x, y) (7.80)
m=1 n=1
It is assumed here that the φmn (x, y) satisfy the boundary conditions
which involve w, w,x , and w,y . The assumed form of the solution (7.80)
is then substituted into the expression for the potential energy Π which,
for a problem of (purely) thermal flexure of a homogeneous isotropic
plate, is given by the following reduced form of (6.130):
b a
K
Π= (w,xx + w,yy )2
0 0 2
+(1 − ν)K(w,xy2
− w,xx w,yy ) (7.81)
#
M #
N
mπx 2nπy
w= cmn sin 1 − cos (7.83)
m=1 n=1
a a
for the transverse deflection, then we satisfy the kinematic boundary con-
ditions for this problem but not the static boundary condition Mx = 0
along the edges x = 0 and x = a, 0 ≤ y ≤ a. Retaining only the
term corresponding to m = 1, n = 1 in (7.83), it is easily verified
that the Rayleigh-Ritz method yields an approximation to w(x, y) in
1
which the maximum deflection, which occurs at x = y = a, is given
2
by 0.0191a2 M T /K. As noted in [217], two-and three-term approxima-
tions using (7.83), yield maximum deflections of 0.0144a2 M T /K and
0.0157a2 M T /k, respectively, while the “exact” value of the maximum
deflection in this case is given (approximately) by 0.0158a2 M T /K.
To implement the Galerkin procedure, we work directly with the rele-
vant differential equation instead of with the associated potential energy;
the equation, for the problem of thermal flexure of an isotropic, homo-
geneous thin plate is just (6.93a) with Φ ≡ 0, i.e.,
K∆2 w + ∆M T − t = 0 (7.84)
© 2001 by Chapm
Remarks: If one incorporates boundary residuals into the Galerkin
procedure it is possible to relax the constraint that the φmn (x, y) satisfy
the static as well as the kinematic boundary conditions. The first vari-
ation δΠ of the total potential energy Π is, for purely thermal flexure
problems, given by the following reduced (and modified) form of (6.131)
b a
δΠ = (K∆2 w + ∆M T − t)δwdxdy (7.87)
0 0
b $ x=a
+ (Mx,x + 2Mxy,y − K̄x )δw x=0
0
x=a
∂w
− (Mx − M̄x )δ dy
∂x x=0
a
$ y=b
+ (My,y + 2Mxy,x − K̄y )δw y=0
0
y=b
∂w
− (My − M̄y )δ dx
∂w y=0
$ x=a,y=b
− (2Mxy − R̄xy )δw x=0,y=b
$ x=a,y=0
+ (2Mxy − R̄xy )δw x=0,y=0 = 0
#
M #
N
δw = δcmn φmn (x, y) (7.88)
m=1 n=1
which is, of course, equivalent to (7.85). One also obtains (7.89) if certain
of the boundary conditions, as noted in [217], are static; however, these
static conditions must be satisfied identically by (7.80). Suppose we
consider, as an example, the case treated earlier in this section by the
Raleigh-Ritz method, i.e., a square plate clamped along two parallel
edges and simply supported along the other two, and subjected to a
uniform thermal moment M T . For this problem, the static boundary
condition Mx = 0 is not satisfied, as already noted, by the assumed form
(7.83) of the solution. The condition (7.87) leads, in this case, to the
following system of equations for the coefficients cmn :
b a
K∆2 w · φmn (x, y)dxdy
0 0
b$ x=a (7.90)
+ (Kw,xx + νKw,yy + M T )φmn,x x=0 dy = 0
0
(m = 1, 2, ..., M ; n = 1, 2, ..., N )
© 2001 by Chapm
representing the first buckling mode. To illustrate the procedure delin-
eated above, we will begin our discussion by presenting three examples
that have been highlighted in Boley and Weiner [221] for isotropic plates
and a rectilinear geometry; we will then proceed to examples involving
circular plates as well as problems for plates with orthotropic material
symmetry.
The first case treated in [221] concerns the buckling of plates sub-
jected to heat conduction (but no transverse loads) with their edges
unrestrained in the plane. We are reminded in [221] of the basic fact
that if the ends of a column are free to displace axially, and the column
is free from axial loads, then the column cannot buckle no matter what
the temperature distribution may be; this is clearly not the case with
plates. Because the plate is assumed to be free of external tractions in
its plane, equilibrium relations of the form
Nx dy = 0 (7.92)
must be satisfied in which the integration extends across the entire plate
along a line given by x = const; a relation such as (7.92) cannot hold
unless Nx > 0 along part of this line while Nx < 0 along its complement,
thus leading to the conclusion that for this class of problems, compres-
sive stresses will always occur in the plane of the plate. A very well
known example of the type referenced above occurs in the often quoted
paper of Gossard, Seide, and Roberts [216], which will be discussed in
some detail later in this chapter; although our focus, with respect to
the discussion of the work in [216], will be on postbuckling behavior, it
should be clear that the buckling problem described, e.g., by the system
(6.93a,b), within the context of small deflection theory is mathemati-
cally isomorphic to the initial buckling problem for the full non-linear
system. Indeed, some specific initial buckling problems for such systems
will be discussed at the end of this section.
A second class of thermal buckling problems in the realm of small
deflection theory, which is discussed in [221] and which is mathematically
similar to the first class of problems, concerns the buckling of plates
which are subjected to heat and loads in their plane with, once again,
their edges unrestrained in the plane of the plate. As an example, we
consider the plate strip of Fig. 7.3, which is loaded at its ends by a
uniformly distributed stress σ0 ; the strip, of width b, is reinforced along
its edges at y = 0, y = b by longitudinals of area A which act as a heat
sink, thus causing the temperature to be higher along the center of the
plate than near its edges. For illustration purposes, the temperature will
be assumed to be uniform across the thickness of the plate and of the
form
2πy
δT (x, y) = c0 − c1 cos (7.93)
b
in the plane of the plate where c0 , c1 are constants which may be chosen
so as to fit empirical data. We consider a single panel of the strip, as
depicted in Fig. 7.3., which extends from x = 0 to x = a; it is assumed
that this panel is at a large enough distance from the ends of the strip so
that the stresses can be taken to be independent of x. Also, we assume
that w = 0 along the line segments x = 0, x = a, for 0 ≤ y ≤ b. As
we have already indicated in the discussion of the procedure for solving
(7.91a,b), the first step in the solution of the problem at hand consists of
determining a stress function Φ from (7.91b) and the pertinent boundary
conditions; these boundary conditions are as follows:
u(0, y) = 0, u(a, y) = u0 , 0 ≤ y ≤ b
(7.94)
v(x, 0) = 0, v(x, b) = v0 , 0 ≤ x ≤ a
where u0 and v0 are constants which are chosen so that
b b
Nx (0, y)dy = Nx (a, y)dy = bhσ0 (7.95)
0 0
© 2001 by Chapm
x
u = (σ0 + αEc0 )
E (7.98c)
v = (−νσ + αEc ) y − (1 + ν)αc1 b sin 2πy
0 0
E 2π b
It is easily seen that (7.98a,b,c) satisfy all the boundary conditions
delineated above provided
a
u0 = (σ0 + αEc0 )
E (7.99)
v0 = (−νσ0 + αEc0 ) b
E
The next step for the problem at hand involves the computation of the
transverse deflection w(x, y) and the corresponding critical combination
of the temperature levels and the applied load. Using the fact that t ≡ 0
in (7.91a), and that M T = 0 for the temperature distribution defined
by (7.93), it is easily seen that the use of (7.98a) in (7.91a) reduces this
equation to
h cos 2πy
∆2 w = σ0 + αEc1 w,xx (7.100)
K b
We consider (7.100) with the conditions relevant for simply supported
edges, namely,
w = w,xx = 0; x = 0, a; 0 ≤ y ≤ b
(7.101)
w = w,yy = 0; y = 0, b; 0 ≤ x ≤ a
for which two special cases are of interest: If c1 = 0 then only the edge
stress distribution σo acts to buckle the panel and only the diagonal
terms in (7.105) survive. In this case, the critical value of σo is given by
the same expression that has already been noted in Chapter 2, namely,
π2 E h
(σo )cr |c1 =0 = −K ( )2 (7.106a)
12(1 − ν 2 ) b
© 2001 by Chapm
seeking the smallest root T1 = Tcr of the resulting infinite determinant;
approximations to Tcr may be obtained from (7.105), with σo = 0, by
retaining only a finite number of rows and columns of the determinant.
By retaining only the element in the first row and the first column of
(7.105), and setting σo = 0, we obtain
αE π2 E h
( )Tcr |σ0 =0 ≈ k ( )2 = −(σ0 )cr |c1 =0 (7.107)
2 12(1 − ν ) b
2
© 2001 by Chapm
Once again, as M T does not vanish along the edges of the plate, such a
solution technique is subject to the criticisms raised earlier. In Fig. 7.6,
we show a nondimensional plot of the deflection (as computed in [221] at
the center of the rectangular plate) against the temperature parameter
a
N T /(N T )cr for various aspect ratios ; here (N T )cr , the value of N T
b
at which buckling occurs is computed to be
a2 π 2
(N T )cr = (1 + ) (7.118)
b2 a2
The (nondimensional) variation of the bending moment Mx in a square
plate is shown in Figs. 7.7 and 7.8 for two different values of the “tem-
perature level” N T /(N T )cr ; such plots are sufficient to determine both
Mx and My throughout the entire plate because of the double symme-
try exhibited by the plates; the results depicted in Figs. 7.7 and 7.8
indicate that the maximum bending moment occurs at the center of the
plate. Finally, in Fig. 7.9 we show, for various aspect ratios, the varia-
tions in Mx at the center of the plate, with the temperature parameter
N T /(N T )cr . All of the results depicted in Figs. 7.6–7.9 are valid only
for values of N T that are sufficiently close to (N T )cr because of the small
deflection assumption employed in their derivation.
For the special case (in rectangular coordinates) in which (6.94) re-
duces to
δT (x, y, z) = To (x, y) (7.119)
so that, by virtue of (6.95),
αEh
NT = To (x, y); M T = 0 (7.120)
1−ν
Equations (6.93a) and (6.93b) for the case of isotropic response, reduce
to
2 Φ + αEhTo = 0 (7.121b)
1 1
2 w = (Nx w,xx + 2Nxy w,xy + Ny w,yy ) ≡ Nij w,ij (7.122)
K K
must be solved subject to the specification of boundary conditions on
w. As an example, we consider a rectangular plate with 0 ≤ x ≤ a, 0 ≤
y ≤ b, which is simply supported along its edges; for this case M T = 0
along the edges of the plate so our previous criticisms do not apply.
One approach to dealing with (7.22) is similar to Maysel’s method,
which was discussed in connection with the thermal bending of plates.
We introduce the Green’s function w∗ , which satisfies the equation
1
w∗ (x, y; ξ1 , ξ2 ) = δ(x − ξ1 )δ(y − ξ2 ) (7.123)
K
in the sense of distributions and the same boundary conditions as w.
Combining (7.122) and (7.123) we obtain
a b
w(x, y) = w∗ (x, y; ξ1 , ξ2 )Nij (ξ1 , ξ2 )w,ij (ξ1 , ξ2 )dξ1 dξ2 (7.124)
o o
Now, for the rectangular plate described above which is simply supported
on all four edges, w∗ (x, y, ξ1 , ξ2 ) is given by (7.73) with ξ → ξ1 , η → ξ2 .
If we assume that the solution of the integral equation (7.125) can be
represented in series form as
∞
#
w(x, y) = Aik sin αi sin βk y, (7.126)
i,k
nπ mπ
αn = , βm = (thus automatically satisfying the boundary condi-
a b
tions of simple support) then by substituting the series representations
© 2001 by Chapm
for w and w∗ into (7.125), and performing elementary computations, we
are led to an infinite system of linear equations for the coefficients Aik
in (7.126) of the form
4 #
Aik + Amn Gnimk =0
abk(αi2 + βk2 ) n,m (7.127)
(i, k = 1, 2, · · · , ∞)
where
Gnimk = αi2 animk + 2αi βk cnimk + βk2 bnimk
with
a b
a = N11 (ξ1 , ξ2 ) sin αi ξ1 sin αn ξ1 sin βk ξ2 sin βm ξ2 dξ1 dξ2
nimk
o o
a b
bnimk = N22 (ξ1 , ξ2 )) sin αi ξ1 sin αn ξ1 sin βk ξ2 sin βm ξ2 dξ1 dξ2
o o
a b
cnimk = N12 (ξ1 , ξ2 ) cos αk ξ1 sin αn ξ1 cos βk ξ2 sin βm ξ2 dξ1 dξ2
o o
The condition for buckling of the plate is, of course, that the (infinite)
determinant of the system of equations (7.117) be zero. A related (clas-
sical) treatment of the thermal buckling of a simply supported isotropic
rectangular plate which employs the Rayleigh-Ritz procedure may be
found in Klosner and Forray [227]; in [227], the temperature distribu-
tion is of the form (7.119) and, in fact, is assumed to be symmetrical
about the centerlines of the plate so that it is representable in the form
nπx mπy
of a double Fourier series in the functions cos cos (the plate
a b
in [227] has length 2a and width 2b).
There are many excellent treatments of the problem of general thermal
deflections of an isotropic elastic circular plate, one of which may be
found in [228]. We assume that the plate has radius b > 0 and is
subjected to the radially symmetric subcase of (6.113), namely,
© 2001 by Chapm
the superscripted o denoting, of course, that we are computing the mid-
dle surface stress field distribution. Because Nr , Nθ , Nrθ , and T1 are
independent of θ, it is easily seen that (7.133) for the transverse deflec-
tion w = w(r) may be rewritten in the form:
d d 1 d dw
(r [ {r }])
dr dr r dr dr
d dT1
= −Γ (r )
dr dr
2
h
(7.135)
αE r d w
+ P0 r − T0 (λ)λdλ
K r 0 dr2
r
αE dw
+ P0 + 2 T0 (λ)λdλ − αET0
r 0 dr
with the constant P0 given by
b
αE
P0 = T0 (λ)λdλ − P (7.136)
b2 0
with the t0j , t1j real constants and m, n arbitrary positive integers. A
solution of (7.137) is now sought in the form of a power series. Substi-
tuting (7.138) in (7.137) and setting
t0j
cj = , dj = jt1j
j+2
we obtain
d 3 w d2 w % #
m
+ 2 + cj (rj − bj )
dr3 dr j=0
(7.139)
1 & dw #
n
hP
+ − 2 = dj rj−1
K r dr j=1
Setting
dw
u(r) = dr
hP # j
m
b0 = K −
cj b
j=1
© 2001 by Chapm
where ξi contains the parameters P, K, h, b and some (or all) of the t0j ,
while δi contains these parameters as well as some (or all) of the t1j .
Thus, (7.143) may be rewritten in the form
where
∞
# ∞
#
w0 (r) = ξj rj , w1 (r) = δj r j (7.145)
j=0 j=0
and
ν
K[w (b) + w (b)] + ΓT1 (b) = 0
b (7.146b)
w(b) = 0
It will be assumed that w0 (b) = 0; as noted in [228], if w0 (b) = 0 then b
is the radius of the clamped plate with given temperature distribution
t0 (r) for which P is a critical buckling load.
In the special case in which m = 0, T0 (r) reduces, in light of (7.138),
to a constant while (7.140) becomes
#
n−1
d2 u 1 du hP 1
+ + − 2 u= dj+1 rj (7.149)
dr2 r dr K r j=0
Using (7.141) again, and working with the special case of the recurrence
relation (7.142) for m = 0, it is not difficult to show that the transverse
deflection assumes the form
+ * 6
hP
w(r) = λ0 J0 r + w10 (r) + κ (7.150)
K
with J0 the Bessel function of the first kind of order zero and w10 (r) the
form assumed by w1 (r) for m = 0. Applying the boundary conditions
(7.146a) we now find as the expression for the transverse deflection of
the circular, isotropic plate clamped along its edge at r = b
+ * 6
w10 (b) hP
w(r) = * + * 6 J0 r
h hP K
J0 b
K K (7.151)
+ * 6
hP
−J0 b + w10 (r) − w10 (b)
K
© 2001 by Chapm
so that M T = 0. The buckling problem (in rectangular coordinates),
either assuming small deflection theory or focusing on the initial buckling
problem, then takes the form (see (6.93a,b))
∆2 Φ = −αEh∆Tm
1 (7.154)
∆2 w = (Φ,yy w,xx − 2Φ,xy w,xy + Φ,xx w,yy )
K
if we again assume that t ≡ 0. Following the stability analysis in [229], we
let T0 represent the maximum value (upper bound) of Tm on the domain
of the plate so that 0 ≤ Tm ≤ T0 and we introduce the dimensionless
temperature parameter T = Tm /T0 so that 0 ≤ T ≤ 1. If b represents
a characteristic length associated with the plate, then in rectangular
coordinates, we may introduce the dimensionless variables ξ = x/b, η =
y/b, in which case (7.154) would assume the form
∆2 φ = −∆T
(7.155)
∆2 w = λ(φ,ηη w,ξξ − 2φ,ξη w,ξη + φ,ξξ w,ηη )
where
φ = Φ/b2 αEhT0 (7.156)
while, by virtue of the definition of K,
2
b
λ = 12(1 − ν 2 ) αT0 (7.157)
h
∂2 ∂2
and, of course, ∆ = + . Following the discussion in [229],
∂ξ 2 ∂η 2
the edge(s) of the plate are assumed to be free of applied forces and
moments so that the buckling (stability) problem consists of finding the
temperature parameter T , which minimizes the value of λ in (7.155);
this may be achieved, e.g., by following the Raleigh-Ritz procedure and
considering the expression
Λ = ΛN /ΛD (7.158)
with
ΛN = (w,ξξ + w,ηη )2 − 2(1 − ν)[w,ξξ w,ηη − w,ξη2
] dξdη
A
ΛD = 2φ,ξη w,ξ w,η − φ,ηη w,ξ
2
− φ,ξξ w,η
2
dξdη
A
where A is the domain occupied by the midplane of the plate. As λ ≤ Λ
for all admissible functions, w = w∗ (ξ, η), which satisfy the edge support
conditions, and, for any such w∗ , ΛN remains constant while ΛD is a
functional of φ, the buckling problem may be cast in the following form:
compute the
∗ ∗ ∗2 ∗2 ∗2
max∗
2φ,ξη w,ξ w,η − φ,ηη w,ξ − φ,ξξ w,η − φ,ξξ w,η dξdη
w
A
1 1 dTm
G (r) + G (r) − 2 G = −Eαh (7.159)
r r dr
and
1
Nr = G, Nθ = G (r) (7.160)
r
Using the plate radius b as a reference length and introducing the nondi-
mensional coordinate ρ = r/b we may rewrite (7.159) in the form
1 1 dT
g (ρ) + g (ρ) − 2 g = − (7.161)
ρ ρ dρ
with
g(ρ) = G(ρ)/EαhbT0 (7.162)
The solution of (7.161) which satisfies the boundary condition Nr (b) = 0
and also satisfies the condition that Nr (0) and Nθ (0) remain bounded,
is 1
1 ρ
g(ρ) = − T (λ)λdλ + ρ T (λ)λdλ (7.163)
ρ 0 0
© 2001 by Chapm
the problem at hand, (7.158) can be written in the following form
2π
[(∆w)2 − (1 − ν)L(w, w)]ρdρdθ
0
2
0
Λ=
2π 2 (7.164)
1 ∂w 1 dg ∂w
− g(ρ) + 2 ρdρdθ
0 0 ρ ∂ρ ρ dρ ∂θ
where
∂ 2 w 1 ∂w 1 ∂2w
∆w = + +
∂ρ2 ρ ∂ρ ρ2 ∂θ2
and
1 ∂w 1 ∂2w ∂2w
L(w, w) = 2 + 2 2
ρ ∂ρ ρ ∂θ ∂ρ2
2
2 (7.165)
1 ∂ w 1 ∂w
−2 − 2
ρ ∂ρ∂θ ρ ∂θ
Following the analysis in [229] we now introduce the admissible functions
w = w∗ (ρ, θ) = wm (ρ) cos mθ, m = 0, 1, 2... (7.166)
which are subject to the boundary conditions
1 ∂mrθ
mr (b) = 0, qr + =0 (7.167)
b ∂θ r=b
where qr , mr , and mrθ are (in r, θ coordinates) the z-direction shear
force, bending moment, and twisting moment, respectively, per unit
length along the edge at r = b. With respect to the set of admissible
functions in (7.166), the boundary conditions (7.167) become
2
d wm dwm
+ν − m wm
2
=0 (7.168a)
dρ2 dρ ρ=1
d3 wm d2 wm dwm
+ 2 + [1 + m2 (2 − ν)]
dρ3 dρ2 dρ (7.168b)
+m (3 − ν)wm ρ=1 = 0
3
© 2001 by Chapm
From (7.163) we compute
ρ 1
dg 1
= 2 T (λ)λdλ − T (ρ) + T (λ)λdλ (7.172)
dρ ρ 0 0
dwm m2 dg 2
Λ0 = g(ρ) + w dρ
dρ ρ dρ m
0
1 2
dwm m2 dg 2
− g(ρ) + w dρ
µ dρ ρ dρ m
and effect the same splitting in both (7.163) and (7.172), i.e., if we set
µ
T (λ)λdλ
I =
0
1 (7.174)
II = T (λ)λdλ
µ
then
1 ρ
g = − T (λ)λdλ + ρI + ρII
ρ 0 , 0≤ρ≤µ (7.175)
ρ
dg = 1 T (λ)λdλ − T (ρ) + I + II
dρ ρ2 0
and
1 ρ 1
g = −ρ T (λ)λdλ + ρII − −ρ I
ρ
µ ρ , µ≤ρ≤1
dg 1 1
= 2 T (λ)λdλ − T (ρ) + + 1 I + II
dρ ρ µ ρ2
(7.176)
If we begin by first considering a radially symmetric buckling mode
then, by (7.166), m = 0 and ΛD reduces to
' µ 2 1
'
' dw0
ΛD ' =− g(ρ) dρ − g(ρ)
' 0 dρ µ
m=0 1 2 (7.177)
dw0
− g(ρ) dρ
µ dρ
An analogous result holds for −g(ρ) with ρ in the interval [0, µ], namely,
we have
ρ
g(ρ) = − (1 − µ2 ), 0 ≤ ρ ≤ µ (7.181)
2
By introducing (7.180) and (7.181) into (7.177) we see that the function
to be maximized becomes
' µ 2
' 1 dwm
ΛD ' = (1 − µ2 ) ρ dρ
m=0 2 0 dρ
2 (7.182)
1 2 1 1 dwm
+ µ −ρ dρ
2 µ ρ dρ
© 2001 by Chapm
As 0 ≤ ρ ≤ 1, and 0 < µ < 1, the expression on the ' right-hand side
'
of (7.182) is positive; for any admissible function ΛD ' may be com-
m=0
puted
' and, moreover, the value of µ can be determined which maximizes
'
ΛD ' .
m=0
If we retain the same temperature field, i.e., T = 1 for 0 ≤ ρ ≤ µ and
T ≡ 0 for µ ≤ ρ ≤ 1 but consider buckling modes which are not radially
symmetric (m = 0) the situation described above changes. In this case,
it follows from (7.175), (7.176) that
1
dg − 2 (1− µ ), 0≤ρ≤µ
2
= 1 1 (7.183)
dρ µ2 + 1 , µ ≤ ρ ≤ 1
2 ρ2
dg dg
so that, as per (7.171a), ≥ 0, for µ ≤ ρ ≤ 1 while ≤ 0, for
dρ dρ
0 ≤ ρ ≤ µ. Using (7.180), (7.181), and (7.183), it follows that the
function to be maximized is given by
µ 2
µ 2
1 dwm wm
ΛD = (1 − µ )2
ρ dρ + m 2
dρ
2 0 dρ 0 ρ
2
1
1 1 dwm
+ µ2 −ρ dρ (7.184)
2 µ ρ dρ
1 2
1 wm
−m2 2
+ 1 dρ
µ ρ ρ
1
ρ(1 − µ2 ), 0 ≤ ρ ≤ µ
2
g(ρ) = 2 (7.186a)
1 µ
−ρ , µ≤ρ≤1
ρ 2
and
1 − µ2 ), 0 ≤ ρ ≤ µ
dg 2 (1
2
= 1 µ (7.186b)
dρ − + 1 , µ≤ρ≤1
µ2 2
µ 2
µ 2
1 dwm wm
ΛD = − (1 − µ )
2
ρ dρ + m 2
dρ
2 0 dρ 0 ρ
2
1
1 2 1 dwm
+ µ − −ρ dρ (7.187)
2 µ ρ dρ
1 2
2 1 wm
+m 2
+1 dρ
µ ρ ρ
© 2001 by Chapm
7.2 Large Deflection Theory
In this section, we relinquish the small deflection assumption that
was imposed in §7.1; we will work within the context of large deflection
theory and will study the buckling and postbuckling behavior of isotropic
and orthotropic elastic plates. For the case of a rectangular plate, our
focus will be on the postbuckling behavior in the isotropic case and on
large thermal deflections in the orthotropic case; postbuckling problems
for rectangular orthotropic plates will be considered in §7.3 within the
context of Berger’s approximation. For the case of an isotropic circular
plate, the emphasis in this section will be on computing the critical
buckling temperature and the corresponding buckling mode; a discussion
of initial buckling, within the context of thermoelasticity theory, for
cylindrically orthotropic circular plates may be found in the paper by
Stavsky [231].
The solution of a large deflection problem for an isotropic rectangular
plate involves the determination of the two unknown functions w(x, y)
and Φ(x, y) from equations (6.93a,b) and suitable boundary support
conditions. In the small deflection case, the buckling problem could be
solved in two steps, namely, one would solve (6.93b) with [w, w] = 0 for
Φ, and then employ that solution in (6.93a) to independently determine
w. In the present situation, a simultaneous solution for both Φ and w
is needed and, in most situations, one must resort to (approximate) se-
ries solution, to iteration schemes, or to numerical methods. An iterative
procedure due to S.R. Boley [232], which has been described in [221], will
be presented for the case of thermal buckling of a rectangular isotropic
plate; in fact, the temperature distribution and applied boundary condi-
tions on the edges are identical with those in (7.93)–(7.95), respectively,
so that for edges at x = 0, a(0 ≤ y ≤ b) and y = 0, b(0 ≤ x ≤ a) simply
supported, i.e. (7.101), the critical combination of applied edge stress
σ0 and “temperature” c1 is obtained by setting the determinant of the
system of homogeneous linear algebraic equations (7.103) equal to zero,
where kmn is given by (7.104).
The iterative procedure consists of determining a set of successive
approximations where the first approximation is the linear solution, i.e.,
those expressions for Φ and w which correspond to the loading that
initiates buckling; these expressions are then employed in the nonlinear
terms of (6.93a,b) in the manner described below, with t ≡ 0, to obtain
new expressions for Φ and w and successive iteratives are computed in
the same fashion. Provided that the actual solution is close to the initial
iterative (i.e., the loading does not greatly exceed the critical load) the
convergence of the iterative scheme may be shown to be rapid [232]; in
fact, as little as two iterations may suffice to produce reasonable results.
Suppose that we retain only two terms of the infinite series for w, i.e.,
of (7.102); then, as shown in [221] the deflected shape at this level of
approximation assumes the form
mπx sin mπy sin 3mπy
w = sin a1 + a3 (7.188)
a a a
a3
As a first approximation we take = L3 with L3 the solution of the
a1
equation which corresponds to either the first or second line of (7.105);
thus
km1 + σ0
L3 = 1 − 2 (7.189)
αEc1
The first approximation for w in the iterative scheme say, w(1) , is then
given by (7.188), with a3 /a1 given by (7.189); it is, therefore, expressed
in terms of an amplitude a1 , which is still indeterminate at this point.
To obtain the second iteration, we substitute the first set of iteratives,
i.e w(1) (x, y), as determined above, and Φ(1) (x, y), as given by (7.97)
into the system (6.93a,b) as follows; first w(1) is used in (6.93b) so as to
produce the equation
1
2 Φ(2) = − Eh[w(1) , w(1) ] − (1 − ν)N T (7.190)
2
for Φ(2) (x, y); the solution of (7.190) which satisfies the same boundary
conditions as Φ(1) (x, y) is given by
σ0 y 2 # nmπy
Φ(2) = − rn cos
2 n=2,4,6
a
# nmπy (7.191)
2mπx
+ cos sn cos
a n=0,2,4
a
with
−αEc1 a2 Ea21
r2 = + (1 − 2L3 )
2m2 π 2 32
Ea21
r = L3 (7.192a)
4
64
2
r = Ea1 L2
6
288 3
© 2001 by Chapm
and
Ea21
s0 = (1 + 9L23 )
32
Ea21
s = L3 (7.192b)
2
16
2
s = − Ea1 L
4 3
400
Next, w(1) and Φ(2) are substituted into the right-hand side of (6.93a)
so as to produce for w(2) (x, y) the equation
(2) (1) (2) (1)
K 2 w(2) = Φ,yy w,xx − 2Φ,xy w,xy
(7.193)
(2) (1)
+Φ,xx w,yy − M T
and
αEc1 a3
+ (km3 + σ0 )
2 a1
a 2 426
(7.194b)
1
= 3(1 − ν 2 )km1 −1+ L3
h 25
1 2 y2 y3
Φ̃0 (x, y) = b EαT0 1 − 3 2 + 2 3
12 b b
x x
x B1 sin hR1 sin R2 + (7.195)
a a
x x
B2 cos hR1 cos R2 + 1
a a
a a
R1 = k1 , R2 = k2 (7.196a)
b b
© 2001 by Chapm
* *
4 105 21
k = · 1 +
1 13 65
* * (7.196b)
4 105 21
2
k = · 1 −
13 65
1 y2 y3
0
σyy ≡ Φ̃0,xx = EαT0 1 − 3 2 + 2 3
12 b b
x
(7.197b)
x x x
× D1 sin hR1 sin R2 + D2 cos hR1 cos R2
a a a a
and
1 y y
0
σxy ≡ −Φ̃0,xy = EαT0 1 −
2
x x
b b
x x (7.197c)
× D3 sin hR1 cos R2 + D4 cos hR1 sin R2
a a a a
with the stress components in the domain −b ≤ y ≤ 0 identical with
those given by (7.197a,b,c). In (7.197a,b,c) the Di s are given by
D1 = B1 (k12 − K22 ) − 2B2 k1 k2
D2 = B2 (k 2 − k 2 ) + 2B1 k1 k2
1 2
(7.198)
D3 = B1 k2 + B2 k1
D4 = B1 k1 − B2 k2
and then minimizing with respect to the unknown coefficients amn , leads
to a set of simultaneous equations constituting a characteristic value
problem of the form
#∞ #∞
π2 K
K pq apq + Kpqmn amn = 0 (7.201)
b2 Eα(T0 )cr h m=1,3,5 n=1,3,5
p = 1, 3, 5...., q = 1, 3, 5, ...
The solutions of (7.201) yield sets of relative values of the coefficients
amn , and associated values of the critical temperature (T0 )cr ; the ex-
pressions for the coefficients kpq , kpqmn in (7.201) are quite involved and
are delineated in Appendix A of [216].
© 2001 by Chapm
If only the terms a11 , a13 , a31 , and a33 are retained in the deflection
function (7.200) then equations (7.201) can be written, for a plate having
an aspect ratio of 1.57, in the matrix form
15.73 22.52 14.88 −5.96 a11 a11
0.504 1.426 0.871 0.377 a13 2 a
= 100π K 13 (7.202)
1.35 3.54 7.42 7.79 a b2 Eα(T0 )cr h a
31 31
−0.0735 0.208 1.043 0.437 a33 a33
π2 K
λ= ,
b2 Eα(T0 )cr h
is obtained by matrix iteration and yields λcr = 5.39. The relative
values of the four coefficients which have been retained in the deflection
function w are given by
a11 1
a13 0.0365
=
a 0.1360
31
a33 0.0042
in which case
πx πy πx 3πy
w0 = a11 cos cos + 0.0365 cos cos
2a 2b 2a 2b
(7.203)
3πx πy 3πx 3πy
+0.1360 cos cos + 0.0042 cos cos
2a 2b 2a 2b
It is easily computed, based on (7.203) that the deflection wc at the
center of the rectangular plate is wc = 1.1767 a11 in which case
wc πx πy πx 3πy
w0 = cos cos + 0.0365 cos cos
3πy (7.204)
1.1767 2a 2b 2a 2b
3πx πy 3πx
+0.1360 cos cos + 0.0042 cos cos
2a 2b 2a 2b
In Table 7.2, we have indicated the convergence of the critical tem-
perature parameter as additional terms are included in the deflection
function; as indicated in [216], retaining further terms in the deflection
function beyond those four already chosen above has a negligible effect
on both the critical temperature and the buckle pattern. The initial
buckling pattern (≡ small deflection buckle pattern) in one quadrant of
the plate is depicted in Fig. 7.13 for a plate with an aspect ratio of 1.57.
The postbuckling behavior of the heated plate which is analyzed in
[216] takes into account, as all analyses of postbuckling behavior must,
the stretching of the plate middle surface due to bending and the cor-
responding changes in the plate stress distribution as the plate deflects;
thus, this analysis is based on the generalized von Karman equations for
the hygrothermal case, i.e., (6.93a,b) with ∆M T = 0, t = 0. Actually,
since the analysis in [216] uses the Airy function Φ̃ based on the local
stress distribution, the relevant form of (6.93a,b) in this case is
K∆2 w = h Φ̃,yy w,xx + Φ̃,xx w,yy − 2Φ̃,xy w,xy
$ 2 (7.205)
∆2 Φ̃ = −Eα 2 T + E w,xy − w,xx w,yy
The resulting stress function Φ̃ is now substituted into the first equation
in (7.205) and the Galerkin approach is again used so as to determine
© 2001 by Chapm
the values of the coefficients amn of the deflection function w; as shown
in [216], this leads to the set of simultaneous equations
a b
mπx nπy
cos cos
−a −b a 2b
K (7.211)
× ∆2 w − Φ̃,yy w,xx − Φ̃,xx w,yy + 2Φ̃,xy w,xy dxdy = 0
n
(m = 1, 3, 5, ..., n = 1, 3, 5, ...)
for the amn . The equations (7.211) are, of course, nonlinear, and their
solution becomes more difficult as the number of terms retained in the
deflection function increases. In [216], therefore, it is assumed that the
shape of the deflected surface of the plate for large deflections may be
taken to be the one the plate has at the onset of buckling; with such an
assumption only the coefficient a11 remains arbitrary, while the ratios
amn /a11 are taken to be those which are given by the initial buckling
solution described above. In lieu of (7.211) we have, therefore, as the
Galerkin-type equation, from which the coefficient a11 may be deter-
mined, the relation
a b K
ws ∆2 ws − Φ̃,yy ws,xx − Φ̃,xx ws,yy
−a −b h (7.212)
+2Φ̃,xy ws,xy dxdy = 0
where
∞
# #∞
amn mπx nπy
ws = cos cos (7.213)
a
m=1,3,5 n=1,3,5 11
2a 2b
In (7.212) the ratios amn /a11 , which were obtained from the initial buck-
ling solution, must also be substituted into the stress function Φ̃.
For a plate with aspect ratio a/b = 1.57 it has already been determined
that
πx πy πx 3πy
ws = cos cos + 0.0365 cos cos
2a 2b 2a 2b
3πx πy 3πx 3πy
+0.1360 cos cos + 0.0042 cos cos
2a 2b 2a 2b
in which case (7.212) yields the following relation between the temper-
ature differential T0 and the center deflection wc :
b2 EαT0 h w2
2
= 5.39 + 1.12(1 − ν 2 ) 2c (7.214)
π K h
At any other point of the plate (assumed to have stress-free edges and
an aspect ratio a/b = 1.57) the deflection assumes the form
9
: 2
: b EαT0 h
:
w ; π 2 K − 5.39
= 0.723 ws (7.215)
h 1.12(1 − ν 2 )
© 2001 by Chapm
The boundary conditions are satisfied by choosing
∞ #
# ∞
rπx sπy
w= ars sin sin (7.217)
r=1 s=1
a b
with T0 the average temperature over the surface of the plate. As the
plate is subjected to the temperature distribution given by (7.219), the
plate will expand and this expansion will result in a distribution of aver-
age edge thrusts which are caused by the restraining spring forces acting
along the edges of the plate; we denote these average edge stresses by
ρ and ξ, respectively, along the edges x = 0, a and y = 0, b. The mag-
nitudes of ρ and ξ depend on the temperature distribution, the plate
dimensions, and the elastic moduli of the plate. Employing (7.217) and
(7.219) it may be shown that the second of the large deflection equations
in (7.205) is satisfied if Φ̃ has the form
∞ ∞
ξx2 ρy 2 # # pπx qπy
Φ̃ = + + cpq cos cos (7.220)
2 2 p=0 q=0
a b
in which
9 ( 2
# p2 b q 2 a
cpq = E Bi + Apq 4 + (7.221)
i=1
a b
where
4α p2 b a2 a
Apq = Tpq ab + (7.222)
π2 a b
while the Bi , i = 1, 2, ...9 are given in [235]; the work in [235] covers the
case of a laterally loaded plate subject to edge thrusts in the plane of the
plate with lateral displacements in the large deflection range but does not
consider temperature effects. Specific expressions for the coefficients brs ,
in terms of the coefficients ars , crs , so that the first equation in (7.205)
is satisfied are quite complex; they are delineated in [234] and will not
be replicated here.
For the plate to be in equilibrium it is necessary that
a
Φ̃,xx − ky v(x, y) dx = 0 at y = 0, b; for 0 ≤ x ≤ a (7.223a)
0
and
b
Φ̃,yy − kx u(x, y) dy = 0 at x = 0, a; for 0 ≤ y ≤ b (7.223b)
0
b (7.226)
1 E 2
v(x, 0) = − Φ̃,xx − ν Φ̃,yy + αET − w,y dy
2E 0 2
© 2001 by Chapm
and
1
v(x, 0) = − ξb − νρb + αET0 b
2E
(7.227b)
∞ ∞ ∞
bE # # # sπ 2 rπx r πx
− ars ars sin sin
4 r=1 s=1 b a a
r =1
where
νky b ky b
λ11 = − , λ12 = 1 +
2E 2E
(7.229a)
λ = − νkx a kx a
22 , λ22 = 1 +
2E 2E
∞ #
# ∞ sπ 2
ky b 1
µ11 = −
αET0 − E a2
2E 8 r=1 s=1 rs b
(7.229b)
1 # # 2 rπ 2
∞ ∞
kx a
µ22 = − 2E αET0 − 8 E
ars
a
r=1 s=1
From (7.228) we obtain the average edge (thrust) stresses in the form
and
ξx2 ρy 2
Φ̃0 = + (7.231)
2 2
and the first equation in (7.205) reduces to the statement that
2 a 2 a 2
Eπ 2 h
1 + = −ρ − ξ (7.232)
12(1 − ν ) a
2 b b
Using (7.229a,b), (7.230a,b) in (7.232), with the summations extending
only over r = s = 1, then yields the following expression for the buckling
temperature
(αT0 )cr = P̂ /Q̂
2 a 2
π2 h 2E 2E
P̂ = 1 + 1 + 1 + − ν 2
12(1 − ν 2 ) a b kx a ky b
a 2 2E a 2 2E
Q̂ = 1 + (1 + ν) + +
b ky b b kx a
(7.233)
If kx = ky = ∞ then (7.233) yields the expression for the buckling
temperature for the case of a nonflexible edge restraint in the form
2 a 2
∞ π2 h
(αT0 )cr = 1+ (7.234)
12(1 + ν) a b
in which case, for a = b and kx = ky
2
2E
1+ − ν2
(αT0 )cr 1 kx a
∞
= · (7.235)
(αT0 )cr 1−ν 2E
1+ν+
kx a
Numerical results have been generated by the authors in [234] for the
case of a uniform temperature distribution T = T0 and a sinusoidal
lateral loading of the form
πx πy
t = b11 sin sin (7.236)
a b
Plots of the center deflection of the plate versus a lateral load parameter
based on q11 are depicted in Fig. 7.17 for various values of T0 , the
© 2001 by Chapm
aspect ratio a/b, and the spring constant kx (kx = ky in the application).
For a uniform lateral loading, the variation of the buckling temperature
ratio (αT0 )cr /(αT0 )∞
cr with the edge spring rate parameter kx a/2E is
depicted in Fig. 7.18 (a = b, kx = ky in the application) while for
the same loading condition the effect of the edge fixity on the critical
temperature is displayed in Fig. 7.19; finally, the center deflection versus
temperature ratio (postbuckling) curves for this case are depicted in Fig.
7.20 for several different values of kx a/2E.
Having discussed several examples of buckling and postbuckling be-
havior for heated isotropic rectangular plates, we now turn to the rec-
tilinear orthotropic case for such a geometry. For this problem, the rel-
evant generalized von Karman equations are given by (6.103a,b) where
we set the applied transverse loading t(x, y) ≡ 0. Athough many of the
papers in the literature, which deal with the buckling behavior of heated,
rectangular, orthotropic plates do so within the context of the Berger’s
approximation, which will be discussed in the next section, there are a
few analyses which deal with the full system (6.103a,b) in the context of
plate bending, e.g., the work of Biswas [236]; this work appears to con-
tain a serious flaw common to much of the literature on thermal bending
and buckling of simply supported plates, as we shall indicate below.
In [236], the author writes (6.103a), with t ≡ 0, in the form
and it is easy to see that, with respect to the notation employed in the
present work
while F = Φ and
β̃1 = c11 α1 + c12 α2
(7.239)
β̃2 = c21 α1 + c22 α2
The β̃i in (7.239) are not to be confused with their previous interpre-
tation as hygroscopic expansion coefficients (where we denoted them as
βi ). To bring the second von Karman equation for this case into line
with the form employed in [236], we multiply (6.103b) through by E2 h
thus producing
E2 E2
Φ,xxxx + − 2ν12 Φ,xxyy + Φ,yyyy
G12 E1
1
= − E2 h[w, w] − (ÑT2 − ν12 ÑT1 ),xx (7.240)
2
E2
− (ÑT1 − ν21 ÑT2 ),yy
E1
while
E1 α1 + E2 α2 ν21
ÑT2 − ν12 ÑT2
=
1 − ν12 ν21
(7.241b)
E1 α1 ν12 + E2 α2
−ν21 NT ≡ E1 α1 NT
1 − ν12 ν21
where
h/2
NT = δT (x, y, z)dz
−h/2
© 2001 by Chapm
if we again identify F = Φ and take
E2
p2 = − 2ν12
G
2
12
q = E2 /E1
(7.244)
λ1 = E2 α2
λ2 = E2 α1
with ∆ = s11 s22 − s212 . It is any easy task to show that, based on (1.61)
and (7.246)
1 1 s12
E1 = , E2 = , ν12 = −
s11 s22 s11 (7.247)
s12 1
ν21 = − , G12 =
s22 s66
We now consider a simply supported orthotropic plate which occupies
the region 0 ≤ x ≤ a, 0 ≤ y ≤ b, − 12 h ≤ z ≤ 12 h in the three-space
and assume a temperature distribution which varies linearly through the
plate of the form
1 T1 − T2
δT (x, y, z) = (T1 + T2 ) + z (7.248)
2 h
so that
1 1
T (x, y, h) = T1 , T (x, y, − h) = T2 (7.249)
2 2
Corresponding to the simple-support conditions, along the edges of the
plate, the author, in [236] looks for plate deflections of the form
πx πy
w(x, y) = w0 sin sin (7.250)
a b
However, by virtue of (6.104), the assumption (7.250) cannot possibly be
valid unless M̃T1 = M̃T2 = 0, i.e., unless MT = 0 (which is not true in view
of (7.248)). For the temperature distribution (7.248), NT,x = NT,y = 0;
Using (7.250) in (7.243) it may be shown that the general solution of
(7.243) is given by an Airy function F (x, y) of the form
2
1 1 1 a 2πx b2 2πy
F = Ax2 + By 2 + E2 hw02 cos + cos (7.251)
2 2 32 b2 a a2 q 2 b
b
s22 E2 hw02 π 2 2πx
v= A− cos
0 h 8b2 a
s12 E2 hw02 π 2 2πy
+ B− cos (7.252b)
h 8a2 q 2 b
2
© 2001 by Chapm
For immovable plate edges (in the plane of the plate) u = v = 0; com-
bining this latter assumption with (7.252a,b) we are led to
with
E2 hπ 2 s11 s22 s22 s11
C1 = −
8∆ a2 q 2 b2
2 (7.253b)
C2 = E2 hπ s11 s22 s22 s12
−
8∆ a2 q 2 b2
for y = 0, b, 0 ≤ x ≤ a, where
h/2
M̃T2 = (c11 α1 + c12 α2 ) δT (x, y, z)dz (7.257a)
−h/2
h/2
M̃T2 = (c21 α1 + c22 α2 ) δT (x, y, z)dz (7.257b)
−h/2
so that
M̃T1 = (c11 α1 + c12 α2 ) h(T1 + T2 ) ≡ A∗
2 (7.258)
M̃ 2 = (c21 α1 + c22 α2 ) h(T1 + T2 ) ≡ B ∗
T
2
Setting D11 = λ1 , D12 = λ2 , D21 = γ1 , D22 = γ2 , the boundary condi-
tions of simple support which are represented by (7.256a,b), (7.257a,b)
become
w(0, y) = w(a, y) = 0
λ1 w,xx (0, y) + λ2 w,yy (0, y) = A∗ (7.259a)
∗
λ1 w,xx (a, y) + λ2 w,yy (a, y) = A
for 0 ≤ y ≤ b and
w(x, 0) = w(x, b) = 0
γ1 w,xx (x, 0) + γ2 w,yy (x, 0) = B ∗ (7.259b)
γ1 w,xx (x, b) + γ2 w,yy (x, b) = B ∗
© 2001 by Chapm
for 0 ≤ x ≤ a. However, as w(0, y) = 0, 0 ≤ y ≤ b, it follows that
w,yy (0, y) = 0, 0 ≤ y ≤ b, and, similarly, w(a, y) = 0, 0 ≤ y ≤ b, implies
that w,yy (a, y) = 0, 0 ≤ y ≤ b. In an analogous manner, we obtain from
w(x, 0) = 0, 0 ≤ x ≤ a, the fact that w,xx (x, 0) = 0 0 ≤ x ≤ a, and from
w(x, b) = 0, 0 ≤ x ≤ a, the conclusion that w,xx (x, b) = 0, 0 ≤ x ≤ a.
Thus, (7.259a,b) imply that along the edges of the plate we must have
w(0, y) = w(a, y) = 0, 0 ≤ y ≤ b
(7.260a)
w(x, 0) = w(x, b) = 0, 0 ≤ x ≤ a
w,xx (0, y) = w,xx (a, y) = A∗ /λ1 , 0 ≤ y ≤ b
(7.260b)
w,yy (x, 0) = w,yy (x, b) = B ∗ /γ2 , 0 ≤ x ≤ a
and
dφ ν
Mr = K + φ − MT∗
dr r
(7.263)
dφ 1
Mθ = K ν + φ − MT∗
dr r
© 2001 by Chapm
where MT∗ is given by (6.111). With the assumptions given above, the
initial buckling equation that follows from the system (6.112) by setting
w = 0 in the second equation of this set, and substituting the resultant
Airy function in the first equation, is
2d φ2
dφ Nr 2 dMT∗
r +r + r − 1 φ = r2 (7.264)
dr2 dr K dr
Nr 1
Replacing ≡ − NT∗ in (7.264) by γ 2 and then setting v(r) = γr we
K K
obtain the equation
d2 φ 1 dφ 1 1 dMT∗
2
+ + 1 − 2
φ= (7.265)
dv v dv v γ dv
in which case,
w(r) = CγJ0 (γr) + C1 (7.268)
K
(Nr )cr = (3.832)2 (7.269a)
a2
so that
K(1 − ν)
(NT )cr = − (3.832)2 (7.269b)
a2
A second case considered in [238] is one for which the temperature dis-
tribution is such that
M0 v2
MT∗ = 1− 2 2 (7.270)
1−ν γ a
2M0
φ = C2 J1 (v) − v (7.272)
γ 3 a2 (1 − ν)
M0 r 2 C2
w = C3 + − J0 (γr) (7.273)
γ a (1 − ν)
2 2 γ
© 2001 by Chapm
The radial and circumferential stress fields in this situation are given,
respectively, by
E Eα
σrr = (err + νeθθ ) − (T0 + zT1 )
1−ν 2 1−ν (7.277)
σθθ =
E
(eθθ + νerr ) −
Eα
(T0 + zT1 )
1−ν 2 1−ν
We note that (6.108) follows from (7.276), (7.277) if (7.277) is integrated
through the thickness of the plate. In (7.277), of course
2
err = v,r + (w,r ) − zw,rr
2 (7.278)
eθθ = 1 v − z w,r
r r
where v = v(r) again denotes the radial displacement of the plate. Com-
bining (7.277) and (7.278) the resultant forces and moments are easily
shown to have the form
E 1 ν
Nr = v,r + (w,r ) + v − (1 + ν)αT0
2
1 − ν2 2 r
(7.279)
Eh 1 1 2
Nθ = v + ν(v,r + (w )) − (1 + ν)αT 0
1 − ν2 r 2 ,r
and
ν
Mr = −K w,rr + w,r + (1 + ν)αT1
r
(7.280)
1
Mθ = −K w,r + νw,rr + (1 + ν)αT1
r
In [222], Nowacki works directly with the equilibrium equations, for the
symmetric situation, in the form
(rNr ),r − Nθ = 0
(rMrr ),r − Mθ − Qr = 0 (7.281)
(rQ),r − (rNr w,r ),r − tr = 0
with Q the shear force and t the external loading of the plate. By
inserting Q from the second equation in (7.281) into the third equation
in this set, and then employing the relations (7.280), one obtains
1
K 4r w + K(1 + ν)α 2r T1 = (rNr w,r ),r + t (7.282)
r
1
with 4r = 2r 2r and 2r = ∂r2 + ∂r . Equation (7.282) is, of course,
r
the form assumed by the first of the von Karman equations in (6.112)
under the assumption of radial symmetry.
From the relations (7.279) one obtains the radial displacement v(r) in
the form
r
v(r) = (Nθ − νNr ) + αrT0 (7.283)
Eh
or, if we use the first of the relations in (7.281) to eliminate the resultant
stress Nθ in (7.283)
r
v(r) = [(rNr ),r − νNr ] + αrT0 (7.284)
Eh
© 2001 by Chapm
in terms of which (7.285) and (7.287) assume the respective forms
1 1 2
ρψ,ρρ + ψ,ρ − ρ ψ = 2 φ
(7.289)
1
ρφ,ρρ + φ,ρ − φ = −kφψ + mρ 2
ρ
while the boundary conditions (7.286) become
1
Then, as φ = w,ρ
a
φ(1) = c(ρ3 − ρ), c = 4w0 /a (7.292)
Substituting φ(1) for φ in the first equation in (7.289) and using the
boundary conditions, yields
c2 7
ψ = 96 (ρ − 4ρ + 6ρ − bρ)
(1) 5 3
(7.293)
5 − 3ν 96 αT0
b= − 2
1−ν c 1−ν
If we now substitute ψ (1) , from (7.293), for ψ in the second equation in
(7.289), multiply this equation by ρ3 − ρ, and integrate from ρ = 0 to
ρ = 1, we obtain
3 k b 1 1
c+c − = m (7.294)
64 24 14 8
4 5 − 3ν
In (7.294) we recall that c = w0 , while the presence of b = −
a 1−ν
96 αT0
accounts for the influence of the temperature distribution on
c2 1 − ν
the plate deflection (which is seen to depend only on the function T0 (r)).
For the special case in which t = 0 (so that m = 0 in (7.294)) it is possible
to obtain the critical buckling temperature in the form
4 h2
(T0 )cr · (7.295)
3(1 + ν) αa2
4 Drr
(T0 )cr · (7.296)
3 Ar ha2
where h/2
Drr = Err zdz
−h/2 (7.297)
Ar = −(αr Err + αθ Erθ )
© 2001 by Chapm
yield exact solutions. Among the many approximate methods which
have been devised to treat large deflection von Karman systems is the
technique first promulgated by Berger in [240] for thin plates subjected
to mechanical loading. As Tauchert [217] notes, “The method is based on
neglecting the strain energy associated with the second invariant of the
strains in the plate middle surface. Although there does not appear to
be any physical justification for this approximation, comparisons of the
results with known solutions indicate that for a broad range of problems
Berger’s approach yields sufficiently accurate results.” In fact, all of the
applications of Berger’s approximation in the literature, where accurate
results appear to be obtained, are associated with problems in which the
edges of the plate are constrained against in-plane movements; Nowinski
and Ohnabe [241] have shown that, for the case of mechanical loading,
the accuracy of Berger’s method is closely tied to this specific type of
boundary condition. In fact, by applying Galerkin’s method to both
the usual von Karman large deflection equations, and the approximate
equations governing large deflections which are obtained from Berger’s
approach, it was determined in [241] that Berger’s method is susceptible
to serious errors if the edge (or edges) of the plate are free to move. It has
been surmised in [217] that the conclusions reached in [241] are almost
certainly extendible to the case of thermal deflections. The accuracy of
Berger’s method was also assessed by Jones, Mazumdar, and Cheung in
[242] for the special case in which t ≡ 0, M T ≡ 0 in (6.93a,b), and a
perturbation technique was proposed for use in those situations where
Berger’s technique may not be sufficiently accurate. We will comment
below on the work in [242] as well as on the work of Banerjee and Datta
[243], who proposed a modified energy expression which leads, as in
the case of Berger’s method, to large deflection plate equations which
are decoupled. Although only mechanical loadings are considered in
[243], the results given for transverse loading of a circular plate are in
good agreement with prior results for this problem for both movable and
immovable edges and an extension of the method to the case of thermal
loading has been indicated.
1 0 2
$2 = $0xx $0yy − (γxy )
4
1 2 1 2
= u,x v,y + u,x w,y + v,y w,x (7.299b)
2 2
1
− u2,y + v,x 2
+ 2u,y v,x + 2u,y w,x w,y + 2v,x w,x w,y
4
Assuming a temperature distribution δT of the form
δT (x, y, z) = T0 (x, y) + g(z)T1 (x, y) (7.300)
Basuli [244] writes (7.298) in the form
b a
1 12
Π= K (w)2 + 2 $21
2 h
0 0
12
−2(1 − ν) 2 $2 + w,xx w,yy − w,xy2
− tw (7.301)
h
Eα
− $1 hT0 − f (h)T1 2 w dxdy
1−ν
© 2001 by Chapm
where h/2
f (h) = g(z)zdz
−h/2
The functional Π is now minimized using the Euler-Lagrange varia-
tional equations
∂Π ∂ ∂Π ∂ ∂Π
− − =0 (7.302a)
∂u ∂x ∂u,x ∂y ∂u,y
∂Π ∂ ∂Π ∂ ∂Π
− − =0 (7.302b)
∂v ∂x ∂v,x ∂y ∂v,y
∂Π ∂ ∂Π ∂ ∂Π ∂ 2 ∂Π
− − +
∂w ∂x ∂w,x ∂y ∂w,y ∂x2 ∂w,xx
(7.302c)
∂ 2 ∂Π ∂ 2 ∂Π
+ + 2 =0
∂x∂y ∂w,xy ∂y ∂w,yy
which yields (upon following Berger [240] and setting $2 = 0) the rela-
tions
∂ Eh ∂ Eh
$1 − N T
= 0, $1 − N T
= 0 (7.303a)
∂x 1 − ν 2 ∂y 1 − ν 2
and
∂2 ∂2
(K 2
w + M T
) + (K 2 w + M T )
∂x2 ∂y 2
∂ Eh
− $1 − N T
w,x (7.303b)
∂x 1 − ν2
∂ Eh
− $1 − N T
w,y − t = 0
∂y 1 − ν2
in the general case and
∂ ∂
{$1 − (1 + ν)αT0 } = 0, {$1 − (1 + ν)αT0 } = 0 (7.304a)
∂x ∂y
∂2 Eα ∂2 Eα
K w+ 2
f (h)T1 + 2 K w + 2
f (h)T1
∂x2 1−ν ∂y 1−ν
E ∂
− {[$1 − (1 + ν)αT0 ]w,x } (7.304b)
1 − ν ∂x
E ∂
− {[$1 − (1 + ν)αT0 ]w,y } − t = 0
1 − ν ∂y
in the special case in which δT (x, y, z) is given by (7.300).
From (7.303a) one easily obtains
Eh
$1 − N T = λ2 , (λ2 const.) (7.305)
1 − ν2
which, upon substitution into (7.303b), yields
K 2 (2 − λ2 )w = t − 2 M T (7.306)
µ2 h2
$1 − (1 + ν)αT0 = , (µ2 const.) (7.307)
12
whose substitution into (7.304b) produces
αE
K 2 (2 − µ2 )w = t − f (h) 2 T1 (7.308)
1−ν
We now consider the application of the system (7.307), (7.308) to the
thermal buckling of a simply supported rectangular plate which occupies
the domain 0 ≤ x ≤ a, 0 ≤ y ≤ b in the x, y plane; in this case, the
assumed temperature distribution has the form given by (7.300).
Remarks: In [244] Basuli writes the conditions of simple support along
the edges of the plate in the form
w = w,xx = 0, x = 0, a; 0 ≤ y ≤ b
(7.309)
w = w,yy = 0, y = 0, b; 0 ≤ x ≤ a
K(w,xx + νw,yy ) + M T = 0, x = 0, a; 0 ≤ y ≤ b
K(w,yy + νw,xx ) + M T = 0, y = 0, b; 0 ≤ x ≤ a
with w = 0 along each of the four edges; thus (7.309) holds, in the case
of simply supported edges if and only if M T (0, y) = M T (a, y) = 0, for
© 2001 by Chapm
0 ≤ y ≤ b, and M T (x, 0) = M T (x, b) = 0, for 0 ≤ x ≤ a. For the
temperature distribution represented by (7.300)
h/2
αE αE
MT = T1 (x, y) g(z)zdz ≡ f (h)T1 (x, y) (7.310)
1−ν −h/2 1−ν
and thus (7.309) represents the conditions of simple support along the
edges of the plate if and only if
T1 (0, y) = T1 (a, y) = 0; 0 ≤ y ≤ b
(7.311)
T1 (x, 0) = T1 (x, b) = 0; 0 ≤ x ≤ a
To maintain the validity of the analysis in [244] we will assume, below,
that the function T1 satisfies (7.311).
Proceeding with the analysis, we assume that each of t(x, y), T1 (x, y)
mπx
may be expanded in a double Fourier series in the products sin
a
nπy
sin so that, in particular, on the right-hand side of (7.308)
b
#∞ # ∞
αE mπx nπy
t− f (h) 2 T1 = qmn sin sin (7.312)
1−ν m=1 n=1
a b
Assuming the edges of the plate are constrained against in-plane motion
is equivalent to requiring that
u(0, y) = u(a, y) = 0; 0 ≤ y ≤ b
(7.314)
v(x, 0) = v(x, b) = 0; 0 ≤ x ≤ a
By substituting (7.312) and (7.313) into (7.308) we obtain, in the usual
manner, the relations
q
wmn = mn
mπ 2 nπ 2 mπ 2 nπ 2
(7.315)
K + + +µ 2
a b a b
so that the deflection has been determined up to the constant µ2 . We
now write out (7.307) in the form
1 2 1 2 µ2 h2
u,x + v,y + w,x + w,y − (1 + ν)αT0 = (7.316)
2 2 12
Integration of (7.316) over the domain of the plate and use of the bound-
ary conditions (7.314), which express the constraint of the plate against
in-plane motion, yields the condition
a b
1 2 µ2 h2
w,x + w,y − (1 + ν)αT0 dxdy =
2
ab (7.317)
0 0 2 12
We now substitute the expansion (7.313) into (7.317) and use standard
results involving trigonometric integrals, e.g.,
a b
mπx nπy 1
cos2 sin2 dxdy = ab
0 0 b b 4
The further substitution of the expressions for the wmn in (7.315) into
(7.318) then produces an algebraic relation from which µ2 may be deter-
mined; once µ2 has been computed (7.313), (7.315) serve to completely
determine the deflection w(x, y).
Remarks: Basuli [244] considers a specific example for which
2πy
T0 (x, y) = T0 = const., T1 (x, y) = T1 cos (7.319)
b
However, while the series on the right hand side of (7.320) converges
to the value of M T (x, y), at each point of continuity of M T in the
© 2001 by Chapm
domain 0 < x < a, 0 < y < b, it will not converge to the values of
M T along the boundary of the domain; it is the function M T (x, y), as
defined by the temperature distribution δT (x, y), and not its Fourier
series representation, which must satisfy the conditions (7.311) so the
criteria for a simply supported rectangular plate, under thermal loading,
reduce to (7.309). In a one-dimensional setting, this phenomenon is
easily represented by the Fourier sine series expansion of f (x) = 1 on
(0, π), i.e.
∞
4#
1= sin(2k − 1)x, 0 < x < π (7.321)
π
k=1
For all x, 0 < x < 1, the series on the right-hand side of (7.321) con-
verges to 1; however, convergence breaks down at x = 0, x = π where
the series converges, by the Fourier theorem, to the average value of the
left-hand and right-hand limits (of the periodic extension, with period
2π, of the odd extension of f (x) to (−π, π)). The situation is depicted
below:
© 2001 by Chapm
Basuli [244] writes the system (7.322), (7.323) in the form
du v 1 2 β 2 h2
+ + w,r − (1 + ν)αT0 = (7.328)
dr r 2 12
2 2
d 1 d d 1 d
+ + − β2 w
dr2 2 dr dr2 r dr
2 (7.329)
1 αE d 1 d
= t− f (h) + T1
K 1−ν dr2 r dr
and obtains, for a uniform load distribution t, that special case of (7.324)
which has the form
t
w = AI0 (βr) + B − r2 (7.330)
4Kβ 2
αE d2 1 d
with t = t− f (h) + T1 . Clearly, t = t if either f (h) =
1−ν dr2 r dr
dw
0 or if T1 is harmonic. With u(b) = 0, and w(b) = (b) = 0, the
dr
solution of the system (7.328), (7.329) follows the same pattern as that
outlined above for temperature distributions more general than (7.300)
and will, therefore, not be repeated here. Somewhat more problematic,
however, is the solution of (7.328), (7.329), which is presented in [244] for
the case of a simply supported circular plate subject to the temperature
distribution (7.300) and the assumption of constrained in-plane motion;
in [244] Basuli takes the conditions of simple support along the edge of
the plate at r = b to have the form w = 0, at r = b, and
d2 w w dw
+ = 0, at r = b (7.331)
dr2 r dr
However, by virtue of (7.323), with w,θ = 0, and R1 = 0, R2 = b, we
have, along the edge at r = b, w(b) = 0 and
2 '
d w 1 dw '' '
K + ' + MT∗ 'r=b = 0 (7.332)
dr2 r dr '
r=b
The problem that arises now is the same one which has been discussed
at length earlier in this section, namely, in view of (6.111)
' Eα h/2
Eα
MT∗ 'r=b = δT (b, z)zdz = f (h)T1 (b) (7.333)
1−ν −h/2 1−ν
so that (7.331) follows from (7.333) if and only if f (h) = 0 or T1 (b) = 0;
under this latter assumption, the results in [244] are valid and the full
solution of the boundary-value problem associated with (7.331), (7.329),
when the edge at r = b is simply supported, assumes the form
r 2
−t b4
w= 2(1 + ν)[I0 (βb) − I0 (βr)] − Hβb 1 −
4Kβ 3 b3 H b
(7.334)
4 2
tb
Kh
1 (1 + ν)αF (b) 4 4
(βb)6 + β b
= 3 h2
1 1 2 I0 (βb)I1 (βb) 4(1 + ν)I2 (βb)
+ [I (βb) − I0 (βb)] +
2
−
2 2 1 βb Hβb
(7.335)
where
H = βbI0 (βb) − (1 − ν)I1 (βb)
b (7.336)
F (b) = T0 (r)rdr
0
Although Basuli’s work [244] extending Berger’s method [240] to non-
isothermal plate deflection problems preceded the work of Pal [247], the
latter author appears not to have been aware of the earlier research in
this direction. Thus, Pal [247] rederives the Berger’s type equations by
minimizing the total potential energy associated with a circular plate
under the assumption of a radially symmetric temperature; these de-
coupled large deflection equations are then solved in [247], for the case
in which t ≡ 0, by using the method of successive approximations. In-
cluded in the analysis in [247] are results for annular plates, i.e., circular
plates with a concentric circular hole. In line with our related discus-
sions in this section, the applicability of the results in [247] for circular
and annular plates with simply supported edges is subject to the same
constraints on the behavior of T1 (r) along the plate edges, which have
been delineated previously. In Pal [247], the temperature distribution is
taken in the form
2z
δT (ζ, z) = T̄0 + T̄1 (1 − ζ 2 ) 1 + (7.337)
3h
© 2001 by Chapm
with r 2
T0 (r) = T̄0 + T̄1 1 − = T1 (r)
a (7.338)
g(z) = 2 · z
3 h
It is clear from (7.337) that T1 (a) = 0 if and only if T̄0 = 0; in [247]
the outer edge of an annular plate is located at r = a while the inner
edge is at r = b (with b = 0 for a solid circular plate). Therefore, in
Pal’s work [247], (7.331) is equivalent to (7.332), with b replaced by a
provided T̄0 = 0. In Figs. 7.23 and 7.24, which are taken from [247],
we show the relation between the temperature rise and the normalized
maximal deflection δ/h(δ = w(0)) at the center of the plate for the
simply supported and clamped edge conditions, respectively; in these
figures h/2
T̃ = δT (r, z)dz
−h/2 (7.339)
h/2
T̄ = δT (r, z)zdz
−h/2
b
Fig. 7.25 depicts, for an annular plate with aspect ratio = 0.4,
a
the relation between the temperature rise and the normalized deflection
δ/h, for the same temperature distribution (7.337); the plate in Fig.
7.25 has both inner and outer edges simply supported. For the same
plate and the same temperature distribution, the analogous results for
an annular plate with both inner and outer edges clamped are depicted
in Fig. 7.26. The results in [247] for the case where the same annual
plate has its inner edge clamped and its outer edge simply supported is
shown in Fig. 7.27. The results shown in Fig. 7.25 (the annular plate
with both edges simply supported) are open to the criticism that even if
T̄0 = 0, so that T1 (a) = 0 is a direct consequence of (7.338), i.e., it does
not necessarily follow that T1 (b) = 0; in fact T1 (b) = 0, when T̄0 = 0, if
and only if a = b which reduces the problem to one for a solid circular
plate of radius a.
Both Biswas [248] and Pal [249] have applied Berger’s approximation
method to linearly elastic thin plates exhibiting anisotropic response.
In [248], Biswas solves the large-deflection problem for a rectilinearly
orthotropic rectangular plate which is simply supported and subjected
to a stationary temperature distribution, while in [249], Pal obtains
solutions for cylindrically orthotropic circular plates using the method
of successive approximations; it is also demonstrated in [249] that the
solutions for a polar orthotropic circular plate reduce to those for an
isotropic circular plate as a special case. We now briefly review the
analysis in [248] and [249] beginning with the work of Biswas [248], which
is, unfortunately, subject, once again, to the same criticism concerning
the interpretation of simple support boundary conditions in thermal
deflection problems. The notation employed by Biswas in [248] parallels
that of his earlier paper [236]; thus, in particular, (7.238) applies, i.e.
Dx = D11 , H = D1 + 2D66 , Dy = D22 (7.340)
1
with D1 = (D12 + D21 ), while the β̃i , i = 1, 2, are given by (7.239).
2
The first strain invariant $1 is given in this situation by
*
0 0 Dy
$1 = $xx + κ$yy , κ = (7.341)
Dx
h/2
and MT , as in (7.237), is given by MT = δT zdz. The full set of von
−h/2
Karman large deflection equations has the form (7.237), (7.240) where
the E1 , E2 , ν12 , ν21 , G12 , and the resultant (thermal) stresses NT , ÑT ,
ÑT2 have been defined in Chapter 6. Neglecting the second strain invari-
ant $2 , Biswas [248] writes the total potential energy for the rectilinearly
orthotropic rectangular plate in the form
1 b a% 2
Π= Dx w,xx + 2D1 w,xx w,yy
2 0 0
12 &
+Dy w,yy 2
+ 4D66 w,xy 2
+ Dx 2 $21 dxdy
h/2 b a % h
&
+ β̃1 $xx δT (x, y, z) + β̃2 $yy δT (x, y, z) dxdydz
−h/2 0 0
(7.342)
The temperature variation in [248] is chosen so as to have the form
z h h
δT (x, y, z) = 2θ , − ≤z≤ (7.343)
h 2 2
in which case
h h
T x, y, = θ, T x, y, − = −θ (7.344)
2 2
Combining (7.342) with (7.343), and using Euler’s equations (7.302a,
b, c), one obtains
∂$1 ∂$1
= 0, =0 (7.345a)
∂x ∂y
© 2001 by Chapm
and
∂2 % &
Dx w ,xx + D 1 w,yy + β̃1 M T
∂x2
∂ % &
+ 2 Dy w,yy + D1 w,xx + β̃2 MT
∂y
∂2 ∂ 12 (7.345b)
+4 2 (D66 w,xy ) − Dx 2 $1 w,x
∂x ∂x h
∂ 12
− Dy 2 $1 κw,y = 0
∂y h
Clearly, (7.345a) implies that
α 2 h2
$1 = const. ≡ (7.346)
12
while (7.345b), in conjunction with (7.346), yields
Dx w,xxxx + 2Hw,xxyy + Dy w,yy
(7.347)
−Dx α2 (w,xx + κw,yy ) + β̃1 MT,xx + β̃2 MT,yy = 0
For the temperature variation given by (7.343) we have
h/2
1
MT ≡ δT (x, y, z)dz = θh2 = const.
−h/2 6
which Biswas [248] expands in the double Fourier series
∞
# #∞
16MT mπx nπy
MT = 2
sin sin (7.348)
m=1,3,... n=1,3...
mnπ a b
Amn
16MT m2 π 2 n2 π 2
β̃1 + β̃2
mnπ 2 a2 b2
= + * 6
m4 π 4 m2 n 2 π 2 n4 π 4 m 2 2
π n 2 2
π D y
Dx 4 + 2H + Dy 4 + Dx α2 + 2
a a2 b2 b a2 b Dx
(7.350)
An additional criticism of the procedure followed in [248] would focus
on the fact that, as MT is constant on the open domain 0 < x < a,
0 < y < b, both MT,xx and MT,yy vanish in (7.347).
The relation (7.350) for the coefficients Amn still contains the unknown
constant α2 , which arises in (7.346). Assuming that
u = 0, on x = 0, a, for 0 ≤ y ≤ b
(7.351)
v = 0, on y = 0, b, for 0 ≤ x ≤ a
in which case, substitution of (7.349) and (7.352) into (7.346) and inte-
gration over the domain 0 ≤ x ≤ a, 0 ≤ y ≤ b, yields
#∞ ∞
# 2 2
1 2 m π n2 π 2 α 2 h2
Amn 2
+ 2
= (7.353)
m=1,3,... n=1,3,...
8 a b 12
Dy H a β̃2
= 0.32, = 0.2, = 20, = 0.5
Dx Dx h β̃1
Fig. 7.28 depicts the predictions for the theory delineated above of
the variation of the central deflection of the plate with changes in the
a 2 β̃ h3
1
temperature parameter θ .
h Dx
© 2001 by Chapm
The counterpart of [248] for the case of heated cylindrically orthotrop-
ic circular plates is the work of Pal [249] who treats both dynamic and
static behavior; Pal [249], as in [247], again uses, as an example, the
temperature variation (7.337). In [249], however, the correct form of the
simply supported boundary condition is used and the resultant problem
r
is treated by employing a series expansion of w in the variable ζ = .
a
For the case of polar orthotropic response, the total potential energy in
the static case assumes the form
2π a
Er h
Π= [ẽ21 − 2(k − νθ )ẽ2 ]rdrdθ
2(1 − νr νθ ) 0 0
2
2 2π a
h 2 1 1
+ w,rr + w,r + 2 w,θθ
12 0 0 r r
1 1
−2(k − νθ ) w,rr w,r + 2 w,θθ
r r
2
2Grθ (1 − νr νθ ) 1 1
− w,rθ − 2 w,θ rdrdθ
Er (k − νθ ) r r
2π a
2 νθ
− (αr + νθ αθ )e0rr + (αθ + νr αr )$0θθ T̃ rdrdθ
h 0 0 νr
(7.354)
with
ẽ1 = e0rr + ke0θθ
Grθ (1 − νr νθ ) 0 2
ẽ2 = e0rr e0θθ − (e ) (7.355a)
2Er (k − νθ ) rθ
1/2
k = (νθ /νr )
and
h/2 h/2
T̄ = δT (r, z)dz, T̃ = δT (r, z)zdz (7.355b)
−h/2 −h/2
The functions ẽ1 and ẽ2 are, respectively, the first-and second-strain in-
variants of the middle surface of the polar orthotropic plate. Introducing
the dimensionless variables
w r T̄ T̃
w̃ = , ζ = , T ∗ = , T̂ = 2 (7.356)
h a h h
and deleting the second strain invariant ẽ2 from (7.354), the Euler-
Langrange equations associated with (7.354) assume the form
1 d d 1 dw̃
ζ ζ
ζ dζ dζ ζ dζ
k 2 − 1 d2 w̃ 1 dw̃
= −
ζ2 dζ 2 ζ dζ
a 2 1 d dw̃
+12 ẽ1 ζ (7.357)
h ζ dζ dζ
1
d ∗ dw̃ d2
− (αr + νθ αθ ) ζT + (αr + νθ αθ ) 2 (T̂ ζ)
ζ dζ dζ dζ
νθ dT̂
− (αθ + νr αr )
νr dζ
dẽ1 1−k (1 − νθ )αr νr − (1 − νr )αθ νθ
+ ẽ1 = T∗
dζ ζ νr ζ
(7.358)
dT ∗
+(αr + νθ αθ )
dζ
Equations (7.357), (7.358) are, therefore, the forms assumed by the
Berger’s equations for the polar orthotropic elastic plate. In terms of
the current set of variables, the boundary conditions along the edge of
the plate at ζ = 1 assume the form
(i) Clamped Plate:
w̃ = 0, w̃,ζ = 0, at ζ = 1 (7.359a)
(ii) Simply Supported Plate:
w̃ = 0
νθ a 2
(7.359b)
w̃,ζζ + w̃,ζ + 12(αr + νθ αθ ) T̂ = 0 at ζ = 1
ζ h
The general temperature variations δT (r, z) in [249] are assumed to be
such that the dimensionless forms of (7.355b), namely T ∗ and T̂ , have
the representations
#∞
T ∗
= c∗j ζ j
j=0,2,4,...
#∞ (7.360)
∗ j
T̂ = dj ζ
j=0,2,4,...
© 2001 by Chapm
and solutions generated by of the system of Berger’s type equations
(7.357), (7.358) are then sought in the form of a series expansion
w̃ = 1 + A2 ζ 2 + A4 ζ 4 + .... (7.361)
with the solutions generated by (7.361) subject to either (7.359a) or
(7.359b); the results are also specialized to the case of an isotropic cir-
cular elastic plate, i.e., νr = νθ = ν, αr = αθ = α, Er = Eθ = E and
for both types of symmetry computations are effected in [249] for the
temperature variation given by (7.337). While the associated analytical
results are too complex to reproduce here, we do depict in Fig. 7.29,
for both the isotropic and orthotropic cases, some of the conclusions
relating temperature increases to deflections at the center of the plate
αθ νθ
for various ratios , , and T̄0 /T̄1 . It may be observed that, in the
αr νr
case when the temperature gradient through the thickness of the plate is
taken into account, the plate starts to deflect at the beginning of heating
without exhibiting the Euler buckling phenomenon at a critical temper-
ature. The case of no temperature gradient through the thickness of the
plate, which corresponds to T̂ = 0, is shown in Fig. 7.29 by the sets of
broken curves (depicting the deflection after buckling).
As we have already indicated, critiques and extensions of Berger’s
approximate method have appeared in several places in the thermal (and
load) buckling literature; we content ourselves here with reviewing just
two such pieces of work, that of Jones, Mazumdar, and Cheung [242], and
that of Banerjee and Datta [243]. The results in [248] cover the case of
dynamic (as well as static) behavior of plates at elevated temperatures,
but we will confine our attention to static situations only; the behavior
of viscoelastic plates within the context of Berger’s approximate scheme
is also discussed in [242] and this analysis will be briefly reviewed in
Chapter 8.
In [242], the authors begin by reviewing the system of Berger’s approx-
imate equations, which are written in the form (compare with (7.305),
(7.306))
MT
K 4 w + λ2 2 w = t − 2 (7.362a)
1−ν
NT 12K
− 2 $1 = λ2 (7.362b)
1−ν h
for isotropic response in rectangular Cartesian coordinates where MT =
h/2 h/2
(1 − ν)M T = αEδT zdz, NT = (1 − ν)N T = αEδT dz, and
−h/2 −h/2
$1 is given by (7.299a); pure thermal buckling corresponds, of course, to
the case in which t ≡ 0, MT = 0. For pure thermal buckling, therefore,
(7.362a) reduces to
K 4 w + λ2 2 w = 0 (7.363)
Equation (7.363) is formally identical to the equation governing the me-
chanical buckling of a rectangular plate subject to uniform compression,
i.e., Nx = Ny = N , Nxy = 0. In order, therefore, for a nontrivial solution
to exist, we must have λ2 = Ncr , Ncr being the critical buckling load.
Taking the corresponding solution for w in the form w = Λwb (x, y),
with Λ an arbitrary parameter, we note that wb is the deformation due
to biaxial loading and, thus, satisfies
Ncr 2
K 4 wb + wb = 0 (7.364)
K
By nondimensionalizing wb so that its maximum value is one, we may
interpret Λ as being the maximum plate deflection. By substituting for
w in (7.362b) the temperature deflection curve may now be obtained
from
NT 12K 1
Ncr = − 2 (u,x + v,y + Λ2 | wb |2 ) (7.365)
1−ν h 2
In fact, for a plate whose edges are restrained in the plane of the plate
(7.365) yields
2b 2a
NT
2 − Ncr dxdy
Λ 1−ν
= 0 0
2b 2a (7.366)
h
6K | wb | dxdy
2
0 0
© 2001 by Chapm
To assess the accuracy of the Berger’s approximation technique, the
authors [242] consider the rectangular plate described above subject to
a temperature variation of parabolic type, i.e.,
2
2
x−a y−b
δT (x, y, z) = T0 + T1 1 − 1− (7.369)
a b
which are precisely the same as the equations for thermal deflections of
a plate strip as derived by Williams [250].
A highlight of the work presented in [242] is the conclusion that the so-
lution of Berger’s equations, for the response of thermally heated plates,
represents the first term in a perturbation solution of the generalized von
Karman equations; this conclusion is arrived at in the following manner:
We write (in rectangular Cartesian coordinates) the governing equations
for the classical analysis of thermally heated isotropic plates in the form
Nx,x + Nxy,y = 0 (7.371a)
Ny,y + Nxy,y = 0 (7.371b)
where e0 is e = $0xx + $0yy , based on the functions u0 (x, y) and v0 (x, y).
© 2001 by Chapm
(ii) For n = 0:
K 4 wn − K̂hen 2 wn
+K̂h[$(n−1)xx wn,xx + $(n−1)yy wn,yy
+2$(n−1)xy wn,xy ] = 0
∂en ∂ ∂ (7.374b)
−
$(n−1)xx + $(n−1)xy = 0
∂x ∂x ∂y
∂en ∂ ∂
− $(n−1)yy + $(n−1)xy = 0
∂y ∂y ∂x
NT
K̂he0 − = λ2 (λ2 const.) (7.375)
1−ν
If the edges of the plate are restrained in the plane of the plate, then
the first equation in (7.374a) becomes
EαT MT
K w0 +
4
h w0 = t −
2 2
(7.376)
1−ν 1−ν
We note that the first relation in (7.374a) and (7.375) are formally equiv-
alent to (7.305), (7.306), i.e., the solution generated by Berger’s approx-
imation method is formally the same as that generated by the zeroth
order perturbation solution of the generalized von Karman system for
thermal deflection of an isotropic plate.
Finally, we consider the work of Banerjee and Datta [243]; in this
paper, a modified energy expression is developed and a new system of
governing differential equations, in decoupled form, is obtained for thin
plates undergoing large deflections. The equations obtained are tested
for circular and square plates with immovable as well as movable edge
conditions under uniform static loads, and an extension to cover the
case of thermal loading is outlined. For both movable and immovable
edges the results obtained appear to be in excellent agreement with other
known results for the same test problems considered. The basic motiva-
tion for the work presented in [243] appears to be the analysis reported
in Nowinski and Ohnabe [241], who establish that the Berger’s approxi-
mation leads to meaningless results for problems involving movable edge
conditions.
For a thin isotropic circular plate of radius b the total potential energy
Π under isothermal conditions is given by
2 2 2
K b d w 2ν dw 1 dw
Π= + +
2 0 dr2 r dr r2 dr
b (7.377)
12
+ 2 e21 + 2(ν − 1)e2 rdr − twrdr
h 0
where e1 and e2 are, respectively, the first and second invariants of the
middle surface strains, i.e.,
2
du 1 dw u
e1 =
+ +
dr 2 dr r
2
(7.378)
e = u du + 1 dw
2 r dr 2 dr
with 2
du u 1 dw
ē1 = +ν + (7.380)
dr r 2 dr
u2
It is now noted that if the expression (1 − ν 2 ) 2 in (7.379) is replaced
4 r
1 dw
by λ , with λ a factor depending on ν, decoupling of (7.379)
4 dr
is possible; carrying out this replacement and using the usual ideas of
the variational calculus, one obtains the following system of decoupled
differential equations from the energy functional Π:
12 d2 w ν dw
w− 2
4
+ Arν−1
h dr2 r dr
2 (7.381a)
6λ dw d2 w t
− 2 w+2 2 =
2
h dr dr K
© 2001 by Chapm
with A a constant of integration to be determined from
2
du u 1 dw
+ν + = Arν−1 (7.381b)
dr r 2 dr
∂ 2 w ∂w ∂ 2 w ∂w t
+2 2
+ 2
=
∂x ∂x ∂y ∂y K
and
2 2
∂u ∂v 1 ∂w v ∂w
A= +ν + + (7.382b)
∂x ∂y 2 ∂x 2 ∂y
∂ 2 w ∂w ∂ 2 w ∂w
+2 +
∂x2 ∂x ∂y 2 ∂y
2 w Eαf (h)
+12T0 α λ(1 − ν 2 ) =− 2 T1
h2 K(1 − ν)
(7.385)
h/2 h/2
where f (h) = zg(z)dz and it is assumed that g(z)dz = 0.
−h/2 −h/2
The constant of integration A is determined by the following thermal
analogue of (7.382b):
2 2
∂u ∂v 1 ∂w v ∂w
+ν + + − (1 + ν)αT0 = A (7.386)
∂x ∂y 2 ∂x 2 ∂y
It does not appear that the equations (7.385), (7.386) have been applied
to any sample problems involving thermal buckling of isotropic rectan-
© 2001 by Chapm
gular plates; nor does it appear that the analogous equations have been
developed for the case of rectilinear orthotropic symmetry.
7.4 Thermal Bending, Buckling, and Postbuckling
Figures, Graphs, and Tables I
Table 7.1 Buckling Loads for the Circular Plate (with permission of
the Taylor & Francis Group plc)
l1 l2 l3 Temperature Distribution
0 0.63 26.6
l4
1 0.59 99.2
l5 l6
2 0.67 20.0
l4
3 0.69 47.1
l5 l6
© 2001 by Chapm
Table 7.2 Critical Buckling Temperature for a Rectangular
Plate (Adopted, in modified form, from [216])
b2 EaT0cr t
Terms retained
π2 D
a11 6.35
© 2001 by Chapm
FIGURE 7.2
Plate containing a heated region AT . (Adopted, in modified
form, from [217].)
FIGURE 7.3
Buckling of a heated plate strip. (From Boley, B.A. and Weiner, J.H.,
Theory of Thermal Stresses, John Wiley & Sons, Inc., N.Y., 1960.
With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 7.4
Critical buckling temperature coefficients
αE kπ 2 E h 2
2
Tcr = 12(1−ν 2 ) ( b ) [221]. (From Boley, B.A. and Weiner, J.H., The-
ory of Thermal Stresses, John Wiley & Sons, Inc., N.Y., 1960. With per-
mission of Elsevier Science.)
FIGURE 7.5
Interaction of temperature level and applied stress in plate buckling.
(From Boley, B.A. and Weiner, J.H., Theory of Thermal Stresses,
John Wiley & Sons, Inc., N.Y., 1960. With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 7.6
Deflection w at the center of the rectangular plate; poisson ratio ν = 1/4.
(From Boley, B.A. and Weiner, J.H., Theory of Thermal Stresses,
John Wiley & Sons, Inc., N.Y., 1960. With permission of Elsevier Science.)
FIGURE 7.7
Distribution of the bending moment Mx in a square plate;
NT /(NT )cr = 0.25, ν = 1/4. (From Boley, B.A. and Weiner,
J.H., Theory of Thermal Stresses, John Wiley & Sons, Inc., N.Y.,
1960. With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 7.8
Distribution of the bending moment Mx in a square plate;
N T /(NT )cr = 0.6, ν = 1/4. (From Boley, B.A. and Weiner, J.H.,
Theory of Thermal Stresses, John Wiley & Sons, Inc., N.Y., 1960.
With permission of Elsevier Science.)
FIGURE 7.9
Bending moment Mx at the center of the rectangular plate. (From Boley,
B.A. and Weiner, J.H., Theory of Thermal Stresses, John Wiley &
Sons, Inc., N.Y., 1960. With permission of Elsevier Science.)
© 2001 by Chapm
FIGURE 7.10
Deflection at the center of the buckled plate. (From Boley, B.A. and
Weiner, J.H., Theory of Thermal Stresses, John Wiley & Sons, Inc.,
N.Y., 1960. With permission of Elsevier Science.)
FIGURE 7.11a
The thermal buckling problem of ref. (Adopted, in modified
form, from [216].)
FIGURE 7.11b
The tentlike temperature distribution. (Adopted, in modified
form, from [216].)
© 2001 by Chapm
FIGURE 7.12
Variation of the normal stress σx . (Adopted, in modified form,
from [233].)
FIGURE 7.13
Small-deflection buckle pattern in one quadrant of a plate of
aspect ratio 1.57. (Adopted, in modified form, from [216].)
© 2001 by Chapm
FIGURE 7.14
Comparison of calculated and experimental deflections at the
plate center. (Adopted, in modified form, from [216].)
FIGURE 7.15a
Longitudinal center line deflection. (Adopted, in modified
form, from [216].)
FIGURE 7.15b
Growth of deflections with temperature. (Adopted, in modi-
fied form, from [216].)
© 2001 by Chapm
FIGURE 7.16
Rectangular plate with edge restraint in the plane of the plate
[234].
FIGURE 7.17
Center deflection versus lateral load parameter [234].
© 2001 by Chapm
FIGURE 7.18
Effect of edge spring rate on buckling temperature ratio [234].
FIGURE 7.19
Effect of edge fixity on buckling temperature [234].
© 2001 by Chapm
FIGURE 7.20
a
Center deflection vs. temperature Ratio, b
= 1 [234].
FIGURE 7.21
Variation of non-dimensional deflections wo /h for different val-
ues of the temperature parameter [236]. (From Biswas, P.,
Nonlinear Analysis of Heated Orthotropic Plates, Indian J.
Pure Appl. Math, 1380, 12, 1981. By permission of the Indian
National Science Academy, New-Delhi)
© 2001 by Chapm
FIGURE 7.22
Simply supported rectangular plate with a non-zero thermal
moment.
FIGURE 7.23
Relation between the temperature rise and the deflection (sim-
ply supported edge) [245]. (From Pal, M.C., Acta Mechanica,
8, 82, 1969. With permission.)
© 2001 by Chapm
FIGURE 7.24
Relation between the temperature rise and the deflection
(clamped edge) [245]. (From Pal, M.C., Acta Mechanica, 8,
82, 1969. With permission.)
FIGURE 7.25
Relation between the temperature rise and the deflection;
b/a = 0.4 with edges simply supported [245]. (From Pal, M.C.,
Acta Mechanica, 8, 82, 1969. With permission.)
© 2001 by Chapm
FIGURE 7.26
Relation between the temperature rise and the deflection;
b/a = 0.4 with both edges clamped [245]. (From Pal, M.C.,
Acta Mechanica, 8, 82, 1969. With permission.)
FIGURE 7.27
Relation between the temperature rise and the deflection;
b/a = 0.4 with the inner edge clamped and the outer edge
simply supported [245]. (From Pal, M.C., Acta Mechanica, 8,
82, 1969. With permission.)
© 2001 by Chapm
FIGURE 7.28
Variation of central deflections with the temperature Parame-
ter [246]. (Biswas, P., Large Deflections of Heated Orthotropic
Plates, Indian J. Pure Appl. Math, 1027, 9, 1978. By permis-
sion of the Indian National Science Academy, New-Delhi)
FIGURE 7.29
Variation of central deflections with and without a temperature
gradient [249]. (Reprinted from the Int. J. Nonlinear Mech., 8,
Pal, M.C., Static and Dynamic Non-Linear Behavior of Heated
Orthotropic Circular Plates, 489-504, 1973, with permission
from Elsevier Science)
© 2001 by Chapm
FIGURE 7.30
Comparison of results for central deflection vs. tem-
perature rise; b/a = 1.0, 1/3, β = 1/(α(1 − ν 2 ))(h/b)2 [242].
(Reprinted from the Int. J. Solids and Structures, 16, Jones,
R., Mazumdar, J., and Y.K. Cheung, Vibration and Buckling
of Plates at Elevated Temperatures, 61-70, 1980, with permis-
sion from Elsevier Science)
Chapter 8
where in the case, e.g., of thermal expansion, the strain tensor compo-
nents $xx , $xy , and $yy are related to the Cartesian stress components
σxx , σxy , and σyy by that special case of (6.4), which has the following
© 2001 by Chapm
form for isotropic response:
1
$xx = (σxx − νσyy ) + αT
E
1
$yy = (σyy − νσxx ) + αT (8.2)
E
$ = 1 + νσ
xy xy
E
In lieu of (8.2) we have, in terms of the Airy function Φ, as defined
by (6.20),
$0 = 1 (Φ − νΦ ) + αT
xx ,yy ,xx
E
1
$0yy = (Φ,xx − νΦ,yy ) + αT (8.3)
E
$0xy = − (1 + ν) Φ,xy
E
Substitution of (8.3) into (8.1) then yields, for the case where an
imperfection wi = wi (x, y) is present in the plate,
2 2 2
∂2w ∂w ∂w
Φ = Eh
2
−
∂x∂y ∂x ∂y
(8.4)
2 2 2
∂ wi 2 ∂ wi ∂ wi
−( ) + − (1 − ν)∆N T
∂x∂y ∂x2 ∂y 2
and
wi
K 1 − c ∆2 w = Φ,yy w,xx − 2Φ,xy w,xy + Φ,xx w,yy (8.7b)
wc
and
1
Φ∗ = Φ (8.9)
h
so that Φ∗ (x, y) is the Airy function associated with the stress tensor
components σxx , σxy , and σyy instead of the stress resultants Nx , Ny ,
αEh
and Nxy . We also assume that δT,z = 0 so that N T = δT (x, y).
1−ν
Employing this latter assumption and (8.8), (8.9) in (8.7a,b), and noting
that αE = α∗ E ∗ , it is easily seen that the governing equations may be
written in the form
2
2 ∗ ∗ ∂2w ∂2w ∂2w
∆ Φ =E − − α∗ E ∗ 2 (δT ) (8.10a)
∂x∂y ∂x2 ∂y 2
K∗
∆4 w = Φ∗,yy w,xx + Φ∗,xx w,yy − 2Φ∗,xy w,xy (8.10b)
h∗
© 2001 by Chapm
Equations (8.10a,b) are identical to the large-deflection equations for
thermal buckling of a flat plate with Young’s modulus E ∗ , thermal ex-
pansion coefficient α∗ , and thickness h∗ . If the initial deflection wi sat-
isfies the same homogeneous boundary conditions as would be satisfied
by the deflection of an initially flat plate, then the solutions of (8.10a,b)
are identical with those for a flat plate of the same aspect ratio but with
E replaced by E ∗ , α by α∗ , and h by h∗ . Thus, in order to analyze an
initially imperfect plate, a flat plate having the same aspect ratio may
be analyzed with the quantities E, α, and h being left arbitrary; then,
everywhere in the resulting expression for the solution, one may replace
2* wi
E by E(1 − {wi2c /wc2 }), α by α/(1 − {wi2c /wc2 }), and h by h 1+ c.
wc
This relatively simple procedure, introduced in [216], then yields the
stresses and deflections for the initially imperfect plate.
For the problem of thermal buckling of a flat isotropic simply sup-
ported rectangular plate of aspect ratio 1.57, subjected to a tentlike
temperature distribution, the relationship between the temperature dif-
ferential T0 and the center deflection wc of the plate was determined to
be given by (7.214). Noting that the plate thickness h also appears in
the expression for the bending stiffness
( K = Eh2 /12(1−ν 2 ), and replac-
w2 2
ing α and h, respectively, by α 1 − icc and h 1 + (wic /wc ) in
w2
(7.214) yields the relation
2 wc − wic
2 2
b2 Eαh wic
T 0 = 5.39 1 − + 1.12(1 − ν ) (8.11)
π2 D wc h2
The Young’s modulus E need not be replaced by E ∗ inasmuch as the
relation (7.214) is independent of E.
The method described above was employed in [216] to compute curves
of center deflection versus average edge compressive stress for the prob-
lem of bending of a square isotropic elastic plate with initial imper-
fections under applied compressive edge stresses; the curves which re-
sulted from this technique turned out to be in excellent agreement with
numerical results obtained from an approximate solution of the von
Karman large deflection equations for initially imperfect plates, with
the agreement existing for all cases in which the initial imperfection
was a half-sine wave deflection in both the x and y directions. Ex-
perimental and theoretical plate center deflection differences wc − wic
have been plotted in Fig. 7.14 as a function of the temperature dif-
ferential T0 with the theoretical curve based on (8.11); on this graph
α = 0.127 × 10−4 in./in. − oF, h = 0.25in., b = 11.25in., wic = 0.045in.,
and ν = 0.33 so that (8.11) has the explicit form
0.045
T0 = 194.1 1 − + 573.6(wc2 − (0.045)2 ) (8.12)
wc
1 2
✸4 (f, g) ≡ ( f )(2 g) + 2 (f 2 g + g 2 f )
2
1
− {4 (f g) + f 4 g + g 4 f }
4
However, for all sufficiently smooth pairs (f, g) it follows that, in fact,
✸4 (f, g) = [f, g]
1 1
✸ (f, g) ≡ [f, g] = f,rr
4
g,r + 2 g,θθ
r r
1 1
−2 f,θ g,θ (8.14)
r r
,r
,r
1 1
+g,rr f,r + 2 f,θθ
r r
and the following special results may be noted:
1
(i) [r, g] = g,rr
r
(ii) [r2 , g] = 2 2 g
1 1
(iii) [r , g] = n(n − 1)r
n n−2
g,r + 2 g,θθ
r r
+nrn−2 g,rr (8.15)
1
(iv) [rh(θ), g] = g,rr (h + h,θθ )
r
2 1
(v) [θ, g] = 2 g,θ
r r ,r
© 2001 by Chapm
where δT is taken, in [253], to be the temperature variation through
the thickness of the plate. The strain compatibility equation (for small
deflections) then assumes, in terms of the Airy function φ(x, y), the form
1 1
(φ,xx − νφ,yy ) + (φ,yy − νφ,xx )
h
,xx h ,yy
(8.19)
1
+2(1 + ν) φ,xy + αE (δT ) = 0
2
h ,xy
© 2001 by Chapm
problems for rectangular plates with an exponential variation in thick-
ness and to circular or annular plates whose thickness varies as a power
of the radius.
Consider a rectangular plate such that
2γπ
4 w + (w,xxy + w,yyy )
a
(8.33)
γ 2 π2 t̄
(w,yy + νw,xx ) = exp (−γπy/a)
a2 K0
with
t̄ = t − (1 + ν) 2 (KκT ) (8.34)
We suppose that the edges of the plate at x = 0 and x = a are simply
supported and that the distributed applied loading is such that
∞
# mπx
βπy/a
t̄ = e qm sin (8.35)
m=1
a
d4 w1 π2 d2 w1
4
+ 2 2β(β − γ) + νγ 2
dx a dx2
4
π 2 1 #
∞ mπx (8.36)
+ 4 β (β − γ) w1 =
2
qm sin
a K0 m=1 a
a4 #
∞
qm mπx
w1 (x) = sin (8.37)
K0 π 4 m=1 (m2 + βγ − β 2 )2 − νm2 γ 2 a
On the other hand, substitution of the sum in (8.35) into (8.33) yields
d4 Ym d3 Ym d 2 Vm
a4 4
+ 2γπa3 3
+ π 2 a2 (γ 2 − 2m2 )
dy dy dy 2 (8.38)
dY m
−2π 3 γm2 a + π 4 m2 (m2 − νγ 2 ) = 0
dy
#
4
Ym = Am,j exp (rm,j πy/a) (8.39)
j=1
and substituting into (8.38) we find that the rm,j must be the roots of
the quartic equation
4
rm 3
+ 2γrm + (γ 2 − 2m2 )rm
2
− 2γm2 rm + m2 (m2 − νγ 2 ) = 0 (8.40)
while the Am,j are determined by the boundary conditions along the
plates edges at y = 0 and y = b.
An analysis similar to that outlined above may be effected to ana-
lyze the plane stress distribution in a rectangular heated plate with an
exponential variation in thickness. Indeed, if h = h0 exp (−λπy/a), a
solution of (8.20), with 2 (δT ) = 0, may be obtained in the form
sin mπx
#∞
φ= Ỹm
cos a
m=1
# 4 (8.41)
sm,j πy/a
Ỹm = Bm,j e
j=1
© 2001 by Chapm
with the sm,j the roots of
s4m + 2λs3m + (λ2 − 2m2 )s2m )s2m − 2λm2 sm + m2 (m2 + νλ2 ) = 0 (8.42)
where
#
4
Rk = Ak,i ργk,i (8.48)
i=1
with the γk,i the roots of the bi-quadratic
Γ(γk ) = L − L{2k + β(1 − ν)}
2 2
+k 2 {k 2 + β(β − 1)(1 − ν) − (β − 2)2 } (8.49)
L = γk (γk + β − 2)
For a circular plate k assumes successive integral values, while for a
sector plate subtending an angle θ0 k = nπ/θ0 . If t̄ in (8.46) can be
expressed in the form of a double summation with respect to arbitrary
parameters χ and k, i.e.,
##
sin
t̄ = t̄χ,k ρx kθ (8.50)
cos
χ k
© 2001 by Chapm
It may be shown that (8.54) yields a stress distribution which produces
a self-equilibrating system of middle surface forces and which satisfies,
therefore, the condition that the edges of the plate are free. Substitution
1
of (8.54), (8.52), and w(r) = − κr2 into (8.28), with 2 (δT ) = 0, then
2
1 2
shows that w = − κr , and the expression in (8.54) for φ, yields a
2
solution of the large deflection problem provided
(1 − ν)r04 κ3
κT − κ = (8.55)
2(7 + ν)h30
In [254], the author gives 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; also examined in [254] is the
behavior of such a plate with an initial spherical curvature, when the fur-
ther possibility of snap-through buckling exists. In the work considered,
the plate thickness and rigidity once again vary, as in (8.52). With the
∂
temperature gradient variation (δT ) through the thickness constant,
∂z
each unrestrained element of the plate would assume a uniform spherical
∂(δT ) αT1
curvature κT = α ≡ where T1 is the temperature difference
∂z h0
across the thickness of the plate at the center of the plate; also, for a
plate subject to such a temperature distribution the deflection of the
plate assumes the form
w = ar2 {1 + b cos 2(θ − θ0 )} (8.56)
with a, b constants depending on the magnitude of κT and the initial
spherical curvature κ0 . If the plate deforms with rotational symmetry,
1
then w assumes the form given in [253], i.e., w = − κr2 ; otherwise
2
(8.56) may be written in the form
1
w = − (κx x2 + κy y 2 ) (8.57)
2
where κx , κy are independent of x and y and are the curvatures in the
x and y directions, respectively. For a suitable choice of the arbitrary
angle θ0 in (8.57), one may achieve κx ≥ κy . Thus, the deflection of
the plate is completely determined by the curvatures κx and κy with
κx = κy = κ in the case of rotational symmetry. Another feature of the
problem under consideration is that while the magnitude of the middle-
surface stresses depends on κT and κo , the distribution of these stresses
depends only on r; in fact, the middle surface forces may be derived, in
the usual way, from an Airy function φ with
1
hσrr = Nr = φ,r
r (8.58)
hσθθ = Nθ = φ,rr
σrθ = 0
where
φ ∼ (1 − ρ2 )3 (8.59)
As φ varies with r in the same way that K does, and φ and K have the
same dimensions, we may write that
φ(r) = βK(r) (8.60)
so that the distribution of middle-surface stresses is determined once the
value of β is known.
b b
The bending stresses σxx and σyy are more conveniently referred to
Cartesian coordinates; they vary linearly through the thickness of the
1
plate with their values on the surface z = h given by
2
b 6 b 6
σxx = 2 Mx and σyy = 2 My (8.61)
h h
The moments Mx , My per unit length are given by
Mx = −K{w,xx + νw,yy + (1 + ν)κT }
(8.62)
My = −K{w,yy + νw,xx + (1 + ν)κT }
so that by combining (8.61), (8.62) we obtain
Eh
b
σxx = [κx + νκy − (1 + ν)κT ]
2(1 − ν 2 )
(8.63)
Eh
σyy
b
= [κy + νκx − (1 + ν)κT ]
2(1 − ν 2 )
At ρ = 1 (the edge of the plate) there are no moments or forces applied;
these conditions are automatically satisfied because the variation of K
(and φ, by virtue of (8.60)) is such that K = K,r = 0 at ρ = 1.
It is convenient to introduce the following non-dimensional quantities,
which are employed in [254]:
r02
{κ̂T , κ̂, κ̂x , κ̂y , κ̂0 } = {κT , κ, κx , κy , κ0 }
h0
(8.64)
r2
{σ̂xx , σ̂yy , σ̂θθ } = 0 2 {σxx , σyy , σrr , σθθ }
Eh0
© 2001 by Chapm
With a small temperature gradient through the thickness of the plate,
the plate deforms into a shallow “saucer” with constant spherical curva-
ture κ; this spherical curvature is initially governed by small deflection
theory so that κ = κT and the plate is stress free. However, as κ in-
creases, middle-surface stresses are developed which stiffen the plate so
that the curvature κ becomes less than κT , which results in the forma-
tion of bending stresses. At a critical value κT = κ∗T middle-surface
stresses dominate the process and for κT > κ∗T the plate is forced into
a shape which is no longer rotationally symmetric. For κT >> κ∗T the
plate approximates a developable surface with parallel generators, i.e.,
the plate curls up about a diameter. To illustrate the development of
this process, we use the large-deflection equations for a plate of variable
thickness, i.e., (8.27), with t ≡ 0, and (8.28), with 2 (δT ) = 0, and will,
henceforth, refer to these respective specializations of (8.27) and (8.28)
as (8.27) and (8.28); these equations are coupled and nonlinear.
However, by virtue of the ansatz (8.60)
1 2 βK0 ∂ 2 1 ∂
2
φ = +
h h0 r04 ∂ρ2 ρ ∂ρ
2
1 ∂ 1 ∂
× + (1 − ρ 2 3
) (8.65)
1 − ρ2 ∂ρ2 ρ ∂ρ
βK0
= 144
h0 r04
In a similar fashion,
1 βK0 1 ∂ 2 1 ∂
,φ = (1 − ρ2 )3
h 4
h0 r0 ρ ∂ρ 2 1−ρ 2 ∂ρ
∂2 2 3 ∂ 1 (8.66)
+ 2 (1 − ρ )
∂ρ ∂ρ 1 − ρ2
βK0
= −24
h0 r04
1
Also, as w = − (κx x2 + κy y 2 ), we have
2
1
[w, w] = κx κy (≡ const.) (8.67)
2
With the assumptions that have been made relative to the form of w
and φ, (8.27) assumes the non-dimensional form
λβ + (1 + ν)κ̂x κ̂y = 0 (8.68)
with λ = 2(7 + ν)/(1 − ν). In [254], the author first treats the case in
which κx = κy = κ; for that situation
2 w = −2κ (8.69)
while
[K, w] = −κ 2 K
(8.70)
[Φ, w] = −βκ 2 K
© 2001 by Chapm
Equation (8.77) may be written in dimensional form so as to express the
deflection at the edge of the plate in terms of h0 , i.e.
1
w|ρ=1 = − κ̂h0 −1.67h0 (8.78)
2
The prebuckling middle surface stresses are given by
(1 − ρ2 )κ̂2
σ̂rr = 4(7 + ν)
(8.79)
(1 − 5ρ2 )κ̂2
σ̂θθ =
4(7 + ν)
while the bending stresses are
1 − ρ2 )κ̂3
b
σ̂xx b
= σ̂yy =− (8.80)
4(7 + ν)
Thus, the middle surface stresses vary as the square of the plate curva-
ture, while the bending stresses vary as the cube of the plate curvature;
the variation of the stress distribution with the magnitude of the tem-
perature gradient is complicated by virtue of the non-linear variation
of κ with κT . To determine the postbuckling behavior of the plate, we
solve (8.68) and (8.73) so as to obtain
β = −(1 − ν)
% &
1
κ̂x = (1 + ν) κ̂T + (κ̂2T − κ̂∗2
T )
1/2
2 (8.81)
% &
1
κ̂ = (1 + ν) κ̂ + (κ̂2 − κ̂∗2 )1/2
y T T T
2
The minimum value of κ̂∗T for which the solution (8.81) exists is κ̂∗T ,
and, at this value, the solutions represented by (8.74), (8.75), and (8.81)
coincide. If |κ̂T | > κ̂∗T then the plate strain energy for the asymmetrical
mode of deformation is less than that for the rotationally symmetric
mode so that the correct solution is given by (8.81) and the solution
given by (8.74), (8.75) then represents an unstable state.
For the postbuckling solution given by (8.81), the fact that β is con-
stant implies that the middle-surface stresses are independent of κT and
are given by
1 − ρ2
σ̂rr =
2(1 + ν)
(8.82)
1 − 5ρ2
σ̂θθ =
2(1 + ν)
while the bending stresses assume the form
1 % &
σ̂xx
b
= − (1 − ρ2 ) κ̂T − (κ̂2T − κ̂∗2
T )1/2
4
% & (8.83)
σ̂ b = − 1 (1 − ρ2 ) κ̂ + (κ̂2 − κ̂∗2 )1/2
yy T T T
4
For |κ̂T | >> κ̂∗T , (8.81) together with (8.83) yield the asymptotic results
κ̂x → (1 + ν)κ̂T + O(1/κ̂T )
(8.84)
κ̂y → 0 + O(1/κ̂T )
and b
σ̂xx → 0 + O(1/κ̂T )
1 (8.85)
σ̂yy
b
→ − (1 − ρ2 )κ̂T + O(1/κ̂T )
2
The asymptotic results delineated in (8.84) and (8.85) are in agreement
with analogous results for inextensional plate theory as given, e.g., in
Mansfield [260]. The variation of the principal plate curvatures with the
temperature gradient through the thickness of the plate is depicted in
Fig. 8.1, while the variation of both the middle surface stresses and the
bending stresses with the temperature gradient through the thickness of
the plate is shown in Fig. 8.2.
For the case in which the circular lenticular plate has a uniform spher-
ical curvature κ0 when κT is zero, and is initially stress-free, the govern-
ing differential equations, assuming an initial deflection wi , are (8.29a),
with 2 (δT ) = 0, and (8.29b), with t ≡ 0, i.e.
1 2 1
2 φ − (1 + ν) ,φ
h h (8.86a)
1
+ E([w, w] − [wi , wi ]) = 0
2
and
2 (K 2 (w − wi )) − (1 − ν)[K, w − wi ]
(8.86b)
+(1 + ν) 2 (KκT ) − [φ, w] = 0
1
wi = − κ0 r02
2
© 2001 by Chapm
1
Taking, again, trial solutions φ = βK, w = − (κx x2 + κy y 2 ), (8.86a)
2
reduces to the (non-dimensional) form
depending on the relative magnitudes of κ̂0 and κ̂T , (8.90b) has either
one or three real roots. Following the work in [254], we set
and note that for J < 0, (8.90b) possesses three real roots; the (alge-
braically) largest and smallest roots correspond to stable configurations,
while the middle root is associated with an unstable configuration. When
κ̂T = 0 there are two stable states for |κ̂0 | > 2λ1/2 , which are given by
κ̂ = κ̂0
% & (8.92)
κ̂ = − 1 κ̂0 + (κ̂20 − 4λ)1/2
2
It is noted, in [254], that when κ̂ and κ̂0 + κ̂T are of opposite sign, and
J = 0, snap-through buckling of the plate will occur and the curvature
1
κ̂ will jump from the unstable value of κ̂ = − {4λ(κ̂0 + κ̂T )}1/2 to the
2
stable value of κ̂ = {4λ(κ̂0 + κ̂T )}1/2 . For J > 0, (8.90b) has only one
real root, but the associated plate configuration is stable only if K < 0
where
1 2(7 + ν)
K = (1 + ν)2 (κ̂0 + κ̂T )2 − κ̂20 − (8.93)
4 1+ν
When K > 0, the solution of (8.87), (8.89) is given by
β = −(1 − ν)
1
κ̂x = (1 + ν)(κ̂0 + κ̂T ) + K 1/2 (8.94)
2
κ̂y = 1 (1 + ν)(κ̂0 + κ̂T ) − K 1/2
2
Therefore, after buckling into a mode which is asymmetrical, the middle-
surface forces remain constant, and independent of κ̂0 ; also, for large
values of |κ̂T | the plate configuration may be approximated by a devel-
opable surface for which
κ̂x → (1 + ν)(κ̂0 + κ̂T ) + O(1/κ̂T )
(8.95)
κ̂y → 0 + O(1/κ̂T )
© 2001 by Chapm
plate is subjected to uniform compression along its edges as well as to a
stationary temperature distribution of the form
in which case,
Eh30
K(r) = r3m ≡ K0 r3m , m < 1 (8.101)
12(1 − ν 2 )
b r
Setting β = , and ρ = , (so that the outer and inner boundaries of
a a
the annular plate correspond, respectively, to ρ = 1 and ρ = β), and
employing (8.94) in the equilibrium equation
© 2001 by Chapm
plastic plates at elevated temperatures, few studies exist that are con-
cerned with the deflection or buckling behavior of viscoelastic or plastic
plates which are thermally loaded. For viscoelastic plates, or plates
subject to temperature-dependent creep buckling, notable contributions
have been made by Ross and Berke [261], Jones and Mazumdar [242],
and Das [262], which will be summarized below. A significant piece of
work in the literature, which is related to the thermal buckling of plates
in the plastic regime, is the paper of Williams [263]; in this paper, the
field equations associated with Neale’s variational theorem [264], [265]
are developed and applied to the problem of thermal buckling and post-
buckling of constrained rectangular plates; the resulting equations are a
generalization of von Karman’s equations in rate form, and, among the
results, which will be described in this section, is the fact that, in the
immediate vicinity of a critical point, the theory predicts a substantial
reduction of the buckling temperature due to plasticity effects. We be-
gin our analysis of these problems by first looking at viscoelastic and
temperature-dependent creep buckling.
In [261], the authors employ the so-called Norton-Bailey power law
for material creep to predict the time-dependent lateral deflection of flat
rectangular plates with a through-thickness steady-state temperature
distribution. The usual Norton-Bailey or power creep law (e.g., Norton
[266]) has the form
$̇ = kσ n (8.108)
where $̇ is the creep strain rate, k is the creep constant, and σ is the
stress. In considering temperature variations, one modifies (8.108) so
that it assumes the form given by Maxwell’s law [267], i.e.,
$̇ = ke−H/RT σ n (8.109)
with H the creep activation energy, R the universal gas constant, and T
the absolute temperature. To study the creep buckling of plates, (8.109)
is replaced by the two-dimensional equations
1
$̇i = kn e−H/RT J2n−1 (2σi − σj ) (8.110)
2
with the subscripts i, j denoting xx, yy in cyclic substitution, σxx , σyy
being the usual in-plane stresses in the x and y directions, respectively,
while J2 is the stress invariant
J2 ≡ σxx
2 2
+ σyy − σxx σyy (8.111)
In [261] a comparison of creep buckling predictions is made for the two
cases where the creep exponent n in (8.110) assumes the values n = 3
and n = 5. For a simply supported flat rectangular plate, such as is
depicted in Fig. 8.7, it is assumed that the deflected shape may be
adequately represented by a half-wave cosine function in each of two
perpendicular directions at all times, in which case, the plate behavior
is determined by the plate deflection at the center of the plate. If the
plate possesses an initial imperfection given by its value at the plate
center, say, wi0 , and a temperature differential is applied between the
sides of the plate, in the thickness direction, the center deflection w0
due to that temperature differential may, as in [268], be approximated
by
α(T0 − Ti )(1 + ν)ab
w0T = (8.112)
π3 h
where Ti , T0 are, respectively, the inner-and outer-plate temperatures.
The central deflection after the application of a through-thickness tem-
perature differential, but prior to the application of in-plane stresses, is
taken to be given by
w0c = wi0 + w0T (8.113)
In [261] an axial force at time t = 0 is applied to the plate. The imme-
diate deflection w0p , which is attained upon adding the axial load, has
been determined, in [269], to have the value
σE
w0p = w0c (8.114)
σE − σ
with σ the applied axial stress and σE the Euler buckling stress; for a
plate (which is isotropic, as we have assumed here) subject to an in-plane
compressive load
2
π2 E b a h2
σE = + (8.115)
12(1 − ν 2 ) a b a2
If the applied stress σ is larger than the material yield stress, that stress
is used in (8.114) in lieu of σE . The stresses and lateral displacements in
πx πy
[261] are assumed to be represented as products of cos cos , e.g.,
a b
πx πy
w(t) = w0 (t) cos cos (8.116)
a b
where w0 (t), the deflection of the center of the plate at time t, is taken
to be the sum of a small deflection ws (t) and a large deflection wl (t);
© 2001 by Chapm
it has been shown, in [270], that for the creep exponent n = 3 these
components of the plate deflection are given by the following expressions
in which Γ = b2 /a2 and k3 is the material creep constant for n = 3:
2
ws (t) 36k3 σ 3 te−2H/3RT0 [e−H/3RT0 + e−H/3RTi ] b
ln =
w0p −H 1 1 π
h2 (2Γ2 + 2Γ + 1] 1 + exp −
3R T0 Ti
(8.117)
and
wl (t) =
4π h2 w0p (Γ2 + Γ + 1)
*
3 16h2
π 2 (Γ2 + Γ + 1) − 81b2 k3 w0p2 σ 3 t e−H/RT0 + e−H/RTi
9
(8.118)
For creep exponent n = 5, on the other hand,
ws (t) 2160k5 σ 5 t 1
ln p = 2
· 2
w0 h (10Γ + 10Γ + 4)
−4H/5RT −H/5RT
−H/5RTi
2 (8.119)
e 0
(e 0
+e ) b
× 1 1 4
π
1+e
−H/5R T − T
0 i
and
1 1 18, 225 k5 σ 5 t 1
p4 − 4
= · 6
· 2
w0 w(t) 32 h (Γ + Γ + 1)3
(8.120)
b 2
× e−H/RT0 + e−H/RTi
π
The predictions embodied in (8.117)–(8.120) were studied, numeri-
cally, in [261] for specific values of the various parameters involved, e.g.,
k3 = 3.05 × 108 /M P a · sec and k5 = 5.71 × 105 /M P a · sec. In Figs.
8.8–8.10 we indicate the type of predictions which follow from the work
in [261], for the lateral deflections as functions of time, and for applied
axial stresses of 6.9, 13.8, and 34.5 MPa, respectively; these figures, for
given values of ∆T and σ, compare the predictions with respect to lat-
eral deflection which are made by the creep power law (8.110) when the
creep exponent n is taken as either 3 or 5. All the curves in Figs. 8.8–
8.10 display the familiar phenomena of increasing strain rate with time.
In Fig. 8.11, we indicate comparative results for the predictions of the
time until the creep deflection reaches a specified value for the creep ex-
ponents n = 3 and n = 5; the chosen value of the fixed creep deflection
in Fig. 8.11 is 5.08 × 10−3 m (which is approximately equal to the plate
thickness). The range of applicability of the results in Figs. 8.8–8.11 is
subject to question, as they correspond to material properties that were
adopted for a constant mean temperature of the plate; in reality, there
is ample experimental evidence to suggest that these properties may be
strongly dependent on temperature.
In [242], Jones, Mazumdar, and Cheung have considered the small
amplitude response of a thermo-rheologically simple viscoelastic plate;
in this case, the model is such that (8.27) is replaced by
NT 2 MT
K(p) 4 w + 2 w = t − p2 (ρhw) − (8.121)
1 − ν(p) 1 − ν(p)
∂
with p ≡ and K(p), ν(p) the viscoelastic time operators correspond-
∂t
ing to the flexual rigidity K and Poisson’s ratio ν for the elastic case.
Equation (8.121) is identical to the governing equation for the analysis
of a viscoelastic plate with in-plane loads
Nx = Ny = NT /(1 − ν(p))
4 W − C 2 W = 0 (8.123)
and where T (x), which represents the magnitude of the deflection, sat-
isfies
NT
K(ρ)T (t) − T (t) =0 (8.124)
C(1 − ν(p))
with c representing a separation constant. Thus, the mode shapes are
time independent and only the amplitudes vary with time. If NT =
© 2001 by Chapm
const. then it may be shown that there exist upper and lower critical
temperatures where the lower critical temperature corresponds to a zero
deflection rate, while the upper critical temperature corresponds to an
infinite deflection. The lower critical temperature is determined by set-
ting p = 0 in (8.124), which yields
NT
C= (8.125)
(1 − ν(0))K(0)
with the notation employed being the same as that in [271]. When the
boundary conditions for the plate contain the Poisson’s ratio ν (as in
the case for a plate with either free or simply supported edges) then
in solving (8.126) or (8.127) one simply replaces ν by ν(0) or ν(∞),
respectively.
For both Kelvin-and Maxwell-type viscoelastic materials, the upper
and lower critical buckling loads are expressible in terms of the elastic
critical loads, with the correlations being indicated in Table 8.1. It may
be shown that the physical interpretation of the critical temperatures is
the same as for the critical buckling loads discussed in [271]. If
∂2w
where w,ij = and we sum, as usual, on repeated indices. Thus,
∂xi ∂xj
with the understanding that ν is time-dependent,
−Ex3
σij = {(1 − ν)w,ij + [νw,kk + (1 + ν)αt τ ]δij } (8.128)
1 − ν2
© 2001 by Chapm
and that the situation depicted in Fig. 8.12 applies, i.e.,
T1 (x1 , x2 ) = T (x1 , x2 , h2 )
(8.134)
T (x , x ) = T (x , x , − h )
2 1 2 1 2 2
12 12
2 τ − 2
(1 + $)τ = − 3 (θ1 − θ2 ) (8.136)
h h
where $ = hλ1 /2λ, or
2 τ − k 2 τ = −β (8.137)
with
12 12
k2 = (1 + $) and β = 3 (θ1 − θ2 ) = const. (8.138)
h2 h
We note that, following the analysis in [262], the fourth-order equation
(8.131) may be replaced by two equations of the second order: using the
fact that
−K(1 + ν) 2 w = M11 + M22
or
M M11 + M22
2 w = − ; M= (8.139)
K (1 + ν)
From (8.131) we then obtain
2 M = m(t)K 2 τ (8.140)
m̃(s)
2 W̃ + τ =0 (8.143)
s
with ∞
W̃ = we−st dt
0 (8.144)
m̃(s) = (1 + ν̃)αt
s being the transform parameter. In equations (8.137) and (8.143) we
introduce the complex coordinate system given by
∂2τ
4 − k 2 τ = −β (8.146a)
∂x3 ∂ x̄3
∂ 2 W̃ m̃(s)
4 + τ =0 (8.146b)
∂x3 ∂ x̄3 s
We now let x3 = f (ξ) be the function which maps the domain of the
¯
plate onto the unit circle so that, in the system of coordinates (ξ, ξ),
(8.146a,b) reduce to
© 2001 by Chapm
with ξ = reiθ , ξ¯ = re−iθ . Assuming that
dx3 dx̄3
= ¯ = a1 = const. (8.148)
dξ dξ
we obtain from (8.147a)
© 2001 by Chapm
a slight modification of the energy potential, as suggested by Neale in
[264], is introduced by Williams [263]. The buckling and postbuckling
equations in [263] are derived following the approach in Danielson [275].
Following the Lagrangian formulation proposed by Neale [264], the
displacement components and relevant Kirchhoff stress components in
[263] assume the form (see Fig. 8.13)
u1 = u(x, y) − zw,x (x, y)
u2 = v(x, y) − zw,y (x, y) (8.161)
u3 = w(x, y)
and
12z
Nij (x, y) + Mij (x, y)
Pij = h2 (8.162)
h
The equations of equilibrium and the constitutive relations governing
u, v, w, Nij , etc., are obtained from the requirement that δI (0) = 0 where
the Lagrangian form of the functional I (0) is
1
I (0) = Ṗij Ėij + Pij vk,i vk,j − W (τ̇ ) dV0
V0 2 (8.163)
∗
− Ḟi vi dS (0) − Ḟi (vi − vi∗ )dS (0)
(0) (0)
SF SV
(0) (0)
In (8.163) SF , SV are the undeformed areas over which the traction
rates Fi∗ and velocities vi∗ are prescribed. The potential W (τ̇ ) is defined
such that the time rate of change of Green’s strain tensor Eij is
∂ ∂W
Eij ≡ Ėij = (8.164)
∂t ∂Pij
The nominal surface tractions per unit undeformed area Fi are given by
(0)
Fj = (Pij + Pik uj,k )ni (8.165)
with n(0) the unit normal vector to the undeformed area. Using the
simplifications indicated in [264], Williams takes
with
1
Eij = (ui,j + uj,i + u3,i u3,j ) (8.167)
2
The effect of a temperature variation is accounted for in Williams [263]
by adding the term αṪ (Ṗ11 + Ṗ22 ) to Neale’s [264] form for W (τ̇ ), i.e.,
1 2 2
W (τ̇ ) = (c11 Ṗ11 + c22 Ṗ22 ) + c12 Ṗ11 Ṗ22 + c13 Ṗ11 Ṗ12
2 (8.168)
2
+c23 Ṗ22 Ṗ12 + c33 P12 + αṪ (Ṗ11 + Ṗ22 )
Following the analysis in Neale [264], the material coefficients cij in [263]
have the form
3k(3J2 )(k−1)/2
G= (8.172)
4J2 E0k
27 3
so that the choice k = 3 leads to G = /E0 ; this restriction to k = 3
4
is then followed in the remainder of the development in [263]. Assum-
ing that the effects of bending and stretching are equally important in
inducing stress in an elastic-plastic plate, that spatial rates of change
are significant over distances of order a, and introducing the associated
© 2001 by Chapm
order-of-magnitude estimates
Mij = hNij O(1)
h h (8.173)
(u, v, w) = h , , 1 O(1)
a a
it may be shown that
h/2
'
'
Ḟj vj dz ' = ± Ṅ11 v̇ + Ṅ12 v̇ − Ṁ11 ẇ1 − Ṁ12 ẇ2
−h/2 x=±a
(8.174a)
∂
+ẇ (Q1 + N11 w1 + N12 w2 )
∂t
h/2 '
'
Ḟj vj dz ' = ± Ṅ12 u̇ + N22 v̇ − Ṁ12 w1 − Ṁ22 ẇ2
−h/2 y=±b
(8.174b)
∂
+ẇ (Q2 + N12 w2 + N22 w2 )
∂t
where
h/2
(Q1 , Q2 ) = (P13 , P23 )dz (8.174c)
−h/2
In writing down (8.174a) and (8.174b), terms of order (h/a)2 have been
neglected. Integrating the bending moment M12 by parts we obtain
'a 'b
' '
Ḟi vi dS = −2[Ṁ12 ẇ]' '
S −a −b
b 'a
'
+ [Ṅ11 u̇ + Ṅ12 v̇ + V̇1 ẇ − Ṁ11 ẇ1 ]' dy (8.175)
−b −a
a 'b
'
+ [Ṅ12 u̇ + Ṅ22 v̇ + V̇2 ẇ − Ṁ22 ẇ2 ]' dx
−a −b
(0) ṄT + Ṅ11 − ν Ṅ22
$̇
11 −
Eh
Ṅ11
(0) ṄT + Ṅ22 − ν Ṅ11
$̇22 −
Eh
Ṅ
22
(0) 2(1 + ν)Ṅ12 Ṅ
2$̇12 −
Eh G 12
= 3 · G
(8.179)
9h
−ẇ − 12(ṀT + Ṁ11 − ν Ṁ22 ) Ṁ11
11
Eh3
Ṁ22
−ẇ22 − 12(ṀT + Ṁ22 − ν Ṁ11 )
Eh 3
Ṁ12
24(1 + ν)Ṁ12
−2ẇ12 −
Eh3
where NT , MT are given by the usual relations, i.e.,
h/2
(NT , MT ) = αE δT (1, z)dz
−h/2
and where the hardening matrix G is defined at the end of this section. If
we let a superposed asterisk again denote quantities which are prescribed
along the boundary of the plate, then the natural boundary conditions
assume the following form:
∗
Ṅ11 = Ṅ11 or u̇ = u̇∗
∗
Ṅ12 = Ṅ12 or v̇ = v̇ ∗
∂
Ṁ11,1 + 2Ṁ12,2 + (N11 w1 + N12 w2 ) = V̇1∗ or ẇ = ẇ∗
∂t
∗
Ṁ11 = Ṁ11 or ẇ1 = ẇ1∗
on x = ±a and (8.180)
∗ ∗
Ṅ12 = Ṅ12 or u̇ = u̇
∗
Ṅ22 = Ṅ22 or v̇ = v̇ ∗
∂
Ṁ22,2 + 2Ṁ12,1 + (N12 w1 + N22 w2 ) = V̇ ∗ or ẇ = ẇ∗
∂t
∗
Ṁ22 = Ṁ22 or ẇ2 = ẇ2∗
© 2001 by Chapm
at the corners of the plate. Taking
V1 = M11,1 + 2M12,2 + N11 w1 + N12 w2
(8.181)
V2 = M22,2 + 2M12,1 + N12 w1 + N22 w2
we find that
Q̇1 = M11,1 + Ṁ12,2
(8.182)
Q̇2 = Ṁ12,1 + Ṁ22,2
and, thus obtain the form which is familiar from classical plate theory,
i.e.,
Ṁαβ,αβ = Q̇α,α (8.183)
Therefore, although constitutive formulation in [264] differs from earlier
formulations of the elastic-plastic deformation problem for plates, one
does recover (in rate form) the equilibrium equations which are familiar
from classical von Karman plate theory.
We now proceed with the derivation of the equations governing the
buckling and postbuckling behavior of heated rectangular plates, assum-
ing that the plate is constrained against displacement along its edges;
such a plate, if elastic, will in general buckle if its edges are clamped
or, if MT = 0, if its edges are simply supported. In [263], attention is
restricted to the case in which MT = 0 and NT = const. over the surface
of the plate (although, in principle, any spatial variation in NT could
be studied); under these restrictions it may be shown that there exists
a prebuckled membrane state given by
u=v=w=0
N11 = N22 = N (NT )
(8.184)
N12 = 0
M11 = M22 = M12 = 0
where
2EGN 2
ṄT + Ṅ 1−ν+ =0 (8.185)
9h2
is a consequence of the constitutive relations. It may be expected that
this prebuckled membrane state persists for NT < (NT )cr with (NT )cr
corresponding to the critical temperature parameter.
To determine (NT )cr , and the behavior of the plate in a neighborhood
of the critical temperature, Williams [263] adopts the procedure given
by Danielson [275], i.e., the time variable is replaced by the distance
parameter s along the load path measured from the critical point. With
∂ ∂ ds
= , we then assume the expansions
∂t ∂s dt
u = su(1) + s2 u(2) + ...
v = sv (1) + s2 v (2) + ... (8.186)
(1) 2 (2)
w = sw +s w + ...
and
(1) (2)
Nαβ = N δαβ + sNαβ + s2 Nαβ + ...
(1) (2)
Mαβ = sMαβ + s2 Mαβ + ...
(8.187)
N = Ncr + N (1) + s2 N (2) + ...
(1) (2)
NT = (NT )cr + sNT + s2 NT + ...
and obtain from the Euler (equilibrium) equations (8.177)-(8.178)
(1) (2)
Nαβ,β = 0 Nαβ,β = 0
(1) (1)
Mαβ,αβ + Ncr δαβ wαβ = 0 (8.188)
M (2) + N δ w(2) + (N (1) + N (1) δ )w(1) = 0
αβ,αβ cr αβ αβ αβ αβ αβ
© 2001 by Chapm
and the components of the symmetric A, B, and D matrices are defined
at the end of this section.
(0)
The strain tensor $αβ is decomposed according to
(0) w,α w β
$αβ = eαβ + (8.192)
2
and, furthermore, we write
(1) (2)
eαβ = seαβ + s2 eαβ + ....
(1) (2)
(8.193)
καβ ≡ w,αβ = sκαβ + s2 καβ + ...
where (k) (k) (k) (k)
e11 = u,1 , e22 = v,2
(8.194)
e(k) = e(k) = 1 u(k) + v (k)
12 21 ,2 ,1
2
The tensor components Aαβγδ , Bαβγδ , Dαβγδ are also expanded as series
in the parameter s, i.e.,
(0) (1) (2)
Aαβγδ = Aαβγδ + sAαβγδ + s2 Aαβγδ + ...
(1) (2)
Bαβγδ = sBαβγδ + s2 Bαβγδ + ... (8.195)
(0) (1) (2)
Dαβγδ = Dαβγδ + sDαβγδ + s2 Dαβγδ + ...
Using the expansions (8.193), (8.195) in the constitutive relations
(8.190), (8.191) it follows that
We now substitute the expansions (8.186), (8.187) into both (8.198) and
(8.200) and subtract the resulting equations so as to obtain
a b
(Nαβ δ ėαβ + N ∗ wα(1) δαβ δ ẇβ − Mαβ δ κ̇αβ
(1) (1)
−a −b
(8.201)
Restricting the strain-rate variations to have the form
(k) (k) (k)
δ $̇αβ = seαβ , δ ẇ,β = ṡw,β , δ ṙαβ = ṡκαβ (8.202)
© 2001 by Chapm
To avoid the possibility that w(k) contains an arbitrary multiple of w(k) ,
Williams [263] imposes the restriction
a b
(k) (k)
δαβ w,α w,β dxdy = 0 (8.204)
−a −b
© 2001 by Chapm
and we have defined
Eh3 /a3
(Ncr , (NT )cr ) = (N̄cr , (N̄T )cr )
1+ν
2
N̄cr
∗ (8.213)
g = Ḡ
1+ν
2 2
1
Ḡ = EG Eh
9 a2
We note that there exists an infinite number of eigenvalues λn of
(8.211) subject to appropriate boundary conditions, such that
1 − ν + g∗
λn = −12N̄cr
2
(8.214)
1 + g∗
We also note that as g ∗ depends on Ncr though (8.213), Ncr is a solution
of the cubic equation
λ2 2 (1 + ν)2 λ2
3
N̄cr + n N̄cr + (1 − ν)N̄cr + n = 0 (8.215)
24 2Ḡ 12
Thus, the minimum buckling stress N̄cr can be determined only after
one examines the roots of (8.215), which correspond to the previously
determined eigenvalues λn . As Williams [263] notes, the limiting case
Ḡ = 0 yields some insight into the dependence (N̄T )cr (G); if we consider
λn as being known, then (8.215) determines the functional relationship
N̄cr (G). Also, as
' 2
' λ2n dg ∗ '' N̄cr (0)
(N̄cr )' =− and ' = (8.216)
Ḡ=0 12(1 − ν) dḠ Ḡ=0 1+ν
it follows, as a consequence of (8.215), that
As (N̄cr )(0) < 0 the effect of the hardening parameter is always to reduce
the buckling temperature and the absolute value of the stress from those
values which apply for a linearly elastic material in a small neighborhood
of G = 0.
The solution of (8.211) which is the easiest to secure is the one which
corresponds to the case in which the plate is simply supported along all
its edges.
Taking
nπ
mπ
w(1) = Amn sin (x + a) sin (y + b) (8.218)
2a 2b
it follows that the boundary conditions along the edges x = ±a, y = ±b
are identically satisfied and that the eigenvalues are
π2 2
λ2mn = (n + m2 φ2 ); φ = a/b (8.219)
4
For ν = 0.3, numerical results for the buckling stress and temperatures
are reported by Williams in [263] and are summarized in tables 8.2
and 8.3. For all the cases shown, the minimum values of stress and
temperature correspond to m = n = 1, i.e., a mode shape with one wave
in each direction. The effect of the hardening parameter is to reduce the
stress and temperature from values which would correspond to G = 0;
this effect is more pronounced, as noted in [263], for long slender plates.
Also, the effect of G is not monotonic: the buckling temperature first
decreases with increasing G and then increases.
For the case in which the edges at x = ±a are simply-supported, while
those at y = ±b are clamped, we may take
nπ y
w(1) = W (ȳ) sin (x + a) ; ȳ = (8.220)
2a h
so that the boundary conditions along x = ±a are satisfied identically,
while w(ȳ) must satisfy
W + (β 2 − n̄2 )W − n̄2 β 2 W = 0 (8.221)
with 2
nπb 2 λn
n̄ = ; n̄ + β 2 = (8.222)
2a φ
The solution of (8.221) has the form
W = A1 sin hn̄ȳ + A2 cos hn̄ȳ + A3 sin β ȳ + Ay cos β ȳ (8.223)
From the clamped conditions along ȳ = ±1, we infer that either the
buckling mode is even (A1 = A3 = 0) with β satisfying
β tan β = −n̄ tan hn̄ (8.224)
or the buckling mode is odd (A2 = A4 = 0) with
tan β tan hn̄
= (8.225)
β n̄
© 2001 by Chapm
Therefore to determine the eigenvalues λn the buckling stress and tem-
perature, (8.224) or (8.225) must be solved for a fixed n̄, for β(n̄). Nu-
merical results for this clamped simply supported case are delineated in
tables 8.4 and 8.5, for ν = 0.3; in all the cases exhibited, the minimum
values of the buckling temperature and stress correspond to the even
mode case (A1 = A3 = 0 in (8.223)) and thus to the smallest root β of
(8.224). Once again, the general effect of G is to reduce the buckling
stress from those values associated with G = 0, but the reduction in the
buckling temperature due to G appears to occur in higher-aspect ratio
plates only and only in a small neighborhood of G = 0.
Williams [263] also treats the initial buckling of a long strip undergoing
both cylindrical and non-cylindrical deformations; for the case of a strip
with simply supported edges the deflection has the form
λx λx
w(1) = As sin + Bs cos (8.226)
a a
where 2a is the width of the strip, while for a strip with clamped edges
at x = ±a
λx λx a2
w(1) = Ac sin + Bc cos + 2 (c1 x + c2 ) (8.227)
a a λ
In the simply supported case we have
π 3π
Bs = 0, As = 0; λ = , , ...
or 2 2 (8.228)
0, Bs = 0, λ = π, 2π, ...
As =
Bc = 0, Ac = 0; λ = π, 2π, ...
or (8.229)
Ac = 0, Bc = 0; tan λ = λ(λ = 4.4934, 7.7252, ...)
Numerical results for these two situations are delineated in tables 8.6 and
8.7 with ν = 0.3; for all the cases depicted, the minimum buckling stress
and temperature are those which are associated with an even mode. The
effect of G is to reduce both the buckling temperature and stress with
the effect now monotonic over the range of G considered.
We now want to briefly summarize some of the results obtained in [263]
relative to the postbuckling behavior of the plate and the nature of the
dependence of that behavior on the hardening parameter G. The first
parameter which arises in a postbuckling calculation is w(2) (s), which is
the initial curvature of the transverse deflection as measured along the
loading curve that initiates at the critical point. As N (1) = 0, it follows
from the equilibrium and constitutive equations that w(2) satisfies the
same equation as w(1) and thus, by virtue of the orthogonality relation
(8.204), w(1) = 0. The next parameter which must be computed is
N (2) . For the specific case of cylindrical deformation of a long strip, the
second-order constitutive equations assume the form
1 (1)2
Eh(u1 + w1 ) = (1 + g ∗ )N11 + (−ν + g ∗ )N22
(2) (2) (2)
2
6g ∗ (4M11 − M11 M22 + M22 )
(1)2 (1) (1) (1)2
+
Ncr h2
6g ∗ (4M22 − M11 M22 + M11 )
(1)2 (1) (1) (1)2
0 = (1 + g ∗ )N22 + (−ν + g ∗ )N11 +
(2) (2)
Ncr h2
(8.230)
However,
so that
1 (1)2
Eh(1 + g ∗ )(u1 + w1 ) = (1 + ν)(1 − ν + 2g ∗ )N11
(2) (2)
2 (8.232)
24(1 + ν)g ∗
[(1 − ν)(1 + 3g ∗ ) + ν 2 + 3g ∗ ]M11
(1)2
+ 2 ∗ 2
Ncr h (1 + g )
(2)
while the second-order equilibrium equation (8.188) requires that N11 =
(2)
const. The form of N11 can be obtained by integrating (8.232) over
−a ≤ x ≤ a and using the fact that u(2) (±a) = 0. In both the clamped
and simply supported cases one obtains (with Bi = Bs or Bc )
(2) 3Ncr Bi2 4g ∗ (1 − ν)(1 + 3g ∗ ) + ν 2 + 3g ∗2
N11 = − 1+
h2 (1 + g ∗ )2 1 − ν + 2g ∗
(8.233)
The form of N (2) may now be computed by employing (8.209); numerical
(2)
results for both N (2) and N11 (with ν = 0.3) are presented in Tables
8.8 and 8.9; Table 8.8 covers the case of a clamped strip , while Table
8.9 addresses the same situation for a simply supported strip. As
(2)
N11 s2 (N11 + N (2) )
=1+ + O(s3 ) (8.234)
Ncr Ncr
© 2001 by Chapm
and N11 + N (2) = O(g ∗ ) > 0, the stress resultant N11 decreases in
(2)
p
σcr 3Et
η = e = f (0.352, 0.500) + (0.324, 0.250) 1 + (8.235)
σcr Es
In (8.236),
Es (1 − νe2 ) 1 1 Es
f= , ν= − − νe (8.236)
E(1 − ν 2 ) 2 2 E
where νe , Es , and Et are the elastic Poisson’s ratio, the secant mod-
ulus, and the tangent modulus of the maternal, respectively. For the
Ramberg-Osgood form (8.170)
2
E 3E P11 9EJ2
=1+ ≡1+
Et E0 E0 E03
2 (8.237)
E E P11 3EJ2
=1+ ≡1+
Es E0 E0 E03
1 2
where Et = Et (J2 ) and J2 = P . However, as P11 = P22 = Ncr /h
3 11
and P12 = 0, at the critical point, one obtains from (8.237)
E E 4g ∗
= 1 + 4g ∗ , =1+ (8.238)
Et Es 3
A comparison of the results in [276] and [263] is presented in Table
a
8.10; in this table, the predictions associated with a long φ = = 3.5 ,
b
(completely) simply supported plate appear in the first column, while
those in the second column correspond to the case of a clamped strip.
As noted by Williams [263], the general behavior indicated in Table
8.10 is the same for both sets of boundary conditions. Concerning the
postbuckling analysis, it should be noted that the condition of loading
(J˙2 > 0) is generally violated once buckling has occurred. In fact, to
first order, the plane stress form for J2 is given by
1 2
J2 = 2
(P + P22 − P11 P22 ) + P12
2
(8.239)
3 11
Thus, as
˙ 1 ∂P11 ∂P22 ∂P12 ds
J2 = (2P11 − P22 ) + (2P22 − P11 ) + 2P12 ,
3 ∂s ∂s ∂s dt
(8.240)
∂P ij 12y (1)
= 3 Mij + O(s)
∂s h (8.241)
N
(2P11 − P22 )(2P22 − P11 ) = cr + O(s),
h
and
(1) (1)
(1) (1) Eh3 w11 + w22
M11 + M22 = − (8.242)
12 1 − ν + 2g ∗
it follows that
EyNcr /3h (1) (1)
J˙2 = − (w + w22 ) + O(s) ṡ (8.243)
1 − ν + 2g ∗ 11
© 2001 by Chapm
(1) (1)
If the first curvature is positive, i.e., w11 + w22 > 0 then, because
Ncr < 0, there will be unloading on the y < 0 (negative) side of the
plate; as indicated in [263], this result is consistent with the expectation
that the bending moments due to buckling will induce tension in the
domain y < 0 and thus produce unloading.
12(2M11 − M22 )2
G11 = (2N11 − N22 )2 +
h2
G12 = (2N11 − N22 )(2N22 − N11 )
12(2M11 − M22 )(2M22 − M11 )
+
h2
12M12 (2M11 − M22 )
G13 = 6 N12 (2N11 − N22 ) +
h2
24(2N11 − N22 )(2M11 − M22 )
G14 =
h2
12[5(N11 M22 + N22 M11 ) − 4(N11 M11 + N22 M22 )]
G15 = = G24
h2
72[N12 (2M11 − M22 ) + M12 (2N11 − N22 )]
G16 = = G34
h2
12(2M22 − M11 )2
G22 = (2N22 − N11 )2 +
h2
12M12 (2M22 − M11 )
G23 = 6 N12 (2N22 − N11 ) +
h2
24(2N22 − N11 )(2M22 − M11 )
G25 =
h2
72[N12 (2M22 − M11 ) + M12 (2N22 − N11 )]
G26 = = G35
h2
2
2 12M12 864N12 M12
G33 = 36 N12 + 2
, G36 =
h h2
12[(2N11 − N22 )2 + 108(2M11 − M22 )2 /5h2
G44 =
h2
(8.244)
12
G45 = bigg[(2N11 − N22 )(2N22 − N11 )
h2
108
+ 2 (2M11 − M22 )(2M22 − M11 )
5h
72[N12 (2N11 − N22 ) + 108M12 (2M11 − M22 )/5h2 ]
G46 =
h2
12[(2N22 − N11 ) + 108(2M22 − M11 )2 /5h2
2
G55 =
h2
72[N12 (2N22 − N11 ) + 108M12 (2M22 − M11 )/5h2 ]
G56 =
h2
2 2
432(N12 + 108M12 /5h2 )
G66 = 2
h
HG13 HG13
1 + HG11 −ν + HG12
2 2
HG23 HG23
1 + HG22
2 2
A=
1 + ν + (HG33 )/2 1 + ν + (HG33 )/2
2 2
1 + ν + (HG33 )/2
(sym)
2
(8.245)
© 2001 by Chapm
G16 G16
G14 G15
2 2
G26 G26
G25
2 2
B=
(8.246)
G36 G36
4 4
G36
(sym)
4
12 12ν HG46 HG46
+ HG44 − 2 + HG45
h2 h 2 2
12 HG56 HG56
+ HG55
h2 2 2
D=
1+ν HG46 1 + ν HG66
6 2 + 6 2 +
h 4 h 4
1+ν HG66
(sym) 6 2 +
h 4
(8.247)
where
GE
H= (8.248)
9h2
where
1 + g ∗ −ν + g ∗ 0 0
1 + g∗ 0 0
1+ν 1+ν
A(0) = (8.250)
2
2
1+ν
(sym)
2
(1) (1)
2(2N11 − N22 + N (1) )
A(1) = HNcr
(sym)
(1) (1) (1) (1)
(N11 + N22 + 2N (1) ) 3N12 3N12
2(2N22 − N11 + N (1) ) 3N12 3N12
(1) (1) (1) (1)
0 0
0
(8.251)
(1) (1) (1) (1) (1) (1)
2(2M11 − M22 ) (M11 + M22 ) 3M12 3M12
12HNcr 2(2M22 − M11 3M12 3M12
(1) (1) (1) (1)
B (1) =
h2 0 0
(sym) 0
(8.252)
and
12
(D(0) , D(1) ) = · (A(0) , A(1) ) (8.253)
h2
where
(1) (1) (1) (1) (1) (1)
(2M11 − M22 )2 (2M11 − M22 )(2M22 − M11
108 (1) (1)
(2M22 − M11 )2
= 2
(2)
D1
5h
(sym)
(1) (1) (1) (1) (1) (1)
3M12 (2M11 − M22 3M12 (2M11 − M22
(1)
M11
(1) (1) (1) (1) (1)
3M12 (2M22 − M11 ) 3M12 (2M22 −
(1)2 (1)2
9M12 9M12
(1)2
9M12
(8.255)
© 2001 by Chapm
(1) (2) (2) (2) (2) (2)
2(2N4 − N22 ) (N11 + N22 ) 3N12 3N12
(2) (2)
2(2N22 −
(2)
N11 )
(2)
3N12 3N12
(2)
D2 = (N )cr
0 0
(sym) 0
(8.256)
and
1 100
1 0 0
= 2(N )cr N (2)
(2)
D3 (8.257)
0 0
(sym) 0
8.4 Thermal Bending, Buckling, and Postbuckling
Figures, Graphs, and Tables II
© 2001 by Chapm
Table 8.2 Buckling Temperature for Simply Supported
Plates (ν = 0.3) [263] (Copyright 1982. From Neale: Plate
equations and applications to buckling of rectangular plates,
J. of Thermal Stresses, 5, 207-243, by Williams, H.E. Repro-
duced by permission of Taylor & Francis, Inc./Routledge, Inc.,
http://www.routledge-ny.com)
(N̄T )cr
Ḡ
a/b
0 0.1 0.2 0.3
−N̄cr
Ḡ
a/b
0 0.1 0.2 0.3
© 2001 by Chapm
Table 8.4 Buckling Temperature for Mixed Boundary Condi-
tion Plates (ν = 0.3) [263] (Copyright 1982. From Neale: Plate
equations and applications to buckling of rectangular plates,
J. of Thermal Stresses, 5, 207-243, by Williams, H.E. Repro-
duced by permission of Taylor & Francis, Inc./Routledge, Inc.,
http://www.routledge-ny.com)
(N̄T )cr
Ḡ
a/b
0 0.1 0.2 0.3
−N̄cr
Ḡ
a/b
0 0.1 0.2 0.3
© 2001 by Chapm
Table 8.6 Buckling Stress and Temperature for the Sim-
ply Supported Strip (λ = π/2) [263] (Copyright 1982.
From Neale: Plate equations and applications to buckling
of rectangular plates, J. of Thermal Stresses, 5, 207-243, by
Williams, H.E. Reproduced by permission of Taylor & Fran-
cis, Inc./Routledge, Inc., http://www.routledge-ny.com)
0 0.2937 0.2056
0.1 0.2910 0.2047
0.2 0.2885 0.2039
0.3 0.2861 0.2031
0 1.175 0.8225
0.1 1.054 0.7841
0.2 0.9860 0.7658
0.3 0.9390 0.7553
Table 8.8 Curvature Parameters for the Clamped Strip
(λ = π, ν = 0.3) [263] (Copyright 1982. From Neale: Plate
equations and applications to buckling of rectangular plates,
J. of Thermal Stresses, 5, 207-243, by Williams, H.E. Repro-
duced by permission of Taylor & Francis, Inc./Routledge, Inc.,
http://www.routledge-ny.com)
(2)
(2) N11 + N (2)
Ḡ −N̄cr N (2) /Ncr −N11 /Ncr −
Ncr
(2)
(2) N11 + N (2)
Ḡ −N̄cr N (2) /Ncr −N11 /Ncr −
Ncr
© 2001 by Chapm
Table 8.10 Comparison with Mechanically Loaded Plates
[263] (Copyright 1982. From Neale: Plate equations and
applications to buckling of rectangular plates, J. of Ther-
mal Stresses, 5, 207-243, by Williams, H.E. Reproduced
by permission of Taylor & Francis, Inc./Routledge, Inc.,
http://www.routledge-ny.com)
(a) (b)
Ncr (Ḡ) Ncr (Ḡ)
Ḡ η η(c)
Ncr (0) Ncr (0)
© 2001 by Chapm
FIGURE 8.2
Variation of middle-surface and bending stresses with the tem-
perature gradient through the thickness (k̂0 = 0) [254]. (From
Mansfield, E. H., “Bending, Buckling, and Curling of a Heated
Thin Plate”, Proc. Roy. Soc., Ser. A, 1334, 316, 1962.
Reprinted by permission of the Royal Society).
FIGURE 8.3
Variation of principal curvatures with the temperature gra-
dient through the thickness (k̂0 = λ1/2 ) [254]. (From Mans-
field, E. H., “Bending, Buckling, and Curling of a Heated Thin
Plate”, Proc. Roy. Soc., Ser. A, 1334, 316, 1962. Reprinted
by permission of the Royal Society).
© 2001 by Chapm
FIGURE 8.4
Variation of principal curvatures with the temperature gradi-
ent through the thickness (k̂0 = 6) [254]. (From Mansfield, E.
H., “Bending, Buckling, and Curling of a Heated Thin Plate”,
Proc. Roy. Soc., Ser. A, 1334, 316, 1962. Reprinted by per-
mission of the Royal Society).
FIGURE 8.5
Variation of principal curvatures with the temperature gradi-
ent through the thickness (k̂0 = 10) [254]. (From Mansfield, E.
H., “Bending, Buckling, and Curling of a Heated Thin Plate”,
Proc. Roy. Soc., Ser. A, 1334, 316, 1962. Reprinted by per-
mission of the Royal Society).
© 2001 by Chapm
FIGURE 8.6
Variation of principal curvatures with the temperature gradi-
ent through the thickness (k̂0 = 12) [254]. (From Mansfield, E.
H., “Bending, Buckling, and Curling of a Heated Thin Plate”,
Proc. Roy. Soc., Ser. A, 1334, 316, 1962. Reprinted by per-
mission of the Royal Society).
FIGURE 8.7
Plate dimensions, applied stresses, and deflections [261].
(Copyright 1981. From Temperature-Dependent Creep Buck-
ling of Plates, J. of Thermal Stresses, 4, 237-247, by Ross, D.A.
and Berke, L. Reproduced by permission of Taylor & Francis,
Inc./Routledge, Inc., http://www.routledge-ny.com)
© 2001 by Chapm
FIGURE 8.8
Creep deflections as a function of temperature differential and
creep exponent for an applied stress of 6.9 M P a(1ksi) [261].
(Copyright 1981. From Temperature-Dependent Creep Buck-
ling of Plates, J. of Thermal Stresses, 4, 237-247, by Ross, D.A.
and Berke, L. Reproduced by permission of Taylor & Francis,
Inc./Routledge, Inc., http://www.routledge-ny.com)
FIGURE 8.9
Creep deflections as a function of temperature differential and
creep exponent for an applied stress of 13.8 M P a(2ksi) [261].
(Copyright 1981. From Temperature-Dependent Creep Buck-
ling of Plates, J. of Thermal Stresses, 4, 237-247, by Ross, D.A.
and Berke, L. Reproduced by permission of Taylor & Francis,
Inc./Routledge, Inc., http://www.routledge-ny.com)
© 2001 by Chapm
FIGURE 8.10
Creep deflections as a function of temperature differential and
creep exponent for an applied stress of 34.5 M P a(5ksi) [261].
(Copyright 1981. From Temperature-Dependent Creep Buck-
ling of Plates, J. of Thermal Stresses, 4, 237-247, by Ross, D.A.
and Berke, L. Reproduced by permission of Taylor & Francis,
Inc./Routledge, Inc., http://www.routledge-ny.com)
FIGURE 8.11
Comparison of predictions for the time until the creep deflec-
tion reaches a specified value for creep exponents n = 3 and
n = 5 [261]. (Copyright 1981. From Temperature-Dependent
Creep Buckling of Plates, J. of Thermal Stresses, 4, 237-247,
by Ross, D.A. and Berke, L. Reproduced by permission of
Taylor & Francis, Inc./Routledge, Inc., http://www.routledge-
ny.com)
© 2001 by Chapm
FIGURE 8.12
The viscoelastic plate [262]. (Reprinted from Int. J. Mech.
Sci., 23, Das, S., Note on Thermal Deflection of Regular Polyg-
onal Viscoelastic Plates, 323-329, 1981, with permission from
Elsevier Science)
FIGURE 8.13
Coordinate system for the elastic-plastic plate [263]. (Copy-
right 1982. From Neale Plate Equations and Applications to
Buckling of Rectangular Plates, J. of Thermal Stresses, 5, 207-
243, by Williams, H.E. Reproduced by permission of Taylor &
Francis, Inc./Routledge, Inc., http://www.routledge-ny.com)
© 2001 by Chapm
FIGURE 8.14
Stress factor for simply supported conditions [263]. (Copy-
right 1982. From Neale Plate Equations and Applications to
Buckling of Rectangular Plates, J. of Thermal Stresses, 5, 207-
243, by Williams, H.E. Reproduced by permission of Taylor &
Francis, Inc./Routledge, Inc., http://www.routledge-ny.com)
FIGURE 8.15
Stress factor for clamped conditions [263]. (Copyright 1982.
From Neale Plate Equations and Applications to Buckling of
Rectangular Plates, J. of Thermal Stresses, 5, 207-243, by
Williams, H.E. Reproduced by permission of Taylor & Francis,
Inc./Routledge, Inc., http://www.routledge-ny.com)
© 2001 by Chapm
References
© 2001 by Chapm
[24] Batdorf, S.B., “Theories of Plastic Buckling,” J. Aero-
naut. Sci., 16, 1949, 405-408.
[25] Bauer, L., Keller, H.B., and Reiss, E., “Multiple Eigenval-
ues Lend to Secondary Buckling,” SIAM Rev., 17, 1975,
101-122.
[26] Berger, M.S., Nonlinearity and Functional Analysis, Aca-
demic Press, N.Y., 1977.
[27] Bijlaard, P.P., “On the Plastic Buckling of Plates,” J.
Aeronaut. Sci., 17, 1950, 742-743.
[28] Bolotin, V., Dynamic Stability of Elastic Systems,
Holden-Day Pub., N.Y., 1964.
[29] Brush, D.O. and Almoth, B.O., Buckling of Bars, Plates,
and Shells, McGraw-Hill, N.Y., 1975.
[30] Budiansky, B., “Dynamic Buckling of Elastic Structures:
Criteria and Estimates,” Proc. Int. Conf. Dynamic Sta-
bility of Structures, Northwestern Univ., Evanston, IL,
1965, 83-106.
[31] Budiansky, B. and Hutchinson, J.W., “Dynamic Buckling
of Imperfection-Sensitive Structures,” in Proc. Int. Cong.
Appl. Mech., Munich, XI, 1964, 636-651.
[32] Bulson, P.S., The Stability of Flat Plates, American Else-
vier, N.Y., 1969.
[33] Chia, C.Y., Nonlinear Analysis of Plates, McGraw-Hill,
N.Y., 1980.
[34] Chow, S-N., Hale, J.K., and Mallet-Paret, J., “Applica-
tions of Generic Bifurcations, I,” Arch. Rat. Mech. Anal.,
59, 1975, 159-188.
[35] Chow, S-N., Hale, J.K., and Mallet-Paret, J., “Applica-
tions of Generic Bifurcations, II,” Arch. Rat. Mech. Anal.,
62, 1976, 209-236.
[36] Ciarlet, P.G., “A Justification of the von Karman Equa-
tions,” Arch. Rat. Mech. Anal., 73, 1980, 349-389.
[37] Cohen, G.A., “Effect of a Nonlinear Prebuckling State on
the Postbuckling Behavior and Imperfection Sensitivity
of Elastic Structures,” AIAA J., 6, 1968, 1616-1620.
[38] Dickey, R.W., Bifurcation Problems in Nonlinear Elastic-
ity, Research Notes in Math., 3, Pitman, N.Y., 1976.
[39] Dym, C.L., Stability Theory and its Applications to Struc-
tural Mechanics, Noordhoff, Amsterdam, 1974.
[40] Fox, D.D., Raoult, A., and Simo, J.C., “A Justification of
Nonlinear Properly Invariant Plate Theories,” Arch. Rat.
Mech. Anal., 124, 1993, 157-199.
[41] Gauss, R.C. and Antman, S.S., “Large Thermal Buckling
of Nonuniform Beams and Plates,” Int. J. Solids Struc-
tures, 20, 1984, 979-1000.
[42] Hill, R., “A General Theory of Uniqueness and Stability
in Elastic/Plastic Solids,” J. Mech. Phys. Solids, 6, 1958,
236-249.
[43] Stavsky, Y. and Hoff, N.J., “Mechanics of Compos-
ite Structures,” in Composite Engineering Laminates,
A.G.H. Dietz, Ed., MIT Press, Cambridge, MA., 1969,
5-59.
[44] Huseyin, K., Nonlinear Theory of Elastic Stability,
Sijthoff & Noordhoff, Amsterdam, 1975.
[45] Hutchinson, J.W., “Imperfection Sensitivity in the Plastic
Range,” J. Mech. Phys. Solids, 21, 191-204.
[46] Hutchinson, J.W., “Post-bifurcation Behavior in the Plas-
tic Range,” J. Mech. Phys. Solids, 21, 1973, 163-190.
[47] Hutchinson, J.W. and Koiter, W.T., “Postbuckling The-
ory,” Appl. Mech. Rev., 23, 1970, 1353-1366.
[48] Knightly, G.H., “An Existence Theorem for the von
Kármán u Equations,” Arch. Rat. Mech. Anal., 27, 1967,
233-242.
[49] Knightly, G.H., “Some Mathematical Problems from
Plate and Shell Theory,” in Nonlinear Functional Analy-
sis and Differential Equations, Dekker, N.Y., 1976.
[50] Knightly, G.H. and Sather, D., “On Nonuniqueness of So-
lutions of the von Karman Equations,” Arch. Rat. Mech.
Anal., 36, 1970, 65-78.
© 2001 by Chapm
[51] Lekhnitskii, S.G., Anisotropic Plates, Gordon and Breach
Pub. Co., N.Y., 1968.
[52] Lekhnitskii, S.G., Theory of Elasticity of an Anisotropic
Elastic Body, Holden-Day, San Francisco, 1963.
[53] List, S., “Generic Bifurcation with Applications to the
von Kármán Equations,” J. Diff. Eqns. 30, 1978, 89-118.
[54] Potier-Ferry, M., “On the Mathematical Foundations of
Elastic Stability Theory,” Arch. Rat. Mech. Anal., 77,
1982, 55-72.
[55] Potier-Ferry, M., “Perturbed Bifurcation Theory,” J.
Diff. Eqns., 33, 1979, 112-146.
[56] Rabinowitz, P., “Applications of Bifurcation Theory,” in
Proc. Symp. Wisc., Oct. 1976, Academic Press, N.Y.,
1977.
[57] Rabinowitz, P., “Some Global Results for Nonlinear
Eigenvalue Problems,” J. Func. Anal., 7, 1971, 487-513.
[58] Sather, D., “Branching of Solutions of Nonlinear Equa-
tions,” Rocky Mtn. J. Math, 3, 1973, 203-250.
[59] Sewell, M.J., “A Survey of Plastic Buckling,” in Stabil-
ity, (H. Leipholz, Ed.), Univ. of Waterloo Press, Ontario,
1972, 85-197.
[60] Stakgold, I., “Branching of Solutions of Nonlinear Equa-
tions,” SIAM Rev., 13, 1971, 289-332.
[61] Stein, M., “Loads and Deformations of Buckled Rectan-
gular Plates,” Tech. Rep. R-40, NASA, 1959.
[62] Thompson, J.M.T. and Hunt, G.W., A General Theory
of Elastic Stability, Wiley, London, 1973.
[63] Thompson, J.M.T. and Hunt, G.W., Elastic Instability
Phenomena, Wiley, N.Y., 1984.
[64] Timoshenko, S.P. and Gere, J.M., Theory of Elastic Sta-
bility, 2nd ed., McGraw-Hill, N.Y., 1961.
[65] Timoshenko, S. and Woinowsky-Krieger, S., Theory of
Plates and Shells, McGraw-Hill, N.Y., 1959.
[66] Troger, H. and Steindl, A., Nonlinear Stability and Bifur-
cation Theory, An Introduction for Engineers and Applied
Scientists, Springer-Verlag, N.Y., 1991.
[67] Vanderbauwhede, A., “Generic Bifurcation and Symme-
try, with an Application to the von Karman Equations,”
Proc. Roy. Soc. Edinburgh, 81A, 1978, 211-235.
[68] Vanderbauwhede, A., “Generic and Nongeneric Bifurca-
tion for the von Karman Equations,” J. Math Anal. Appl.,
66, 1978, 550-573.
[69] Föppl, A., Vorlesungen über Technische Mechanik, 5,
1907, 132.
[70] Uthgenannt, E. and Brand, R., “Postbuckling of Or-
thotropic Annular Plates,” J. Appl. Mech., 40, 1973, 559-
564.
[71] Coffin, D., “Flange Buckling and the Deep-Drawing of
Thermoplastic Composite Sheets,” Ph.D. thesis, Univer-
sity of Delaware (Mech. Eng.), 1993.
[72] Friedrichs, K.O. and Stoker, J.J., “The Nonlinear Bound-
ary Value Problem of the Buckled Plate,” Amer. J. Math,
63, 1941, 839-888.
[73] Johnson, M. and Urbanik, T.J., “A Nonlinear Theory for
Elastic Plates with Application to Characterizing Paper
Properties,” J. Appl. Mech., 51, 1984, 146-152.
[74] Johnson, M. and Urbanik, T.J., “Buckling of Axially
Loaded, Long Rectangular Paperboard Plates,” Wood
and Fiber Science, 19, 1987, 135-146.
[75] Schleicher, F., “Die Knick Spannungen von eingespan-
nten rechteckigen Platten,” Mitt. Forschungsanstatt.,
Gutchoffnungshütte Konzerno, 1, 1931.
[76] Levy, S., “Buckling of Rectangular Plates with Built-in
Edges,” J. Appl. Mech., 9, 1942, A 71.
[77] Bryan, G.H., “On the Stability of a Plane Plate under
Thrusts in its own Plane with Applications to the Buck-
ling of the Sides of a Ship,” Proc. London Math Soc., 22,
1891, 54-67.
© 2001 by Chapm
[78] Prezemiencki, J.S., “Buckling of Rectangular Plates Un-
der Bi-Axial Compression,” J. Roy. Aero. Soc., 59, 1955,
566-568.
© 2001 by Chapm
[103] Tvergaard, V., “Effect of Plasticity on Post-Buckling Be-
havior,” Lecture Notes in Physics, vol. 288, Springer-
Verlag, N.Y., 1987, 143-183.
© 2001 by Chapm
[124] Berger, M.S., “On von Karman’s Equations and the
Buckling of a Thin Elastic Plate, I: The Clamped Plate,”
Comm. Pure Appl. Math, 20, 1967, 687-719.
[125] Berger, M.S. and Fife, P.C., “On von Karman’s Equations
and the Buckling of a Thin Elastic Plate,” Bull. A.M.S.,
72, 1966, 1006-1011.
© 2001 by Chapm
[147] Sherbourne, A.N., “Elastic Postbuckling Behavior of a
Simply Supported Circular Plate,” J. Mech. Eng. Sci., 3,
1961, 133-141.
[148] Stoker, J.J., Nonlinear Elasticity, Gordon and Breach
pub., N.Y., 1968.
[149] Bodner, S.R., “The Postbuckling Behavior of a Clamped
Circular Plate,” Poly. Inst. Brooklyn, XII, 1955, 397-401.
[150] Woinowsky-Krieger, S., “Buckling Stability of Circular
Plates with Circular Cylindrical Aeolotropy,” Ingenieur-
Archiv, 26, 1958, 129-131.
[151] Mossakowski, J., “Buckling of Circular Plates with Cylin-
drical Orthotropy” Arch. Mech. Stos. (Poland), 12, 1960,
583-596.
[152] Iwinski, T. and Nowinski, J., “The Problem of Large
Deflections of Orthotropic Plates,” Arch. Mech. Stos.
(Poland), 7, 1957, 593-602.
[153] Pandalai, K.A.V. and Patel, S.A., “Buckling of Or-
thotropic Circular Plates,” J. Royal Aerospace Soc., 69,
1965, 279-280.
[154] Huang, C.L., “Nonlinear Axisymmetric Flexural Vibra-
tion Equations of a Cylindrically Anisotropic Circular
Plate,” J. AIAA, 10, 1972, 1378-1379.
[155] Sherbourne, A.N. and Pandey, M.D., “Postbuckling of
Polar Orthotropic Circular Plates—Retrospective,” J.
Eng. Mech. Div., ASCE, 118, 1992, 2087-2103.
[156] Raju, K.K. and Rao, G.V., “Finite-Element Analysis of
Postbuckling Behavior of Cylindrically Orthotropic Cir-
cular Plates,” Fibre Sci. Tech., 19, 1983, 145-154.
[157] Turvey, G.J. and Drinali, H., “Elastic Postbuckling of
Circular and Annular Plates with Imperfections,” Proc.
3rd Int. Conf. Composite Struct., Appl. Sci. Pub., 1985,
315-335.
[158] Carrier, G.F., “Stress Distribution in Cylindrically Ae-
olotropic Plates,” J. Appl. Mech. Trans. ASME, 10, 1943,
A117-A122.
[159] Huang, C.L. and Sandman, B.E., “Large Amplitude Vi-
brations of a Rigidly Clamped Circular Plate,” Int. J.
Nonlinear Mech., 6, 1971, 451-466.
[160] Swamidas, A.S. and Kunukkasseril, V.X., “Buckling of
Orthotropic Circular Plates,” AIAA Journal, 11, 1963,
1633-1636.
[161] Antman, S.S. and Negrón-Marrero, P.V., “The Remark-
able Nature of Radially Symmetric Equilibrium States of
Aeolotropic Nonlinearly Elastic Bodies,” J. Elasticity, 18,
1987, 131-164.
[162] Yamaki, S., “Buckling of a Thin Annular Plate Under
Uniform Compression,” J. Appl. Mech., 25, 1958, 267-
273.
[163] Olsson, R.G., “Über axialsymmetrische Knickung dünner
Kriesringplatten,” Ingenieur-Archiv, 8, 1937, 449-452.
[164] Schubert, “Die Beullast dünner Kreisringplatten, die am
Aussen und Innenrand gleichmässigen Druckerfahren,”
Zeitschrift Für Angewandte Mathematik und Physik, 25,
1947, 123-124.
[165] Mansfield, E.H., “On the Buckling of an Annular Plate,”
Quart. J. Mech. Appl. Math., XIII, 1960, 16-23.
[166] Majumdar, S., “Buckling of a Thin Annular Plate Un-
der Uniform Compression,” AIAA Journal, 9, 1971, 1701-
1707.
[167] Machinek, A. and Troger, H., “Postbuckling of Elastic
Annular Plates,” Dynamics and Stability of Systems, 3,
1988, 79-88.
[168] Machinek, A.K. and Steindl, A., “Behandlung des Beul-
problems der Kreisringplatte mit Hilfe eines numerischen
Verfahrens,” Zeitschrift für Angewandte Mathematik und
Mechanik 66, 1986, T61-T63.
[169] Uthgenannt, E. and Brand, R., “Buckling of Orthotropic
Annular Plates,” AIAA Journal, 8, 1970, 2102-2104.
[170] Woinowsky-Krieger, S., “ÜberdieBeulsicherheit von
Kreisplatten mit Kreiszylindrischer Aeolotropic,”
Ingenieur-Archiv, 26, 1958, 129-133.
© 2001 by Chapm
[171] Rozsa, M., “Stability Analysis of Thin Annular Plates
Compressed Along the Outer or Inner Edge by Uniformly
Distributed Radial Forces,” Acta Technica Academiae
Scientiarum Hungaricae, 53, 1966, 359-377.
[173] Dumir, P.C., Nath, Y., and Gandhi, M., “Non-Linear Ax-
isymmetric Static Analysis of Orthotropic Thin Annular
Plates,” Int. J. Non-Linear Mechanics, 19, 1984, 255-272.
[176] Alwar, R.S. and Reddy, B., “Large Deflection Static and
Dynamic Analysis of Isotropic and Orthotropic Annular
Plates,” Int. J. Non-Linear Mech., 14, 1979, 347-359.
© 2001 by Chapm
[195] Minguet, P.J., Dugundji, J., and P. Lagace, “Postbuck-
ling Behavior of Laminated Plates Using a Direct Energy
Minimization Technique,” AIAAJ., 27 1989, 1785-1792.
[196] Engelstad, S.P., Reddy, J.N., and N.F. Knight, Jr., “Post-
buckling Response and Failure Predictions of Graphite-
Epoxy Plates Loaded in Compression,” AIAA J., 30,
1992, 2106-2113.
[204] Bauer, L., Keller, H.B., and E.L. Reiss, “Multiple Eigen-
values Lead to Secondary Bifurcation,” SIAM Review, 17,
1975, 101-122.
© 2001 by Chapm
[219] Misra, J.C., “On the Bending of a Heated Rectangular
Anisotropic Plate with Two Opposite Edges Simply Sup-
ported and Two Others Built In, Due to Different Tem-
perature Distributions on the Plane Surfaces,” Bull. Cal.
Math. Soc., 63, (1971), 183-190.
[220] Pell, W.H., “Thermal Deflections of Anisotropic Thin
Plates,” J. Soc. Indust. Appl. Math., IV, (1946), 27-44.
[221] Boley, B.A. and J.H. Weiner, Theory of Thermal Stresses,
(1960), Wiley Publishing, N.Y.
[222] Nowacki, W., Thermoelasticity, (1962), Pergamon Press,
Oxford, U.K.
[223] Hetnarski, R.B., “Basic Equations of the Theory of Ther-
mal Stresses,” in Thermal Stresses, vol. I, R.B. Hetnarski,
Ed., (1986), Elsevier Science Publishers.
[224] Kovalenko, A.D., Thermoelasticity, (1969), Wolters-
Noordhoff Publishing, Gröningen.
[225] Forray, M. and M. Newman, “Bending of Circular Plates
Due to Asymmetric Temperature Distribution,” J. of the
Aerospace Sciences, (1961), 773-778.
[226] Misra, J.C., “Note on the Thermal Bending of a Simply
Supported Rectangular Plate of Anisotropic Material,”
Indian J. Mech. Nath, 11, (1973), 33-36.
[227] Klosner, J.M. and M.J. Forray, “Buckling of Simply Sup-
ported Plates Under Arbitrary Symmetrical Temperature
Distributions,” J. of the Aerospace Sciences, 25, (1958),
181-184.
[228] Forray, M. and M. Newmann, “Axisymmetric Large De-
flections of Circular Plates Subjected to Thermal and Me-
chanical Loads,” J. of the Aerospace Sciences, 29, (1962),
1060-1066.
[229] Bednarczyk, H. and M. Richter, “Buckling of Plates Due
to Self-Equilibrated Thermal Stresses,” J. of Thermal
Stresses, 8, (1985), 139-152.
[230] Bargmann, H.W., “Thermal Buckling of Elastic Plates,”
J. of Thermal Stresses, 8, (1985), 71-98.
[231] Stavsky, Y., “Thermoelastic Stability of Laminated Or-
thotropic Circular Plates,” Acta Mechanica, 22, (1975),
31-51.
[232] Boley, S.R., “A Procedure for the Approximate Analy-
sis of Buckled Plates,” J. of the Aerospace Sciences, 22,
(1955), 336-338.
[233] Heldenfels, R. and W.M. Roberts, “Experimental and
Theoretical Determination of Thermal Stresses in a Flat
Plate,” NACA Tech. Note 2769, 1952.
[234] Wilcox, M.W. and L.E. Clemmer, “Large Deflection
Analysis of Heated Plates,” J. Eng. Mech. Div., ASCE,
6, (1964), 165-189.
[235] Levy, S., “Bending of Rectangular Plates with Large De-
flections,” NACA Tech. Note 846, 1942.
[236] Biswas, P., “Nonlinear Analysis of Heated Orthotropic
Plates,” Indian J. Pure Appl. Math., 12, (1981), 1380-
1389.
[237] Mansfield, E.H., “The Large Deflection Behavior of a
Thin Strip of Lenticular Section,” Quart. J. Mech. Appl.
Math., 12, (1959), 421-430.
[238] Sarkar, S.K., “Buckling of a Heated Circular Plate,” In-
dian. J. Math. Mech., V, (1967), 39-42.
[239] Vinokurov, S.G., “Thermal Stresses in Plates and Shells,”
(in Russian), Izv. Kas. Filyala, AN SSSR (1953).
[240] Berger, H.M., “A New Approach to the Analysis of Large
Deflection of Plates,” J. Appl. Mech., 22, (1955), 465-472.
[241] Nowinski, J.L. and H. Ohnabe, “On Certain Inconsisten-
cies in Berger Equations for Large Deflections of Plastic
Plates,” Int. J. Mech. Sci., 14, (1972), 165-170.
[242] Jones, R., Mazumdar, J., and Y.K. Cheung, “Vibration
and Buckling of Plates at Elevated Temperatures,” Int.
J. Solids Structures, 16, (1980), 61-70.
[243] Banerjee, B. and S. Datta, “A New Approach to an Anal-
ysis of Large Deflections of Thin Elastic Plates,” Int. J.
Non-Linear Mech., 16, (1981), 47-52.
© 2001 by Chapm
[244] Basuli, S., “Large Deflection Plate Problems Subjected to
Normal Pressure and Heating,” Indian J. Math. Mech.,
VI (1968), 22-30.
[245] Pal, M.C., “Large Deflections of Heated Circular Plates,”
Acta Mechanica, 8, (1969), 82-103.
[246] Biswas, P., “Large Deflections of Heated Orthotropic
Plates,” Indian J. Pure Appl. Math., 9, (1978), 1027-1032.
[247] Pal, M.C., “Large Amplitude Free Vibration of Circular
Plates Subjected to Aerodynamic Heating,” Int. J. Solid
Structures, 6, (1970), 301-314.
[248] Biswas, P., “Thermal Buckling of Orthotropic Plates,” J.
Appl. Mech., 21, (1996), 361-363.
[249] Pal, M.C., “Static and Dynamic Non-Linear Behavior of
Heated Orthotropic Circular Plates,” Int. J. Non-Linear
Mech., 8, (1973), 489-504.
[250] Williams, M.L., “Large Deflection Analysis for a Plate
Strip Subjected to Normal Pressure and Heating,” J.
Appl. Mech., (1955), 458-465.
[251] Schmidt, R., “On Berger’s Method in the Non-Linear
Theory of Plates,” J. Appl. Mech., 41, (1974), 521-523.
[252] Schmidt, R. and D. DaDeppo, “A New Approach to the
Analysis of Shells, Plates, and Membranes with Finite
Deflections,” Int. J. Non-Linear Mechanics, 9, (1974),
409-419.
[253] Mansfield, E.H., “On the Analysis of Elastic Plates of
Variable Thickness,” Quart. J. Mech. Appl. Math., XV,
(1962), 167-183.
[254] Mansfield, E.H., “Bending, Buckling, and Curling of
a Heated Thin Plate,” Proc. Roy. Soc. Ser. A., 1334,
(1962), 316-329.
[255] Biswas, P., “Buckling of a Heated Thin Annular Circu-
lar Plate of Variable Thickness,” Indian J. of Theoretical
Physics, 25, (1977), 107-112.
[256] Mendelson, A. and M. Herschberg, “Analysis of Elas-
tic Thermal Stresses in Thin Plates with Spanwise and
Chordwise Variations of Temperature and Thickness,”
NACA Tech. Note 3778, 1956.
[257] Z. Thrun, “O Odksztalceniach i naprezeniach temioznyeh
w cienkich plytach prostokatnych i kolowych o zmionuej
grubosei,” Rozprawy Inzynierskie, tom IV, zeszyt 4, 523-
41 Warsaw 1956.
[258] Olsson, R.G., “Beigung der Rechteckplatte bei linear
voranderlicker Biegungssteifigkeit,” Ingenieur-Archiv, 5
(1934), 363.
[259] Olsson, R.G., “Unsymmetrische Biogung der Kreis-
ringplate von quadratisch veranderlicker Steifikeit,”
Ingenieur-Archiv, 10 (1939) 14, 11 (1940), 25, and 13
(1942-3), 147.
[260] Mansfield, E.H., “The Inextensional Theory for Thin Flat
Plates,” Quart. J. Mech. Appl. Math., 8, (1955), 338-352.
[261] Ross, D.A. and L. Berke, “Temperature-Dependent Creep
Buckling of Plates,” J. of Thermal Stresses, 4, (1981),
237-247.
[262] Das, S., “Note on Thermal Deflection of Regular Polyg-
onal Viscoelastic Plates,,” Int. J. Mech. Sci., 23, (1981),
323-329.
[263] Williams, H.E., “Neale Plate Equations and Applica-
tions to Buckling of Rectangular Plates,” J. of Thermal
Stresses, 5, (1982), 207-243.
[264] Neale, K.W., “A General Variational Theorem for the
Rate Problem in Elasto-Plasticity,” Int. J. Solids Struc-
tures, 8, (1972), 865-876.
[265] Neale, K.W., “Effect of Imperfections on the Plastic
Buckling of Rectangular Plates,” J. Appl. Mech., 42,
(1975), 115-120.
[266] Norton, R.N., Creep in Steel at High Temperatures,
(1929), 115-120.
[267] Rabotnov, Yu N., Creep Problems in Structural Members,
(1969), North Holland Publishing, Amsterdam.
© 2001 by Chapm
[268] Timoshenko, S. and S. Woinowsky-Kreiger, Theorem of
Plates and Shells, 2nd ed., (1959), McGraw-Hill, N.Y.
[269] Hoff, N.J. and I.M. Levi, “Short Cuts in Creep Buckling
Analysis,” Int. J. Solids Structures, 8, (1972), 1103-1114.
[270] Ross, D.A., “Creep Buckling of a Flat Plate with
Through-Thickness Temperature Variation,” NASA
Tech. Memo. 81627, 1980.
[271] Deleeuw, S.L. and G.E. Mase, “Behavior of Viscoelastic
Plates Under the Action of Inplane Forces,” Proc. 4th
U.S. Nat. Cong. Appl. Mech., (1962), 999-1005.
[272] Mayers, J. and E. Nelson, “Elastic and Maximum
Strength Analysis of Postbuckled Rectangular Plates
Based Upon Modified Versions of Reissner’s Variational
Principle,” SUDAAR Report #262, (1966), Stanford Uni-
versity.
[273] Turvey, G.J., “Axisymmetric Elastic-Plastic Flexure of
Circular Plates in the Large Deflection Regime,” Proc.
Inst. Civil Eng., 67, (1979), 81-92.
[274] Myszkowski, J., “Endliche Durchbiegungen Beliebig
Eingespannier Dunner Kriesund Kriersring platten
im Plastischen Materialbereich,” Ingenieur-Archiv, 40,
(1971), 1-13.
[275] Danielson, D.A., “Theory of Shell Stability,” in Thin-
Shell Structures: Theory, Experiment, and Design, Y.C.
Fung and E.E. Sechler, Eds., (1974), Prentice-Hall, N.J.
[276] Gerard, G. and H. Becker, Handbook of Structural Sta-
bility, Part I: Buckling of Flat Plates, NACA Tech. Note
3781, 1957.