10

Some Exact Solutions for Specially

Orthotropic Laminates
The equations that govern the behavior of a symmetric composite plate
subjected to loads and/or displacements that vary along the edges of the plate
were derived in Chap. 9. In this chapter, exact solutions to the governing
equations will be obtained for a special class of symmetric laminates called
specially orthotropic laminates. The solutions presented allow calculation
of the deections caused by the combined eects of in-plane loads, transverse
loads, and changes in temperature or moisture content. Buckling caused by
in-plane loads and/or temperature changes is also discussed.
1 EQUATIONS OF EQUILIBRIUM FOR A SPECIALLY
ORTHOTROPIC LAMINATE
The equations of equilibrium for a symmetric composite laminate were
presented in Sec. 2.2 of Chap. 9. Boundary conditions consistent with thin-
plate theory were then discussed in Sec. 3 of Chap. 9. Unfortunately, exact
solutions to the equations of equilibrium that satisfy specied boundary con-
ditions often cannot be found. Sometimes, the diculty in obtaining an exact
solution is because of the boundary conditions involved; diculties of this
sort are encountered for both isotropic and anisotropic plates. In other cases,
the diculties are strictly because of the anisotropic nature of composites.
527
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
We have previously restricted our discussion to symmetric stacking
sequences, and so B
ij
M
T
ij
M
M
ij
= 0 in all cases considered. In addition,
in this chapter, we will further restrict our discussion to the so-called specially
orthotropic laminates. A specially orthotropic laminate is one for which A
16
A
26
D
16
D
26
N
T
xy
N
M
xy
= 0. It turns out that this is a very re-
strictive limitation. Indeed, a review of the stacking sequences presented in
Sec. 7 of Chap. 6 will reveal that there are only three stacking sequences that
eliminate these coupling stinesses and thermal/moisture resultants and can
therefore be classied as specially orthotropic. These are:

Unidirectional [0j]
n
laminates (described in Sec. 7.1 of Chap. 6).

Unidirectional [90j]
n
laminates (described in Sec. 7.1 of Chap. 6).

Symmetric cross-ply laminates, e.g., [(0/90)

n
]
s
(described in Sec. 7.2 of
Chap. 6).
It is mentioned in passing that the denition of a specially orthotropic
laminate is inconsistently used in the literature, and is sometimes relaxed
from the denition just described. Specically, the denition of a specially
orthotropic laminate may be relaxed if the laminate is symmetric and is
subjected to in-plane loading only. In problems of this type, the coupling
stinesses D
16
and D
26
play no role. Therefore researchers studying the
behavior of symmetric composites subjected to in-plane loads and displace-
ments (only) often specify that a specially orthotropic laminate is one in which
A
16
A
26
N
T
xy
N
M
xy
= 0, but place no restriction on the values of D
16
or D
26
. For example, Lekhnitski [1] and Savin [2] have obtained solutions that
allow calculation of the in-plane stresses induced in regions near elliptical,
rectangular, or triangular holes in anisotropic composite plates, and these
solutions are valid if the laminate is symmetric and subjected to in-plane loads
only. The solutions by Lekhnitski and Savin are said to be applicable to
specially orthotropic laminates, but in fact only require that A
16
A
26

N
T
xy
N
M
xy
= 0. Hence both symmetric balanced laminates and symmetric
angle-ply laminates (see Sec. 7 of Chap. 6) are specially orthotropic under this
relaxed denition. Nevertheless, in this chapter, we consider situations in
which the composite laminate is subjected to both in-plane and out-of-plane
loading, and the therefore the bending stinesses of the laminate plays an
important role. Therefore we must maintain the more restrictive denition
and stipulate that a specially orthotropic laminate must have A
16
A
26

D
16
D
26
N
T
xy
N
M
xy
= 0.
For a specially orthotropic laminate, the equations of equilibrium for
a symmetric laminate [i.e., Eq. (8) of Chap. 9] are simplied further still,
and become:
A
11
@
2
u
o
@x
2
A
12
A
66

@
2
v
o
@x@y
A
66
@
2
u
o
@y
2
0 1a
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
A
12
A
66

@
2
u
o
@x@y
A
22
@
2
v
o
@y
2
A
66
@
2
u
o
@x
2
0 1b
A
11
@u
o
@x
A
12
@v
o
@y
N
T
xx
N
M
xx
_ _
@
2
w
@x
2
_ _
1c
A
12
@u
o
@x
A
22
@v
o
@y
N
T
yy
N
M
yy
_ _
@
2
w
@y
2
_ _
2 A
66
@u
o
@y

@v
o
@x
_ _ _ _
@
2
w
@x@y
_ _
D
11
@
4
w
@x
4
2D
12
2D
66

@
4
w
@x
2
@y
2
D
22
@
4
w
@y
4
qx; y 0
In the following sections, we will rst obtain solutions for plate
deections under relatively simple loading conditions. It will then be seen
that solutions for simple loading condition form the basis for predicting
displacements caused by complex transverse loads.
There are, in essence, two techniques that can be used to obtain exact
solutions for problems involving arbitrary transverse loads. Both methods
place limitations on the type of boundary conditions that can be considered.
The rst approach was developed by Navier in about 1820, and is known as
the Navier solution [3]. The Navier solution is applicable if all four edges of
the specially orthotropic rectangular plate are simply supported. The Navier
solution applied to the case of a simply supported specially orthotropic panel
subjected to arbitrary transverse loading will be described in Sec. 6. The
second exact solution technique was developed by Levy in 1899, and is known
as the Levy solution [3]. In this case, two opposite edges of the plate must be
simply supported, while the boundary conditions on the remaining two edges
are arbitrary and may be clamped or free, for example. The Levy solution
technique will not be discussed in this textbook, and the interested reader is
referred to Refs. 15 for a discussion of this method.
Anal preliminary comment is that thin composite plates (or in fact any
thin-walled structure) are prone to buckling if subjected to compressive in-
plane loads and/or high shear loads. Buckling of specially orthotropic
laminates is discussed in Secs. 7 and 8, respectively.
2 IN-PLANE DISPLACEMENT FIELDS IN SPECIALLY
ORTHOTROPIC LAMINATES
The overall goal in this chapter is to predict deections induced in specially
orthotropic composite plates that are simply supported along all four edges.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
As discussed in Sec. 3 of Chap. 9, four types of simple supports may be dened
for a thin plate. For the edge x=a, these are:
S1: wa; y 0 M
*
xx
a; y 0 u
o
a; y u
x
o
y v
o
a; y v
x
o
y
S2: wa; y 0 M
*
xx
a; y 0 N
*
xx
a; y N
x
xx
y v
o
a; y v
x
o
y
S3: wa; y 0 M
*
xx
a; y 0 u
o
a; y u
x
o
y N
*
xy
a; y N
x
xy
y
S4: wa; y 0 M
*
xx
a; y 0 N
*
xx
a; y N
x
xx
y N
*
xy
a; y N
x
xy
y
Hence the denition of a simply supported edge requires that either in-plane
displacements, in-plane stress resultants, or some combination thereof must
be specied along the edge, in addition to the requirement that out-of-plane
displacements and bending moments vanish along the edge.
The following sequence of events is assumed to occur during fabrication
and assembly of a simply supported composite plate. We assume that the
laminate is cured at an elevated temperature, and that the laminate is stress-
and strain-free at the cure temperature.* Following cure, the laminate is
cooled to roomtemperature, and therefore midplane displacements (as well as
thermal stress resultants and associated ply strains and stresses) are induced
during cooldown to room temperature. The laminate is then trimmed to the
desired dimensions a b and assembled in a surrounding structure that
provides simple supports along all four edges. According to this scenario then,
midplane displacements have already been induced within the laminate prior
to assembly in the simple supports. Whether additional in-plane displace-
ments subsequently occur, due to application of q(x,y), a further change in
temperature, and/or a change in moisture content, depends on the type of
simple supports involved. That is, the development of additional in-plane
displacements depends on whether simple supports of type S1, S2, S3, or S4
are imposed along each edge of the plate.
All problems considered here will be based on the following. First, we
assume that opposite edges of the plate are subjected to the same type of
simple support. For example, if the edge x = 0 is subjected to the type S1
simple support, then by assumption the edge x = a is also subjected to type
S1 supports. Second, we assume that stress resultants N
xx
, N
yy
, and/or N
xy
applied to opposite edges of the plate (if any) are identical and uniformly
distributed along the edge. This loading condition is precisely equivalent to
*As mentionedin Section 6.2 of Chap. 6, in practice the stress- andstrain-free temperature is often
2050jC below the final cure temperature. This complication has been ignored throughout this
text, and it is assumed that the laminate is stress- and strain-free at the cure temperature.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
that assumed during the development of CLT in Chap. 6. Therefore we can
develop general expressions giving the in-plane displacement elds caused by
any combination of N
ij
, DT, and/or DM using a standard CLT analysis. We
will then specialize these general expressions to correspond to type S1, S2, S3,
or S4 simple supports.
To begin this process, we relate midplane strains and curvatures to stress
resultants using Eq. (45) of Chap. 6. For a symmetric specially orthotropic
laminate subjected to uniform in-plane stress resultants, a uniform change in
temperature, and/or a uniform change in moisture content, Eq. (45) of Chap.
6 becomes:
e
o
xx
e
o
yy
c
o
xy
j
xx
j
yy
j
xy
_

_
_

a
11
a
12
0 0 0 0
a
12
a
22
0 0 0 0
0 0 a
66
0 0 0
0 0 0 d
11
d
12
0
0 0 0 d
12
d
22
0
0 0 0 0 0 d
66
_

_
_

_
N
xx
N
T
xx
N
M
xx
N
yy
N
T
yy
N
M
yy
N
xy
0
0
0
_

_
_

_
As in earlier chapters, we assume innitesimal strain levels. Therefore mid-
plane strains are related to midplane displacements according to Eq. (10) of
Chap. 6:
e
o
xx

@u
o
@x
e
o
yy

@v
o
@y
c
o
xy

@u
o
@y

@v
o
@x
Consequently, midplane displacement elds are given by:
@u
o
@x
a
11
N
xx
N
T
xx
N
M
xx
a
12
N
yy
N
T
yy
N
M
yy
2a
@v
o
@y
a
12
N
xx
N
T
xx
N
M
xx
a
22
N
yy
N
T
yy
N
M
yy
2b
@u
o
@y

@v
o
@x
a
66
N
xy
2c
Integrating Eq. (2a) with respect to x, we nd:
u
o
x; y a
11
N
xx
N
T
xx
N
M
xx
a
12
N
yy
N
T
yy
N
M
yy

_ _
x f
1
y k
1
Where f
1
( y) is an unknown function of y (only) and k
1
is an unknown constant
of integration. Similarly, integrating Eq. (2b) with respect to y we nd:
v
o
x; y a
12
N
xx
N
T
xx
N
M
xx
a
22
N
yy
N
T
yy
N
M
yy

_ _
yf
2
x k
2
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
where f
2
(x) is an unknown function of x (only) and k
2
is a second unknown
constant. Without a loss in generality, we assume that midplane displace-
ments are zero at the origin (i.e., let u
o
= v
o
= 0 at x = y = 0), and con-
sequently we conclude k
1
= k
2
= 0. Substituting the above expressions for
u
o
(x,y) and v
o
(x,y) into Eq. (2c), we nd:
@u
o
@y

@v
o
@x

@f
1
@y

@f
2
@x
a
66
N
xy
Because the terms that appear on the right side of the equality (i.e., a
66
and
N
xy
) are known constants, it follows that f
1
and f
2
must be at most linear
functions of y and x, respectively:
f
1
y k
3
y
f
2
x k
4
x
Hence we can write:
k
3
k
4
a
66
N
xy
Constants k
3
and k
4
can take on any value as long as they sum to the product
(a
66
N
xy
). To determine particular values convenient for our use, we now
require that the innitesimal rotation vector in the xy plane, x
xy
(which
represents rigid body motion of the plate), is zero. The innitesimal rotation
vector is given by [see Ref. 3]:
x
xy

1
2
@u
o
@y

@v
o
@x
_ _
Requiring that x
xy
=0 leads to:
k
3
k
4

1
2
a
66
N
xy
Combining the preceding results, we conclude that in-plane midplane dis-
placement elds induced in a symmetric specially orthotropic composite panel
by the combination of uniform in-plane stress resultants, a uniform change in
te mperature, and/or a uniform change in moisture contents are given by:
u
o
x; y a
11
N
xx
N
T
xx
N
M
xx
a
12
N
yy
N
T
yy
N
M
yy

_ _
x
1
2
a
66
N
xy
y
v
o
x; y
1
2
a
66
N
xy
x a
12
N
xx
N
T
xx
N
M
xx

_
a
22
N
yy
N
T
yy
N
M
yy
y
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
These expressions can be simplied through the use of the eective thermal
expansion coecients of the laminate, dened by Eq. (73a) of Chap. 6, and the
eective moisture expansion coecients, dened by Eq. (76a) of Chap. 6. For
a specially orthotropic laminate, these become:
a
xx

1
DT
a
11
N
T
xx
a
12
N
T
yy
_ _
a
yy

1
DT
a
12
N
T
xx
a
22
N
T
yy
_ _
a
xy
0
b
xx

1
DM
a
11
N
M
xx
a
12
N
M
yy
_ _
b
yy

1
DM
a
12
N
M
xx
a
22
N
M
yy
_ _
b
xy
0
Hence we see that the midplane displacement elds can be written as:
u
o
x; y a
11
N
xx
a
12
N
yy
DTa
xx
DMb
xx
_
x
1
2
a
66
N
xy
y 3a
v
o
x; y
1
2
a
66
N
xy
x a
12
N
xx
a
22
N
yy
DTa
yy
DMb
yy
_
y 3b
Note that the displacement elds are independent of the transverse load
q(x,y). As pointed out in Sec. 2 of Chap. 9, u
o
(x,y) and v
o
(x,y) are predicted to
be independent of q(x,y) because we have assumed displacement gradients are
small, such that gradients squared can be ignored [e.g., (@w/@x)
2
c0]. Analy-
ses that account for large displacement gradients are not considered in this
text. If we had included large gradients in our analysis, then expressions for
u
o
(x,y) and v
o
(x,y) corresponding to Eqs. (3a) and (3b) would depend on
transverse load q(x,y).
Direct substitution of Eqs. (3a) and (3b) will reveal that these equations
satisfy the equations of equilibrium, Eqs. (1a) and (1b). In the following
sections, we will use these expressions to specify in-plane displacement elds
associated with simple supports of type S1 through S4.
3 SPECIALLY ORTHOTROPIC LAMINATES SUBJECT
TO SIMPLE SUPPORTS OF TYPE S1
In this section, we consider the specially orthotropic plate shown in Fig. 1.
The plate is assumed to be rectangular with thickness t and in-plane
dimensions a b. All four edges of the plate are subject to simple supports
of type S1, where it is assumed that the laminate was mounted in the structure
that imposes type S1 supports following cooldown from the cure temperature
to room temperature. After assembly, the plate is subjected to a uniform
change in temperature and a transverse loading that varies over the xy plane
according to:
qx; y q
o
sin
px
a
_ _
sin
py
b
_ _
4
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
We will not consider a change in moisture content (i.e., let DM=0), although
from earlier discussion it should be clear that a change in moisture content
can be accounted for (in a mathematical sense) using the same techniques
used to model uniform changes in temperature.
Let us dene DT
c
as the change in temperature from cure to room
temperature:
DT
c
T
RT
T
C
where T
RT
= room temperature and T
C
= the cure temperature. For a
symmetric specially orthotropic laminate, the in-plane displacements caused
by DT
c
can be calculated using Eqs. (3a) and (3b) with N
xx
=N
yy
=N
xy
=0:
u
c
o
x; y a
xx
DT
c
x 5a
v
c
o
x; y a
yy
DT
c
_ _
y 5b
These displacements are induced before assembly of the simply supported
plate. Note that these in-plane displacement elds satisfy the rst two
equations of equilibrium, Eqs. (1a) and (1b). A type S1 simple support will
simply maintain these displacements during subsequent loading and/or
Figure 1 Thin rectangular plate of thickness t and in-plane dimensions a b,
subjected to a transverse load q(x,y) = q
o
{sin[(px)/a]}{sin[(py)/b]}. All four
edges of the plate are subject to simple supports of type S1.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
temperature changes. Therefore type S1 boundary conditions for all four
edges of the plate become:
For x 0 For x a :
w0; y 0 wa; y 0
M
*
xx
0; y D
11
@
2
w
@x
2
D
12
@
2
w
@y
2
0 M
*
xx
a; y D
11
@
2
w
@x
2
D
12
@
2
w
@y
2
0
u
o
0; y u
x
o
y 0 u
o
a; y u
x
o
y a
xx
DT
c
a
v
o
0; y v
x
o
y a
yy
DT
c
_ _
y v
o
a; y v
x
o
y a
yy
DT
c
_ _
y
For y 0 For y b :
wx; 0 0 wx; b 0
M
*
yy
x; 0 D
11
@
2
w
@x
2
D
12
@
2
w
@y
2
0 M
*
yy
x; b D
11
@
2
w
@x
2
D
12
@
2
w
@y
2
0
u
o
x; 0 u
y
o
x a
xx
DT
c
x u
o
x; b u
x
o
x a
xx
DT
c
x
v
o
x; 0 v
y
o
x 0 v
o
x; b v
x
o
y a
yy
DT
c
_ _
b
We wish to determine the out-of-plane displacement eld w(x,y) that
satises these boundary conditions as well as the equations of equilibrium,
when the plate is subjected to a transverse loading q(x,y) and/or a further
change in temperature. Guided by the functional form of the transverse
pressure [i.e., Eq. (4)], we assume that the out-of-plane displacement eld is
given by:
wx; y c sin
px
a
_ _
sin
py
b
_ _
6
where c is an unknown constant. Substituting this assumed form into the
boundary conditions will reveal that they are identically satised for any value
of c. Hence the value of constant c must be determined by enforcing the third
equation of equilibrium, Eq. (1c). We perform the following operations:
(a) substitute Eqs. (4), (5a), (5b), and (6) into the third equation of
equilibrium, Eq. (1c);
(b) write the thermal stress resultants N
xx
T
and N
yy
T
in terms of eec-
tive thermal expansion coecients [using Eq. (73b) of Chap. 6]; and
then
(c) solve the resulting expression for constant c.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Following this process we nd:
c
q
o
p
4
D
11
a
4

2
a
2
b
2
D
12
2D
66

D
22
b
4

DT
c
DT
p
2
1
a
2
A
11
a
xx
A
12
a
yy
_

1
b
2
A
12
a
xx
A
22
a
yy
_
_ _ _ _
7a
Notice that the temperature change as dened in earlier chapters (DT) ap-
pears in Eq. (7a). That is:
DT current temperature cure temperature
Also note that if the temperature is not changed following assembly in the
type S1 simple supports, then DT=DT
c
, and the eects of temperature cancel
in Eq. (7a).
Results from thin-plate theory are often expressed in terms of the so-
called plate aspect ratio, R = a/b. Equation (7a) can be rewritten using the
aspect ratio as follows:
c
q
o
R
4
b
4
p
4
D
11
2R
2
D
12
2D
66
R
4
D
22

DT
c
DTa
2
p
2
a
xx
A
11
R
2
A
12
_ _
a
yy
A
12
R
2
A
22
_ _ _ _
_ _
7b
The predicted out-of-plane deection is obtained by combining either
Eq. (7a) or (7b) with Eq. (6). Using Eq. (7b) for example, we have:
wx; y
q
o
R
4
b
4
sinpx=asinpy=b
p
4
D
11
2R
2
D
12
2D
66
R
4
D
22

DT
c
DTa
2
p
2
a
xx
A
11
R
2
A
12
_ _
a
yy
A
12
R
2
A
22
_ _ _ _
_ _
8
Equations (5a), (5b), and (8) give the predicted displacement eld induced in
the plate and represent the solution to this problem.
To summarize, we have considered a symmetric specially orthotropic
plate subjected to type S1 simple-supports. We have assumed that the plate is
mounted within simple supports while at roomtemperature. The laminate has
therefore likely experienced a change in temperature prior to assembly be-
cause modern composites are typically cured at an elevated temperature. The
change in temperature associated with cooldown to room temperature is
represented by DT
c
. After assembly, the plate is subjected to a sinusoidally
varying transverse loading and/or the temperature is changed away from
room temperature. The resulting displacement elds are given by Eqs. (5a),
(5b), and (8). A typical application of this solution is discussed in Sample
Problem 1.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
As a closing comment, it should be kept in mind that we have not yet
considered the possibility of buckling. The solution presented here is not valid
if the change in temperature is such that the resulting in-plane stress resultants
are compressive and have a magnitude large enough to cause buckling. The
phenomenon of buckling induced by a change in temperature is called
thermal buckling and will be discussed in Sec. 8.
Sample Problem 1
A [0
2
/90)
2
]
s
graphiteepoxy laminate is cured at 175jC and then cooled to
room temperature (20jC). After cooling, the at laminate is trimmed to in-
plane dimensions of 300 150 mmand mounted in an assembly that provides
type S1 simple supports along all four edges. The x axis is dened parallel to
the 300 mm edge (i.e., a = 0.3 m; b = 0.15 m). The laminate is then subjected
to a transverse pressure given by q(x,y) = 40 {sin[(px)/a]}{sin[(py)/b]} (kPa)
and a uniform temperature change. No change in moisture content occurs
(DM = 0). Plot the maximum out-of-plane displacement as a function of
temperature, over the range 50jC<T<20jC. Use the properties listed for
graphiteepoxy in Table 3 of Chap. 3, and assume each ply has a thickness of
0.125 mm.
Solution. The rectangular plate is a 12-ply laminate with total thickness
t = 12(0.125 mm) = 1.5 mm and aspect ratio R = a/b = (0.3 m)/(0.15 m) =
2.0. Out-of-plane displacements are given by Eq. (8), and hence elements of
the [ABD] matrix are required. Based on the properties listed in Table 3 of
Chap. 3 for graphiteepoxy and the specied stacking sequence, the [ABD]
matrix is:
ABD
1:76 10
6
4:52 10
6
0 0 0 0
4:52 10
6
95:6 10
6
0 0 0 0
0 0 19:5 10
6
0 0 0
0 0 0 40:1 0:848 0
0 0 0 0:848 10:8 0
0 0 0 0 0 3:66
_

_
_

_
where the units of A
ij
are Pa m and the units of D
ij
are Pa m
3
.
We also require the eective thermal expansion coecients. Based on
the properties listed in Table 3 of Chap. 3 for graphiteepoxy and the specied
stacking sequence, these are:
a
xx
0:29 lm=m jC a
yy
2:44 lm=m jC
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Because the laminate is specially orthotropic, the eective shear thermal
expansion coecient is zero (a
xy
= 0). The cooldown from the cure temper-
ature to room temperature is:
DT
c
20jC 175jC 155jC
Following assembly, the temperature ranges from room temperature to as
low as 50jC. Therefore:
255jC < DT < 155jC
Note that the temperatures to be considered are all at or below room tem-
perature. Because the eective thermal expansion coecients are algebrai-
cally positive, if the laminate were not constrained by the S1 simple supports,
it would tend to contract as temperature is lowered. Because it is in fact
constrained by the simple supports, the in-plane stress resultants that develop
as temperature is lowered tend to be tensile. Therefore thermal buckling is not
of concern. Substituting all known values, Eq. (8) becomes:
wx; y
324
p
4
278
88:6DT
c
DT
p
2
_ _
sin
px
0:30
_ _
sin
py
0:15
_ _
meters
This expression can be used to calculate the out-of-plane displacement in-
duced at any point (x,y) over the surface of the plate. The maximum out-of-
Figure 2 Maximum out-of-plane displacement for the graphiteepoxy plate
considered in Sample Problem 1 as a function of temperature.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
plane displacement occurs at the center of the plate, i.e., at x = a/2 = 0.15 m
and y = b/2 = 0.075 m. Because DT
c
= 20jC175jC = 155jC, and tem-
peratures ranging from50jC<T<20jCare to be considered, the quantity
(DT
c
DT) ranges from 70jC z(DT
c
DT) z0jC in this problem. A plot
of maximumout-of-plane displacement as a function of temperature is shown
in Fig. 2. At room temperature (20jC), a maximum deection of 12 mm is
predicted. As wouldbe expected, the plate becomes stier as the temperature is
decreased, in the sense that out-of-plane displacements are decreased because
of the in-plane tensile loads that develop as temperature is decreased. At the
lowest temperature considered (50jC), a maximum deection of 3.7 mm is
predicted.
4 SPECIALLY ORTHOTROPIC LAMINATES SUBJECT
TO SIMPLE SUPPORTS OF TYPE S4
In this section, we consider the specially orthotropic plate shown in Fig. 3. As
in the preceding section, the plate is rectangular with thickness t and in-plane
dimensions a b. However, we now assume that each edge of the plate is
subject to simple supports of type S4, rather than type S1. Hence displacement
Figure 3 Thin rectangular plate of thickness t and in-plane dimensions a b,
subjected to a transverse load q(x,y) = q
o
sin(
px
a
)sin(
py
b
). All four edges of the
plate are subject to simple supports of type S4 (compare with Figure 1).
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
elds along the plate edges are not required to take on any specied value.
Rather, we assume that a uniform normal stress resultant is applied to each
edge, as shown in Fig. 3. Note that shear loading is not considered: N
xy
=
N
yx
= 0. The plate is also subjected to a uniform change in temperature, DT,
and a transverse loading that varies over the xy plane according to:
qx; y q
o
sin
px
a
_ _
sin
py
b
_ _
9
We will not consider any change in moisture content, i.e., DM = 0. Also, we
will not consider the possibility of buckling in this section, although buckling
is a possibility if either of the in-plane normal stress resultants is compressive.
Buckling under type S4 simple supports will be discussed in Sec. 9.
The boundary conditions that dene type S4 simple support were
discussed in Sec. 3.3.2 of Chap. 9. Because the stress resultants applied along
each edge are uniform and constant, we can write:
N
*
xx
0; y N
*
xx
a; y N
xx
N
*
yy
x; 0 N
*
yy
x; b N
yy
N
*
xy
0; y N
*
xy
a; y N
*
yx
x; 0 N
*
yx
x; b 0
Because we have limited consideration to symmetric specially orthotropic
laminates, A
16
A
26
D
16
D
26
B
ij
N
T
xy
N
M
xy
M
T
ij
M
M
ij
=0.
Therefore the boundary conditions can be written as:
For x = 0,a:
w0; y wa; y 0 10a
M
*
xx
0; y M
o
xx
a; y D
11
@
2
w
@x
2
D
12
@
2
w
@y
2
0 10b
N
*
xx
0; y N
*
xx
a; y A
11
@u
o
@x
A
12
@v
o
@y
N
T
xx
N
xx
10c
N
*
xy
0; y N
*
xy
a; y A
66
@u
o
@y

@v
o
@x
_ _
0 10d
For y = 0,b:
wx; 0 wx; b 0 11a
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
M
*
yy
x; 0 M
*
yy
x; b D
12
@
2
w
@x
2
D
22
@
2
w
@y
2
0 11b
N
*
yy
x; 0 N
*
yy
x; b A
12
@u
o
@x
A
22
@v
o
@y
N
T
yy
N
yy
11c
N
*
yx
x; 0 N
*
yx
x; b A
66
@u
o
@y

@v
o
@x
_ _
0 11d
Using Eqs. (3a) and (3b) for the assumed conditions (i.e., N
xy
=DM=0), the
in-plane displacements elds are:
u
o
x; y a
11
N
xx
a
12
N
yy
DTa
xx
_
x 12a
v
o
x; y a
12
N
xx
a
22
N
yy
DTa
yy
_
y 12b
Let us conrm that these equations satisfy the appropriate boundary con-
ditions. Substituting Eqs. (12a) and (12b) into boundary condition (10c) and
rearranging, we nd that the following expression must be satised:
N
xx
A
11
a
11
A
12
a
12
N
yy
A
11
a
12
A
12
a
22

DTA
11
a
xx
A
12
a
yy
N
T
xx
N
xx
Because the laminate is specially orthotropic,
A
A
11
A
12
0
A
12
A
22
0
0 0 A
66
_
_
_
_
a A
1

A
22
A
11
A
22
A
2
12

A
12
A
11
A
22
A
2
12

0
A
12
A
11
A
22
A
2
12

A
11
A
11
A
22
A
2
12

0
0 0
1
A
66
_

_
_

_
Therefore by direct substitution we nd:
A
11
a
11
A
12
a
12
1
A
11
a
12
A
12
a
22
0
Also, from Eq. (73b) of Chap. 6, we nd (for A
16
= 0):
N
T
xx
DTA
11
a
xx
A
12
a
yy

Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.

Hence the boundary condition represented by Eq. (10c) is satised identically
by the in-plane displacement elds. A similar process can be used to conrm
that the boundary condition given by Eq. (11c) is also satised. Both Eqs.
(10d) and (11d) are satised as well, because @u
o
/@y = @v
o
/@x = 0.
Now consider out-of-plane displacements w(x,y). Guided by the func-
tional form of the transverse pressure [i.e., Eq. (9)], we once again assume the
out-of-plane displacement eld is given by:
wx; y c sin
px
a
_ _
sin
py
b
_ _
13
where c is an unknown constant. Substituting this assumed form into
boundary conditions [Eqs. (10a), (10b), (11a), and (11b)] will reveal that they
are identically satised for any value of c. Hence the value of constant c must
be determined by enforcing the third equation of equilibrium. Substituting
Eqs. (12a), (12b), and (13) into the third equation of equilibrium, Eq. (1c),
and solving for constant c, we nd:
c
q
o
p
4
1
a
4
D
11

2
a
2
b
2
D
12
2D
66

1
b
4
D
22

1
p
2
a
2
N
xx

1
p
2
b
2
N
yy
_ _
14a
Using the denition of the plate aspect ratio, R = a/b, this result can also be
written as:
c
q
o
R
4
b
4
p
4
D
11
2R
2
D
12
2D
66
R
4
D
22

a
2
p
2
N
xx
N
yy
R
2
_ _
_ _ 14b
The predicted out-of-plane deection is obtained by combining either Eq.
(14a) or (14b) with Eq. (13). Using Eq. (14b) for example, we have:
wx; y
q
o
R
4
b
4
sinpx=asinpy=b
p
4
D
11
2R
2
D
12
2D
66
R
4
D
22

a
2
p
2
N
xx
N
yy
R
2
_ _
_ _ 15
Equations (12a), (12b), and (15) give the predicted displacement elds in-
duced in the plate and represent the solution to this problem.
To summarize, we have found the displacement elds induced in a
symmetric specially orthotropic type S4 simply supported plate subjected to a
sinusoidally varying transverse load, a uniform change in temperature DT,
and uniform stress resultants N
xx
and N
yy
. A typical application of this
solution is discussed in Sample Problem 2. It should be kept in mind that we
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
have not considered the possibility of buckling in this section. The solution we
have obtained is not valid if N
xx
and/or N
yy
are compressive and have
magnitudes large enough to cause buckling. Buckling under type S4 simple
supports will be discussed in Sec. 9.
Sample Problem 2
A [(0
2
/90)
2
]
s
graphiteepoxy laminate is cured at 175jC and then cooled to
room temperature (20jC). After cooling, the at laminate is trimmed to in-
plane dimensions of 300150 mm and mounted in an assembly that provides
type S4 simple supports along all four edges. The x axis is dened parallel to
the 300-mm edge (i.e., a = 0.3 m; b = 0.15 m). The laminate is then subjected
to a uniform in-plane tensile loading (i.e., N
xx
= N
yy
) and transverse pressure
given by q(x,y) = 40 {sin [(px)/a]} {sin[(py)/b]} (kPa). Temperature remains
constant and no change in moisture content occurs (DM = 0).
(a) Plot the out-of-plane displacements induced along the centerline
dened by y = 0.075 m, if in-plane loads N
xx
= N
yy
= 50 kN/m
are applied.
(b) Plot the maximum out-of-plane displacement as a function of in-
plane loads, over the range 0<(N
xx
= N
yy
)<70 kN/m.
(c) Compare the maximum out-of-plane displacement caused by the
specied transverse load at room temperature if the plate is sub-
ject to
(i) type S1 simple supports (as discussed in Sec. 3), and
(ii) type S4 simple supports, with N
xx
= N
yy
= 0
Use the properties listed for graphiteepoxy in Table 3 of Chap. 3, and assume
each ply has a thickness of 0.125 mm.
Solution. A [(0
2
/90)
2
]
s
graphiteepoxy laminate was also considered in
Sample Problem 1, and numerical values for the [ABD] matrix are listed
there. As before, the 12-ply laminate has a total thickness t = 1.5 mm and
aspect ratio R = a/b = 2.0. Using these laminate stinesses, dimensions, and
the specied transverse loading, Eq. (15) becomes:
wx; y
324
p
4
278
0:090
p
2
N
xx
4N
yy
_ _
_ _
_

_
_

_
sin
px
0:3
_ _
sin
py
0:15
_ _
meters
This expression can be used to calculate the out-of-plane displacement
induced at any point (x,y) over the surface of the plate.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Part (a). Using N
xx
N
yy
= 50 kN/m and y = 0.075 m, out-of-plane dis-
placements are:
wx; y 1:30 10
3
sin
px
0:3
_ _
meters
A plot of these displacements over the length of the plate (0<x<0.3m is
showninFig. 4(a). Displacements are zeroat the edges denedby x=0, 0.3 m,
as dictated by the specied boundary conditions. As would be expected
because of symmetry, out-of-plane displacement is maximum at the center
of the plate, and equals 1.3 mm for the loading considered.
Part (b). The maximum out-of-plane displacement occurs at the center of
the plate, i.e., at x = a/2 = 0.15 m and y = b/2 = 0.075 m, and at this point
the maximum out-of-plane displacement is given by:
wj
max

324
p
4
278
0:090
p
2
N
xx
4N
yy
_ _
_ _
_

_
_

_
meters
Aplot of maximumout-of-plane displacement as a functionof in-plane tensile
loads is shown in Fig. 4b. As would be intuitively expected, the plate is
stiened by the application of in-plane loading. That is, the maximum out-of-
plane displacement is decreased as in-plane tensile loads are increased. A
maximumdeection of 12 mmoccurs when N
xx
=N
yy
=0, whereas the max-
imum deection is reduced to 0.96 mm when N
xx
= N
yy
= 70 kN/m.
Part (c). This same panel and transverse loading was considered in
Sample Problem 1, except type S1 simple supports were assumed. Thus in
Sample Problem 1 in-plane displacements were xed and were not allowed
to change when the transverse load was applied. In contrast, type S4 simple
supports are assumed in this problem; in-plane stress resultants are specied
rather than in-plane displacements.
Referring to the results presented in these two sample problems, we
nd that identical deections are predicted, despite the dierences in
boundary conditions. That is, a maximum deection of 12 mm is predicted
at room temperature for type S1 condition, and an identical 12 mm
deection is predicted if N
xx
= N
yy
= 0 for type S4 conditions. This result
may seem nonintuitive and (rigorously speaking) is incorrect. That is, for
type S4 boundary conditions, a transverse loading will cause a change in in-
plane displacements. Therefore one might anticipate that the out-of-plane
displacement for type S4 conditions would be increased, relative to type S1
conditions. However, the relative increase is very small if displacement
gradients are small. Thus the relative increase in out-of-plane displacements
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Figure 4 Out-of-plane displacements induced in the graphiteepoxy plate
considered in Sample Problem 2. (a) Out-of-plane displacements induced along
centerline y = 0.075 m when N
xx
= N
yy
= 50 kN/m, (b) Maximum out-of-plane
displacements as a function of in-plane loads N
xx
= N
yy
.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
for type S4 conditions is not predicted because we have based our analysis
on innitesimal strains. The consequences of the innitesimal strain as-
sumption were alluded to in Sec. 2.2 of Chap. 9. It was noted there that this
assumption ultimately leads to the conclusion that in-plane displacement
elds u
o
(x,y) and v
o
(x,y) are independent of the transverse load, q(x,y). The
comparison between the results of Sample Problems 1 and 2 presented here
is an illustration of this independence.
Of course, results for the two dierent boundary conditions are iden-
tical because we have considered the case in which N
xx
= N
yy
= 0. If N
xx
and/or N
yy
p 0, then the transverse displacements for a plate supported by
type S4 simple supports is quite dierent from that of a plate supported by
type S1 supports.
5 SPECIALLY ORTHOTROPIC LAMINATES WITH TWO
SIMPLY SUPPORTED EDGES OF TYPE S1 AND TWO
EDGES OF TYPE S2
In this section, we consider the specially orthotropic plate shown in Fig. 5. As
in the preceding section, the plate is rectangular with thickness t and in-plane
Figure 5 Thin rectangular plate of thickness t and in-plane dimensions a x b,
subjected to a transverse load q(x,y) = q
o
{sin[(px)/a]}{sin[(py)/b]}. Edges x =
0,a are subject to simple supports of type S2, whereas edges y =0,b are subject
to simple support of type S1 (compare with Figs. 1 and 3).
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
dimensions a b. We now assume that the two edges x = 0,a are subjected to
simple supports of type S2. That is, we specify that these two edges are
subjected to known stress resultants N
xx
and known displacements v
o
. In
contrast, the twoedges y =0,b are subject tosimple supports of type S1, which
means that we require these edges to maintain known in-plane displacements
u
o
and v
o
. The plate is also subjected to a uniform change in temperature, DT,
and a transverse loading that varies over the xy plane according to:
qx; y q
o
sin
px
a
_ _
py
b
_ _
16
We will not consider a change in moisture content, i.e., DM = 0. Also, we will
not consider the possibility of buckling in this section, although buckling is a
possibility if N
xx
is compressive or if the change in temperature DT tends to
cause the laminate to expand.
As in earlier sections, we assume the laminate was cured at an elevated
temperature and cooled to room temperature prior to assembly in a sur-
rounding structure that provides simple supports. Consider the midplane
strains v
o
(x,y) induced during cooldown. Using Eq. (3b), this can be written
(with N
xx
= N
yy
= N
xy
= DM = 0):
v
o
x; y DT
c
a
yy
y 17
We can use this expression to specify known values of v
o
along all four edges
of the plate. Because a known stress resultant N
xx
is applied to the two edges
x = 0 and x = a, all quantities necessary to dene the type S2 simple support
along these two edges are known. To dene the type S1 simple supports along
the edges y = 0,b, we must specify known values of u
o
. Toward that end,
equate Eqs. (3b) and (17):
1
2
a
66
N
xy
x a
12
N
xx
a
22
N
yy
DT a
yy
_
y DT
c
a
yy
y
This relation must be satised along the edges y =0,b. Substituting y =0, it is
seen that N
xy
must vanish (N
xy
= 0). Substituting y = b, and solving for N
yy
,
we obtain:
N
yy

a
yy
DT
c
DT a
12
N
xx
a
22
18
Equation (18) gives the stress resultant N
yy
that must be provided by the
simple supports along the y = 0,b so as to maintain the stipulated value of v
o
represented by Eq. (17). Note that if temperature does not change from room
temperature (i.e., if DT =DT
c
), and if no stress resultant N
xx
is applied (i.e.,
if N
xx
= 0), then N
yy
= 0, as would be expected.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
We can now calculate the displacement u
o
implied by these conditions,
using Eq. (3b):
u
o
x; y a
11
N
xx
a
12
a
yy
DT
c
DT a
12
N
xx
a
22
_ _
DT a
xx
_ _
x
This expression can be written as:
u
o
x; y N
xx
a
11

a
2
12
a
22
_ _

a
12
a
22
a
yy
DT
c
DT a
xx
DT
_ _
x 19
Equations (17) and (19) can be used to specify the displacement boundary
conditions along all four edge of the plate. A summary of all boundary con-
ditions associated with the problem considered in this section is:
For x = 0,a
w0; y wa; y 0 20a
M
*
xx
0; y M
*
xx
a; y D
11
@
2
w
@x
2
D
12
@
2
w
@y
2
0 20b
N
*
xx
0; y N
*
xx
a; y A
11
@u
o
@x
A
12
@v
o
@y
N
T
xx
N
xx
20c
v
o
0; y v
o
a; y DT
c
a
yy
y 20d
For y = 0,b:
wx; 0 wx; b 0 21a
M
*
xx
x; 0 M
*
yy
x; b D
12
@
2
w
@x
2
D
22
@
2
w
@y
2
0 21b
u
o
x; 0 u
o
x; b
N
xx
a
11

a
2
12
a
22
_ _

a
12
a
22
a
yy
DT
c
DT a
xx
DT
_ _
x 21c
v
o
x; 0 0 v
o
x; b DT
c
a
yy
b 21d
Now consider out-of-plane displacements w(x,y). Guided by the func-
tional formof the transverse pressure [i.e., Eq. (16)], we once again assume the
out-of-plane displacement eld is given by:
wx; yc sin
px
a
_ _
sin
py
b
_ _
22
where c is an unknown constant. Substituting this assumed form into
boundary conditions Eqs. (20a), (20b), (21a), and (21b) will reveal that they
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
are identically satised for any value of c. Hence the value of constant c must
be determined by enforcing the third equation of equilibrium. Substituting
Eqs. (19), (17), and (22) into the third equation of equilibrium, Eq. (1c), and
solving for constant c, we nd:
c
q
o
p
4
D
11
a
4

2D
12
2D
66

a
2
b
2

D
22
b
4

N
xx
p
2
1
a
2

A
12
A
11
b
2
_ _

a
yy
A
12
A
22
A
2
12

p
2
b
2
A
11
DT
c
DT
_ _
23
Using the denition of the plate aspect ratio, R = a/b, this result can also be
written as:
c
q
o
R
4
b
4
p
4
D
11
2R
2
D
12
2D
66
R
4
D
22

R
4
b
2
p
2
N
xx
1
R
2

A
12
A
11
_ _

a
yy
A
12
A
22
A
2
12

A
11
DT
c
DT
_ _ _ _
24
The predicted out-of-plane deection is obtained by combining either Eq.
(23) or (24) with Eq. (22). Using Eq. (24), for example, we have:
wx; y
q
o
R
4
b
4
sinpx=asinpy=b
p
4
D
11
2R
2
D
12
2D
66
R
4
D
22

R
4
b
2
p
2
N
xx
1
R
2

A
12
A
11
_ _

a
yy
A
12
A
22
A
2
12

A
11
DT
c
DT
_ _ _ _
25
Equations (17), (19), and (25) give the predicted displacement elds induced
in the plate and represent the solution to this problem.
To summarize, we have found the displacement elds induced in a
symmetric specially orthotropic plate subjected to a sinusoidally varying
transverse load and a uniform change in temperature DT. The two edges x =
0,a are subject to simple supports of type S2, whereas the two edges y = 0,b
are subject to simple supports of type S1. Atypical application of this solution
is discussed in Sample Problem 3. It should be kept in mind that the
possibility of buckling has not been considered. Buckling is a possibility if
N
xx
is compressive or if the change in temperature DT tends to cause the
laminate to expand.
Sample Problem 3A
[(0
2
/90)
2
]
s
graphiteepoxy laminate is cured at 175jC and then cooled to
room temperature (20jC). After cooling, the at laminate is trimmed to in-
plane dimensions of 300 150 mmand mounted in an assembly that provides
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
type S2 simple supports along the two edges x = 0 and x = a, and type
S1 simple supports along the two edges y = 0 and y = b. The x axis is
dened parallel to the 300-mm edge (i.e., a = 0.3 m; b = 0.15 m). The
laminate is then subjected to a uniformin-plane tensile loading N
xx
along x =
0,a and a transverse pressure given by q(x,y) = 40 {sin[(px)/a]}{sin[(py)/b]}
(kPa).
(a ) Plot the maximumout-of-plane displacement as a function of tensile
load over the range 0<N
xx
< 70 kN/m, assuming temperature
remains constant at room temperature.
(b) Plot the maximum out-of-plane displacement as a function of
temperature, over the range 50jC<T<20jC, assuming a con-
stant in-plane tensile load N
xx
= 50 kN/m.
Use the properties listed for graphiteepoxy in Table 3 of Chap. 3, and assume
each ply has a thickness of 0.125 mm.
Solution. A [(0
2
/90)
2
]
s
graphiteepoxy laminate was also considered in
Sample Problem 1, and numerical values for the [ABD] matrix and eective
thermal expansion coecients are listed there. As before, the 12-ply laminate
has a total thickness t = 1.5 mm and aspect ratio R = a/b = 2.0. Using these
laminate stinesses, dimensions, and the specied transverse loading, Eq. (25)
becomes:
wx; y
324
p
4
278
0:099
p
2
N
xx

82:5
p
2
DT
c
DT
_ _
_

_
_

_
sin
px
0:3
_ _
sin
py
0:15
_ _
meters
This expression can be used to calculate the out-of-plane displacement
induced at any point (x,y) over the surface of the plate.
Part (a). The maximum out-of-plane displacement occurs at the center of
the plate, i.e., at x = a/2 = 0.15 m and y = b/2 = 0.075. Because the plate
remains at room temperature, DT =DT
c
, under these conditions the
maximum out-of-plane displacement is given by:
wj
max

324
p
4
278
0:099
p
2
N
xx
_ _
meters
A plot of maximum out-of-plane displacement as a function of N
xx
is shown
in Fig. 6(a). As would be intuitively expected, the plate is stiened as N
xx
is
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Figure 6 Out-of-plane displacements induced in the graphiteepoxy plate
considered in Sample Problem 3. (a) Maximum out-of-plane displacements at
room temperature as a function of in-plane load N
xx
. (b) Maximum out-of-plane
displacements as a function of temperature (N
xx
= 50 kN/m).
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
increased. A maximum deection of 12 mm occurs when N
xx
= 0, whereas
the maximum deection is reduced to 3.4 mm when N
xx
= 70 kN/m.
Part (b). As before, the maximumout-of-plane displacement occurs at x =
a/2 = 0.15 m and y = b/2 = 0.075. Because a constant in-plane tensile load
N
xx
= 50 kN/m is applied, the maximum out-of-plane displacement is given
by:
wj
max

324
p
4
781
82:5
p
2
DT
c
DT
_ _
meters
Aplot of maximumout-of-plane displacement as a function of temperature is
shown in Fig. 6(b). At roomtemperature (20jC), a maximumdeection of 4.3
mm is predicted. As would be expected, the plate becomes stier as the
temperature is decreased. Out-of-plane displacements are decreased because
of the in-plane tensile load N
yy
that develops as temperature is decreased. At
the lowest temperature considered (50jC), a maximumdeection of 2.4 mm
is predicted.
6 THE NAVIER SOLUTION APPLIED TO A SPECIALLY
ORTHOTROPIC LAMINATE SUBJECT TO SIMPLE
SUPPORTS OF TYPE S4
In Sec. 4, we developed the solution for a specially orthotropic laminate
subjected to homogenous simple supports of type S4 along all four edges,
uniformin-plane loads N
xx
and N
yy
, a uniformtemperature change DT, and a
sinusoidal transverse loading given by:
qx; y q
o
sin
px
a
_ _
sin
py
b
_ _
In-plane and transverse displacements caused by this thermomechanical
loading are given by Eqs. (12a), (12b), and (15).
Now recall from Sec. 4 of Chap. 9 that any transverse loading may be
represented in terms of the double-Fourier series given by Eq. (14) of Chap. 9,
repeated here for convenience:
qx; y

l
m1

l
n1
q
mn
sin
mpx
a
_ _
sin
npy
b
_ _
9:14
Thus any transverse loading can be viewed as the sum of a large number of
sinusoidal load components. The displacements caused by an arbitrary
transverse loading applied to a plate with simple supports of type S4 can
therefore be obtained using the same approach as that used in Sec. 4. The only
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
dierences are that the transverse load is given by Eq. 14 of Chap. 9 and the
transverse deections w(x,y) are assumed to be of the form:
wx; y

l
m1

l
n1
c
mn
sin
mpx
a
_ _
sin
npy
b
_ _
26
where c
mn
are unknown constants to be determined using the equations of
equilibrium. The assumed form and solution for in-plane displacements
u
o
(x,y) and v
o
(x,y) remain unchanged and are given by Eqs. (12a) and
(12b). Substituting Eqs. (26) and Eq. (14) of Chap. 9 into the third equation
of equilibrium, Eq. (1c), and equating coecients (following the same
procedures as used in Sec. 4), we obtain:
c
mn

q
mn
a
4
b
4
p
4
D
11
mb
4
2D
12
2D
66
mnab
2
D
22
na
4

ab
p
_ _
2
N
xx
mb
2
N
yy
na
2
_ _
_ _
27a
Using the denition of the plate aspect ratio, R=a/b, this result can also be
written as:
c
mn

q
mn
R
4
b
4
p
4
D
11
m
4
2D
12
2D
66
mnR
2
D
22
nR
4

a
2
p
2
N
xx
m
2
N
yy
nR
2
_ _
_ _
27b
Substituting this result into Eq. (26) completes the solution to the problem.
To summarize, the midplane displacements induced in a symmetric specially
orthotropic laminate subjected to an arbitrary transverse loading given by
Eq. (15) of Chap. 9, constant and uniform in-plane loads N
xx
and N
yy
, a
uniformtemperature change DT, and homogenous simple supports of type S4
along all four edges are given by:
u
o
x; y
A
22
N
xx
A
12
N
yy
A
11
A
22
A
2
12
_ _
a
xx
DT
_ _
x 28a
v
o
x; y
A
11
N
yy
A
12
N
xx
A
11
A
22
A
2
12
_ _
a
yy
DT
_ _
y 28b
wx; y
R
4
b
4
p
4

l
m1

l
n1
q
mn
sinmpx=asinnpy=b
D
11
m
4
2D
12
2D
66
mnR
2
D
22
nR
4

a
2
p
2
N
xx
m
2
N
yy
nR
2
_ _
_ _
28c
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
The method of using a double-Fourier series expansion to represent the
transverse loading and out-of-plane displacements was rst suggested by
Navier in about 1820, and is known as the Navier solution. An application of
this technique is illustrated in Sample Problem 4. The solution presented here
is for a plate subjected to type S4 simple supports, but the same approach may
be used to study simply supported plates with type S1 supports (as in Sec. 3),
or plates with mixed simple supports (as in Sec. 5).
As previously mentioned, a second technique known as the Levy
solution [15] can also be used to obtain exact solutions for symmetric
specially orthotropic laminates. While the Navier solution requires that all
four edges be simply supported, the Levy solution only requires that two
opposite edges of the plate are simply supported; the boundary conditions on
the remaining two edges are arbitrary and may be clamped or free, for
example. The Levy solution technique will not be discussed in this textbook,
and the interested reader is referred to the references cited for a discussion of
this approach.
Sample Problem 4
A [(0
2
/90)
2
]
s
graphiteepoxy laminate is cured at 175jC and then cooled to
room temperature (20jC). After cooling, the at laminate is trimmed to in-
plane dimensions of 300150 mm and mounted in an assembly that provides
type S4 simple supports along all four edges. The x axis is dened parallel to
the 300 mm edge (i.e., a = 0.3 m; b = 0.15 m). The laminate is then subjected
to a uniform in-plane tensile loading N
xx
= N
yy
= 50 kN/m and uniform
transverse loading q(x,y) = 100 kPa. The temperature remains constant and
no change in moisture content occurs (DM = 0). Plot the out-of-plane dis-
placements induced along the centerline dened by y = 0.075 m. Use the
properties listed for graphiteepoxy in Table 3 of Chap. 3, and assume each
ply has a thickness of 0.125 mm.
Solution. A [(0
2
/90)
2
]
s
graphiteepoxy laminate was also considered in
Sample Problem 1, and numerical values for the [ABD] matrix are listed
there. As before, the 12-ply laminate has a total thickness t = 1.5 mm and
aspect ratio R = a/b = 2.0.
The double-Fourier series expansion of a uniform transverse loading
was discussed in Sec. 4 of Chap. 9. The coecients in the Fourier series
expansion were found to be:
q
mn

16q
o
p
2
mn
; m; n odd integers
Combining these coecients with Eq. (28c) allows prediction of out-of-plane
displacements. A plot of these displacements along the centerline of the plate
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
dened by y =b/2 =0.075 mis shown in Fig. 7. Curves are shown based on a
1-term expansion (i.e., m,n = 1), a 4-term expansion (m,n = 1,3), and a 9-
termexpansion (m,n =1,3,5). As would be expected due to symmetry, out-of-
plane displacement is maximum at the center of the plate. The solution
rapidly converges. The maximum displacement predicted on the basis of a 1-,
4-, and 9-term expansion equals 5.27, 4.71, and 4.78 mm, respectively. If 100
terms were used (m,n = 1,3,. . .,19), the maximum predicted displacement is
4.77 mm.
7 BUCKLING OF RECTANGULAR SPECIALLY ORTHOTROPIC
LAMINATES SUBJECT TO SIMPLE SUPPORTS OF TYPE S4
This section is devoted to the phenomenon known as plate buckling. A
brief discussion of what the term buckling means in the context of a thin
plate is in order. Consider an initially at symmetric laminate subjected to
constant and uniformboundary edge loads N
x
xx
N
x
xx
=N
xx
and N
y
yy
=
N

yy
= N
yy
(shear resultants are assumed zero: N
xy
= N
yx
= 0). As we have
seen in preceding sections, coupling stinesses B
ij
= 0 for all symmetric lam-
inates. Hence, according to our earlier analyses, we would not expect these
edge loads to cause out-of-plane displacements because in-plane loads and
out-of-plane displacements are (apparently) uncoupled for a symmetric
laminate.
Figure 7 Out-of-plane displacements induced in the graphiteepoxy plate con-
sidered in Sample Problem 4, along centerline y = 0.075 m.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
This statement is always true if both N
xx
and N
yy
are tensile; a at
symmetric laminate subjected to tensile edge loads N
xx
and/or N
yy
will remain
at regardless of the magnitude of load (unless, of course, the loads are high
enough to cause fracture). However, suppose that either N
xx
or N
yy
(or both)
are compressive. If the magnitude of the compressive load(s) is (are) relatively
low, then the initially at laminate remains at; that is, only in-plane
displacements u
o
(x,y) and v
o
(x,y) are induced and out-of-plane displacement
remain zero: w(x,y)=0. However, as the compressive load(s) is (are) in-
creased, the laminate may exhibit a sudden out-of-plane displacement, w(x,y)
p 0. This is the phenomenon is known as buckling.
The load level at which out-of-plane displacements initially occur is
called the critical buckling load. In general, a thin plate does not collapse at the
critical buckling load, and can support a further increase in load. However,
out-of-plane displacements rapidly increase as in-plane loads are increased
beyond the initial buckling load, signaling imminent structural failure.
The out-of-plane displacements that occur at the onset of buckling
exhibit a characteristic pattern. This characteristic pattern is called the
buckling mode. A typical buckling mode is illustrated in Fig. 8. A three-
dimensional view of the buckling mode is shown in Fig. 8(a), while a 2-Dview
(which can be thought of as a topographical map of out-of-plane displace-
ments) is shown in Fig. 8(b). In the discussion to follow, predicted buckling
modes will be illustrated in a form similar to Fig. 8(b).
The buckling mode exhibited by a given laminate depends on the
stiness of the laminate, plate aspect ratio, applied loading, and the boundary
conditions. Aparticular buckling mode is described in terms of the number of
points of relative maximum out-of-plane displacements in the x and y
directions. For the buckling mode shown in Fig. 8, there are two points of
relative maximum displacement in the x direction, whereas there is only one
point of relative maximum displacement in the y direction. Hence the pattern
shown in Fig. 8 is called mode [2,1].
In this section, we consider buckling of an initially at rectangular
specially orthotropic laminate subjected to simple supports of type S4 and
uniformedge loads N
x
xx
N
x
xx
N
xx
and N
y
yy
N
y
yy
=N
yy
, at least one
of which is compressive. If the laminate was cured at an elevated temperature
and then cooled to room temperature, then pre-existing thermal stress re-
sultants associated with cooldown, N
T
xx
and N
T
yy
, are present. It is assumed
that no transverse loading is applied and that no change moisture content
occurs: q(x,y) = DM = 0.
Because the applied edge loads are constant and uniform, prior to
buckling the internal stress resultants at all points within the laminate are also
constant and uniform, and are equal to the external edge loads. Hence, prior
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
to buckling, the internal stress resultants at all points within the laminate are
uniform and given by:
N
*
xx
x; y A
11
@u
o
@x
A
12
@v
o
@y
N
T
xx
N
x
xx
N
x
xx
N
xx
29a
N
*
yy
x; y A
12
@u
o
@x
A
22
@v
o
@y
N
T
yy
N
y
yy
N
y
yy
N
yy
29b
N
*
xy
x; y A
66
@u
o
@y

@v
o
@x
_ _
0 29c
It is emphasized that Eqs. (29a), (29b), and (19c) are valid prior to buckling.
After buckling has occurred and signicant out-of-plane displacements have
Figure 8 A typical mode [2,1] buckling mode, caused in a rectangular plate by
a uniaxial compressive loading N
xx
.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
developed, the internal stress and moment resultants may no longer be
constant and uniform, and may not equal the applied edge loads.
The remaining type S4 simple support boundary conditions are:
Along x = 0:
wx; y 0 30a
M
*
xx
0; y D
11
@
2
w0; y
@x
2
D
12
@
2
w0; y
@y
2
0 30b
Along x = a:
wa; y 0 31a
M
*
xx
a; y D
11
@
2
wa; y
@x
2
D
12
@
2
wa; y
@y
2
0 31b
Along y = 0:
wx; 0 0 32a
M
*
yy
x; 0 D
12
@
2
wx; 0
@x
2
D
22
@
2
wx; 0
@y
2
0 32b
Along y = b:
wx; b 0 33a
M
*
yy
x; b D
12
@
2
wx; b
@x
2
D
22
@
2
wx; b
@y
2
0 33b
The equation of equilibrium governing out-of-plane displacements, Eq. (1c),
is [with q(x,y) = DM = 0]:
A
11
@u
o
@x
A
12
@v
o
@y
N
T
xx
_ _
@
2
w
@x
2
_ _
34
A
12
@u
o
@x
A
22
@v
o
@y
N
T
yy
_ _
@
2
w
@y
2
_ _
2 A
66
@u
o
@y

@v
o
@x
_ _ _ _
@
2
w
@x@y
_ _
D
11
@
4
w
@x
4
2D
12
2D
66

@
4
w
@x
2
@y
2
D
22
@
4
w
@y
4
0
Equations (29a), (29b), and (29c) are valid up to the onset of buckling; that is,
we assume that internal stress resultants are constant and uniform as the
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
buckling phenomenon is approached. Hence substituting Eqs. (29a), (29b),
and (29c) into Eq. (34), we have:
D
11
@
4
w
@x
4
2D
12
2D
66

@
4
w
@x
2
@y
2
D
22
@
4
w
@y
4
N
xx
@
2
w
@x
2
_ _
N
yy
@
2
w
@y
2
_ _
35
Notice that thermal stress and moment resultants do not appear in Eq. (35).
This reveals that the buckling response of symmetric specially orthotropic
laminates is independent of temperature, when subject to simple supports of
type S4. As will be discussed in Sec. 8, if the laminate were subject to
boundary conditions that constrain in-plane displacements, e.g., simple
supports of type S1, then buckling may be caused by a temperature change.
We now assume out-of-plane displacements are given by:
wx; y c sin
mpx
a
_ _
sin
npy
b
_ _
36
where mand n are positive integers and c is an unknown constant representing
the maximum out-of-plane deection. It is easy to show that Eq. (36) satises
the boundary conditions, Eqs. (30a), (30b), (31a), (31b), (32a), (32b), and
(33a), (33b), so the next step is to evaluate the conditions under which the
equation of equilibrium governing out-of-plane deections is satised. Sub-
stituting Eq. (36) into Eq. (35) and simplifying, we nd that in order for
equilibrium to be maintained, it is required that:
D
11
mp
a
_ _
4
2D
12
2D
66

mp
a
_ _
2
np
b
_ _
2
D
22
np
b
_ _
4
N
xx
mp
a
_ _
2
N
yy
np
n
_ _
2
37
The unknown constant c has canceled out and does not appear in the
requirement for equilibrium, Eq. (37). Referring to Eq. (36), note that integers
m and n dictate the spatial variation of the out-of-plane displacement eld,
i.e., m and n dene how w(x,y) varies with x and y, and consequently dene
the buckling mode. Constant c represents the magnitude of w(x,y). Hence
while we are able to determine m and n and thus predict the buckling mode
on the basis of Eq. (37), we cannot predict the magnitude of out-of-plane
displacements.
Although Eq. (37) is valid for any combination of N
xx
and N
yy
, it is
convenient to rearrange Eq. (37) for three dierent loading conditions. As
explained above, N
xx
and/or N
yy
must be compressive to cause buckling.
Assume for the moment that a constant transverse tension N
yy
z0 is applied,
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
in which N
xx
case must be compressive to cause buckling. Solving Eq. (37) for
N
xx
, we nd:
N
xx

p
2
m
2
a
2
b
2
_
D
11
mb
4
2D
12
2D
66
mnab
2
38
D
22
na
4
N
yy
na
2
b
p
_ _
2
_
Using the denition of the plate aspect ratio, R = a/b, this result can also be
written as:
N
xx

p
2
ma
2
_
D
11
m
4
2D
12
2D
66
mnR
2
39
D
22
nR
4
N
yy
naR
p
_ _
2
_
Hence, given some value for N
yy
z0 and assumed integer values for m and n,
either Eq. (38) or (39) can be used to calculate a corresponding value for N
xx
.
Because m and n can be any combination of positive integers, there are an
innite number of values for N
xx
that satisfy Eq. (38) or (39). The critical
buckling load, denoted N
xx
c
, corresponds to the particular combination of m
and n that leads to the value of N
xx
with lowest magnitude. The combination
of m and n that correspond to this lowest load dene the predicted critical
buckling mode.
The following observations are based on inspection of Eq. (38) or (39).
Because we have assumed for the moment that N
yy
z 0, all variables that
appear on the right side of the equality sign are algebraically positive. Hence
N
xx
c
must be algebraically negative, i.e., the critical buckling load N
xx
c
is
predicted to be compressive, as would be expected. Secondly, the minimum
magnitude of N
xx
c
will always correspond to n = 1 because N
yy
z 0. Finally,
note that if the constant transverse load N
yy
is increased, then the magnitude
of N
xx
c
is increased. That is, a transverse tension will cause an increase in the
critical buckling load.
As a second loading condition of interest, let us now assume that a
constant N
xx
z 0 is applied, and that a compressive load in the y direction
causes buckling. Solving Eq. (37) for N
yy
, we nd:
N
yy

p
2
n
2
a
4
b
2
_
D
11
mb
4
2D
12
2D
66
mnab
2
40
D
22
na
4
N
xx
mab
p
_ _
2
_
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Using the plate aspect ratio, R = a/b, this result can be written as:
N
yy

p
2
nbR
2

2
_
D
11
m
4
2D
12
2D
66
mnR
2
D
22
nR
4
N
xx
ma
p
_ _
2
_
41
As before, m and n can be any combination of positive integers, and therefore
there are an innite number of values for N
yy
that satisfy Eq. (40) or (41). The
critical buckling load N
yy
c
corresponds to the particular combination of m
and n that leads to the value of N
yy
with lowest magnitude, and the
combination of m and n that correspond to this load dene the predicted
buckling mode. Because it has been assumed that N
xx
z0, m = 1 in all cases,
and if the constant tensile load N
xx
is increased, the magnitude of the critical
buckling load N
yy
c
will increase.
Finally, consider a third loading condition in which buckling is caused
by the simultaneous increase in both N
xx
and N
yy
. Further, assume the two
loads are linearly related. That is, assume N
yy
= kN
xx
, where k is a known
constant. Substituting this relation into Eq. (37) and solving for N
xx
, we nd:
N
xx

p
2
ab
2
D
11
mb
4
2D
12
2D
66
mnab
2
D
22
na
4
_ _
mb
2
kna
2
_ _
42
As before, m and n can be any combination of positive integers, and therefore
there are an innite number of values for N
xx
that satisfy Eq. (42). The critical
buckling condition, dened by the two simultaneous loads, N
xx
c
and N
yy
c
=
kN
xx
c
, corresponds to the particular combination of m and n that leads to the
value of N
xx
(and N
yy
) with lowest magnitude. The combination of m and n
that correspond to this load condition denes the predicted buckling mode.
Note that if k < 0, then N
xx
and N
yy
are of opposite algebraic signs. This
implies that there are two distinct buckling load conditions, one in which
(N
xx
< 0, N
yy
>0), and a second in which (N
xx
>0, N
yy
<0).
Numerical examples illustrating buckling predictions for symmetric
specially orthotropic simply supported laminates will be discussed in Sample
Problems 5 and 6. Two important concluding comments are made in passing.
First, from the preceding discussion, the reader may have inferred that
buckling is only caused by N
xx
, by N
yy
, or by some combination thereof.
This is not the case. Other forms of loading may cause buckling, e.g., a shear
load N
xy
. In this chapter, only buckling caused by N
xx
and/or N
yy
is con-
sidered. The reader interested in buckling caused by other types of loading is
referred to Refs. 4 and 5.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Second, recall that the analysis presented above does not allow pre-
diction of the magnitude of out-of-plane displacements induced at buckling.
Buckling of thin plates is rarely catastrophic, and in general nite out-of-
plane displacements occur at and above the critical buckling load. That is, a
thin plate does not collapse once the buckling load is reached and can support
a further increase in load, albeit at a reduced level of stiness. Methods to
predict the magnitude of out-of-plane displacement for load levels at or above
the initial buckling load are known as post-buckling analyses, and are not
discussed in this text.
Sample Problem 5
A [(0
2
/90)
2
]
s
graphiteepoxy laminate is cured at 175jC and then cooled to
room temperature (20jC). After cooling, the at laminate is trimmed to in-
plane dimensions of 300 150 mmand mounted in an assembly that provides
type S4 simple supports along all four edges. The x axis is dened parallel to
the 300 mm edge (i.e., a = 0.3 m; b = 0.15 m).
(a) Predict the critical buckling load N
xx
c
and mode for this laminate, if
0 V N
yy
V 400 kN/m.
(b) Predict the critical buckling load N
yy
c
and mode for this laminate, if
0 V N
xx
V 400 kN/m.
Use the properties listed for graphiteepoxy in Table 3 of Chap. 3, and assume
each ply has a thickness of 0.125 mm.
Solution. A [(0
2
/90)
2
]
s
graphiteepoxy laminate was also considered in
Sample Problem 1, and numerical values for the [ABD] matrix are listed
there. As before, the 12-ply laminate has a total thickness t = 1.5 mm and
aspect ratio R = a/b = 2.0.
Part (a). It is noted that n = 1, because N
yy
z 0. Equation (39) becomes in
this case:
N
xx

p
2
0:09m
2
40:1m
4
65:34m
2
172:8 N
yy
0:6
p
_ _
2
_ _
A plot of the critical buckling load for 0 V N
yy
V 400 kN/m is presented in
Fig. 9(a). As expected, N
xx
c
is increased as N
yy
is increased. A change in
buckling mode also occurs as N
yy
is increased. The plate buckles in mode
[2,1] over the range 0 V N
yy
< 35 kN/m, in mode [3,1] over the range 35 kN/
m V N
yy
< 150 kN/m, and in mode [4,1] over the range 150 kN/m V N
yy
< 400 kN/m. These buckling modes are illustrated in Fig. 9(b,c,d),
respectively.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
Part (b). It is noted that m=1, because N
xx
z 0. Equation (41) becomes in
this case:
N
yy

p
2
0:36n
2
40:1 65:34n
2
172:8n
4
N
xx
0:09
p
2
_ _
A plot of the critical buckling load for 0 V N
xx
V 400 kN/m is presented in
Fig. 10(a). As expected, N
yy
c
is increased as N
xx
is increased. A change in
Figure 9 Buckling response of the [(0
2
/90)
2
]
s
graphiteepoxy plate considered
in Sample Problem 5.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
buckling mode also occurs as N
xx
is increased. The plate buckles in mode [1,1]
over the range 0 VN
xx
<72 kN/m, and in mode [1,2] over the range 72 kN/m
V N
xx
< 400 kN/m. These buckling modes are illustrated in Fig. 10(b,c),
respectively.
Note that the magnitudes of N
yy
c
calculated in part (b) are far lower than
those calculated for N
xx
c
in part (a). This pronounced dierence is largely
because of the stacking sequence involved. For the [(0
2
/90)
2
]
s
laminate under
Figure 10 Buckling response of the [(0
2
/90)
2
]
s
graphiteepoxy plate consid-
ered in Sample Problem 5.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
consideration, 8 of 12 plies are 0j plies (i.e., plies with bers parallel to the x
axis). Hence the resistance to buckling because of a compressive load N
xx
is
far higher than resistance to buckling because of a compressive N
yy
.
Sample Problem 6
A structure is being designed that will involve a [(0
2
/90)
2
]
s
graphiteepoxy
laminate with a width (in the y direction) of 150 mm. The length of the panel
(in the x direction) has not yet been established, and could be anywhere from
150 to 750 mm. During service, the panel will be subjected to a compressive
load N
xx
(only), and simple supports of type S4 along all four edges. Buckling
is therefore of concern. Predict the buckling load and mode for the panel, for
any panel length ranging from 150 to 750 mm.
Solution. A [(0
2
/90)
2
]
s
graphiteepoxy laminate was also considered in
Sample Problem 1, and numerical values for the [ABD] matrix are listed
there. As before, the 12-ply laminate has a total thickness t = 1.5 mm.
Buckling loads and modes will be predicted using Eq. (39). According to the
problem statement, b = 0.15 m, and 0.15 m<a <0.75 m. The plate aspect
Figure 11 Predicted buckling loads and modes as a function of aspect ratio for
the [(0
2
/90)
2
]
s
graphiteepoxy panel considered in Sample Problem 6.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
ratio therefore varies over 1 VRV5. Because transverse loading is zero
(N
yy
= 0), n = 1 and Eq. (39) becomes:
N
xx

p
2
ma
2
40:1m
4
16:34mR
2
10:8R
4
_ _
Aplot of the predicted critical buckling load over the specied range in aspect
ratio is presented in Fig. 11. The buckling mode is predicted to increase as the
aspect ratio increases: mode [1,1] is predicted over the range 1 <R<1.96,
mode [2,1] is predicted over the range 1.96 <R<3.40, mode [3,1] is predicted
for 3.40 <R<4.81, and mode [4,1] is predicted for 4.81 <R<5.00. Still
higher buckling modes would occur at higher aspect ratios.
The predicted buckling load generally decreases with aspect ratio,
although a local maximum in the buckling load occurs at each aspect ratio
corresponding to a change in mode shape. At an aspect ratio R = 1 (i.e., for a
square plate), buckling is predicted to occur at N
xx
c
= 29.5 kN/m.
8 THERMAL BUCKLING OF RECTANGULAR SPECIALLY
ORTHOTROPIC LAMINATES SUBJECT TO SIMPLE
SUPPORTS OF TYPE S1
Buckling caused by direct application of uniform external edge loads N
xx
and/or N
yy
was considered in the Sec. 7. The analysis was conducted for a
specially orthotropic laminate subjected to inhomogeneous simple supports
of type S4. The initially at symmetric laminate was subjected to uniform
boundary edge loads N
x
xx
N
x
xx
and N
y
yy
= N
y
yy
. Because we specied
in-plane edge loads, we did not specify in-plane edge displacements. In es-
sence, edge displacements were allowed to vary with changes in the specied
edge loads.
We now wish to consider buckling caused by a dierent mechanism.
Namely, we wish to consider buckling caused by a change in temperature. In
this case, we will specify inhomogeneous simple supports of type S1. That is,
we will specify in-plane edge displacements, but will not specify in-plane edge
loads. These boundary conditions were also considered in Sec. 3. As was
discussed there, we assume that pre-existing midplane displacements (as well
as ply strains) are induced prior to assembly of the laminate in the simply
supported conguration. That is, midplane displacements are induced during
cooling from an elevated cure temperature to room temperature. For a
symmetric specially orthotropic laminate, the midplane displacements caused
by cooling are given by Eq. (5a), (5b):
u
c
o
x; y a
xx
DT
c
x
v
c
o
x; y a
yy
DT
c
y
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
where a
xx
and a
yy
are the eective thermal expansion coecients of the
laminate and DT
c
is the change in temperature from cure to room temper-
ature. Therefore boundary conditions along all four edges are:
For x 0 : For x a :
w0; y 0 wa; y 0
M
*
xx
0; y 0 M
*
xx
a; y 0
u
o
0; y u
x
o
y0 u
o
a; y u
x
o
y a
xx
DT
c
a
v
o
0; y v
x
o
y a
yy
DT
c
y v
o
0; y v
x
o
y a
yy
DT
c
y
For y 0 : For y b :
wx; 0 0 wx; b 0
M
*
yy
x; 0 0 M
*
yy
x; b 0
u
o
x; 0 u
y
o
x a
xx
DT
c
x u
o
x; b u
y
o
x a
xx
DT
c
a
v
o
x; 0 v
y
o
x 0 v
o
x; b v
y
o
y a
yy
DT
c
b
After assembly, we assume the laminate is subjected to a uniform change in
temperature. The change in temperature is referenced to the strain-free tem-
perature (assumed to be the cure temperature) and is represented by DT =
(current temperature) (cure temperature). If the laminate was not simply
supported and instead was free to expand or contract, then a uniform change
in temperature would simply cause a change in midplane displacements and
no external edge loads would result. Because the laminate is instead subject to
simple supports of type S1, no changes in midplane displacements are allowed
to occur and external edge loads develop as temperature changes. Depending
on the magnitude of the temperature change, these thermally induced edge
loads may cause the laminate to buckle.
Although the edge loads are thermally induced, they are external
mechanical loads nevertheless and consequently the analysis presented in
Sec. 7 is still applicable. The onset of thermal buckling occurs if Eq. (37) is
satised, repeated here for convenience:
D
11
mp
a
_ _
4
2D
12
2D
66

mp
a
_ _
2
np
b
_ _
2
D
22
np
b
_ _
4
N
xx
mp
a
_ _
2
N
yy
np
b
_ _
2
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
In the present case, loads N
xx
and N
yy
are caused by a deviation from room
temperature, and are given by:
N
xx
DT
c
DTA
11
a
xx
A
12
a
yy
43a
N
yy
DT
c
DTA
12
a
xx
A
22
a
yy
43b
Substituting Eqs. (43a) and (43b) into Eq. (37), solving for the temperature
dierence, and using the denition of the plate aspect ratio, R = a/b, we
obtain:
DT
c
DT
p
2
b
2
R
4
D
11
m
4
2D
12
2D
66
mnR
2
D
22
nR
4
n
2
A
12
a
xx
A
22
a
yy

m
R
_ _
2
A
11
a
xx
A
12
a
yy

_
_

_ 44a
It is convenient to express the temperature dierences involved as:
DT
c
DT T
RT
T
where T
RT
=room temperature (i.e., the temperature at which the laminate
was mounted in the simple support xture); T = current temperature.With
this change in notation we have:
T
p
2
b
2
R
4
D
11
m
4
2D
12
2D
66
mnR
2
D
22
nR
4
n
2
A
12
a
xx
A
22
a
yy

m
R
_ _
2
A
11
a
xx
A
12
a
yy

_
_

_T
RT
44b
Equations (44a) and (44b) are entirely equivalent, and both are based on the
assumption that the simply supported laminate is assembled at room temper-
ature, T
RT
. In practice, it is conceptually simplest to use Eq. (44b). By
specifying a particular laminate and room temperature, all variables on the
right side of the equality sign in Eq. (44b) are known, except for variables m
and n. Because m and n can be any combination of positive integers, there are
an innite number of values of temperature T that will satisfy Eq. (44b). The
temperature at which thermal buckling will occur, denoted T
bk
, corresponds
to the particular combination of m and n that leads to the value of T of lowest
magnitude. The combination of m and n that correspond to this lowest
temperature dene the predicted critical thermal buckling mode.
Now, based on the results of Sample Problem 6, one might anticipate
that the thermal buckling mode exhibited would vary as a function of aspect
ratio. That is, based on earlier analyses, one would expect that the values of m
and n that correspond to the critical thermal buckling load would vary with
aspect ratio. Equations (44a) and (44b) allow for such dependence. However,
if physically reasonable material properties are used during the evaluation of
Eqs. (44a) and (44b), then numerical experiments show that m = n = 1 in all
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
cases. Thus a thermal buckling mode [1,1] is always predicted when realistic
material properties are used (at least for the simply supported boundary
conditions considered in this section), regardless of the aspect ratio or
stacking sequence considered.
Assuming that this observation always holds true and that m = n = 1,
Eq. (44b) reduces to:
T
bk

p
2
b
2
R
4
D
11
2D
12
2D
66
R
2
D
22
R
4
A
12
a
xx
A
22
a
yy

1
R
2
A
11
a
xx
A
22
a
yy

_
_

_ T
RT
This latter result shows that the thermal buckling temperature decreases with
an increase in aspect ratio. In the limit (i.e., as R!l), the thermal buckling
temperature becomes:
T
bk

p
2
D
22
b
2
A
12
a
xx
A
22
a
yy

T
RT
Sample Problem 7
Two [0
2
/90)
2
]
s
graphiteepoxy laminates are cured at 175jC and then cooled
to room temperature (20jC). After cooling, one laminate is trimmed to in-
plane dimensions of 300150 mm, whereas the second is trimmed to in-plane
dimensions of 3000150 mm. Both laminates are then mounted in assemblies
that provide type S1 simple supports along all four edges, and subjected to a
uniform increase in temperature. Determine the temperature at which each
plate will buckle. Use the properties listed for graphiteepoxy in Table 3 of
Chap. 3, and assume each ply has a thickness of 0.125 mm.
Solution. A [(0
2
/90)
2
]
s
graphiteepoxy laminate was also considered in
Sample Problem 1, and numerical values for the [ABD] matrix and eective
thermal expansion coecients are listed there. The aspect ratios involved in
this problem are:
R 300 mm=150 mm 2
and
R 3000 mm=150 mm 20:
The temperature at which thermal buckling is predicted to occur is obtained
through application of Eq. (44b). It is found that both laminates are predicted
to buckle in mode [1,1]. For the laminate with aspect R = 2, thermal buckling
is predicted to occur when temperature is raised to 51jC, whereas for the
laminate with aspect ratio R = 20, thermal buckling is predicted at a
temperature of 40jC.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.
9 COMPUTER PROGRAM SPORTHO
The computer program SPORTHO has been developed to supplement the
material presented in this chapter. This program can also be downloaded
at no cost from the following website: http://depts.washington.edu/amtas/
computer.html.
Program SPORTHO is applicable to symmetric specially orthotropic
laminates. The program can be used to calculate the transverse deections
according to the analyses presented in Secs. 3 through 6, to calculate buckling
loads according to the analysis presented in Sec. 7, or to calculate the
temperature at which thermal buckling will occur, according to the analysis
presented in Sec. 8. The user is prompted to input all information necessary to
perform these calculations. Properties of up to ve dierent materials may be
dened. Numerical values must be dened using a consistent set of units in all
cases. For example, the user must input elastic moduli for the composite
material system(s) of interest. Using the properties listed in Table 3 of Chap. 3
and based on the SI system of units, the following numerical values would be
input for graphiteepoxy:
E
11
170 10
9
Pa E
22
10 10
9
Pa v
12
0:30 G
12
13 10
9
Pa
Because 1 Pa = 1 N/m
2
, all other lengths must be input in meters. For ex-
ample, ply thickness must be input in meters (not millimeters). Atypical value
would be t
k
=0.000125 m(corresponding to a ply thickness of 0.125 mm). In-
plane plate dimensions must also be input in meters.
If the English system of units were used, then the following numerical
values would be input for the same graphiteepoxy material system:
E
11
25:010
6
psi E
22
1:510
6
psi v
12
0:30 G
12
1:910
6
psi
In this case, all lengths would be input in inches.
REFERENCES
1. Lekhnitskii, S.G. Anisotropic Plates, translated by S. W. Tsai and T. Cheron,
Taylor and Francis Books Ltd, London, UK, ISBN 0-677-20670-4, 1968.
2. Savin, G.N. Stress Concentration Around Holes; New York: Pergamon Press,
1961.
3. Timoshenko, S.; Woinowsky-Krieger, S. Theory of Plates and Shells; New York,
NY: McGraw-Hill Book Co (ISBN 0-07-0647798), 1987.
4. Whitney, J.M. Structural Analysis of Laminated Anisotropic Plates; Lancaster,
PA: Technomic Pub Co (ISBN 87762-518-2).
5. Buckling and Postbuckling of Composite Plates. Turvey, G.J.; Marshall, I.H.,
Eds; Chapman and Hall: New York, NY, 1995.
Copyright 2004 by Marcel Dekker, Inc. All Rights Reserved.