https://ntrs.nasa.gov/search.jsp?

R=19990052629 2018-02-03T10:46:54+00:00Z

Elastic Buckling of Orthotropic Plates under

Varying Axial Stresses
A. BADIR, H. HU AND A. OIALLO

ABSTRACT

The elastic buckling load of simply supported rectangular Orthotropic plates subjected to a
second degree parabolic variation of axial stresses in the longitudinal direction is calculated
using analytical methods. The variation of axial stresses is equilibrated by nonuniform shear
stresses along the plate edges and transverse normal stresses. The influence of the aspect ratio
is examined, and the results are compared with plates subjected to uniform axial stresses.

INTRODUCTION

In most studies of stability of plates, the axial stress has been taken as uniform compression
throughout flat rectangular plates (Timoshenko and Gere, 1961; Allen and Bulson, 1980;
Whitney, 1987). Buckling of isotropic plates under a compressive stress that varies linearly
from one loaded edge to the other has been studied by Libove, Ferdman and Reusch (1949).
Cases of practical interest exist, however, in which the axial stress is not uniform but varies
from tension at both loaded edges to compression in the middle. An example is the stability of
the crown of the hat stiffened panel, a candidate configuration of the upper and lower skin of
the Blended Wing Body (BWB) Aircraft The BWB Aircraft is an advanced long-range ultra-
high-capacity airliner with the principal feature being the wide double-deck body which is
blended into the wing (Popular Science Magazine, 1995).
In the present paper, analytical methods are used to investigate the local stability of the hat
crown plate in order to minimize its weight while optimizing its buckling strength. A varying
tension-compression-tension stress is induced in the crown of the stiffeners. The axial stresses
vary longitudinally due to bending, and the change in stresses is equilibrated by nonuniform
shear stresses along the plate edges and transverse normal stresses as shown in Fig. 1.

ANALYTICAL MODEL

The distribution of the axial stresses is given by

Ashraf Badir. Assistant Professor, Department of Engineering, Clark Atlanta University, Atlanta, GA
30314
Hurang Hu, Research Associate, Department of Engineering, Clark Atlanta University, Atlanta, GA 30314
Abdouramane Diallo, Graduate Student, Department of Mathematics, Clark Atlanta University, Atlanta.
GA 30314
23
 6 6
- uh uJ
tf*} *(*} (1)

and the expressions for the shear and transverse normal stresses T^ (*,>>) and oy(y), are
obtained by integrating the plane stress equilibrium equations

=0 (2)
dx dy

=0 (3)
By dx

and are found to be

A -)
a a
(4)

(5)

•ox(x)

ox(x)

Figure 1. Plate subjected to tensUe/cornpression/tensile longitudinal stresses

equilibrated by shear stresses T C x , ? ) and transverse stresses Oy(y
 ndies unaer varying Axial Stresses 25

The differential equation of the deflection surface of the orthotropic plate can be written as
follows:

<?4w <92 Vf ^2w w (6)

D,,—£ + 2(D, 2 +2D 66 ) T T+£>7?
2 T-^^T- ^^-^ ^2 2
&V ^ <?y

where DtJ, w, Nx, N^, and Ny are the bending stiffnesses of the orthotropic plate (Jones,
1975), vertical displacement, axial, shear and transverse stress resultants, respectively A
solution to Eq. (6) that satisfies the boundary conditions cannot be obtained in closed form
Thus, approximate solutions are sought. Two different displacement fields are considered- (1)
Double Fourier series, using Rayleigh-Ritz method, and (2) a polynomial in the longitudinal
direction, using nalerkin method. A summary of the analytical models is described in the
next sections, the influence of the aspect ratio is examined, and the results are compared with
plates subjected to uniform axial loads.

DOUBLE FOURIER SERIES (RAYLEIGH-RITZ METHOD)

The total potential energy functional of the orthotropic plate is given by

\2

(7)

where t, a, and b represent the thickness, length and width of tlie plate, respectively.
Of all conceivable buckling patterns satisfying boundary conditions, the actual buckling
pattern is that for which the potential energy U, as given by Eq. (7), is a minimum. The
deflection surface of the simply supported skin in bending may be represented by a Double
Fourier series as

M N . mitx . nny
.sin
1 sin—— (8)
m=ln=l a b

which satisfies the simply supported boundary conditions. Rayleigh-Ritz method requires that
the coefficients amn be chosen such that U is a minimum. That is, they must satisfy the
equations

= 1,2...AO
(9)
26 ANALYSIS AND DESIGN

The deflection expression (8) can be splitted into two independent subsets, one
corresponding to even values of n and the other, to odd values of n. The first subset
corresponds to buckling that is antisymmetric about the line v = b/2 and the second, to
buckling that is symmetric about that line. The second subset gives lower buckling loads, and
attention is hereinafter confined to it. Similar argument can be made to the buckling pattern in
the longitudinal direction (subscript m). Therefore, the buckling pattern is chosen as
M N mnx . nicy
I I i sin—-
m=l,3,5 . n=l,3,5.. a b
(10)

Substituting Eqs. (1), (4-5), and (10) into the energy expression (7), while using Eq. (9),
we obtain the following relations
M N M N

y=1.3,5... (ID
« = 1,3,5...

where

—1, m = i

(m M |m ± /I = even

Cy = n>- 12 i | i
7
-6, m = i,n = ;
%mni/
2 2-TT, m * i,n * 7,|m ±i| = ev«n,jn± jj = even
and
K-- F-' )
(12)

P and A/o represent the aspect ratio and buckling load, respectively. A nondimensional
buckling coefficient k may be expressed as
 Elastic Buckling of Orthotropic Plates under Varying Axial Stresses 27

k= (13)
2
A l*

Relations (11) represent a set of (M +1)(N +1) / 4 linear equations, the smallest eigenvalue
of Eqs. (11) gives the buckling load.

POLYNOMIAL (GALERKIN METHOD)

We seek a nontrivial solution to the deflection w which satisfies equation (6) and the
boundary conditions. For simply supported plates, the solution of Eq. (6) may be represented

M N
+*y £I*
(14)

in which £ = jt/flJ TJ = ylb ^d the coefficients A^ - ^4 are determined by satisfying the

four simply supported boundary conditions at the two loaded sides.

The expression for the Galerkin equations is given by

'#w

=0

(15)

Theses relationships represent a set of MxN homogeneous algebraic equations. In order

to avoid a trivial solution (qmn =0), the nondimensional buckling load k is chosen such that
the determinant of the coefficient matrix vanishes. The minimum value of k that satisfies the
requirement is the buckling load.

TABLE I. MATERIAL PROPERTIES

E1( EM GU \)n
9.25 Msi 4.67 Msi 2.27 Msi 0.397
28 ANALYSIS AND DESIGN

TABLE D. CONVERGENCE OF BUCKLING COEFFICIENT k FOR A SQUARE PLATE

M/N 1 3 5 7
1 1.75 1.69 1.65 1.65
3 1.72 1.65 1.65 1.65
5 1.72 1.65 1.65
7 1.72 1.65
9 1.72

RESULTS

First, a convergence study for Rayleigh-Ritz method is performed. Orthotropic plates with
material properties as listed in Table I have been studied. For a square plate, buckling
coefficients k versus increasing values of M and N are listed in Table n. It can be seen that
when M=N=3, k reaches its final value. Convergence of the solution in this case is faster than in
the case of long strips. For example a four-term series (M = N = 3, note that only odd M and
N are considered) converges to the final k value compared to a 12-term solution
(M = 23, N = 1) for a long strip with ft = 20.
The buckling coefficient k versus aspect ratio ft is plotted in Fig. 2. Figure 2 shows that
for plates with aspect ratio less than six, the buckling coefficient k increases as ft increases.
However, for long strips (0>6), k does not significantly change and takes a value of
approximately 5.75.

5
4
3
2
1

1 2 3 4 5 6 7 8 91 0
a/b

Figure 2. Buckling coefficient versus aspect ratio

 Elastic Buckling of Orthotropic Plates under Varying Axial Stresses 29

TABLE HI. CONVERGENCE OF BUCKLING COEFFICIENT FOR A

SQUARE PLATE USING GALERKIN METHOD.

M k
1 2.54
2 1.96
3 1.79
4 1.68
5 1.66

Galerkin method is also used to calculated the buckling load. The accuracy of the
polynomial representation of the displacement field in the longitudinal direction, given by Eq
(14), is examined for a square plate. The values of the buckling coefficient k are listed in Table
nCcbnverging to a value of 1.66 (M=5, N=l) compared to 1.65 as obtained using Fourier
Series (Raylcigh-Ritz method, Table II). For M=5 and JV=1, the displacement w, Eq. (14)
takes the following form

- 22? (16)

The buckling load intensity (load per unit width) calculated based on a parabolic and
uniform load distribution is listed in Table IV. The parabolic load values represent the buckling
load intensity at mid-span of the plate (i.e., ftr. or kD[ \n2/2b2).

Table IV shows that the buckling load intensity is influenced by both the load distribution
and the aspect ratio. For a square plate, the buckling load calculated based on a uniform stress
distribution is 244% higher than die buckling load calculated based on a
tensik/comptessive/kensile parabolic load distribution. The difference is only 5% (reduction)

TABLE IV. COMPARISON OF BUCKLING COEFFICIENTS FOR VARYING

AXIAL STRESS AND UNIFORM STRESS

Parabolic Load* Uniform % cliff.

Ob/in) Load*
(Ih/in)
Fourier Polyno- Exact
Series mial
a=b=150in 1.35' 1.36T 4.64 +244%
(a/b=l)
a=150 in. 5773* • •• 5482 -5%
b=4.3in.
(a/b=35)
* Associated with shear and transverse stresses as given in Eqs. (4) and (5).
* No shear and transverse stresses.
* Buckling Load intensity at mid-span (x=a/2). -"
 30 ANALYSIS AND DESIGN

for long strips. One may expect that a tensile stress at the edges as shown in Fig. 1 will tend to
restore the plate to its stable position and therefore increases its buckling load. However, the
parabolic stress distribution" is associated with both compressive and shear stresses The
transverses compressive stresses Gy(y) have a detrimental effect on the buckling load for
square plates, while their effect is negligible for long strips.

CONCLUSIONS
The elastic buckling load of orthotropic plates subjected to a parabolically varying tension-
compression-tension longitudinal stress, equilibrated by nonuniform shear stresses and
transverse compressive stresses is analyzed. It is shown that the buckling load depends on the
aspect ratio for plates with aspect ratio less than six. For square plates, the buckling load due to
a uniform stress distribution is 244% higher than the buckling load calculated based on a
~tensile/compressive/tensile parabolic load distribution. However, the difference is only 5%
(reduction) for long strips.

ACKNOWLEDGMENT

This work is supported by the NASA Langley Research Center under grant NAG-1-1714
This support is gratefully acknowledged.

REFERENCES
Allen, H. G., and P S Bulson, 1980. "Background to Buckling," McGraw-Hill Book Company

Jones, R. M. 1975. "Mechanics of Composite Materials," McGraw-Hill, New York

Libove, C., S. Ferdman, and J. Reusch. 1949. "NACA Tech. Note No. 1891."

Popular Science Magazine, April, 1995

Timoshenko, S P and J. M. Gere, 1961. "Theory of Elastic Stability," 2nd ed McGraw-Hill Book
Company.

Whitney, J M. 1987. "Structural Analysis of Laminated Anisotropic Plates," Technomic Publishing

Company.