ON TRANSVERSE BENDING OF PLATES,

INCLUDING THE EFFECT OF

TRANSVERSE SHEAR DEFORMATIONS
E , REISSNER
Department of Applied Mechanics and Engineering Sciences, University of California,
San Diego, La Jolla, CA92037,U.S.A.
(Received 8 July 19741

In this paper we return once more to the problem of transverse bending of homogeneous
transversely isotropic plates, including the effect of transverse shear deformation. Our object is
to describe an approach to the problem which reproduces our earlier equations of
two-dimensional plate theory[2] exactly, while at the same time leading to new supplement~y
information concerning certain practically important fhree-dimensional aspects of the problem.
The significance of this supplementary information will be shown for the example of the problem
of pure bending of an infinite plate with a circular hole.
Our starting point for the derivation of two-dimensional plate theory is once more the
conventional system of equilibrium equations of three-dimensiona theory
uXx.=
+ ryx,r+ T~.~= 0, etc.

(1)

together with stress strain relations which are here taken in the form

(2)
and
rA.2 +

w,, =

zz
v,, + w., = 22
G
G

WJ

a2

=--v,

Ez

a; + u,

Ez

(3)

and together with face boundary conditions

z =+;h;

uz =O,

rrx=O,

7zr=O,

(4)

where for simplicitys sake, and with no significant effect on the conclusions to be established, we
omit consideration of loads distributed over the faces of the plate.
In our earlier work121 we obtained a system of equations of two-Dimensions theory upon
introducing approximate stress distributions
a;

Mx z

=hZ/6h/23

7x2

=&[l-($y],

etc.

(5)

+Prepared with the. support of the Office of Naval Research and the National Science Foundation. The author wishes to
thank Gary E. Lee for assistance in obtaining the results presented in this paper.
569

S&S Vol. 11 No. 5-D

E. REISSNER

570

into a suitable minimum complementary energy condition, and upon evaluating this condition in
conjunction with an application of the Lagrange multiplier concept.
In what follows we retain our assumption that the dist~bution of transverse shears may be
taken to be parabolic, approximately. However, we abandon our assumption that ax, 0; and T,,
vary linearly across the thickness (which is an entirely adequate approximation for the derivation
of our approximate two-dimensional theory through use of the minimum complementary energy
condition, without however being equally adequate for the determination of the effect of
transverse shearing strains on the dependence of ax, a, and 7XYon the thickness coordinate z).
Instead we proceed as follows. We take account of the fact that, by equ~lib~um considerations,
the transverse normal stress a, is small compared to the transverse shearing stresses G, and Tag,
and these in turn are small compared to uX,uY and rXY,by omitting 0; altogether in the stress
strain relations and by approximating T,, and 7Yiin these relations as follows
UJ +

w.x -3v,
2/&j

(,_$, )

v.z+ w*x2&

(l-4$ )

(6)
(7)

and
ux
-_=E

u.x+ m,y cr,_ v,, + V&.x ?XY= u .I 4-v .x

j-v

z----- l--v2 E 2(1+v)

We now integrate (7) in the form

w = W(x,y)-

t %+jz
v*

(8)

(9)

and introduce this relation into the approximate transverse shear stress strain equations (6). From
these we obtain

(10)
with a corresponding equation for u,*. We next observe that in these equations the stress sum
0; + uYcontributes an effect of the same order of magnitude as the stresses T*=and Tag.Therefore
it is legitimate to here approximate a; + a, in the form

(11)
It turns out that, numerically, the effect of the (0; + a,)-term in (10) will be negligibly small
compared to the effect of the rXz-term, for normal ranges of values of v,GIE,.
We will limit
ourselves here to a consideration of the problem subject to this restriction and replace equation
(I0)by the simplified relation
U=-W,,+~

again with a corresponding relation for v,,.

3vx
(

l-45

(12)

571

On transverse bending of plates

Having (12) we conclude immediately that to the same degree of approximation

together with

Introduction of (13) into equations (8), and observation of the equilibrium equation
V,., + V,,, = 0, gives as new supplementary information concerning the three-dimensional
aspects of the problem of plate bending expressions for bending and twisting stresses of the form
(14a)
(14b)
(14c)
Having equations (14) we obtain expressions for bending and twisting couples in the usual
way, with D = Eh3/12(1 - v*), in the form
61-v
W.xx f vW,yy -s+&

ikf, =-D

My = -D W.,, + VW,, -is

[

KY = -(l-

v)D[ W,,, -;&-(Vm

(15d

7
KY]

+ Vd

(15b)

(15c)

and we note that these equations are in exact agreement with the corresponding relations (II) to
(IV) in[2], upon introducing the stipulation that p = 0, and upon setting E = 2(1+ v)G in
equations (15).t
For what follows it is of importance to compare the contents of (15) with the expressions for
the surface values of the stresses oX,a,, T*, as given by equations (14). Evidently, we have from
equations (14),
ux

h
_lEh
=+21_v2
2

(r- >

UW

w,, + VW,,, - J$L]

tWe note in passing the ~ssibility of introducing effective rototio~

~splacement com~nents rp, and pp,,so that

and therewith, in place of equations (15),

Mx = D(cp,x f ucp,,), na, = D@,.., f WA

2% = (I- G&x.,

+ cp,,f.

572

E. REISSNER

(16b)

and the noteworthy difference between the right hand sides of (15) and (16) is the absence in
equations (16) of the factor 6/5 which occurs in front of the terms with V, and V, in equations
(15). To show the significance of this difference we consider in what follows its effect on the
solution of the problem of the stress concentration due to the presence of a circular hole in the
otherwise uniform transverse pure bending of an infinite plate, as previously considered in [2].
Introducing polar coordinates r, 0 and making use of the notation and also of the results in[2]
the problem which is now being considered is defined by means of the boundary conditions

).

H,,

=so:

M,=O,

r=a:

H,=O,

M,y,

v, = 0

(1%)

V,=O.

(17b)

To be determined on the basis of these boundary conditions are in particular a stress couple
concentration factor ke, defined by k, = Me(a, n/2)/M0, and in addition to this a maximum-stress
concentration factor kg, defined by k$ = a,(a, n/2, h/2)/n, where u. = 6Molh2.
In order to evaluate ke and ki we make use of expressions for Me and ue(h/2) which in
accordance with equations (15) and (16), in association with equation (IIIa) in[21, are

(18)
(19)
In this we now make use of the expressions for W, V, and V, which were obtained in[2] for the
boundary value problem defined by the boundary conditions (17) and by the associated
two-dimensional differential equations. We furthermore make use of various relations and
determinations carried out in[2]. These leave us with expressions for Me(a, 13)and ue(a, 0, h/2)
which are of the following form
M&

2u+ v)Kz(l*)
0) = MO 1-cOS~@(~
+ v)K2(p)
+2~~(~)
t

(20)

and
~@(aJG)=+-cos2e

2(1+ V)K&) - 4K&))/3P

.
(1+v)K*(P)+2Ko(P)
I

(21)

In this K,, K, and K2 are modified Bessel functions and p = VlOalh.

Equations (20) and (21) imply as expressions for k, and kg
k = 3(1+ v)&+2Ko
B

(l+v)Kz+2Ko

k~=3(1+~)K~+2Ko-4K~/3~
(1+ v)Kz+2Ko

(22)

513

On transverse bending of plates

Having equations (22) we are now in a position to compare our results with the results of an
exact analysis of this stress concentration problem, within the framework of three-dimensional
elasticity theory, by Alblas [ 11.It was found by Alblas that his exact values for the stress couple
concentration factor kB were in close agreement with the corresponding approximate values
in [2]. Alblas also obtained what amounted to exact values for the maximum-stress concentration
factor kT, and he showed that, in the range of values of the ratio 2alh which he considered kg
may differ from k, by as much as 10 per cent (111,p. 99), consistent with the observation in[2]
that if standard plate theory gives a value of 1.5 and the present theory a value 2-O(for the factor
of concentration of stress as approximated by the stress couple concentration factor) then it is
believed that the actual value lies in between 190 and 2*10. In this context it is of special
significance that the new approximate maximum-stress concentration factor kQ, in equation (22),
does in fact give numerical values which are in excellent agreement with Alblas values in the
2a/h-range considered by him, with the differences between Alblas exact values of kg and the
present approximate values being considerably smaller than 1 per cent, in accordance with the data
in Table 1.
Table 1. Values of stress couple concentration
factors kB and of maximum-stress concentration
factors kg for transverse pure bending of an
infinite isotropic homogeneous plate with a
circular hole, as a function of the diameterthickness ratio 2a/h, with Poissons ratio Y= l/4
kX

k.
2a
h
o-3
0.5

I
2
3
4
5
m

2.268
2.045
1960
I.914
1.8%
1.I69

2.50
2.243
2.038
1.956
I.912
I.885
1.769

2.052
1.938
1.865
1.841
1.830
1.769

2.41
2.18
2.038
1.922
1.875
1.850
1.835
1.769

REFERENCES
1. J. B. Alblas, Theorie van de Driedimensionale Spanningstoestand in een Doorboorde Plaat. Dissertation, Delft (1957).
2. E. Reissner, The effect of transverse shear deformation on the bending of elastic plates. J. Appl. Mech. 12, A-69-A-77
(1945).