INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING. VOL.

15, 1t87-120O (1980)

A PENALTY PLATE-BENDING ELEMENT FOR THE

ANALYSIS OF LAMINATED ANISOTROPIC COMPOSITE
PLATES
J. N. R E D D Y t
School of Aerospace, Mechanical and Nuclear Engineering, University of Oklahoma, Norman, Oklahoma, W.S.A.

SUMMARY
A Co(penalty)finite element is developed for the equations governingthe heterogeneouslaminated plate
theory of Yang, Norris and Stavsky. The YNS theory is a generalization of Mindlin’s theory for
homogeneous, isotropicplates to arbitrarily laminated anisotropic plates and includes shear deformation
and rotary inertia effects. The present element can also be used in the analysis of thin plates by
appropriatelyspecifying the penalty parameter. A variety of problems are solved,including those for which
solutions are not available in the literature, to show the material effects and the parametric effects of plate
aspect ratio, length-to-thicknessratio, lamination scheme, number of layers and lamination angle on the
deflections, stresses, and vibration frequencies. Despite its simplicity, the present element gives very
accurate results.

INTRODUCTION
Over the past few years composites, especially fibre-reinforced laminates, have found increasing
application in many engineering structures. The fibre-reinforced composites possess two
desirable features: one is their high stiffness-to-weight ratio, and the other is their anisotropic
material property that can be tailored through variation of the fibre orientation and stacking
sequence-a feature which gives the designer an added degree of flexibility.
Recent developments in the analysis of plates laminated of fibre-reinforced materials indicate
that the thickness effect on the behaviour of the plate is more pronounced than in isotropic
plates. The classical thin plate theory assumes that normals to the midsurface before defor-
mation remain straight and normal to the midsurface after deformation, implying that trans-
verse shear deformation effects are negligible. As a result, the free vibration frequencies, for
example, calculated using the thin plate theory are higher than those obtained by the Mindlin
plate theory,’ which includes transverse shear and rotary inertia effects; the deviation increases
with increasing mode number. Higher order linear theories that include transverse shear effects
have also appeared (see Reissner’ and Lo, Christensen and Wu3). Elasticity solutions by Pagano
and his associatese7 indicate the inadequacy of the classical laminated plate theory (e.g.
Reissner and Stavsky* Dong, Pister and Taylorg and Bert and Mayberry,” in which the classical
Kirchoff-Love kinematic assumptions are adopted and effects of transverse shear deformations
are neglected). The transverse shear deformation effects are even more pronounced, due to the
low transverse shear modulus relative to the in-plane Young’s moduli, in t& case of filamentary
composite plates. A reliable prediction of the response characteristics of high modulus
composite plates requires the use of shear deformable theories.
A number of shear deformable theories for laminates have been proposed to date. The first
such theory for laminated isotropicplates is due to Stavsky.” The theory has been generalized to
t Current Address: Professor, EngineeringScience and Mechanics, Virginia Polytechnic Institute and State University,
Blacksburg, Va. 24061, U S A .
0029-5981/80/0815-1187$01.00 Received 2 June 1979
0 1980 by John Wiley & Sons, Ltd. Revised 25 October 1979
1187
1188 J. N. REDDY

laminated anisotropic plates by Yang, Norris and Stavsky.” A review of various other theories,
for example, the effective stiffness theory of Sun and Whitney,13 the higher-order theory of
Whitney and Sun14 and the three-dimensional elasticity theory of Srinivas and Rae," can be
found in the paper by Bert.I6 It has been shown, see for example the papers by Sun and
WhitneyI3 and Srinivas and Rao” that the Yang-Norris-Stavsky (YNS) theory is adequate for
predicting the overall behaviour such as transverse deflections and natural frequencies (first few
modes) of laminated anisotropic plates.
The first application of the YNS theory is apparently due to Whitney and Pagano,I7 who
considered cylindrical bending of antisymmetric cross-ply and angle-ply plate strips under
sinusoidal load distribution and free vibration of antisymmetric angle-ply plate strips. Fortier
and Rossettos” analysed free vibration of thick rectangular plates of unsymmetric cross-ply
construction, while Sinha and Rath’’ considered both vibration and buckling for the same type
of plates. Recently, Bert and ChenZ0presented, using the YNS theory, a closed-form solution
for the free vibration of simply supported rectangular plates of antisymmetric angle-ply
laminates.
While considerable effort has been expended in the finite element analysis of isotropic plates,
only limited investigations of laminated anisotropic plates can be found in the literature. Pryor
and Barker” and Barker, Lin and Dara22 used the conventional displacement finite element
method to analyse thick laminated plates. The element has 7 degrees-of-freedom (three
displacements, two rotations and two shear rotations) per node. Exploiting the symmetries
exhibited by anisotropic plates, Noor and mat her^'^''^ studied the effects of shear deformation
and anisotropy on the accuracy and convergence of several shear-flexible displacement finite
element models based on a form of Reissner’s plate theory. The elementz3 used involved 80
degrees-of-freedom per element. The conventiona1 finite element, when applied to relatively
thick laminated plates, either has failed to predict accurately the local deformations and stresses
of a plate under bending or is too expensive to use due to the large number of degrees-of-
freedom involved for even relatively simple problems. Mau, Tong and Pian2’ have employed the
so-called hybrid stress finite element method to analyse composite plates including shear
deformation. Most recently, Panda and Natarajan26used, following Mawenya and D a ~ i e sthe ,~~
quadratic shell element of Ahmad, Irons and Zienkiewicz” with the same normal rotation
through the thickness to claim improved accuracy over Mawenya and D a ~ i e s . ~The’ ‘thickness
concept’ mentioned there is essentially the same as that used in the YNS theory.12 The authors
were primarily concerned with the accuracy of the element, and no attempt was made to solve
new problems for which there do not exist any closed-form or exact solutions.
The present paper is concerned with the development of a simple Coelement for the YNS
theory of laminated composite plates. The penalty function concept of Courant29 (also see
Zienkiewicz3’) is used to develop the finite element model. The element contains 5 degrees-of-
freedom (three displacements and two slopes, i.e. shear rotations), per node. The accuracy of the
element is demonstrated via problems for which the exact solutions and numerical results are
available, and results are also presented for a variety of problems for which solutions are not
available in the literature.
LAMINATED PLATE THEORY OF YANG-NORRIS-STAVSKY (YNS)
Consider a plate of constant thickness h composed oE a finite number, L, of thin anisotropic
layers oriented at angles el, 02,. . . , OL. The origin of the co-ordinate system is located within the
middle plane (x-y) with the z-axis being normal to the mid-plane. The material of each layer is
assumed to possess a plane of elastic symmetry parallel to the xy-plane. We shall denote the
middle plane with R.
 PENALTY PLATE-BENDING ELEMENT 1189

The YNS theory is based on the following assumed displacement field:

where u, u and w are the displacement components in the x, y and z directions, respectively, t is
the time, uo and uo are the in-plane (stretching) displacements of the middle plane, and & and * y
are the shear rotations.
The equations of motion associated with YNS theory are

aN1 dN6 a2u

-+- =p ~ R- +
ax ay at at2
aN6 a N z a2v
-+-==+R-
ax a y at at2
aQx
-+ aQy a2w
-=p7-P
ax ay at

where

p'"' being the material density of layer m,and P = P(x, y ) is the transversely distributed load.
Assuming that the only plane of symmetry existing is in the plane of the plate, the plate
constitutive equations can be written as
>
U0.X

V0.Y

W.Y + *Y
w,x + (1x
(5)
u0.y + U0.X
*x,x

*Y.Y

*X.Y + *Y.X ,
1190 J. N. REDDY

The material components Aij,Bij and Dii are given by

I h12

c hi2
Aij= k , k d i i , Aii= J
-hl2
a!,!”)
dz, (i, i = 4,5), (Y = 6 - i, p = 6 -i

The stiffness coefficients a$“’

depend on the material properties and orientation of the mth
layer. The parameters kiare the shear correction coefficients.
The strain energy associated with equations (2) and ( 5 ) is the sum of the bending energy, Ut,
and shear energy, U2. The complete expressions for U1 and Uzare given in equations (18) and
(19), respectively, in Appendix I.

PENALTY FUNCTION FORMULATION O F T H E EQUATIONS

The assumption of the classical thin plate theory that the normals to the midsurface before
deformation remain straight and normal to the midsurface after deformation implies that

If we substitute for (I, and $, from (7) into equation (16) of Appendix I, we obtain the strain
energy U = U1 associated with the classical thin plate theory. In that case, U1 involves the
second-order derivatives of the transverse deflection, and the associated (conventional) finite
element formulation results in complicated elements (with many degrees-of -freedom).
Approaches that have been taken to relax the continuity requirements placed on the shape
functions in the displacement formulation of the thin plate theory include, in addition to the
nonconforming, hybrid and mixed formulations, the ‘discrete Kirchhoff hypothesis’ of
Wempner, Oden and K ~ o s s ,Fried,32~~ and the ‘residual energy balancing’ and ‘reduced
integration’ techniques of Fried,” Zienkiewicz, Taylor and Too34and Hughes, Taylor and
Kanokn~kulchai.~’ The present penalty function method is a formalization and extension of
these ideas to the shear deformable theory of laminated composite plates.
The problem of finding the static solution (u, 0,w ) to the thin plate equations can be viewed as
one of finding the critical points of the total potential energy r 1= U1+ V, where U,is the strain
energy given by (18) in terms of u, u and w, and V is the potential energy due to applied loads.
Alternatively, the problem can also be viewed as one of finding (u, 0,w, II,, r),) subject to the
constraint conditions in (7). To incorporate the constraints, one can use the Lagrange multiplier
method, or the penalty function method.
If the Lagrange multiplier method is used, we have

where A, ane A, are the Lagrange multipliers. Comparing the Euler equations of rL = UL+ V
with those of w = U + V, we see that the Lagrange multipliers are given by

Thus, ULis equivalent to the strain energy U of the shear deformable theory.
 PENALTY PLATE-BENDING ELEMENT 1191

If the penalty function method is used, the modified functional is given by 7rp = Ul + Up+ V,
wherein the penalty functional Upis chosen to be

where E : and E : are the penalty parameters. Clearly, in the limits E ~ E~, + 00, the constraints are
satisfied exactly. As opposed to the Lagrange multiplier method the constraints are satisfied only
approximately, and no additional variables are introduced in the penalty method. Comparing
the Euler equations of the functional 7rp with the equations of the YNS theory, we see the
correspondence:

E: = k:A44, & 1 & 2= kl k&, E$ = k22255

This correspondence implies that for very large values of cirthe equations govern the thin plate
theory, and for values of E~ given in (1l ) , the equations coincide with the YNS theory.

FINITE ELEMENT MODELS

Here we present a (semidiscrete) finite element model based on 7rp(u,u, w,t,bx, I&). We assume,
over each element n,, the same kind of interpolation for all of the variables:
n n
ub = uSNr, ub = urNr, etc. ( n = nodes per element) (12)
i i

where NF are the element interpolation (or shape) functions, and U T and 0; are the nodal values
of ub and ub, respectively.. Substituting (12) into the first variation of 7ri(u, u, w ,&, $,), and
collecting the coefficientsQf the variations, Sui, Sui, etc., we obtain
[M‘]{A‘}+[ K e ] { A e=} {F‘)
where

P[S01 “OI
\
P[S01 R‘ [So]
\
P[S01
\
symmetric mO1
\
m01
The elements of the stiffness matrix, [K], and the mass matrix, [MI are given in equations (20)
in Appendix 11.
For free vibration problems, equation (13) becomes
([K‘]-w*[M‘]){A‘}= (0) (15)
1192 J. N. REDDY

where w is the frequency of the natural vibration, For static analysis, {A} is set to zero. The
element stiffness matrices are assembled in the usual manner, and boundary conditions of the
problem are imposed before solving for {A} or wn.
In the present study linear ( n = 4) and quadratic ( n = 8)elements of the Serendipity family are
used. The element stiffness matrices for these elements are of order 20x20, and 40x40,
respectively.

NUMERICAL EXAMPLES AND DISCUSSION

The (penalty) finite element developed herein was employed in the bending and free vibration
analyses of a variety of layered composite rectangular plates. Most of the numerical results
presented here were obtained using a uniform mesh of 2 x 2 quadratic b e . 8-node quadrilateral)
elements in the quarter-plate. Computations were conducted on an IBM 370/158 computer.
In the following analyses two types of boundary conditions, simply supported and clamped,
and two types of orthotropic materials were used. The co-ordinate system and boundary
conditions are shown in Figure 1. The material properties used are ( G l z= G13; v12= ~ 1 ~ ) ;
Material I: E1/E2= 25, G12/E2= 0.5, G23/E2 = 0.2, v12 = 0.25
Material 11: E 1 / E 2= 40, G12/EZ
= 0.6, G23/E2 = 0.5, v12= 0.25
A value of 5 / 6 was used for the shear correction coefficients (see W h i t n e ~ ~ ~ ) .

layer n

u=JI,=o

5
v=(J =o V'Q =o
Y Y

Figure 1 . Co-ordinate system, finite element mesh and boundary conditions

Bending analysis
First the effect of the 'reduced integration' on the bending deflections and stresses is examined
using a four-layer (equal thickness), cross-ply (0°/900/900/00),simply supported (SS) plate
(material I) subjected to sinusoidal loading (SSL),
lrx lry
P = Po cos--0s -
a b
This problem is equivalent to the three-layer, the two outer layers are of thickness h/4 each and
the middle layer is of thickness h / 2 , cross-ply (O0/9Oo/O0) plate. For this problem the closed-
form solution to the equations governing the YNS theory can be obtained (see Reddy3').
The percentage error (between the closed-form solution and the finite element solution) in the
centre deflection as a function of the side-to-thickness ratio ( a / h )is plotted in Figure 2(a) for
 PENALTY PLATE-BENDING ELEMENT 1193

/v I'
L
Q -- quadratic
linear
- F - f u l l integr.
16
Q1F -' /!
,+L4F R - reduced i n t . -
1 1 ; Q l R = 1x1-Q-R, etc.
\ / /
'-
- #'
/
,/

/
L /
L
0
L
aJ Q2R
a 8
cn
m,
c L2R
E
aJ
u
L
0
Q3F
- 4
I- Q3R L4R \ Q2F7

0
a/h 3 10 20 30 40 50

(a) I Error = (CF-FEN) x 1 O O / C F

--
r-
3.- L-linear
L- 1inear
3
+ 9-quadrati c
I
(w) L22

2*t
U
a
u 2.
-
1

w
Zi
-0
aI
L

L
k0 1.-
W

L22( stress

0).
0
a/h 10 20 30 40 50
(b) Error = ((FEM)red.-(FEM)fU1l
FEM)red.-(~~~4)fU11

Figure 2. Effect of the mesh, element type and reduced integration on the accuracy of solutions (CF-closed form,
FEM = finite element method)

various meshes of the linear and quadratic elements. For the linear element the reduced
integration involves the use of 1x 1 Gauss rule and the full integration involves the use of 2 x 2
Gauss rule. Similarly, for the quadratic element the reduced integration involves the use of 2 x 2
Gauss rule and the full integration involves the use of 3 x 3 Gauss rule. From Figure 2(a) it is
clear that use of the full integration results, for any mesh and element, in larger error in the
deflection than when the reduced integration is used. The error decreases with a refined mesh
1194 J. N. REDDY

and/or a higher order element. It is also clear that the percentage error for thick plates
( a / h < 2 5 ) does not decrease at the same rate as for thin plates ( a / h> 2 5 ) when the mesh is
refined. The difference between the normalized centre deflection obtained by the reduced
integration and that obtained by the full integration is plotted against the side-to-thickness ratio
in Figure 2(b). It is clear that the difference approaches zero as a / h approaches the value of 5 .
That is, reduced integration has little or no effect on the deflections of thick plates. Further, the
normal stresses (computed at the Gaussian points) are much less affected by the reduced
integration. It should be noted that we cannot play with the values of penalty parameters in the
present problem-(they are defined by equation (11))-for we are interested in the thick plate
theory. Guided by these observations, only 2 x 2 mesh of quadratic elements with reduced
integration was used in the remaining analyses.
To assess the numerical convergence of the present element, the percentage error in the
central deflection and in the maximum normal stress are plotted against the number of elements
in the mesh in Figure 3. It is clear that the rate of convergence is faster for thin plates than for
thick plates, and for quadratic elements than for linear elements.

40 I I I

R - reduced integration
l3 30 F - full integration
E
.r

2
L
L
al
0)
20
m
ml
L Liner elements (a/h = 10)-F
K
aJ
V
L
2 10

Linear elements ( a / h = 100)-R

i 4 Number o f elements 16

Figure 3. Numerical convergence of the penalty finite elements

Figure 4 shows the bending deflection versus the side-to-thickness ratio for the same problem
using 2 x 2 Gauss rule. This result is in excellent agreement with the closed-form solution of
Whitne~.~' Note that the effect of the thickness shear is more pronounced for side-to-thickness
ratios less than 25. Figure 4 also shows the stresses, ex,eYand f X yfor the four-layer (equal
thickness), cross-ply (Oo/900/90"/O"),simply-supported square plate under sinusoidal loading.
To see the effect of loading and material on the deflection, the same problem was solved using
material I1 and uniform loading, and material I1 and sinusoidal loading. Note that decreasing the
ratio E I / E 2from 40 to 25 has the same effect as using the uniform loading in place of sinusoidal
loading. Bending deflections and stresses are presented in Table I for a (four-ply
(0°/900/900/00),material I) clamped plate under sinusoidal loading, and simply-supported plate
under point load at the centre.
 PENALTY PLATE-BENDING ELEMENT 1195

14. - I I I I I

l (0"/90"/90"/0",S S )

12. -
N
m
0
n
\
0
(u

x
10. -
N
r
0
I1

I0 8. -
-ru
f

0
a
\
0
0 6. -
7

X
m
r
N
Y
3 4. -
I1

13

2. -

a/h -+ 10 20 30 40 50

Figure 4. Bending deflections and stressesvs. side-to-thicknessratio for four-layer,cross-ply,simply-supportedsquare

plates

Table I. Four-layer (0°/900/900/0") square plate (material I)

Clamped plate under sinusoidal loading Simply-supportedplate under point load

Normalized Normalized maximum stress, 6 Normalized Normalized maximum stress, 6

center center
a/h deflection, w Fa: r U: *TF, deflection, GJ Tut Fa,"
C
T,y *
5 12.9544 0.4284 0.5084 0.00900 105.448 2.3692 3.4564 0.0874
10 6.6956 0.4896 0.3538 0.01083 42.015 2.6500 2.7160 0.0765
20 4.9415 0.5150 0.2880 0.01319 25.344 2.7900 2.3258 0.0712
25 4.7208 0.5184 0.2787 0-01378 23.287 2.8144 2.2560 0.07029
50 4-4222 0.5232 0.2662 0.01484 20.493 2.8536 2.1296 0.06904
100 4.3567 0.5255 0.2634 0.01519 19.763 2.8640 2.0590 0.06916

6 = W(E2h3/Poo4)103,b -- oh2/Poa2, A = (0.0528.0.0528, h / 2 ) .

B = (0.0528.0.0528. h/4). C = (0.4472.0.4472, h/2).
1196 J. N. REDDY

To further illustrate the accuracy of the present element, two problems for which exact”7
three-dimensional elasticity solutions and finite element solution^^^'^^ are available, were solved
and results are summarized in Tables I1 and I11 (for material I). Table I1 contains the normalized
bending deflections and stresses for three-layer, cross-ply (Oo/900/Oo),simply-supported square
plate subjected to sinusoidal loading. The outer layers are each h/4, and the middle layer is h / 2

Table 11-Three-layer (O0/9Oo/O0)simply-supported square plate subjected to sinusoidal loading (material

I, t i = t 3 = h/4, t 2 = h/2)
Normal stresses, C+ (top and bottom):
Normalized
centre a b h
Source deflection, CI

5.0 Present FEM 2.964 0-419 0.500 0-0280

6.25 Present FEM 2,300 0.444 0.443 0.0262
10 Pagano & Hatfield7 1.709 0.559 0.403 0.0276
Present FEM 1.534 0.484 0.350 0.0234
Panda and Natarajan26 1.448 0.532 0.307 0.0250
Mawenya and Davies” 2.034 0.542 - 0.0292
12.5 Present FEM 1.546 0.496 0.322 0.0224
20 Pagano and Hatfield7 1,189 0.543 0.309 0.0230
Present FEM 1.136 0.511 0.287 0.0214
Panda and Natarajanz6 1.114 0.557 0.307 0.0231
Mawenya and DaviesZ7 1.273 0-546 - 0.0239
25 Present FEM 1.086 0.515 0.277 0.02 11
50 Pagano and Hatfield7 1.031 0.539 0.276 0.0216
Present FEM 1.019 0.520 0-265 0.0207
Panda and Natarajan26 1.016 0.565 0.287 0.0225
Mawenya and D a v i e ~ ~ ~ 1-048 0.550 - 0.0221
100 Pagano and Hatfield’ 1.008 0.539 0,271 0.0214
Present FEM 1.005 0.523 0.263 0.0207
Panda and Natarajan26 1.003 0.566 0.284 0.0223
Mawenva and D a v i e ~ ~ ~ 1.015 0.551 - 0.0213
Classical plate theory 1.000 0.539 0.269 0.0213
~~ ~

KJ= w , ( ~ ~ / P ~ u ~5)-, r . ~ h ~ / ! ‘ o ~ ’ , a = {4GI2+ [El + (1 + 2 v 1 2 ) E2 ] /( 1 -~ , ~ v ~ ~ ) } r ~ / 1 2 .

+ Computed at the Gaussian points in the present study.

thick (i.e. sandwich construction equivalent to the four-layer cross-ply problem considered
earlier). Table I11 contains similar information for three-layer (equal thickness), cross-ply
(Oo/900/Oo),simply-supported rectangular ( b / u = 3) plate under sinusoidal loading. Present
solutions are compared with exact solutions of Pagano’ and Pagano and Hatfield7, and the finite
element solutions of Panda and Natarajan26 and Mawenya and D a ~ i e s . * While
~ all three finite
element solutions are fairly close to the exact, it is clear that the present solution is the closest to
the exact solution for the deflection for all ratios of a / h . It should be remembered that the
present element is based on the thick plate theory, not on the three-dimensional elasticity
theory. Since the stresses in the present study are computed at the Gaussian points, it is expected
that they would not be very close to the exact solutions. However, the magnitudes of the stresses
are in the vicinity of the exact values.
Figure 5 shows the normalized bending deflections versus the ratio u / h for the three-layer,
cross-ply, simply-supported square plate under sinusoidal loading. For material I the same
 PENALTY PLATE-BENDING ELEMENT 1197

Table 111. Three-layer (O0/9Oo/O0)simply-supported rectangular plate (b/a = 3) subjected to sinusoidal

loading (material I)

Normalized stress, Cr (top and bottom):

Normalized
centre O,O,h
T rX( -f)
7)OOh fly( 2‘a b2’ i)
f7.~( h
Source deflection, +
5 Present FEM 1.695 0.598 0.0691 0.0179
6.25 Present FEM 1.267 0.601 0.0540 0.0134
10 Exact: Pagano’ 0.919 0.725 0.0435 0-0123
Present FEM 0.802 0.603 0.0364 0.0102
Panda and Natarajan’“ 0.752 0.653 0.0367 0.0105
Mawenya and Davies” 1.141 0.685 - 0.0141
12.5 Present FEM 0.694 0604 0.0322 0.0094
20 Exact: Pagano’ 0.610 0-650 0.0299 0.0093
Present FEM 0.578 0.605 0.0276 0.0086
Panda and Natarajan26 0.565 0.654 0-0287 0.0091
Mawenya and D a ~ i e s ~ ~ 0.664 0.65 1 - 0.0099
25 Present FEM 0.551 0.605 0.0264 0.0084
50 Exact: Pagano’ 0.520 0.628 0-0259 0.0084
Present FEM 0.515 0.604 0.025 1 0.008 1
Panda and Natarajan26 0.5 13 0.654 0.0264 0.0087
Mawenya and D a v i e ~ ~ ~ 0.529 0.640 - 0.0087
100 Exact: Pagano’ 0.508 0.624 0.0253 0.0083
Present FEM 0.506 0.603 0.0253 0-0080
Panda and Nataraian’“ 0.505 0.654 0.0261 0.0086
Mawenya and Dahes’’ 0,510 0.638 - 0,0085
Classical plate theory 0.503 0.623 0.0252 0.0083
6 = w100E2h3/Poa4, 5=Crh2/Poa2.
+ Computed at the Gaussian points in the present study.

4.
I I 1 I

+ Oo/900/Oo, Material I
3. - -0- Oo/9l0/Oo, Material I
f
m -0- Oo/900/Oo, Material I 1
0
a.
m. 2. tl=t3 W 4 , t Z = h / 2)
r
N
W
P
W
I3
1. -
- c

I I I 1

a/h + 10 20 30 40 50
Figure 5. Bending deflections vs. side-to-thickness ratio for three-layer, cross-ply, simply-supportedsquare plate under
sinusoidal loading
1198 J. N. REDDY

problem was solved with layer orientation of O0/9lo/O0(the middle layer is now oriented at 91")
to see the effect of slight variation (introduced, say, in manufacturing) in the orientation of the
layers on the deflection. Note that the slight error in the angle causes slight variation in the
deflection only at higher values of a / h (i.e. for thin plates).
Figure 6 shows plots of bending deflection versus the side-to-thickness ratio for two-layer,
cross-ply (0"/90")square plate (material 11) under sinusoidal and uniform loadings, and for
four-layer, symmetric angle-ply (45"/-45"/-45"/45") square plate (materials I and 11) under
sinusoidal loading. It is clear that the effect of shear deformation is quite significant in (cross-ply,
as well as angle-ply) composites with side-to-thickness ratio, a / h < 20.

I 1 I I
1
1 -0- oo/900, UDL, Flat. I 1
-
-0- 45°/-450/-450/450, SSL, N a t I-

--c oo/900, SSL, Y a t . I1 -

-8- 45"/-450/-450/45", SSL, M a t 11

I I I I I 1 1
a/h -+ 10 20 30 40 50
Figure 6. Bending deflections vs. side-to-thickness ratio for two-layer cross-ply (0"/90")
and four-layer, angle-ply,
simply-supported square plate

Bending deflections and stresses versus side-to-thickness ratio are shown in Figure 7 for two-
and four-layer angle-ply (45"/-45"/45"/-45") square plates. For the same plate thickness, a
two-layer, angle-ply plate undergoes larger deflection than the four-layer, angle-ply plate.
The stresses ex= eYand f x yare shown only for the four-layer, angle-ply plate under uniform
loading. The stresses corresponding to the sinusoidal loading (for the same problem) are scaled
with respect to the stresses associated with the uniform loading. In the case of sinusoidal loading
both normal stress and shear stress increase with decreasing side-to-thickness ratio.

Free vibration analysis

Figure 8 shows plots of the nondimensionalized fundamental frequencies versus side-to-
thickness ratio for four-layer, angle-ply (symmetric and antisymmetric) square plates. The
 12. I I 1 1 1 I I 1 I
20.0
4-layer, SSL, Material I 1
10. + 2-layer, SSL, Fhterial I 1 -
--c 4-layer, UDL. Material 1 1 19.8
N
f -m- 2-layer, S S L , Material I 4
rn 8. - 0
0
a.
\
0- N
\ r
rv 19.6
0 N

m z
r 6. - x
N
Y 0
I1
I1 19.4
13 10

4. -
4

19.2

2. -
0
19. 1 1 I I I 13.0
a/h - 10 20 30 40 50 a/h - 10 20 30 40 50

( a ) Deflections (b) Stresses (4-layer, Material 11)

Figure 7. Bending deflections and stresses vs. side-to-thickness ratio for angly-ply, simply-supported square plates
1200 J. N. REDDY

I I I I I

24.

20.

16.

12.

8.

Figure 8. Fundamental frequencies vs. side-to-thickness ratio for four-layer, angle-ply square plates

results obtained for simply-supported (material 11) plate is in fair agreement with the closed-
form solution of Bert and Chen," with the maximum error (at a / h = 50) being 4.6 per cent. The
present study predicts higher frequencies, with the deviation decreasing with a / h . Figure 8 also
shows the plot of fundamental frequencies for the symmetric angle-ply (45"/-45"/-45"/45",
material I). Incidentally, this plot is in excellent agreement with that in Figure 5 of Whitney and
Pagano." However, the figure caption there (i.e. in Reference 17) says that the result was
obtained for four-layer, antisymmetric angle-ply (45"/-45"/45"/-45"), simply-supported
square plate (material 11). As pointed out by Bert and Chen," and confirmed by the present

r I I I I

22. .
-
CI

N
m
Oo/900/900/O", M a t e r i a l I 1 -
(b/a=l)
14. -
4
-0- Oo/900/Oo, Materi a1 I I (b/a=3)'
oo/900/Oo,M a t e r i a1 I I (b/a=l1.
10. I I I I

a/h + 10 20 30 40 50
Figure 9. Fundamental frequencies vs. side-to-thickness ratio for cross-ply plates
 PENALTY PLATE-BENDING ELEMENT 1201

22.
-0- (e/-e/e/-e, Material 11)

k/:::j
-0- (e/-e/e, Material I )

p
18.

N
4 14.
I1
4

10. I 1
e -, 10" 20" 30" 40" 50"
Figure 10. Fundamental frequencies vs. angle of orientation for angle-ply, simply-supported square plates ( a / h = 10)

study, the plot shown in Figure 5 of Reference 17 does not correspond to the antisymmetric
angle-ply plate. To identify the result with the right problem the author experimented with
material I and with clamped boundary conditions, for which results are also shown in Figure 8.
Obviously, none of these come close to that presented by Whitney and Pagano." Thus, Figure 5
of Whitney and Pagano" corresponds to four-layer, symmetric angle-ply ( 4 5 O / - 4 5 O / - 4 5 O / 4 5 O ) ,
simply-supported square plate with layers made of material I.
Similar results are presented in Figure 9 for three- and four-layer, cross-ply simply-supported
plates (material 11). Interestingly, the three-layer and four-layer, cross-ply square plate have
almost the same fundamental frequencies for a / h < 15. Figure 10 shows plots of nondimen-
sionalized fundamental frequencies versus the angle of orientation for three- and four-layered
square plates (material 11, and a / h = 10). Finally, Figure 11 shows the effect of the number of
layers on the fundamental frequency of layered angle-ply (45"/-45"/ + / - / . . . ) square and
rectangular plates (material 11). For a number of layers greater than six, the fundamental
frequency is virtually the same.

:c
I I

10. - 0 - 0 - 0 - 0 4 - ~

b/a = 1, *' b/a = 3

9.
1
\

I3
41
R)
01
2

8.
-- r - - - O - I
7.
1202 J. N. REDDY

CONCLUSIONS
Using the penalty function concept of Courant29to the equations governing the thin plate
theory, a shear deformable theory for layered composite plates that resembles the YNS theory12
is presented. A finite element model based on the penalty/YNS theory is developed herein and
applied to the bending and free vibration analyses of rectangular composite plates with various
edge conditions and loadings. The present finite element is very simple and computationally
efficient. The numerical results are compared with those obtained by other finite element
methods and with exact solutions. The present element, despite its simplicity in formulation and
programming, gives very accurate results.
Application of the element to nonlinear (in the von Karman sense) and bimodulus (i.e.
different elastic properties in tension and compression)plate problems was investigatea recently
by the However, its application to a nonlinear, shear deformable theory of
composite plates is still awaiting (see Reddy and Chao41).

ACKNOWLEDGEMENTS

Support of this work by the Office of Naval Research through Contract NO0014 78-C-0647 is
gratefully acknowledged. The author is also thankful to Professor C. W. Bert for helpful
comments and to Mr. W. C. Chao for computational assistance.

APPENDIX I: STRAIN ENERGY AND KINETIC ENERGY ASSOCIATED WITH

THE YNS THEORY
The strain energy and the kinetic energy associated with equations (2) and ( 5 ) are given by

T =.!
2
In
{ ($)* + (
p[ $)2 + ($ ) 2 ] +I[( $)I z)2]
+( + 2R[
?! ! ! a" + %
at at
"1)
at at
dx dy (17)

where
 PENALTY PLATE-BENDING ELEMENT 1203

Note that the quantities in the square brackets of U2 are the shear forces Q, and Qy,
respectively.
1204 J. N. REDDY

APPENDIX 111: NOMENCLATURE

Aii,Bii,Dii =Extensional, flexural-extensional and flexural stiffness (i, j = 1,2,6).
a, b = Plate planform dimensions in x, y directions.
E l , E2= Layer elastic moduli in directions along fibres and normal to them, respec-
tively.
F, = Force components in the finite element formulation (i = 1,2, . . . , 5 ) .
GI2,G13,GZ3= Layer in-plane and thickness shear moduli.
h =Total thickness of plate.
Z = Rotary inertia coefficient per unit midplane area of lamina.
ki = Shear correction coefficients associated with the yz and xz planes, respectively
(i = 1,2).
K ; B = Element stiffness coefficients (i, j = 1 , 2 , . . ., 8 ; a,p = 1,2,. . . , 5 ) .
L =Total number of layers in the plate.
Mi, Ni= Stress couple and stress resultant, respectively (i = 1,2,6).
M &= Element mass coefficients (i, j = 1,2, . . . , 8 ) .
Nr = Element shape functions (i = 1,2, . . . , 8 ) .
n = Nodes per element.
p = Laminate normal inertia coefficient per unit midplane area.
P = Transversely distributed load.
Po = Intensity of transversely distributed load.
Q,, Q, = Shear stress resultants.
Qij= Plane stress reduced stiffness coefficients (i, j = 1,2,6)
R = Laminate rotary-normal coupling inertia coefficient per unit midplane area.
S g = Element matrices in FEM formulation (i, j = 1,2, . . . , 8 ; 5, = 0, x, y).
f = Time.
u, u, w = Displacement components in x, y, z directions, respectively.
uo, uo = In-plane displacements in x , y directions.
ui, ui = Nodal values of displacements u, u (i = 1,2, . . . , 8 ) .
U, Ul, U2,UL= Strain energies.
V = Potential energy.
x, y, z = Position co-ordinates in Cartesian system.
yxy,yx2,yyz= Shear strains.
{A} = Column of vector of generalized nodal displacements.
E~ = Penalty parameters (i = 1,2).
E,, E,, E= = Normal strains.
Om = Orientation of mth laminate ( m = 1,2, . . . ,L).
A,, A y = Lagrange multipliers.
p = Density of the material.
r,rp= Total potential energy functionals.
a,,ay, a2= Normal stresses.
7xy,T ~ T~~ ~ =, Shear stresses.
JI,, $y = Slope functions.
o = Fundamental frequency of free vibration.

REFERENCES
1. R. D. Mindlin, ‘Influenceof rotary inertia and shear on flexural motions of isotropic, elastic plates’, J. Appl. Mech.
18,31-38 (1951).
2. E. Reissner, ‘On transverse bending of plates, including the effect of transverse shear deformation’, Inr. J. Sol.
Srruer. 11, 569-573 (1975).
 PENALTY PLATE-BENDING ELEMENT 1205

3. K. H.Lo, R. M. Christensen and E. M. Wu, ‘A higher-order theory of plate deformation: part 1, homogeneous
plates; part 2, laminated plates’, J. Appl. Mech., 44, 662-668, 669-676 (1977).
4. N. J. Pagano, ‘Exact solutions for composite laminates in cylindrical bending’, 1. Comp. Marer., 3(3), 398411
(1969).
5. N. J. Pagano, ‘Exact solutions for rectangular bidirectional composites and sandwich plates’, I. Comp. Muter., 4,
20-34 (1970).
6. N. J. Pagano and S.D. Wang, ‘Further study of composite laminates under cylindrical bending’, J. Comp. Muter. 5.
521-528 (1971).
7. N. J. Pagano and S.J. Hatfield, ‘Elastic behavior of multilayer bidirectional composites’. A I A A J., 10, 931-933
(1972).
8. E. Reissner and Y. Stavsky, ‘Bending and stretching of certain types of heterogeneous aerotropic elastic plates’, J.
Appl. Mech., 28.402408 (1961).
9. S . B.Dong, K. S.PiEter and R. L.Taylor, ‘Onthe theory of laminated anisotropic shells and plates’, J. Aerospace
Sci., 29,969-975 (1962).
10. C . W. Bert and B. L. Mayberry, ‘Free vibrations of unsymmetrically laminated anisotropic plate with clamped
edges’, J. Comp. Maw., 3, 282-293 (1969).
11. Y. Stavsky, ‘Onthe theory of symmetrically heterogeneous plates having the same thickness variation of the elastic
moduli’, in Topicsin Appl. Mech. (Eds. D. Abir, F. Ollendorff and M. Reiner). American Elsevier, New York, 1965,
p. 105.
12. P. C. Yang, C. H. Norris and Y.Stavsky, ‘Elastic wave propagation in heterogeneous plates’, Inr. J. Sol. Srrucr., 2,
665-684 (1966).
13. C. T. Sun and J. M. Whitney, ‘Theories for the dynamic response of laminated plates’, AIAA J., 11, 178-183
(1973).
14. J. M. Whitney and C. T. Sun,‘A higher order theory for extensional motion of laminated composites’, J. Sound Vib.,
30,85-97 (1973).
15. S. Srinivas and A. K. Rao. ‘Bending. vibration, and buckling of simply supported thick orthotropic rectangular
plates and laminates’. Int. 1.Sol. Srrucr., 6 , 1463-1481 (1970).
16. C. W. Bert, ‘Analysis of plates’, in SrructzualDesignand Analysis, part I, (Ed. C. C. Chamis), Academic Press, New
York, 1974.
17. J. M. Whitney and N. J. Pagano, ‘Shear deformation in heterogeneous anisotropic plates’, J. Appl. Mcch., 37,
1031-1036 (1970).
18. R. C. Fortier and J. N. Rossettos, ‘Onthe vibration of shear deformable curved anisotropic composite plates’, J.
Appl. Mech., 40,299-301 (1973).
19. P. K. Sinha and A. K. Rath, ‘Vibration and buckling of cross-ply laminated circular cylindrical panels’, Aeronaut.
Qly., 26,211-218 (1975).
20. C. W. Bert and T. L. C. Chen, ‘Effect of shear deformation on vibration of antispmetric angle-ply laminated
rectangular plates’, Int. J. Sol. Srruct., 14,465473 (1978).
21. C. W. Pryor Jr. and R. M. Barker, ‘A finite element analysis including transverse shear effects for applications to
laminated plates’, AIAA J., 9,912-917 (1971).
22. R. M. Barker, F. T. Lin and J. R. Dara, ‘Three-dimensional finite-element analysis of laminated composites’.
National symp. Computerized Srructural Analysis and Design, George Washington University, 1972.
23. A. K.Noor and M. D. Mathers, ‘Anisotropy and shear deformation in laminated composite plates’. A I A A 1..14,
282-285 (1976).
24. A. K. Noor and M. D. Mathers. ‘Finite element analysis of anisotropic plates’, Inr. 1. num. Merh. Engng, 11,
289-307 (1977).
25. S. T.Mau. P. Tong and T. H. H. Pian, ‘Finite element solutions for laminated thick plates’, J. Comp. Muter., 6 ,
304-311 (1972).
26. S. C. Panda and R. Natarajan, ‘Finite element analysis of laminated composite plates’, Inr. J. nurn. Merh. Engng, 14,
69-79 (1979).
27. A. S.Mawenya and J. D. Davies, ‘Finite element bending analysis of multilayer plates’. Int. J. num. Merh. Engng. 8,
215-225 (1974).
28. S.Ahmad, B. M. Irons and 0.C. Zienkiewicz, ‘Analysis of thick and thin shell structures by curved finite elements’,
Int. J. num. Merh. Engng, 2,419-451 (1970).
29. R. Courant, Calculus of Variationsand Supplementary Notes and Exercise, revised and amended by J. Moser. New
York University, 1956.
30. 0. C. Zienkiewicz, ‘Constrained variational principles and penalty analysis function methods in finite elements’,
Conf. on Numer. Solution of Differential Equation, Dundee, Lecture Notes on Marhematics. Springer-Verlag.
Berlin 1973.
31. G. Wempner, J. T. Oden and D. A. Kross,‘Finite element analysis of thin shells’, Proc. ASCE, 1.EngngMech. Dio.,
95, 1273-1294 (1968).
32. I. Fried, ‘Shear in c“ and C’plate bending elements’, Int. J. Sol. and Struct.. 9.449-460 (1973).
33. I. Fried, ‘Residual energy balancing technique in the generation of plate bending finite elements’, Comp. and Smtct.,
4,771-778 (1974).
34. 0 .C . Zienkiewicz. R. L. Taylor and J. M. Too,‘Reduced integration technique in general analysis of plates and
shells’, Int. J. nurn. Meth. Engng., 3, 575-586 (1971).
1206 J . N. REDDY

35. T. J. R. Hughes, R. L. Taylor and W. Kanoknukulchai. ‘A simple and efficient finite element for plate bending’, Inr.
J. num. Merh. Engng, 11, 1529-1543 (1977).
36. J. M. Whitney, ‘Stress analysis of thick laminated composite and sandwich plates’, J. Comp. Mater., 6, 426-440
(1972).
37. J. N. Reddy, ‘Comparisons of closed form and finite-element solutions for bending and free vibrations of layered
composite rectangular plates’, Rep. OW-AMNE-79-20,School of Aerospace, Mechanical, and Nuclear
Engineering, University of Oklahoma, Norman, OK.
38. J. M.Whitney, ‘The effect of transverse shear deformation on the bendingof laminated plates’, J. Comp. Mater. 3,
534-547 (1969).
39. J. N. Reddy, ‘Simple finite elements with relaxed continuity for non-linear analysis of plates’, Proc. 3rd Infernat.
Conf. in Ausfralia on Finite Element Methods, University of New South Wales, Sydney, July 2-6 1979.
40. J. N. Reddy and C. W. Bert, ‘Analysis of plates constructed of fiber-reinforced bimodulus composite material’, in
Mechanics of Bimodulus Materials (Ed.. C . W. Bert), ASME Winter Annual Meeting, New York, 1979,pp. 67-83.
41. J. N. Reddy and W. C. Chao, ‘Large deflection and large amplitude free vibrations of laminated plates’. Symp.
ComputarionalMerhs. in Nonlinear Smicrural Solid Mechanics, George Washington University. 1980.