Journal of Composite

Materials http://jcm.sagepub.com/

Exact Solutions for Composite Laminates in Cylindrical Bending

N.J. Pagano
Journal of Composite Materials 1969 3: 398
DOI: 10.1177/002199836900300304

The online version of this article can be found at:

http://jcm.sagepub.com/content/3/3/398

Published by:

http://www.sagepublications.com

On behalf of:

American Society for Composites

Additional services and information for Journal of Composite Materials can be found at:

Email Alerts: http://jcm.sagepub.com/cgi/alerts

Subscriptions: http://jcm.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://jcm.sagepub.com/content/3/3/398.refs.html

>> Version of Record - Jan 1, 1969

What is This?
Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014
 Exact Solutions for Composite Laminates
in Cylindrical Bending
N. J. PAGANO

Air Force Materials Laboratory

Nonmetallic Materials Division
Wright-Patterson AFB, Ohio
and
Washington University
St. Louis, Mo.

(Received May 7, 1969)

Limitations of classical laminated plate theory are investigated by

comparing solutions of several specific boundary value problems in this
theory to the
corresponding theory of elasticity solutions. The general
class of problems treated involves the geometric configuration of any
number of isotropic or orthotropic layers bonded together and subjected
to cylindrical bending. In general it is found that conventional plate
theory leads to a very poor description of laminate response at low
span-to-depth ratios, but converges to the exact solution as this ratio
increases. The analysis presented is also valid in the study of sand-
wich plates under cylindrical bending.

INTRODUCTION
that classical laminated
plate theory (CPT)
A [1,21 possessesrecognized
LTHOUGH it is
certain deficiencies, it has wide usage the basis for as

the analysis and design of structural laminates. Among the characteristics

of CPT which limit its generality in the description of laminate response
are the following:

a) the assumption of linear in-plane displacements through the thick-

ness, in particular, for laminates in which the stiffness properties
vary drastically from layer to layer
b) the presence of only 2 boundary conditions per edge in the bending
theory which precludes the precise calculation of boundary layer
effects, such as stress concentration factors
c) the neglect of shear deformation, implied by the Kirchhoff hypothe-
sis (normals remain normal)
d) the assumption of a state of plane stress in the constitutive relations,

J COMPOSITE MATERIALS, Vol. 3 (July 1969), 398

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

 which eliminates the possibility of rigorous calculation of inter-
laminar stresses.
While Whitney [3] has presented an approximate method to incorporate
the influence of shear deformation on plate deflection in composite lami-
nates, it is important to assess the limitations imposed by the CPT assump-
tions on the computation of stress, as well as displacement. In this paper
we shall construct the exact solutions (within the linear theory of elasticity)
of a number of specific boundary value problems of unidirectional and
bidirectional (0°-90°) layered anisotropic systems under cylindrical bend-
ing and compare them to the corresponding CPT solutions. In subsequent
papers, we shall examine the response of plates with finite boundaries
and more general stacking geometries to establish more conclusive
evidence regarding the credibility and limitations of CPT.

ELASTICITY SOLUTION

Consider laminate composed of m orthotropic layers such that the

a
various axes of material symmetry are parallel to the plate axes x, y, z
(see fig. 1). The body is in a state of plane strain with respect to the xy
plane and is simply supported on the ends x = 0 and x= t. We shall deter-
mine the response of the plate under a normal traction Œy = q (x) applied
on the upper surface. Since each layer is orthotropic, the constitutive

equations are given by [4]

where Ri, are the reduced compliance coefficients for plane strain, defined
by

and SlJ are the compliances with respect to the axes of material symmetry.
The remaining governing equations are the equations of equilibrium

where differentiation is denoted by a comma, and the strain-displacement

relations

In plane strain problems, the above stresses are accompanied by the

399

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

component az, where

and all stress, strain, and displacement components are functions of x

and y alone.
The prescribed boundary conditions on the upper and lower surfaces
are given by

while the simple support conditions are simulated by

We now introduce an index to identify the layers, where the top layer
corresponds to i =
a local coordinate system x,, yi, z,
1. We also construct
on the center line of the ith
layer the end x = 0. The thickness of the ith
at
layer designated hz.
is as To establish continuity of traction and displace-
ment at the interfaces between layers we must satisfy the conditions
/ 7-B / 7- B

Recognizing that almost any load function can be expressed in the form of
a Fourier series, we let

where q~ is a constant and

The solution of the boundary value problem described by eqs. (1)-(10)

can be found by putting

400

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

which satisfy the equations of equilibrium (3) identically. Using eqs.
(1), (4), and (11), we find that the functions £ (y) are defined by the solu-
tion of the ordinary differential equation

which is expressed by

where A,, are constants, provided that the m,2 are all distinct (for a given
i). The various m,, are given by

where

The composite material properties considered later on are such that

Consequently, the mj, are all real and distinct and the coefficients A,, are
all real. The stress components now take the form

while sufficiently general expressions for the displacement components are

The boundary conditions (7) are identically satisfied by the functions

401

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

 Figure 1. Plate notation.

given in eqs. (16) and (17). Satisfaction of the remaining boundary condi-
tions (6) and interface continuity conditions (8) lead to a system of 4m
simultaneous equations for the 4m constants A,~. These equations are
easily solved on a computer for given material and geometrical parameters.
The stress and displacement fields are then given by eqs. (16) and (17).
In the event that b, in eq. (15) vanishes for a given layer, which occurs
for material that is isotropic or transversely isotropic in the xy plane, the
expression for fi(y) given by eq. (13) is no longer valid. In this case, the
functions become

where m12 is given by eq. (14) with bi = 0. The stresses are as defined in
eqs. (11), while the displacements take the form

where the integral in the above equation represents the indefinite integra-
tion of eq. (18). Equations (18) and (19) can be utilized, for example, in the
description of the core of a sandwich plate.

CPT SOLUTION
In order to conform to the customary nomenclature adopted in thin
plate theory and to avoid the introduction of an unconventional notation
for the constitutive coefficients, we now reverse the y and z coordinate
axes. Thus, from this point on, the z axis will be oriented upward and the

plate extends from to 00 in the y direction. The position of the origin

-~

shown in Fig. 1 is unaltered.

For the case of cylindrical bending considered here, the governing
equations of CPT reduce to [2]

402

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

where w is the plate deflection and uo is the central plane displacement
in the x-direction. Also

and Qz, are the reduced stiffness coefficients for plane stress [4,5]. Letting
q =
qo sin px, we find that

where

Therefore, the only non-vanishing strain component E,x is given by

and from the layer constitutive relations we can determine the in-plane
stresses as

According to the assumptions employed in CPT, the stress components Œz

and T.~z, as determined from the constitutive relations, would vanish,
hence we compute these stresses by use of the equilibrium equations

which lead to the expressions

where H, and L, are constants to be determined such that the boundary

403

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

conditions for Œz and T xz on z =± 2 and
the stress continuity conditions at
the interfaces are satisfied. Although there are 2(m + 1) boundary and
interface continuity conditions for Πand Txz, only 2m of these are inde-
pendent owing to the approximations employed in CPT, thus these condi-
tions allow for the computation of the 2m constants H, and L,.

DISCUSSION
In order to compare the CPT results to the corresponding exact solu-
tions, we consider layers of square symmetric unidirectional fibrous
composite material possessing the following stiffness properties, which
simulate a high modulus graphite/epoxy composite,

where L signifiesthe direction parallel to the fibers, T the transverse

direction, and vLT is the Poisson ratio measuring strain in the transverse
direction under uniaxial normal stress in the L direction. The highly
anisotropic nature of this material represents a severe test for CPT.
Three separate geometrical configurations are considered, namely:
(1) a unidirectional plate with the fibers oriented in the x-direction
(2) a bidirectional (coupled) laminate with the T and L directions
aligned parallel to x in the top and bottom layers, respectively, the
layers being of equal thickness
(3) symmetric 3-ply laminate with layers of equal thickness-the L
a

direction coincides with x in the outer layers, while T is parallel

to x in the central layer.

In each case we let n = 1 that q qo sin

so

will correspond to that discussed in connection with CPT.

=

7rX
and the notation employed

The functions of prime interest in the present case are the longitudinal
stress Œx, transverse stresses Txz and Œz, and plate deflection w (the stress
component Œy is very small in the examples considered here). The elas-
ticity and CPT solutions for these functions are illustrated in Figs. 2-4.
As the two solutions merge for sufficiently thin plates, the trend is indicated
by plotting the respective results for various span-to-depth ratios (S)
whenever CPT exhibits a significant deviation from the exact solution for
a particular quantity. As a practical consideration, a span-to-depth ratio
of 4 is the lowest value selected for detailed analysis, although the elas-
ticity solution is valid for any value of S. The following normalized
quantities are defined in connection with Figs. 2-4,

404

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

In the various curves, the solid line indicates the elasticity solution while
the CPT result is denoted by a dashed line.
The relationship of the maximum deflection w to S is shown for the
three configurations in Figs. 2(a), 3(a), and 4(a). As would be expected,
CPT underestimates the plate deflection and gives a very poor estimate
for relatively low values of S, showing the necessity of incorporating the
influence of shear deformation in this range. At S = 20, for which CPT
would involve negligible error in isotropic media, the errors in cases
(1) and (3) are approximately 20%, the error growing significantly with
lower values of S. For higher values of S, the exact solution approaches
the CPT result asymptotically.
The distribution of 0--x given by the two theories is represented in
Figs. 2(b), 3(b), and 4(b) for S 4. For cases (1) and (3) the results for
=

S =
10 are displayed in Figs. 2(d) and 4(d) as an indication of the degree
in which the CPT result approaches the exact solution. CPT is in sub-
stantial agreement with elasticity theory in case (2) for S = 10. All of the
(rx curves are in accord with the fact that the normal stresses predicted by
CPT and elasticity theory on any plane x constant are equipollent (as
=

are the two solutions for Txz) in these problems.

The distribution of Txz is shown in Figs. 2(c), 3(c), and 4(c) for S = 4.
The only significant difference between the two solutions occurs in case
(3). Hence, in Fig. 4(e), we present the case for S 10, where CPT is in =

good agreement with the exact solution. Similar curves for o-x are given
in Figs. 3(d) and 4(f). For the problems studied here, the CPT approxima-
tion for these transverse stresses, which are responsible for delamination
failures, converges to the exact solution much more rapidly than its pre-
diction for plate deflection w.
It is also of interest to study the validity of the Kirchhoff hypothesis,
in particular for cases (2) and (3), where the body is laminated. The nor-
malized horizontal displacement at the end x 0 is represented in Figs. =

3(e) and 4(g) for S 4 and compared to the corresponding CPT calcula-
=

tion. The response in case (3), the 3-ply laminate, exhibits a very striking
departure from the prediction of CPT. Figure 4(g) indicates that the as-
sumption of linear displacement within each layer may be reasonable,

405

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

 Figure 2. Comparison of elasticity and CPT solutions for unidirectional plate, Case (1).

406

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

Figure 3. Comparison of elasticity and CPT solutions for two-ply laminate, Case (2).

407

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

 Figure 4. Comparison of elasticity and CPT solutions for three-ply laminate, Case (3).

408

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

 409

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

however the deformed configuration of the original normal cannot be
described by the Kirchhoff hypothesis. The straightening of the deformed
normal with increasing S is illustrated in Figs. 4(h) and 4(i), where S 6 =

and 10, respectively.

CONCLUSIONS
In conclusion, we have presented an approach to define the elasticity
solutions for laminates consisting of arbitrary numbers of orthotropic or
isotropic layers in cylindrical bending. While we have limited our dis-
cussion to a particular loading condition, more general loadings can be
treated by Fourier analysis. Since the solutions are exact within the as-
sumptions of linear elasticity, they are free from the simplifying assump-
tions imposed by CPT. Thus there need be no distinction between
thick and thin plates and shear deformation is automatically taken account
of. The only thickness effect involved in practice arises if the end condi-
tions (at x = 0, t) are such that the stresses on these planes are distributed
in a different manner than given in our solution. As a practical considera-
tion, however, the detailed distribution of these stresses in an actual
structural component would not appear to be known in most cases.
The type of analysis carried out here can also be employed to study
the response of sandwich plates with orthotropic face sheets and may
shed light on the assumptions made in the classical analysis of such
structural elements. In fact, one would anticipate the response of a sand-
wich plate to be quite similar to that discussed in case (3), where the stiff-
ness of the central layer in the longitudinal direction is quite small. The

treatment presented here can also be applied in the analysis of typical

composite flexure specimens and are in essential agreement with the con-
clusions of [6], which were derived in a more elementary manner.
We have seen that the CPT solution for stresses and displacements
converges to the exact solution with increasing span-to-depth ratio in all
examples treated here, although the computed CPT stresses converge
more rapidly than the displacement functions. Based on the evidence pre-
sented here, the use of CPT in the elastic design of very thin (high S)
bodies, such as composite skins in aircraft structures, appears adequate.
It should be noted that the stress resultants Nx, Qx, and moment
resultant Mx in the examples considered here are independent of the
assumptions employed in the derivation of CPT, i.e., they can be com-
puted directly from the plate equilibrium equations and boundary con-
ditions. Therefore, it will be informative to consider certain problems
relating to plates with finite dimensions in the y-direction to ascertain the
influence of the CPT assumptions on stress and moment resultants, and
in turn, on the internal stress fields. It is also important to include layers in

410

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014

which the shear coupling coefficients are non-zero. Such a catalogue of
exact solutions are necessary in order to rigorously describe the range of
applicability of CPT. They can also give insight into the basic assump-
tions required in the formulation of more general theories of flat plates,
and will be the subject of future works in this area.

ACKNOWLEDGMENT
The author wishes to express his appreciation to Lt. S. O. Fought
(ASNCD) for the computer analysis required in the numerical solutions.

REFERENCES
1. E. Reissner and Y. Stavsky, "Bending and Stretching of Certain Types of Heterogeneous
Aeolotropic Elastic Plates,"
. Appl. Mech., Vol. 28 (1961), p. 402.
J
2. S. B. Dong, K. S. Pister, and R. L. Taylor, "On the Theory of Laminated Anisotropic Shells
and Plates," J. Aero. Sci., Vol. 28 (1962), p. 969.
3. J. M. Whitney, "Effect of Transverse Shear Deformation on the Bending of Laminated
J. Composite Materials, Vol. 3 (1969), p. 000
Plates,"
4. S. W. Tsai, "Introduction to Mechanics of Composite Materials, Part II—Theoretical
Aspects," Air Force Materials Laboratory Report, AFML-TR-66-149 (1966).
5. S. W. Tsai and N. J. Pagano, "Invariant Properties of Composite Materials," in Composite
Materials Workshop, S. W. Tsai, J. C. Halpin, and N. J. Pagano, eds., Technomic Publish-
ing Company, Inc., Stamford, Conn. (1967), p. 233.
6. N. J. Pagano, "Analysis of the Flexure Test of Bidirectional Composites," J. Composite
Materials, Vol. 1 (1967), p. 336.

411

Downloaded from jcm.sagepub.com at St Petersburg State University on February 11, 2014