Composites Science and Technology 61 (2001) 1271–1283

www.elsevier.com/locate/compscitech

A new layerwise trigonometric shear deformation theory

for two-layered cross-ply beams
Rameshchandra P. Shimpi *, Yuwaraj M. Ghugal
Aerospace Engineering Department, Indian Institute of Technology Bombay, Powai, Mumbai, 400 076, India

Received 9 February 2000; received in revised form 5 December 2000; accepted 13 February 2001

Abstract
A new layerwise trigonometric shear deformation theory for the analysis of two-layered cross-ply laminated beams is presented.
The number of primary variables in this theory is even less than that of first-order shear deformation theory, and moreover, it
obviates the need for a shear correction factor. The sinusoidal function in terms of thickness coordinate is used in the displacement
field to account for shear deformation. The novel feature of the theory is that the transverse shear stress can be obtained directly
from the use of constitutive relationships, satisfying the shear-stress-free boundary conditions at top and bottom of the beam and
satisfying continuity of shear stress at the interface. The principle of virtual work is used to obtain the governing equations and
boundary conditions of the theory. The effectiveness of the theory is demonstrated by applying it to a two-layered cross-ply lami-
nated beam. # 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Shear deformation; Laminated thick beam; Transverse shear stress; Interface shear continuity; Cross-ply beam

1. Introduction Bresse [1], Rayleigh [2], and Timoshenko [3] were the
pioneer investigators who included refined effects such as
The use of fiber-reinforced composite laminates has the shear deformaton and rotatory inertia in the beam the-
greatly increased in weight sensitive applications such as ory. Timoshenko showed that the effect of transverse shear
aerospace and automotive structures because of their high is much greater than that of rotatory inertia on the response
specific strength and high specific stiffness. The increased of transverse vibration of prismatic beams. This theory is
use of laminated beams in various structures has stimu- now widely referred to as the Timoshenko beam theory in
lated considerable interest in their accurate analysis. On the literature. In this theory, transverse shear strain dis-
account of their low ratio of transverse shear modulus to tribution is assumed to be constant through the beam
the in-plane modulus, shear deformation effects are more thickness and, thus, requires a shear correction factor to
pronounced in the composite beams subjected to trans- appropriately represent the strain energy of deformation.
verse loads. Kant and Manjunatha [4], Manjunatha and Kant [5],
It is well-known that the classical Euler–Bernoulli Maiti and Sinha [6] and Vinayak et al. [7] used the
theory of beam bending, also known as elementary the- equivalent single layer, displacement based, higher-order
ory of bending (ETB), disregards the effects of the shear shear deformation theories (HSDT) in the analysis of
deformation. The theory is suitable for slender beams symmetric and unsymmetric laminated beams and
but not for thick or deep beams since it is based on the employed the finite-element method as a solution tech-
assumption that the transverse normal to the neutral axis nique. These theories are the special cases of Lo et al. [8]
remains so during bending and after bending, implying higher-order theory.
that the transverse shear strain is zero. Since the theory Levy [9] and Stein [10] developed refined plate the-
neglects the transverse shear deformation, it leads to less ories expressing the displacement field in terms of
accurate results in the case of isotropic thick beams and trigonometric functions to represent the thickness effect
more so in the case of laminated composite thick beams, and approximated the shear stress distribution through
where shear effects are significant. the thickness.
Recently, Liu and Li [11] presented an overall com-
* Corresponding author. parison of laminate theories based on displacement
E-mail address: rpshimpi@aero.iitb.ac.in (R.P. Shimpi). hypothesis emphasizing the importance of layerwise
0266-3538/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PII: S0266-3538(01)00024-0
1272 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283

Nomenclature 
EQL
zx Non-dimensional transverse shear
stress obtained from the equilibrium
b Width of beam equations
D Flexural rigidity 
z Non-dimensional transverse normal
D
Modified flexural rigidity coefficient stress
as defined in Appendix
D1, D2, D3, D
3 Constants as defined in Appendix Superscripts
E (1), E (2) Young’s moduli of layer 1 and layer CR Constitutive relationships
2, respectively EQL Equilibrium equations
G(1), G(2) Shear moduli of layer 1 and layer 2,
respectively Acronyms
h Depth (i.e. thickness) of beam ESCBP Exact solution of cylindrical bending
L Span of the beam of plate
S Aspect ratio (i.e. ratio of span to CPT Classical plate theory
depth of beam) ETB Elementary theory of beam-bending
x, y, z Rectangular coordinates FSDT First-order shear deformation theory

Neutral axis coefficient as defined in FEM Finite element method
Appendix HSDT Higher-order shear deformation
l1, l2 Coefficients as defined in Appendix theory
u
Non-dimensional inplane displace- HOST Higher-order shear deformation
ment theory
w
Non-dimensional transverse displace- HST Higher-order shear deformation
ment theory

x Non-dimensional inplane stress LTSDT-I Layerwise trigonometric shear defor-

CR
zx Non-dimensional transverse shear mation theory, Model I
stress obtained from the constitutive LTSDT-II Layerwise trigonometric shear defor-
relationships mation theory, Model II

theories and also presented a series of quasi-layerwise ories is scarce. These facts have been commented upon
theories. by Vinson and Chou [17] and Icardi [18]. Thus, there is
Lu and Liu [12] developed an interlaminar shear stress a need for a new layerwise refined shear deformation
continuity theory for composite laminates by using Her- theory with a minimum number of displacement vari-
mite cubic shape functions. The theory is layer dependent ables, for shear-deformable laminated beams.
and the number of degrees of freedom involved is very More recently, Shimpi and Ghugal [19] developed a sim-
high and hence it is computationally complicated. Results ple layerwise trigonometric shear deformation theory
of single-layered, two-layered and three-layered cross- (designated here as LTSDT-I) for flexural analysis of cross-
ply laminates for cylindrical bending were presented. ply laminated beams. However, the displacement model of
To improve the accuracy of the transverse stress pre- the theorysuffered fromthe defectinthattherewas anunba-
diction, layerwise higher-order theories based on lancedsmallresultantforcealongthelengthwisedirectionof
assumed displacements for individual layers, have been the beam. Removing this deficiency, an improved layerwise
proved to be very promising techniques in the flexural displacementmodelis presented in this paper and will be
analysis of thick laminates. Such theories were developed referred to as LTSDT-II.
and used by Lu and Liu [12], Li and Liu [13], Ambart- The constitutive relationships in respect of transverse
sumyan [14], Reddy [15] and Reddy and Robbins [16]. shear stress and shear strain in each layer are satisfied
A study of the literature indicates that, most of the and also interface shear stress continuity is satisfied in
layerwise theories have been developed for symmetric the proposed theory. In order to verify the accuracy of
cross-ply laminates subjected to cylindrical bending. the theory, it has been applied to two-layered cross-ply
Furthermore, higher order theories even with more than [90/0] laminated beam.
three displacement variables (e.g. there are three in the
case of FSDT) appear to be insufficient and inefficient
when the laminated beam is unsymmetric, if the unsym- 2. Theoretical formulation
metry in the lay-up is not properly accounted for in the
theory. It is further noted that the research work dealing The theoretical formulation of a cross-ply laminated
with unsymmetric laminated beams using refined the- beam based on certain kinematical and physical
 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283 1273

assumptions is presented. The variationally correct dw

forms of differential equations and boundary conditions, uð1Þ ðx; zÞ ¼ ðz 
hÞ
dx
based on the assumed displacement field are obtained 

 z=h 

using the principle of virtual work. þ h C1 þ C2 sin ðxÞ ð3Þ
The beam under consideration consists of two layers: 2 0:5 þ

layer 1 and layer 2. dw
Layer 1 (90
layer) occupies the region: uð2Þ ðx; zÞ ¼ ðz 
hÞ
dx


04x4L; b=24y4b=2; h=24z40 ð1Þ  z=h 

þ h C3 þ sin ðxÞ ð4Þ
2 0:5 

Layer 2 (0
layer) occupies the region:
wðxÞ ¼ wðxÞ ð5Þ
04x4L; b=24y4b=2; 04z4h=2 ð2Þ
Here u(1) and u(2) are the inplane displacement com-
where x, y, z are Cartesian coordinates, L is the length, ponents in the x direction, superscripts 1 and 2 refer to
b is the width and h is the total depth of beam. The layer 1 and layer 2, w(x) is the transverse displacement
beam is subjected to transverse load of intensity q(x) per in the z direction and C1, C2, C3 and
are the constants
unit length of the beam. The beam can have any mean- (given in the Appendix). The function (x) is a rotation
ingful boundary conditions. function or the warping function of the cross-section of
the beam.
2.1. Assumptions made in the theoretical formulation
2.3. Superiority of the present theory
1. The axial displacement consists of two parts:
The present theory is a displacement-based layerwise
theory, and layerwise higher-order or refined shear-
(a) displacement which is analogous to that given
deformation theories are known to be successful tech-
by elementary theory of bending;
niques for improving the accuracy of displacement and
(b) displacement due to shear deformation, which
stresses [11].
is assumed to be sinusoidal in nature with respect
The kinematics of the present theory is much richer
to thickness coordinate, such that maximum shear
than those of the higher order shear deformation the-
stress occurs at neutral axis as predicted by the
ories available in the literature, because if the trigono-
elementary theory of bending of beam.
metric term (involving thickness coordinate z) is expanded
in power series, the kinematics of higher order theories
2. Displacement u is such that the resultant of
(which are usually obtained by power series in thickness
inplane stress,  x, acting over the cross-section
coordinate z) are implicitly taken into account to good
should result in only bending moment and should
deal of extent.
not result in force in the x direction.
Also, it needs to be noted that every additional power
3. The transverse displacement is assumed to be a
of thickness coordinate in the displacement field of
function of longitudinal length coordinate of beam.
other higher-order theories of Lo et al. [8] type not only
4. The displacements are small compared to beam
introduces additional unknown variables in those the-
thickness.
ories but these variables are also difficult to interpret
5. The layers are perfectly bonded to each other.
physically [16]. Thus use of the sine term in the thick-
6. The stacking sequence of layers is such that there
ness coordinate (in the kinematics) enhances the rich-
is no bending-twisting coupling.
ness of the theory, and also results in the reduction of
7. The body forces are ignored in the analysis. (The
the number of unknown variables as compared to other
body forces can be effectively taken into account
theories [5,7] without loss of physics of the problem in
by adding them to the external forces.)
modelling.
8. One-dimensional constitutive laws are assumed for
The number of primary variables in this theory is even
each layer.
less than that of first-order shear-deformation theory,
9. The beam is subjected to lateral load only.
and moreover, it obviates the need of a shear correction
factor. Thus, the displacement field chosen is superior to
2.2. The displacement field those of others.

Based on the before mentioned assumptions, the dis- 2.3.1. Strains

placement field of the present layerwise beam theory is Normal and transverse shear strains for layers 1
given as below: and 2
1274 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283

@uð1Þ d2 w (i.e. w and ), we obtain the governing equations and

ðx1Þ ¼ ¼ ðz 
hÞ 2 the associated boundary conditions. The governing
@x

dx 
ð6Þ equations are as follows:
 z=h 
d
þ h C1 þ C2 sin
2 0:5 þ
dx d4 w d3  q
ð2 Þ 2  D 1  ¼0 ð15Þ
@u dw dx4 dx3 D
ðx2Þ ¼ ¼ ðz 
hÞ 2
@x

dx
ð7Þ
 z=h 
d d3 w d2 
þ h C3 þ sin  D þ D3  ¼ 0 ð16Þ
2 0:5 
dx dx3
2
dx2


ð1Þ @uð1Þ dw C2  z=h 

zx ¼ þ ¼ cos  ð8Þ
@z dx 1 þ 2
2 0:5 þ
The associated boundary conditions obtained are of

 the following form:
ð2Þ @uð2Þ dw   z=h 

zx ¼ þ ¼ cos  ð9Þ
@z dx 1  2
2 0:5 
d3 w d2 
3
 D1 2 ¼ 0 or w is prescribed ð17Þ
dx dx
2.3.2. Stresses
According to assumption (8), one-dimensional con- d2 w d dw
stitutive laws are used to obtain the normal bending and 2
 D1 ¼ 0 or is prescribed ð18Þ
dx dx dx
transverse shear stresses for layers 1 and 2.
 d2 w d
d2 w  D2 ¼ 0 or  is prescribed ð19Þ
xð1Þ ¼ Eð1Þ ðx1Þ
¼E ð1Þ
 ðz 
hÞ 2 dx 2 dx
dx

  ð10Þ
 z=h 
d
þ h C1 þ C2 sin where D, D1, D2, D3 are the constants as defined in
2 0:5 þ
dx
 Appendix.
d2 w Thus the variationally consistent governing differ-
xð2Þ ¼ Eð2Þ ðx2Þ ¼ Eð2Þ  ðz 
hÞ 2
dx ential equations and boundary conditions are obtained.

  ð11Þ
The flexural behaviour of beam is given by the solution
 z=h 
d
þ h C3 þ sin of these equations and simultaneously satisfaction of
2 0:5 
dx
the associated boundary conditions.


ð1Þ G ð1 Þ  C 2  z=h 

zx ¼ Gð1Þ zx
ð1 Þ
¼ cos  ð12Þ
1 þ 2
2 0:5 þ

3. Illustrative example


ð2Þ Gð2Þ   z=h 

zx ¼ Gð2Þ zx
ð2 Þ
¼ cos  ð13Þ A simply supported two-layered [90/0] composite
1  2
2 0:5 
beam, wherein layers 1 and 2 occupy the regions given
by expressions (1) and (2), respectively, is considered for
2.4. Governing equations and boundary conditions detailed numerical study. The beam is subjected to sin-
gle sinusoidal load q(x)=qo sin (x=L), acting in the z
Using the expressions (6)–(13) for strains and stresses direction, where qo is magnitude of the sinusoidal load-
and principle of virtual work, variationally consistent ing at midspan. This is the same problem, which was
differential equations and boundary conditions for the also considered by Pagano [20]. The material of the
beam under consideration are obtained. The principle of beam layers is a carbon-fibre/epoxy uni-directional
virtual work when applied to the beam leads to: composite. The following has been assumed:
ð x¼L ð z¼0
 ð1 Þ ð1 Þ ð1Þ ð1Þ
Eð2Þ G ð1 Þ G ð2 Þ
b x  x þ zx zx dx dz ¼ 25; ¼ 0:20; ¼ 0:02
x¼0 z¼h=2 Eð1Þ Eð1Þ Eð2Þ
ð x¼L ð z¼þh=2

þb xð2Þ  ðx2Þ þ zx
ð2 Þ ð2 Þ
zx dx dz ð14Þ Superscripts (1) and (2) refer to layers 1 and 2,
x¼0 z¼0
ð x¼L respectively.
 q w dx ¼ 0
x¼0 3.1. The solution scheme
where the symbol  denotes the variational operator.
Employing Green’s theorem in Eq. (14) successively The following is the form of the solution for w(x) and
and collecting the coefficients of the primary variables (x) that satisfies boundary conditions perfectly:
 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283 1275


wðxÞ ¼ w1 sin
x q0 S2 Eð1Þ
z 
L xð1Þ ¼ 
l1  l2
b 2 D
Eð2Þ h

 ð24Þ
x  z=h 
 x
ðxÞ ¼ 1 cos C1 þ C2 sin sin
L 2 0:5 þ
L

q0 S2
z 
where w1 and 1 are constants. xð2Þ ¼ 
l1  l2
Using the assumed solution scheme and the governing b 2 D
h

 ð25Þ
equations, the unknown functions w(x) and (x) can be  z=h 
 x
C3 þ sin sin
found. Expressions for transverse displacement w and 2 0:5 
L
function  are given as follows:

q0 hS4 l1 x 3.4. Expressions for transverse shear stresses derived

w¼ sin ð20Þ
Eð2Þ b4 D
L from the constitutive relationships, tCR
zx

q0 S 3 l2 x It may be noted that it is possible to obtain transverse

¼ cos ð21Þ shear stress,  zx, by using either the constitutive rela-
Eð2Þ b3 D
L CR
tionships or the equilibrium equations. Notation zx
denotes zx obtained by using constitutive relationships.
where b, D
, E(2), h, S, l1 and l2 are defined in the
CR
 ð1Þ CR
Nomenclature and/or Appendix. Substituting expres- zx ¼ zx ifh=24z40
sions for w and  given by Eqs. (20) and (21) in Eqs.  ð2Þ CR
(3)–(5), final expressions for in-plane displacements can ¼ zx if 04z4h=2
be obtained.  ð1Þ CR  ð2Þ CR
where zx and zx are as follows:
3.2. Expressions for inplane displacement, u



ð1Þ CR q0 S3 Gð1Þ
zx ¼
u ¼ uð1Þ ifh=24z40 b 3 D
Eð2Þ
ð2Þ



¼u if 04z4h=2 C2 l2   z=h 
x
 cos cos ð26Þ
1 þ 2
2 0:5 þ
L
where u(1) and u(2) are as follows:


q0 h S3 
ð2Þ CR q0 S3 Gð2Þ
u ð1 Þ
¼ zx ¼
Eð2Þ b 3 D
 b 3 D
Eð2Þ


 


z  z=h 
 x l2   z=h 
x
 
l1 þ C1 þ C2 sin l2 cos  cos cos ð27Þ
h 2 0:5 þ
L 1 þ 2
2 0:5 
L
ð22Þ
q0 h S3
u ð2 Þ ¼
Eð2Þ b 3 D
 3.5. Expressions for transverse shear stress, tEQL
zx , and


 
z  z=h 
 x transverse normal stress,  z, obtained from the
 
l1 þ C3 þ sin l2 cos equilibrium equations
h 2 0:5 
L
ð23Þ
Substituting expressions for w and  given by Eqs. The following equilibrium equations of two-dimen-
(20) and (21) in Eqs. (10)–(13), the final expressions for sional elasticity, ignoring body forces, are used to
in-plane and transverse shear stresses can be obtained. obtain transverse shear and transverse normal stresses.

@x @zx
þ ¼0 ð28Þ
3.3. Expressions for inplane stress, sx @x @z
@zx @z
þ ¼0 ð29Þ
@x @z
x ¼ xð1Þ ifh=24z40
To obtain  zx, we integrate Eq. (28) layerwise, w.r.t
¼ xð2Þ if 04z4h=2
the thickness coordinate z and impose the following
where xð1Þ and xð2Þ are as follows: boundary condition at bottom of the beam
1276 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283

 ð2 Þ
 q0 S
zx ¼0 ð30Þ ð2Þ EQL
z¼h=2 zx ¼


8 
 b
 D
 9
>
> z 1 z 2 1
>
>
and the stress-continuity condition >
<
h  2 h þ 8  2 l1 >
= x



 cos
  ð2Þ >
> z 1 1  2
 z=h 
> L
ð1 Þ
zx ¼ zx ð31Þ > þ C3  
: cos l2 >
>
;
z¼0 z¼0 h 2  2 0:5 

ð35Þ
at interface to get constants of integrations and to
maintain the continuity of transverse shear stress at
q0 Eð1Þ
layer interface. zð1Þ ¼ 

Eð2Þ
To obtain  z, we substitute the expressions obtained 8 
b D
  9
for shear stresses Eq. (29) and integrate layerwise, w.r.t >
> 1 z 3
z 2

z 1
z >
>
>
>    l1 >
>
the thickness coordinate z and impose the following >
> 6 h 2 h 2 h 8 h >
>
>
>

   >
>
>
> Eð 2Þ
1
C
z 2
z  >
>
boundary condition at bottom of the beam >
> 1 >
>
>
> þ  l 1  þ l 2 >
>
>
> Eð 1Þ 24 8 2 h h >
>
 >
>

 >
>
ð2 Þ >
< 2 >
= x
zz ¼0 ð32Þ 1 þ 2
 z=h 

z¼h=2 þC2 sin l2 sin
>  2 0:5 
> L
>
>

 >
>
>
> 2 >
>
and the stress-continuity condition >
> 1 þ 2
 
>
>
>
> C2 sin l2 >
>
>
>  2 0:5 þ
>
>
>
> " # >
>
  ð2 Þ >
ð2Þ 
>
2 n
o >
>
ð1 Þ >
> E C 1  2
 
>
>
zz z¼0
¼ zz z¼0
ð33Þ >
> 
3
þ 1 sin l >
>
: Eð 1Þ 8  2 0:5 

2 ;

ð36Þ
at interface to obtain constants of integrations and to
maintain the continuity of transverse normal stress at
layer interface. q0
zð2Þ ¼ 
Using the above procedure, expressions are obtained b D

8
  9
for transverse shear zx EQL
and transverse normal  z >
> 1 z 3

z 2 1
z
1 >
stresses in their following regions of through-thickness >
>    þ l1 >
>
>
>
> 6 h 2 h 8 h 8 24 >
>
variation: < C 
z2
z
1
> >
= x
3
(Notation zxEQL
denotes shear stress  zx as obtained by    þ l2 sin
>
> 2 h h 4 >
> L
using equilibrium equations). >
>
 2 
 >
>
>
> 1  2
 z=h 
>
>
>
: 1  sin l2 >
;
 ð1Þ EQL  2 0:5 

EQL
zx ¼ zx if h=24z40 ð37Þ
 ð2Þ EQL
¼ zx if 04z4h=2
z ¼ zð1Þ if h=24z40 Thus expressions for displacements and stresses have
been obtained in their explicit forms.
¼ zð2Þ if 04z4h=2

 ð1Þ EQL  ð2Þ EQL ð1Þ

where zx , zx , z and zð2Þ are as follows: 4. Numerical results


ð1Þ EQL q0 S Eð1Þ The results obtained for displacements and stresses at
zx ¼
b  D
Eð2Þ salient points are presented in Tables 1–6 and in Figs. 1–
8


 9
> z 1
z2 Eð2Þ 1
>
3 in non-dimensional parameters.
>
>
 þ  l >
>
>
> h 2 h E ð1Þ 8 2
1 >
>
>
> 

 >
> 4.1. Non-dimensional displacements and stresses
>
>
 >
>
>
> z 1 þ 2
 z=h 
>
>
>
< þ C 1  C 2 cos l 2>
=
h  2 0:5 þ
x The results for inplane displacement, transverse dis-


 cos
>
> 1 þ 2
 
>
> L placement, inplane and transverse stresses are presented
>
> þC2 cos l2 >
>
>
>  2 0:5 þ
>
> in the following non-dimensional form in this paper.
>
>


 > >
> >
>
>
> Eð2Þ C3 1  2
 
 >
>
>
: þ cos l 2 ;
Eð1Þ 2  2 0:5 
Eð1Þ bu 100Eð1Þ bh3 w bx
u
¼ ; w
¼ ; 
x ¼ ;
ð34Þ qo h qo L 4 qo
 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283 1277

bzx bz effect in beam theory. Results of present


zx ¼ ; 
z ¼ LTSDT-II model and HOSTB5 beam model of
qo qo
Kant and Manjunath [4] are comparable within
3.46%. The maximum value of this displacement
The percentage difference (% Diff.) in results obtained using LTSDT-II has been found to decrease by
by models of other researchers with respect to the cor- 0.43% as compared to that of LTSDT-I for the
responding results obtained by the present theory aspect ratio 4.
(LTSDT-II) is calculated as follows: (b) For the aspect ratio 10, the value of this dis-
placement obtained by LTSDT-I is found to
%Diff: ¼
increase by 0.11% when compared to LTSDT-II.
value obtained by other modelcorresponding value by LTSDT-II
value by LTSDT-II
 100 The values by Lu and Liu [12] and Pagano [20]
are found to decrease by 2.58% for aspect ratio
10.
(c) The ETB underestimates the value by 23.28% as
5. Discussion compared to that of LTSDT for the aspect ratio
4. For aspect ratio 10, ETB underestimates the
For the problem under consideration, it may be noted result by 4.66% compared to LTSDT-II. In gen-
that the exact results are not available in the literature eral, ETB underpredicts the inplane displace-
to the best of the authors’ knowledge. In the absence of ments as compared to those given by higher
exact results, researchers have compared their results order models.
with the exact solution for the cylindrical bending of
plate (ESCBP) obtained by Pagano [20] for composite 5.2. Transverse displacement
laminates.
Kant and Manjunatha [4], Manjunatha and Kant [5], The variation of maximum nondimensionalised
Maiti and Sinha [6] and Vinayak et al. [7] used the transverse displacement with various aspect ratios, is
equivalent single layer (ESL) theories in the analysis of shown in Fig. 1. The comparison of maximum trans-
laminated beams and provided the finite element solu- verse displacements for the aspect ratios of 4 and 10 is
tions. presented in Tables 2 and 3. It should be noted that the
It is well known that the cylindrical-plate-bending deflection of cylindrical bending of a plate strip is
problem, though near to a beam-bending problem, is a somewhat smaller than the corresponding beam deflec-
different problem altogether, in which the Poisson’s tion due to the plate action in plate strip.
effect figures. In this paper also, results obtained by Lu
and Liu [12] and Pagano [20] are quoted for the sake of (a) It is seen from the comparison that the results for
the record. Also, results using elementary theory of transverse displacement of present model are
beam bending are quoted. comparable with the results of higher order
In LTSDT-II, stress/strain relationships are satisfied
in both the layers and the interface continuity for
transverse shear stress is maintained, also the closed
form solution has been obtained for the problem under
consideration. Because of these reasons, it is believed
that the LTSDT-II results are superior.

5.1. In-plane displacement

The comparison of results of maximum non-

dimensionalised inplane displacement and the percen-
tage difference in results of other models with respect to
results of present model is shown in Table 1 for the
aspect ratios 4 and 10. These displacements are calcu-
lated at left end of the beam (i.e. at x=0).

(a) Results of in-plane displacement of other higher-

order models and the present model are on the
high side compared to the exact results of
cylindrical bending of plate for the aspect ratio 4 Fig. 1. Variation of maximum transverse deflection w
at x=0.5 L with
as a consequence of the neglect of the Poisson’s aspect ratio S.
1278 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283

models of Lu and Liu [12], Vinayak et al. [7] and placement by 11.64% as compared to that of the
the finite element solution of first order shear present model. It is seen from the Fig. 2 that the
deformation theory by Maiti and Sinha [6]. LTSDT-II converges to the ETB results asymp-
However, higher order model of Maiti and Sinha totically as the aspect ratio increases.
underestimates the deflection by 25.49% as
compared to that given by the present model for 5.3. In-plane stress
the aspect ratio 4 and this may be due to the lack
of C1 continuity in the finite element formula- This component of stress is directly evaluated using
tion. the constitutive law and strain displacement relations.
(b) For aspect ratio 10 results of LTSDT-II model, The comparison of maximum nondimensionalised
HOSTB5 model of Manjunatha and Kant [5], Lu inplane stresses obtained by the present theory and
and Liu [12], model and Pagano [20] are in close other refined theories is presented in the Tables 2 and 3
agreement with each other. No significant differ- for the aspect ratios of 4 and 10 respectively. Fig. 2
ence is observed between the value of transverse shows through the thickness variation of inplane stress
displacement obtained by LTSDT-II and for the aspect ratio 4, at midspan of beam.
LTSDT-I for the aspect ratio 4 and 10.
(c) The ETB underestimates the results for trans- (a) It can be seen that the maximum value of inplane
verse displacement by 44.60% as compared to stress obtained by LTSDT-II is comparable with
result of LTSDT-II for the aspect ratio 4; and for the values of Manjunatha and Kant [5], Vinayak
the aspect ratio 10, ETB underestimates the dis- et al. [7], Lu and Liu [12] and Pagano [20] for the

Table 1
Comparison of non-dimensional in-plane displacement (u
)

For S=4 when x=0 and z= h=2 For S=10 when x=0 and z= h=2
a
Source Model u
Percentage difference u
Percentage differencea

Present LTSDT-II 5.0353 0.00 63.3050 0.00

Shimpi and Ghugal [19] LTSDT-I 5.0571 0.43 63.3790 0.11
Kant and Manjunatha [4] HOSTB5 4.8611 3.46 – –
Lu and Liu [12] HSDT 4.7143 6.37 61.6667 2.58
Pagano [20] ESCBP 4.5000 10.63 61.6667 2.58
Euler–Bernoulli ETB 3.8627 23.28 60.3540 4.66
a
Percentage difference quoted is with respect to the corresponding value obtained using LTSDT-II.

Table 2
Comparison of non-dimensional transverse displacement (w
) and in-plane stress (
x )

For S=4

Source Model when x=0.5 L when x=0.5 L and z= h=2 when x=0.5 L and z=h=2
a a
w
% Diff. 
x % Diff. 
x % Diff.a

Present LTSDT-II 4.7437 0.00 3.9547 0.00 30.5650 0.00

Shimpi and Ghugal [19] LTSDT-I 4.7431 0.01 3.9650 0.26 30.2980 0.87
Manjunatha and Kant [5] HOSTB5 4.2828 9.71 3.7490 5.20 27.0500 11.50
FEM
Maiti and Sinha [6] HST 3.5346 25.49 2.3599 40.33 25.7834 15.64
FEM
Maiti and Sinha [6] FSDT 4.7898 0.97 3.1514 20.31 29.1144 4.75
FEM
Vinayak et al. [7] HSDT 4.5619 3.83 4.0000 1.15 27.0000 11.66
FEM
Lu and Liu [12] HSDT 4.7773 0.71 3.5714 9.69 30.0000 1.85
Pagano [20] ESCBP 4.3276 8.77 3.8359 3.00 30.0290 1.75
Euler and Bernoulli ETB 2.6281 44.60 3.0386 23.16 27.9540 8.54
a
Percentage difference quoted is with respect to the corresponding value obtained using LTSDT-II.
 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283 1279

Table 3
Comparison of non-dimensional transverse displacement (w
) and in-plane stress (
x }

For S=4

Source Model when x=0.5 L when x=0.5 L and z= h=2 when x=0.5 L and z=h=2

w
% Diff.a 
x % Diff.a 
x % Diff.a

Present LTSDT-II 2.9744 0.00 19.8880 0.00 177.140 0.00

Shimpi and Ghugal [19] LTSDT-I 2.9743 0.00 19.8999 0.06 176.870 0.15
Manjunatha and Kant [5] HOSTB5 2.8986 2.55 19.7100 0.90 173.000 2.34
FEM
Lu and Liu [12] HSDT 3.0000 0.86 20.0000 0.56 175.000 1.21
Pagano [20] ESCBP 2.9569 0.59 20.0000 0.56 175.000 1.21
Euler and Bernoulli ETB 2.6281 11.64 18.9910 4.51 174.710 1.37
a
Percentage difference quoted is with respect to the corresponding value obtained using LTSDT-II.

Fig. 2. Variation of inplane stress 
x through the thickness for aspect

ratio 4 at x=0.5 L.

Fig. 3. Variation of transverse shear stress 
zx through the thickness

aspect ratio of 4. Results of higher order model
aspect ratio 4 at x=0.0.
of Maiti and Sinha [6] in both the layers deviate
significantly (by 40.33 and 15.64%) from those
of the present model, however, their FSDT solu- inplane stress by 23.16% as compared to the
tion gives comparable value of this stress in the value obtained by the present model for the
0o layer. LTSDT-I has exhibited an increase in aspect ratio of 4 in the 90
layer. In the 0
layer,
value of this stress in the 90
layer by 0.26% and ETB underpredicts the value of inplane stress by
decrease in the value by 0.87% in 0
layer as 8.54% as compared to value of LTSDT-II. The
compared to LTSDT-II for the aspect ratio 4. values of inplane stress in 90 and 0
layers given
(b) For the aspect ratio of 10, the stresses given by by the present model are in close agreement with
LTSDT-II, higher order models and cylindrical those of Pagano [20].
bending of plate (Pagano [20]) are in good
agreement with each other with negligible varia- 5.4. Transverse shear stress
tion in the results. However, ETB underestimates
the value of inplane stress by 4.51% in the 90
The transverse shear stresses are obtained directly
layer and by 1.37% in the 0
layer as compared from the constitutive relationships (
CR
zx ) or, alter-
to the maximum value of inplane stress given by natively, from the equilibrium equations of two dimen-
present model. No significant difference is found sional elasticity (
EQL
zx ). The transverse shear stress
between the values of inplane stress for the satisfies the stress free boundary conditions on the top
aspect ratio 10 in both the layers. and bottom surfaces of the beam and continuity at layer
(c) It is observed from the Tables 2 and 3 that the interface when these stresses are obtained by both the
ETB underestimates the maximum value of above mentioned approaches.
1280 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283

The through thickness variation of shear stress for the 0.81%, however, the model of Maiti and Sinha
aspect ratio 4 is given in the Fig. 3, in which the varia- [6] underpredicts this value by 4.77% for the
tion obtained by using constitutive relations of the pre- aspect ratio 4. LTSDT-I has exhibited a decrease
sent theory is denoted by LTSDT-II- CR curve and the in this value by 2.1% as compared to that of
same obtained by using equilibrium equations is deno- LTSDT-II for the aspect ratio 4. The use of con-
ted by LTSDT-II- EQL curve. The comparison of stitutive relations showed the underestimation of
maximum nondimensionalised shear stresses for aspect shear stresses in 90
layer and overestimation of
ratios 4 and 10 is presented in the Tables 4 and 5 same in the 0
layer as compared to those obtained
respectively. by direct integration of equilibrium equations for
the aspect ratio 4.
(a) The maximum transverse shear stress obtained (b) It may be seen from the Fig. 3 that the through
by HOSTB5 model of Manjunatha and Kant [5] thickness distribution of this stress obtained by
via constitutive relations underpredicts the value use of constitutive relations is more closer to that
by 35.54% while that obtained by direct integra- of ETB in the 0
layer. The results of LTSDT-II
tion overestimates the value by 5.44% for the using constitutive relations are comparable with
aspect ratio 4. The model of Vinayak et al. [7] those of FSDT and ETB for the aspect 4. How-
overpredicts the value by 2.67%, and that is ever, ETB overestimates this value by 9.96%
given by Pagano [20] is comparable within when compared with that of LTSDT-II obtained

Table 4
Comparison of maximum transverse shear stress 
CR EQL
zx , 
zx and transverse normal stress (
z )

For S=4

Source Model when x=0.0 and z=
h when x=0.0 and z=
h when x=0.5 L and z=h=2


CR
zx % Diff.a 
EQL
zx % Diff.a 
z % Diff.a

Present LTSDT-II 2.9895 0.00 2.6784 0.00 1.00 0.00

Shimpi and Ghugal [19] LTSDT-I 2.9886 0.03 2.6222 2.10 0.98 2.00
Manjunatha and Kant [5] HOSTB5 1.9270 35.54 2.8240 5.44 1.00 0.00
FEM
Maiti and Sinha [6] HST – – 2.4252 9.45 – –
FEM
Maiti and Sinha [6] FSDT 2.8467 4.77 – – – –
FEM
Vinayak et al. [7] HSDT – – 2.7500 2.67 1.00 0.00
FEM
Lu and Liu [12] HSDT 2.7925 6.59 – – – –
Pagano [20] ESCBP – – 2.7000 0.81 1.00 0.00
Euler and Bernoulli ETB – – 2.9453 9.96 1.00 0.00
a
Percentage difference quoted is with respect to the corresponding value obtained using LTSDT-II.

Table 5
Comparison of non-dimensional transverse shear stress 
CR EQL
zx , 
zx and transverse normal stress (
z )

For S=4

Source Model when x=0.0 and z=
h when x=0.0 and z=
h when x=0.5 L and z=h=2


CR
zx % Diff. a

EQL
zx % Diff. a

z % Diff.a

Present LTSDT-II 7.6948 0.00 7.2638 0.00 1.00 0.00

Shimpi and Ghugal [19] LTSDT-I 7.6945 0.00 7.2406 0.32 0.99 1.00
Manjunatha and Kant [5] HOSTB5 4.9130 36.15 7.2820 0.25 1.00 0.00
FEM
Lu and Liu [12] HSDT 7.3000 5.13 – – – –
Pagano [20] ESCBP – – 7.3000 0.49 1.00 0.00
Euler and Bernoulli ETB – – 7.3643 1.38 1.00 0.00
a
Percentage difference quoted is with respect to the corresponding value obtained using LTSDT-II.
 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283 1281

Table 6
Comparison of variation of non-dimensional transverse normal stress (
z ) through the thickness of beam (
z at x=0.5 L and S=4)

LTSDT-II LTSDT-I HOSTB5 ETB ESCBP

z/h Present Shimpi and Ghugal [19] Manjunatha and Kant [5] Pagano [20]

0.5 1.0000 0.9803 1.0000 1.0000 1.0000

0.4 0.9886 0.9688 0.9869 0.9911 0.9737
0.3 0.9571 0.9373 0.9552 0.9660 0.9605
0.2 0.9097 0.8897 0.9087 0.9274 0.9211
0.1 0.8500 0.8298 0.8508 0.8777 0.8684
0.0 0.7815 0.7609 0.7848 0.8197 0.7894
0.1 0.6525 0.6393 0.6581 0.6951 0.6711
0.2 0.4574 0.4500 0.4578 0.4869 0.4605
0.3 0.2499 0.2467 0.2432 0.2590 0.2500
0.4 0.0768 0.0759 0.0716 0.0754 0.0789
0.5 0.0000 0.0000 0.0000 0.0000 0.0000

by equilibrium equations, for the aspect ratio 4.

(c) For aspect ratio 10, the model of Manjunatha for the aspect ratio of 4. An improved through the
and Kant [5], using constitutive relations, under- thickness variation of this stress is obtained by the pre-
estimates the maximum value of this stress by sent model as compared to that given by the LTSDT-I
36.15% and that of Lu and Liu [12] under- model. The results for this stress component obtained
predicts it by 5.13% as compared to that by the present model LTSDT-II are in good comparison
obtained by LTSDT-II. The value of this stress with those of HOSTB5 model of Manjunatha and Kant
in LTSDT-II, obtained by equilibrium equa- [5], ETB, and exact solution of cylindrical bending of
tions, is in excellent agreement with other solu- plate obtained by Pagano [20].
tions as can be seen from Table 5 for aspect ratio The above discussion in respect of the numerical
10. example, validates the efficacy of the LTSDT-II model.

This difference in transverse shear stresses obtained

by use of constitutive relations (CR) and equilibrium 6. Conclusions
equations (EQL) is well known in the bending analysis
of laminated beams/plates [5,21]. Since the displacement In this paper a new model has been presented for
solution can not give these stresses accurately when cross-ply [90/0] laminated composite beam. This model
constitutive relations are used, many authors [22] sug- is an improvement over the earlier model developed by
gested use of stress equilibrium equations of elasticity to the authors. The earlier model suffered from the defect
obtain the transverse shear and transverse normal in that there was an unbalanced small resultant force
stresses through the thickness of the laminated beams along the lengthwise direction of the beam. This defi-
more accurately, but their approach has been adversely ciency has been removed in the present model. The dis-
commented upon by Liu and Li [11], Bisegna and Sacco placement based model LTSDT-II presented in this
[23]. Furthermore, this approach is more tedious com- paper has following advantages:
pared to the approach using constitutive relations.
The discrepancy in the shear stresses obtained by 1. It is a displacement based layerwise model.
constitutive relations and by integration of equilibrium 2. The number of displacement variables is less than
equations is much less in the present case as compared that in first order shear deformation theory. It
to other attempts [5]. Furthermore, overall distribution contains only two variables.
of the shear stress obtained via constitutive relations is 3. Transverse shear stress satisfies continuity condi-
reasonably accurate. tions at layer interface and zero shear stress con-
The difference between results is due to the two dif- ditions at top and bottom surfaces of the beam.
ferent approaches used to evaluate the stresses. 4. The model does not require the use of shear cor-
rection factor.
5.5. Transverse normal stress 5. Constitutive relations are satisfied in both the lay-
ers in respect of inplane stress and transverse shear
The transverse normal stress is obtained by direct stress.
integration of equilibrium equations of two dimensional 6. It is possible to get reasonable accuracy of trans-
elasticity. The comparison of through thickness varia- verse shear stress even when the shear stress is
tion of this stress component is presented in the Table 6 obtained by the use of constitutive relations.
1282 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283

7. The governing differential equations and the (c) Constants C1, C2, C3 are as follows:
boundary conditions are variationally consistent.
8. The use of the present model gives accurate results,

Eð2Þ
as has been seen from the numerical example stu- C1 ¼
Eð1Þ þ Eð2Þ
died. 8

 9
> 
 
>
>
> sin  C 2 sin >
>
>
> 1  2
1 þ 2
>
>
<
ð1Þ 

>
>
=
Appendix E 1 þ 2


 þ2 C2 ð2Þ cos
>
>
E   1 þ 2
>
>
>
>

 >
>
(a) Layer 1 integration constants A1, A2, A3, A4 are as >
> 2 1  2
>
>
: cos ;
follows:  1  2

 


3 ð1Þ 1

2 ð2 Þcos
A1 ¼ b h E þ þ G 0:5 þ

24 4 2 C 2 ¼ ð1 Þ
1  2

G 0:5 


cos
8
 9 1þ2

>
> 1
>
>
>
> C1 þ >
>
9
>
> 8 2 > 8


>
<
2 
 >
>
= > 
>
1 þ 2

>
> sin  C 2 sin >
>
3
A2 ¼ b h E ð1Þ
þC2 1 þ sin >
> 1
 2

1 þ 2
>
>
> >
 >
<  >
=
>
>
>
 
1 þ2
> >
> Eð1 Þ 1 þ 2


>
> 1 þ 2

>
> C3 ¼ þ2 C 2 cos
> > Eð1Þ þ Eð2Þ >
: þ
C 2

cos
1 þ 2

; >
>
ð2Þ 

>
  1
þ 2
 > >
>
>
> E 1  2

> >
>
: 2 cos ;
E ð 1 Þ  1  2

8 2

9
>
> C1 1 þ 2

> >
<  2C1 C2 cos =
3 ð1Þ 2 

1 þ 2

A3 ¼ b h E
> C 2
1 þ 2

> (d) Constants D, D
, D1, D2, D3, D
3 are as follows:
>
:þ 1þ sin >
;
4  0:5 þ

D ¼ ðA1 þ A2 Þ ¼ D
Eð2Þ b h3




b h Gð1Þ C2  2 1 þ 2


ð1Þ 


A4 ¼ 1þ sin E 1

2 1

2
4 1 þ 2
 0:5 þ


D¼ þ þ þ  þ
Eð2Þ 24 4 2 24 4 2



A2 þ B2 A3 þ B3
(b) Layer 2 integration constants B1, B2, B3, B4 are as D1 ¼ ; D2 ¼ ;
A1 þ B1 A2 þ B2
follows:

A4 þ B4
  D3 ¼
1

2 A2 þ B2
B1 ¼ b h3 Eð2Þ  þ
24 4 2
D
3 ¼ D3 h2
8
 9
>
> 1
1 >
>
>
> C3  >
> (e) Constants S,
, l1, l2, are as follows:
>
> 8 2 >
>
>
<
2 h
i >
=
3 ð2Þ 1  2


B2 ¼ b h E þ 1 þ sin
> 1  2
>
>
>
>
 

>
>
>
S ¼ L=h
>
> 1  2

>
>
>
: 
cos >
;

 1  2 1 Eð2Þ  Eð1Þ

¼
8 2

9 4 Eð2Þ þ Eð1Þ
>
> C3 1  2

 >>
< þ 2 C3 cos = h i
3 ð2Þ 2 
  1  2
 l1 ¼ 1 þ D1 = D2  D1 þ D
3 2
B3 ¼ b h E

>
> 1 1  2

>
>
:þ 1  sin ;
4  0:5 
h i
l2 ¼ 1= D2  D1 þ D
3 2



b h Gð2Þ
 2 1  2


B4 ¼ 1 sin
4 1  2
 0:5 
¼ S=
 R.P. Shimpi, Y.M. Ghugal / Composites Science and Technology 61 (2001) 1271–1283 1283

References [12] Lu X, Liu D. An interlaminar shear stress continuity theory for

both thin and thick composite laminates. ASME Trans Journal
of Applied Mechanics 1992;59:502–9.
[1] Bresse JAC. Cours de mecanique applique. Paris: Mallet-Bache- [13] Li X, Liu D. Zig-zag theory for composite laminates. AIAA
lier, 1859. Journal 1994;33:1163–5.
[2] Rayleigh JWSLord. The theory of sound. London: Macmillan [14] Ambartsumyan SA. Theory of anisotropic plates., Technomic
Publishers, 1877. Publication, Stamford, CT, 1969 (Translated from Russian edi-
[3] Timoshenko SP. On the correction for shear of the differential tion by T. Cheron and edited by J. E. Ashton).
equation for transverse vibrations of prismatic bars. Philosophi- [15] Reddy JN. A generalisation of two-dimensional theories of lami-
cal Magazine, Series 6 1921;41:742–6. nated composite plates. Communications in Applied Numerical
[4] Kant T, Manjunatha BS. Refined theories for composite and Methods 1987;3:173–80.
sandwich beams with C0 finite elements. Computers and Struc- [16] Reddy JN, Robbins DH Jr. Theories and computational models
tures 1989;33:755–64. for composite laminates. Applied Mechanics Reviews 1994;
[5] Manjunatha BS, Kant T. Different numerical techniques for the 47:147–69.
estimation of multiaxial stresses in symmetric / unsymmetric [17] Vinson JR, Chou TW. Composite materials and their use in
composite and sandwich beams with refined theories. Journal of structures. London: Applied Science, 1975 (Chapter 6, pp. 234–
Reinforced Plastics and Composites 1993;12:2–37. 235).
[6] Maiti DK, Sinha PK. Bending and free vibration analysis of [18] Icardi U. Eight-noded zig-zag element for deflection and stress
shear deformable laminated composite beams by finite element analysis of plates with general lay-up. Composites Part B
method. Composite Structures 1994;29:421–31. 1998;29B:411–25.
[7] Vinayak RU, Prathap G, Naganarayana BP. Beam elements [19] Shimpi RP, Ghugal YM. A layerwise trigonometric shear defor-
based on a higher order theory — I: formulation and analysis of mation theory for two layered cross-ply laminated beams. Jour-
performance. Computers and Structures 1996;58:775–89. nal of Reinforced Plastics and Composites 1999;18:1516–42.
[8] Lo KH, Christensen RM, Wu EM. A higher order theory for [20] Pagano NJ. Exact solution for composite laminates in cylindrical
plate deformations, Part 2: laminated plates. ASME Trans Jour- bending. Journal of Composite Materials 1969;3:398–411.
nal of Applied Mechanics 1977;44:669–76. [21] Cho M, Parmerter RR. Efficient higher order composite plate
[9] Levy M. Memoire sur la theorie des plaques elastiques planes. theory for general lamination configurations. AIAA Journal
Journal des Mathematiques Pures et Appliquees 1877;30:219– 1993;31:1299–306.
306. [22] Lo KH, Christensen RM, Wu EM. Stress solution determination
[10] Stein M. Nonlinear theory for plates and shells including the for high order plate theory. International Journal of Solids and
effects of transverse shearing. AIAA Journal 1986;24:1537–44. Structures 1978;14:655–62.
[11] Liu D, Li X. An overall view of laminate theories based on dis- [23] Bisegna P, Sacco E. A layerwise laminate theory rationally
placement hypothesis. Journal of Composite Materials 1996; deduced from the three-dimensional elasticity. Trans ASME,
30:1539–61. Journal of Applied Mechanics 1997;64:538–45.