COMPDYN 2011

III ECCOMAS Thematic Conference on

Computational Methods in Structural Dynamics and Earthquake Engineering
M. Papadrakakis, M. Fragiadakis, V. Plevris (eds.)
Corfu, Greece, 25–28 May 2011

EFFECT OF HIGHER ORDER SHEAR DEFORMATION ON THE

NONLINEAR DYNAMIC ANALYSIS OF LAMINATED COMPOSITE
PLATE UNDER IN-PLANE LOADS
Dr. Haider K. Ammash

Iraq , Al-Qadissiya University , Engineering College ,Civil

Engineering

[email protected]

Keywords: Dynamic analysis, Nonlinear analysis, Composite plate, Finite element method

Abstract: A nonlinear finite element method is adopted for the large displacement dynamic
analysis of anisotropic plates under in-plane compressive loads. The analysis is based on the
two-dimensional layered approach with higher order shear deformation theory with five,
seven, nine, and eleven degrees of freedom per node, nine-node Lagrangian isoparametric
quadrilateral elements are used for the discretization of the laminated plates. A complete
bond between the layers is assumed (no delamination occurs). A consistent mass matrix is
used in the present study. Damping property is considered by using Rayleigh type damping
which is linearly related to the mass and the stiffness matrices. Newmark integration method
and Harmonic acceleration method are used for solving the dynamic equilibrium equations.
The effects of number of layers, damping factor, and number of degree of freedom per node
on the large displacement dynamic analysis are considered. From the present study, noticed
that the central deflection increase with increasing the degree of freedom per node.
 Haider K. Ammash

1.GENERAL
Certain civil engineering structures are designed to carry their own dead load plus
superimposed loads which are immovable and unvarying with time, that is, superimposed
static loads. In such cases, the stress analysis involves only principles of statics. More often
the design of a civil engineering structure involves not only static loads but also superimposed
loads which are either moving or movable and may vary with time as in superimposed
dynamic loads. In such cases, the stress analysis properly should involve principles of
dynamics to determine the effect of dynamic loading. However, in many of these cases,
experience has shown that the dynamic effect makes a minor contribution to the total load
which must be provided for the design and therefore the dynamic effect need not be evaluated
precisely. In such cases, the dynamic effect may be handled by the use of an equivalent static
load, or by an impact factor or by a modification of the factor of safety(3).
There have been a number of developments which have led to growing interest in a
more precise evaluation of the effects produced by the dynamic portion of the loading.
Among these are the imposition of more severe live load conditions (that is, machinery and
vehicles moving at high speeds), the construction of high towers and long bridges involving
more severe and important wind-loading conditions, the necessity of developing blast
resistant constructions, and the desire to improve earthquake resistance of constructions.
These are some aspects where it may be necessary to consider more precisely the response
produced by dynamic loading.
The ability of thin-walled structures to absorb the energy of dynamic transient loading
has led to its utilization for several classes of important structures, such as aerodynamic
structures, power plant structure, bridge structures, etc. These types of structures are designed
under these loads to maintain the overall structural integrity with irreversible deformation
analysis. In the present study a computational modeling is developed for the nonlinear
dynamic analysis of laminated composite plates using finite element method. The dynamic
equilibrium Equation and the derivation of mass and damping matrices will be presented. A
Newmark direct time integration method is adopted. In [1993], Kommineni and Kant
presented a Co-continuous finite element formulation of a higher order displacement theory
for predicting linear and geometrically nonlinear behavior in the sense of von-Karman
transient response of composite and sandwich plates. Azevedo and Awruch [1999] presented
a geometric nonlinear dynamic analysis of plates and shells using eight-node hexahedral
isoparametric elements. The main features of their study are: (1) the element matrices were
obtained by using reduced integrations with hourglass control; (2) an explicit Taylor-Galerkin
scheme was used to carry out the dynamic analysis by solving the corresponding equations of
motion in terms of velocity components. Tao, et al. [2004] presented a simple solution of the
dynamic buckling of stiffened plates under impact loading. Based on large deflection theory,
a discretely stiffened plate model had been used. The tangential stresses of stiffeners and
their in-plane displacements were neglected.

2. LAMINATED PLATE THEORIES

A laminated plate is a series of laminas bonded together to act as an integral structural
element. Thus, a laminate is not a material but instead a structural element with essential
features of both material properties and geometry. The stiffness and strength of such a
composite material with structural configuration are obtained from the properties of the
constituent laminas, and thus the macromechanical behavior of a laminate is the main topic of
this section. The lamination so described can be considered as a single layer with "rule of
mixtures" representation of the interaction between the multiple laminas in a plate or shell(6).

2
 Haider K. Ammash

Figure (1): Laminated plate with several lamina orientations(6)

In the analysis of the laminated plates, there are two categories of theories, equivalent
single layer and three dimensional elasticity theories. In the first category, the material
properties of the constituent layer are smeared to form a hypothetical single layer whose
properties are equivalent to through thickness integrated sum of its constituents, and this
category contains classical lamination theory, first order shear deformation theory, and higher
order shear deformation theory as will be given in the following section:

2.1 Classical lamination theory

Classical laminated plate theory is also often called "classical laminated theory (CLT)" which
is based on the Kirchhoff-Love hypothesis for plates and shells (6). The assumptions of
classical laminated plate theory are as follows:
1- The plate is thin. That is the thickness (h) is small compared to the other physical
dimensions.
2- The displacements u( x , y , z ), v ( x , y , z ) and w ( x , y , z ) are small compared to the plate
thickness.
3- The in-plane strains ε xo , ε oy and γ xy
o
are small compared to unity.
4- The transverse normal stress σ z is negligible.
5- The transverse shear stresses τ xz , τ yz are negligible.

2.2 First order shear deformation theory (FSDT)

Timoshenko deep beam theory, which includes transverse shear deformation and rotary
inertia effect, has been extended to isotropic plates by Reissner and Mindlin, and to laminate
anisotropic plates by Yang, et. al. and their theory, also called "First order shear deformation
theory (FSDT)", takes into account the effect of transverse shear deformation and assume it
constant through the plate thickness. Thus, a shear correction factor is used (3). The
assumptions of First order shear deformation theory (FSDT) are as follows:
1- The in-plane displacements are linear functions of z (plane cross sections remaining plane
after deflection).
2- The displacements u( x , y , z ), v ( x , y , z ) and w ( x , y , z ) are small compared to the plate
thickness.
3- The in-plane strains ε x , ε y and γ xy are small compared to unity.
4- The transverse normal stress σ z is negligible.
5- The transverse shear stresses τ xz , and τ yz are considered to be constant through the plate
thickness.

3
 Haider K. Ammash

2.3 Higher order shear deformation theory (HSDT)

In general, a layered composite plate exhibits coupling between the in-plane displacements,
transverse displacements and shear rotations. However, due to the low transverse shear
modulus relative to the in-plane Young's modulus of each lamina, the transverse shear
deformation effect is more pronounced in composite than in isotropic plates. Hence, several
types of shear deformation theories have been introduced.
The higher order shear deformation theories are more efficient to represent the
transverse shear deformation, through-thickness displacement and strains. The assumption of
a higher order plate theory can also be used within the equivalent layer formulation (6).
The assumptions of higher order shear deformation are as follows:
1- The plate may be moderately thick.
2- The in-plane displacements u( x , y , z ), v ( x , y , z ) are cubic functions of z.
3- The transverse shear stresses τ xz , and τ yz are parabolic in z, no shear correction factor is
necessary.
4- The in-plane stresses σ x , σ y , and τ xy are cubic functions of z.
5- The normals to the mid-surface before deformation are straight, but not necessarily
remain normal to it the mid-surface after deformation.
6- The transverse normal stress σ z is negligible.
Figure (2) briefly shows, the basic difference between the classical and the first order
theories with the higher order theories.

Undeformed x,u
A z
x
z,w
∂w ∂x
CLT
(uo,wo)
(u,w) ∂w ∂x

ϕx
FSDT
uo,wo
(u,w) ∂w ∂x
ϕx
HSDT uo,wo
(u,w) ∂w ∂x

Figure (2): Kinematics of deformation of a plate in various plate theories(1)

All the prescribed theories are considered in the present study in order to study the effect of
these theories on the accuracy and the time consumption in the analysis. In the present study,
three types of displacement equations were considered. Firstly the displacement
representation for this theory with five degrees of freedom per node is as follows:
u ( x , y , z , t ) = u o ( x , y , t ) + zθ x ( x , y , t )
v ( x , y , z , t ) = v o ( x , y , t ) + zθ y ( x , y , t ) (1)
w (x, y, z, t ) = w o (x, y, t )

4
 Haider K. Ammash

in which t denotes the time; and uo, vo, and wo are the components of the mid-plane
displacements for a generic point (x,y,z) having displacements u, v, and w in x, y, and z
directions, respectively. Here, θx and θy are rotations of transverse normals in the (xz) and
(yz) planes, respectively.
The strain-displacement relations after differentiating Equation (1) are:
∂u
εx = = ε xo + zκ x
∂x
∂v
εy = = ε oy + zκ y
∂y
∂u ∂v
γ xy = + = γ xy
o
+ zκ xy (2)
∂y ∂x
∂u ∂w
γ xz = + =ϕx
∂z ∂x
∂v ∂w
γ yz = + =ϕy
∂z ∂y
where
∂uo ∂v ∂u ∂v
ε xo = , ε oy = o , γ xy
o
= o+ o
∂x ∂y ∂y ∂x
∂θ x ∂θ y ∂θ ∂θ y
κx = ,κy = , κ xy = x +
∂x ∂y ∂y ∂x
(3)
∂w o
ϕx =θx +
∂x
∂w o
ϕy =θy +
∂y
All the strains above are defined in the middle plane of the laminate and substitution
these Equations into the stress-strain relations. Secondly the higher order shear deformation
theory (HSDT) with seven degrees of freedom per node was considered. The strain
expressions derived from the displacement field was considered by Mallikarjuna, and Kant
[1988], and by Ali [2004] with seven degrees of freedom per node as follows:

u( x , y , z , t ) = uo ( x , y , t ) + zθ x ( x , y , t ) + z 3 θ*x ( x , y , t )
v ( x , y , z , t ) = v o ( x , y , t ) + zθ y ( x , y , t ) + z 3 θ*y ( x , y , t ) (4)
w( x, y, z, t ) = wo ( x, y, t )
in which (u, v, w, θx, and θy) are defined previously, θ*x and θ*y are the corresponding higher
order terms in Taylor's series expression and also defined at the middle plane. The strain-
displacement relations after differentiating Equation (4) are:

5
 Haider K. Ammash

∂u
εx = = ε xo + zκ x + z 3κ x*
∂x
∂v
εy = = ε yo + zκ y + z 3κ *y
∂y
∂u ∂v
γ xy = + = γ xyo + zκ xy + z 3κ xy
*
(5)
∂y ∂x
∂u ∂w
γ xz = + = ϕ x + z 2ϕ x*
∂z ∂x
∂v w
γ yz = + = ϕ y + z 2ϕ *y
∂z ∂y
where the parameters ( ε ox , ε oy , γ oxy , κ x , κ y , κ xy , ϕ x , ϕ y ) are defined previously.
* *
∂θ*x * ∂θ y * ∂θ*x ∂θ y
κ *x = , κy = , κ xy = +
∂x ∂y ∂y ∂x
(6)
ϕ *x = 3θ*x
ϕ *y = 3θ*y
Also, all the strains above are defined in the middle-plane of the laminate and
substitution these Equations into the stress-strain relations.

Thirdly, Higher order shear deformation theory (HSDT) with nine degrees of freedom per
node was considered. The strain expressions derived from the displacement field were
considered by [Ali, 2004] with nine degrees of freedom per node as follows:

u( x , y , z , t ) = uo ( x , y , t ) + zθ x ( x , y , t ) + z 2 uo* ( x , y , t ) + z 3 θ *x ( x , y , t )
v ( x , y , z , t ) = v o ( x , y , t ) + zθ y ( x , y , t ) + z 2 v o* ( x , y , t ) + z 3 θ *y ( x , y , t ) (7)
w ( x, y, z, t ) = w o ( x, y, t )
in which the parameters (u, v, w, θx, θy, θ*x , and θ*y ) are defined previously, uo* , and v o* are
the corresponding higher order terms in Taylor's series expression and they are also defined at
the middle plane. The strain-displacement relations after differentiating Equation (7) are:

∂u *
εx = = ε ox + zκ x + z 2 ε ox + z 3 κ *x
∂x
∂v *
εy = = ε oy + zκ y + z 2 ε oy + z 3 κ *y
∂y
∂u ∂ v *
γ xy = + = γ oxy + zκ xy + z 2 γ oxy + z 3 κ *xy (8)
∂ y ∂x
∂u ∂w
γ xz = + = ϕ x + zγ oxz + z 2 ϕ*x
∂z ∂x
∂v ∂w
γ yz = + = ϕ y + zγ oyz + z 2 ϕ*y
∂z ∂ y

6
 Haider K. Ammash

where ( ε ox , ε oy , γ oxy , κ x , κ y , κ xy , κ *x , κ *y , κ *xy , ϕ x , ϕ y , ϕ*x , ϕ*y ) are defined previously.

∂u*o o* ∂v*o o* ∂u*o ∂v*o

ε ox* = , εy = , γ xy = +
∂x ∂y ∂y ∂x
(9)
γ oxz* = 2u*o
γ oyz* = 2v*o
Also, all the strains above are defined in the middle-plane of the laminate. By
substitution from Equation (8) into the stress-strain relations, after complete integration, the
stress-resultant/strain relations of the laminate are as follows:

E16 ⎤ ⎡ ε x ⎤
o
⎡ N x ⎤ ⎡ A11 A12 A16 D11 D12 D16 B11 B12 B16 E11 E12
⎢ ⎥ ⎢ ⎥⎢ o ⎥
⎢ N y ⎥ ⎢ A12 A22 A26 D12 D22 D26 B12 B22 B26 E12 E 22 E26 ⎥ ⎢ ε y ⎥
⎢ N xy ⎥ ⎢ A E66 ⎥ ⎢⎢ γ xy ⎥⎥
o
A26 A66 D16 D26 D66 B16 B26 B66 E16 E 26
⎢ * ⎥ ⎢ 16 ⎥ o*
⎢ N x ⎥ ⎢ D11 D12 D16 F11 F12 F16 E11 E12 E16 G11 G12 G16 ⎥ ⎢ ε x ⎥
⎢ N* ⎥ ⎢ ⎢ ⎥
⎢ *y ⎥ ⎢ D12 D22 D26 F12 F22 F26 E12 E22 E26 G12 G22 G26 ⎥ ⎢ ε oy* ⎥
⎥ o*
⎢ N xy ⎥ ⎢ D16 D26 D66 F16 F26 F66 E16 E26 E66 G16 G26 G66 ⎥ ⎢ γ xy ⎥
⎢ ⎥=⎢ ⎢ ⎥ (10)
⎢ M x ⎥ ⎢ B12 B12 B16 E11 E12 E16 D11 D12 D16 F11 F12 F16 ⎥ ⎢ κ x ⎥
⎥
⎢ M y ⎥ ⎢B B22 B26 E12 E22 E26 D12 D22 D26 F12 F22 F26 ⎥ ⎢ κ y ⎥
⎢ ⎥ ⎢ 12 ⎥⎢ ⎥
M
⎢ xy ⎥ ⎢ B16 B26 B66 E16 E26 E66 D16 D26 D66 F16 F26 F66 ⎥ ⎢ κ xy ⎥
⎢ M* ⎥ ⎢E ⎢ ⎥
E12 E16 G11 G12 G16 F11 F12 F16 H11 H12 H16 ⎥ ⎢ κ*x ⎥
⎢ x ⎥ ⎢ 11 ⎥ *
⎢ M *y ⎥ ⎢ E12 E22 E26 G12 G22 G26 F12 F22 F26 H12 H 22 H 26 ⎥ ⎢ κ y ⎥
⎢ * ⎥ ⎢ ⎢ ⎥
E E26 E66 G16 G26 G66 F16 F26 F66 H16 H 26 H 66 ⎥⎦ ⎢ κ*xy ⎥
⎣⎢ M xy ⎦⎥ ⎣ 16 ⎣ ⎦
and,

⎡Q x ⎤ ⎡ A55 A45 B55 B45 D55 D45 ⎤ ⎡ ϕ x ⎤

⎢Q ⎥ ⎢A ⎢ ⎥
⎢ y⎥ ⎢ 45 A44 B45 B44 D45 D44 ⎥⎥ ⎢ ϕ y ⎥
⎢ Sx ⎥ ⎢ B55 E45 ⎥ ⎢ γ xz ⎥
o*
B45 D55 D45 E55
⎢ ⎥= ⎢ ⎥ ⎢ o* ⎥ (11)
⎢Sy ⎥ ⎢ B45 B44 D45 D44 E45 E44 ⎥ ⎢ γ yz ⎥
⎢Q*x ⎥ ⎢ D55 F45 ⎥ ⎢ ϕ x ⎥
*
D45 E55 E45 F55
⎢ *⎥ ⎢ ⎥ *⎥
⎢
⎢⎣Q y ⎥⎦ ⎢⎣ D45 D44 E45 E44 F45 F44 ⎥⎦ ⎢⎣ ϕ y ⎥⎦

All coefficients in A, B, D, E, F, G, and H groups are defined as follows:

NL
Aij = ∑ Qij ( hL − hL−1 ) i, j = 1, 2, 6 or i, j= 4, 5 (12 a)
L=1

NL
Bij = (1 2) ∑ Qij ( hL2 − hL2 −1 ) i, j = 1, 2, 6 or i, j= 4, 5 (12 b)
L=1

NL
Dij = (1 3) ∑ Qij ( hL3 − hL3 −1 ) i, j = 1, 2, 6 or i, j= 4, 5 (12 c)
L=1
NL
E ij = (1 4) ∑ Qij ( hL4 − hL4 −1 ) i, j = 1, 2, 6 or i, j= 4, 5 (12 d)
L=1

7
 Haider K. Ammash

NL
Fij = (1 5) ∑ Qij ( hL5 − hL5 −1 ) i, j = 1, 2, 6 or i, j= 4, 5 (12 e)
L=1
NL
G ij = (1 6) ∑ Qij ( hL6 − hL6 −1 ) i, j = 1, 2, 6 (12 f)
L=1
NL
H ij = (1 7 ) ∑ Q ij ( hL7 − hL7 −1 ) i, j = 1, 2, 6 (12 g)
L =1

1
2
h0
h1
h2
Middle surface
h
hn-3
hn-2
hn
hn-1
n
Layer number
Figure (3): Geometry of an NL-layered laminate [Jones,1999]()

DYNAMIC EQUILIBRIUM EQUATION

The dynamic equilibrium Equations are obtained by using the principle of virtual work which
states that for any arbitrary kinematically consistent set of displacements, the internal virtual
work done by stresses through virtual strains must be equal to that done by the external forces
irrespective of the material behavior as [Cook, 1995](8):

∫ (dε ) σ dv = ∫ (du ) Pt ds + ∫ (du ) ( Pb − ρu&& − Cu& )dv

T T T
(13)
v st v
where du is a vector of virtual displacements, dε is the vector of associated virtual strains
and σ is the vector of actual stresses. The term Pt is a vector of surface tractions acting on
the portion s t of the boundary S . Vectors Pb , ρ u
&& and C u& are the body, inertial and damping
.
forces respectively. The symbol ( ) denotes differentiation with respect to time. ρ is the mass
density and C is the damping parameter.
For the finite element representation, the displacements, velocities and accelerations
&& can be defined in terms of the nodal variables d , d& and d&& by the expressions:
u, u& and u

m
u = ∑ N i (ξ, η) d i = Nd (14)
i =1 , du = N δd
m
u& = ∑ N i (ξ , η) d& i = N d& (15)
i =1
m
&& = ∑ N i (ξ , η) d&&i = N d&&
u (16)
i =1

8
 Haider K. Ammash

m
where u = ∑ N i (ξ ,η ) d i = Nd , Ni is the shape functions for i node, and m is the number of
i =1
nodes.
With standard strain-nodal displacement matrix [B] , the virtual strain vector can be
related to the nodal displacements as:
m
dε = ∑ [B ]i δd i = [B ]δd (17)
i =1
Upon substitution of Equations (14-17) into Equation (13) then:

δd T [[M ]d&& + [C ]d& + [K ]d ] = δd T { f e (t )} (18)

in which the mass matrix [M ] , the damping matrix [C ] , the stiffness matrix [K ] and the
external applied vector { f e (t )} have the following element contributions:

[M e ] = ∫ N T ρ N dV (19)
Ve
[C e ] = ∫ N T C N dV (20)
Ve

[K e ] = ∫ [B]T [D][B]dV (21)

Ve

∫N ∫ NPb dV
T
f e (t ) = Pt ds + (22)
se Ve

where s e and Ve denote the surface and volume of the element under consideration. As δd T is
arbitrary, then Equation (18) may be written as:

[M ]{d&&}+ [C ]{d& }+ [K ]{d } = { f e (t )} (23)

Equation (23) is the dynamic equilibrium Equation for a single or multi-degree of freedom
system.

FORMULATION OF ELEMENT MASS MATRIX

The kinetic energy of the element (e) can be expressed as follows:
1 &T
2 ∫A
TI e = d [m ] d& dA{} {} (24)
The velocity vector within an element is discretized such that:

{d&} = ∑ N i {d&}, NN: number of nodes.

NN
(25)
i =1
By substituting Equation (25) into Equation (24), one gets:

TI e =
1 NN &
∑d
2 i =1
{ }T ∫ N iT [m ]N i dA{d&} (26)
A
Thus,
1 1
[M ]e = ∫ [N ]T [m ][N ]dA = ∫ ∫ [N ]TJ[M ][N ] dξ dη (27)
A −1−1

9
 Haider K. Ammash

The mass matrix for nine degrees of freedom per node is:

⎡ I1 0⎤
⎢ I1 ⎥
⎢ ⎥
⎢ I1 ⎥
⎢ ⎥
⎢ I2 ⎥
[m ]9×9 = ⎢ I2 ⎥ (28)
⎢ ⎥
⎢ I3 ⎥
⎢ I3 ⎥
⎢ ⎥
⎢ I4 ⎥
⎢0 I 4 ⎥⎦
⎣
For layered plates, the element mass matrix can be written as follows:

NL
[M ] = ∑ [M ]e (29)
L=1
where in the above Equation (28), I1, I2, I3, and I4 are translation inertia, rotary inertia, and
respectively higher order inertia terms, and these are given by:

NL hL
(I1 , I 2 , I 3 , I 4 ) = ∑ ∫ (1, z 2 , z 4 , z 6 ) ρ L dz (30)
L=1 hL −1

where ρ L is material density of L-th layer.

FORMULATION OF DAMPING PROPERTIES

The most common form for the representation of the damping matrix [C] is the so-called
Rayleigh-type damping(3) which was given as;

[C ] = a o [M ] + a1 [K ] (31)
in which (ao and a1) are arbitrary proportionality factors, which make the damping matrix
satisfy the orthogonality condition with respect to the modal matrix [Φ] in the same way of
the orthogonality conditions for the mass and stiffness matrices that is(5):

{Φ}T [M ]{Φ} = [I ]
{Φ}T [K ]{Φ} = [Λ ] (32)
{Φ}T [C ]{Φ} = 2[γ ][Λ ]1/ 2
where
{Φ} : The modal matrix whose columns represent the natural modal shapes and the
superscript ( T ) denotes transpose.
[I ] : Identity matrix.
[Λ ] : Spectral matrix, which is a diagonal matrix with elements representing the
squares of the natural frequencies ( ω 2i ).

10
 Haider K. Ammash

[γ ] : Modal damping matrix which is also a diagonal matrix with elements

representing the damping ratios for the system modes ( γ i )
Premultiplying Equation (37) by {Φ} and postmultiplying it by {Φ} yields:
T

{Φ}T [C ]{Φ} = a o {Φ}T [M ]{Φ} + a1 {Φ}T [K ]{Φ} (33)

Substituting Equations (38) into Equation (39) gives;

2[γ ][Λ ] = a o [I ] + a1 [Λ ]
1/ 2
(34)
The two factors, ao and a1 can be determined by specifying the damping ratios for two
modes for example 1 and 2, and substituting into Equation (34) as(12):
2γ 1ω1 = a o + a1ω12 (35)
2γ ω = a + a ω 2
2 2 o 1 2
(36)
where ω1 and ω2 are the natural frequencies for modes 1 and 2 respectively. By solving the
above two Equations one can get:

2ω1ω 2 (ω 2 γ 1 − ω1 γ 2 )
ao = (37)
(ω 22 − ω12 )
2(ω 2 γ 2 − ω1 γ 1 )
a1 = (38)
(ω 22 − ω12 )
Then, the values of ao and a1 are substituted into Equation (31) to get the required damping
matrix.

FORCED VIBRATION ANALYSIS

The calculation of the nonlinear dynamic response of structure of structures including
instability or buckling phenomena has received considerable attention and a good amount of
literature has appeared on this subject. The nonlinear dynamic analysis depends largely on
solving the following Equations:

[M ] {d&&( t )}+ [C ] {d& ( t )}+ [K T ] {d ( t )} = {F ( t )} (39)

in which [K T ] is the tangent stiffness matrix of the plate (or structure) and depends on the
current displacements and stresses. The most conventional implicit time integration
procedures is Newmark method. After solving Equation (39) at time (t+∆t) for
displacements, velocities, and accelerations, the following equation as:

( [K T ] + ao [M ] + a1 [C ]){d }t + ∆t = {F (t )}t + ∆t + [M ](a 2 {d&}t + a3 {d&&}t )

+ [C ](a 4 {d&}t + a5 {d&&}t )
(40)
For convenience, the following is used:

[K T ]eff = [K T ] + ao [M ] + a1 [C ] (41)
and,
{F ( t )}eff = {F ( t )}t + ∆t + [M ](a 2 {d& }t + a 3 {d&&}t ) + [C ](a 4 {d& }t + a 5 {d&&}t ) (42)

11
 Haider K. Ammash

So, Equation (42) may be written in the form:

[K T ]eff {d }t + ∆t = {F (t )}eff (43)
For a linear system, [K T ]eff will be constant during the analysis at any time, while in the
nonlinear analysis, [K T ]eff is a function of current displacement vector {d } . Therefore, an
iterative procedure must be used to define [K T ]eff . In the nonlinear analysis, it is more useful
to put Equation (43) in increment form. For such purpose, Equation (43) may be rewritten as:

[Kˆ ]{∆d } = {∆Fˆ (t )}

T (44)
[ ] { }
in which K̂ T is the effective stiffness matrix and ∆Fˆ ( t ) is the effective load vector.
Equation (44) is solved by an iterative procedure like Equation (40). It may be noted that
Equation (40) may be used for solving linear problems, while for nonlinear problems,
Equation (44) should be used.
Solving Equation (44) for {∆d }, approximate values for accelerations, velocities and
displacements may be given as:
d&&{ }t + ∆t o
= a {∆d } − a d& − a d&& 2{ }t 3 { }t
{d&}t + ∆t = a1{∆d } − a 4 {d& }t − a 5 {d&&}t (45)
{d }t + ∆t = {d }t + {∆d } ;
where
1 γ 1 1
ao = , a1 = , a2 = , a3 = −1
β ( ∆t ) 2
β ( ∆t ) β ( ∆t ) 2β
γ ∆t ⎛ γ ⎞
a4 = − 1, a 5 = ⎜⎜ − 2 ⎟⎟
β 2 ⎝β ⎠

APPLICATIONS AND DISCUSSIONS

Several plates are analyzed to study the different effects on the large displacement dynamic
behavior of plates with some comparison with other researchers.
Comparison with available theoretical investigation of composite plate
Clamped supported square angle-ply laminated plate under transverse suddenly applied
constant dynamic loading

A square angle-ply (0o/45o/90o/core/90o/45o/30o/0o) sandwich laminated plate with clamped

edges and subjected to a suddenly applied uniformly transverse load was analyzed and
compared with Kommineni and Kant [1993](). The following layer material properties are
used in the analysis: for face sheets (Graphite/epoxy prereg system) (E1=130.8 GPa; E2=10.6
GPa G12=G13=6 GPa; G23=3.9 GPa; v12=0.28; and ρ=15.8 kN.sec2/m4); for core sheet (US
Commercial al. honeycomb, ¼ in cell size, 0.003 in foil) (G13=0.5206 GPa; G23=0.1772 GPa;
ρ=1.009 kN.sec2/m4). The time step is (∆t=0.000025 sec),and applied load (q=50 kN/m2).
The geometry properties are (a=1.0 m, a/b=1, and h=0.01m, at top three stiff layers, thickness
of each layer=0.025 h, at bottom four stiff layer, thickness of each layer=0.08125 h, and
thickness of core=0.6 h). Kommineni and Kant used nine-node isoparametric Lagrangian

12
 Haider K. Ammash

elements with nine-node degrees of freedom per node and divided the full plate into (4×4)
element mesh.
In the present study, the full laminated plate is modeled by (4×4) element mesh with
nine-node isoparametric Lagrangian element and nine degrees of freedom per node. A
consistent mass matrix and Newmark integration method with α=1/2, and β=1/4 were used
in the present study.
Figure (4) shows the time history curve for the clamped angle-ply laminated plate
under transverse suddenly applied load. From this figure, it can be noticed that good
agreement with other study exists with a difference not exceeding (1%).
1.0
Kommineni and Kant
0.9
Present study
Deflection ratio (w/h)

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
0.0 0.5 1.0 1.5 2.0 2.5
3
Time ×10 (sec)
Figure (4): Central deflection ratio-time curve of a clamped square sandwich composite plate under transverse constant
dynamic loading,(b/h=100, ∆t=0.000025sec, q=50 kN/m2)

A simply supported square plate with slenderness ratio (b/h=100), and with symmetric
cross-ply and antisymmetric cross-ply arrangements, were chosen to study the effect of
number of layers on the large displacement dynamic behavior of a laminated composite plate
under in-plane constant dynamic loading. The initial imperfection is (wo/h= 0.1) by which the
shape is considered to be a sinusoidal curve.

Figures (5) and (6) present the time history curve and show that for the same volume of
the plate, the response (deflection) will decrease about (15%) for the symmetric cross-ply and
about (29%) for the antisymmetric cross-ply plates where with increasing the number of
layers (3-10) for the symmetric cross-ply and (2-10) for the antisymmetric cross-ply
arrangements, the stiffness increase may be related to the increase of the number of the
reinforced layers. Thus, extension and bending stiffness will increase; and therefore, the
amplitude will decrease. Also, the increase of the number of layers will give a better
distribution of orthogonal stiffness through the depth. From these figures, it can be seen that
the increase of the number of layers more than (8 layers) for the symmetric cross-ply and the
antisymmetric cross-ply plates have slight effect on increasing the stiffness of the plate.

13
 Haider K. Ammash

3.0
2.5
Deflection ratio (w/h) 2.0
1.5
1.0
0.5
0.0
-0.5
Symmetric cross-ply Px
-1.0
(0/90/0)
-1.5
(0/90/90/0)
-2.0 (0/90/0)
Px

Load
2
-2.5 (0/90/0/90) 2
-3.0 (0/90/0/90/0) 2
Time
-3.5
0 5 10 15 20 25
Time ×103 (sec)
Figure (5): Effect of number of layers on the large displacement analysis of symmetric cross-ply laminated
plate under in-plane constant dynamic loading ratio (Px/Pu=0.4), (b/h=100, t=0.0001, wo/h=0.1)

3.5
Antisym. cross-ply
3.0
(0/90) 1
2.5
Deflection ratio (w/h)

(0/90)2
2.0
(0/90) 3
1.5
(0/90) 4
1.0
(0/90) 5
0.5
0.0 Px
-0.5
-1.0
Px
Load

-1.5
-2.0 Time
-2.5
0 5 10 15 20 25
Time ×103 (sec)
Figure (6): Effect of number of layers on the large displacement analysis of antisymmetric cross-ply laminated
plate under in-plane constant dynamic loading ratio (Px/Pu=0.4), (b/h=100, t=0.0001, wo/h=0.1)

To study the effect of shear deformation on the large displacement dynamic analysis
of a laminated composite plate under in-plane constant dynamic loading, a simply supported
square plate with slenderness ratio (b/h=20), and with symmetric cross-ply antisymmetric
cross-ply arrangements and with eight layers was analyzed. The initial imperfection is (wo/h=
0.1) by which the shape is considered to be a sinusoidal curve.

14
 Haider K. Ammash

Figures (7) and (8) present the time history curves for the symmetric cross-ply, and for
the antisymmetric cross-ply laminated composite plates by taking the through-thickness shear
deformation through the degrees of freedom of the element. From these figures, it can be
noticed that increasing the number of degrees of freedom per node from five degrees to nine
degrees will increase the central deflection about (16%) for symmetric cross-ply and about
(20%) for antisymmetric cross-ply plates.

0.2
Deflection ratio (w/h)

0.1

0.0
Symmetric cross-ply
Px
Load
5 DOF Px
7 DOF
9 DOF
Time
-0.1
0 2 4 6 8 10
3
Time ×10 (sec)
Figure (7): Effect of transverse shear deformation on the large displacement analysis of symmetric cross-ply
laminated plate under in-plane constant dynamic loading ratio (Px/Pu=0.3), (b/h=20, t=0.0001,
wo/h=0.1,Pu=18563 kN/m)

0.4
Antisymmetric cross-ply
Px
5 DOF
Deflection ratio (w/h)

0.3 7 DOF
9 DOF
Px
Load

0.2
Time

0.1

0.0

-0.1
0 2 4 6 8 10
Time ×103 (sec)
Figure (8): Effect of transverse shear deformation on the large displacement analysis of antisymmetric cross-
ply laminated plate under in-plane constant dynamic loading ratio (Px/Pu=0.3), (b/h=20, t=0.0001,
wo/h=0.1,Pu=16347 kN/m)

To study the effect of damping on the large displacement elastic-plastic dynamic

behavior of composite plates, two examples are considered. The first one is a simply

15
 Haider K. Ammash

supported square plate with symmetric cross-ply lamination with eight layers and under in-
plane dynamic loading. The second one is a simply supported square plate with
antisymmetric cross-ply lamination with eight layers and under in-plane dynamic loading.
Different values of damping factor (0.05-0.1) are considered in the present study. The initial
imperfection shape is considered to be a sinusoidal curve. The following geometry and layer
material properties of high graphite epoxy are used in the analysis: (E1=172.5 GPa; E2=7.08
GPa; G12=G13=3.45 GPa, G23=1.38 GPa; ρ=15.8 kN.sec2/m4)(). The geometry properties are
(a=1.0 m, a/b=1).
Figure (9) and (10) present the time history curve for a simply supported square plate
with symmetric and antisymmetric cross-ply lamination under in-plane constant loading. It is
noticed that the response (deflection) decreases with the increase of the damping factor. Also,
the plate shows no oscillation about the static deflection position, this means that the plate is
under the critical damping ratio
4

3
Deflection ratio (w/h)

2
Px

Symmetric cross-ply
1
Px
Load

γ=0.05

γ=0.075

0 γ=0.10
Time

-1
0 5 10 15 20 25
3
Time ×10 (sec)
Figure (9):Effect of damping factor on the large displacement analysis of a simply supported square symmetric
cross-ply plate under in-plane constant dynamic loading, (b/h=100, t=0.0001, wo/h=0.1, Px/Pu=0.65,
Pu=972.4 kN/m )
4
Deflection ratio (w/h)

Px
1 Antisymmetric cross-ply
γ=0.05
Px
Load

γ=0.075
0 γ=0.10

Time
-1
0 5 10 15 20 25
Time ×103 (sec)

Figure (10):Effect of damping factor on the large displacement analysis of a simply supported square
antisymmetric cross-ply plate under in-plane constant dynamic loading, (b/h=100, t=0.0001, wo/h=0.1,
Px/Pu=0.65, Pu=960 kN/m)

16
 Haider K. Ammash

CONCLUSIONS
A nonlinear finite element method is adopted for the large displacement dynamic analysis of
anisotropic plates under in-plane compressive load. Damping property is considered by using
Rayleigh type damping which is linearly related to the mass and the stiffness matrices.
Newmark integration method is used for solving the dynamic equilibrium equations. The
effects of initial imperfection, orthotropy of individual layers, fiber’s orientation angle, type
of loading, damping factor, and on the large displacement dynamic analysis are considered.
The conclusion it is shown that the antisymmetric cross-ply laminated plate has a damping
rate faster than the symmetric cross-ply laminated plate and if damping is considered and if
the response of the plate shows no oscillation about the static deflection position, it means that
the damping factor is below the critical damping factor. So, noticed that the central deflection
increasing with increasing the degree of freedom per node.

REFERENCES
[1] Ali, N. H., “Finite Element Dynamic Analysis of Laminated Composite Plates Including
Damping Effect”, M.Sc. Thesis, University of Babylon, Hilla, Iraq, 2004.
[2] Akay, H. “Dynamic Large Deformation Analysis of Plates Using Mixed Finite Elements”
Comp. & Struct., Vol.11, 1980, pp1-11.
[3] Ammash, H. K., “Nonlinear Static and Dynamic Analysis of Laminated Plates Under In-
plane Forces”, Ph.D. Thesis, University of Babylon, Hilla, Iraq, 2008.
[4] Azevedo, R.L. and Awruch, A.M. “Geometric Nonlinear Dynamic Analysis of Plates and
Shells Using Eight-Node Hexahedral Finite Element with Reduced Integration”, J. Braz.
Soc. Mech. Sci., Vol.21, No.3, 1999, pp.1-22.
[5] Bathe, K.J., and Ozdemir, H. “Elastic-Plastic Large Deformation Static and Dynamic
Analysis.”, Comp. & Struct., Vol.6, No.2, 1975, pp81-92.
[6] Jones, R.M., “Mechanics of Composite Materials”, Second Edition, Taylor and Francis
Inc., U.S.A., 1999.
[7] Kao, R., “Nonlinear Dyanmic Buckling of Spherical Caps with Initial Imperfections”,
Comp. & Struct., Vol.12, 1980, pp49-63.
[8] Kaw, A., “Mechanics of Composite Materials”, Second Edition, Taylor and Francis
Group, LLC, 2006.
[9] Khante, S. N., Rode, V., and Kant, T., “Nonlinear Transient Dynamic Response of
Damping Plates Using a Higher Order Shear Deformation Theory”, Nonlinear Dynamics,
Vol.47, 2007, pp38-403.
[10] Kommineni, J. R., and Kant, T. “Geometrically Non-linear Transient Co Finite Element
Analysis of Composite and Sandwich Plates with a Refined Theory.” Struct. Eng. And
Mech., Vol.1, No.1, 1993, pp87-102.
[11] Pica, A., Wood, R.D., and Hinton, E. “Finite Element Analysis of Geometrically
Nonlinear Plate Behavior Using a Mindlin Formulation.” Comp. & Struct., Vol.11, 1979,
pp.203-215.
[12] Pytet, M., “Introduction to Finite Element Vibration Analysis”, 1990.
[13] Tao, Z., Tu-guang, L., Yao,Z., and Jio-zhi,L. “Nonlinear Dynamic Buckling of
Stiffened Plates under In-plane Impact Load.”, J. Zhejiang University Science, Vol.5,
No.5, 2004, pp609-617.
[14] Weller, T., Abramovich, H., and Yaffe, R., “Dynamic Buckling of Beams and Plates
Subjected to Axial Impact”, Comp. & Struct., Vol.32, No.3/4, 1989, pp835-851.

17