You are on page 1of 42

Author's Accepted Manuscript

Non-linear static bending and forced vibra-


tions of rectangular plates retaining non-
linearities in rotations and thickness defor-
mation
Farbod Alijani, Marco Amabili

www.elsevier.com/locate/nlm

PII: S0020-7462(14)00197-8
DOI: http://dx.doi.org/10.1016/j.ijnonlinmec.2014.10.003
Reference: NLM2379

To appear in: International Journal of Non-Linear Mechanics

Received date: 30 June 2014


Revised date: 29 September 2014
Accepted date: 3 October 2014

Cite this article as: Farbod Alijani, Marco Amabili, Non-linear static bending
and forced vibrations of rectangular plates retaining non-linearities in
rotations and thickness deformation, International Journal of Non-Linear
Mechanics, http://dx.doi.org/10.1016/j.ijnonlinmec.2014.10.003

This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal
pertain.
NON-LINEAR STATIC BENDING AND FORCED VIBRATIONS OF
RECTANGULAR PLATES RETAINING NON-LINEARITIES IN
ROTATIONS AND THICKNESS DEFORMATION

Farbod Alijani, Marco Amabili

Department of Mechanical Engineering, McGill University


817 Sherbrooke Street West, Montreal, Canada H3A 0C3

Number of Pages = 26+Figures


Number of Figures=7

Professor Marco Amabili, Canada Research Chair (Tier 1)


Department of Mechanical Engineering
McGill University
Macdonald Engineering Building
817 Sherbrooke Street West
Montreal, PQ, Canada H3A 0C3
Fax: +1 514 398 7365
Ph: +1 514 398 3068
email: marco.amabili@mcgill.ca

1
Abstract

Geometrically non-linear static bending and forced vibrations of rectangular plates are studied

allowing full non-linear terms associated with Green-Lagrange strain-displacement relations,

second-order thickness stretching, third-order shear deformation and rotary inertia by using

seven independent parameters to describe the shell kinematics. In particular, in addition to non-

linearities in membrane and transverse deflection, non-linear terms associated with rotations and

thickness deformation parameters are also included. In order to obtain the governing equations of

motion, the three-dimensional constitutive equations are used, removing the assumption of zero

transverse normal strain. The boundary conditions of the plate are assumed to be simply

supported immovable and the equations of motion are derived by using a Lagrangian approach.

The numerical solutions are obtained by using pseudo arc-length continuation and collocation

scheme. In order to compare the non-linear static response, another analysis has also been carried

out by using the finite element code ANSYS and three-dimensional solid modeling. Results

reveal that the new theory with full geometric non-linearities provides significant accuracy

improvement for rotational and thickness deformation parameters, and, unlike other shear

deformation theories, predicts the correct thickness stretching along the plate.

Keywords

Non-linear static bending; Non-linear vibrations; Thickness deformation; Full geometric nonlinearities

2
1. Introduction

Rectangular plates that are made of soft materials, like rubbers and biological tissues present

significant thickness deformation while subjected to large deflections. Moreover, in very high-

frequencies, particular natural modes with prevalent dynamic variation of the thickness can be

observed. Therefore, in order to accurately predict the thickness deformation of the structures,

theories should be developed that allow thickness stretching in addition to shear deformation.

It is well-known that two-dimensional (2D) plate theories are treated as particular cases of three-

dimensional (3D) continuum theory of plates by eliminating the thickness coordinate z via a

priori integration in the thickness direction. Among 2D theories, the simplest theory is the

classical, also known as Kirchhoff theory, which neglects shear deformation and thus transverse

shear and transverse normal strains. To overcome this limitation, 2D shear deformation theories

have been developed. These theories can be divided into first-order and higher-order shear

deformation theories [1, 2]. In the first category, a shear correction factor is required for the

equilibrium since a uniform shear strain is assumed through the plate thickness. However, in the

latter this limitation is removed by assuming a realistic shear stress distribution through the plate

thickness and satisfying the condition of zero shear stresses at the top and bottom surfaces of the

structure.

An extensive literature survey on the classical and shear deformation theories of plates and shells

has been provided by Reddy and Arciniega [3] and Carrera [4]. The work of Carrera [5] also

elaborates the derivation of the governing equations and the finite element matrices of some of

the most well-known plate and shell theories given in Ref [4]. These review studies show the

diversity of classical and shear-deformation theories that have been developed over the past

years (e.g. [6-10]). Among them, Reddy’s non-linear third-order shear deformation theory has

3
received considerable attention in the literature [10]. In fact, this theory has been the basis of

many studies dealing with non-linear bending, buckling and vibrations of traditional, composite

and functionally graded (FG) plates (e.g. [11-13]). Reddy’s third-order theory accounts for

strains of the von Kármán type only and assumes a displacement field in which in-plane

displacements are cubic functions of the thickness coordinate and the transverse displacement is

constant through the thickness.

Even though higher-order shear deformation theories predict the global behavior of plates and

shells with enough accuracy, they are unable to reproduce the transverse displacement in the

thickness direction and fail to satisfy the continuity conditions of transverse shear stresses. To

overcome this problem, layer-wise and Zig-Zag theories have been developed [14, 15]. A

historical review of the Zig-Zag theories for multilayered plates and shells can be found in

Carrera [16].

In recent years, special attention has been devoted to higher-order 3D theories of shells and

plates accounting the thickness stretching effect. In particular, these new theories retain the effect

of transverse normal strain εzz and therefore provide a better evaluation of the structure’s

thickness variation comparing to the known 2D theories. Moreover, they provide the possibility

of using 3D material laws within the analysis. Among these works, one may refer to Carrera and

Brischetto [17, 18] who elaborated the effect of thickness stretching on linear bending and

vibrations of isotopic and multi-layered rectangular plates and shells, respectively. In particular,

they found that retaining the transverse normal strain εzz in theories that have linear transverse

displacement expansion in the thickness direction could result in a phenomenon known as

thickness locking or Poisson locking when 3D material laws are used. This phenomenon was

found to be due to the fact that linear displacement field theories result in a constant transverse

4
normal strain through the thickness that is not consistent with the physical transverse normal

strain obtained from the plane stress condition. To resolve this issue, they recommended using

2D material laws for classical and shear deformation theories that have constant or linear

transverse displacement expansion. New thickness stretching theories have also been proposed

by Neves et al. [19, 20] by using sinusoidal and hyperbolic functions for membrane

displacements to study the statics and free vibrations of rectangular plates, respectively. General

non-linear shear deformation theories with von Kármán type non-linearities and thickness

stretching effect for FG plates are presented in Reddy [21] and Reddy and Kim [22]. In

particular, by considering a third-order shear deformation theory in which the transverse

displacement was expanded up to the 2nd order in terms of thickness coordinate, Reddy [21] used

Hamilton’s principle to obtain the equations of motion and boundary conditions of FG

rectangular plates. Reddy and Kim [22] extended the theory of Ref [21] by using the modified

couple stress theory.

The new 3D theories are also believed to play an important role in studying the non-linear statics

and dynamics of highly stretched plates and shells [23-25]. In particular, Amabili [23, 24]

developed new thickness stretching and high-order shear deformation theories for shells of

generic shape by using 6 (linear transverse displacement in the thickness coordinate) and 8

parameters (cubic transverse displacement in the thickness coordinate), respectively, that was

shown to provide significant accuracy improvement for highly pressurized laminated composite

circular cylindrical shells. By using different higher-order thickness stretching theories, Alijani

and Amabili [25] discussed for the first time the role of transverse normal strain and stress on the

geometrically non-linear statics and dynamics of FG rectangular plates. Unlike other works, in-

5
plane non-linearities were retained in the strain-displacement relations in Refs [23-25] in order to

improve the accuracy during large deformations.

There are also a few 3D theories in the literature in which full non-linear terms associated with

Green-Lagrange strain-displacement relations are used [26, 27]. These works are based on a

displacement field in which the membrane displacements are cubic functions and the transverse

displacement is linear and therefore their analyses inherently results in some inaccuracy in the

evaluation of natural frequencies for anisotropic materials [25].

In the present paper, geometrically non-linear static bending and vibrations of rectangular plates

are discussed by considering full non-linear terms associated with Green-Lagrange strain-

displacement relations and using 2nd order thickness deformation theory to avoid inaccuracies for

anisotropic materials linked to the 1st order thickness deformation theory. In particular, in

addition to non-linearities in membrane and transverse deflection, non-linear terms associated

with rotations and thickness deformation parameters are also included. To the best of authors’

knowledge, no complete research has yet been performed in the literature in this regard and

therefore it is believed that this work is the first one that discusses the effect of rotational

geometric non-linearities in depth. In order to obtain the governing equations of motion, the

three-dimensional constitutive equations are used, removing the assumption of zero transverse

normal strain. The boundary conditions of the plate are assumed to be simply supported

immovable and the equations of motion are derived by using a Lagrangian approach. A

numerical bifurcation analysis is carried out by using pseudo arc-length continuation and

collocation scheme to obtain the non-linear static bending and forced vibration response of the

plate. Another analysis has been carried out by using the finite element code ANSYS via three-

dimensional solid modeling. Great attention has been given to obtaining the response of rotations

6
and thickness deformation parameters. It has been shown that the new theory with full non-

linearities leads to a thickness deformation and rotation response that is practically identical to

3D modeling in ANSYS and very different from the previous 3D theories which neglect non-

linearities in rotations and thickness deformation. Moreover, it is found that the new theory gives

weaker hardening behavior compared to conventional higher-order and thickness deformation

theories in case of flat plates and weaker softening behavior in case of plates with initial

deflection due to pressure. It is believed that the utilization of this theory is necessary for

structural elements that incorporate high-thickness deformation and require accurate modeling of

the variation of normal displacement through the thickness such as hyperplastic materials and

biological soft tissues.

2. New non-linear theory with thickness and shear deformation

Figure 1 shows a rectangular plate with in-plane dimensions a and b and thickness h considered

in an orthogonal coordinate system (O; x, y, z). The displacements of an arbitrary point of

coordinates (x, y) on the middle surface of the plate are denoted by u, v and w, in the x, y and z

directions, respectively. The displacements (u1, u2, u3) of a generic point of the plate are related

to the middle surface displacements by

u1 = u + zφ1 + z 2ψ 1 + z 3γ 1 , (1a)

u2 = v + zφ2 + z 2ψ 2 + z 3γ 2 , (1b)

u3 = w + z χ1 + z 2 χ 2 , (1c)

where φ1 and φ2 are the rotations of the transverse normals at z = 0 about the y and x axes,

respectively, and χ1 and χ 2 are parameters linked to the thickness deformation per unit

thickness. Moreover, ψ1, ψ2, γ1, γ2, are functions to be determined in terms of u, v, w, φ1, φ2, χ1

7
and χ 2 which are the seven independent parameters that describe the plate deformation. Here the

same displacement field introduced by Reddy [21] has been used.

By using equation (1c), the thickness stretching Δ h at a specific point (x, y) of the middle surface

can be obtained as

Δ h = u3 z =h /2 − u3 z =− h /2 = h χ1 . (2)

The Green’s strains, considering geometric non-linear terms, can be obtained as [1, 2]

∂u1 1 ⎡⎛ ∂u1 ⎞ ⎛ ∂u2 ⎞ ⎛ ∂u3 ⎞ ⎤


2 2 2

ε xx = + ⎢⎜ ⎟ +⎜ ⎟ +⎜ ⎟ ⎥, (3a)
∂x 2 ⎢⎣⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎥⎦

∂u2 1 ⎡ ⎛ ∂u1 ⎞ ⎛ ∂u2 ⎞ ⎛ ∂u3 ⎞ ⎤


2 2 2

ε yy = + ⎢ + + ⎥, (3b)
∂y 2 ⎢⎣ ⎜⎝ ∂y ⎟⎠ ⎜⎝ ∂y ⎟⎠ ⎜⎝ ∂y ⎟⎠ ⎥⎦

∂u1 ∂u2 ∂u1 ∂u1 ∂u2 ∂u2 ∂u3 ∂u3


ε xy = + + + + . (3c)
∂y ∂x ∂ x ∂y ∂x ∂ y ∂x ∂ y

∂u3 1 ⎡⎛ ∂u1 ⎞ ⎛ ∂u2 ⎞ ⎛ ∂u3 ⎞ ⎤


2 2 2

ε zz = + ⎢⎜ ⎟ ⎜+ ⎟ ⎜+ ⎟ ⎥, (3d)
∂z 2 ⎢⎣⎝ ∂z ⎠ ⎝ ∂z ⎠ ⎝ ∂z ⎠ ⎥⎦

∂u1 ∂u3 ∂u1 ∂u1 ∂u2 ∂u2 ∂u3 ∂u3


γ xz = + + + + , (3e)
∂z ∂x ∂x ∂z ∂ x ∂z ∂ x ∂z

∂u2 ∂u3 ∂u1 ∂u1 ∂u2 ∂u2 ∂u3 ∂u3


γ yz = + + + + . (3f)
∂z ∂y ∂y ∂z ∂y ∂z ∂y ∂z

Thus, the corresponding strain-displacement equations can be written in the following form:

ε xx = ε x,0 + z ( kx(0) + z kx(1) + z2kx(2) ) , (4a)

ε yy = ε y ,0 + z ( k y(0) + z k y(1) + z 2k y(2) ) , (4b)

γ xy = γ xy,0 + z ( kxy(0) + z kxy(1) + z2kxy(2) ) , (4c)

ε zz = ε z,0 + z kz(0) , (4d)

8
γ xz = γ xz,0 + z2 kxz(1) , (4e)

γ yz = γ yz,0 + z2 k yz(1) , (4f)

where the terms in equations (4) have to be determined. In particular, ε x ,0 , ε y ,0 , ε z ,0 , γ xy ,0 , γ xz ,0

and γ yz ,0 represent the middle surface strains, k x( 0) , k x(1) , k x( 2 ) , k y(0) , k y(1) , k y(2) and k z( 0) the

change in curvature in directions x, y and z respectively, and k xy(0) , k xy(1) , k xy(2) , k xz(1) and k yz(1) the

change in torsion.

By imposing the condition of zero shear stresses at the top and the bottom surfaces of the plate

[2], one obtains

γ xz z =± h /2 = 0 , γ yz z =± h /2 = 0 . (5a, b)

By collecting linear and cubic terms in z in equations (5a) and (5b), each into two separate

equations (since these expressions change sign if evaluated at z = ± h / 2 ), ψ 1 and ψ 2 can be

obtained as follows

1 ∂χ1 1 ∂χ1
ψ1 = − , ψ2 = − . (6a, b)
2 ∂x 2 ∂y

Moreover, equations (5a) and (5b) give for zero order and quadratic terms in z evaluated at

z=±h/2 the following expressions for γ 1 and γ 2

4 ⎛ ∂w h 2 ∂χ 2 ∂u ∂v ∂w ⎞
γ1 = − 2 ⎜
+ φ + + φ1 + φ2 + χ1 , (7a)
∂x ⎟⎠
1
3h ⎝ ∂x 4 ∂x ∂x ∂x

4 ⎛ ∂w h 2 ∂χ 2 ∂u ∂v ∂w ⎞
γ2 = − 2 ⎜ + φ2 + + φ1 + φ2 + χ1 . (7b)
3h ⎝ ∂y 4 ∂y ∂y ∂y ∂y ⎟⎠

At this point it is possible to obtain the expressions of the middle surface strains and changes in

curvature and torsion in equations (4). Substituting equations (1a-c) into (3), keeping non-linear

9
terms in u, v, w, φ1, φ2 and χ1 till order zero in z and then making use of equations (6) and (7), the

strain-displacement relations of the new theory can be obtained as follows:

∂u 1 ⎡⎛ ∂u ⎞ ⎛ ∂v ⎞ ⎛ ∂w ⎞ ⎤
2 2 2

ε x,0 = + ⎢⎜ ⎟ + ⎜ ⎟ + ⎜ ⎟ ⎥ , (8a)
∂x 2 ⎣⎢⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎦⎥

∂v 1 ⎡ ⎛ ∂u ⎞ ⎛ ∂v ⎞ ⎛ ∂ w ⎞ ⎤
2 2 2

ε y ,0 = + ⎢⎜ ⎟ + ⎜ ⎟ + ⎜ ⎥, (8b)
∂y 2 ⎢ ⎝ ∂y ⎠ ⎝ ∂y ⎠ ⎝ ∂y ⎟⎠ ⎥
⎣ ⎦

1 2
ε z ,0 = χ1 +
2
( φ1 + φ22 + χ12 ) , (8c)

∂u ∂v ∂u ∂u ∂v ∂v ∂w ∂w
γ xy ,0 = + + + + , (8d)
∂y ∂x ∂x ∂y ∂x ∂y ∂x ∂y

∂w ∂u ∂v ∂w
γ xz ,0 = φ1 + + φ1 + φ2 + χ1 , (8e)
∂x ∂x ∂x ∂x

∂w ∂u ∂v ∂w
γ yz ,0 = φ1 + + φ1 + φ2 + χ1 , (8f)
∂y ∂y ∂y ∂y

∂φ1
k x(0) = , (9a)
∂x

∂φ2
k y(0) = , (9b)
∂y

k z(0) = 2 χ2 , (9c)

∂φ1 ∂φ2
k xy(0) = + , (9d)
∂y ∂ x

1 ∂ 2 χ1
k x(1) = − , (9e)
2 ∂x 2

1 ∂ 2 χ1
k y(1) = − , (9f)
2 ∂y 2

10
∂ 2 χ1
k (1)
=− , (9g)
∂x∂y
xy

4 ⎛ ∂w ⎞
k xz(1) = − 2 ⎜
+ φ1 ⎟ , (9h)
h ⎝ ∂x ⎠

4 ⎛ ∂w ⎞
k yz(1) = − 2 ⎜
+ φ2 ⎟ , (9i)
h ⎝ ∂y ⎠

1 ∂2 χ2 4 ⎛ ∂ 2 w ∂φ1 ⎞
k x(2) = − − ⎜ + ⎟, (9j)
3 ∂x 2 3h 2 ⎝ ∂x 2 ∂x ⎠

1 ∂2 χ2 4 ⎛ ∂ 2 w ∂φ2 ⎞
k y(2) = − − ⎜ + ⎟, (9k)
3 ∂y 2 3h 2 ⎝ ∂y 2 ∂y ⎠

2 ∂2 χ2 8 ∂2w 4 ⎛ ∂φ ∂φ ⎞
k xy(2) = − − 2 − 2 ⎜ 1 + 2 ⎟. (9l)
3 ∂x∂y 3h ∂x∂y 3h ⎝ ∂y ∂x ⎠

In the present study, the acronym 2TS-NL is used to distinguish the new theory from the

conventional higher-order shear and thickness deformation theories, where 2TS stands for 2nd

order thickness stretching theory and NL refers to the presence of non-linearities in rotations and

thickness deformation. Equations (8a-f) give non-linear middle surface strains and equations (9a-

l) give the changes in curvature and torsion of the middle surface, obtained retaining only linear

terms. In fact, non-linear terms in the changes in curvature and torsion play a very small role, at

least for moderate deformations, and therefore are neglected. It is worth noting that the strain-

displacement relations of the conventional 2nd order thickness stretching theory 2TS [25] could

be obtained from equations (7) and (8) simply by removing the non-linearities in φ1, φ2 and χ1 .

Particularly, note that the difference between 2TS and 2TS-NL is the presence of additional non-

linear terms in ε z ,0 , γ xz ,0 , γ yz ,0 .

11
3. Kinetic and elastic strain energies of rectangular plates

The kinetic energy of an isotropic rectangular plate including rotary inertia can be obtained as
h/2 a b
1
Tp = ∫ ∫ ∫ ρ ( u12 + u22 + u32 ) d x d y d z , (10)
2 −h / 2 0 0

where overdot indicates differentiation with respect to time, ρ is the mass density of the plate

and u1, u2 and u3 are the displacements given by equations (1a) to (1c).

The elastic stain energy of the plate is given by


a b h /2
1
U p = ∫∫ ∫ (σ xxε xx +σ yyε yy +σ zzε zz +τ xyγ xy +τ xzγ xz +τ yzγ yz ) d z d y d x , (11)
2 0 0 −h /2

where σ ij and τ ij are normal and shear Kirchhoff stresses that can be obtained in terms of Green’s

strains by using the following three-dimensional constitutive relations [21]

⎧σ xx ⎫ ⎛ c11 c12 c12 0 0 0 ⎞ ⎧ε xx ⎫


⎪σ ⎪ ⎜ ⎟ ⎪ε ⎪
⎪ yy ⎪ ⎜ c12 c11 c12 0 0 0 ⎟ ⎪ yy ⎪
⎪⎪σ zz ⎪⎪ ⎜ c12 c12 c11 0 0 0 ⎟ ⎪⎪ε zz ⎪⎪
⎨ ⎬=⎜ ⎟⎨ ⎬, (12)
τ
⎪ ⎪ ⎜
yz 0 0 0 c22 0 0 ⎟ ⎪γ yz ⎪
⎪τ xz ⎪ ⎜ 0 0 0 0 c22 0 ⎟ ⎪γ xz ⎪
⎪ ⎪ ⎜ ⎟⎪ ⎪
⎪⎩τ xy ⎪⎭ ⎝ 0 0 0 0 0 c22 ⎠ ⎪⎩γ xy ⎪⎭

in which

c11 =
(1 − ν ) E , c12 =
νE
, c22 =
E
, (13 a-c)
1 − ν − 2ν 2
1 − ν − 2ν 2
2 (1 + ν )

where E is the Young modulus and ν is the Poisson ratio. By substituting equations (12) and (13)

into (11), the strain energy of the plate will be simplified to

12
{
ab
2(ν −1) (ε x2,0 + ε y2,0 + εz2,0 ) + ( 2ν −1) (γ xy2 ,0 + γ xz2 ,0 + γ yz2 ,0 )
Eh
Up = ∫∫
4( 2ν +ν −1) 0 0
2

− 4ν (εx,0ε y,0 + εx,0εz,0 + ε y,0ε z,0 ) +


h2 ⎡
12 ⎣
(
2(ν −1) kx(0) + ky(0) + kz(0) + 2εx,0 kx(1) + 2ε y,0 ky(1)
2 2 2

)
+ ( 2ν −1) kxy(0) + 2( 2ν −1) ( γ xy,0 kxy(1) + γ xz,0 kxz(1) + γ yz,0 kyz(1) ) − 4ν ( εx,0 ky(1) + ε y,0 kx(1) + ε z,0 ( kx(1) + ky(1) )
2

(14)
(0)
x
(0)
y
(0)
x
(0)
z
h4
(0)
y
80 ⎣
(0)
z (
+k k + k k + k k ) ⎦⎤ + ⎡2(ν −1) kx(1) + ky(1) + ( 2ν −1) kxy(1) + kxz(1) + kyz(1)
2 2

) ( 2 2 2

)
+ ( 4ν −1) ( kx(0) kx(2) + ky(0) ky(2) + kxy(0) kxy(2) ) − 4ν ( kx(0) ky(2) + ky(0) kx(2) + kx(1) ky(1) + kx(2) kz(0) + ky(2) kz(0) )

+
h6 ⎡
448 ⎣
(2 2

)
2(ν −1) kx(2) + ky(2) + ( 2ν −1) kxy(2) − 4ν kx(2) ky(2) ⎤ dx dy .
2

4. Boundary conditions and discretization

The boundary conditions are assumed to be simply supported immovable. Therefore, the

following geometric boundary conditions should be satisfied at the plate edges [25]

u = 0 , v = 0 , w = 0, φ2 = 0, χ 2 = 0, at x = 0, a, (15a-e)

u = 0 , v = 0 , w = 0, φ1 = 0, χ 2 = 0, at y = 0, b. (16a-e)

For this boundary condition the thickness deformation is not zero ( χ1 ≠ 0 ) at the edges.

Therefore, in order to satisfy the geometric boundary conditions, the following set of expansions

are considered

M N
u ( x , y , t ) = ∑∑ um ,n (t ) sin(2 mπ x / a ) sin( nπ y / b ) , (17a)
m =1 n =1

M N
v ( x , y , t ) = ∑∑ vm ,n ( t ) sin( mπ x / a ) sin(2 nπ y / b ) , (17b)
m =1 n =1

M N
w( x, y, t ) = ∑∑ wm ,n (t ) sin(mπ x / a) sin(nπ y / b) , (17c)
m =1 n =1

13
M N
φ1 ( x, y, t ) = ∑∑ φ1 (t ) cos( mπ x / a ) sin( nπ y / b) ,
m ,n
(17d)
m =1 n =1

M N
φ2 ( x, y , t ) = ∑∑ φ2 (t ) sin( mπ x / a ) cos( nπ y / b) ,
m ,n
(17e)
m =1 n =1

M N
χ 2 ( x, y , t ) = ∑∑ χ 2 (t ) sin(mπ x / a ) sin(nπ y / b) ,
m ,n
(16f)
m =1 n =1

where m and n are the number of half-waves in the x and y directions, respectively; t is time and

um,n (t ) , vm,n (t ) , wm,n (t ) , φ1 (t ) , φ2 (t ) and χ 2 (t ) are the generalized coordinates that are
m ,n m ,n m ,n

unknown functions of time. M and N indicate the number of terms in the expansion of in-plane

displacements and are usually greater than M and N, which are the required terms in the

expansion of transverse displacement, rotations and thickness deformation parameter χ 2 .

Since χ1 is not zero at the edges, its necessary trigonometric terms will be obtained by evaluating

ε zz in the equivalent plane-stress problem. In fact, in the generalized plane-stress state, ε zz is

ν
ε zz = −
1 −ν
(ε x ,0 + ε y ,0 ) + O ( z ). (18)

After replacing the right hand side of equation (18) by equations (8a, b) and (17a-c), collecting

zero order terms in z and then using equation (8c), one can find that χ1 should include the

following necessary trigonometric terms in case of immovable plates

M N
χ1 ( x, y , t ) = ∑∑ χ1(1) (t ) cos( mπ x / a ) cos( nπ y / b)
m ,n
m =0 n =0
M N M N
(19)
+ ∑∑ χ (2)
1m , n (t ) cos( mπ x / a ) sin( nπ y / b) + ∑∑ χ (3)
1m , n (t )sin(mπ x / a ) cos( nπ y / b) ,
m =1 n =1 m =1 n =1

where M and N are the number of cos( mπ x / a ) cos( nπ y / b) terms in the expansions of χ1 .

14
5. Virtual work and Lagrange equations of motion

In the presence of transverse concentrated harmonic force of magnitude f applied at the center

of the plate (a/2, b/2) and uniform pressure P , both positive in z direction, the virtual work done

by the external forces will be


b a b a
W = ∫ ∫ f wδ ( x − a / 2)δ ( y − b / 2) cos(Ω t ) d x d y + ∫ ∫ p w d x d y
0 0 0 0
(20)
⎡ M N
⎤ 4a b M N w2i −1,2 j −1
= f cos(Ω t ) ⎢ ∑∑ (−1)i +1 (−1) j +1 w2i −1,2 j −1 ⎥ + 2 P ∑∑ ,
⎣ i =1 j =1 ⎦ π i =1 j =1 ( 2i − 1)( 2 j − 1)

where Ω is the excitation frequency and δ is the Dirac delta function. In equation (20) it has

been assumed that the pressure maintains the vertical direction and does not follow the

deformation of the plate; also it is assumed that there is no change in the area of application of

the pressure (in case of deflection, the middle surface of the plate gets stretched and the

application of the same pressure gives a larger force resultant due to the surface increase). These

approximations are correct only for moderate deflections of the plate.

The following notation is introduced for brevity

{
q = um,n / h, vm,n / h, wm,n / h, φ1m ,n , φ2m ,n ,
(21)
}
T
χ1 , χ 2 h
m ,n m ,n
for m = 0,…, M or M or M , n = 0,…, N or N or N .

It can be observed that there is a big difference in the order of magnitude of independent

variables (e.g. wm,n is of O(h) whereas χ 2m ,n is of O(h-1)) which can result in ill-conditioned mass

matrix. In order to avoid matrix ill-conditioning, variables are made dimensionless with respect

to the plate thickness h and a high numerical accuracy is used.

The generic element of the time-dependent vector q is referred to as qj; the dimension of q is

N max , which is the number of degrees of freedom (dofs) used in the mode expansion.

15
The Lagrange equations of motion are

d ⎛ ∂ TP ⎞ ∂ TP ∂ U P ∂ W
⎜ ⎟⎟ − + = , j = 1,… , N max . (22)
d t ⎜⎝ ∂ q j ⎠ ∂ q j ∂ q j ∂ q j

Equation (22) gives a system of non-linear ordinary differential equations with quadratic and

cubic non-linear terms as follows

Mq+Cq +[K+N 2 (q)+N 3 (q,q)]q=p+f cos(Ω t) , (23)

where M is a non-diagonal mass matrix as a result of inertia coupling in the equations of motion.

C is the viscous damping matrix which is added to the equations of motion to describe

dissipation, K is the linear stiffness matrix, N2 gives the quadratic non-linear stiffness terms and

N3 represents the cubic non-linear terms. Moreover, p and f are vectors containing the

contribution of the generalized forces of the distributed pressure p and transverse harmonic force

f , respectively.

In order to perform numerical integration, equation (23) is multiplied by the inverse of mass

matrix and then recast into first-order ordinary differential equations by introducing the dummy

vector y as follows

q = y,
(24)
y = -M -1Cy - [M -1K + M -1 N 2 (q ) + M -1 N 3 (q, q)]q + M -1 p + M -1f cos(Ω t ),

-1
where M C is the dissipation term, which is assumed to be given by

⎡2ω1 ζ 1 ... 0 ⎤
⎢ ⎥
M-1C = ⎢ ⎥. (25)
⎢ 0 ... 2ωN max ζ Nmax ⎥⎦

16
In equation (25) ωi are the natural frequencies and ζ i the corresponding modal damping ratios of

each generalized coordinate. Matrix (25) is assumed to be diagonal in order to use modal

damping.

6. Results and discussion

The equations of motion have been obtained by using Mathematica software to perform

analytical surface integrals of trigonometric functions. Moreover, 100 digits of accuracy has been

used to avoid matrix ill-conditioning.

Calculations have been performed for thin (h=0.001m) and thick (h=0.01m) aluminium

rectangular plates with a=b=0.1 m, E=70 GPa, ν=0.3, ρ=2702 Kg/m3. In order to carry out the

non-linear analyses, a base of 37 dofs has been used for the conventional higher-order shear

deformation theory (HSDT) that neglects thickness deformation. This model includes: w1,1 , w1,3,

w3,1 , w3,3 , w1,5 , w5,1 , w3,5 , w5,3 , w5,5 , u2,1 , u4,1 , u6,1 , u2,3 , u4,3 , v1,2 , v1,4 ,v1,6 , v3,2 , v3,4 , φ11,1 ,

φ1 , φ1 , φ1 , φ1 , φ1 , φ1 φ1 , φ1 , φ2 , φ2 , φ2 , φ2 , φ2 , φ2 , φ2 , φ2 , φ2 . In Refs [25] and


1,3 3,1 3,3 1,5 5,1 3,5 5,3 5,5 1,1 1,3 3,1 3,3 1,5 5,1 3,5 5,3 5,5

[28] it was shown that a numerical model with a base of 15 dofs that included one transverse

generalized coordinate, two rotations and twelve in-plane modes gives acceptable convergent

results for HSDT and for immovable plates. In the present study, in order to increase the

accuracy, the number of dofs has been increased to 37. Accordingly, for the 2TS and 2TS-NL

theories the number of dofs become 60 including the same base of 37 dofs plus χ1(1)
0 ,0
, χ1(1)
0 ,2
, χ1(1)
2 ,0

, χ1(1)
2 ,2
, χ1(22,1) , χ1(42,1) , χ1(62,1) , χ1(22,3) , χ1(42),3 , χ1(3)
1,2
, χ1(3)
1,4
, χ1(3)
1,6
, χ1(3)
3,2
, χ1(3)
3,4
, χ21,1 , χ 21,3 , χ 23,1 , χ 23,3 , χ 21,5 ,

χ2 , χ2 , χ2 , χ2 .
5,1 3,5 5,3 5,5

17
Here it should be noted that, unlike theories that consider linear normal displacement field and

consequently uniform normal strain, the new theories (2TS and 2TS-NL) present very accurate

evaluation of the natural frequencies even for anisotropic materials when 3D material laws are

used. In fact, for FGM plates it was observed in [25] that accurate natural frequencies are not

obtained by using theories that consider linear normal displacement field, due to the fact that

normal strain varies through the thickness for anisotropic materials.

The non-linear equations of motion have been integrated by using the software AUTO [29] for

continuation and bifurcation analysis by using the pseudo-arclength continuation and collocation

method, starting at zero pressure and zero external harmonic force from the trivial solution

(undeformed state). Before the bifurcation analysis begins, the equations of motion are made

dimensionless. In particular, the displacements are divided by the thickness of the plate, h, and

the transverse normal coordinate χ2 is multiplied by h. The time is divided by the period of the

f
resonant mode ω1,1 and the non-dimensional force and pressure are defined as f = and
hω μ
2
1,1 1,1

P a4
p= , where μ1,1 is the modal mass of mode (1, 1). The continuation begins by considering
E h4

the uniform external pressure amplitude ( P ) as the bifurcation parameter. The solution starts

from the un-deformed state of the plate at P = f = 0 and is incremented to obtain the deformed

state due to the external pressure. At the end of the first stage, the non-linear static behaviour due

to the external pressure is obtained. The vibration analysis begins over the statically bent

configuration and by considering the amplitude of the harmonic force ( f ) as the bifurcation

parameter at fixed excitation frequency far from resonance. Once the desired force amplitude is

18
reached, the bifurcation continues by spanning the excitation frequency Ω in the neighbourhood

of the resonant frequency to obtain the frequency-amplitude response of the plate.

The non-linear static deformation of the thin plate with immovable simply supported boundary

conditions is shown in Figure 2. In particular, Figure 2(a) compares the non-linear deflection of

the center of the plate versus the applied pressure when different theories are used. Included in

the Figure are also the results obtained by using the finite element software ANSYS via 3D

modeling and by using SOLID 185 element. The solution provided by ANSYS for linear elastic

materials is accurate only for relatively small strains; a convergence analysis has been performed

by increasing the number of elements both in-plane and in the thickness direction. Table 1 shows

the convergence rate of the mid-plane deflection obtained by ANSYS versus the number of

elements used for mesh generation for the thick aluminum plate. It could be seen that the mesh

generated by 40 elements in-plane directions and 10 elements in the thickness could yield

acceptable result. Thus, in order to obtain the non-linear deflection and thickness deformation

with accuracy, the 3D plate has been meshed by at least 16000 elements. Moreover, the

boundary conditions were applied only on the nodes of the mid-plane. Figure 2(a) shows that all

theories (HSDT, 2TS and 2TS-NL) give similar non-linear deflection curves practically identical

to ANSYS-3D for large deflections (w≈15h). Figure 2(b) shows that unlike central deflection, for

the maximum rotation φ1 at x=0, y=b/2, the result obtained from 2TS-NL is different from HSDT

and 2TS. It is evident that for large deflections (w≥5h) the curve of HSDT and 2TS are diverging

from the ANSYS-3D solution while 2TS-NL is almost identical to the finite element result. As

evidenced by Figure 2(b), non-linearities in rotation could play an important role in the

convergence of rotational coordinates for very large deflections. Note that numerical simulations

are performed assuming linear material and neglecting the possibility of having large strains and

19
therefore hyperelastic behavior; hyperelasticity could become necessary to model soft materials

for very large deflections.

Table 2 summarizes the results of Figure 2(a, b) and has been provided in order to have a better

comparison between different theories and the results obtained by ANSYS.

Figures 3 show the parameters related to thickness deformation of the thin plate subjected to

uniform pressure. It is interesting to see that the results obtained from conventional 2TS are very

different from 2TS-NL that retains non-linearities in rotation and thickness deformation. The

results are particularly different for χ1(1)


m ,n
which are the generalized coordinates corresponding to

cos( mπ x / a ) cos( nπ y / b) in the thickness deformation response. Here it should be noted that the

presence of χ1(1)
m ,n
terms are necessary in the numerical analysis in order to obtain convergent

results [25]. Moreover, it is worth noting that it is only through the non-linear analysis that

thickness variation coordinates are activated. The reason is because in linear analysis and for

isotropic materials, the transverse deflection is only coupled to rotations and χ2 which play no

role in obtaining the thickness stretching of the plate. Therefore, in order to obtain Δ h , a non-

linear analysis should be carried out. Figure 3(e) shows the thickness deformation of the plate

along y=b/2. The large difference for thickness deformation parameters χ1(1)
m ,n
that was observed

in Figures 3(a) to 3(d) gives a large difference in the evaluation of the thickness deformation. In

fact, it can be seen that even though the thickness deformation at the centre of the plate is the

same for all the theories, far from the centre and close to the edges this deformation is very

different when non-linearities are retained in rotations and thickness deformation. In particular,

the result obtained from conventional HSDT and 2TS is very different from ANSYS-3D while

2TS-NL predicts correct thickness deformation along the plate. Note that the small difference

20
that is seen between ANSYS-3D and 2TS-NL close to the edges is due to the fact that the

maximum number of in-plane half-waves that is considered in the analysis is 6. In fact, in order

to obtain more accurate results at the edges this number should be greatly increased.

Similar analysis has been performed for the thick immovable plate and its corresponding results

are shown in Figures 4 and 5. In particular, Figures 4(a) and 4(b) show the maximum deflection

and rotation φ1 at the centre and the edge, respectively. Here, it can be seen that even though for

small deflection all the curves are identical, for w≥h the curve of 2TS-NL is more similar to

ANSYS-3D. The same scenario could be perceived for the rotation as the 2TS-NL result is

converging to ANSYS-3D.

The thickness deformation and the variation of normal displacement at the centre and close to the

edge for the thick plate are shown in Figures 5(a) to 5(c), respectively. Once again it appears

from Figure 5(a) that only 2TS-NL is globally predicating the thickness deformation correctly. In

fact, for 2TS and HSDT the maximum thickness variation is found in the centre. Conversely, for

2TS-NL and ANSYS-3D thickness stretching is minimum at the centre and maximum close to

the edges. Moreover, Figures 5(b) and 5(c) show that even though 2TS is predicating the normal

displacement through the thickness correctly at the centre, it gives incorrect distribution close to

the edge. In contrast, the difference between the normal displacement predicted by ANSYS and

2TS-NL close to the edge is almost negligible. Note that the results depicted in Figures 5

correspond to static deflection of the order of plate thickness (w ≈ h) and for non-dimensional

pressure p=50.

In case of non-linear forced vibrations, comparisons between 2TS and HSDT with a reference

solution provided by Shen [30] in case of immovable functionally graded (FG) rectangular plates

21
has previously been performed by the same authors in Ref [25] showing close agreement and

thus are not repeated here.

The non-linear forced vibration response of the thick aluminium square plate is shown in Figures

6 and 7 using HSDT, 2TS and 2TS-NL theories and for damping ratio equal to 0.01. In

particular, Figure 6 shows the frequency-response curves of the flat plate while Figure 7 displays

the effect of initial pressure on the non-linear forced vibration curves. In Figure 6, the plate is

assumed to be under non-dimensional concentrated harmonic force f=0.004 applied at the centre.

It is evident from Figures 6(a-f) that 2TS-NL presents weaker hardening response comparing to

HSDT and 2TS. It can be also observed that close to the resonance peak at Ω =1.55ω1,1, the

difference between 2TS and 2TS-NL is significant for the thickness deformation coordinates (see

Figs 6(e) and 6(f)). Moreover, the additional peak that is seen in the response of other

generalized coordinates close to Ω =1.55ω1,1 is due to a 3:1 internal resonance between the

fundamental mode and mode (1, 3) and a 5:1 internal resonance between modes (1, 1) and (3, 3).

Note that the responses obtained by using 2TS and HSDT for the flat plate are practically

identical.

Non-linear forced vibration response of the thick pressurized plate is shown in Figure 7 for p=50

and f=0.0843. In this case, it is assumed that the plate is primarily subjected to uniform pressure

and the non-linear forced vibration response is obtained over the initial deformed state due to

uniform pressure. Figure 7 shows that the presence of static pressure gives a significant

qualitative change in the trend of non-linearity, changing the behaviour from hardening to

softening. The reason for this change of behaviour is because pressurized plates are no more flat

and have a shape close to a shallow shell and therefore present larger natural frequencies. Figure

7 is obtained by normalizing the excitation frequency with respect to the fundamental natural

22
frequency of the pressurized plate and subtracting the value of the static deflection due to the

pressurization. It can be observed that similar to the flat plate, 2TS-NL is predicting weaker non-

linearity. Moreover, as a result of different static response, as evidenced by Figures 4 and 5, the

dynamic state in 2TS-NL begins from a different deflection, rotation and thickness deformation

value. In other words, the static deformation, specifically thickness deformation coordinates, are

so different in 2TS-NL that they greatly affect the non-linear frequency response curves even-

though the global behaviour predicted by both theories are the same.

7. Conclusions

Geometrically non-linear static and vibrations of rectangular plates are studied by considering

full non-linear terms associated with Green-Lagrange strain-displacement relations and using 2nd

order thickness deformation theory. In particular, in addition to non-linearities in membrane and

transverse deflection, non-linear terms associated with rotations and thickness deformation

parameters are included. Great attention has been given to obtain the non-linear response of

rotations and thickness deformation parameters. It has been shown that the theory with full non-

linearities leads to a thickness deformation and rotation response that is identical to 3D modeling

in ANSYS and different from the previous 3D theories which neglect these geometric non-

linearities. Particularly, very different results are obtained for thickness deformation parameters

and it is found that only the new theory predicts the stretching profile correctly along the plate.

Moreover, it is revealed that the new theory gives weaker hardening behavior comparing to

conventional higher-order and thickness deformation theories in case of flat plates and weaker

softening behavior in case of pressurized plates. It is believed that the utilization of this theory

could be necessary for structural elements that incorporate high-thickness deformation and

23
require accurate modeling of the variation of normal displacement through the thickness such as

hyperplastic materials and biological soft tissues.

Acknowledgments

The authors acknowledge the financial support of the NSERC Discovery Grant, Canada

Research Chair Program and the NPRP program of Qatar National Research Foundation

(QNRF).

References
[1] J.N. Reddy, Mechanics of Laminated Composite Plates and Shells: Theory and Analysis, 2nd

edition, CRC Press, Boca Raton, FL, USA, 2004.

[2] M. Amabili, Nonlinear Vibrations and Stability of Shells and Plates, Cambridge University

Press, New York, USA, 2008.

[3] J.N. Reddy, R. A. Arciniega, Shear deformation plate and shell theories: From Stavsky to

present, Mechanics of Advanced Materials and Structures 11 (2004), 535-582.

[4] E. Carrera, Theories and finite elements for multilayered, anisotropic, composite plates and

shells, Archives of Computational Methods in Engineering, 9 (2002) 87-140.

[5] E. Carrera, Theories and finite elements for multilayered, anisotropic, composite plates and

shells: A unified compact formulation with numerical assessment and benchmarking,

Archives of Computational Methods in Engineering, 10 (2003) 215-296.

[6] A.v. Krishna Murthy, Higherorder theory for vibration of thick plates, AIAA Journal, 18

(1977) 1823-1824.

[7] K.H. Lo, R.M Christensen, E.M. Wu, A higher-order theory of plate deformation-part 1:

homogeneous plates. Journal of Applied Mechanics , 44 (1977) 663-668.

24
[8] K.H. Lo, R.M Christensen, E.M. Wu, A higher-order theory of plate deformation-part 2:

laminated plates. Journal of Applied Mechanics , 44 (1977) 669-676.

[9] J.N. Reddy, A simple higher-order theory of laminated composite plates, Journal of Applied

Mechanics, 51 (1984) 745-752.

[10] J.N. Reddy, A general non-linear third-order theory of plates with moderate thickness,

International Journal of Non-Linear Mechanics, 25 (1990) 677-686.

[11] G.N. Praveen, J.N. Reddy, Nonlinear transient thermoelastic analysis of functionally

graded ceramic-metal plates, International Journal of Solids and Structures, 35 (1998) 4457-

4476.

[12] F. Alijani, M. Amabili, Non-linear dynamic instability of functionally graded plates in

thermal environments, International Journal of Non-Linear Mechanics, 50 (2013) 109-126.

[13] F. Alijani, M. Amabili, Theory and experiments for nonlinear vibrations of imperfect

rectangular plates with free edges, Journal of Sound and Vibtration, 332 (2013) 3564-3588.

[14] J.N. Reddy, A generalization of two dimensional theories of laminated composite plates,

Communications in Applied Numerical Methods, 3 (1987) 173-180.

[15] M. Di Sciuva, An improved shear deformation theory for moderately thick multilayered

anisotropic shells and plates, Journal of Applied Mechanics, 54 (1987) 589-596.

[16] E. Carrera, Historical review of Zig-Zag theories for multilayered plates and shells,

Applied Mechanics Reviews, 56 (2003) 287-308.

[17] E. Carrera,S. Brischetto, Analysis of thickness locking in classical, refined and mixed

multilayered plate theories, Composite Structures 82 (2008) 549-562.

[18] E. Carrera, S. Brischetto, Analysis of thickness locking in classical, refined and mixed

theories for layered shells, Composite Structures 85 (2008) 83-90.

25
[19] A.M.A. Neves, A.J.M Ferreira,E. Carrera,C.M.C. Roque,M. Cinefra,R.M.N Jorge,

C.M.M Soares. A quasi-3D sinusoidal shear deformation theory for the static and free

vibration analysis of functionally graded plates, Composites: Part B, 43 (2012) 711-725.

[20] A.M.A. Neves, A.J.M Ferreira,E. Carrera,C.M.C. Roque,M. Cinefra,R.M.N Jorge,

C.M.M Soares. A quasi-3D hyperbolic shear deformation theory for the static and free

vibration analysis of functionally graded plates. Composite Structures 94 (2013)1814-1825.

[21] J.N. Reddy, A general nonlinear third-order theory of functionally graded plates,

International Journal of Aerospace and Lightweight Structures, 1 (2011) 1-21.

[22] J.N. Reddy,J. Kim, A nonlinear modified couple stress-based third-order theory of

functionally graded plates, Composite Structures, 94 (2012) 1128-1143.

[23] M. Amabili, A new nonlinear higher-order shear deformation theory with thickness

variation for large-amplitude vibrations of laminated doubly curved shells, Journal of Sound

and Vibration 332 (2013) 4620-4640.

[24] M. Amabili, A non-linear higher-order thickness stretching and shear deformation theory

for large-amplitude vibrations of laminated doubly curved shells, International Journal of

Non-linear Mechanics, 58 (2014) 57-75.

[25] F. Alijani, M. Amabili, Effect of thickness deformation on large-amplitude vibrations of

functionally graded rectangular plates, Composite Structures, 113 (2014) 89-107.

[26] P. Dash, B.N. Singh, Geometrically nonlinear bending analysis of laminated composite

plate, Communications in Nonlinear Science and Numerical Simulation, 15 (2010) 3170-

3181.

[27] P.Dash, B.N. Singh, Buckling and post-buckling of laminated composite plates,

Mechanics Research Communications, 46 (2012) 1-7.

26
[28] M. Amabili, S. Farhadi, Shear deformable versus classical theories for nonlinear

vibrations of rectangular isotropic and laminated composite plates, Journal of Sound and

Vibration 320 (2009) 649-667.

[29] E.J. Doedel, A.R. Champneys, T.F. Fairgrieve, Y.A. Kuznetsov, B. Sandstede, X. Wang,

AUTO 97: Continuation and Bifurcation Software for Ordinary Differential Equations (with

HomCont), Concordia University, Montreal, Canada, 1998.

[30] H. S. Shen, Functionally Graded Materials, Nonlinear Analysis of Plates and Shells. Florida:

CRC Press, 2009.

27
Table 1. Maximum mid-plane deflection of thick aluminium square plate (a=0.1m, h=0.01m)
versus the number of elements used in ANSYS -3D using SOLID 185 element for dimensionless
pressure p=50.

Number of
Elements
w(x=a/2 , b/2) /h
(x×y×z)
20×20×2 1.038
40×40×4 1.055
80×80×8 1.076
40×40×10 1.078
100×100×10 1.082

28
Table 2. Comparison between different theroeis and ANSYS-3D in predicting the deflection

and rotation response of of the thin aluminium square plate (a=0.1m, h=0.001m) subjected to

uniform pressure.

P a4
Dimensionless pressure ( p = )
E h4
Solution Method 142.86 1428.6 7142.86 14286 71428.6 142860
2TS-NL 1.34 3.28 5.53 6.98 11.95 15.05
w(x=a/2 , b/2) /h 2TS 1.34 3.28 5.53 6.98 11.94 15.04
ANSYS-3D 1.33 3.21 5.52 6.97 12.14 15.61
2TS-NL -0.045 -.012 -0.208 -0.26 -0.411 -0.485
φ1 (x=0,y=b/2) 2TS -0.045 -0.12 -0.213 -0.268 -0.45 -0.56
ANSYS-3D -0.046 -0.119 -0.211 -0.266 -0.42 -0.484

29
z b
y
a

x h

Fig. 1. Geometry of the rectangular plate.


Mid-plane

30
16
ANSYS-3D
14

w (x=a/2, b/2 ) /h 12

10 HSDT, 2TS & 2TS-NL

0
0 5 10 15
4 4 4
P a /(E h ) x 10 (a)

-0.1

-0.2
φ 1 (x=0,y=b/2 )

2TS-NL
-0.3

-0.4

ANSYS-3D
-0.5

-0.6 HSDT & 2TS


0 5 10 15
4 4 4
P a /(E h ) x 10 (b)

Fig. 2. Nonlinear static response of thin aluminium square plate (a=0.1m, h=0.001m) subjected
to uniform pressure. (a) Maximum transverse deflection w(x=a/2 , b/2) /h ; (b) maximum rotation
φ1 (x=0,y=b/2) .

31
0 0

-0.002 -0.002

2TS
-0.004 -0.004 2TS
Max (χ 11 0,0 )

Max (χ 11 0,2 )
)

)
(

-0.006

(
-0.006

-0.008 2TS-NL
-0.008

-0.01 2TS-NL
-0.01

-0.012 -0.012
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
4 4
P a /(E h ) 4 4
(a) P a /(E h )
(b)

32
0.035

0.03

0.025
2TS-NL

Max (χ 11 2,2 )
0.02

)
( 0.015

0.01 2TS

0.005

0
0 1000 2000 3000 4000 5000
4 4
P a /(E h ) (c)

0.018

0.016 2TS-NL

0.014

0.012
Max ( χ 12 2,1 )

2TS
)

0.01
(

0.008

0.006

0.004

0.002

0
0 1000 2000 3000 4000 5000
4 4
P a /(E h ) (d)

33
-3
x 10
-1

-2
HSDT & 2TS
-3 ANSYS-3D

-4

-5
Δ h /h 2TS-NL
-6

-7

-8

-9

-10
0 0.2 0.4 0.6 0.8 1
x /a
(e)
Fig. 3. Thickness deformation in thin Aluminium square plate (a=0.1m, h=0.001m) subjected to
uniform pressure. (a) Maximum of χ1(1) 0,0
; (b) maximum of χ1(1)
0,2
; (c) maximum of χ1(1)
2,2
; (d)
maximum of χ1(2)
2,1
; (e) thickness stretching ( Δh h ) at y=b/2 and non-dimensional pressure
p = 1428.6 .

1.6
ANSYS-3D
1.4
2TS-NL
1.2
/h
(x=a/2, b/2 )

1 HSDT & 2TS

0.8
w

0.6

0.4

0.2

0
0 50 100 150
4 4
P a /(E h ) (a)

34
0

-0.1

-0.2
φ 1 (x=0,y=b/2 )

2TS-NL
-0.3

-0.4
ANSYS

-0.5
HSDT & 2TS

0 50 100 150
4 4
P a / (E h ) (b)

Fig. 4. Nonlinear static response of thick aluminium square plate (a=0.1m, h=0.01m) subjected
to uniform pressure. (a) Maximum deflection w(x=a/2 , b/2) /h ; (b) maximum rotation φ1 (x=0,y=b/2) .

-0.01

-0.02

-0.03
Δ h /h

-0.04 ANSYS
HSDT
-0.05 2TS
2TS-NL
-0.06

-0.07
0 0.2 0.4 0.6 0.8 1
x/a
(a)

35
0.5
ANSYS
0.4
2TS
0.3 2TS-NL

0.2

0.1

z /h
0

-0.1
-0.2

-0.3
-0.4

-0.5
-20 -15 -10 -5 0 5
(u3-w)/h -3
x 10 (b)

0.5
0.4 ANSYS
2TS
0.3 2TS-NL
0.2
0.1
z/h

0
-0.1
-0.2
0.3
-0.4
-0.5
-0.015 -0.01 -0.005 0 0.005 0.01
(u3-w)/h
(c)
Fig. 5. (a) Thickness deformation in thick aluminium square plate when p = 50 ; (b) variation of
the normal displacement across the thickness at x/a=y/b=0.5, non-dimensional pressure p = 50 ;
(c) variation of the normal displacement across the thickness at x/a=y/b=0.1, p = 50 .

36
1.2

1
2TS-NL
/h)

0.8
2TS
1,1
Max (w

0.6 HSDT

HSDT & 2TS


0.4

0.2

0
0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
Ω/ω
1,1 (a)

0.04

0.035

0.03 2TS-NL

0.025
Max (w 1,3 / h)

2TS

0.02

0.015 HSDT

0.01

0.005

0
0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
Ω/ω
1,1 (b)

37
0.12

0.1

0.08 0.1

/ h)
2TS

3,3 0.06
Max (w
0.05 2TS-NL
HSDT
0.04
0
1.45 1.5 1.55
0.02

0
1 1.2 1.4 1.6
Ω/ω
1,1 (c)

0.35

0.3
2TS-NL
0.25
)
1 1,1

0.2
HSDT & 2TS
Max ( φ

0.15

0.1

0.05

0
1 1.2 1.4 1.6
Ω/ω
1,1 (d)

38
0.045

0.04
0.035
0.04
0.03 2TS-NL
0.03

Max ( χ (1) )
0.025

10,2
0.02
0.02 2TS
0.015 0.01
0.01 0
1.54 1.545 1.55
0.005
0

-0.005
1 1.2 1.4 1.6
Ω/ω
1,1 (e)

0.5

0.4
Max ( χ 12,2 )
(1)

0.3

2TS-NL
0.2
2TS
0.1

0
1 1.2 1.4 1.6
Ω / ω 1,1
(f)

Fig. 6. Nonlinear forced vibration response of thick aluminium square plate with immovable
edges (a=0.1, b=0.1 h=0.01) using different shear deformation theories. Dimensionless force is
0.004 and damping ratio is 0.01 for all generalized coordinates. (a) Maximum of w1,1/h ; (b)
maximum of w1,3/h; (c) maximum of w3,3/h;(d) maximum of φ11,1 ; (e) maximum of χ1(1)
0,2
; (f)

maximum of χ1(1)
2,2
.

39
2TS-NL
1.6

1.5

/h )
1,1 1.4 2TS
Max (w

1.3

1.2

1.1

0.9 0.95 1 1.05


ω /ω
1,1 (a)

-0.01

-0.02 2TS

-0.03
Max (χ 1 0,2 )
(1)

-0.04

-0.05

-0.06 2TS-NL

-0.07

-0.08
0.9 0.95 1 1.05
ω /ω 1,1
(b)

41
0.5

0.45 2TS-NL

0.4

0.35

Max ( χ (1) )
1 2,2
0.3

0.25
2TS
0.2

0.15

0.1
0.9 0.95 1 1.05
ω /ω
1,1 (c)

Fig. 7. Nonlinear forced vibration response of pressurized thick aluminium square plate with
immovable edges (a=0.1, b=0.1 h=0.01; p=50, f=0.0843; ζ=0.01. (a) Maximum of w1,1/h ; (b)
maximum of χ1(1)
0,2
; (f) maximum of χ1(1)
2,2
.

Highlights
• Geometrically non-linear static bending and vibrations of rectangular plates are studied.
• In addition to non-linearities in membrane and transverse deflection, non-linear terms
associated with rotations and thickness deformation are included.
• Results reveal that the theory with full geometric non-linearities provides significant
accuracy improvement for rotational and thickness deformation parameters.

42

You might also like