Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
By
DEBABRATA PODDER
ROLL NO.: 000810402005
REGN. NO.: 81964 of 2001 - 2002
EXAM. ROLL NO.: M4CIV 10-05
At the
Department of Civil Engineering
Faculty of Engineering and Technology
Jadavpur University
Kolkata – 700 032
DEPARTMENT OF CIVIL ENGINEERING
FACULTY OF ENGINEERING AND TECHNOLOGY
JADAVPUR UNIVERSITY
KOLKATA – 700 032
CERTIFICATE OF RECOMMENDATION
This is to certify that the thesis titled, “A Study on the Dynamic Instability of Cylindrical
Shell due to Parametric Excitation”, that is being submitted by Debabrata Podder
(Roll no. 000810402005) to Jadavpur University for the partial fulfillment of the
requirements for awarding the degree of Master of Civil Engineering (Structural Engineering)
is a record of bona fide research work carried out by him under my direct supervision &
guidance.
The work contained in the thesis has not been submitted in part or full to any other university
or institution or professional body for the award of any degree or diploma.
Countersigned by
_______________________ ________________________
Prof. S. Chakrabarti Prof. N. Chakraborti
Head of the Department Dean, FET
Department of Civil Engineering Jadavpur University
Jadavpur University Kolkata 700032
Kolkata 700032
(ii)
DEPARTMENT OF CIVIL ENGINEERING
FACULTY OF ENGINEERING AND TECHNOLOGY
JADAVPUR UNIVERSITY
KOLKATA – 700 032
CERTIFICATE OF APPROVAL
2.
(iii)
ACKNOWLEDGEMENT
_______________________________
Debabrata Podder
Jadavpur University, Kolkata M.C.E. (Structural Engineering)
Date: Roll No.:- 000810402005
Regn. No.:- 81964 of 2001-’02
Exam. Roll No.:- M4CIV 10-05
(iv)
ABSTRACT
Oscillatory instability of shell structures has been a major cause of concern in many
branches of engineering. The dynamic instabilities are the result of pulsating time varying
loads mainly inertial or thermal in origin. The greatest danger posed by such instabilities is
that the failure is very quick and abrupt. In parametric instability the rate of increase in
amplitude is generally exponential and thus potentially dangerous, while in typical resonance
due to external excitation the rate of increase in response is linear. More over damping
reduces the severity of typical resonance, but may only reduce the rate of increase during
parametric resonance. Parametric instability occurs over a region of parameter space and not
at discrete points. It may occur due to excitation at frequencies remote from the natural
frequencies. Researchers have worked in understanding the behavior of such instabilities.
With this particular concept in mind, a theoretical formulation has been developed in
the present work for the analysis of singly curved surfaces subjected to in plane periodic
loading and undergoing parametric excitation. A four noded iso-parametric shell element
having five mechanical degrees of freedom per node, using Mindlin and Reissener’s shallow
shell theory has been developed in MATLAB platform. The first order shear deformation and
effect of rotary inertia has been considered. The results obtained by the present FE code for
static, free vibration and buckling analysis are verified with the ANSYS finite element
software. Parametric instability studies have been carried out for cylindrical shells having
different fibre orientations, various geometric properties with different R/a ratio. A
generalized Rayleigh proportional damping has been considered for all the cases to study the
shift of the stability point with respect to frequency ratio in various cases. The obtained
results are discussed in detail and conclusions highlighting the important findings are made.
(v)
CONTENTS
CERTIFICATE OF RECOMMENDATION II
ACKNOWLEDGEMENT IV
ABSTRACT V
SYMBOLS IX-X
1.4 SHELLS 7
1.4.1 Shell as a Structural Form 7
1.4.2 Parametric Excitation on Shell Structure 7
(vi)
CHAPTER-2: CONSTITUTIVE EQUATIONS 14-25
2.1 INTRODUCTION 14
3.1 INTRODUCTION 26
4.1 INTRODUCTION 44
(vii)
4.3 FREE VIBRATION ANALYSIS 47-50
4.3.1 Isotropic cantilever shell 47-48
4.3.2 Composite cantilever shell 48-50
REFERENCES 69-72
(viii)
SYMBOLS
{} Curvatures
{N} Force resultant vector (N.m-1)
{M} Moment resultant vector (N.m.m-1)
T Kinetic energy of the system (N.m)
U Potential energy of the system (N.m)
[B] Strain-displacement matrix
[K] Elastic stiffness matrix
[K ] Geometric stiffness matrix
[Z] Position matrix
[N] Shape function matrix
[J] Jacobian matrix
[M] Mass matrix
{p} Mechanical load vector (N)
{U} Displacement vector (m)
{U} Velocity vector (m.s-1)
(ix)
[T] Transformation matrix
[C] Damping matrix
L Length of the shell (m)
a Width of the shell
R Radius of curvature of shell (m)
n Natural frequency (Rad.s-1)
[ ] Inertia matrix
(x)
LIST OF FIGURES
(xi)
for fibre orientation 45/0/45 and R/a=50, L=2m, a=1m,
t=0.03m for E-Glass Epoxy composite. 52
4.9 Stability plot for cylindrical cantilever Aluminum shell with and without
damping (R/a = 25, L=2.0 m,a=1.0 m, thickness = 0.03 m,
E = 70 GPa, ν = 0.3) 54
4.12 Stability plots for different fibre orientations 90/ /90 with R/a ratio 15 and
100, L=2m, a=1m, t=0.03 m for cantilever composite shell. 59
4.14 Stability plots for different fibre orientations / / with R/a ratio 15 and 100,
L=2m, a=1m, t=0.03 m for cantilever composite shell. 61
4.18 Stability plots for different geometries with R/a ratio 15 and 100, t = 0.03 m,
fibre orientation 30/30/30 for a cantilever composite shell. 65
(xii)
LIST OF TABLES
(xiii)
CHAPTER – 1
INTRODUCTION
Failures in many engineering structures fall into one of the two simple categories: (1)
material failure and (2) structural instability due to form failure. In material failure, the
stresses in the structure exceed the specified safe limit, resulting in the formation of cracks
which cause failure. Material failure can be adequately predicted by analyzing the structure
on the basis of equilibrium conditions or equations of motion that are written for the initial
undeformed configuration of the structure. By contrast, the prediction of failures due to
structural instability requires equations of equilibrium or motion to be formulated on the basis
of deformed configuration of the structure. In many instances, instability is not directly
associated with the failure of the overall structure. For example, if the skin of a plate or shell
like structure wrinkles or locally buckles, the entire structure does not fail. However, if a
portion of the structural element between two adjacent parts becomes unstable, the entire
structure fails catastrophically. Thus, stability plays a very important part in designing a
structure.
The load at which a structure becomes unstable can be, in the simplest approach,
regarded as independent of the material strength or yield limit; it depends on loading of the
structure, structural geometry and size, especially slenderness, and is governed primarily by
the stiffness of the material, characterized, for example, by the elastic modulus. Failures of
elastic structures due to structural instability have their primary cause in geometric effects:
the geometry of deformation introduces nonlinearities that amplify the stresses calculated on
the basis of the initial undeformed configuration of the structure. In the structural instability
due to form failure, though the stresses may not exceed the safe value, the structure may not
able to maintain its original form. Here, the structure does not fail physically but may deform
to some other shape due to intolerable external disturbance.
The instability problems, according to the type of loading can be divided into two
categories: (1) static instability and (2) dynamic instability. If the loading are static in nature
the instability problems encountered are called static instability and if the loading varies with
time it is called dynamic instability.
1
reduce the bending stiffness to zero for some physically possible deformation mode. If the
membrane forces are reversed – that is, made tensile rather than compressive – bending
stiffness is effectively increased. This is called stress stiffening effect. The effect of
membrane forces are accounted for by a matrix [k ] that augments the conventional stiffness
matrix [k]. Matrix [k ] has given various names, as follows: initial stress stiffness matrix,
differential stiffness matrix, geometric stiffness matrix, and stability coefficient matrix.
In certain cases time varying loads acts axially on the structures which may lead to
loss of dynamic equilibrium of the system resulting in instability of the system. Among the
problems of the dynamic stability of structures probably the best known subclass can be
constituted by the problems of parametric excitation, or parametric resonance. If the ordinary
resonance of forced vibrations occurs when the natural and exciting frequencies are equal
(primary resonance), then parametric resonance occurs when the exciting frequency is equal
to double the frequency of free vibrations (principal parametric resonance). Another essential
difference of parametric resonance lies in the possibility of exciting vibrations with
frequencies smaller than the frequency of the principal resonance. Finally, qualitatively new
in parametric resonance is the existence of continuous regions of excitation (regions of
dynamic instability). A typical example is the initially straight prismatic column whose two
ends are simply supported and upon which a periodic axial compressive load is acting. Such a
column is known to develop lateral oscillations if its straight-line equilibrium is disturbed.
Depending upon the magnitude and the frequency of the pulsating axial load, the linear Hill
or Mathieu equation defining the lateral displacements of the column may yield bounded or
unbounded values for these displacements.
In the introduction to the first English edition of their monumental textbook entitled
Engineering Dynamics, Biezeno and Grammel [1] explain that, following Kirchhoff’s
definition, dynamics is the science of motion and forces, and thus includes statics, which is
the study of equilibrium, and kinetics, which treats of the relationship between forces and
motion. Dynamics is generally accepted as the antonym of statics in everyday usage, and this
is the sense in which it is used in the title of International Conference on Dynamic Stability
of Structures.
2
equations of motion; such an investigation is designated by Ziegler as a stability analysis with
the aid of the kinetic criterion. Excluding parametric resonance the other types of dynamic
instability problems are listed as follows:
Interaction between the non-conservative aerodynamic forces and the elastic structure
of airplanes and missiles can give rise to theoretically interesting and practically important
problems. They are dealt with, as a rule, by specialists known as aeroelasticians.
Of the many possible time-dependent loading conditions not yet mentioned, one,
buckling under the conditions prevailing in the ordinary testing machine, presents special
interest. In industry, most compressed structural elements are designed on the basis of Euler’s
theory of buckling, or with the aid of one of the modifications of Euler’s theory to account for
inelastic behavior. The practical suitability of these theories is judged, as a rule, on the basis
of a comparison with buckling loads obtained in conventional mechanical or hydraulic testing
machine. In 1949 Hoff J. Nicholas [2] drew attention to the fact that the behavior of dynamic
system consisting testing machine and test column does not necessarily agree with that of a
compressive element in an airplane hitting the ground or in a bridge subjected to dead and
live loads; nor do the initial and boundary conditions assumed in Euler’s theory agree with
those prevailing in the testing machine.
3
1.3 PARAMETRIC EXCITATION
The motion of a particle of mass m attached to one end of a mass-less rod of length l,
while the other end of the rod is attached to a point under the influence of a prescribed
acceleration as shown in Figure 1.1 is considered. Applying Newton’s second law of motion
in the direction perpendicular to the rod leads to
(1.1)
Hence, (1.2)
4
which is an equation with variable coefficients. For small oscillations about , the above
(1.3)
5
Figure-1.2. Stability Diagram
6
1.4 SHELLS
Thin shells are an example of strength through form as opposed to strength through
mass. The effort in design is to make the shell as thin as practical requirements will permit so
that the dead weight is reduced and the structure functions as a membrane free from large
bending stresses. By this means, a minimum of materials is used to the maximum structural
advantage. Shells of double curvature are among the most efficient of known structural
forms. Most shells occurring in nature are doubly curved. Shells of eggs, nuts, and the human
skull are commonplace examples. These naturally occurring shells are hard to crack or break.
The problem of parametric resonance can occur in various shell type of structures
when the structural system is under time-varying loads, especially periodic loads. Some
common examples are given below:
The shell of the rocket can be under parametric excitation or resonance when
the huge amount of energy releases from it’s backwards and as a result the
body of the rocket goes upwards. Here the reaction force due to the release of
that energy can be treated as time varying loads.
7
1.5 LITERATURE REVIEW
Melde [4] tied a string between a rigid support and the extremity of the prong of a
massive tuning fork of low pitch. He observed that the string could be made to oscillate
laterally, although the exciting force is longitudinal, at one half the frequency of the fork
under a number of critical conditions of string mass and tension and fork frequency and
loudness.
Strutt [5] provided a theoretical basis for these observations and performed further
experiments with a string attached to one end of the prong of a tuning fork.
The results of Strutt [5] were amplified by Stephenson [6] and he observed the
possibility of exciting vibrations when the frequency of the applied axial excitation is a
rational multiple of the fundamental frequency of the lateral vibration of the string.
Stephenson [6] seems to be the first to point out that a column under the influence of a
periodic load may be stable even though the steady value of the load is twice that of the Euler
load.
8
Krylov and Bogoliubov [15] used the Galerkin procedure to determine the dynamic
response of a column with arbitrary boundary conditions under the influence of
multiharmonic axial forces.
A number of physical systems contain pipes conveying fluid. The velocity often has
an unsteady component induced by the pumps. Thus parametric and combination instabilities
might occur in such pipes. These were studied theoretically by Chen [22], Ginsberg [23],
Paidoussis and Issid [24], Bohn and Herrmann [25], and Paidoussis and Issid [26].
P. Bhattacharya, S. Homann and M. Rose [27] studied the effects of piezo actuated
damping on parametrically excited laminated composite plates with feedback control
methodology. The study showed that piezo electric damping plays a positive role on the
stability behavior of laminated plates.
A number of theories exist for layered anisotropic shells. Many of this theories were
developed originally for thin shells, and are based on the Kirchhoff-Love kinematic
hypothesis that straight lines normal to the undeformed mid-surface remain straight and
normal to the middle surface after deformation and undergo no thickness stretching. Surveys
of various shell theories can be found in the works of Naghdi [28] and Bert [29] and a
detailed study of thin ordinary (i.e., not laminated) shells can be found in the monographs by
Kraus [30], Ambartsumyan [31], and Vlasov [32].
9
Dong, Pister, and Taylor [33] formulated a theory of thin shells laminated of
anisotropic material that is an extension of the theory developed by Stavsky [34] for
laminated anisotropic plates to Donnell’s shallow shell theory. Cheng and Ho [35] presented
an analysis of laminated anisotropic cylindrical shells using Flugge’s shell theory. A first
approximation theory for the un-symmetric deformation of non-homogeneous, anisotropic,
elastic cylindrical shells was derived by Widra and Chung [36] by means of the asymptotic
integration of the elasticity equations. For a homogeneous, isotropic material, the theory
reduces to Donnell’s equations.
All the theories listed above are based on Kirchhoff-Love’s hypotheses, in which the
transverse shear deformation is neglected. These theories, known as the Love’s first
approximation theories, are expected to yield sufficiently accurate results when
However, the application of such theories to layered anisotropic composite shells could lead
to 30% or more errors in deflections, stresses, and frequencies.
Whitney and Sun [37] developed a shear deformation theory for laminated
cylindrical shells that includes transverse shear deformation and transverse normal strain as
well as expansional strains. Recently, Reddy [38] presented a generalization of Sander’s shell
theory (1959) to laminated, doubly-curved anisotropic shells. The theory accounts for
transverse shear strains and the Von-Karman (nonlinear) strains.
S.K. Sahu and P.K.Dutta [39] studied the dynamic stability of curved panels with
cutouts. In their literature the parametric instability behavior of curved panels with cutouts
subjected to in-plane static and periodic compressive edge loadings were studied using finite
element analysis. The first order shear deformation theory was used to model the curved
panels, considering the effects of transverse shear deformation and rotary inertia. The theory
used was the extension of dynamic, shear deformable theory according to the Sander's first
approximation for doubly curved shells, which can be reduced to Love's and Donnell's
theories by means of tracers. The effects of static load factor, aspect ratio, radius to thickness
ratio, shallowness ratio, boundary conditions and the load parameters on the principal
instability regions of curved panels with cutouts were studied in detail using Bolotin's
method.
10
point and patch tensile edge loadings by using the finite element method. The effect of first
order shear deformation theory was used to model the doubly curved panels.
Vibration, buckling and dynamic stability of cracked cylindrical shells was studied by
M. Javidruzi, A. Vafai, J.F. Chen and J.C. Chilton [41]. This paper presents a finite
element study on the vibration, buckling and dynamic stability behavior of a cracked
cylindrical shell with fixed supports and subject to an in plane compressive/tensile periodic
edge load. The effects of crack length and orientation were analyzed.
S.N. Patel, P.K. Datta and A.H. Sheikh [42] studied the Buckling and dynamic
instability analysis of stiffened shell panels. The static and dynamic instability characteristics
of stiffened shell panels subjected to uniform in-plane harmonic edge loading were
investigated in this paper. The eight-noded iso-parametric degenerated shell element and a
compatible three-noded curved beam element were used to model the shell panels and the
stiffeners, respectively.
11
1.6 OBJECTIVE AND SCOPE OF THE PRESENT WORK
From the critical review of the existing literatures in the field of dynamic instability of
orthotropic shells under parametric excitation done in the previous section, it is found that a
very few literature is available in the area of parametric instability of orthotropic shell
structures and the effect of damping on instability behavior is almost non-existent.
The present work has been developed in the MATLAB environment using four noded
iso-parametric element. The various numerical results are compared with ANSYS finite
element software. The different aspects of the present work are presented systematically in
various sections which have been briefly outlined in the next section.
With the objectives stated above, the present work is presented in five chapters.
Chapter 1: Introduction
A general introduction to the present work, types of dynamic instability, a short note
on parametric excitation is given. A literature review is presented to give a broad
understanding of the previous works related to the present work. Finally the objective
and scope of the present work is presented.
In this chapter a general introduction to composite materials, lamina and laminate are
given. Constitutive equations of laminated composites and displacement modeling
have been shown. Finally the energy formulation for the present physical problem has
given.
12
Chapter 3: Finite Element Formulation
In this chapter, a brief discussion on the element shape functions is given. The strain-
displacement equations are derived. The structural stiffness matrix, mass matrix,
geometric stiffness matrix are derived. Finally the stability equations are presented.
This chapter presents the results for static, buckling and free vibration analysis using
developed FE code and compares with ANSYS finite element software. Parametric
instability of a cylindrical shell has been carried out both for isotropic and composite
cases.
Chapter 5: Conclusions
Major conclusions drawn from the present work along with a brief outline on the
scope of future work on the present investigation are highlighted in this chapter.
13
CHAPTER-2
CONSTITUTIVE EQUATIONS
2.1 INTRODUCTION
A composite material consists of two or more materials and offers a significant weight
saving in a structure in view of its high strength to weight and high stiffness to weight ratios.
Further, in a fibrous composite, the mechanical properties can be varied as required by
suitably orienting the fibres. In such a material the fibres are the main load bearing members,
and the matrix, which has low modulus and high elongation, provides the necessary
flexibility and keeps the fibres in position and protects them from environment.
Depending on the matrix material, composites can be further classified into the
following categories:
Depending on the reinforcement used, they can be classified into the following groups:
A lamina, in general is a thin sheet with fibres oriented in some direction. Such a sheet
can be characterized as two dimensional, with orthogonal material properties. It is, however,
14
not capable of carrying any load. Hence, for practical purposes, a structure consisting of
several laminae (laminate) is used.
The stiffness of a composite laminate changes with the change in ply orientation. The
particular axis which is chosen for conveniently solving the problem, is known as the loading
axis or the reference axis and for the fibre reinforced composites, another axis system which
is parallel and perpendicular to the fibre orientation is convenient for the calculation of
material properties, is known as the principal material axis. The axis system as described
above is shown below in Figure 2.2
15
Figure-2.2 Axis system of a unidirectional stressed Lamina
σ1 Q11 Q12 0 0 0 ε1
σ2 Q12 Q22 0 0 0 ε2
τ 12 = 0 0 Q66 0 0 γ 12 (2.1)
τ 13 0 0 0 Q44 0 γ 13
τ 23 0 0 0 0 Q55 γ 23
where, [Q] i,j=1,2,6,4,5 are reduced stiffness for plane stress and are defined in terms of the
engineering constants as
E1
Q11 = (2.2)
1 − υ12υ 21
E2
Q22 = (2.3)
1 − υ12υ 21
υ12 E 2 υ 21 E1
Q12 = Q21 == = (2.4)
1 − υ12υ 21 1 − υ12υ 21
16
The relationships between stresses of the principal material axis and the reference axis are
The relationships between strains of the principal material axis and the reference axis are
where, [T] is the transformation matrix and is given by the following equation
m2 n2
0 0 mn
2 2
n m 0 0 − mn
5 0 0 m −n 0 (2.8)
0 0 n m 0
− 2mn 2mn 0 0 (m − n 2 )
2
where,
Now when a lamina is loaded in the reference axis X-Y, the relationship between stresses in
the reference X-Y axis and that in the principal material axis 1-2 is given by
2346&7 5 9- 234
-&1 (2.9)
σx m2 n2
0 0 − mn Q11 Q12 0 0 0 ε1
σy 2
n m 2
0 0 mn Q12 Q22 0 0 0 ε2
i.e. τ xy 0 0 m n 0 0 0 Q66 0 0 γ 12 (2.10)
τ xz 0 0 −n m 0 0 0 0 Q44 0 γ 13
τ yz 2mn − 2mn 0 0 (m − n 2 )
2
0 0 0 0 Q55 γ 23
Substituting strains in the 1-2 axis in terms of X-Y reference axis Equation 2.9 will take the form as
2346&7 5 9-
: ;&< 5 284=&> i , j = 1,2,6,5,4 (2.11)
17
Equation- 2.11 will take the form as
′
where [Q ] i,j = 1,2, 6,4,5 are the transformed reduced stiffnesses which are given in terms of
reduced stiffnesses, Qij as
Q'16 = ( Q11 - Q12 - 2 Q66 ) nm3 + ( Q12 - Q22 + 2 Q66 ) n3m (2.16)
Q' 26 = ( Q11 - Q12 - 2 Q66 ) n3m + ( Q12 - Q22 +2 Q66 ) nm3 (2.17)
18
Normal to the mid-surface before deformation remains straight but is not necessarily
normal to the mid-surface after deformation.
19
Employing first order shear deformation theory the displacement u, v, w on the shell can be
expressed as,
u = u0 + zθ y ; v = v0 − zθ x ; w = w0 (2.22)
where, u, v, w are the translational displacement along X, Y and Z axes at a distance z from
the mid-plane while the notations with suffix ‘0’ denotes the same at mid-plane, and θ x, θ y
are rotation of shell element about X and Y axis respectively.
∂u 0 W0 ∂θ y
ε xx = ε x0 + zκ x = − +z
∂x R x ∂x
∂v 0 W0 ∂θ
ε yy = ε y0 + zκ y = − −z x
∂y R y ∂y
∂v0 ∂u 0 2w0 ∂θ y ∂θ x
γ xy = γ xy0 + zκ xy = + − +z − (2.23)
∂x ∂y R xy ∂y ∂x
∂w0 ∂u 0 ∂w0
γ zx = γ zx0 + zκ zx = + = +θy
∂x ∂z ∂x
where, 866 & 877 are the normal strains in X and Y directions respectively, and ?67 & ?7@ & ?@6 are
the shear strains in X-Y, Y-Z, and X-Z plane respectively.
B 7 B 6 B 7 B 6
A6 C A7 C A67
B= B> B> B=
and the mid-plane strains are expressed in terms of the mid-plane displacements as,
∂u 0 W0 ∂v 0 W0 ∂v0 ∂u 0 2 w0
ε x0 = − ε y0 = − γ xy0 = + −
∂x R x ∂y R y ∂x ∂y R xy
; ;
∂w0 ∂v 0 ∂w0 ∂w ∂u ∂w
γ yz0 = + = −θx γ zx0 = 0 + 0 = 0 + θ y
∂y ∂z ∂y ; ∂x ∂z ∂x
20
2.7 STRESS -STRAIN
STRAIN RELATIONS FOR A LAMINATE
The stresses in the kth layer of a laminated composite can be expressed in terms of the
laminate strains and curvatures as
21
The resultant forces and moments acting on a laminate are obtained by integrating the
stresses in each layer or lamina through the laminate thickness, as given by
t/2 t/2
Ni = σ i dz ; M i = σ i zdz (2.25)
−t / 2 −t / 2
Ni is the force resultant (force per unit length) of the cross section of the laminate and Mi is
the moment resultant (moment per unit length) as shown in the Figure-2.6.
The total of force and moment resultants for an n-layered laminate can be defined as
Nx σx Mx σx
t/2 t/2
Ny = σy dz ; My = σy zdz (2.26)
−t / 2 −t / 2
N xy σ xy k
M xy σ xy k
The integration indicated in the above equations can be rearranged to take advantage of the
fact that the stiffness matrix for a lamina is constant within the lamina. Thus, the stiffness
matrix goes outside the integration over layer, but is within the summation of force and
moment resultants for each layer. When the lamina stress-strain relations are substituted,
Nx N
Q'11 Q'12 Q'16 ε xo κx
Zk Zk
Ny = Q'12 Q' 22 Q' 26 ε y o dz + κ y zdz (2.27)
Zk −1 Zk −1
k =1
N xy
k
Q'16 Q' 26 Q' 66 k
γ xy o κ xy
Mx N
Q'11 Q'12 Q'16 ε xo κx
Zk Zk
My = Q'12 Q' 22 Q' 26 ε y o zdz + κ y z 2 dz (2.28)
Zk −1 Zk −1
k =1
M xy
k
Q'16 Q' 26 Q' 66 k
γ xy o κ xy
o o o
However, we should recall that ε x , ε y , γ xy , κ x , κ y and κ xy are not functions of z but are
mid-plane values so can be removed from under the summation signs. Thus the above
equations can be written as
22
Mx B11 B12 B16 ε xo D11 D12 D16 κx
o
M y = B12 B22 B26 ε y + D12 D22 D26 κy (2.30)
M xy B16 B26 B66 γ xy o D16 D26 D66 κ xy
where,
Where, Aij are extensional stiffnesses, Bij are coupling stiffnesses, and Dij is the bending
stiffnesses and are given as below.
n
Aij = (Q ' ij ) k ( Z k − Z k −1 ) (2.33)
k =1
1 n
Bij = (Q 'ij ) k ( Z 2 k − Z 2 k −1 ) (2.34)
2 k =1
1 n
Dij = (Q 'ij ) k ( Z 3 k − Z 3 k −1 ) (2.35)
3 k =1
23
The stress resultants Mxz, Myz for the laminates can be written in terms of the constitutive
matrix for the shear [C]s and shear strain γ xz , γ yz as,
M xz A A45 γ xz
= 44 (2.36)
M yz A54 A55 γ yz
Aij, approximated as
where, Qoff, ij = off axis reduced stiffness (elastic constants) for plane stress.
where, T and U are the kinetic and potential energies of the system. Now the associated
Euler- Lagrange’s equation for n d.o.f system is,
d ∂T ∂
.
− (T − U ) = QI , i = 1,2,... (2.39)
dt ∂ q ∂qi
In the above equation, T is the kinetic energy and U is the potential energy of the system and
qi’s are the generalized coordinates.
The strain energy of the system due to mechanical strain can be expressed as,
1 n
UM =
2 V
{ε } {σ }dV
k T k
(2.40)
k =1
24
The work done due to mechanical loading is given by,
WM = u T {F }dv (2.41)
V
where,
u = Displacement d.o.f in x-direction
F = Mechanical Force vector
= Stress
= Strain.
Now, the functional I can be expressed in the variational form for a time interval from t0 to t1
by the use of Equation-2.40 and 2.41 as,
t1 t
1 n
1 1
δI = −δ dt
2
{ε } {σ }dv
k T k
−
2v
{u ′}T [ρ ]{u ′}dv + dt {δu}T {Fm }dA (2.42)
t0 k =1 v t0 A
Now if the structural system is subjected to a pulsating axial compressive force P(t) = P + P
s t
cos t, acting along its undeformed axis where is the excitation frequency of the dynamic
load component, P is the static and P is the amplitude of the time dependent component of
s t
the load. In this case a residual strain will exist in the structural system. For this case, the
strain energy of the system can be expressed as follows:
U M = U L + U NL (2.43)
where, UL and UNL is the linear and nonlinear strain energy respectively. The expression for
UNL is given in chapter-3 and the expression for UL is given in Equation 2.40. UNL is the
nonlinear strain energy due to compressive residual forces.
25
CHAPTER – 3
3.1 INTRODUCTION
The finite element method is the most popular numerical tool available to the present
day engineer. It is quite versatile and is being used to solve a wide variety of engineering
problems. The following are the basic steps involved in the finite element analysis.
26
3.2 FORMULATION
The finite element formulation for the present case has been described in the
following subsections.
In the present work an attempt has been made to develop a four node isoparametric
2-D finite element with five mechanical degrees of freedom (u, v, w, θ x , θ y ) per node based
on Mindlin-Reissner shallow shell theory.
The mapping between the natural coordinates (D, E) and the physical coordinates
(X,Y) is related by certain functions Ni(D,E). Thus, if (D, E) coordinates of a point in the
natural coordinate system is known, then the coordinates of the corresponding point in the
physical coordinate system (X, Y) is given by
4
X = N i xi
i =1
4
Y= N i yi ; where, xi and yi are the nodal co-ordinates of the element.
i =1
Similarly, the components of displacement vector at any point of the element are given by
4 4 4 4 4
u= N i ui v= N i vi w= N i wi θx = N iθ x i θy = N iθ y i
i =1 i =1 i =1 i =1 i =1
where, u, v, w, x
, y
having a subscript ‘i’ are the nodal displacements.
27
The element displacements as given in equation can be written in the matrix form as follows:
u Ni 0 0 0 0 ui
v 0 Ni 0 0 0 vi
w = 0 0 Ni 0 0 wi
θx 0 0 0 Ni 0 θ xi
θy 0 0 0 0 Ni θ yi
The shape functions for an four noded quadrilateral element is given by,
1 1
N 1 = (1 − ξ )(1 − η ) ; N 2 = (1 + ξ )(1 − η ) ;
4 4
1 1
N 3 = (1 − ξ )(1 + η ) ; N 4 = (1 + ξ )(1 + η ) ;
4 4
where, i
and i
are the values of natural co-ordinates at node ‘i’.
The transformation from the local ( - ) co-ordinate to the global (x-y) co-ordinate is carried
out as follows:
N iX −1 N iξ
= [J ] ;
N iY N iη
X iξ Yiξ
where, [J ] = (3.3)
X iη Yiη
28
3.2.2 Strain-Displacement Relations
The strains for the shell element can be calculated by using the constitutive equations
given below:
εx ε x0 kx
εy ε y0 ky
γ xy = γ xy0 + Z k xy
γ yz γ yz0 0
γ xz γ xz0 0
0
(3.4) where, x, y, xy, yz,are the strains at a distance z from the mid-plane of the shell,
xz x ,
0 0 0 0
y , xy , yz , xz are the mid-plane strains and x, y, xy are the shell curvatures.
εx
εy
γ xy = [ Z ]5 x 8 [ B ]8 x 5 {d }5 x1 (3.5)
γ yz
γ xz
where, [Z] is the position matrix and {d} is the nodal d.o.f as given below:
1 0 0 z 0 0 0 0 u0
0 1 0 0 z 0 0 0 v0
[Z ] = 0 0 1 0 0 z 0 0 ; {d } = w0 (3.6a)
0 0 0 0 0 0 1 0 θx
0 0 0 0 0 0 0 1 θy
29
The above [B] matrix can be divided into two parts, bending and shear strain-displacement
matrices.
Bending strain-displacement matrix consists of bending strains, shear strains and shell
curvatures.
∂N i Ni
0 − 0 0
∂x Rx
∂N i N
0 − i 0 0
∂y Ry
∂N i ∂N i N
4 −2 i 0 0
[ B ]b = ∂y ∂x R xy (3.7)
i =1 ∂N i
0 0 0 0
∂x
∂N i
0 0 0 − 0
∂y
∂N i ∂N i
0 0 0 −
∂x ∂y
∂N i
4 0 0 − Ni 0
[ B]s = ∂y (3.8)
i =1 ∂N i
0 0 0 − Ni
∂x
The strain-displacement relationship for the shell in the matrix operator form is derived from
Equation 2.23 and is expressed as:
∂N i Ni
0 − 0 0
∂x Rx
∂N i N
0 − i 0 0
ε x0 ∂y Ry u0
∂N i ∂N i 2Ni
ε y0 − 0 0 v0
∂y ∂x Ry
γ xy0 w0
∂N i
γ yz0 0 0 − Ni 0 θx
0
= ∂y
γ zx ∂N i θy
0 0 0 Ni 1
κx ∂x { }2 (3.9)
κy ∂N i { }3
0 0 0 0
κ xy ∂x { }4
∂N i
0 0 0 − 0
∂y
∂N i ∂N i
0 0 0 −
∂x ∂y i =1to 4
30
which can also be written as,
{ε } = [∂ ][ N ]{d }
0
e (3.10)
{ε } = [ B]{d }
0
e (3.11)
The strain energy due to mechanical strain for shell as expressed in Equation 2.4 is
given as,
1 n
UM =
2 V
{ε } {σ }dV
k T k
k =1
Now combining Equation 2.4 and Equation 2.12 the potential energy for the laminate due to
strain is given by,
1 n
UM =
2 V
{ε } [Q ]{ε }dV
k T k k
(3.12)
k =1
a b
1
UM = {d e }T [ B ]T [C ][ B ]{d e }dxdy (3.13)
200
where, [C] is the material matrix, given in Equation 2.32 and 2.36.
1 (3.14)
UM = {d e }T [ K M ]{ d e }
2
Here, [KM] is the element stiffness (mechanical) matrix and is given by,
a b
[K M ] = [ B ]T [C ][ B ]dxdy
(3.15)
0 0
31
To obtain the numerical value of the element stiffness, a numerical integration technique using Gauss-
Quadrature Method is adopted. A reduced order 2 x 2 integration and 1 x1 integration is carried out
for bending part and shear part respectively to avoid shear locking in thin shells.
The element stiffness matrix given in equation (3.15) can be written in the natural coordinate system
as,
1 1
[K M ] = [ B ]T [C ][ B ][ J ]dξdη . (3.16)
−1 −1
The equations of motion can be expanded and without considering the external force
effect can be written as,
∂σ xx ∂τ yx ∂τ zx ∂ 2u
+ + =ρ 2 (3.17a)
∂x ∂y ∂z ∂t
∂τ xy ∂σ yy ∂τ zy ∂ 2v
+ + =ρ (3.17b)
∂x ∂y ∂z ∂t 2
∂τ xz ∂τ yz ∂τ zz ∂2w
+ + =ρ 2 (3.17c)
∂x ∂y ∂z ∂t
Substituting the expanded form of u, v and w according to Equation 2.22 one can express the
equations of motion as follows,
∂σ xx ∂τ yx ∂τ zx ∂ 2u0 ∂ 2θ y
+ + =ρ + z (3.18a)
∂x ∂y ∂z ∂t 2 ∂t 2
∂τ xy ∂σ yy ∂ 2 v0
∂τ zy ∂ 2θ x
+ + = ρ[ −z 2 (3.18b)
∂x ∂y ∂z ∂t 2 ∂t
2
∂τ xz ∂τ yz ∂σ zz ∂ w0
+ + =ρ (3.18c)
∂x ∂y ∂z ∂t 2
32
∂ 2u0 ∂ 2θ y
= I1 2 + I 2 (3.19)
∂t ∂t 2
Similarly, Equations 3.18b and 3.18c are integrated with respect to z and Equations 3.18a to
3.18c are integrated with respect to z . Finally, the inertia matrix [J] is expressed as,
2
I1 0 0 0 I2
0 I1 0 − I2 0
[ρ ] = 0 0 I1 0 0 (3.20)
0 − I2 0 I3 0
I2 0 0 0 I3
h
2
with I 1 , I 2 , I 3 = ρ .( z ).(1, z, z 2 )dz
h
−
2
1
{ } [N ] [ρ ][N ]{d }dx.dy
T T
T= de e
2A
Or,
1
T=
2
{ } [M ]{d }
de
T
e (3.21)
Material nonlinearity
Contact nonlinearity
Geometric nonlinearity
33
The nonlinear strains which are also known as the Green-Lagrangian strains are given as,
2 2 2
∂u 1 ∂u ∂v ∂w
εx = + + +
∂x 2 ∂x ∂x ∂x
2 2 2
∂v 1 ∂u ∂v ∂w
εy = + + + (3.23)
∂y 2 ∂y ∂y ∂y
∂u ∂v ∂u ∂u ∂v ∂v ∂w ∂w
γ xy = + + + +
∂y ∂x ∂x ∂y ∂x ∂y ∂x ∂y
Total strain energy for the shell element can be written as,
The nonlinear strain energy per unit volume for shell can be expressed as,
-
(Strain energy) NL= 3, 8KL
1
where 3, is the initial stress developed due to the external loading and 8KL is the nonlinear strain
vector.
ε xNL
3, 36M 37, 367, , ε NL = ε yNL (3.24)
γ xyNL
2 2 2
1 ∂u ∂v ∂w
ε yNL = + + (3.25)
2 ∂y ∂y ∂y
∂u ∂ u ∂ v ∂v ∂w ∂w
γ xyNL = + +
∂ x ∂ y ∂ x ∂ y ∂ x ∂y
34
After expanding,
2 2 2
1 ∂u 0 ∂v 0 ∂w 0 w 0 ∂u 0 w ∂v 0 ∂w u ∂u 0 ∂ θ y
ε xNL = + + −2 −2 0 − 2 0 0 + 2z
2 ∂x ∂x ∂x R x ∂x R xy ∂x ∂x R x ∂x ∂x
2 2 2
∂θ y w0 ∂v 0 ∂θ x ∂θ x w0 u 0 θ y θ y ∂w0 ∂θ y ∂θ x θy
− − + + − + z2 + +
∂x R x ∂x ∂x ∂x R xy R x R x R x ∂x ∂x ∂x Rx
2 2 2
w w0 u
+ 0 + + 0 (3.26a)
Rx R xy Rx
2 2 2
1 ∂u 0 ∂v0 ∂w0 w0 ∂u 0 w ∂v ∂w v ∂u 0 ∂θ y
ε yNL = + + −2 − 2 0 0 − 2 0 0 + 2z
2 ∂y ∂y ∂y R xy ∂y R y ∂y ∂y R y ∂y ∂y
2 2 2
∂θ y w0 ∂v 0 ∂θ x ∂θ x w0 v 0 θ x θ x ∂w0 ∂θ y ∂θ x θx
− − + − + + z2 + +
∂y R xy ∂y ∂y ∂y R y R y R y R y ∂y ∂y ∂y Ry
2 2 2
w0 w v
+ + 0 + 0 (3.26b)
R xy Ry Ry
∂ u 0 ∂u 0 ∂v 0 ∂v 0 ∂w 0 ∂w 0 ∂u 0 w 0 ∂v 0 w 0 ∂v 0 w 0 ∂w 0 v 0 ∂w 0 u 0
γ xyNL = + + − − − − −
∂x ∂y ∂ x ∂y ∂x ∂y ∂x R xy ∂x R y ∂y R xy ∂x R y ∂y R x
∂u 0 w 0 ∂u 0 ∂θ y ∂θ y ∂u 0 ∂θ y w 0 ∂ v 0 ∂θ x ∂θ x ∂ v 0 ∂θ x w 0 ∂θ x w 0
− +z + − − − + +
∂y R x ∂x ∂y ∂x ∂y ∂x R xy ∂x ∂y ∂x ∂y ∂x R y ∂y R xy
∂w 0 θ x u 0 θ x θ y ∂w 0 θ y v 0 w 0 ∂ θ y ∂θ y ∂ θ y ∂θ x ∂ θ x θ y θ x
+ − − + − + z2 + −
∂x R y R x R y R x ∂y R x R y R x ∂y ∂x ∂y ∂ x ∂y Rx R y
w0 w0 u 0 v 0 w0 w0
+ + + (3.26c)
R xy R y R x R y R x R xy
{ε NL } = 1 [ R]{d } (3.27)
2
35
where,
O, P, Q, Q, Q, 6 7
2*4 NO,&6 O,&7 P,&6 P,&7 Q,&6 Q,&7 6&6 6&7 7&7 7&6 S
R6 R7 R6 R7 R67 R7 R6
and [R] is obvious from Equation 3.26. Now {d} can also be written as,
where,
N i,x 0 0 0 0
N i, y 0 0 0 0
0 N i,x 0 0 0
0 N i, y 0 0 0
0 0 N i,x 0 0
0 0 N i, y 0 0
0 0 0 N i,x 0
0 0 0 N i, y 0
0 0 0 0 N i,x
0 0 0 0 N i, y
Ni
4 0 0 0 0
[Gi ]i =1to 4 = Rx
i =1 Ni
0 0 0 0
Ry
Ni
0 0 0 0
Rx
Ni
0 0 0 0
Ry
Ni
0 0 0 0
R xy
Ni
0 0 0 0
Ry
Ni
0 0 0 0
Rx
Then the potential energy due to residual stresses for the element can be written as,
1
U NL = {d e }T [G ]T [ R ]T {σ 0 } (3.28)
2V
36
Since, [R]T23, 4 23, 4 T 2*U 4
1 1
U NL = {d e }T [G ]T [σ 0 ][G ]{d e }dv = {d e }T [ K σ ]{d e } (3.29)
2V 2
in which [K ]= [G ]T [σ 0 ][G ]dv is the geometric stiffness matrix due to residual stresses.
V
The geometric stiffness matrix can also be expressed in terms of residual stress resultants and
the local coordinates of the element as,
1 1
[Kσ ] = [G ]T [ S σ ][G ] J dξdη (3.30)
−1 −1
In the expression of the [S ] matrix Nx , Ny , Nxy, Mx, My, Mxy are the force and moment
resultants and were expressed using standard composite laminate modeling and the rest of the
terms like N x , N y , N xy in the stress matrix are given by,
ε x0
ε y0
Nx
γ xy0
Ny = [T ] (3.31)
κx
N xy
κy
κ xy
where,
1 n
1 n
Dij = (Q 'ij ) k ( Z 3 k − Z 3 k −1 ) Dij = (Q 'ij ) k ( Z 4 k − Z 4 k −1 )
3 k =1 and 4 k =1
37
Where, the stress matrix [S ] is given by,
The potential energy of the system due to external mechanical loading for each
element is written as,
P= {d e }T [N ]T {p}dA (3.33)
A
And {p} is the force vector specified on the surface at z = h n and is given by,
{ p} = [ p x py pz p yz p xz p xy ]
Now, combining all the terms for potential energy as well as the energy due to inertia
the total energy for each element is given as,
1
{d e }[B ]T [D ][B ]{d e }dA − {d e } [N ]T {P}dA − 1 d e
{ } [N ] [ρ ]{d }dA
T T T
∏e = e (3.34)
2A A
2A
L=T-V
The Hamilton’s principle states that the variation of the Lagrangian during any time
interval t0 to t1 must be equal to zero, i.e.
t1
δ Le dt = 0 (3.35)
t0
Substituting from Equation 3.34 variational principle is applied to Equation 3.35 and on
integrating the last term by parts with respect to time one obtains,
t1 t1
(3.36)
The first term on the RHS in the Equation 3.36 vanishes at the limits because of the
agreement that { de} =0 at t=t0 and t=t1.
39
The final form of Equation 3.37 can therefore be expressed as given in the following equation
in the element level,
[M ]{d }+ [K ]{d } = {P }
e
e
e
e
e
(3.37)
The Equation 3.37 is derived for conservative system. In case of non conservative system
damping matrix introduce in Equilibrium Equations and expressed as
MU + CU + KU = P (3.38)
where,
M= mass matrix
C= damping matrix
K=stiffness matrices:
U = displacement vectors
U = velocity vectors
U = acceleration vectors
If residual stresses 3, are present due to some axial external loading in the system or
structure, one have to consider the geometric stiffness part along with the normal stiffness
matrix. The effect of membrane forces are accounted for by geometric stiffness matrix. The
compressive axial loading reduces the bending stiffness. If the membrane forces are reversed-
that is made tensile rather than compressive- bending stiffness is effectively increased. This
effect is called stress stiffening. For compressive membrane forces Equation 3.37 will take
the form as below,
[M ]{d }+ [[ K
e
e
e
]
] − [ K σ ] {d e } = 0 (3.39)
If the axial external loading is dynamic and harmonic in nature, the loading function can be
expressed as
+ +, +- (3.40)
where, P0 is the constant load part and P1 is the time varying part of that dynamic loading.
40
3.4.8 Stability equations
In the present case the eigen bending vector (which is same as that of the buckling
shape for a cantilever structure) is used to uncouple the governing equation. . In the next step
the uncoupled equation is solved using the method of strained parameter and the detailed
procedure is shown. It must be noted that in the present case the constant part of the loading
is not considered and the time dependent loading generating stiffness is only taken into
consideration.
As a first step the generalized displacement {d} are transformed into the model co-ordinate
using the transformation, {d (t)} = V 2O 4 (3.41)
where [ψ ] is the modal state vector.
a
The natural eigen frequency ^_1 ` and ωσ =
Kσ
allows to rewrite Equation 3.42 in the
b M
form
XYZ c XZ
^_1 +^1 ] ./ O (3.43)
X Y bX
Let
d
"
then,
du du dt ∗ θ du
= = (3.44a)
dt dt ∗ dt 2 dt ∗
d 2u d du θ dt ∗ d du θ θ d 2 u θ 2
= = ∗ = (3.44b)
dt 2 dt ∗ dt ∗ 2 dt dt dt ∗ 2 2 dt ∗2 4
d 2 u θ 2 C θ du
∗2
+
dt 4 M 2 dt ∗
2
( 2
)
+ ω n + Pωσ cos(θt ) u (t ) = 0 (3.45a)
d 2 u 2C du 4 2 4 2
∗2
+ ∗
+ 2 ω n + P 2 ωσ cos(2t ∗ ) u (t ) = 0 (3.45b)
dt Mθ dt θ θ
41
Equation 3.45b can be written as,
( 2
)
u + 2 µ ∗u + δ ∗ + 2ψ ∗ cos( 2t ∗ ) u (t ) = 0 (3.45c)
where,
2
4 C 2 Pω σ
δ∗ = 2
ωn 2 , µ ∗ = and ψ ∗ =
θ Mθ θ2
In the FE formulation, Rayleigh proportional damping has been considered, which has the
following form:
C= + K, where and are constants. After application of the weighted modal matrix â
here results in
a T Ca = αa T Ma + βa T Ka = αI + βΛ
(3.46)
Now,
C αM + β K α β K
µ∗ = = = +
Mθ Mθ θ Mθ (3.47)
for the condition of parametric resonance, = 2 n , now putting in Equation 3.47 one gets,
α βω n
µ∗ = +
2ω n 2 (3.48)
To evaluate and 4 % damping for the first mode and 6 % for the second mode has been
considered and putting them in Equation 3.46 two simultaneous algebraic equation has been
solved. Now after getting the value of and and putting them in Equation 3.48, * has been
calculated to provide damping to the first mode only, i.e. for n = 1 in the present case.
42
The Floquet theory together with the strained parameter approach [44] is applied to Equation
3.45c. This results for small parameters ψ ∗ in the expansions
1 2
0 − ψ ∗ + ...
2
1
δ∗ = (
∗2
1 ± ψ − 4µ ∗2 2
)
1 2
− ψ ∗ + ...
8 (3.49)
1
1 2 1 ∗2 2 2
4+ ψ∗ ± ψ − 16 µ ∗ + ...
6 16
In the derivation the unknown entities, δ ∗ and u(t*) are expanded in a power series with
respect to ψ ∗ under the additional assumption of a proportional Rayleigh damping
coefficient. This assumption is justified, because the stability is increased for higher damping.
The truncated series of Equation 3.49 are good approximation to the stability border. This can
be verified by a direct evaluation of the Floquet theory in combination with numerical
solution of the differential Equation 3.45c. The expansion around zero is independent of the
damping coefficient. The expansion one and four correspond to excitation frequencies θ
around two times the natural frequency and around the natural frequency itself respectively.
In the present work, the second stability equation has been considered from Equation 3.49 for
stability study.
43
CHAPTER-4
4.1 INTRODUCTION
Based on the finite element formulation derived in the earlier chapter, a program has
been developed in the MATLAB platform. Using the developed code, some problems have
been solved for static, free vibration and buckling analysis of cylindrical shell structures and
the results are compared with those using ANSYS finite element software. Thereafter, studies
have been carried out to understand the instability behavior of isotropic as well as laminated
shell structures with different fibre orientations and with various radius of curvature to width
ratio. In case of stability studies a generalized Rayleigh proportional damping has been
considered for all the cases.
In this section static analysis results are presented for cantilever shell. Mesh
convergence study of transverse tip deflection has been done for the isotropic case.
An isotropic cantilever shell for various radius of curvature to width ratio is subjected
to transverse unit load distributed equally at the free edge. The geometric and material
properties used for the shell are given below.
Material properties:
Material: Aluminum
Geometric properties:
Length, L = 2 m
Width, a = 0.5 m
Thickness, t = 0.02 m
44
Table 4.1 Mesh convergence study of non-dimensional transverse tip deflection (w/) for an
Isotropic cantilever shell (L=2m a=0.5m, t=0.02m of R/a = 30)
It can be seen from the above table that the result converges as we increase number of
elements. But the computational time also increases as a result. So keeping two things in
mind i.e. computational time and accuracy level, an optimized mesh size distribution (20 x 20
for the present study) have been taken.
It is observed from Table 4.2 that as the R/a ratio increases, w / at the mid node of the
free edge also increases. This is because, as R/a ratio increases the shell becomes shallower
and as a result the stiffness gets reduced accordingly. In the case of plate, the maximum
deflection occurs as there is no curvature resulting in it being the least stiff among all the
cases. The finite element model of the cantilever shell is shown in Figure- 4.1. Here the
maximum error is of the order 0.78 % for R/a ratio 15.
45
Figure- 4.1 Finite element model of a cantilever shell subjected to unit mechanical transverse
load at the free edge. The mesh consists of (20 x 25) elements.
In the present study, two typical two layered and eight layered laminates are
considered. The finite element model has been shown above in Figure-4.1. The material and
geometrical properties used for the analysis are presented below.
Material properties:
Material: Carbon Fiber Composite
Young’s Modulus, E1 = 105 GPa; E2 = 6.13 GPa
Poisson’s Ratio, 12 = 0.317; Shear Modulus, G12 = 2.28 GPa; G13=G23=2.28 GPa
Geometric properties:
Length, L = 2 m
Width, a = 1 m
Thickness, t = 0.03 m
Fibre orientation:
2- Layered laminate = 450/-450; 8- layered laminate = (00/900/450/-450) s.
46
The non-dimensional transverse displacements obtained at x = 0.5 m, y = 0 m from
the present FE model and from ANSYS are given in Table 4.3. Here (20 x 25) element
meshing has been considered in the FE modeling.
Table 4.3 Non-dimensional transverse deflection (w/) of at x = 0.5 m and y = 0 m of a
cantilever shell (L= 2m, a=1m, t=0.03 m for R/a = 50).
450/-450 0.28706
0.29043
(00/900/450/-450)s 0.07886
0.07938
Note: w/ = (w.t5.E11)/ (F.a4)
It is observed from Table 4.3 that, as the number of layer is increasing w / at x = 0.5 m
and y = 0 m, also significantly getting reduced, which indicates the stiffness of the structure is
increasing. Here the maximum error is of the order 1.16 % for 450/-450 laminate.
The free vibration analysis has been carried out both for isotropic as well as
composite shells. The finite element model and the boundary conditions are shown in Figure-
4.1.
The material properties are same as given in problem 4.2.1 and the geometric
properties are same as given in problem 4.2.2. The results obtained by the present F.E
modeling are compared with ANSYS and is given in Table 4.4. Here, the non-dimensional
ρ
frequency is taken as, ω = ωL2
E11t 2
Table 4.4 Mesh convergence study of non-dimensional first natural frequency of isotropic
cantilever shell (L= 2m, a=1m, t=0.03 m).
47
It is observed from Table-4.4 that as R/a ratio reduce the non-dimensional first natural
a
frequency increases correspondingly. It is known that natural frequency, ^ ` ; where K is
b
the stiffness and M is the mass of the element. As the stiffness is only variable here, non-
dimensional natural frequency will increase if the stiffness of the element also increases
which describes the situation here, i.e. stiffness is increasing with the corresponding
reduction of R/a ratio.
4.3.2 Composite cantilever shell
The geometric properties are same as given in problem 4.2.2. The results obtained by
the present FE modeling are compared with ANSYS and is given in Table 4.5. The material
properties are as follows:
Density, J He \ f g
Table 4.5 Non-dimensional first natural frequency of composite cantilever shell (L= 2m,
a=1m, t=0.03 m.)
PRESENT FINITE
FIBRE
R/a RATIO ELEMENT ANSYS (ω )
ORIENTATION
(ω )
90/0/90
1.0081 1.0045
45/0/45
50 0.5384 0.5372
0/90/45/-45/45/90/0
0.6849 0.6825
ρ
Here, the non-dimensional frequency is taken as, ω = ωL2
E11t 2
It is observed from Table 4.5 that when most of the fibers are oriented along perpendicular to
X- axis (i.e. 90/0/90) the natural frequency show higher value than it is oriented along other
angle to the X- axis (i.e. 45/0/45). This is because the stiffness is getting higher for the first
case. Here it can be also observed that as the numbers of layers are increasing the first natural
frequency value is also increasing in the present case.
48
Figure- 4.2 Mode shape for 1st natural frequency (19.93 rad/sec) for fibre orientation 45/0/45
and R/a=50, L=2m, a=1m, t=0.03m for E-Glass Epoxy composite.
Figure- 4.3 Mode shape for 2nd natural frequency (105.96 rad/sec) for fibre orientation
45/0/45 and R/a=50, L=2m, a=1m, t=0.03m for E-Glass Epoxy composite.
49
Figure- 4.4 Mode shape for 3rd natural frequency (223.83 rad/sec) for fibre orientation
45/0/45 and R/a=50, L=2m, a=1m, t=0.03m for E-Glass Epoxy composite.
The material and the geometric properties are same as taken in section-4.3.1. The
results obtained by the present FE modeling are compared with ANSYS and is given in Table
N L2
4.6. Here, the non-dimensional buckling load is taken as, λ = x 3
E11t
Table 4.6 Non-dimensional first buckling load factor of isotropic cantilever shell (L= 2m,
a=1m, t=0.03 m).
PRESENT FINITE
R/a RATIO ELEMENT ANSYS (λ )
(λ )
15
0.2309 0.2292
50
0.2145 0.2144
100
0.2132 0.2133
50
Figure- 4.5 Finite element model of a cantilever shell subjected to unit mechanical
compressive load at the free edge. The mesh consists of (20 x 25) elements.
It is observed from Table 4.6 that the first buckling load factors increases
correspondingly with the decrement of the R/a ratio, i.e. with the increment of the stiffness.
The maximum error is of the order of 0.74 % for R/a ratio 15.
The material and geometric properties are same as given in section 4.3.2. The results
obtained by the present F.E modeling are compared with ANSYS and is given in Table 4.8.
Table 4.7 Non-dimensional first buckling load factor of composite cantilever shell for R/a
ratio 50 (L= 2m, a=1m, t=0.03 m).
PRESENT FINITE
FIBRE ORIENTATION ELEMENT ANSYS (λ )
(λ )
90/0/90
0.9839 0.9409
45/0/45
0.2614 0.2433
0/90/45/-45/45/90/0
0.4529 0.4360
51
Figure- 4.6 Mode shape for 1st buckling load factor (14116) for fibre orientation 45/0/45 and
R/a=50, L=2m, a=1m, t=0.03m for E-Glass Epoxy composite.
Figure- 4.7 Mode shape for 2nd buckling load factor (128080) for fibre orientation 45/0/45
and R/a=50, L=2m, a=1m, t=0.03m for E-Glass Epoxy composite.
52
Figure- 4.8 Mode shape for 3rd buckling load factor (358300) for fibre orientation 45/0/45
and R/a=50, L=2m, a=1m, t=0.03m for E-Glass Epoxy composite.
The developed FE code seems to be working well for static, free vibration and
buckling analysis. With this background, stability studies due to parametric excitation have
been carried out in the subsequent sections.
53
4.4.1 Isotropic cantilever shell
In this subsection an aluminum shell having R/a ratio 25, of the following particular
material and geometric properties has been considered. The shell is subjected to an axial load
and the stability curves with and without damping is plotted in Figure-4.9.
Material properties:
Material: Aluminium
Geometric properties:
Length, L = 2 m
Width, a = 1 m
Thickness, t = 0.03 m
Figure-4.9 Stability plot for cylindrical cantilever Aluminum shell with and without damping
(R/a = 25, L=2.0 m, a=1.0 m, thickness = 0.03 m, E = 70 GPa, ν = 0.3)
54
In Figure-4.9 the firm and dashed boundary line between stable and unstable region
depicts the boundary lines without and with damping, respectively. It is clear from the figure
that the presence of damping cuts off that part of the regions of instability which borders on
the axis of the ordinate, i.e. the damping decreases the unstable region by lifting it from delta
axis and narrowing its boundaries in the xi-delta plane.
In this section the study of the fibre orientation, effect of thickness, effect of various
geometric properties on the stability under separate subsections has been carried out for
different R/a ratios of a cylindrical cantilever shell. The laminated material properties are
given as follows:
Density, J He \ f g
90/ / 90
a) 90/0/90
b) 90/30/90
c) 90/90/90
/ /
a) 30/30/30
b) 45/45/45
c) 90/90/90
CASE-1
Considering the first case i.e. 0/ /0, the stability charts are given both for R/a ratio 15
and 100 with L=2m, a=1m, t=0.03m in Figure-4.10. As mentioned above same amount of
Rayleigh proportional damping i.e. 4 % for the first mode and 6 % for the second mode has
been introduced for all the cases and after studying the shift of the stability point in various
cases a plot has been shown in Figure-4.11.
55
(a) 0/0/0, R/a = 15 (b) 0/0/0, R/a = 100
Figure-4.10 Stability plots for different fibre orientations 0/ /0 with R/a ratio 15 and 100,
L=2m, a=1m, t=0.03 m for cantilever composite shell.
56
Figure-4.11 Variation of stability point for different fibre orientations 0/ /0 with R/a ratio 15
and 100.
Table 4.8 Non-dimensional 1st natural frequency and buckling load factor for 0/ /0
orientation for cylindrical shell of L=2m, a=1m and t=0.03m.
NON NON
FIBRE DIMENSIONAL DIMENSIONAL
R/a RATIO NATURAL BUCKLING LOAD
ORIENTATION
FREQUENCY (ω ) (λ )
0/0/0
0.4820 0.01844
15
0/30/0 0.4866
0.01966
0/90/0 0.02410
0.5312
0.05420
0/0/0 0.4617
100 0.04540
0/30/0 0.4633
0.05840
0/90/0 0.4943
k Kp LY
Where, ^
h ^i1 `j Y n as given in section 4.3.2 and o as given in section 4.4.2.
lmm lYY q
57
Itt can be observed from Figure-4.11
4.11 that,, with the variation of ‘ ’ the stability point
int
(xi-value)
alue) does not change in each case of R/a ratio. This can be explained after noticing
Table 4.8.
4.8 From that table one observes that, with the variation of ‘ ’ the first non non-
dimensional natural frequency
frequency and buckli
buckling
ng load factor does not change significantly in each
case of R/a ratio. The
he natural frequency does not change significantly which signifies that the
structural stiffness does not get altered significantly and hence the results
results.
The xi-value
xi value is nothing but a function (from Equation
quation-3.45c)
3.45c) of the ratio of buckling
load to the eigen frequency. Because the ratio of the change in the buckling load and the
eigen frequency is similar, that is why ‘xi’( *) remains constant. Hence the stability shi
shift
ft for
such a fibre orientation does not show any change.
The lower xixi-value
value for R/a ratio 15 than R/a ratio 100 suggests that, though the
damping is same for both the cases the portion of the instability zone in case of R/a ratio 100
gets much more reduced than the other case case i.
i.e. this damping is much more effective in case
of R/a ratio 100. That is why the xi xi-value shows higher value in this case. Alternately
lternately it can
be said that, because the shell has higher stiffness in casecase of R/a ratio 15, the value of the
stability point gets reduced in this case.
CASE-2
Now the fibre orientation 90/ /90 is being considered here. The geometric and
material properties are same as case-1.
case The various cases of stability plot have been shown
below.. Finally a plot has been given to show the shift of xi-value
xi value for different values.
Figure-4.12 Stability plots for different fibre orientations 90/ /90 with R/a ratio 15 and 100,
L=2m, a=1m, t=0.03 m for cantilever composite shell.
Figure-4.13 Variation of stability point for different fibre orientations 90/ /90 with R/a ratio
15 and 100.
59
Like Figure-4.11, it can also be seen in Figure-4.13 that the stability point does not
vary at all with the variation of i.e. the same explanation is also applicable here like as case-
1. But after comparing Figure-4.11 and 4.13 one thing can be noted that, in case of Figure-
4.13 the xi-value are much lower than the values shown in Figure-4.11. Hence from this one
thing can be concluded that, 90/ /90 fibre orientation is stiffer than 0/ /0 fibre orientation.
CASE-3
In this case / / fibre orientation has been considered. The geometric and material
properties are same as case-1 and 2. The various cases of stability plot have been shown
below. Finally a plot has been given to show the shift of xi-value for different values.
(a) 30/30/30, R/a = 15, = 18.39 rad/sec (b) 30/30/30, R/a = 100, = 17.57 rad/sec
(c) 45/45/45, R/a = 15, = 20.83 rad/sec (d) 45/45/45, R/a = 100, = 19.87 rad/sec
60
(e) 90/90/90, R/a = 15, = 38.97 rad/sec (f) 90/90/90, R/a = 100, = 37.79 rad/sec
Figure-4.14 Stability plots for different fibre orientations / / with R/a ratio 15 and 100,
L=2m, a=1m, t=0.03 m for cantilever composite shell.
Figure-4.15 Variation of stability point for different fibre orientations / / with R/a ratio 15
and 100
It can be observed from the result presented in Figure-4.15 that the fibre orientation
affects the stability plot significantly for / / fibre orientation. This can be attributed to the
fact that with the change in the fibre orientation the stiffness of the structure changes. This
61
can be observed by the marked change of the eigen frequencies given in Figure-4.14. As it
can be noted from Figure-4.15 that as the fibre orientation is going towards 900 the eigen
frequency is also increasing correspondingly i.e. the stiffness of the structure is also
increasing. That is why the xi-value is decreasing correspondingly or it can be said that, the
structure is becoming more unstable with the application of same amount of damping.
(a) R/a = 15, t = 0.002 m, = 5.09 rad/sec (b) R/a = 100, t = 0.002 m, = 1.36 rad/sec
(c) R/a = 15, t = 0.0075 m, = 6.56 rad/sec (d) R/a = 100, t = 0.0075 m, = 4.33 rad/sec
62
(e) R/a = 15, t = 0.015 m, = 9.88 rad/sec (f) R/a = 100, t = 0.015 m, = 8.57 rad/sec
Figure-4.16 Stability plots for different thickness variation with R/a ratio 15 and 100, L=2m,
a=1m and fibre orientation 0/0/0 for a cantilever composite shell.
Figure-4.17 Variation of stability point for different thickness values with R/a ratio 15 and
100, fibre orientation – 0/0/0, L = 2 m, a = 1 m of a cantilever composite cylindrical shell.
63
The stability plot for thickness 0.03 m has been shown earlier in Figures-4.10 (a) and
(b). It can be observed from the result presented in Figure-4.17 that the thickness variation
affects the stability plot significantly for 0/0/0 fibre orientation. This can be attributed to the
fact that with the change in thickness the stiffness of the structure also changes. This can be
observed by the marked change of the eigen frequencies in Figure-4.16 and Figure-4.10 (a) &
(b). As it can be noted from Figure-4.16 that, as the thickness is increasing the eigen
frequency is also increasing correspondingly i.e. the stiffness of the structure is also
increasing. That is why the xi-value is decreasing correspondingly or it can be said that the
same amount of damping is becoming ineffective with the increment of the thickness.
(a) L = 0.5m, a =1m, = 299.17 rad/sec (b) L= 0.5m, a =1m, = 293.16 rad/sec
(c) L = 1m, a =1m, = 74.45 rad/sec (d) L = 1m, a =1m, = 71.98 rad/sec
64
(e) L = 2 m, a =1m, = 18.39 rad/sec (f) L = 2 m, a =1m, = 17.57 rad/sec
Figure-4.18 Stability plots for different geometries with R/a ratio 15 and 100, t = 0.03 m,
fibre orientation 30/30/30 for a cantilever composite shell.
Figure-4.19 Variation of stability point for different L/a ratios with R/a ratio 15 and 100,
fibre orientation 30/30/30, t = 0.03 m of a cantilever composite cylindrical shell.
65
It can be seen from Figure-4.19 that the stability point shifts considerably with the
change in L/a ratio. This can be observed by the marked change of the eigen frequencies in
Figure-4.18. It is not clear to the present researchers what may be the exact reason for this
particular behavior. The change of geometry from square to rectangular plate, other stress-
strain relations and Poisson-ratio influence likely causes these effects.
66
CHAPTER-5
CONCLUSIONS
With the presently developed code, static, free-vibration and buckling analysis of
isotropic and composite shell structures have been carried out. By observing the results it is
very clear that the presently developed code gives extremely comparable results with FE
software like ANSYS. So keeping this thing in mind, parametric stability studies were carried
out for isotropic as well as composite cylindrical shell structures. Parametric instability
studies have been done with different fibre orientations, varying thickness and geometry and
with different R/a ratios in the present work.
Based on the analysis carried out in the earlier chapter the important conclusions
made are listed below:
The results are close to ANSYS result with 20 x 20 mesh in the mesh convergence
study for non-dimensional transverse tip deflection, in case of isotropic cantilever
shell.
Different R/a ratio and various fibre orientations play an important role in the static
transverse deflection of the shell. For lower R/a ratio, the transverse tip deflection is
lower, which signifies that the shell stiffness is higher in this case.
If the fibre orientation is along the perpendicular direction with the plane of curvature,
the shell shows higher natural frequency, and higher buckling load which signifies
higher stiffness.
Region of instability reduces with application of damping.
From Figure-4.11 and 4.13 it can be concluded that for 0/ /0 and 90/ /90
corresponding fibre orientations stability point does not change at all with the
variation of ‘ ’, i.e. for these fibre orientations the natural frequency and buckling
load factor do not vary significantly, which signifies the structural stiffness does not
get altered significantly. But for 0/ /0 fibre orientation the stability point shows higher
value than 90/ /90 fibre orientation, which signifies that because of the lower stiffness
the instability zone of 0/ /0 fibre oriented structure can be reduced much more than
90/ /90 fiber oriented structure with application of same amount of damping in both
cases.
67
With the application of same amount of damping, lower R/a ratio shows lower
stability point than the higher R/a ratio, which signifies that if the stiffness of the
structure increases the instability zone of the structure gets less reduced.
From Figure-4.15 it can be concluded that as the fibre orientation is along the
perpendicular direction with the plane of curvature, xi-value decreases
correspondingly or it can be said that, the structure is becoming more unstable with
the application of same amount of damping.
From Figure-4.17 it can be seen that as the ‘t/a’ ratio increases, the instability zone of
the structure increases with the application of same amount of damping. It is also
interesting to note that as the ‘t/a’ ratio is going towards a higher value (0.03 in this
case) the R/a ratio does not play any role in the parametric instability of the structure,
i.e. the xi-value converges to a same value as ‘t/a’ ratio increases.
From Figure-4.19 it can be seen that, as the ‘L/a’ ratio increases the xi-value also
increases, which signifies that the same amount of damping becomes more effective
with the increment of the ‘L/a’ ratio. It can also be seen from the figure that, in lower
‘L/a’ ratio the ‘R/a’ ratio plays insignificant role.
The work reported in this thesis is a limited part of a vast area of research on
structural stability studies due to parametric excitation and it’s control. There are number of
complex problems that need to be solved in the present area of research. Some of the
important aspects pertaining to the present work that need attention are listed below:
68
REFERENCES
[1] BIEZENO, C.B. and GRAMMEL, Engineering Dynamics, First English Edition,
Blackie and Son, London(1955); First German Edition entitled Technishe Dynamik,
Julius Springer, Berlin(1939).
[2] NICHOLAS J. HOFF, The process of the Buckling of Elastic Columns, PIBAL
Report No. 163, Polytechnic Institute of Brooklyn(1949). Also printed in J. Appl.
Mech., 18, 68(1951).
[3] FARADAY, M., 1831, “On a peculiar class of acoustical figures and on certain forms
assumed a group of particles upon vibrating elastic surfaces”, Philosophical
Transactions of the Society of London, 121, pp 299-318.
[4] MELDE, F. (1859) "Über Erregung stehender Wellen eines fadenförmigen Körpers"
[On the excitation of standing waves on a string], Annalen der Physik und Chemie
(Ser. 2), vol. 109, pages 193-215.
[5] STRUTT, J.W. (Lord Rayleigh) (1887) "On the maintenance of vibrations by forces
of double frequency, and on the propagation of waves through a medium endowed
with periodic structure", Philosophical Magazine, vol.24, pages 145-159.
[7] RAMAN, C. V., 1912 Proc. Indian Assoc. for the cultivation of Sci. Bull. 6
[8] BELIAEV, N.M. 1924, “Die stabilität prismatischer stäbe unter die wirkung
veränderlicher längskräfta” sbosnik, inshenernye soorushenija I stroitelnaja
mechanika”, sbosnik, inshenernye soorushenija i stroitelnaja mechanika”, 149-167,
Leningrad.
[9] ANDRONOV, A., & M.A. LEONTOVICH (1927) On the vibrations of systems with
periodically varying parameters. Zh. Russk. Fiz. –Khim. Obshch., 59,429-
443,259,267.
[10] LUBKIN, S. and J.J. STOKER, Stability of columns and strings under periodically
varying forces, Quarterly of Appl. Math., 1, 215(1943).
69
[11] METTLER, E Biegeschwingungen eines Stabes unter pulsierrender Axiallast,
Mitteilungen des Forschungs institutes des Gutehoffnunghütte-Konzerns, 8,
1(1940).
[12] GOLDENBLATT, I.I., “The Effect of Initial Forces on the Hydroelastic Vibration
and Stability of planar Curved Tubes” Modern Problems of Vibrations and
Resistance in Engineering Construction , 1947
[15] KRYLOV, N.M. and BOGOLIUBOV, N.N. Calculations of the vibrations of frame
construction with consideration of normal forces and with the help of the methods on
nonlinear mechanics, in Investigation of Vibration of Structures, ONTI Kharkov/Kiev,
1935, p. 5.
[16] CHELOMEI, V.N. (1939). The Dynamic Stability of Elements of Aircraft Structures.
Aeroflot, MOSCOW. 259, 266, 267, 268.
[19] BONDERENKO, G. V. (1936). The Hill Differential Equation and Its uses in
Engineering Vibration Problems. Akademü Nauk SSSR, MOSCOW. 259, 267,283,
295.
[20] AFSAR, K.R. and MASOUD, K.K., 1994, “Damping of parametrically excited single
degree of freedom system”, Int. Journal of nonlinear mechanics, 29, pp 421-428.
[21] HSU, C.S. On approximating a general linear periodic system, J. Math. Anal. Appl.
45 (1974) 234–251.
[22] CHEN, S.S. Flow Induced Vibration of Circular Cylindrical Structures, Hemisphere
Publishing Corporation,New York, 1987.
70
[23] GINSBERG, J.H. The dynamic stability of a pipe conveying a pulsatile 6ow, Internat.
J. Eng. Sci. 11 (1973)1013–1024.
[24] ISSID, N.T. PA DOUSSIS, M.P. Dynamic stability of pipes conveying 6uid, J. Sound
Vib. 33 (1974) 267–294.
[25] BHON, M. P., and HERRMANN, G. (1974). The dynamic behavior of articulated
pipes conveying fluid with periodic flow rate. J. Appl. Mech., 41, 55-62, 269.
[26] PA DOUSSIS, M.P. and ISSID, N.T. Dynamic stability of pipes conveying 6uid, J.
Sound Vib. 33 (1974) 267–294.
[28] NAGHDI, P. M., “A Survey of Recent Progress in the Theory of Elastic Shells”,
Appl.
[29] BERT, C. W., “Analysis of Shells,” Chapter 5 in Structural Design and Analysis, Part
I, C.C. Chamis(ed.), Vol.7 of Composite Materials, L. J. Broutman and R.H.
Krock(eds.), Academic Press, New York , pp. 207-258(1974b).
[30] KRAUS, H., Thin Elastic Shells, Wiley, New York, 1967.
[32] VLASOV, V. Z., General Theory of Shells and Its Applications in Engineering
(Translation of Obshcaya teoriya obolocheck i yeye prilozheniya v tekhnike), NASA
TT F-99, National Aeronautics and Space Administration, Washington, D.C., 1964.
[33] DONG, S. B., PISTER, K. S., and TAYLOR, R. L., “ On the Theory of Laminated
Anisotropic Shells and Plates,” J. Aerosp. Sci., Vol. 29 , pp. 969-975(1962).
71
[35] CHENG, Z. Q., He, L. H. and KITIPORNCHAI, S. Influence of Imperfect Interfaces
on Bending and Vibration of Laminated Composite Shells, Int. J. Solid Struct., vol.
37, pp. 2127–2150, 2000.
[36] WIDRA, G.E.O. and CHUNG, S.W., “A Theory for Non-Homogeneous Anisotropic
Cylindrical Shells,” Z. Angew. Math. Phys., Vol. 21, pp. 378-399(1970).
[37] WHITNEY, J.M. and SUN, C.T., “ A Refined Theory for Laminated Anisotropic,
Cylindrical Shells,” J. App. Mech., Vol. 41, pp. 471-476(1974).
[39] DATTA, P. K. and SAHU, S. K. “ Dynamic Stability of Curved Panels with Cutouts”
Journal of Sound and Vibration, 2003, Vol. 251, Issue 4, P 683-696.
[40] RAVI KUMAR,L.,DATTA, P.K., and PRABHAKARA, D.L., “Tension buckling and
dynamic stability behavior of laminated composite doubly curved panels subjected to
partial edge loading” Composite Structures 60 (2003) 171–181.
[42] PATEL,S.N., DATTA, P.K., and SHEIKH, A.H., “Buckling and dynamic instability
analysis of stiffened shell panels” Thin-Walled Structures 44 (2006) 321–333.
[43] BERT, C.W., and BIRMAN, V., “Parametric Instability of Thick, orthotropic,
[44] NAYFEH, A. H. and D. T. MOOK, ‘Nonlinear Oscillations’, John Willy and Sons,
New York.
72