Professional Documents
Culture Documents
F. A. Fazzolari
J. R. Banerjee
M. Boscolo
City University
School of Engineering and Mathematical Sciences
10 Northampton Square
London EC1V 0HB
UNITED KINGDOM
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing the burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson
Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply
with a collection of information if it does not display a currently valid OMB control number.
PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From – To)
09 December 2013 Final Report 20 September 2010 – 19 December 2013
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
FA8655-10-1-3084
Dynamic Stiffness Modeling of Composite Plate and Shell Assemblies 5b. GRANT NUMBER
Grant 10-3084
5c. PROGRAM ELEMENT NUMBER
61102F
6. AUTHOR(S) 5d. PROJECT NUMBER
F. A. Fazzolari
J. R. Banerjee 5d. TASK NUMBER
M. Boscolo
AFRL-AFOSR-UK-TR-2013-0061
12. DISTRIBUTION/AVAILABILITY STATEMENT
14. ABSTRACT
This grant sought to develop the dynamic stiffness method for composite shell assemblies. In the first part an exact dynamic
stiffness element based on higher order shear deformation theory and extensive use of symbolic algebra is developed for the
first time to carry out buckling analysis of composite plate assemblies. The principle of minimum potential energy is applied to
derive the governing differential equations and natural boundary conditions. The effects of significant parameters such as
thickness-to-length ratio, orthotropy ratio, number of layers, lay-up and stacking sequence and boundary conditions on the
critical buckling loads and mode shapes are investigated. In the second part of the grant an exact free vibration analysis of
laminated composite doubly-curved shallow shells was carried out by combining the dynamic stiffness method (DSM) and a
higher order shear deformation theory (HSDT) for the first time. The Wittrick-Williams algorithm is used as a solution technique
to compute the eigenvalues of the overall DS matrix.
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18, NUMBER 19a. NAME OF RESPONSIBLE PERSON
ABSTRACT OF PAGES Matt Snyder
a. REPORT b. ABSTRACT c. THIS PAGE
UNCLAS UNCLAS UNCLAS
SAR 72 19b. TELEPHONE NUMBER (Include area code)
+44 (0)1895 616420
Standard Form 298 (Rev. 8/98)
Prescribed by ANSI Std. Z39-18
Dynamic Stiffness Modelling of Composite Plate
and Shell Assemblies
Report 6/6
(3rd year report)
March 2013
1
PhD Candidate, Tel.: +44(0)20 7040 8483, E-mail: fiorenzo.fazzolari.1@city.ac.uk
2
Professor, Tel.: +44(0)20 7040 8924, E-mail: j.r.banerjee@city.ac.uk
3
Research Fellow, Tel.: +44(0)20 7040 8928, E-mail: marco.boscolo.1@city.ac.uk
Part I vi
Summary vii
1 Introduction 1
2 Theoretical formulation 3
2.1 Displacement field and derivation of governing differential equations . . . 3
2.2 Dynamic stiffness method . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Lèvy-type closed form exact solution and DS formulation . . . . . . . . . 7
2.4 Assembly procedure, boundary conditions and similarities with FEM . . . 14
2.5 Application of the Wittrick-Williams Algorithm . . . . . . . . . . . . . . . 15
Part II 20
Summary 21
4 Introduction 22
5 Theoretical Formulation 23
5.1 Governing differential equations . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2 Dynamic Stiffness Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.3 DS formulation and Lèvy-type closed form exact solution . . . . . . . . . 29
5.4 Assembly procedure and similarities with FEM . . . . . . . . . . . . . . . 38
7 Conclusions 46
References 51
ii
iii
Nomenclature
A Area
A, B, D, E, F , H Membrane, coupling and bending stiffness
b Width of the place
C Constitutive matrix of laminate
E Elastic modulus
f Elements of the dynamic stiffness matrix
G Shear modulus
h Thickness of plate
I0 , I1 , I2 , I3 , I4 , I5 , I6 Inertias of plate
j Number of natural frequencies lying below the trial frequency
j0 Number of clamped-clamped natural frequencies
k Shear correction factor
K Dynamic stiffness matrix
L Length of the plate
M , N , Q, P Moment and force resultants using HSDT
M, N , Q, P Amplitude of moment and force resultants using HSDT
Nl Number of layers
rp Roots of the polynomial for the in-plane case
ro Roots of the polynomial for the out-of-plane case
s Elements of the dynamic stiffness matrix
T Kinetic Energy
t Time
U Potential Energy
U, V, W Displacements amplitudes
u, v, w Displacements in the x, y, and z direction respectively
η Displacement vector
x, y, z Rectangular Cartesian coordinate system
iv
Symbols
α Half sin-wave
ε, ε Strain and strain vector
φ Rotations
Φ Amplitudes of rotation
ρ Density of plate material
σ, σ Stress and stress vector
ν Poisson’s ratio
vi
An exact dynamic stiffness element based on higher order shear deformation theory
and extensive use of symbolic algebra is developed for the first time to carry out buckling
analysis of composite plate assemblies. The principle of minimum potential energy is
applied to derive the governing differential equations and natural boundary conditions.
Then by imposing the geometric boundary conditions in algebraic form the dynamic
stiffness matrix, which includes contributions from both stiffness and initial pre-stress
terms, is developed. The Wittrick-Williams algorithm is used as solution technique to
compute the critical buckling loads and mode shapes for a range of laminated composite
plates. The effects of significant parameters such as thickness-to-length ratio, orthotropy
ratio, number of layers, lay-up and stacking sequence and boundary conditions on the
critical buckling loads and mode shapes are investigated. The accuracy of the method
is demonstrated by comparing results whenever possible with those available in the
literature.
vii
1 Introduction
Several methodologies have been developed over the years to solve the elastic stability
problem. A simplified approach to calculate the ith critical load, is to consider the critical
load as the load at which more than one infinitesimally adjacent equilibrium configu-
rations exist that can be identified with the ith bifurcation point (Euler’s method) [1].
In a linearized structural stability analysis, the determination of the critical load leads
to a linear eigenvalues problem. The bifurcation method can be successfully used par-
ticularly for plates, when the critical equilibrium configuration shows a slight geometry
change as the critical buckling load is reached. However, as explained by Leissa [2],
linearized stability analysis is meaningful, if and only if, the initial in-plane loading does
not produce an out-of-plane deformation. Furthermore, there are many cases in which
Euler’s method may fail, in particular when thin-walled structures like shells exhibit
the snap-buckling phenomenon. In such cases, the most general approach, based on the
solution of the complete equilibrium and stability equations [3, 4] is preferred.
Amongst a wide class of methodologies employed to analyze the elastic stability of ad-
vanced composite structures, the DSM is probably the most accurate and computation-
ally efficient option. The DSM based on Lèvy-type closed form solution for plates [5] is
indeed an exact approach to the solution procedure. Wittick [6] laid the groundwork of
the DSM for plates. The basic assumption in this work is that the deformation of any
component plate varies sinusoidally in the longitudinal direction. Using this assump-
tion, a stiffness matrix may be derived that relates the amplitudes of the edge forces and
moments to the corresponding edge displacements and rotations for a single component
plate. For the exact DSM, this stiffness matrix is derived directly from the equations of
equilibrium that describe the buckling behavior of the plate. Essentially, Wittrick [6] de-
veloped an exact stiffness matrix for a single isotropic, long flat plate subject to uniform
axial compression. His analysis basically used classical plate theory (CPT). Wittrick
and Curzon [7] later extended this analysis to account for the spatial phase difference
between the perturbation forces and displacements which occur at the edges of the plate
during buckling due to the presence of in-plane shear loading. This phase difference was
accounted for by defining the magnitude of these quantities using complex quantities.
Wittrick [8] then extended his analysis further to consider flat isotropic plates under
any general state of stress that remains uniform in the longitudinal direction (i.e., com-
binations of bi-axial direct stress and in-plane shear). A method very similar to that
described in [6] was also presented by Smith in [9] for the bending, buckling, and vibra-
tion of plate-beam structures. Following these developments, Williams [10] presented
two computer programs, GASVIP and VIPAL to compute the initial buckling stress of
prismatic plate assemblies subjected to uniform longitudinal stress or uniform longitu-
dinal compression, respectively. GASVIP was used to set up the overall stiffness matrix
for the structure, and VIPAL demonstrated the use of substructuring. Next, Wittrick
and Williams [11] reported on the VIPASA computer code for the buckling analysis of
prismatic plate assemblies. This code allowed for analysis of isotropic or anisotropic
plates using a general state of stress (including in-plane shear). The complex stiffnesses
described in [12] were incorporated in VIPASA, as well as allowances were made for
eccentric connections between component plates. This code also implemented an algo-
rithm, referred to as the Wittrick-Williams algorithm [13] for determining any buckling
load for any given wavelength. The development of this algorithm was necessary be-
cause the complex stiffnesses described above are transcendental functions of the load
factor and half wavelength of the buckling modes of the structure which make a deter-
minant plot cumbersome and unfeasible. Viswanathan and Tamekuni [14, 15] presented
an exact DSM based upon CPT for the elastic stability analysis of composite stiffened
structures subjected to biaxial in-plane loads. The structure was idealized as an as-
semblage of laminated plate elements (flat or curved) and beam elements. Tamekuni,
and Baker extended this analysis in [16] considering long curved plates subject to any
general state of stress, together with in-plane shear loads. Anisotropic material prop-
erties were also allowed. This analysis utilized complex stiffnesses as described in [12].
The works described in [9, 13, 16] are more or less similar. The differences are discussed
in [11]. Williams and Anderson [17] presented modifications to the eigenvalue algorithm
described in [13]. Further modifications presented in [17] allowed the buckling mode
corresponding to a general loading to be represented as a series of sinusoidal modes in
combination with Lagrangian multipliers to apply point constraints at any location on
edges. the DS matrix for laminated composite plates for buckling analysis. This useful
extension is of considerable theoretical and computational complexity as will be shown
later. The research is particularly relevant when analysing thick composite plates for
their buckling characteristics.
2 Theoretical formulation
2.1 Displacement field and derivation of governing differential equa-
tions
In the derivation that follows, the hypotheses of straightness and normality of a trans-
verse normal after deformation are assumed to be no longer valid for the displacement
field which is now considered to be a cubic function in the thickness coordinate, and
hence the use of higher order shear deformation theory (HSDT). This development is in
sharp contrast to earlier developments based on CPT and FSDT and no doubt a signif-
icant step forward. The deformation pattern through thickness of the plate is shown in
Fig. 1. A laminated composite plate composed of Nl layers is considered in order to make
the theory sufficiently general. The integer k is used as a superscript denoting the layer
number which starts from the bottom of the plate. The kinematics of deformation of a
transverse normal using both first order and higher order shear deformation are shown
in Fig. 1. After imposing the transverse shear stress homogeneous conditions [18, 19] at
the top/bottom surface of the plate, the displacements field are given below in the usual
form:
z
FSDT
HSDT x
x,y
3 ∂w0 (x, y, t)
u (x, y, z, t) = u0 (x, y, t) + z φx (x, y, t) + c1 z φx (x, y, t) +
∂x
3 ∂w 0 (x, y, t) (1)
v (x, y, z, t) = v0 (x, y, t) + z φy (x, y, t) + c1 z φy (x, y, t) +
∂y
w (x, y, z, t) = w0 (x, y, t)
where Πk is the total potential energy for the kth layer of the composite plate. The first
variation can be expressed as:
δΠk = δU k + δV k (4)
where δU k k
is the virtual potential strain energy, δV is the virtual potential energy due
to external loadings, and assume the following form:
Z Z Z Z
T
δU k = δεk σ k dΩk dz, δV k = δεnl nl k
xx σ̃x0 + δεyy σ̃y0 dΩ dz (5)
Ωk zk Ωk zk
σ̃x0 and σ̃y0 denote the in-plane initial stresses. The non-linear strains εnl nl
xx and εyy are
approximated with the Von Karman’s non-linearity:
1 1
εnl
xx = (w, x)2 εnl
yy = (w, y)2 (7)
2 2
The subscript T signifies an array transposition and δ the variational operator. Consti-
tutive and geometrical relationships are defined respectively as:
k
σ k = C̃ εk ε = Dη (8)
k
where C̃ is the plane stress constitutive matrix and D is the differential matrix (see
Appendix A for details). Substituting Eq. (42) into the Eq. (48) and imposing the
condition in Eq. (46), the equations of motion are obtained after extensive algebraic
manipulation as:
δu0 : A11 u0,xx + A12 v0,yx + A16 (u0,yx + v0,xx ) + B11 φx,xx + B12 φy,yx + B16 (φx,yx + φy,xx ) + E11 c2 φx,xx
+ E11 c2 w0,xxx + E12 c2 φy,yx + E12 c2 w0,yyx + E16 c2 φx,yx + E16 c2 φy,xx + 2 E16 c2 w0,xyx + A16 u0,xy
+ A26 v0,yy + A66 (u0,yy + v0,xy ) + B16 φx,xy + B26 φy,yy + B66 (φx,yy + φy,xy ) + E12 c2 (φx,xy + w0,xxy )
+ E26 c2 (φy,yy + w0,yyy ) + E66 c2 (φx,yy + φy,xy + 2 w0,xyy ) = 0
δv0 : A16 u0,xx + A26 v0,yx + A66 (u0,yx + v0,xx ) + B16 φx,xx + B26 φy,yx + B66 (φx,yx + φy,xx ) + E16 c2 φx,xx
+ E16 c2 w0,xxx + E26 c2 φy,yx + E26 c2 w0,yyx + E66 c2 φx,yx + E66 c2 φy,xx + 2 E66 c2 w0,xyx + A12 u0,xy
+ A22 v0,yy + A26 (u0,yy + v0,xy ) + B12 φx,xy + B22 φy,yy + B26 (φx,yy + φy,xy ) + E12 c2 (φx,xy + w0,xxy )
+ E22 c2 (φy,yy + w0,yyy ) + E26 c2 (φx,yy + φy,xy + 2 w0,xyy ) = 0
(9)
δw0 : A44 (φy,y + w0,yy ) + A45 (φx,y + w0,xy ) + D44 c1 (φy,y + w0,yy ) + D45 c1 (φx,y + w0,xy )
+ A45 (φy,x + w0,xy ) + A55 (φx,x + w0,xx ) + D45 c1 (φy,x + w0,xy ) + D55 c1 (φx,x + w0,xx )
+ D44 c1 (φy,y + w0,yy ) + D45 c1 (φx,y + w0,xy ) + F44 c21 (φy,y + w0,yy ) + F45 c21 (φx,y + w0,xy )
+ D45 c1 (φy,x + w0,xy ) + D55 c1 (φx,x + w0,xx ) + F45 c21 (φy,x + w0,xy ) + F55 c21 (φx,x + w0,xx )
− E11 c2 u0,xxx − E12 c2 v0,xxy − E16 c2 (u0,xxy + v0,xxx ) − F11 c2 φx,xxx − F12 c2 φy,xxy
− F16 c2 (φx,xxy + φy,xxx ) − H11 c22 (φx,xxx + w0,xxxx ) − H12 c22 (φx,xxy + w0,xxyy )
− H16 c22 (φx,xxy + φy,xxx + 2w0,xxxy ) − 2 E16 c2 u0,xxy − 2 E26 c2 v0,xyy − 2 E66 c2 (u0,xyy + v0,xxy )
− 2 F16 c2 φx,xxy − 2 F26 c2 φy,xyy − 2 F66 c2 (φx,xyy + φy,xxy ) − 2 H16 c22 (φx,xxy + w0,xxxy )
− 2 H26 c22 (φy,xyy + w0,xyyy ) − 2 H66 c22 (φx,xyy + φy,xxy + 2 w0,xxyy ) − E12 c2 u0,xyy − E22 c2 v0,yyy
− E26 c2 (u0,yyy + v0,xyy ) − F12 c2 φx,xyy − F22 c2 φy,yyy − F26 c2 (φx,yyy + φy,xyy )
− H12 c22 (φx,xyy + w0,xxyy ) − H22 c22 (φy,yyy + w0,yyyy ) − 2 H26 c22 (φx,yyy + φy,xyy + 2 w0,xyyy )
= Ñx0 w0,xx + Ñy0 w0,yy
δφx : B11 u0,xx + B12 v0,yx + B16 (u0,yx + v0,xx ) + D11 φx,xx + D12 φy,xy + D16 (φx,yx + φy,xx )
+ F11 c2 (φx,xx + w0,xxx ) + F12 c2 (φy,yx + w0,yyx ) + F16 c2 (φx,yx + φy,xx + 2 w0,xyx )
+ B16 u0,xy + B26 v0,yy + B66 (u0,yy + v0,xy ) + D16 φx,xy + D26 φy,yy + D66 (φx,yy + φy,xy )
+ F16 c2 (φx,xy + w0,xxy ) + F26 c2 (φy,yy + w0,yyy ) + F66 c2 (φx,yy + φy,xy + 2 w0,xyy )
+ E11 c2 u0,xx + E12 c2 v0,yx + E16 c2 (u0,yx + v0,xx ) + F11 c2 φx,xx + F12 c2 φy,xy + F16 c2 (φx,yx + φy,xx )
+ H11 c22 (φx,xx + w0,xxx ) + H12 c22 (φy,yx + w0,yyx ) + H16 c22 (φx,yx + φy,xx + 2 w0,xyx )
+ E16 c2 u0,xy + E26 c2 v0,yy + E66 c2 (u0,yy + v0,xy ) + F16 c2 φx,xy + F26 c2 φy,yy + F66 c2 (φx,yy + φy,xy )
+ H16 c22 (φx,xy + w0,xxy ) + H26 c22 (φy,yy + w0,yyy ) + H66 c22 (φx,yy + φy,xy + 2 w0,xyy )
− A45 (φy + 2 w0,y ) − A55 (φx + 2 w0,x ) − 2 D45 c1 (φy + 2 w0,y ) − 2 D55 c1 (φx + 2 w0,x )
− F45 c21 (φy + 2 w0,y ) − F55 c21 (φx + 2 w0,x ) = 0
δφy : B16 u0,xx + B26 v0,yx + B66 (u0,yx + v0,xx ) + D16 φx,xx + D26 φy,xy + D66 (φx,yx + φy,xx )
+ F16 c2 (φx,xx + w0,xxx ) + F26 c2 (φy,yx + w0,yyx ) + F66 c2 (φx,yx + φy,xx + 2 w0,xyx )
+ B12 u0,xy + B22 v0,yy + B26 (u0,yy + v0,xy ) + D12 φx,xy + D22 φy,yy + D26 (φx,yy + φy,xy )
+ F12 c2 (φx,xy + w0,xxy ) + F22 c2 (φy,yy + w0,yyy ) + F26 c2 (φx,yy + φy,xy + 2 w0,xyy )
+ E16 c2 u0,xx + E26 c2 v0,yx + E66 c2 (u0,yx + v0,xx ) + F16 c2 φx,xx + F26 c2 φy,xy + F66 c2 (φx,yx + φy,xx )
+ H16 c22 (φx,xx + w0,xxx ) + H26 c22 (φy,yx + w0,yyx ) + H66 c22 (φx,yx + φy,xx + 2 w0,xyx )
+ E12 c2 u0,xy + E22 c2 v0,yy + E26 c2 (u0,yy + v0,xy ) + F12 c2 φx,xy + F22 c2 φy,yy + F26 c2 (φx,yy + φy,xy )
+ H12 c22 (φx,xy + w0,xxy ) + H22 c22 (φy,yy + w0,yyy ) + H26 c22 (φx,yy + φy,xy + 2 w0,xyy )
− A44 (φy + 2 w0,y ) − A45 (φx + 2 w0,x ) − 2 D44 c1 (φy + 2 w0,y ) − 2 D45 c1 (φx + 2 w0,x )
− F44 c21 (φy + 2 w0,y ) − F45 c21 (φx + 2 w0,x ) = 0
δu0 : Nxx = A11 u0,x + B11 φx,x + E11 c2 φx,x + E11 c2 w0,xx + A12 v0,y + B12 φy,y + E12 c2 φy,y + E12 c2 w0,yy
+ A16 u0,y + A16 v0,x + B16 φx,y + B16 φy,x + E16 c2 φx,y + E16 c2 φy,x + 2 E16 c2 w0,xy
δv0 : Nxy = A16 u0,x + B16 φx,x + E16 c2 φx,x + E16 c2 w0,xx + A26 v0,y + B26 φy,y + E26 c2 φy,y + E26 c2 w0,yy
+ A66 u0,y + A66 v0,x + B66 φx,y + E66 c2 φy,x + E66 c2 φx,y + E66 c2 φy,x + 2 E66 c2 w0,xy
δw0 : Qx = H11 c22 φx,xx + H11 c22 w0,xxx + E11 c2 u0,xx + F11 c2 φx,xx + E12 c2 v0,yx + F12 c2 φy,yx
+ H12 c22 φy,yx + H12 c22 w0,yyx + 2 E16 c2 u0,xy + 2 F16 c2 φx,xy + 2 H16 c22 φx,xy + E16 c2 u0,yx
+ E16 c2 v0,xx + F16 c2 φx,yx + H16 c22 φx,yx + H16 c22 φy,xx + 2 H16 c22 w0,xxy + 2 E26 c2 v0,yy
+ 2 F26 c2 φy,yy + 2 H26 c22 w0,yyy + 4 H66 c22 w0,xyy + 2 H26 c22 φx,yy + 2 H26 c22 φy,xy + 2 E66 c2 u0,yy
+ 2 E66 c2 v0,xy + 2 F66 c2 φx,yy + 2 F66 c2 φy,xy − 2 D45 c1 φy − 2 D45 c1 w0,y − F45 c21 φy
− F45 c21 w0,y − A55 φx − A55 w0,x − D55 c1 φx − 2 c1 w0,x − F55 c21 φx − F55 c21 w0,x
δφy : Mxx = D11 φx,x + H11 c22 φx,x + H11 c22 w0,xx + B11 u0,x + E11 c2 u0,x + 2 F11 c2 φx,x + F11 c2 w0,xx
+ F11 c2 w0,xx + B12 v0,y + D12 φy,y + F12 c2 φy,y + F12 c2 w0,yy + E12 c2 v0,y + F12 c2 φy,y + H12 c22 φy,y
+ H12 c22 w0,yy + B16 u0,y + B16 v0,x + D16 φx,y + D16 φy,x + F16 c2 φx,y + F16 c2 φy,x + 2 F16 c2 w0,xy
+ E16 c2 u0,y + E16 c2 v0,x + F16 c2 φx,y + F16 c2 φy,x + H16 c22 φx,y + H16 c22 φy,x + 2 H16 c22 w0,xy
δφx : Mxy = D16 φx,x + H16 c22 φx,x + H16 c22 w0,xx + B16 u0,x + E16 c2 u0,x + 2 F16 c2 φx,x + F16 c2 w0,xx
+ F16 c2 w0,xx + B26 v0,y + D12 φy,y + F26 c2 φy,y + F26 c2 w0,yy + E26 c2 v0,y + F26 c2 φy,y + H26 c22 φy,y
+ H26 c22 w0,yy + B66 u0,y + B66 v0,x + D66 φx,y + D66 φy,x + F66 c2 φx,y + F66 c2 φy,x + 2 F66 c2 w0,xy
+ E66 c2 u0,y + E66 c2 v0,x + F66 c2 φx,y + F66 c2 φy,x + H66 c22 φx,y + H66 c22 φy,x + 2 H66 c22 w0,xy
δw0,x : Pxx = H11 c22 φx,x + H11 c22 w0,xx + E11 c2 u0,x + F11 c2 φx,x + E12 c2 v0,y + F12 c2 φy,y + H12 c22 φy,y
+ H12 c22 w0,yy + E16 c2 u0,y + E16 c2 v0,x + F16 c2 φx,y + F16 c2 φy,x + H16 c22 φx,y + H16 c22 φy,x
+ 2 H16 c22 w0,xy
(10)
where the suffix after the comma denotes the partial derivative with respect to that
variable and
Nl Z
X
k
1 z, z 2 , z 3 , z 4 , z 6 dz
(Aij , Bij , Dij , Eij , Fij , Hij ) = C̃ij
k
k=1 z
(11)
XNl Z
ρk 1 z, z 2 , z 3 , z 4 , z 6
(I0 , I1 , I2 , I3 , I4 , I6 ) = dz
k=1 zk
are laminate stiffnesses and rotatory inertial terms, respectively with i and j varying
form 1 to 6. The in-plane loadings can be defined as Ñx0 = λ Nx0 and Ñy0 = λ Ny0 ,
where Nx0 , Ny0 are the initial in-plane loadings and λ is a scalar load factor, c1 has
already been defined (see Eq. (37)) and c2 = − h42 .
(i) Seek a closed form analytical solution of the governing differential equations of the
structural element.
(ii) Apply a number of general boundary conditions in algebraic forms that are equal
to twice the number of integration constants; these are usually nodal displacements
and forces.
(iii) Eliminate the integration constants by relating the amplitudes of the harmonically
varying nodal forces to those of the corresponding displacements which essentially
generates the dynamic stiffness matrix, providing the force-displacement relation-
ship at the nodes of the structural element.
Referring to the equations of motion Eqs.(9), an exact solution can be found in Lèvy’s
form for symmetric, cross ply laminates. For such laminates B = E = 0, and C̃16 k =
k k
C̃26 = C̃45 = 0 and the out-of-plane displacements are uncoupled from the in-plane ones.
L21 = c1 Dx3 (F11 + c1 H11 ) + Dx −A55 − c2 (2D55 + c2 F55 ) − α2 c1 (F12 + 2F66 + c1 H12 + 2c1 H66 )
L22 = −A55 − c2 (2D55 + c2 F55 ) + Dx2 D11 + 2c1 F11 + c21 H11 − α2 D66 + 2c1 F66 + c21 H66
L23 = Dx −αD12 − αD66 − 2αc1 F12 − 2αc1 F66 − αc21 H12 − αc21 H66
L31 = −α A44 + c2 (2D44 + c2 F44 ) + α2 c1 (F22 + c1 H22 ) + αc1 Dx2 (F12 + 2F66 + c1 H12 + 2c1 H66 )
d
where Dx = dx and Aij , Bij , Cij , Dij , Eij , Fij , Hij have already been defined in Eq.
(57). Expanding the determinant of the matrix in Eq. (13) the following differential
equation is obtained:
where
Ψ = Wm , Φym , Φxm (16)
Using a trial solution eλ in Eq. (63) yields the following auxiliary equation:
a 1 λ8 + a 2 λ6 + a 3 λ4 + a 4 λ2 + a 5 = 0 (17)
Substituting µ = λ2 , the 8th order polynomial of Eq. (17) can be reduced to a quartic
as:
a 1 µ4 + a 2 µ3 + a 3 µ2 + a 4 µ + a 5 = 0 (18)
the four roots for the quartic equation are given by:
1 s8 s6 1√
r
µ1 = −s1 − −s5 + s2 − √ − − s9
2 4 s9 3 a 1 s7 2
1 s8 s6 1√
r
µ2 = −s1 + −s5 + s2 − √ − − s9
2 4 s9 3 a 1 s7 2
(19)
1 s8 s6 1√
r
µ3 = −s1 − −s5 + s2 + √ − + s9
2 4 s9 3 a 1 s7 2
1 s8 s6 1√
r
µ4 = −s1 + −s5 + s2 + √ − + s9
2 4 s9 3 a 1 s7 2
where
a1 4 a3 a22
s1 = , s2 = − + , s3 = 2 a3 − 9 a2 a3 a4 − 72 a22 a5 + 27 a22 a5 + 27 a1 a24 ,
4 a2 3 a21 2 a21
√ 13
s3 − 4 s4
1 s3 +
s4 = a23 − 3 a2 a4 + 12 a1 a5 , s5 = ,
a1 32
√ 3 √ 3
a2
3
4 a2 a3 8 a4 s2 s6
s6 = 22 s4 , s7 = 32 s5 a1 , s8 = − , s9 = s5 + +
a1 a21 a1 2 3 s7 a 1
(20)
Eqs. (66) into (13) the following relationships are obtained using symbolic computation:
A1 = δ1 B1 , A2 = −δ1 B2 , C1 = γ 1 B 1 , C2 = −γ1 B2
A3 = δ2 B3 , A4 = −δ2 B4 , C3 = γ 2 B 3 , C4 = −γ2 B4
(22)
A5 = δ3 B5 , A6 = −δ3 B6 , C5 = γ 3 B 5 , C6 = −γ3 B6
A7 = δ4 B7 , A8 = −δ4 B8 , C7 = γ 4 B 7 , C8 = −γ4 B8
where
h
2 2 2 4 2 2
δi = − − A55 α D22 − 2A55 c2 D44 − 2α c2 D22 D55 − 4c2 D44 D55 − α D22 D66 − 2α c2 D44 D66 − 2A55 α c1 F22
2 4 2 3 2 2 2 2 3
− 4α c1 c2 D55 F22 − 2α c1 D66 F22 − A55 c2 F44 − 2c2 D55 F44 − α c2 D66 F44 − α c2 D22 F55 − 2c2 D44 F55
2 2 4 4 2 4 2 2 2
− 2α c1 c2 F22 F55 − c2 F44 F55 − 2α c1 D22 F66 − 4α c1 c2 D44 F66 − 4α c1 F22 F66 − 2α c1 c2 F44 F66
2 2 2 2 4 2 2 2 2 4 3 4 2 2 2
− A55 α c1 H22 − 2α c1 c2 D55 H22 − α c1 D66 H22 − α c1 c2 F55 H22 − 2α c1 F66 H22 − α c1 D22 H66
− 2α c1 c2 D44 H66
4 3 2 2 2 4 4 2 2 2
− 2α c1 F22 H66 − α c1 c2 F44 H66 − α c1 H22 H66 − A44 A55 + 2c2 D55 + c2 F55 + α (D66 + 2c1 F66 + c1 H66 )
2
+ A44 (D11 + c1 (2F11 + c1 H11 ))µi + A55 (D66 + c1 (2F66 + c1 H66 )) + 2c2 (D11 D44 + D55 D66 + c1 (2D44 F11
2
+ 2D55 F66 + c1 D44 H11 + c1 D55 H66 )) + c2 (D11 F44 + D66 F55 + c1 (2F11 F44 + 2F55 F66 + c1 F44 H11 + c1 F55 H66 ))
2 2
−α (D12 − D11 (D22 + c1 (2F22 + c1 H22 )) + 2D12 (D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 ))) + c1 (4F12 (D66 + c1 F12 )
− D22 (2F11 + c1 H11 ) + c1 (8F12 F66 + 2D66 H12 − 2F11 (2F22 + c1 H22 ) + c1 (−2F22 H11 + H12 (4(F12 + F66 ) + c1 H12 )
i
2 4
− c1 H11 H22 + 4F12 H66 + 2c1 H12 H66 )))) µi − (D11 + c1 (2F11 + c1 H11 ))(D66 + c1 (2F66 + c1 H66 ))µi
h
2 2 2
/ α (D12 + D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 )))µi A44 + 2c2 D44 + c2 F44 + α c1 (F22 + c1 H22 )
2 2 2
− c1 (F12 + 2F66 + c1 H12 + 2c1 H66 )µi − µi A55 + 2c2 D55 + c2 F55 + α c1 (F12 + 2F66 + c1 H12 + 2c1 H66 )
i
2 2 2 2 2
− c1 (F11 + c1 H11 )µi A44 + c2 (2D44 + c2 F44 ) + α (D22 + 2c1 F22 + c1 H22 ) − (D66 + 2c1 F66 + c1 H66 )µi
h
2 2 2 2 2
γi = α A44 A55 + 2A55 c2 D44 + 2A44 c2 D55 + 4c2 D44 D55 + A44 α D66 + 2α c2 D44 D66 + A55 α c1 F22 + 2α c1 c2 D55 F22
4 2 3 2 2 2 3 2 2 4
+ α c1 D66 F22 + A55 c2 F44 + 2c2 D55 F44 + α c2 D66 F44 + A44 c2 F55 + 2c2 D44 F55 + α c1 c2 F22 F55 + c2 F44 F55
2 2 4 2 2 2 2 2 2 2 4 2
+ 2A44 α c1 F66 + 4α c1 c2 D44 F66 + 2α c1 F22 F66 + 2α c1 c2 F44 F66 + A55 α c1 H22 + 2α c1 c2 D55 H22 + α c1 D66 H22
2 2 2 4 3 2 2 2 2 4 3 2 2 2 4 4
+ α c1 c2 F55 H22 + 2α c1 F66 H22 + A44 α c1 H66 + 2α c1 c2 D44 H66 + α c1 F22 H66 + α c1 c2 F44 H66 + α c1 H22 H66
+ A55 (D12 + D66 ) − A44 (D11 + c1 (2F11 + c1 H11 )) − 2c2 (D11 D44 − D55 (D12 + D66 ) + c1 (2D44 F11 − D55 F12 + c1 D44 H11
2
+ c1 D55 H66 )) + c2 (−D11 F44 + (D12 + D66 )F55 − c1 (2F11 F44 − F12 F55 + c1 F44 H11 + c1 F55 H66 )) + c1 (A55 (F12 − c1 H66 )
2 2
+ α (−D11 F22 + D12 (F12 + 2F66 + c1 (H12 + 2H66 )) + c1 (2F12 − D11 H22 − 2F11 (F22 + c1 H22 ) + F12 (4F66 + 3c1 H12
2
+ 4c1 H66 ) + c1 (H12 (2F66 + c1 H12 ) − H11 (F22 + c1 H22 ) + 2c1 H12 H66 )))) µi + c1 (D11 (F12 + 2F66 ) − D12 (F11 + c1 H11 )
i
4
− D66 (F11 + c1 H11 ) + c1 (D11 (H12 + 2H66 ) + c1 H11 (−F12 + c1 H66 ) + F11 (2F66 + c1 (H12 + 3H66 ))))µi
h
2 2 2 2 2
/ µi A44 A55 − A44 α D12 + A55 α D22 + 2A55 c2 D44 − 2α c2 D12 D44 + 2A44 c2 D55 + 2α c2 D22 D55 + 4c2 D44 D55
2 2 2 4 2 2 4 2
− A44 α D66 − 2α c2 D44 D66 − A44 α c1 F12 + α c1 D22 F12 − 2α c1 c2 D44 F12 + 2A55 α c1 F22 − α c1 D12 F22 + 4α c1 c2 D55 F22
4 2 2 2 3 2 2 2 2 2 2 2
− α c1 D66 F22 + A55 c2 F44 − α c2 D12 F44 + 2c2 D55 F44 − α c2 D66 F44 − α c1 c2 F12 F44 + A44 c2 F55 + α c2 D22 F55
3 2 2 4 4 4 2 4 2 4 3 2 2
+ 2c2 D44 F55 + 2α c1 c2 F22 F55 + c2 F44 F55 + 2α c1 D22 F66 + 2α c1 F22 F66 + α c1 D22 H12 + α c1 F22 H12 + A55 α c1 H22
4 2 2 2 4 2 4 3 2 2 2 2 2 4 2
− α c1 D12 H22 + 2α c1 c2 D55 H22 − α c1 D66 H22 − α c1 F12 H22 + α c1 c2 F55 H22 + A44 α c1 H66 + 2α c1 D22 H66
2 2 4 3 2 2 2 4 4 2
+ 2α c1 c2 D44 H66 + 3α c1 F22 H66 + α c1 c2 F44 H66 + α c1 H22 H66 − A55 (D66 + c1 (2F66 + c1 H66 )) + c2 (D66 F55
+ c1 (F11 F44 + 2F55 F66 + c1 F44 H11 + c1 F55 H66 )) + 2c2 (c1 D44 (F11 + c1 H11 ) + D55 (D66 + c1 (2F66 + c1 H66 )))
2
+ c1 (A44 (F11 + c1 H11 ) + α (−F12 (D12 + 2c1 F12 ) − 2D12 F66 + D22 (F11 + c1 H11 ) + c1 (−4F12 F66 + F11 (2F22 + c1 H22 )
2
− D12 (H12 + 2H66 ) + c1 (2F22 H11 − H12 (3F12 + 2F66 + c1 H12 ) + c1 H11 H22 − 4F12 H66 − 2c1 H12 H66 )))) µi
i
4
+ c1 (F11 + c1 H11 )(D66 + c1 (2F66 + c1 H66 ))µi
(23)
with i = 1, 2, 3, 4. The procedure leading to Eqs. (22) and (23) must be undertaken
with sufficient care, because if wrong equations are chosen from Eq. (66) to obtain the
relationship connecting different sets of constant, numerical instability can occur. When
10
Eqs. (22) are substituted into Eqs. (66) a solution in terms of only 8 constants can be
formulated for Wm (x), Φxm (x) and Φym (x), respectively. Thus
The expressions for forces and moments can also be found in the same way by substi-
tuting Eqs. (69) into Eqs. (10) and using symbolic computation. In this way
Qx (x, y) = eµ1 x (B1 + B2 e−2 µ1 x )(A55 + A55 δ1 µ1 + 2 c2 (D55 + D55 δ1 µ1 ) + c22 (F55 + δ1 F55 µ1 )
Mxx (x, y) = eµ1 x (B1 + B2 e−2 µ1 x )(α2 c1 δ1 (F12 + c1 H12 ) + α γ1 (D12 + c1 (2 F12 + c1 H12 )) − µ1 (D11
+ c1 (2 F11 + c1 H11 + δ1 F11 µ1 + c1 δ1 H11 µ1 )))+
eµ2 x (B3 + B4 e−2 µ2 x )(α2 c1 δ2 (F12 + c1 H12 ) + α γ2 (D12 + c1 (2 F12 + c1 H12 )) − µ2 (D11
+ c1 (2 F11 + c1 H11 + δ2 F11 µ2 + c1 δ2 H11 µ2 )))+
eµ3 x (B5 + B5 e−2 µ3 x )(α2 c1 δ3 (F12 + c1 H12 ) + α γ3 (D12 + c1 (2 F12 + c1 H12 )) − µ3 (D11
+ c1 (2 F11 + c1 H11 + δ3 F11 µ3 + c1 δ3 H11 µ3 )))+
eµ4 x (B7 + B8 e−2 µ4 x )(α2 c1 δ4 (F12 + c1 H12 ) + α γ4 (D12 + c1 (2 F12 + c1 H12 )) − µ4 (D11
+ c1 (2 F11 + c1 H11 + δ4 F11 µ4 + c1 δ4 H11 µ4 ))) sin (α y) = Mxx sin (α y)
11
Mxy (x, y) = eµ1 x (B1 + B2 e−2µ1 x )(γ1 (D66 + c1 (2 F66 + c1 H66 )) µ1 + α (D66 + c1 (2 F66 + c1 H66
+ 2 δ1 F66 µ1 + 2 c1 δ1 H66 µ1 )))+
eµ2 x (B1 + B2 e−2µ2 x )(γ2 (D66 + c1 (2 F66 + c1 H66 )) µ2 + α (D66 + c1 (2 F66 + c1 H66
+ 2 δ2 F66 µ1 + 2 c1 δ2 H66 µ2 )))+
eµ3 x (B1 + B2 e−2µ3 x )(γ3 (D66 + c1 (2 F66 + c1 H66 )) µ3 + α (D66 + c1 (2 F66 + c1 H66
+ 2 δ3 F66 µ3 + 2 c1 δ3 H66 µ2 )))+
eµ4 x (B1 + B2 e−2µ4 x )(γ4 (D66 + c1 (2 F66 + c1 H66 )) µ4 + α (D66 + c1 (2 F66 + c1 H66
+ 2 δ4 F66 µ1 + 2 c1 δ4 H66 µ4 ))) cos (α y) = Mxy cos (α y)
Pxx (x, y) = eµ1 x (−B1 + B2 e−2 µ1 x ) (α2 c1 δ1 H12 + α γ1 (F12 + c1 H12 ) − µ1 (F11 + c1 H11 (1 + δ1 µ1 )))+
eµ2 x (−B1 + B2 e−2 µ2 x ) (α2 c1 δ2 H12 + α γ2 (F12 + c1 H12 ) − µ2 (F11 + c1 H11 (1 + δ2 µ2 )))+
eµ3 x (−B1 + B2 e−2 µ3 x ) (α2 c1 δ3 H12 + α γ3 (F12 + c1 H12 ) − µ3 (F11 + c1 H11 (1 + δ3 µ3 )))+
eµ4 x (−B1 + B2 e−2 µ4 x ) (α2 c1 δ4 H12 + α γ4 (F12 + c1 H12 ) − µ4 (F11 + c1 H11 (1 + δ4 µ4 )))
sin (α y) = Pxx sin (α y)
(25)
At this point, zero boundary conditions are generally used to eliminate the constants
when using the classical method which establishes the stability equation for a single
individual plate. By contrast, the development of the dynamic stiffness matrix entails
imposition of general boundary conditions in algebraic form and widens the possibility
of the analysis of multi-plate systems. In order to develop the dynamic stiffness matrix,
the following boundary conditions are applied next.
x=0 : Wm = Wm1 , Φxm = Φx1 , Φym = Φy1 , Wm,x = Wm1, x
x=b : Wm = Wm2 , Φxm = Φx2 , Φym = Φy2 , Wm,x = Wm2, x
(26)
x=0 : Qx = −Qx1 , Mxx = −Mxx1 , Mxy = −Mxy1 , Pxx = −Pxx1
x=b : Qx = Qx2 , Pxx = Pxx2 , Mxy = Mxy2 , Pxx = Pxx2
By substituting Eq. (26) into Eq.(69), the following matrix relations for the displace-
ments are obtained:
W1 δ1 −δ1 δ2 −δ2 δ3 −δ3 δ4 −δ4
B1
Φx 1 1 1 1 1 1 1 1 1 B2
Φy1 γ1 −γ1 γ2 −γ2 γ3 −γ3 γ4 −γ4 B3
W1 , x fb1µ −f1 f2 −f2 f3 −f3 f4 −f4 B4
= −b µ b µ −b µ b µ −b µ b µ −b µ
W2 δ1 e o1 −δ1 e o1 δ2 e o2 −δ2 e o2 δ3 e o3 −δ3 e o3 δ4 e o4 −δ4 e o4 B5
Φx 2 eb µo1 −b µo1 b µo2 −b µo2 b µo3 −b µo3 b µo4 −b µo4
−e e −e e −e e −e B6
Φy2 γ1 eb µo1 −γ1 e−b µo1 γ2 eb µo2 −γ2 e−b µo2 γ3 eb µo3 −γ3 e−b µo3 γ4 eb µo4 −γ4 e−b µo4 B7
W2 , x f1 eb µo1 −f1 e−b µo1 f2 eb µo2 −f2 e−b µo2 f3 eb µo3 −f3 e−b µo3 f4 eb µo4 −f4 e−b µo4 B8
(27)
where
fi = δ i µi ; with i = 1, 2, 3, 4
Equations (82) and (27) can be written as
δ = AC (28)
12
By applying the same procedure for forces and moments, i.e. substituting Eq. (26) into
Eq.(25) the following matrix relations are obtained:
Q1 Q1 Q2 Q2 Q3 Q3 Q4 Q4
Qx
1 B1
Mxx1 T1 −T1 T2 −T2 T3 −T3 T4 −T4 B2
Mxy1 −I1 −I1 −I2 −I2 −I3 −I3 −I4 −I4 B3
Pxx1 L1 −L1 L2 −L2 L3 −L3 YL −L4 B4
=
Q1 eb µo1 −Q1 e−b µo1 Q2 eb µo2 −Q2 e−b µo2 Q3 eb µo3 −Q3 e−b µo3 Q4 eb µo4 −Q4 e−b µo4
Q x2 B5
−T1 eb µo1 T1 e−b µo1 −T2 eb µo2 T2 e−b µo2 −T3 eb µo3 T3 e−b µo3 −T4 eb µo4 T4 e−b µo4
Mxx2 B6
Mxy2 I1 eb µo1 I1 e−b µo1 I2 eb µo2 I2 e−b µo2 I3 eb µo3 I3 e−b µo3 I4 eb µo4 I4 e−b µo4 B7
Mxx2 −L1 eb µo1 L1 e−b µo1 −L2 eb µo2 L2 e−b µo2 −L3 eb µo3 L3 e−b µo3 −L4 eb µo4 L4 e−b µo4 B8
(29)
where
Qi = −A55 (1 + δi µoi ) − 2c2 (D55 + D55 δi µoi ) − c22 (F55 + δi F55 µoi ) − c1 (αγi (F12 + 2F66 + c1 H12
+ 2c1 H66 )µoi − µ2oi (F11 + c1 H11 + c1 δi H11 µoi ) + α2 (2F66 + 2c1 H66 + c1 δi H12 µoi + 4c1 δi H66 µoi ))
Ti = α2 c1 δi (F12 + c1 H12 ) + αγi (D12 + c1 (2F12 + c1 H12 )) − µoi (D11 + c1 (2F11 + c1 H11
+ δi F11 µoi + c1 δi H11 µoi ))
Ii = γ1 (D66 + c1 (2F66 + c1 H66 ))µoi − α(D66 + c1 (2F66 + c1 H66 + 2δi F66 µoi + 2c1 δi H66 µoi ))
Li = c1 (α2 c1 δi H12 + αγi (F12 + c1 H12 ) − µoi (F11 + c1 H11 (1 + δi µoi ))) with i = 1, 2, 3, 4
(30)
13
Finally the dynamic stiffness matrix related to the force and displacement vectors can
be written as follows:
Qx 1 sqq sqm sqt sqh fqq fqm fqt fqh W1
Mxx1 smm smt smh −fqm fmm fmt fmh Φx1
Mxy1 stt sth fqt −fmt ftt fth Φy1
−fqh −fth
Pxx1
shh fmh fhh
W 1, x
= (34)
Qx 2 Sym sqq −sqm sqt −sqh
W2
Mxx2 smm −smt smh Φx2
Mxy2 −sth Φy2
stt
Pxx2 shh W 2, x
14
Because of the similarities between DSM and FEM, DS elements can be implemented in
FEM codes and thus the accuracy of results can be increased substantially.
1 2 N
z
j = j0 + s(K ∗ ) (36)
where j0 is the number of critical buckling loads of all single strip elements within the
structure which are still lower than the trial buckling load (λ∗ ) when their opposite
sides are fully clamped. It is necessary to account for this clamped-clamped critical
buckling loads because exact buckling analysis using DSM allows an infinity number of
critical buckling loads to be accounted for when all the nodes of the structures are fully
clamped.(i.e. in the overall formulation K δ = 0, these critical buckling loads correspond
15
y
Ny0
L Nx0
x
b
for moderately thick (a/h = 10) simply-supported cross-ply square plates uniaxially
loaded in the x direction is carried out and the results are shown in Table 1 for different
orthotropy ratos. The dimensionless critical buckling load, obtained using HSDT within
the framework of the DSM are in excellent agreement when compared with the 3D
elasticity solution and the results also lead to the same findings of the classical Lèvy-
type closed form solution. Note that for all practical purpose, it is only the first buckling
load that matters. Therefore only the first critical loads is presented in this paper. As
expected the percentage error, with respect to the 3D elasticity solution, increases when
increasing the orthotropic ratio. In Table 2 the dimensionless critical buckling load for
the same case study of Table 1 is computed but taking into account the effects of the
length-to-thickness and orthotropy ratios and boundary conditions (see Fig. 4). At a
16
y y y
x x x
y y y
x x x
2
Table 1: Dimensionless uniaxial buckling load (along x direction) Ncr = N̄cr E2b h3 , for
simply supported cross-ply square plates with b/h = 10, E1 /E2 = open, G12 /E2 =
G13 /E2 = 0.6, G23 /E2 = 0.5, ν12 = ν13 = 0.25.
Classical Lèvy’s solution HSDT 5.393 (1.68) 15.298 (1.86) 23.340 (2.01)
FSDT 5.399 (1.79) 15.351 (2.21) 23.453 (2.50)
CLPT 5.754 (8.48) 19.712 (31.2) 36.160 (58.0)
17
fixed length-to-thickness ratio, the dimensionless critical buckling load increases when
increasing the orthotropic ratio for all the considered boundary conditions. A similar
behavior can be observed when varying the length-to-thickness ratio but by fixing the
orthotropic ratio. Understandably, the largest dimensionless critical buckling load is
2
Table 2: Dimensionless uniaxial buckling load (along x direction) Ncr = N̄cr E2b h3 ,
for simply supported cross-ply square plates, stacking sequence [0◦ /90◦ /90◦ /0◦ ] and
E1 /E2 = open, G12 /E2 = G13 /E2 = 0.6, G23 /E2 = 0.5, ν12 = ν13 = 0.25.
given by the boundary condition S-C-S-C and the lowest by S-F-S-F. In Table 3 the
18
results are given for composite plates that are uniaxially loaded in the y direction, instead
of the x direction for different values of length-to-thickness and orthotropy ratios. The
dimensionless critical buckling load is generally lower for all the boundary conditions
but for for the case with one or two sides free, namely, S-S-S-F and S-F-S-F, it decreases
significantly.
2
Table 3: Dimensionless uniaxial buckling load (along y direction) Ncr = N̄cr E2b h3 ,
for simply supported cross-ply square plates, stacking sequence [0◦ /90◦ /90◦ /0◦ ] and
E1 /E2 = open, G12 /E2 = G13 /E2 = 0.6, G23 /E2 = 0.5, ν12 = ν13 = 0.25.
19
20
21
4 Introduction
Aerospace structures are generally made up of thin-walled cylindrical or spherical shell
components. The application of composite shell structures is justified because of the
extraordinary load-carrying capability. Furthermore, their structural efficiency is char-
acterized by a high stiffness-to-weight and strength-to-weight ratios. Due to the harsh
environment conditions in which aerospace structures work a comprehensive and accu-
rate investigation of their dynamic behavior is required. During the last decades many
theories have been developed to this purpose. In particular, most of the classical theo-
ries are mainly based on Kirchhoff-Love’s hypotheses [25,26]. It is well-known that these
shell theories neglecting the transverse shear and normal stresses lead to accurate results
when high values of length-to-thickness ratio are considered and when 3D local effects
are not present. When moderately thick or thick shells are investigated the inclusion
of transverse shear and normal stresses are mandatory. However, on the other hand, a
refinement in results accuracy cannot be only achieved by virtue of an enhancement in
the kinematics description of the displacement model, but combining it with an adequate
description of the curvature terms h/Ri with i = α, β. In this regards, some interesting
observations have been provided by Qatu [27] and, Carrera [28, 29] and recently Fazzo-
lari [5,30]. In order to provide solutions of the Governing Differential Equations (GDEs)
of practical interest, considerable efforts were focused and devoted to their simplifica-
tion. As a result, the GDEs of shallow shells were derived. Most notably, it worths
highlighting the articles of Donnell [31, 32], Mushtari [33, 34] and Vlasov [35, 36]. The
hypothesis of thinness was discarded in the shell theories proposed by Flügge [37, 38],
Lur’e [39] and Byrne [40], where higher-order expansion of the reciprocal of the Lamé
parameters was considered. Other refinements of the developed shell theories were pro-
posed by Sanders [41]. Additional effects in the development of shell theories were taken
into account by Whitney and Sun [42], Librescu [43], Gulati and Essemberg [44], Zucas
and Vinson [45] and Ambartsumian [46–53] amongst others. Additional references can
be found in Naghdi [54], Ambartsumian [55] and Bert [56–58]. Reddy [59] proposed a
generalization of Sander’s theory to anisotropic doubly-curved shells. The application
of layer-wise theories for shell structures can be found in the papers presented by Hsu
and Wang [60], Cheung and Wu [61], Barbero et al. [62] and Carrera [28, 29, 63]. Re-
views on finite element shell formulations have been given by Denis and Palazzotto [64]
and Di and Ramm [65]. An exhaustive review on classical theories can be found in
Librescu [43]. As regards the use of approximation methods, Qatu and Asadi [66] ad-
dressed the vibration analysis of doubly-curved shallow shells with arbitrary boundary
conditions by using the Ritz method with algebraic polynomial displacement functions.
Asadi et al. [67] employed a 3D and several shear deformation theories in order to
carry out static and vibration analysis of thick deep laminated cylindrical shells. Fer-
reira et al. [68] used a wavelet collocation method for the analysis of laminated shells.
The same author [69] combined a sinusoidal shear deformation theory with the radial
basis functions collocation method to deal with static and vibration analyses of lami-
nated composite shells. Tornabene et al. [70, 71] studied the free vibration behavior of
22
5 Theoretical Formulation
In the derivation that follows, the hypotheses of straightness and normality of a trans-
verse normal after deformation are assumed to be no longer valid for the displacement
field which is now considered to be a cubic function in the thickness coordinate and
hence the use of higher order shear deformation theory (HSDT). The composite circular
cylindrical shell is assumed to be composed of Nl layers so that the theory is sufficiently
general. The geometry of the shell and the coordinate system are shown in Fig. 5.
The integer k is used as a superscript denoting the layer number where the numbering
starts from the bottom. After imposing the transverse shear stress homogeneous con-
ditions [19, 86] at the top/bottom surface of the shell, the displacements field is in the
23
where u, v, w are general displacements within the shell in the α, β, and z directions,
respectively, whereas u0 , v0 , w0 are the corresponding displacements of the reference
surface (mid-plane Ω), φα , φβ are the rotations of the normals to the α, β, respectively,
Aα , Aβ are the surface metrics and Rα and Rβ are the radii of curvature in the α and β
directions. The geometry of the shell is completely described by the parameters given in
Fig. 5. Most notably, r (α, β) is the position vector of a point on the middle surface Ω of
z z
Middle Surface W
n
n
p
z p
p r,a r,b
dr
p
h h
a a
R b
x3 r b
x3
r r + dr
e3 e3
e2 x2 x2
e1 e2
e1
x1 x1
the shell, R (α, β, z) is the position vector of a generic point within the volume occupied
by the shell. At each point P of the middle surface n (α, β) indicates the unit normal
vector. Subscripts and superscripts k refers that quantity at layer level. The square of
line elements in orthogonal curvilinear coordinates is therefore defined as:
2 2 2
ds2k = dr k · dr k = Hαk dαk2 + Hβk dβk2 + Hzk dzk2 (38)
24
where !
zk zk
Hαk = Akα 1+ k Hβk = Akβ 1+ k Hzk = 1 (41)
Rα Rβ
are referred to as Lamé parameters and Akα and Akβ are the first fundamental magnitude
of the first fundamental form dΩk . Attention is herein focused on shells with a constant
curvature, i.e., doubly-curved shells (cylindrical, spherical, toroidal geometries)
for which
k k z
Aα = Aβ = 1. In the case of a swallow shell the approximation 1 + Rα ≈ 1 and
1 + Rzβ ≈ 1 is generally valid and leads to reasonably reliable results. The strain-
displacement relationships referred to an orthogonal curvilinear coordinate system give
rise to the following deformation field [19, 68]:
where
∂u0 w ∂u0 w ∂u0 ∂v0
ε0αα = + , ε0ββ = + , γαβ0
= +
∂α Rα ∂β Rβ ∂β ∂α
0 ∂w0 0 ∂w 0
γαz = + φα , γβz = + φβ
∂α ∂β
0 ∂φα 0 ∂φβ 0 ∂φα ∂φβ
kαα = , kββ = , kαβ = + (43)
∂α ∂β ∂β ∂α
2
∂ 2 w0
2 ∂φα ∂ w0 2 ∂φβ
kαα = c1 + , k ββ = c 1 +
∂α ∂α2 ∂β ∂β 2
∂ 2 w0
2 ∂φβ ∂φα 1
w0 1 w0
kβα = c1 + +2 , kαz = c2 φ α + , kβz = c2 φ β +
∂α ∂β ∂α∂β ∂α ∂β
25
where Lk is the Lagrangian for the kth layer of the composite shell. The first variation
can be expressed as
δLk = δT k − δU k (47)
where δU k is the virtual strain energy, δT k is the virtual kinetic energy, and assume the
following form
Z Z z k+1
k T
δU = δεk σ k dΩk dz;
Ωk zk
(48)
Z Z z k+1
δT k = ρk δ η̇ T η̇ dΩk dz
Ωk zk
where stresses σ k and strains εk are defined in Eq. (45) and η is the displacement
vector given by
T
η= u v w (49)
ρk denotes mass density while an over dot denotes differentiation with respect to time.
The symbol δ represents the variational operator. Imposing the condition in Eq. (46),
the equations of motion are derived
∂ 2 u0 ¯ ∂ 2 φα
2
∂Nα α ∂Nα β ∂ ∂ w0
δu0 : + = I0 2
+ I1 2
− I3 ;
∂α ∂β ∂t ∂t ∂α ∂t2
∂ 2 v 0 ¯ ∂ 2 φβ ∂ ∂ 2 w0
∂Nα β ∂Nβ β
δv0 : + = I0 + I1 − I3 ;
∂α ∂β ∂t2 ∂t2 ∂β ∂t2
∂ Pα α ∂ 2 Pβ β ∂ 2 Pα β
2
∂Qα α ∂Qα β ∂Kα α ∂Kβ β
δw0 : + + c2 + − c1 + +2 −
∂α ∂β ∂α ∂β ∂α2 ∂β 2 ∂α β
2
∂ ∂ 2 v0 h ∂ ∂2φ
Nα α Nβ β ∂ ∂ u0 ¯ α
− = −c1 I3 + − c 1 I4 +
Rα Rβ ∂α ∂t2 ∂β ∂t2 ∂α ∂t2
∂ ∂ 2 φβ i ∂ 2 w0
2 2
∂ 2 ∂ 2 w0
2 ∂ ∂ w0
+ I0 − c 1 I6 + ;
∂β ∂t2 ∂t2 ∂α2 ∂t2 ∂β 2 ∂t2
(50)
26
∂ 2 u0
∂Mα α ∂Mα β ∂Pα α ∂Pα β
δφα : + − Qα α − c2 Kα α + c1 + = I¯1 +
∂α ∂β ∂α ∂β ∂t2
∂ 2 φα ∂ 2 w0
¯ ∂
I2∗ + c I
1 4 ;
∂t2 ∂α ∂t2
∂ 2 v0
∂Mβ α ∂Mβ β ∂Pα β ∂Pβ β
δφβ : + − Qβ β − c2 Kβ β + c1 + = I¯1 +
∂α ∂β ∂α ∂β ∂t2
2
∂ ∂ 2 w0
∗ ∂ φβ ¯
I2 + c 1 I4
∂t2 ∂β ∂t2
The natural boundary conditions are
where
Pn = n2α Pα α + n2β Pβ β + 2 nα nβ Pα β
Pns = (Pβ β − Pα α ) nα nβ + Pα β n2α − n2β
Mn = n2α Mα α + n2β Mβ β + 2 nα nβ Mα β
(52)
Mns = (Mβ β − Mα α ) nα nβ + Mα β n2α − n2β
Qn = Qα α n α + Qβ β n β
Kn = Kα α nα + Kβ β nβ
and c1 = − 3 4h2 , c2 = − h42 . Moreover, nα , nβ are the direction cosines of the unit normal
on the boundary of the laminate. The force and moment resultants appearing in Eqs.
(50) and (51) are defined as follows:
Nl Z z k+1
X
(Ni , Mi , Pi ) = σik 1, z, z 3 dz with i = αα, ββ, βα
k=1 zk
Nl Z z k+1
X
k
(Qαα , Kαα ) = σαz 1, z 2 dz (53)
k
k=1 z
N l Z z k+1
X
k
(Qββ , Kββ ) = σβz 1, z 2 dz
k=1 zk
27
{M } = [B] ε0 + [D] k 0 + [F ] k 2
{P } = [E] ε0 + [F ] k 0 + [H] k 2
(56)
{Q} = [As ] γ 0 + [Ds ] k 1
{K} = [Ds ] γ 0 + [F s ] k 1
XNl Z z k+1
Asij , Dij
s
, Fijs = k
1 z 2 , z 4 dz
C̃ij with i, j = 4, 5
k=1 zk
(57)
For cross-ply stacking sequences an exact solution can be sought in Navier’s or Lèvy’s
k = C̃ k = C̃ k = 0. Moreover, if the stacking sequence is
form. For such laminates C̃16 26 45
symmetric the coupling elastic coefficients Bij and Eij (see Eq. (57)) are zero.
28
(i) Seek a closed form analytical solution of the governing differential equations of
motion of the structural element under consideration in free vibration.
(ii) Apply a number of general boundary conditions that are equal to twice the num-
ber of integration constants; these are usually nodal displacements and forces in
algebraic forms.
(iii) Eliminate the integration constants by relating the amplitudes of the harmonically
varying nodal forces to those of the corresponding displacements which essentially
generates the frequency-dependent dynamic stiffness matrix, providing the force-
displacement relationship of the nodal lines.
29
(60) in the GDEs a set of ordinary differential equations (ODEs) is obtained which can
be written in matrix forms as follows:
L11 L12 L13 L14 L15
Um 0
L21 L22 L23 L24 L25 Vm 0
L31 L32 L33 L34 L35 Wm = 0 (61)
L41 L42 L43 L44 L45 Φα 0
L51 L52 L53 L54 L55 Φβ 0
30
Ai = γi Ci = γi χi Di ; Bi = δi Ci = δi χi Di ; C i = χi D i ; Ei = φi Di
(67)
where
λi A212 α2 Rα + A11 −A22 α2 + I0 ω 2 + A66 λ2i Rα + A22 A66 α2 Rβ + A12 A66 α2 Rα + I0 ω 2 Rβ + A66 λ2i Rβ
γi = −
(A22 α2 − I0 ω 2 ) (A66 α2 − I0 ω 2 ) + A212 − A11 A22 + 2A12 A66 α2 + (A11 + A66 )I0 ω 2 λ2i + A11 A66 λ4i
α −(A12 + A66 )λ2i (A11 Rα + A12 Rβ ) − A66 α2 − I0 ω 2 − A11 λ2i (A12 Rα + A22Rβ )
δi = −
(A12 αλi + A66 αλi )2 − A66 α2 − I0 ω 2 − A11 λ2i −A22α2 + I0 ω 2 + A66 λ2i
(68)
31
χi = − − α2 (D12 + D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 )))2 λ2i + − A55 − c2 (2D55 + c2 F55 ) − α2 (D66 + c1 (2F66
+ c1 H66 )) + (I2 + c1 (2I4 + c1 I6 ))ω 2 + (D11 + c1 (2F11 + c1 H11 ))λ2i A44 + c2 (2D44 + c2 F44 ) + α2 (D22
+ c1 (2F22 + c1 H22 )) − (I2 + c1 (2I4 + c1 I6 ))ω 2 − (D66 + c1 (2F66 + c1 H66 ))λ2i / − λi A44 + c2 (2D44
+ c2 F44 ) + α2 (D22 + c1 (2F22 + c1 H22 )) − (I2 + c1 (2I4 + c1 I6 ))ω 2 − (D66 + c1 (2F66 + c1 H66 ))λ2i A55
+ 2c2 D55 + c22 F55 + c1 α2 (F12 + 2F66 + c1 (H12 + 2H66 )) − (I4 + c1 I6 )ω 2 − (F11 + c1 H11 )λ2i + α2 (D12
+ D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 )))λi A44 + 2c2 D44 + c22 F44 + c1 α2 (F22 + c1 H22 ) − (I4 + c1 I6 )ω 2
− (F12 + 2F66 + c1 (H12 + 2H66 ))λ2i
1
φi = − A55 + 2c2 D55 + α2 D66 + c22 F55 + 2α2 c1 F66
α(D12 + D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 )))λi
+ α2 c21 H66 − I2 ω 2 − 2c1 I4 ω 2 − c21 I6 ω 2 − D11 λ2i − 2c1 F11 λ2i − c21 H11 λ2i − α2 (D12 + D66 + c1 (2F12 + 2F66
+ c1 (H12 + H66 )))2 λ3i − A55 − c2 (2D55 + c2 F55 ) + c1 − α2 (F12 + 2F66 + c1 (H12 + 2H66 )) + (I4 + c1 I6 )ω 2 + (F11
+ c1 H11 )λ2i / − λi A44 + c2 (2D44 + c2 F44 ) + α2 (D22 + c1 (2F22 + c1 H22 )) − (I2 + c1 (2I4 + c1 I6 ))ω 2
− (D66 + c1 (2F66 + c1 H66 ))λ2i A55 + 2c2 D55 + c22 F55 + c1 α2 (F12 + 2F66 + c1 (H12 + 2H66 )) − (I4 + c1 I6 )ω 2
− (F11 + c1 H11 )λ2i + α2 (D12 + D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 )))λi A44 + 2c2 D44 + c22 F44 + c1 α2 (F22
+ c1 H22 ) − (I4 + c1 I6 )ω 2 − (F12 + 2F66 + c1 (H12 + 2H66 ))λ2i − λi A55 + c2 (2D55 + c2 F55 ) + α2 (D66 + c1 (2F66
+ c1 H66 )) − (I2 + c1 (2I4 + c1 I6 ))ω 2 − (D11 + c1 (2F11 + c1 H11 ))λ2i A44 + c2 (2D44 + c2 F44 ) + α2 (D22 + c1 (2F22
+ c1 H22 )) − (I2 + c1 (2I4 + c1 I6 ))ω 2 − (D66 + c1 (2F66 + c1 H66 ))λ2i − A55 − c2 (2D55 + c2 F55 ) + c1 − α2 (F12
+ 2F66 + c1 (H12 + 2H66 )) + (I4 + c1 I6 )ω 2 + (F11 + c1 H11 )λ2i / − λi A44 + c2 (2D44 + c2 F44 ) + α2 (D22
+ c1 (2F22 + c1 H22 )) − (I2 + c1 (2I4 + c1 I6 ))ω 2 − (D66 + c1 (2F66 + c1 H66 ))λ2i A55 + 2c2 D55 + c22 F55 + c1 α2 (F12
+ 2F66 + c1 (H12 + 2H66 )) − (I4 + c1 I6 )ω 2 − (F11 + c1 H11 )λ2i + α2 (D12 + D66 + c1 (2F12 + 2F66 + c1 (H12
+ H66 )))λi A44 + 2c2 D44 + c22 F44 + c1 α2 (F22 + c1 H22 ) − (I4 + c1 I6 )ω 2 − (F12 + 2F66 + c1 (H12 + 2H66 ))λ2i
with i = 1, · · · , 12. When Eqs. (67) are substituted into Eqs. (66) a solution in terms
of only twelve integration constants is obtained. Thus
12
X 12
X
Um (α) = γ i Ci eλi α = γ i χi D i e λ i α
i=1 i=1
12
X 12
X
Vm (α) = δi Ci eλi α = δ i χi D i e λ i α
i=1 i=1
X12 12
X 12
X
Wm (α) = χi D i e λ i α , Φα (α) = Di eλi α , Φβ (α) = φi Di eλi α
i=1 i=1 i=1
(69)
The expressions for the generalized forces can then be found in the same way by sub-
stituting Eqs. (69) into Eqs. (51) written in terms of displacements. Thus, retaining
32
the terms according to the symmetric cross-ply composite shallow shells the following
expressions are derived:
∞ h
X 12
X 12
X
Nαα (α, β) = A11 λi γi χi Di eλi α − A12 θ̂ δ i χi D i e λ i α
m=1 i=1 i=1
12 12
A12 X A11 X i
+ λi χi D i e λ i α + λi χi Di eλi α sin(θ̂ β) = Nαα sin(θ̂ β)
Rβ Rα
i=1 i=1
12 12
∞
" #
X X X
Nββ (α, β) = A66 θ̂ γi χi Di eλi α + A66 λi δi χi Di eλi α cos(θ̂ β)
m=1 i=1 i=1
= Nββ cos(θ̂ β)
∞ h
X 12
X 12
X 12
X
Qα (α, β) = H11 c22 λ2i Di eλi α + H11 c22 λ3i χi Di eλi α + F11 c2 λ2i Di eλi α
m=1 i=1 i=1 i=1
12
X 12
X 12
X
− F12 c2 λi φi Di eλi α − H12 c22 λi φi Di eλi α − H12 c22 λi χi Di eλi α
i=1 i=1 i=1
X12 12
X 12
X
+ 4 H66 c22 λi χi Di eλi α − 2 F66 c2 Di eλi α − 2 F66 c 2 λi φ i D i e λ i α
i=1 i=1 i=1
12
X 12
X 12
X
− A55 Di eλi α − A55 λi χi Di eλi α − 2 D55 c 2 Di eλi α
i=1 i=1 i=1
X12 12
X 12
X i
− 2 D55 c2 λi χi Di eλi α − F55 c22 Di eλi α − F55 c22 λi χi Di eλi α
i=1 i=1 i=1
× sin(θ̂ β) = Qα sin(θ̂ β)
∞ h
X 12
X 12
X 12
X
Mαα (α, β) = D11 λi Di eλi α + H11 c22 λi Di eλi α + H11 c22 λ2i χi Di eλi α
m=1 i=1 i=1 i=1
12
X 12
X 12
X
+ 2 F11 c2 λi Di eλi α + 2 F11 c2 λ2i χi Di eλi α − D12 φi Di eλi α
i=1 i=1 i=1
X12 12
X 12
X
− 2 F12 c2 φi Di eλi α − F12 c2 χi Di eλi α − H12 c22 φi Di eλi α
i=1 i=1 i=1
12
X i
− H12 c22 χi Di eλi α sin(θ̂ β) = Mαα sin(θ̂ β)
i=1
33
∞ h
X 12
X 12
X 12
X
Mββ (α, β) = D12 λi φi Di eλi α + D66 Di eλi α + D66 φi Di eλi α
m=1 i=1 i=1 i=1
12
X 12
X 12
X
+ D12 λi φi Di eλi α + 2 F66 c2 Di eλi α + 2 F66 c 2 φi Di eλi α
i=1 i=1 i=1
X12 12
X X12
+ 2 F66 c2 λi χi Di eλi α + H66 c22 Di eλi α + H66 c22 φi Di eλi α
i=1 i=1 i=1
X12 i
+ 2 H66 c22 λi χi Di eλi α cos(θ̂ β) = Mββ cos(θ̂ β)
i=1
(70)
∞ h
X 12
X 12
X 12
X
Pαα (α, β) = H11 c22 λi Di eλi α + H11 c22 λ2i χi Di eλi α F11 c 2 λi D i e λ i α
m=1 i=1 i=1 i=1
12
X 12
X 12
X i
− F12 c2 φi Di eλi α − H12 c22 φi Di eλi α − H12 c22 χi Di eλi α
i=1 i=1 i=1
× sin(θ̂ β) = Pαα sin(θ̂ β)
Zero boundary conditions are then generally imposed to eliminate the constants in the
classical method in order to establish the frequency equation for a single plate element.
By contrast, the development of the dynamic stiffness matrix entails imposition of gen-
eral boundary conditions in algebraic form which has much wider implications. Thus
in order to develop the dynamic stiffness matrix the following boundary conditions are
applied:
Generalized displacements
α=0 : U m = U m1 ; Vm = Vm 1 ; W m = W m1 ; Φαm = Φα1 ; Φβm = Φβ1 ;
Wm,α = Wm1, α
(71)
α=L : U m = U m2 ; Vm = Vm 2 ; W m = W m2 ; Φαm = Φα2 ; Φβm = Φβ2 ;
Wm,α = Wm2, α
Generalized forces
α=0 : Nαα = −Nαα1 ; Nββ = −Nββ1 ; Qα = −Qα1 ; Mαα = −Mαα1 ;
Mαβ = −Mαβ1 ; Pαα = −Pαα1 ;
(72)
α=L : Nαα = Nαα2 ; Nββ = Nββ2 ; Q α = Q α2 , Mαα = Mαα2 ;
Mαβ = Mαβ2 ; Pαα = Pαα2
34
δ = AD (73)
where
T
δ = U m1 Vm 1 W m1 Φα1 Φβ1 Wm1, α Um2 Vm 2 W m2 Φα2 Φβ2 Wm2, α
T
D= D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12
(74)
and
A11 A21 A31 A41 A51 A61 A71 A81 A91 A10
1 A11
1 A12
1
A12 A22 A32 A42 A52 A62 A72 A82 A92 A10
2 A11
2 A12
2
A13 A23 A33 A43 A53 A63 A73 A83 A93 A10
3 A11
3 A12
3
A14 A24 A34 A44 A54 A64 A74 A84 A94 A10
4 A11
4 A12
4
A15 A25 A35 A45 A55 A65 A75 A85 A95 A10
5 A11
5 A12
5
A16 A26 A36 A46 A56 A66 A76 A86 A96 A10 A11 A12
A=
6 6 6
(75)
A17 A27 A37 A47 A57 A67 A77 A87 A97 A10 A11 A12
7 7 7
A18 A28 A38 A48 A58 A68 A78 A88 A98 A10 A11 A12
8 8 8
A19 A29 A39 A49 A59 A69 A79 A89 A99 A10 A11 A12
9 9 9
A110 A210 A310 A410 A510 A610 A710 A810 A910 A10 A11 A12
10 10 10
A111 A211 A311 A411 A511 A611 A711 A811 A911 A10 A11 A12
11 11 11
A112 A212 A312 A412 A512 A612 A712 A812 A912 A10
12 A11
12 A12
12
where:
Ai1 = (−1)i+1 χi γi , Ai2 = (−1)i+1 χi δi , Ai3 = (−1)i+1 χi
Ai4 = 1, Ai5 = (−1)i+1 φi , Ai6 = χi λi ,
Ai7 = (−1)i+1 χi γi e(a λi ) , Ai8 = (−1)i+1 χi δi e(a λi ) , Ai9 = (−1)i+1 χi e(a λi )
Ai10 = e(a λi ) , Ai11 = (−1)i+1 φi e(a λi ) , Ai12 = χi λi e(a λi )
(76)
with i = 1, · · · , 12. By applying the same procedure for boundary conditions of the
generalized forces and thus exploiting Eq. (72), the following relationship is obtained:
F = RD (77)
where
T
F = Nαα1 Nββ1 Q α1 Mαα1 Mαβ1 Pαα1 Nαα2 Nββ2 Q α2 Mαα2 Mαβ2 Pαα2
T
D= D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12
(78)
35
and
R11 R21 R31 R41 R51 R61 R71 R81 R91 R10
1 R11
1 R12
1
R12 R22 R32 R42 R52 R62 R72 R82 R92 R10
2 R11
2 R12
2
R13 R23 R33 R43 R53 R63 R73 R83 R93 R10
3 R11
3 R12
3
R14 R24 R34 R44 R54 R64 R74 R84 R94 R10
4 R11
4 R12
4
R15 R25 R35 R45 R55 R65 R75 R85 R95 R10
5 R11
5 R12
5
R16 R26 R36 R46 R56 R66 R76 R86 R96 R10 R11 R12
R= 6 6 6
(79)
R17 R27 R37 R47 R57 R67 R77 R87 R97 R10
7 R11
7 R12
7
R18 R28 R38 R48 R58 R68 R78 R88 R98 R10 R11 R12
8 8 8
R19 R29 R39 R49 R59 R69 R79 R89 R99 R10 R11 R12
9 9 9
R110 R210 R310 R410 R510 R610 R710 R810 R910 R10 R11 R12
10 10 10
R111 R211 R311 R411 R511 R611 R711 R811 R911 R10 R11 R12
11 11 11
R112 R212 R312 R412 R512 R612 R712 R812 R912 R10
12 R11
12 R12
12
where
1 1
Ri1 = (−1)i+1 A12 α χi δi − A12 χi γi λi − A12 χi − A11 χi ;
Rβ Rα
Ri2 = (−1)i+1 − A66 α χi γi − A66 χi δi λi ;
Ri3 = −A55 − c2 (2 D55 + c2 F55 ) − 2 α2 c1 F66 − 2 α2 c21 H66 − χi (A55 + c2
(2 D55 + c2 F55 )) λi − α2 c21 χi H12 λi − 4 α2 c21 χi H66 λi − α c1 F12 φi λi
− 2 α c1 F66 φi λi − α c21 H12 φi λi − 2 α c21 H66 φi λi + c1 F11 λ2i + c21 H11 λ2i
+ c21 χi H11 λ3i ;
Ri4 = (−1)i+1 α2 c1 χi F12 + α2 c21 χi H12 + α D12 φi + 2 α c1 F12 φi + α c21 H12 φi
− D11 λi − 2 c1 F11 λi − c21 H11 λi − c1 χi (F11 + c1 H11 ) λ2i ;
Ri5 = −α D66 − 2 α c1 F66 − α c21 H66 − 2 α c1 χi F66 λi − 2 α c21 χi H66 λi − (D66
+ c1 (2 F66 + c1 H66 )) φi λi ;
Ri6 = −c1 (−α2 c1 χi H12 − α F12 φi − α c1 H12 φi + F11 λi + c1 H11 λi + c1 χi H11 λ2i );
1
Ri7 = (−1)i+1 − A12 α χi δi e(a λi ) + A11 χi e(a λi ) γi λi + A12 χi e(a λi )
Rβ
1
+ A11 χi e(a λi ) ;
Rα
Ri8 = (−1)i+1 A66 α χi e(a λi ) γi + A66 χi δi e(a λi ) λi ;
Ri9 = e(a λi ) (A55 + c2 (2 D55 + c2 F55 )) + 2 α2 c1 e(a λi ) F66 + 2 α2 c21 e(a λi ) H66
+ χi e(a λi ) (A55 + c2 (2 D55 + c2 F55 )) λi + α2 c21 χi e(a λi ) H12 λi + 4 α2 c21 χi e(a λi )
× H66 λi + α c1 e(a λi ) F12 φi λi + 2 α c1 e L λi F66 φi λi + α c21 e(a λi ) H12 φi λi
+ 2 α c21 e(a λi ) H66 φi λi − c1 e(a λi ) F11 λ2i − c21 e(a λi ) H11 λ2i − c21 χi e(a λi ) H11 λ3i ;
36
Ri10 = (−1)i+1 − α2 c1 χi e(a λi ) F12 − α2 c21 χi e(a λi ) H12 − α D12 e(a λi ) φi
− 2 α c1 e(a λi ) F12 φi − α c21 e(a λi ) H12 φi + D11 e(a λi ) + 2 c1 e(a λi ) F11 λi
+ c21 e(a λi ) H11 λi + c1 χi e(a λi ) (F11 + c1 H11 ) λ2i ;
Ri11 = α D66 e(a λi ) + 2 α c1 e(a λi ) F66 + α c21 e(a λi ) H66 + 2 α c1 χi e(a λi ) F66 λi (80)
+ 2 α c21 χi e(a λi ) H66 λi + e(a λi ) (D66 + c1 (2 F66 + c1 H66 )) φi λi ;
Ri12 = (−1)i+1 c1 (−α2 c1 χi e(a λi ) H12 − α e(a λi ) F12 φi − α c1 e(a λi ) H12 φi
+ e(a λi ) F11 λi + c1 e(a λi ) H11 λi + c1 χi e(L λi ) H11 λ2i )
with i = 1, · · · , 12. Now by eliminating the constants vectors D in the Eqs. (73) and
(77) the dynamic stiffness matrix which links the forces and moments vector F with the
generalized displacements vector δ is derived:
F = K δ, K = R A−1 (81)
i.e.
K11 K12 K13 K14 K15 K16 K17 K18 K19 K110 K111 K112
K21 K22 K23 K24 K25 K26 K27 K28 K29 K210 K211 K212
K31 K32 K33 K34 K35 K36 K37 K38 K39 K310 K311 K312
K41 K42 K43 K44 K45 K46 K47 K48 K49 K410 K411 K412
K51 K52 K53 K54 K55 K56 K57 K58 K59 K510 K511 K512
K61 K62 K63 K64 K65 K66 K67 K68 K69 K610 K611 K612
K= (82)
K71 K72 K73 K74 K75 K76 K77 K78 K79 K710 K711 K712
K81 K82 K83 K84 K85 K86 K87 K88 K89 K810 K811 K812
K91 K92 K93 K94 K95 K96 K97 K98 K99 K910 K911 K912
1
K10 2
K10 3
K10 4
K10 5
K10 6
K10 7
K10 8
K10 9
K10 10
K10 11
K10 12
K10
1 2 3 4 5 6 7 8 9 10 11 12
K11 K11 K11 K11 K11 K11 K11 K11 K11 K11 K11 K11
1
K12 2
K12 3
K12 4
K12 5
K12 6
K12 7
K12 8
K12 9
K12 10
K12 11
K12 12
K12
The above dynamic stiffness matrix will now be used in conjunction with the Wittrick-
Williams algorithm [85] to analyze assemblies of laminated composite cylindrical and
spherical shallow shells to investigate their free vibration characteristics based on HSDT.
Explicit expressions for each shell element of the DS matrix were obtained via symbolic
computation using Mathematicar . They are far too extensive and voluminous to report
in this paper. The correctness of these expressions was further checked by implement-
ing them in a MATLABr program and then carrying out a wide range of numerical
simulations.
37
a
b
Rb Ra
z
Global dynamic
stiffness matrix
1
a
N b
like the FEM, a mesh is required in the DSM, but it should be noted that the latter is
mesh independent in the sense that additional elements are required only when there is
a change in the geometry of the structure. A single DS laminate element is enough to
compute any number of its natural frequencies to any desired accuracy, which, of course,
is impossible in the FEM. However, for the type of structures under consideration DS
shell elements do not have point nodes, but have line nodes instead. In this particular
case, no change in geometry along the longitudinal direction is admitted. This is in
addition to the assumed simple support boundary conditions on two opposite sides,
inherent in DSM for shell elements at present. The other two sides of the shell can
have any boundary conditions. The application of the boundary conditions of the global
dynamic stiffness matrix involves the use of the so-called penalty method. This consists
of adding a large stiffness to the appropriate leading diagonal term which corresponds
to the degree of freedom of the node that needs to be suppressed. It is thus possible to
apply free (F), simple support (S) and clamped (C) boundary conditions on the structure
by penalizing the appropriate degrees of freedom. Clearly for simple support boundary
38
condition, the generalized displacement amplitudes Vi , Wi and Φyi are assigned zero
values. On the other hand, for clamped boundary condition Ui , Vi , Wi , Φxi , Φyi and
Wxi on the boundary are assigned zero values. Of course for the free-edge boundary
condition stress resultants are assigned zero values and then no penalty will be applied
at the generalized displacement amplitudes. Because of the similarities between DSM
and FEM, DS elements can be implemented in FEM codes to enhance the accuracy of
results in FEM very considerably.
q
a2 ρ
Table 4: Fundamental circular frequency parameter ω̂ = ω h E2 , of square cylindrical
and spherical shells with staking sequence [0◦ /90◦ /0◦ ] length-to-thickness ratio a/h = 10
and radius-to-length ratio Rβ /a = 20.
Cylindrical shell
Khdeir and Reddy [87] 11.793 13.825 3.789 15.999 4.322 6.089
Present HSDT DSM 11.793 13.825 3.789 15.999 4.322 6.089
results perfectly match each other for all the considered boundary conditions. In Tables
5 and 6 further assessments have been carried out. Most notably, the fundamental circu-
lar frequency parameter has been computed for three and four-layer symmetric cross-ply
and moderately thick spherical and cylindrical shallow shells, respectively. Results are
compare with several meshless formulations proposed by Ferreira el. all [68, 88, 89], in
particular a radial basis formulation (RBF) based on a HSDT, a wavelet collocation
(WLC) formulation based on a FSDT and a radial basis collocation (RBFC) solution
39
q
a2 ρ
Table 5: Dimensionless fundamental circular frequency parameter ω̂ = ω h E2 , of
square spherical and cylindrical shells with staking sequence [0◦ /90◦ /0◦ ],
length-to-
thickness ratio a/h = 10 and varying the radius-to-length ratio Rβ /a and R/a.
Rβ /a
5 10 20 50 100 Plate
Cylindrical
HSDT RBF [68] 11.851 11.808 11.797 11.794 11.793 11.793
FSDT WLC [88] 12.214 12.176 12.166 12.163 12.163 12.163
SSDT RBFC [89] 11.923 11.915 11.913 11.912 11.912 11.912
HSDT DSM 11.846 11.804 11.793 11.790 11.790 11.790
q
a2 ρ
Table 6: Dimensionless fundamental circular frequency parameter ω̂ = ω h E2 , of
square spherical and cylindrical shells with staking sequence [0◦ /90◦ /90◦ /0◦ ], length-to-
thickness ratio a/h = 10 and varying the radius-to-length ratio Rβ /a and R/a.
Rβ /a
5 10 20 50 100 Plate
Cylindrical
HSDT RBF [68] 11.838 11.794 11.783 11.780 11.779 11.779
FSDT WLC [88] 12.279 12.240 12.230 12.228 12.227 12.227
SSDT RBFC [89] 11.901 11.887 11.884 11.883 11.883 11.883
HSDT DSM 11.832 11.790 11.780 11.777 11.777 11.776
40
q
a2 ρ
Table 7: First five circular frequency parameters ω̂ = ω h E2 , of square cylindrical
shells with staking sequence [0◦ /90◦ /0◦ ] length-to-thickness ratio a/h = 10 and varying
the radius-to-length ratio Rβ /a.
first six natural frequencies are computed and compared with the 3D elasticity solution
and many other theories, which are refined in the displacement field or in the curvature
description. As can be seen from the Table the proposed DSM formulation provide the
best accuracy for the fundamental natural frequency with respect to the 3D elasticity
solution, but for higher frequencies there is an increase in the error percentage. This
slight loss of accuracy is due to the fact that approximated curvature descriptions have
been employed. In Table 12 the first three circular frequency parameters of three-layer
symmetric cross-ply shallow spherical shells are calculated. The boundary conditions
SSSS, SCSS and SCSC are taken into account. The investigation is carried out for dif-
ferent values of the radius-to-length ratio and for moderately thick and thin spherical
panels. In Table 13 the effect of the orthotropic ratio on the dimensionless fundamen-
tal frequency parameter is studied. The boundary conditions SSSS, SCSS and SCSC
41
Figure 7: First six mode shapes of a symmetric cross-ply cylindrical shell with SCSC boundary
condition.
42
Figure 8: First six mode shapes of a symmetric cross-ply cylindrical shell with SFSC boundary
condition.
43
Figure 9: First six mode shapes of a symmetric cross-ply cylindrical shell with SFSF boundary
condition.
44
q
a2 ρ
Table 8: First five circular frequency parameters ω̂ = ω h E2 , of square cylindrical
shells with staking sequence [0◦ /90◦ /0◦ ] length-to-thickness ratio a/h = 100 and varying
the radius-to-length ratio Rβ /a.
q
a2 ρ
Table 9: First five circular frequency parameters ω̂ = ω h E2 , of square cylindrical
shells with staking sequence [0◦ /90◦ /90◦ /0◦ ] length-to-thickness ratio a/h = 10 and
varying the radius-to-length ratio Rβ /a.
45
q
a2 ρ
Table 10: First five circular frequency parameters ω̂ = ω h E2 , of square cylindrical
shells with staking sequence [0◦ /90◦ /90◦ /0◦ ] length-to-thickness ratio a/h = 100 and
varying the radius-to-length ratio Rβ /a.
are accounted for, a symmetric cross-ply [0◦ /90◦ /90◦ /0◦ ] square spherical shallow shell
(R/a = 50) is analyzed. As expected the fundamental frequency parameter increases
when increasing both the orthotropic ratio and the length-to-thickness ratio, for all the
considered boundary conditions. Finally, in Fig. 10 the first six mode shapes of SCSC
square spherical shallow shells are represented. The geometrical characteristics and the
lamination scheme are equal to the ones used for the cylindrical shallow shells shown in
Figs. 7, 8 and 9.
7 Conclusions
A detailed literature reviews on the dynamic stiffness method for both buckling of lam-
inated composite plates and free vibration of doubly curved shallow shells have been
presented. In particular, in the first part of the report the stability equations have been
obtained using the principle of minimum potential energy and the dynamic stiffness
matrix has been derived for laminated composite plate elements based on the HSDT.
The element stiffnesses have been implemented in a computer program and results for
composite plate assemblies have been obtained and validated.
In the second part, an exact free vibration analysis of laminated composite shallow
shells has been carried out by combining for the first time the dynamic stiffness method
and a higher order shear deformation theory. The effect of several parameters such as
length-to-thickness ratio and radius-to-length ratio, orthotropic ratio, stacking sequence
and number of layers on the dimensionless circular frequency parameters has been in-
vestigated in details. Results have been compared with those available in the literature
46
CST [91] 0.53263 (−1.370)† 0.59041 (−1.063) 0.59080 (−1.014) 0.68486 (−1.197) 0.76020 (−1.065) 0.76260 (−1.383)
47
FSTD [92] 0.50211 (+4.438) 0.56247 (+3.720) 0.56248 (+3.828) 0.65706 (+2.911) 0.73915 (+1.734) 0.74035 (+1.575)
HSTD [92] 0.50223 (+4.415) 0.56276 (+3.670) 0.56277 (+3.779) 0.65788 (+2.790) 0.73966 (+1.666) 0.74081 (+1.514)
FSTD (DT‡ ) [93] 0.52864 (−0.611) 0.58954 (−0.914) 0.58954 (−0.798) 0.68370 (−1.025) 0.75974 (−1.004) 0.75974 (−1.002)
FSTD (ST∗ ) [93] 0.52830 (−0.546) 0.58853 (−0.741) 0.58853 (−0.626) 0.68232 (−0.822) 0.75818 (−0.796) 0.75818 (−0.795)
HSDT DSM 0.52795 (−0.480) 0.58899 (−0.820) 0.58982 (−0.846) 0.68562 (−1.309) 0.75989 (−1.023) 0.76032 (−1.080)
Material and geometric properties: a = b = 1.0118, h = 0.0191, R = 1.91, E = 1, ρ = 1, ν = 0.3
ω3D −ω
† Error% = ω3D
× 100; ‡ Donnell approximation; ∗ Sanders approximation
q
a2 ρ
Table 12: First three circular frequency parameters ω̂ = ω h E2 , of square spherical
shells with staking sequence [0◦ /90◦ /0◦ ] and varying the radius-to-length R/a and the
length-to-thickness a/h ratios.
q
a2 ρ
Table 13: Fundamental circular frequency parameter ω̂ = ω h E2 , of square spherical
shells with staking sequence [0◦ /90◦ /90◦ /0◦ ], R/a = 50 and varying orthotropic E1 /E2
and the length-to-thickness a/h ratios.
48
Figure 10: First six mode shapes of a symmetric cross-ply spherical shell with SCSC boundary
condition.
49
50
References
[1] S. P. Timoshenko. Theory of elastic stability. McGraw Hill, New York, 1961.
[2] A.W. Leissa. Condition for laminated plates to remain flat under inplane loading.
Composite Structures, 6:261–270, 1986.
[3] D. Bushnell. Computerized buckling analysis of shells. M. Nihhoff, Dordrecht, The
Netherlands, 1985.
[4] E. Riks. Buckling, Encyclopedia of Computational Mechanics,edited by E. Stein,
R. de Borst and T.J.R. Hughes. Vol.2 Wiley, New York, 2004.
[5] F. A. Fazzolari, M. Boscolo, and J. R. Banerjee. An exact dynamic stiffness ele-
ment using a higher order shear deformation theory for free vibration analysis of
composite plate assemblies. Composite Structures, 96:262–278, 2013.
[6] W. H. Wittrick. A Unified Approach to the Initial Buckling of Stiffened Panels in
Compression. Aeronautical Quarterly, 19:265–283, 1968.
[7] W. H. Wittrick and Curzon P. L. V. Stability Functions for the Local Buckling of
Thin Flat-Walled Structures with the Walls in Combined Shear and Compression.
Aeronautical Quarterly, 19:327–351, 1968.
[8] W. H. Wittrick. General Sinusoidal Stiffness Matrices for Buckling and Vibration
Analysis of Thin Flat-Walled Structures. International Journal of Mechanical Sci-
ences, 10:949–966, 1968.
[9] C. S. Smith. Bending, Bucking and Vibration of Orthotropic Plate-Beam Structures.
Journal of Ship Research, 12:249–268, 1968.
[10] F. W. Williams. Computation of Natural Frequencies and Initial Buckling Stresses
of Prismatic Plate Assemblies. Journal of Sound and Vibration, 21:87–106, 1972.
[11] W.H. Wittrick and F.W. Williams. Buckling and vibration of anisotropic or isotropic
plate assemblies under combined loadings. International Journal of Mechanical
Sciences, 16(4):209–239, 1974.
[12] R.J. Plank and W. H. Wittrick. Buckling Under Combined Loadings of Thin Flat-
Walled Structures By a Complex Finite-Strip Method. International Journal of
Numerical Methods in Engineering, 8:323–339, 1974.
[13] F. W. Williams and W. P. Howson. Compact computation of natural frequencies
and buckling loads for plane frames. International Journal of Numerical Methods
in Engineering, 11:1067–1081, 1977.
[14] A. V. Viswanathan and M. Tamekuni. Elastic Buckling Analysis for Composite
Stiffened Panels and other Structures Subjected to Biaxial In-plane Loads. NASA
Technical Report CR-2216, 1973.
51
52
[29] E. Carrera. Multilayered shell theories accounting for layerwise mixed description,
part 2: Numerical evaluations. AIAA Journal, 37(9):1117–1124, 1999.
[30] F. A. Fazzolari and E. Carrera. Coupled thermoelastic effect in free vibration analy-
sis of anisotropic multilayered plates by using an advanced variable-kinematics Ritz
formulation. European Journal of Mechanics Solid/A,, Accepted for publication,
2012.
[31] L. H. Donnell. Stability of thin walled tubes under torsion. Technical Report 479,
NACA, 1933.
[32] L. H. Donnell. A discussion of thin shell theory. In Proceedings of the Fifth Inter-
national Congress for Applied Mechanics, 1938.
[34] K. M. Mushtari. Certain generalizations of the theory of thin shells, (in Russian).
Izv. Fiz. Mat. Odd. pri Kazan. Univ., Vol. 11 (8), 1938.
[37] W. Flügge. Statik und Dynamik der Schalen. Julius Springer, Berlin (Reprinted by
Edwards Brothers Inc., Ann Arbor, Mich., 1943), 1934.
[39] A. I. Lur’e. General theory of elastic shells. Prikl. Mat. Mekh., 4(1):7–34, 1940.
[40] R. Byrne. Theory of small deformations of a thin elastic shell. Seminar Reports in
Math., Univ. of Calif. Pub. in Math., N. S., 2(1):103–152, 1944.
53
[46] S. A. Ambartsumian. Some main equations of the theory of thin layered shell. DAN
ArmSSR, 8(5):–, 1948.
[57] C. W. Bert. Dynamic of composite and sandwich panels - part I. Shock & Vibration
Digest, 8(10):37–48, 1976.
[58] C. W. Bert. Dynamic of composite and sandwich panels - part II. Shock & Vibration
Digest, 8(11):15–24, 1976.
[59] J.N. Reddy. A simple higher order theory for laminated plates. J. Appl. Mech.,
51:745752, 1984.
[61] Y. K. Cheung. The Finite Strip Method in Structural Analysis. Pergamon Press,
Oxford, England, 1st edition, 1976.
54
[65] S. Di and E. Ramm. Hybrid stress formulation for higher-order theory of laminated
shell analysis. Computer Methods in Applied Mechanics and Engineering, 109:356–
359, 1993.
[66] M. S. Qatu and E. Asadi. Vibration of doubly curved shallow shells with arbitrary
boundaries. Applied Acoustics, 73:21–27, 2012.
[67] E Asadi, W. Wenchao, and M. S. Qatu. Static and vibration analyses of thick
deep laminated cylindrical shells using 3d and various shear deformation theories.
Composite Structures, 94:494–500, 2012.
[70] F. Tornabene. 2-D GDQ solution for free vibrations of anisotropic doubly-curved
shells and panels of revolution. Composite Structures, 93:1854–1876, 2011.
[72] F. A. Fazzolari and E. Carrera. Advanced variable kinematics Ritz and Galerkin
formulation for accurate buckling and vibration analysis of laminated composite
plates. Composite Structures, 94(1):50–67, 2011.
[74] F. A. Fazzolari and E. Carrera. Advances in the Ritz formulation for free vibration
response of doubly-curved anisotropic laminated composite shallow and deep shells.
Composite Structures, 101:111–128, 2013.
55
[76] F. A. Fazzolari and E. Carrera. Free vibration analysis of sandwich plates with
anisotropic face sheets in thermal environment by using the hierarchical trigono-
metric Ritz formulation. Composites Part B: Engineering, 50:67–81, 2013.
[80] A.Y.T. Leung. Dynamic stiffness analysis of laminated composite plates. Thin-
Walled Structures, 25:109–133, 1996.
[82] R. S. Langley. A dynamic stiffness technique for the vibration analysis of stiffened
shell structures. Journal of Sound and Vibration, 156(3):521–540, 1992.
[84] D. Tounsi, J. B. Casimir, and M. Haddar. Dynamic stiffness formulation for circular
rings. Computers & Structures, 112113:258–265, 2012.
[85] W. H. Wittrick and F. W. Williams. A general algorithm for computing natural fre-
quencies of elastic structures. Quarterly Journal of mechanics and applied sciences,
24(3):263–284, 1970.
[86] E. Carrera. On the use of transverse shear stress homogeneous and non-
homogeneous conditions in third-order orthotropic plate theory. Composite Struc-
tures, 77:341–352, 2007.
[87] J.N. Reddy and D. Phan. Stability and vibration of isotropic , orthotropic and
laminate plates according to a higher-order shear deformation theory. Journal of
Sound and Vibration, 98(2):157170, 1985.
56
[91] S. C. Fan and M. H. Luah. Free vibration analysis of arbitrary thin shell structures
by using spline finite element. Journal of Sound and Vibration, 179(5):763–776,
1995.
[92] R. K. Khare, T. Kant, and A. K. Garg. Free vibration of composite and sandwich
laminates with a higher-order facet shell element. Composite Structures, 65(3–
4):405–418, 2004.
57
where Dx and Dy are the derivatives in x and y respectively and c1 = − 3 4h2 . The
constitutive equations in the lamina reference system can be written, in terms of reduced
stiffness coefficients, as:
"
" σ1 # C̃11 C̃12 0 0 0 ε1
#
σ2 C̃12 C̃22 0 0 0 ε2
τ12 = 0 0 C̃66 0 0 γ12 (A.3)
τ23 0 0 0 C̃44 0 γ23
τ13 0 0 0 0 C̃55 γ13
where the C̃ij are expressed in terms of stiffness coefficients Cij , as:
2 2
C13 E1 C13 C23 ν12 E2 C23 E2
C̃11 = C11 − = , C̃12 = C12 − = , C̃22 = C22 − =
C33 1 − ν12 ν21 C33 1 − ν12 ν21 C33 1 − ν12 ν21
C̃44 = C44 = G23 , C̃55 = C55 = G13 C̃66 = C66 = G12
(A.4)
where E1 is the elastic modulus in the fibre direction, E2 the elastic modulus in per-
pendicular to the fibre, ν12 and ν21 = ν12 E2 /E1 the Poisson’s ratios, G12 = G13 and
G23 the shear modulus of each single orthotropic lamina. If the lamina is placed at an
angle θ in the laminate or global reference system, the equation need to be transformed
as follows:
C 11 =C̃11 C 4 + 2(C̃12 + 2C̃66 )S 2 C 2 + C̃22 S 4
C 12 =(C̃11 + C̃22 − 4C̃66 )S 2 C 2 + C̃12 (S 4 + C 4 )
C 16 =(C̃11 − C̃12 − 2C̃66 )SC 3 + (C̃12 − C̃22 + 2C̃66 )S 3
C 22 =C̃11 S 4 + 2(C̃12 + 2C̃66 )S 2 C 2 + C̃22 C 4
C 26 =(C̃11 − C̃12 − 2C̃66 )S 3 C + (C̃12 − C̃22 + 2C̃66 )SC 3 (A.5)
2 2 4 4
C 66 =(C̃11 + C̃22 − 2C̃12 − 2C̃66 )S C + C̃66 (S + C )
C 44 =C̃44 C 2 + C̃55 S 2
C 55 =C̃44 S 2 + C̃55 C 2
C 45 =(C̃55 − C̃44 )CS
58
where C = cos (θ) and S = sin (θ). This leads to the constitutive equation for the k-th
lamina in the laminate or global reference system:
C 11 C 12 C 16 0 0
σxx εxx
σyy
C 12 C 22 C 26 0 0
εyy
τxy =
C 16 C 26 C 66 0 0
γxy
(A.6)
τyz 0 0 0 C 44 C 45 γyz
τxz 0 0 0 C 45 C 55 γxz
that in compact form can be written for each k-th lamina as:
k
σ k = C εk (A.7)
59
2 2 2 2 2 2 2 2 2 3 2 2 3 2 2 2
a2 = −c1 A44 + α D22 F11 + 2α c1 D12 F11 F12 − α c1 D11 F12 + 2α c1 F11 F12 − 2α c1 F11 F22 + 4α c1 D12
2 2 2 2 2 2 3 2 2 2 2 2 2 2
F11 F66 − 4α c1 D11 F12 F66 − 4α c1 D11 F66 − 8α c1 F11 F66 + A44 c1 D11 H11 − α c1 D12 H11 + α c1 D11 D22 H11
2 2 2 3 2 3 2 4 2 2 3 2 4
− 2α c1 D12 D66 H11 − 2α c1 D12 F12 H11 − 4α c1 D66 F12 H11 − α c1 F12 H11 + 2α c1 D11 F22 H11 − 4α c1 F12 F66
2 4 2 2 2 2 3 2 3 2 3 2 4
H11 − 4α c1 F66 H11 + 2α c1 D11 D66 H12 + 2α c1 D12 F11 H12 + 4α c1 D66 F11 H12 − 2α c1 D11 F12 H12 + 2α c1 F11
2 4 2 4 2 2 4 2 2 4 2 2 2 3
F12 H12 + 4α c1 F11 F66 H12 − α c1 D11 H12 − α c1 F11 H22 + α c1 D11 H11 H22 + 4α c1 D11 D66 H66 + 4α c1 D12 F11
2 3 2 3 2 4 2 4 2 4
H66 + 8α c1 D66 F11 H66 − 4α c1 D11 F12 H66 + 2α c1 D12 H11 H66 + 4α c1 D66 H11 H66 − 2α c1 D11 H12 H66 + A55 D11
2 2 2
(D66 + c1 (2F66 + c1 H66 )) + c2 D11 D66 F55 + 2c1 D11 F55 F66 + c1 −F11 F44 + D11 F44 H11 + D11 F55 H66 + 2c2
2 2 2 2
−c1 D44 F11 + D11 c1 D44 H11 + D55 D66 + 2c1 F66 + c1 H66 + (D11 + c1 (2F11 + c1 H11 ))(D66 + c1 (2F66 + c1 H66 ))λNx0
2 2 2 2 2 2 2
a3 = 2α c2 D12 D55 − 2α c2 D11 D22 D55 − 4c2 D11 D44 D55 − 2α c2 D11 D44 D66 + 4α c2 D12 D55 D66 − 4α c1 c2 D12 D44 F11
2 2 2 4 2 2
2
− 8α c1 c2 D44 D66 F11 + 4α c1 c2 D11 D44 F12 + 4α c1 c2 D12 D55 F12 + 8α c1 c2 D55 D66 F12 + 2α c1 D22 F11 F12 + 4α
2 4 2 2 2 2 2 4 3 3 2 4 2
c1 c2 D44 F11 F12 − 2α c1 D12 F12 + 2α c1 c2 D55 F12 − 4α c1 F12 − 4α c1 c2 D11 D55 F22 − 2α c1 D12 F11 F22 − 2
4 2 4 2 4 3 3 2 2 2 2
α c1 D66 F11 F22 + 2α c1 D11 F12 F22 + 4α c1 F11 F12 F22 − 2c2 D11 D55 F44 − α c2 D11 D66 F44 − 2α c1 c2 D12 F11
2 2 2 2 2 2 2 2 2 2 2 2 3
F44 − 4α c1 c2 D66 F11 F44 + 2α c1 c2 D11 F12 F44 + 2α c1 c2 F11 F12 F44 + α c2 D12 F55 − α c2 D11 D22 F55 − 2c2
2 2 2 2 2 2 2 2 2 2 2 2
D11 D44 F55 + 2α c2 D12 D66 F55 + 2α c1 c2 D12 F12 F55 + 4α c1 c2 D66 F12 F55 + α c1 c2 F12 F55 − 2α c1 c2 D11
4 2 4 2 2 2 4 2 2 2
F22 F55 − c2 D11 F44 F55 + 4α c1 c2 D11 D44 F66 + 4α c1 D22 F11 F66 + 8α c1 c2 D44 F11 F66 − 8α c1 D12 F12 F66 + 8α c1
4 3 2 4 2 4 3 2 2 2 2 2
c2 D55 F12 F66 − 16α c1 F12 F66 + 4α c1 D11 F22 F66 + 12α c1 F11 F22 F66 + 2α c1 c2 D11 F44 F66 + 4α c1 c2 F11 F44
2 2 2 4 2 2 2 2 2 4 3 2 2 2 2 2 2 2
F66 + 4α c1 c2 F12 F55 F66 − 8α c1 D12 F66 + 8α c1 c2 D55 F66 − 16α c1 F12 F66 + 4α c1 c2 F55 F66 − 4α c1
4 2 2 2 4 3 4 3 4 3
c2 D12 D44 H11 − α c1 D22 D66 H11 − 8α c1 c2 D44 D66 H11 + 2α c1 D22 F12 H11 − 2α c1 D12 F22 H11 − 4α c1 D66 F22
4 4 2 2 2 2 2 2 4 3 4 4
H11 + 2α c1 F12 F22 H11 − 2α c1 c2 D12 F44 H11 − 4α c1 c2 D66 F44 H11 + 2α c1 D22 F66 H11 + 4α c1 F22 F66 H11 + 2
4 2 2 4 2 4 2 4 3 4 3 4 3
α c1 D12 H12 − 2α c1 D11 D22 H12 + 4α c1 D12 D66 H12 − 2α c1 D22 F11 H12 + 4α c1 D12 F12 H12 + 8α c1 D66 F12
4 4 2 4 3 4 4 4 4 4 4 2 4 4
H12 − 2α c1 F12 H12 − 2α c1 D11 F22 H12 − 2α c1 F11 F22 H12 − 8α c1 F12 F66 H12 − 8α c1 F66 H12 + 2α c1 D12
2 4 4 2 2 2 4 2 4 3 4 3 4 3
H12 + 4α c1 D66 H12 − 2α c1 c2 D11 D55 H22 − α c1 D11 D66 H22 − 2α c1 D12 F11 H22 − 4α c1 D66 F11 H22 + 2α c1
4 4 2 2 2 4 3 4 4 4 4
D11 F12 H22 + 2α c1 F11 F12 H22 − α c1 c2 D11 F55 H22 + 2α c1 D11 F66 H22 + 4α c1 F11 F66 H22 − 2α c1 D12 H11 H22
4 4 4 2 2 4 2 2 2 2 2 4 2
− 4α c1 D66 H11 H22 + 4α c1 D12 H66 − 4α c1 D11 D22 H66 − 2α c1 c2 D11 D44 H66 − 4α c1 c2 D12 D55 H66 + 8α c1 D12
2 2 4 3 4 3 4 3 4 3
D66 H66 − 8α c1 c2 D55 D66 H66 − 4α c1 D22 F11 H66 + 8α c1 D12 F12 H66 + 16α c1 D66 F12 H66 − 4α c1 D11 F22 H66
4 4 2 2 2 2 2 2 2 2 2 4 4 4 4
− 2α c1 F11 F22 H66 − α c1 c2 D11 F44 H66 − 2α c1 c2 D12 F55 H66 − 4α c1 c2 D66 F55 H66 − α c1 D22 H11 H66 + 4α c1
4 4 4 4 2 2 2 2
D12 H12 H66 + 8α c1 D66 H12 H66 − α c1 D11 H22 H66 + α D12 λNx0 − α D11 D22 λNx0 − 2c2 D11 D44 λNx0 + 2α D12 D66
2 2 2
λNx0 − 2c2 D55 D66 λNx0 − 2α c1 D22 F11 λNx0 − 4c1 c2 D44 F11 λNx0 + 4α c1 D12 F12 λNx0 + 4α c1 D66 F12 λ
2 2 2 2 2 2 2 2
Nx0 + 4α c1 F12 λNx0 − 2α c1 D11 F22 λNx0 − 4α c1 F11 F22 λNx0 − c2 D11 F44 λNx0 − 2c1 c2 F11 F44 λ
2 2 2 2 2
Nx0 − c2 D66 F55 λNx0 + 4α c1 D12 F66 λNx0 − 4c1 c2 D55 F66 λNx0 + 8α c1 F12 F66 λNx0 − 2c1 c2 F55 F66 λ
2 2 2 2 3 2 2 2 2
Nx0 − α c1 D22 H11 λNx0 − 2c1 c2 D44 H11 λNx0 − 2α c1 F22 H11 λNx0 − c1 c2 F44 H11 λNx0 + 2α c1 D12 H12 λ
2 2 2 3 2 3 2 4 2 2 2
Nx0 + 2α c1 D66 H12 λNx0 + 4α c1 F12 H12 λNx0 + 4α c1 F66 H12 λNx0 + α c1 H12 λNx0 − α c1 D11 H22 λ
2 3 2 4 2 2 2 2 3
Nx0 − 2α c1 F11 H22 λNx0 − α c1 H11 H22 λNx0 + 2α c1 D12 H66 λNx0 − 2c1 c2 D55 H66 λNx0 + 4α c1 F12 H66
2 2 2 4 2 2
λNx0 − c1 c2 F55 H66 λNx0 + 2α c1 H12 H66 λNx0 − A44 A55 D11 + 2c2 D11 D55 + c2 D11 F55 + α (2c1 (F11 (D12 + 2D66
60
− c1 (F12 + 2F66 )) + c1 (D12 + 2D66 )H11 ) + D11 (D66 + c1 (−2(F12 + F66 ) + c1 H66 ))) + (D11 + c1 (2F11 + c1 H11 ))λNx0
2 2
+ A55 − c2 D11 (2D44 + c2 F44 ) + α D12 − D11 (D22 + c1 (2F22 + c1 H22 )) + 2D12 (D66 + c1 (F12 − c1 H66 ))
2 2
+ c1 c1 (F12 + 2F66 ) + 4D66 (F12 − c1 H66 ) − (D66 + c1 (2F66 + c1 H66 ))λNx0 − α (D11 + c1 (2F11 + c1 H11 ))
2 4 2 4 2 2 4 2
a4 = 4A55 α c2 D12 D44 − 2α c2 D12 D44 + 2α c2 D11 D22 D44 + 8α c2 D12 D44 D55 + A55 α D22 D66 + 8A55 α c2 D44 D66
4 4 2 2 4 4 4
− 4α c2 D12 D44 D66 + 2α c2 D22 D55 D66 + 16α c2 D44 D55 D66 + 4α c1 c2 D22 D44 F11 − 2A55 α c1 D22 F12 − 4α c1 c2 D12
4 4 6 2 2 4 2 2 4 4
D44 F12 − 4α c1 c2 D22 D55 F12 − 8α c1 c2 D44 D66 F12 − α c1 D22 F12 − 2α c1 c2 D44 F12 + 2A55 α c1 D12 F22 + 4α c1
4 4 4 2 6 2 4 2
c2 D12 D55 F22 + 4A55 α c1 D66 F22 + 8α c1 c2 D55 D66 F22 − 2A55 α c1 F12 F22 + 2α c1 D12 F12 F22 − 4α c1 c2 D55 F12
6 3 2 6 2 2 6 3 2 2 2 4 2 2 4 2 2
F22 + 2α c1 F12 F22 − α c1 D11 F22 − 2α c1 F11 F22 + 2A55 α c2 D12 F44 − α c2 D12 F44 + α c2 D11 D22 F44 + 4α
3 2 2 4 2 2 3 4 2 4 2
c2 D12 D55 F44 + 4A55 α c2 D66 F44 − 2α c2 D12 D66 F44 + 8α c2 D55 D66 F44 + 2α c1 c2 D22 F11 F44 − 2α c1 c2 D12
4 2 4 2 2 2 2 3 4 2 2 3 4
F12 F44 − 4α c1 c2 D66 F12 F44 − α c1 c2 F12 F44 + 4α c2 D12 D44 F55 + α c2 D22 D66 F55 + 8α c2 D44 D66 F55 − 2α
2 4 2 4 2 4 2 2 2 4 2
c1 c2 D22 F12 F55 + 2α c1 c2 D12 F22 F55 + 4α c1 c2 D66 F22 F55 − 2α c1 c2 F12 F22 F55 + 2α c2 D12 F44 F55 + 4α
4 4 4 6 2 4 2 4 2
c2 D66 F44 F55 − 2A55 α c1 D22 F66 − 4α c1 c2 D22 D55 F66 − 4α c1 D22 F12 F66 − 8α c1 c2 D44 F12 F66 − 4A55 α c1 F22
6 2 4 2 4 2 2 4 2 4 2 2
F66 + 4α c1 D12 F22 F66 − 8α c1 c2 D55 F22 F66 − 4α c1 c2 F12 F44 F66 − 2α c1 c2 D22 F55 F66 − 4α c1 c2 F22 F55
6 2 2 4 2 2 6 3 2 4 2 2 2 4 2 6 4
F66 − 4α c1 D22 F66 − 8α c1 c2 D44 F66 − 8α c1 F22 F66 − 4α c1 c2 F44 F66 + 2α c1 c2 D22 D44 H11 − α c1
2 4 2 2 6 2 6 3 6 3 6 3
F22 H11 + α c1 c2 D22 F44 H11 + 2α c1 D22 D66 H12 − 2α c1 D22 F12 H12 + 2α c1 D12 F22 H12 + 4α c1 D66 F22 H12
6 4 6 4 6 4 2 4 2 6 2 2 6 2
+ 2α c1 F12 F22 H12 + 4α c1 F22 F66 H12 − α c1 D22 H12 + 2A55 α c1 D12 H22 − α c1 D12 H22 + α c1 D11 D22 H22
4 2 4 2 6 2 4 2 6 3 6 3
+ 4α c1 c2 D12 D55 H22 + 4A55 α c1 D66 H22 − 2α c1 D12 D66 H22 + 8α c1 c2 D55 D66 H22 + 2α c1 D22 F11 H22 − 2α c1
6 3 6 4 2 4 2 2 4 2 2 6 4
D12 F12 H22 − 4α c1 D66 F12 H22 − α c1 F12 H22 + 2α c1 c2 D12 F55 H22 + 4α c1 c2 D66 F55 H22 − 4α c1 F12 F66 H22
6 4 2 6 4 4 2 4 2 4 2 6 2
− 4α c1 F66 H22 + α c1 D22 H11 H22 + A55 α c1 D22 H66 + 4α c1 c2 D12 D44 H66 + 2α c1 c2 D22 D55 H66 + 4α c1 D22
4 2 6 3 6 3 6 3 4 2 2
D66 H66 + 8α c1 c2 D44 D66 H66 − 4α c1 D22 F12 H66 + 4α c1 D12 F22 H66 + 8α c1 D66 F22 H66 + 2α c1 c2 D12 F44 H66
4 2 2 4 2 2 6 4 6 4 6 4 2
+ 4α c1 c2 D66 F44 H66 + α c1 c2 D22 F55 H66 − 2α c1 D22 H12 H66 + 2α c1 D12 H22 H66 + 4α c1 D66 H22 H66 + A55 α
2 2 4 2
D22 λNx0 + 2A55 c2 D44 λNx0 + 2α c2 D22 D55 λNx0 + 4c2 D44 D55 λNx0 + α D22 D66 λNx0 + 2α c2 D44 D66 λNx0
2 2 4 2 3 2
+ 2A55 α c1 F22 λNx0 + 4α c1 c2 D55 F22 λNx0 + 2α c1 D66 F22 λNx0 + A55 c2 F44 λNx0 + 2c2 D55 F44 λNx0 + α
2 2 2 3 2 2 4 4
c2 D66 F44 λNx0 + α c2 D22 F55 λNx0 + 2c2 D44 F55 λNx0 + 2α c1 c2 F22 F55 λNx0 + c2 F44 F55 λNx0 + 2α c1
2 4 2 2 2 2 2
D22 F66 λNx0 + 4α c1 c2 D44 F66 λNx0 + 4α c1 F22 F66 λNx0 + 2α c1 c2 F44 F66 λNx0 + A55 α c1 H22 λNx0
2 2 4 2 2 2 2 4 3 4 2
+ 2α c1 c2 D55 H22 λNx0 + α c1 D66 H22 λNx0 + α c1 c2 F55 H22 λNx0 + 2α c1 F66 H22 λNx0 + α c1 D22
2 2 4 3 2 2 2 4 4 2
H66 λNx0 + 2α c1 c2 D44 H66 λNx0
+ 2α c1 F22 H66 λNx0
+ + α c1 c2 F44 H66 λNx0
+α α c1 H22 H66 λNx0
A55 (D66 + c1 (2F66 + c1 H66 )) + 2c2 (D11 D44 + D55 D66 + c1 (2D44 F11 + 2D55 F66 + c1 D44 H11 + c1 D55 H66 ))
2 2 2
+ c2 (D11 F44 + D66 F55 + c1 (2F11 F44 + 2F55 F66 + c1 F44 H11 + c1 F55 H66 )) − α D12 − D11 (D22 + c1 (2F22
+ c1 H22 )) + 2D12 (D66 + c1 (2F12 + 2F66 + c1 (H12 + H66 ))) + c1 (4F12 (D66 + c1 F12 ) − D22 (2F11 + c1 H11 )
+ c1 (8F12 F66 + 2D66 H12 − 2F11 (2F22 + c1 H22 ) + c1 (−2F22 H11 + H12 (4(F12 + F66 ) + c1 H12 ) − c1 H11 H22
2 4 2
+ 4F12 H66 + 2c1 H12 H66 ))) λNy0 + A44 2A55 α (D12 + 2D66 ) + α − D12 + D11 D22 + c1 2D22 F11 − 4D66
2
F12 − c1 (F12 + 2F66 ) + c1 D22 H11 + 4c1 D66 H66 − 2D12 (D66 + c1 (F12 − c1 H66 )) + A55 λNx0 + c2 (2D55 + c2
2 2
F55 )λNx0 + α 4c2 D55 (D12 + 2D66 ) + 2c2 (D12 + 2D66 )F55 + (D66 + c1 (2F66 + c1 H66 ))λNx0 + (D11 + c1
(2F11 + c1 H11 ))λNy0
2 2 2 2 2 2 2 2
a5 = −α A55 + 2c2 D55 + c2 F55 + α D66 + 2c1 F66 + c1 H66 α A44 D22 + 2c2 D22 D44 + c2 D22 F44 + α c1
2 2 2
−F22 + D22 H22 + A44 + c2 (2D44 + c2 F44 ) + α D22 + 2c1 F22 + c1 H22 λNy0
(B.1)
61
62