a unified chebyshev–ritz formulation for vibration analysis of composite laminated deep open...
TRANSCRIPT
Arch Appl Mech (2014) 84:441–471DOI 10.1007/s00419-013-0810-1
ORIGINAL
Tiangui Ye · Guoyong Jin · Zhu Su · Xingzhao Jia
A unified Chebyshev–Ritz formulation for vibration analysisof composite laminated deep open shells with arbitraryboundary conditions
Received: 8 October 2013 / Accepted: 27 November 2013 / Published online: 21 December 2013© Springer-Verlag Berlin Heidelberg 2013
Abstract In this paper, a unified Chebyshev–Ritz formulation is presented to investigate the vibrations ofcomposite laminated deep open shells with various shell curvatures and arbitrary restraints, including cylindri-cal, conical and spherical ones. The general first-order shear deformation shell theory is employed to includethe effects of rotary inertias and shear deformation. Under the current framework, regardless of boundaryconditions, each of displacements and rotations of the open shells is invariantly expressed as Chebyshevorthogonal polynomials of first kind in both directions. Then, the accurate solutions are obtained by using theRayleigh–Ritz procedure based on the energy functional of the open shells. The convergence and accuracy ofthe present formulation are verified by a considerable number of convergence tests and comparisons. A varietyof numerical examples are presented for the vibrations of the composite laminated deep shells with variousgeometric dimensions and lamination schemes. Different sets of classical constraints, elastic supports as wellas their combinations are considered. These results may serve as reference data for future researches. Paramet-ric studies are also undertaken, giving insight into the effects of elastic restraint parameters, fiber orientation,layer number, subtended angle as well as conical angle on the vibration frequencies of the composite openshells.
Keywords Vibration analysis · Chebyshev–Ritz formulation · Laminated open cylindrical shells · Laminatedopen conical shells · Laminated open spherical shells · Arbitrary boundary conditions
1 Introduction
Composite laminated deep open shells, consisting of segments of shells of revolution (e.g., open cylindrical,conical and spherical shells), have a wide range of engineering applications, particularly in aerospace crafts,military hardware and civil constructions. The composite laminated open shells in these applications can besubjected to various boundary conditions, such as classical restraints, elastic supports and their combinations.Understanding the vibration characteristics of these shell components is particularly important for engineersto design suitable structures with low vibration and noise radiation characteristics. Compared with many otherkinds of open shells, the open cylindrical, conical and spherical shells are most frequently encountered in theengineering practices. Therefore, the present work is focused on the vibration analysis of these three types ofopen shells with arbitrary boundary conditions.
In the past few decades, extensive researches regarding the vibration analysis of composite laminatedshells have been carried out. Recently, a modified Fourier method had been used to predict the vibration
T. Ye · G. Jin (B) · Z. Su · X. JiaCollege of Power and Energy Engineering, Harbin Engineering University,Harbin 150001, People’s Republic of ChinaE-mail: [email protected]; [email protected]
T. YeE-mail: [email protected]
442 T. Ye et al.
characteristics of thin and moderately thick composite cylindrical shells with general elastic restraints andinternal line supports by Jin et al. [1,2]. Qu et al. [3] analyzed the free and forced vibrations of compositelaminated shells of revolution, in which a modified variational principle in conjunction with a multi-segmentpartitioning technique is employed to derive the formulation based on the first-order shear deformation theory.Exhaustive descriptions of various theoretical formulations and solution methods on this subject are availablein the review articles [4–7] and monographs [8–10]. However, most of these researches are restricted to theclosed shells (shells of revolution) and the information available for the vibration characteristics of compositelaminated deep open shells; especially, the open conical and spherical ones is very limited despite their practicalimportance. The most likely reason for this lacuna lies in the analytical difficulties involved: for a completelyclosed shell such as circular cylindrical, circular conical and spherical shells, the assumed 2D displacementfield can be reduced to a quasi 1D problem through Fourier decomposition of the circumferential wave motion.However, for an open shell, the assumption of whole periodic wave numbers in the circumferential directionis inappropriate, and thus, a set of complete three-dimensional analysis is required and resort must be made toa full two-dimensional solution scheme. Such a scheme will inevitably be complicated further by the depen-dence of the circumferential arc length on its meridional location [11]. This forms a major deterrent so thatthe analyses of open shells have not been widely available.
To overcome the difficulties, the shallow shell theories that are based on the general shell equations andcertain additional assumptions have been developed. A few of notable works on the free vibration analyses ofcomposite shallow shells can be found in Refs. [12–27]. Recently, the vibration behaviors of thin compositelaminated shallow shells with general elastic boundary conditions were reported by Ye et al. [28] using amodified Fourier series method. However, the shallow shell theories are limited to the open shells which havesmall curvatures (i.e., large radius of curvature). It has been proven that the shallow shell theories will giveinaccurate results for the lower frequencies when applied to deep open shells [8]. Researches on the vibrationcharacteristics of deep open shells are still limited in the open literature. Among those available, Bardell etal. [29] studied the vibration of a general three-layer conical sandwich panel based on the h-p version of thefinite element method. The static and free vibration analysis of laminated shells is performed by radial basisfunctions collocation, according to a layerwise deformation theory by Ferreira et al. [14]. A general surveyand comparison for variety of simply supported shallow spherical, cylindrical, plate and saddle panels in rect-angular planform was made by Chern and Chao [30]. By using the finite element method and classical shelltheories, Selmane and Lakis [31] investigated the dynamic and static behaviors of thin, elastic, anisotropicand non-uniform open cylindrical shells. Free vibration of simply supported laminated spherical panels withrandom material properties is reported by Singh et al. [32] based on the high-order shear deformation shallowtheory. The influence of classical boundary conditions and transverse shear on the vibration of angle-ply lam-inated plates and cylindrical panels was investigated by Soldatos and Messina [33]. This analysis was basedon the unified shear deformable Love-type theory and Ritz method. The vibration characteristics of twistedcantilevered conical composite shells were reported by Lee et al. [34] and Hu et al. [35]. Also, vibration of can-tilevered laminated composite shallow conical shells was presented by Lim et al. [36]. In contrast to the openconical and spherical shells, the researches on the open cylindrical ones are considerable. The free vibrationof two-side simply supported laminated cylindrical panels is analyzed by Zhao et al. [37,38] via the mesh-free kp-Ritz method. Comprehensive studies of thin, laminated cylindrical panels are conducted by Bardellet al. [39] by using the h − p version of the FEM. The natural frequencies of thin two-side simply supportedlaminated cylindrical panels were calculated by Bercin [40]. Some other contributors on this subject are thefollowing: Messina and Soldatos [41,42], Selmane and Lakis [43], Lee and Reddy [44], Messina [45], etc.
In view of the aforementioned issues and concerns, it should be emphasized that most of the existingcontributions were restricted to composite laminated open shells with large radius of curvature or subjectedto a limited set of classical supports. The vibration behaviors of composite laminated deep open shells withcircumferentially varying geometry and different combinations of classical and non-classical boundary con-straints have until now remained unsolved. Moreover, most of the available solution procedures in the openliterature are often only customized for a specific set of restraint conditions, which may not be appropriate forpractical applications because there are hundreds of different combinations of boundary conditions for an openshell. Developing an accurate, robust and efficient method which is capable of simplifying solution algorithms,reducing model input data and universally dealing with various boundary conditions is still of great interest toboth researchers and engineers.
The main purpose of this study is to complement the vibration investigations of composite laminated deepopen shells with arbitrary boundary conditions and develop a unified and accurate formulation to provide someuseful results for the titled problem, which may be used as reference data for future researchers. Under the cur-
Arbitrary boundary conditions 443
rent formulation, the first-order shear deformation shell theory based on the general shell equations is employedto include the effects of rotary inertias and shear deformation. Each displacement and rotation of the open shells,regardless of boundary conditions, is invariantly expressed as Chebyshev polynomials of first kind in both direc-tions. Thereby, all the Chebyshev expanded coefficients are treated equally and independently as the generalizedcoordinates and solved directly by using the Rayleigh–Ritz procedure. The convergence and accuracy of thepresent formulation are checked by a considerable number of convergence tests and comparisons. A variety ofnumerical examples are presented for the free vibration of shallow and deep composite laminated cylindrical,conical and spherical panels with various geometric dimensions and lamination schemes. Different combina-tions of classical boundary conditions (e.g., completely free, shear-diaphragm restrained, simply supported andclamped) and uniform elastic restraints as well as their combinations are considered in the investigation. Para-metric studies are also undertaken, giving insight into the effects of elastic restraint parameter, fiber orientation,layer number, subtended angle as well as conical angle on the vibration frequencies of the composite open shells.
2 Theoretical formulations
2.1 Preliminaries
A doubly curved composite laminated deep open shell with length Lα , width Lβ and uniform total thickness hshown in Fig. 1 is selected as the analysis model. Let (α, β and z) denote the orthogonal curvilinear coordinatesystem such that α and β curves are lines of curvature on the middle surface (z = 0). Rα and Rβ represent thevalues of the principal radii of curvature of the middle surface. The considered open shell is composed of NLorthotropic layers. Unless otherwise stated, all the layers are assumed to be of equal thickness (h/NL). Thecharacters Zk and Zk+1 are used to indicate the distances from the undersurface and the top surface of the k’thlayer to the referenced plane. According to the FSDT assumptions, the displacement field in the shell space isgiven as follows:
u(α, β, z, t) = u0(α, β, t)+ zφα(α, β, t)
v(α, β, z, t) = v0(α, β, t)+ zφβ(α, β, t)
w(α, β, z, t) = w0(α, β, t) (1)
where t is time variable; u, v andw are the generalized displacements along the α, β and z coordinates, respec-tively. u0, v0 andw0 are the displacements of a point on the middle surface, and φα and φβ separately representthe rotations at z = 0 of transverse normal to the mid-surface with respect to theβ, α coordinates. The strain dis-placement relations at any point in the shell space with reference to curvilinear coordinate system are as follows:
εα = ε0α + zχα, εβ = ε0
β + zχβ, γαβ = γ 0αβ + zχαβ
γαz = ∂w0
∂α+ φα, γβz = ∂w0
∂β+ φβ (2)
z
w
vu
Rα
Rβ
β
α
(b)(a)
Lα
Lβ
Middle surface
h
k+1
k
Fig. 1 Geometry and notations of a laminated open shell: a coordinates; b lamination scheme
444 T. Ye et al.
x
zz
θ θ
θ0 θ0
θ0
ϕ0
ϕ1
ϕ
θ
ϕc z
o
o
oL
LR R0
R1
z
x
R
(a) (b) (c)
Fig. 2 Coordinate systems and geometric parameters of three types of open shells: a open cylindrical shell; b open conical shell;c open spherical shell
where ε0α, ε
0β and γ 0
αβ denote the normal and shear strains in the (α, β, z) coordinate system. χα, χβ and χαβ arethe curvature and twist changes; the transverse shear strains γαz and γβz are constant through the thickness. Theparameters (ε0
α, ε0β, γ
0αβ, χα, χβ, χαβ) are defined in terms of the middle surface displacements and rotation
components as [8]:
ε0α = 1
A∂u0∂α
+ v0AB
∂A∂β
+ w0Rα, χα = 1
A∂φα∂α
+ φβAB
∂A∂β
ε0β = 1
B∂v0∂β
+ u0AB
∂B∂α
+ w0Rβ, χβ = 1
B∂φβ∂β
+ φαAB
∂B∂α
ε0αβ = 1
A∂v0∂α
− u0AB
∂A∂β
+ 1B∂u∂β
− v0AB
∂B∂α, χαβ = 1
A∂φβ∂α
− φαAB
∂A∂β
+ 1B∂φα∂β
− φβAB
∂B∂α
(3)
where the quantities A and B are the Lamé parameters. According to Fig. 2, the coordinate systems andgeometric parameters (Rα, Rβ, A and B) of the deep open cylindrical, conical and spherical shells underconsideration are the following:
(a) Cylindrical shells: α = x, β = θ, Rα = ∞, Rβ = R, A = 1, B = R;(b) Conical shells: α = x, β = θ, Rα = ∞, Rβ = x tan α0, A = 1, B = x sin α0;(c) Spherical shells: α = ϕ, β = θ, Rα = R, Rβ = R, A = R, B = R sin ϕ.
Considering Eq. (2), the stresses for the k’th layer are determined by the generalized Hooke’s law, namely:
⎧⎨
⎩
σασβταβ
⎫⎬
⎭=⎡
⎢⎣
Qk11 Qk
12 Qk16
Qk12 Qk
22 Qk26
Qk16 Qk
26 Qk66
⎤
⎥⎦
⎧⎨
⎩
εαεβγαβ
⎫⎬
⎭,
{τβzταz
}
=[
Qk44 Qk
45
Qk45 Qk
55
]{γβzγαz
}
(4)
in which σα and σβ are the normal stresses in the α, β directions, respectively. τβz, ταz and ταβ are the corre-
sponding shear stresses in the curvilinear coordinate system. Qki j (i, j = 1, 2, 4 ∼ 6) are the lamina stiffness
coefficients, they are defined as follows:
⎡
⎢⎢⎢⎢⎢⎢⎣
Qk11 Qk
12 0 0 Qk16
Qk12 Qk
22 0 0 Qk26
0 0 Qk44 Qk
45 0
0 0 Qk45 Qk
55 0
Qk16 Qk
26 0 0 Qk66
⎤
⎥⎥⎥⎥⎥⎥⎦
= T
⎡
⎢⎢⎢⎢⎣
Qk11 Qk
12 0 0 0Qk
21 Qk22 0 0 0
0 0 Qk44 0 0
0 0 0 Qk55 0
0 0 0 0 Qk66
⎤
⎥⎥⎥⎥⎦
TT (5)
Arbitrary boundary conditions 445
where Qki j (i, j = 1, 2, 4 ∼ 6) are the material constant of the k’th layer in the laminate coordinate system.
For the orthotropic material, they are known in terms of the engineering constants as:
Qk11 = E1
1 − μ12μ21, Qk
12 = μ12 E2
1 − μ12μ21= Qk
21, Qk22 = E2
1 − μ12μ21
Qk44 = G23, Qk
55 = G13, Qk66 = G12 (6)
where E1 is the longitudinal modulus, E2 is the transverse modulus, and μ12 is the major Poisson’s ratio.The Poisson’s ratio μ21 are determined by equation μ12 E2 = μ21 E1. G12,G13 and G23 are shear moduli. Itshould be noted that by letting E1 = E2,G12 = G13 = G23 = E1/(2+2μ12), the present formulation can bereadily used to analyze the isotropic deep open shells with general boundary restraints. T is the transformationmatrix, which is defined as:
T =
⎡
⎢⎢⎢⎣
cos2 ϑ sin2 ϑ 0 0 −2 sin ϑ cosϑsin2 ϑ cos2 ϑ 0 0 2 sin ϑ cosϑ
0 0 cosϑ sin ϑ 00 0 − sin ϑ cosϑ 0
sin ϑ cosϑ − sin ϑ cosϑ 0 0 cos2 ϑ − sin2 ϑ
⎤
⎥⎥⎥⎦
(7)
where ϑ denotes the included angle between the material coordinate of the k’th layer and the α-axis of theshell. The force and moment resultants are obtained by integrating the stresses over the shell thickness:
⎡
⎣NαNβNαβ
⎤
⎦ =h/2∫
−h/2
⎡
⎣σασβταβ
⎤
⎦dz,
⎡
⎣Mα
Mβ
Mαβ
⎤
⎦ =h/2∫
−h/2
⎡
⎣σασβταβ
⎤
⎦zdz,
[Qα
Qβ
]
=h/2∫
−h/2
[ταzτβz
]
dz (8)
The constitutive equations relating the force and moment resultants to the strains and curvatures of the referencesurface are given in the matrix form:
⎡
⎢⎢⎢⎢⎢⎣
NαNβNαβMα
Mβ
Mαβ
⎤
⎥⎥⎥⎥⎥⎦
=
⎡
⎢⎢⎢⎢⎢⎣
A11 A12 A16 B11 B12 B16A12 A22 A26 B12 B22 B26A16 A26 A66 B16 B26 B66B11 B12 B16 D11 D12 D16B12 B22 B26 D12 D22 D26B16 B26 B66 D16 D26 D66
⎤
⎥⎥⎥⎥⎥⎦
⎡
⎢⎢⎢⎢⎢⎢⎣
ε0α
ε0β
γ 0αβ
χαχβχαβ
⎤
⎥⎥⎥⎥⎥⎥⎦
,
[Qβ
Qα
]
= kc
[A44 A45A45 A55
] [γβzγαz
]
(9)
where quantities (Nα, Nβ, Nαβ) are the normal and shear force resultants, and (Mα,Mβ,Mαβ) denote thebending and twisting moment resultants. Qβ and Qα are the transverse shear force resultants. kc is the shearcorrection factor. The existence of the shear correction factor is to make sure that the strain energy due to thetransverse shear stresses in Eq. (8) equals the strain energy due to the true transverse shear stresses predictedby the three-dimensional elasticity theory [10]. In the present analysis, the shear correction factor is typicallytaken at 5/6. Ai j , Bi j and Di j are the stretching, stretching-bending coupling and bending stiffness. They aredefined as follows:
Ai j =NL∑
k=1
Qki j (Zk+1 − Zk), (i, j = 1, 2, 6); Ai j = kc
NL∑
k=1
Qki j (Zk+1 − Zk), (i, j = 4, 5)
Bi j = 1
2
NL∑
k=1
Qki j
[Z2
k+1 − Z2k
], Di j = 1
3
NL∑
k=1
Qki j
[Z3
k+1 − Z3k
], (i, j = 1, 2, 4, 5, 6) (10)
It should be noted that for open shells, which are symmetrical with respect to their middle surface, Bi j = 0. Inthis part, the kinematic relations and stress resultants of the composite laminated open shell based on generalshell equations are presented, and in the next subsection, the energy expressions of the open shell are obtainedby using these relations and resultants.
446 T. Ye et al.
2.2 Energy expressions
It is well known that the exact solutions are generally available only for the open shells that are simply sup-ported along at least one pair of opposite edges. For other boundary conditions, however, one may have to useapproximate methods such as the Rayleigh–Ritz method [47]. Since the main purpose of the present work is todevelop an efficient formulation, which can universally deal with various boundary conditions, the Rayleigh–Ritz method is employed due to its simplicity, stability and efficiency in numerical implementation. Therefore,the energy functional of the open shell is established firstly. The strain energy (UV ) of the considered openshell is defined as follows:
UV = 1
2
∫
S
{Nαε
0α + Nβε
0β + Nαβγ
0αβ + Mαχα + Mβχβ + Mαβχαβ + Qαγαz + Qβχβz
}dS (11)
where S represents the middle surface area of the shell. Substituting Eqs. (3) and (9) into Eq. (11), the strainenergy expression of the open shell is divided into three parts (i.e., UV = Us+Ub+Ubs) and written in terms ofshell displacements and rotations. Considering the Lamé parameter A for the open shells under considerationis constant, i.e., ∂A/∂α = 0 and ∂A/∂β = 0, these energy expressions are written as follows:
Us = 1
2
∫ ∫
⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩
A11
(1A∂u0∂α
+ w0Rα
)2 + A22
(1B∂v0∂β
+ u0AB
∂B∂α
+ w0Rβ
)2
+2A12
(1B∂v0∂β
+ u0AB
∂B∂α
+ w0Rβ
) (1A∂u0∂α
+ w0Rα
)
+2A16
(1A∂u0∂α
+ w0Rα
) (1A∂v0∂α
+ 1B∂u0∂β
− v0AB
∂B∂α
)
+2A26
(1B∂v0∂β
+ u0AB
∂B∂α
+ w0Rβ
) (1A∂v0∂α
+ 1B∂u0∂β
− v0AB
∂B∂α
)
+A66
(1A∂v0∂α
+ 1B∂u0∂β
− v0AB
∂B∂α
)2 + A44
(1B∂w0∂β
+ φβ
)2
+2A45
(1B∂w0∂β
+ φβ
) (1A∂w0∂α
+ φα
)+ A55
(1A∂w0∂α
+ φα
)2
⎫⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎬
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎭
ABdαdβ (12)
Ub = 1
2
∫ ∫
⎧⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎩
D11
(1A∂φα∂α
)2 + 2D12
(1B∂φβ∂β
+ φαAB
∂B∂α
) (1A∂φα∂α
)
+D22
(1B∂φβ∂β
+ φαAB
∂B∂α
)2 + D66
(1A∂φβ∂α
+ 1B∂φα∂β
− φβAB
∂B∂α
)2
+2D16
(1A∂φα∂α
) (1A∂φβ∂α
+ 1B∂φα∂β
− φβAB
∂B∂α
)
+2D26
(1B∂φβ∂β
+ φαAB
∂B∂α
) (1A∂φβ∂α
+ 1B∂φα∂β
− φβAB
∂B∂α
)
⎫⎪⎪⎪⎪⎪⎪⎬
⎪⎪⎪⎪⎪⎪⎭
ABdαdβ (13)
Ubs =∫ ∫
⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩
[B11
(1A∂φα∂α
)+ B12
(1B∂φβ∂β
+ φαAB
∂B∂α
)] (1A∂u0∂α
+ w0Rα
)
+B16
(1A∂φβ∂α
+ 1B∂φα∂β
− φβAB
∂B∂α
) (1A∂u0∂α
+ w0Rα
)
+[
B12
(1A∂φα∂α
)+ B22
(1B∂φβ∂β
+ φαAB
∂B∂α
)] (1B∂v0∂β
+ u0AB
∂B∂α
+ w0Rβ
)
+B26
(1A∂φβ∂α
+ 1B∂φα∂β
− φβAB
∂B∂α
) (1B∂v0∂β
+ u0AB
∂B∂α
+ w0Rβ
)
+[
B16
(1A∂φα∂α
)+ B26
(1B∂φβ∂β
+ φαAB
∂B∂α
)] (1A∂v0∂α
+ 1B∂u0∂β
− v0AB
∂B∂α
)
+B66
(1A∂φβ∂α
+ 1B∂φα∂β
− φβAB
∂B∂α
) (1A∂v0∂α
+ 1B∂u0∂β
− v0AB
∂B∂α
)
⎫⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎬
⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎭
ABdαdβ (14)
where Us,Ub and Ubs indicate the stretching, bending and bending–stretching coupling energy expressions,respectively. The kinetic energy (T ) of the open shell is written as follows:
T = 1
2
∫ ∫⎧⎪⎨
⎪⎩
I0
(∂u0∂t
)2 + 2I1
(∂u0∂t
) (∂φα∂t
)+ I2
(∂φα∂t
)2 + I0
(∂v0∂t
)2
+2I1
(∂v0∂t
) (∂φβ∂t
)+ I2
(∂φβ∂t
)2 + I0
(∂w0∂t
)2
⎫⎪⎬
⎪⎭ABdαdβ (15)
and the inertia terms are the following:
[I0, I1, I2] =⎛
⎜⎝
NL∑
k=1
Zk+1∫
Zk
ρk[1, z1, z2]dz
⎞
⎟⎠ (16)
Arbitrary boundary conditions 447
in which ρk is the mass of the k’ th layer per unit middle surface area. The arbitrary boundary conditions ofthe open shell are implemented by introducing three groups of linear springs (ku, kv, kw) and two groups ofrotational springs (Kα, Kβ), which are distributed uniformly along the boundary at each of the shell ends.The stiffness of the boundary springs can take any value from zero to infinity to better model many real-worldrestraint conditions. For instance, the clamped restraint condition is essentially obtained by setting the springstiffness substantially larger than the bending rigidity of the involved open shell. For the sake of simplicity,kuψ, kvψ , kwψ , K α
ψ and K βψ(ψ = α0, β0, αL and βL) are used to indicate the stiffness (per unit length) of the
boundary springs and the subscripts α0, β0, αL and βL separately denote the springs distributed at the edgesα = 0. β = 0, α = Lα and α = Lβ . Hence, the potential energy (Psp) stored in the boundary springs is
Psp = 1
2
∫⎧⎨
⎩
[kuα0u2
0 + kvα0v20 + kwα0w
20 + K α
α0φ2α + K β
α0φ2β
] ∣∣∣α=0
+[kuαLu2
0 + kvαLv20 + kwαLw
20 + K α
αLφ2α + K β
αLφ2β
] ∣∣∣α=Lα
⎫⎬
⎭Bdβ
+1
2
∫ { [kuβ0u2
0 + kvβ0v20 + kwβ0w
20 + K α
β0φ2α + K β
β0φ2β ]β=0
+[kuβLu2
0 + kvβLv20 + kwβLw
20 + K α
βLφ2α + K β
βLφ2β ]β=Lβ
}
Adα (17)
Thus, the Lagrangian energy functional (L) of the shell can be defined in terms of the aforementioned energyexpressions as follows:
L = T − Us − Ub − Psp (18)
Once the Lagrangian energy functional of the open shell is established, the following task is to construct a setof appropriate admissible displacement functions and determine these functions.
2.3 Admissible displacement functions
The selection of appropriate admissible displacement functions is of particular importance in the Rayleigh–Ritzmethod because the accuracy of the solution will usually depend upon how well the actual displacement canbe faithfully represented by them [47]. The main advantage of the Lagrangian energy functional given in Eq.(18) is that the choice of the admissible displacement functions of the open shells is considerably simplified,and any continuous, independent, complete basis functions may be employed to produce accurate results. Thereason lies in the fact that the geometric boundary conditions in an open shell are relaxed and enforced throughboundary springs, which can be seen as penalty parameters [49,50], and there is no need to explicitly satisfy theessential and natural conditions on these boundaries for the admissible displacement functions in advance [51].In this work, the displacements and rotation components of the open shell are generally expanded, regardlessof boundary conditions, as Chebyshev polynomials of first kind, written as:
u0(α, β, t) =∞∑
m=0
∞∑
n=0
AmnTm(α)Pn(β)ejωt ; φα(α, β, t) =
∞∑
m=0
∞∑
n=0
DmnTm(α)P(β)ejωt
v0(α, β, t) =∞∑
m=0
∞∑
n=0
BmnTm(α)P(β)ejωt ; φβ(α, β, t) =
∞∑
m=0
∞∑
n=0
EmnTm(α)P(β)ejωt
w0(α, β, t) =∞∑
m=0
∞∑
n=0
CmnTm(α)P(β)ejωt
(19)
where Amn, Bmn,Cmn, Dmn and Emn are the corresponding Chebyshev expanded coefficients; Tm(α) andPn(β) are the mth and nth order Chebyshev polynomial for the displacement components in the α and βdirections, respectively. The Chebyshev polynomials of first kind are written as follows [52]:
T0(α) = 1, T1(α) = α, Tm(α) = 2αTm−1(α)− Tm−2(α) (m > 2)
P0(β) = 1, P1(β) = β, Pn(β) = 2βPn−1(β)− Pn−2(β) (n > 2) (20)
It should be remarked here that the Chebyshev polynomials of first kind given in Eq. (20) are complete andorthogonal polynomials defined on the interval of [−1, 1]. Thus, a coordinate transformation from α and β (forα ∈ [0, Lα] and β ∈ [0, Lβ ], to α, β ∈ [−1, 1]) needs to be introduced to implement the present analysis, i.e.,
448 T. Ye et al.
α = Lα(α + 1)/2 and β = Lβ(β + 1)/2. It is stressed that only finite terms of the displacement expressionsare used in actual calculations due to the limited speed, the capacity and the numerical accuracy of computers.For the sake of computational simplicity, it is assumed that the numbers of polynomial terms truncated forthe displacements and rotation components are uniformly chose as M and N . The good accuracy and fastconvergence behavior of the Chebyshev series solution will be shown later.
2.4 Solution procedure
With the admissible displacement functions and energy expressions of the open shell given in previous sub-sections, the remained task is to determine the unknown expansion coefficients in the admissible displacementfunctions. In the present work, the Rayleigh–Ritz procedure is adopted to obtain the vibration results of the openshell. Substituting the energy expressions and the displacement functions of the open shell into Eq. (18) and min-imizing the total expression of the Lagrangian energy functional with respect to the undetermined coefficients:
∂L
∂α= 0, α = Amn, Bmn,Cmn, Dmn, Emn, (21)
a total of 5 ∗ (M + 1) ∗ (N + 1) equations are obtained. They can be summed up in a matrix form:⎛
⎜⎜⎜⎜⎝
⎡
⎢⎢⎢⎢⎣
Ku,u Ku,v Ku,w Ku,α Ku,β
KTu,v Kv,v Kv,w Kv,α Kv,β
KTu,w KT
v,w Kw,w Kw,α Kw,β
KTu,α KT
v,α KTw,α Kα,α Kα,β
KTu,β KT
v,β KTw,β KT
α,β Kβ,β
⎤
⎥⎥⎥⎥⎦
− ω2
⎡
⎢⎢⎢⎢⎣
Mu,u 0 0 Mu,α 00 Mv,v 0 0 Mv,β
0 0 Mw,w 0 0MT
u,α 0 0 Mα,α 00 MT
v,β 0 0 Mβ,β
⎤
⎥⎥⎥⎥⎦
⎞
⎟⎟⎟⎟⎠
G = 0 (22)
where superscript T represents the transposition operator. The elements of the sub-matrices Ki, j and Mi, j(i, j = u, v and w) are given in Appendices A and B. G represent the collection of the undetermined coeffi-cients and is given as follows:
G = [Gu0 ,Gv0 ,Gw0 ,Gφα ,Gφβ]T
(23)
where
Gu = [A00, . . . , A0n, . . . , A0N , A10, . . . , A1n, . . . , Amn, . . . , AM N]
Gv = [B00, . . . , B0n, . . . , B0N , B10, . . . , B1n, . . . , Bmn, . . . , BM N]
Gw = [C00, . . . ,C0n, . . . ,C0N ,C10, . . . ,C1n, . . . ,Cmn, . . . ,CM N]
Gφα = [D00, . . . , D0n, . . . , D0N , D10, . . . , D1n, . . . , Dmn, . . . , DM N]
Gφβ = [E00, . . . , E0n, . . . , E0N , E10, . . . , E1n, . . . , Emn, . . . , EM N]
The natural frequencies and mode shapes of the open shell can now be easily obtained by solving a stan-dard matrix eigenproblem. Once the coefficients in the Chebyshev expansions are solved from Eq. (22), thedisplacements and rotation components of the shell can be directly determined by using Eq. (19). It should benoted that Eq. (22) represents free vibration analysis of the open shell. When the forced vibration is involved,by adding the work done by external force in the right-hand side of the Lagrangian energy functional, apply-ing the Rayleigh–Ritz procedure and summing the loading vector F on the right side of Eq. (22), then thecharacteristic equation for the forced vibration of the open shell is readily obtained.
3 Numerical examples and discussion
In this section, free vibration characteristics of the composite laminated deep open cylindrical, conical andspherical shells are investigated. In summary, the investigation is arranged as follows: first, the number ofterms in the displacement expressions is investigated and the comparisons between the present results andthe available benchmarks are made, aiming to verify the convergence and accuracy of the Chebyshev–Ritzformulation. The effects of the elastic restraint parameters are also studied. Then, a variety of new vibration
Arbitrary boundary conditions 449
results including frequencies and modes shapes for the composite laminated open cylindrical and conical shellswith various combinations of classical and uniform elastic restraints as well as different shell parameters (e.g.,geometric parameters, material properties and lamination schemes) are presented. Illustrative figures are givento show the effects of the fiber orientation, layer number, conical angle and subtended angle on the vibrationfrequencies of the shells as well. Finally, illustrative examples are presented for the open spherical shells andsome further numerical results for moderately thick composite laminated open spherical shells with variousboundary conditions and shell parameters are given.
3.1 Convergence study
In this subsection, the convergence of the proposed Chebyshev–Ritz method and effects of the elastic restraintparameters are investigated. For all numerical examples in this subsection, unless otherwise stated, thematerial constants of the open shells are the following: for the cylindrical shells: E1 = 60.7 GPa , E2 =24.8 GPa , μ12 = 0.23,G12 = G13 = G23 = 12 GPa , ρ = 1700 kg/m 3; for the conical shells: E1 =E2 = 70 GPa , μ12 = 0.3,G12 = G13 = G23 = 26.9 GPa , ρ = 2,700 kg/m 3; and for the spherical shells:E1 = 138 GPa , E2 = 8.96 GPa , μ12 = 3,G12 = G13 = 7.1 GPa ,G23 = 3.9 GPa , ρ = 1, 500 kg/m 3.
In Tables 1, 2 and 3, the convergence of the frequencies of the aforementioned open cylindrical, conical andspherical shells with completely free boundary condition is presented, respectively. The geometric dimensionsof these shells are as follows: for the cylindrical shell: R = 2 m , L = 1 m , h = 0.01 m , θ0 = 28.6◦; for theconical shell: R0 = 0.16 m , L = 1.12 m , h = 0.002 m , ϕc = 26.5◦, θ0 = 180◦; and for the spherical shell:R = 2 m , h = 0.01 m , ϕ0 = 75.5◦, ϕ1 = 104.5◦, θ0 = 28.6◦ and 180◦. In all the following computations, thezero frequencies corresponding to the rigid body modes were omitted from the results. Excellent convergenceof frequencies can be observed in the tables. In order to verify the accuracy of the formulation, the numericalresults reported by Zhao et al. [37] by using mesh-free method, Messina and Soldatos [42] based on HSDT aswell as Qatu and Leissa [46] based on shallow shell theory are included in Table 1. These comparisons showedthe present solutions are in good agreement with the reference results, although different theories and methodswere employed in the literature. In Table 2, the h-p FEM and experiment results reported by Bardell et al. [11]are listed as well. A consistent agreement between the present solutions and the reference data is seen from
Table 1 Convergence and comparison of frequency parameters Ω = ωL2α/h
√ρ/E1 of a four-layered, angle-ply
[−60◦/60◦/60◦ −60◦] e-glass/epoxy open cylindrical shell (R = 2 m , L/R = 0.5, h/L = 0.01, θ0 = 28.6◦)
Terms (M × N ) Mode number
1 2 3 4 5 6 7 8 9 10
11 × 11 3.2440 5.5872 8.3682 11.114 12.495 15.270 17.799 21.996 22.052 26.00212 × 12 3.2422 5.5869 8.3672 11.112 12.494 15.269 17.793 21.986 22.044 26.00213 × 13 3.2422 5.5868 8.3639 11.109 12.488 15.269 17.792 21.986 22.039 25.99914 × 14 3.2404 5.5866 8.3629 11.107 12.487 15.269 17.786 21.977 22.032 25.99915 × 15 3.2404 5.5865 8.3598 11.105 12.481 15.268 17.785 21.976 22.027 25.99616 × 16 3.2404 5.5863 8.3590 11.103 12.481 15.268 17.785 21.976 22.021 25.996Zhao et al. [37] 3.3016 5.7328 8.5087 11.133 12.626 15.724 18.231 22.129 22.257 25.828Messina and Soldatos [42] 3.2498 5.5910 8.3873 11.137 12.533 15.328 17.895 22.072 22.142 26.036Qatu and Leissa[46] 3.2920 5.7416 8.5412 11.114 12.591 15.696 18.221 22.058 22.194 25.871
Table 2 Convergence and comparison of frequencies (Hz) for an isotropic open conical shell (R0 = 0.16 m , L = 1.12 m , h =0.002 m , θ0 = 180◦, ϕc = 26.5◦)
Terms (M × N ) Mode number
1 2 3 4 5 6 7 8 9 10
13 × 13 4.273 8.201 11.12 20.68 21.80 32.75 47.16 47.21 64.77 67.8214 × 14 4.273 8.200 11.12 20.68 21.79 32.75 46.67 47.17 64.27 67.1315 × 15 4.273 8.200 11.12 20.67 21.78 32.74 46.66 47.13 63.02 67.0416 × 16 4.273 8.199 11.12 20.67 21.78 32.74 46.65 47.13 62.99 66.9117 × 17 4.273 8.199 11.12 20.67 21.77 32.74 46.65 47.12 62.92 66.9018 × 18 4.273 8.198 11.12 20.67 21.77 32.74 46.65 47.12 62.92 66.89ANSYS [11] 4.31 8.58 11.28 20.63 21.92 32.71 46.68 47.20 63.16 66.46Experiment [11] 4.5 8.9 11.5 20.9 21.7 33.2 46.6 47.4 58.6 63.7
450 T. Ye et al.
Table 3 Convergence and comparison of frequency parameters Ω = ωL2α/h
√ρ/E1 of a four-layered, angle-ply
[30◦/−30◦/−30◦/30◦] graphite/epoxy open spherical shell (R = 2 m , h = 0.01 m , ϕ0 = 75.5◦, ϕ1 = 104.5◦)
Terms (M × N ) Mode number (θ0 = 29◦) Mode number (θ0 = 180◦)
1 2 3 4 5 1 2 3 4 5
12 × 12 2.334 3.391 5.995 6.877 8.507 0.0715 0.2150 0.2960 0.4711 0.651313 × 13 2.334 3.391 5.990 6.871 8.497 0.0715 0.2149 0.2960 0.4709 0.651314 × 14 2.333 3.390 5.989 6.869 8.495 0.0715 0.2149 0.2958 0.4708 0.650915 × 15 2.333 3.389 5.985 6.865 8.488 0.0715 0.2149 0.2958 0.4707 0.650816 × 16 2.332 3.389 5.985 6.863 8.487 0.0715 0.2149 0.2957 0.4707 0.650617 × 17 2.332 3.389 5.985 6.863 8.487 0.0715 0.2148 0.2957 0.4704 0.6506Qatu [8] 2.283 3.323 5.871 6.781 8.394 – – – – –
the table. In Table 3, a comparison is made with the frequency parameters Ω = ωL2/h√ρ/E1 obtained by
Qatu [8] by using the classical shallow theory. The symbols “–” are missing data that were not considered byQatu. The discrepancy between the present and the reference results is acceptable. The small discrepancy inthe results may be attributed to different shell theories used in the literature. It has been proven that the shallowshell theories will give inaccurate results at the lower frequencies when applied to deep open shells.
Then, the influence of the elastic restraint stiffness on the open shells is investigated. For simplicity andconvenience in the analysis, five non-dimensional elastic restraint parametersΓu, Γv, Γw, Γα andΓβ , which aredefined as the ratios of the corresponding spring stiffness to the bending stiffness D = E1h3/12(1 −μ12μ21),respectively, are introduced here, i.e., Γu = ku/D, Γv = kv/D, Γw = kw/D, Γα = Kα/D and Γβ = Kβ/D.Also, a frequency parameter�Ω , which is defined as the difference of the dimensionless frequency parametersΩ = ωL2
α/h√ρ/E1 to those of the elastic restraint parameters Γλ(λ = u, v, w, α and β ) equal to 10−1, i.e.,
�Ω = ΩΓλ −ΩΓλ=10−1 are used in the calculations. In Figs. 3 and 4, variation in the 1st, 3rd and 5th mode fre-quency parameters�Ω versus the elastic restraint parametersΓλ for three-layered, cross-ply [0◦/90◦/0◦] opencylindrical, conical and spherical shells with various elastic restraints are presented. The geometric dimensionsof these shells are the following: for the cylindrical shell: R = 1 m, L/R = 2, h/R = 0.1, θ0 = 120◦; forthe conical shell: R0 = 1 m , L/R0 = 2, h/R0 = 0.1, ϕc = 45◦, θ0 = 120◦; and for the spherical shell:R = 1 m , h/R = 0.1, ϕ0 = 30◦, ϕ1 = 90◦, θ0 = 120◦. In Fig. 3, the composite open shells are clampedat edge α = 0, free at edges β = 0 and β = Lβ , while the other edge is elastically restrained by only onekind of spring components with various stiffness (denoted by C–F–Fe–F). It is clearly that in a certain range,the frequency parameters �Ω increase rapidly as the elastic restraint parameters Γλ increase. And beyondthis range, there is little variation in the frequency parameters. For the sake of completeness, in Fig. 4, theshells is assumed to be clamped at edge α = 0 and free at edges β = 0 and β = Lβ as usual, but the otheredge is restrained by all the five groups boundary springs in which four groups of them with infinite stiffness(107 D) and the rest one is assigned at arbitrary stiffness (denoted by C–F–Ce–F). The similar tendency asFig. 3 is seen in the figures. This study shows the active ranges of the elastic restraints stiffness on the vibrationcharacteristic of the composite open shells varied with spring components. In this case, they can be defined asΓu : 101 − 104, Γv : 100 − 103, Γw : 10−1 − 103, Γα : 10−1 − 102 and Γβ : 10−1 − 102, respectively.
3.2 Open shells with arbitrary boundary conditions
The present subsection contains comparisons and new results for the composite laminated shallow and deepopen shells with various boundary conditions and different shell parameters. As pointed out by Qatu [8], thethick shallow and deep open shells can have 24 possible classical boundary conditions at each edge, which willresult in a high number of combinations of boundary conditions. Besides, open shells with elastic constraintsare often encountered and such constraints can be generalized by introducing translational and rotationalsprings, which are distributed uniformly along the boundary at each of the shell ends [8]. And it is impossibleto undertake an all-encompassing survey of every case of the aforementioned boundary conditions. Therefore,only four typical classical boundaries, i.e., completely free (F), shear-diaphragm supported (SD), simply sup-ported (S) and complete clamped (C), and three uniform elastic restraint conditions, i.e., E1, E2 and E3, areconsidered in the current investigation. Taking edge α = 0, for example, the corresponding spring stiffnessfor these seven types of boundaries are the following:
Arbitrary boundary conditions 451
10-1
101
103
105
107
0
0.2
0.4
0.6
0.8
1
Elastic restraint parameter: Log(Γu)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(a)Cylindrical shell
10-1
101
103
105
107
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Elastic restraint parameter: Log(Γu)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(a)Conical shell
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5
4
Elastic restraint parameter: Log(Γu)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(a)Spherical shell
10-1
101
103
105
107
0
1
2
3
4
5
6
7
Elastic restraint parameter: Log(Γv)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(b)
10-1
101
103
105
107
0
1
2
3
4
5
6
Elastic restraint parameter: Log(Γv)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(b)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5
Elastic restraint parameter: Log(Γv)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(b)
10-1
101
103
105
107
0
1
2
3
4
5
6
Elastic restraint parameter: Log(Γw)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(c)
10-1
101
103
105
107
0
1
2
3
4
5
Elastic restraint parameter: Log(Γw)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(c)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5
Elastic restraint parameter: Log(Γw)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(c)
10-1
101
103
105
107
0
0.2
0.4
0.6
0.8
Elastic restraint parameter: Log(Γα)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(d)
10-1
101
103
105
107
0
0.2
0.4
0.6
0.8
1
Elastic restraint parameter: Log(Γα)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(d)
10-1
101
103
105
107
0
0.1
0.2
0.3
0.4
0.5
Elastic restraint parameter: Log(Γα)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(d)
10-1
101
103
105
107
0
0.5
1
1.5
2
Elastic restraint parameter: Log(Γβ)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(e)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
Elastic restraint parameter: Log(Γβ)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(e)
10-1
101
103
105
107
0
0.2
0.4
0.6
0.8
1
Elastic restraint parameter: Log(Γβ)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(e)
Fig. 3 Relationships of frequency parameters �Ω with elastic restraint parameter Γλ for certain three-layered, cross-ply[0◦/90◦/0◦] open shells with C–F–Fe-F boundary condition: a Γu ; b Γv ; c Γw; d Γα ; e Γβ
452 T. Ye et al.
10-1
101
103
105
107
0
0.2
0.4
0.6
0.8
1
1.2
Elastic restraint parameter: Log(Γu)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(a)Cylindrical shell
10-1
101
103
105
107
0
0.1
0.2
0.3
0.4
0.5
Elastic restraint parameter: Log(Γu)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(a)Conical shell
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5
Elastic restraint parameter: Log(Γu)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(a)Spherical shell
10-1
101
103
105
107
0
1
2
3
4
5
Elastic restraint parameter: Log(Γv)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(b)
10-1
101
103
105
107
0
1
2
3
4
Elastic restraint parameter: Log(Γv)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(b)
10-1
101
103
105
107
0
1
2
3
4
5
Elastic restraint parameter: Log(Γv)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(b)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5
Elastic restraint parameter: Log(Γw)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(c)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5
4
Elastic restraint parameter: Log(Γw)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(c)
10-1
101
103
105
107
0
0.5
1
1.5
Elastic restraint parameter: Log(Γw)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(c)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
Elastic restraint parameter: Log(Γα)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(d)
10-1
101
103
105
107
0
0.5
1
1.5
Elastic restraint parameter: Log(Γα)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(d)
10-1
101
103
105
107
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Elastic restraint parameter: Log(Γα)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(d)
10-1
101
103
105
107
0
0.5
1
1.5
2
2.5
3
3.5x 10
-3
Elastic restraint parameter: Log(Γβ)
Freq
uenc
y pa
ram
eter
: ΔΩ
1st mode3rd mode5th mode
(e)
10-1
101
103
105
107
0
1
2
3
4
5
6
x 10-3
Elastic restraint parameter: Log(Γβ)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(e)
10-1
101
103
105
107
0
2
4
6
8
10
12
14x 10
-4
Elastic restraint parameter: Log(Γβ)
Fre
quen
cy p
aram
eter
: ΔΩ
1st mode3rd mode5th mode
(e)
Fig. 4 Relationships of frequency parameters �Ω with elastic restraint parameter Γλ for certain three-layered, cross-ply[0◦/90◦/0◦] open shells with C–F–Fe–F boundary condition: a Γu ; b Γv ; c Γw; d Γα ; e Γβ
Arbitrary boundary conditions 453
(a) F edge: Nα = 0, Nαβ = 0, Qα = 0,Mα = 0,Mαβ = 0 or kuα0 = kvα0 = kwα0 = K α
α0 = K βα0 = 0
(b) SD edge: Nα = 0, v = 0, w = 0,Mα = 0,Mαβ = 0 or kvα0 = kwα0 = 107 D, kuα0 = K α
α0 = K βα0 = 0
(c) S edge: u = 0, v = 0, w = 0,Mα = 0, φβ = 0 or kvα0 = kwα0 = kuα0 = K β
α0 = 107 D, K αα0 = 0
(d) C edge: u = 0, v = 0, w = 0, φα = 0, φβ = 0 or kuα0 = kvα0 = kwα0 = K α
α0 = K βα0 = 107 D
(e) E1 edge: only transverse direction is elastically restrained, i.e.:
kwα0 = 10D, kuα0 = kvα0 = K α
α0 = K βα0 = 107 D
(f) E2 edge: only rotation is elastically restrained, i.e.:
K αα0 = D, ku
α0 = kvα0 = kwα0 = K βα0 = 107 D
(g) E3 edge: transverse direction and rotation are elastically restrained, i.e.:
kwα0 = 10D, K αα0 = D, ku
α0 = kvα0 = K βα0 = 107 D
where D = E1h3/12(1−μ12μ21) is the flexural stiffness of the shell. The appropriateness of defining theaforementioned seven types of boundary conditions in terms of boundary spring components will be shownlater. In order to easily refer to the boundary conditions of an open shell, a clockwise notation starting formα = 0 is employed. For instance, the symbol “F–C–S–E1” represents an open shell free at edge α = 0,clamped at edge β = 0, simply supported at edge α = Lα and elastically constrained at edge β = Lβ .
3.2.1 Open cylindrical and conical shells with arbitrary boundary conditions
As the first case, Table 4 shows the comparison of the non-dimensional frequency parameters Ω =ωL2
α
√ρ/E1 Rh of a two-layered, cross-ply [90◦/0◦]open cylindrical shell subjected to SD–SD–SD–SD bound-
ary conditions, with results provided by Messina and Soldatos [41] based on the conjunction of Ritz methodand the Love-type version of a unified shear deformable shell theory. The shell parameters used in the com-parison are as follows: E1 = 10 GPa , E2 = 25E2, μ12 = 0.25,G12 = G13 = 0.5E2,G23 = 0.2E2, ρ =1700 kg/m 3, R = 5 m , L/R = 5, θ0 = 60◦. The first eight frequencies and four sets of thickness radiusratios, i.e., h/R = 0.1, 0.05, 0.02 and 0.01, are performed in the comparison. It is clearly evident that thepresent solutions are generally in good agreement with the reference results, although a different shell theorywas employed by Messina and Soldatos [41]. The differences between these two results are very small and donot exceed 0.74% for the worst case. In order to further verify the present formulation and validate the accuracyof the proposed Chebyshev–Ritz method, in Table 5, the present solutions for dimensionless frequency param-etersΩ = ωL2
α
√ρh/D of certain four-layered, symmetrically laminated composite open cylindrical shells are
compared with the results reported by Bardell et al. [39] by using the h− p version of the finite element method.The geometric and material constants of the shells are the following: E1 = 10 GPa , E2 = 15.4E2, μ12 =0.3,G12 = G13 = G23 = 0.8E2, ρ = 1,500 kg/m 3, R = 5 m , R/L = 5, h/L = 0.01, θ0 = 11.5◦. Fivelamination schemes, i.e., [0◦], [30◦/− 30◦/− 30◦/30◦], [45◦/− 45◦/− 45◦/45◦], [60◦/− 60◦/− 60◦/60◦]and [90◦/−90◦/−90◦/90◦] are conducted in the comparison. It can be seen that a good agreement is obtainedbetween the current solutions and those given by Bardell et al. [39]. The open cylindrical shells can be regardedas the special cases of the open conical shells. For the general composite open conical shells, there are nosuitable results for comparison in the literature. Therefore, to ensure the accuracy of the proposed Cheby-shev–Ritz formulation, the illustrative examples are solved for isotropic conical panels. The comparison ofthe lowest six frequency parameters Ω = ωL2
α
√ρh/D for an isotropic open conical shell with S–S–S–S and
C–C–C–C boundary conditions is given in Table 6. The shell parameters used in the comparison are as follows:E = 70 GPa , μ12 = 0.3, ρ = 1,500 kg/m 3, R0 = 1 m , L/R0 = 2, h/R0 = 0.006, θ0 = 60◦, ϕc = 90◦.The solutions are compared with the results reported by Bardell et al. [11] by using the hp-FEM as well as those
454 T. Ye et al.
Table 4 Comparison of frequency parameters Ω = ωL2α
√ρ/E1 Rh of an unsymmetrically laminated, two-layered open cylin-
drical shell with SD–SD–SD–SD boundary conditions (L/R = 5, θ0 = 60◦)
Lamination schemes h/R Method Mode number
1 2 3 4 5 6 7 8
[90◦/0◦] 0.10 Present 7.025 7.645 8.882 10.89 13.52 14.05 16.66 20.27Ref. [41] 7.025 7.702 8.932 10.94 13.57 14.05 16.74 20.40Difference (%) 0.00 0.74 0.56 0.46 0.37 0.00 0.48 0.64
0.05 Present 5.922 8.145 9.934 11.12 14.44 18.02 19.87 21.83Ref. [41] 5.932 8.153 9.935 11.12 14.45 18.03 19.87 21.87Difference (%) 0.17 0.10 0.01 0.00 0.07 0.06 0.00 0.18
0.02 Present 5.049 9.773 14.84 15.37 15.71 15.96 17.11 18.75Ref. [41] 5.049 9.774 14.84 15.40 15.71 16.01 17.13 18.77Difference (%) 0.00 0.01 0.00 0.19 0.00 0.31 0.12 0.11
0.01 Present 5.724 11.03 12.27 13.03 14.62 17.60 20.31 20.92Ref. [41] 5.724 11.03 12.27 13.03 14.62 17.61 20.31 20.95Difference (%) 0.00 0.00 0.00 0.00 0.00 0.06 0.00 0.14
obtained from the finite element analyses (ANSYS). It is observed that the present analysis agree well withthe reference. The difference between these two results is very small and is less than 1.12 % for the worst case.
The excellent agreement between the present results and the referential ones obtained in Tables 1–2 andTables 4–6 indicates that the proposed Chebyshev–Ritz method is able to calculate the vibration frequenciesof the composite laminated open cylindrical and conical shells with sufficient accuracy. It also verified thatthe defining of the four types of classical boundaries at the beginning of this subsection is appropriate. Havinggained confidence in the present method, some further numerical results for the composite laminated opencylindrical and conical shells with different sets of classical, elastic boundary conditions and their combina-tions as well as different shell parameters, such as geometric properties, lamination schemes, are given in thefollowing discussions.
Tables 7 and 8 give the first five non-dimensional frequenciesΩ = ωL2α
√ρh/D of three-layered, cross-ply
[0◦/90◦/0◦] open cylindrical and conical shells with various boundary conditions and subtended angles. Thematerial properties and geometric dimensions of the shells are as follows: E1 = 10 GPa , E2 = 15E2, μ12 =0.25,G12 = G13 = G23 = 0.5E2,ρ = 1,500 kg/m 3; for the cylindrical panel: R = 1 m , L/R = 2, h/R =0.1; for the conical panel: R0 = 1 m , L/R0 = 2, h/R0 = 0.1, ϕc = 45◦. These results may serve as bench-mark values for future researches. It is obvious from the tables that the increase in the subtended angle willresult in an increase in the natural frequencies of the shells. Meanwhile, we can see that an open shell withgreater restraining rigidity will have higher natural frequencies. As aforementioned, once the coefficients inthe admissible displacement expressions of an open shell are solved from Eq. (22), its mode shapes can bedirectly determined. To enhance the understanding of the vibration behaviors of the open cylindrical and con-ical shells, the first five mode shapes for the shells with C–C–C–C restraint conditions are present in Figs. 5and 6, which is constructed in three-dimension views. Next, Table 9 compares the lowest five dimensionlessfrequenciesΩ = ωL2
α
√ρh/D of a two-layered, cross-ply [0◦/90◦] open conical shell with various subtended
angles and cone angles. The shell subjected to SD–SD–SD–SD, S–S–S–S and C–C–C–C boundary conditionsare considered in the investigation. The material and geometric constants of the shell are the same as theaforementioned three-layered conical panel (Table 8). The results show that the C–C–C–C conical panel hasthe higher, frequency values than the SD–SD–SD–SD one. In addition, the increase in the subtended angleand cone angle will result in an increase in the natural frequencies.
One of the main advantages of the composite laminated structures is that desired mechanical parameters canbe obtained by appropriately selecting their lamination schemes, such as fiber orientations, number of layersand stacking sequences. Therefore, it is of particular importance to understand the effects of lamination schemeon the vibration characteristics of a composite laminated open shell. Figure 7 depicts the variation in the firstthree dimensionless frequenciesΩ = ωL2
α
√ρh/D of a full clamped, three-layered composite open cylindrical
shell with varying fiber orientations in the middle layer, while the top and bottom layers of the shell are ofuniform principal direction which is paralleled to α-axis (denoted by [0◦/ϑ/0◦]). The shell parameters are thesame as the aforementioned three-layered, cross-ply [0◦/90◦/0◦] open cylindrical shell employed in Table 7.Four subtended angle configurations, i.e., θ0 = 60◦, 90◦, 120◦, 150◦ are considered in the study. As it can beseen, the effects of the fiber orientations on the frequency parameters varied with the subtended angles. Whenθ0 = 60◦ and θ0 = 90◦, the frequency parameters Ω increase steadily as ϑ becomes larger. However, when
Arbitrary boundary conditions 455
Tabl
e5
Com
pari
son
offr
eque
ncy
para
met
ersΩ
=ω
L2 α√ ρ
h/
Dof
four
-lay
ered
open
cylin
dric
alsh
ells
with
C–C
–C–C
boun
dary
cond
ition
s(R
=5
m,
R/
L=
5,h/
L=
0.01,
θ 0=
11.5
◦ )
Mod
e[0◦
][30
◦ /−
30◦ /
−30◦/30
◦ ][45
◦ /−4
5◦/
−45◦/45
◦ ][60
◦ /−
60◦ /
−60
◦ /60
◦ ][90
◦ /−
90◦ /
−90
◦ /90
◦ ]Pr
esen
tR
ef.[
39]
Pres
ent
Ref
.[39
]Pr
esen
tR
ef.[
39]
Pres
ent
Ref
.[39
]Pr
esen
tR
ef.[
39]
128
.13
28.2
731
.43
31.4
938
.84
38.8
948
.81
48.8
760
.89
60.9
82
31.0
831
.20
38.7
938
.83
46.1
646
.26
52.6
852
.89
63.2
463
.88
342
.30
42.6
156
.00
56.3
755
.55
55.7
059
.88
60.1
764
.28
64.2
04
60.0
260
.55
57.2
157
.61
65.9
766
.38
67.5
467
.86
70.1
170
.77
564
.74
65.0
470
.44
71.0
977
.87
78.6
677
.18
77.7
471
.36
71.5
06
68.0
768
.76
78.4
579
.57
88.0
089
.02
87.1
487
.90
81.5
182
.21
456 T. Ye et al.
Tabl
e6
Com
pari
son
offr
eque
ncy
para
met
ersΩ
=ω
L2 α√ ρ
h/
Dof
anis
otro
pic
open
coni
cals
hell
with
S–S–
S–S
and
C–C
–C–C
boun
dary
cond
ition
s(R
0=
1m,
L/
R0
=2,
h/
R0
=0.
006,θ 0
=60
◦ ,ϕ
c=
90◦ )
Mod
eS–
S–S–
SC
–C–C
–C
Pres
ent
hp-F
EM
[11]
Err
or(%
)A
NSY
S[1
1]E
rror
(%)
Pres
ent
hp-F
EM
[11]
Err
or(%
)A
NSY
S[1
1]E
rror
(%)
119.1
019.2
0.52
19.1
0.00
35.0
935.2
0.32
35.0
0.25
243.6
343.6
0.06
43.5
0.30
66.1
966.3
0.16
66.0
0.29
350.8
150.9
0.18
50.7
0.21
74.5
974.9
0.41
74.4
0.26
478.9
378.9
0.04
78.6
0.42
108.
1810
8.4
0.20
107.
70.
455
83.1
683.3
0.16
82.9
0.32
112.
8511
3.2
0.31
112.
20.
586
100.
7810
1.1
0.32
100.
60.
1813
3.39
134.
91.
1213
3.0
0.29
Arbitrary boundary conditions 457
Tabl
e7
The
low
estfi
vefr
eque
ncy
para
met
ersΩ
=ω
L2 α√ ρ
h/
Dof
ath
ree-
laye
red,
cros
s-pl
y[0◦/90
◦ /0◦
]ope
ncy
lindr
ical
shel
lwith
vari
ous
boun
dary
cond
ition
san
dsu
bten
ded
angl
es(R
=1
m,
L/
R=
2,h/
R=
0.1)
Subt
ende
dan
gle
(θ0)
Mod
eB
ound
ary
cond
ition
s
S–S–
S–S
C–C
–C–C
E1–
E1–
E1–
E1
E2–
E2–
E2–
E2
E3–
E3–
E3–
E3
F–E
1–F
–E
1F–
E2–F
–E
2F-
E3–F
–E
3SD
-E1–S
D–
E1
SD-E
2–S
D–
E2
45◦
173.2
910
83.2
629
38.2
183
80.2
406
36.9
915
32.2
969
78.7
490
31.8
954
37.8
709
79.8
828
278.0
822
93.8
046
51.6
517
85.9
480
50.9
290
37.1
930
79.2
152
36.8
846
56.4
151
85.5
810
384.8
691
101.
206
70.2
437
88.4
240
69.9
175
52.4
520
83.3
029
52.2
645
78.1
625
87.5
019
485.8
746
111.
677
78.4
578
96.1
813
71.0
621
75.5
263
83.6
009
68.1
201
85.3
006
95.3
444
510
4.99
611
3.75
085.4
237
107.
406
80.0
494
77.4
524
85.0
636
70.6
430
88.0
511
100.
979
90◦
125.9
226
36.2
073
27.8
660
30.5
844
25.8
649
15.9
876
22.8
582
15.3
691
25.8
452
29.5
411
241.7
629
51.4
391
32.2
137
46.0
887
30.7
796
23.6
851
27.2
626
23.0
811
30.7
598
45.3
682
348.8
777
59.0
643
46.4
152
52.0
340
45.6
829
25.5
137
42.2
955
25.3
875
49.6
361
50.4
902
457.3
390
68.3
664
47.1
679
61.0
707
46.1
159
27.8
199
42.3
096
27.7
377
50.0
194
50.6
410
573.3
012
83.4
734
50.4
757
79.6
108
48.4
787
43.7
127
44.4
612
43.1
826
50.0
733
59.7
677
135◦
123.5
462
28.4
288
25.3
762
24.8
346
23.3
236
8.90
282
9.59
418
8.10
483
22.7
972
23.6
472
223.9
800
31.7
386
26.3
539
26.9
701
24.3
611
14.7
709
18.9
623
14.5
617
23.8
653
25.7
078
336.7
580
46.0
048
32.8
598
40.9
773
31.6
645
15.6
516
19.7
151
15.4
631
31.6
216
33.6
603
446.0
628
55.3
353
41.5
853
48.4
516
40.1
247
19.1
403
20.0
410
18.8
048
33.6
603
40.2
035
548.5
158
55.6
624
43.1
197
49.8
557
42.1
916
24.4
959
33.4
695
24.4
958
40.6
088
46.8
978
180◦
120.7
458
26.7
045
23.9
731
22.4
267
21.7
196
4.95
173
4.81
124
4.43
420
20.9
804
20.9
466
222.7
066
27.8
521
24.2
230
24.1
403
22.1
716
9.43
695
10.7
630
9.10
214
21.3
473
22.8
068
328.2
867
33.3
990
30.4
373
30.0
170
28.9
710
9.54
649
10.7
902
9.11
705
25.2
452
25.2
452
432.5
518
39.8
891
30.8
290
35.5
486
29.3
558
14.1
193
15.9
960
14.1
171
28.6
804
29.1
538
544.4
454
52.1
301
38.1
696
46.8
712
37.0
584
16.0
470
19.9
382
15.7
937
29.2
386
34.6
057
215◦
120.1
659
25.9
464
23.1
276
21.7
914
20.8
369
2.85
403
2.66
187
2.55
135
19.9
462
20.1
962
220.9
320
26.1
552
23.3
816
21.9
780
21.2
288
5.78
266
6.22
486
5.50
721
20.1
962
20.2
222
324.8
366
31.2
471
27.1
244
27.1
241
25.2
318
6.27
179
6.46
487
5.86
619
20.2
662
20.5
008
428.4
455
31.6
396
28.5
512
29.1
824
26.9
505
10.1
379
11.4
349
10.0
328
24.8
487
25.9
197
532.3
505
38.3
294
33.3
688
34.6
933
32.3
519
11.0
966
12.3
305
10.6
821
26.4
381
28.3
208
458 T. Ye et al.
Tabl
e8
The
low
est
five
freq
uenc
ypa
ram
eter
sΩ=ω
L2 α√ ρ
h/
Dof
ath
ree-
laye
red,
cros
s-pl
y[0◦/90
◦ /0◦
]ope
nco
nica
lsh
ell
with
vari
ous
boun
dary
cond
ition
san
dsu
bten
ded
angl
es(R
0=
1m,
L/
R0
=2,
h/
R0
=0.
1,ϕ
c=
45◦ )
Subt
ende
dan
gle
(θ0)
Mod
eB
ound
ary
cond
ition
s
S–S–
S–S
C–C
–C–C
E1−
E1–
E1–
E1
E2–
E2–
E2–
E2
E3–
E3–
E3−
E3
F-E
1–F
–E
1F-
E2–F
–E
2F-
E3–F
–E
3SD
-E1–S
D–
E1
SD-E
2–S
D–
E2
45◦
129.9
531
37.8
306
28.8
198
34.1
433
27.4
694
20.8
519
22.8
218
19.9
823
28.7
481
30.9
805
232.5
418
44.1
757
31.4
313
37.6
354
30.8
347
22.2
939
23.6
698
22.2
018
31.2
546
36.4
947
350.0
048
58.3
630
39.4
802
52.2
366
38.8
227
29.9
684
35.2
348
29.7
024
44.4
719
50.1
424
452.4
097
67.4
064
42.8
628
59.4
472
41.4
628
35.5
898
42.1
113
35.0
797
47.3
355
58.3
983
554.1
443
70.3
470
45.0
697
62.7
026
44.0
163
36.7
203
44.7
549
36.2
172
50.9
200
60.7
600
90◦
116.7
023
24.1
034
19.5
931
19.1
932
17.7
335
7.24
791
7.25
426
6.91
093
17.8
157
17.7
056
219.8
104
27.6
709
21.2
025
22.8
720
19.5
758
10.4
828
11.0
037
10.2
467
19.3
284
20.4
278
329.9
098
37.8
984
28.9
555
34.3
015
27.8
802
17.4
350
18.3
831
16.9
269
25.2
241
25.4
229
433.3
470
39.2
989
31.3
100
34.4
347
30.8
400
18.5
249
18.7
199
18.4
149
29.3
059
33.1
349
540.4
619
49.8
597
33.0
667
42.6
416
32.4
185
19.7
646
20.6
545
19.2
693
36.4
992
40.8
596
135◦
115.2
810
22.5
655
17.9
102
17.4
019
15.9
187
3.48
686
3.45
428
3.30
912
15.6
272
15.6
413
216.5
357
22.8
877
18.3
448
18.0
846
16.5
362
5.17
746
5.21
019
5.03
296
16.2
094
16.4
474
319.9
434
26.7
792
22.2
250
22.3
346
20.7
278
9.59
960
9.65
161
9.31
237
18.8
795
18.8
827
422.9
775
30.1
860
23.5
819
25.7
042
22.2
468
10.5
397
11.3
673
10.3
621
20.9
632
20.9
373
529.8
987
37.7
534
29.1
453
33.2
195
28.1
523
13.5
280
14.3
986
13.1
983
23.8
348
25.1
830
180◦
114.8
195
21.8
554
17.3
066
16.7
122
15.2
739
1.88
869
1.83
388
1.78
656
14.9
136
14.9
495
215.1
287
21.9
830
17.3
403
16.7
362
15.3
764
2.86
904
2.85
679
2.79
299
14.9
552
14.9
825
317.4
443
24.5
192
19.7
835
19.7
118
18.0
272
5.61
851
5.56
669
5.46
221
16.3
091
16.3
458
419.3
675
25.0
376
20.5
301
20.8
068
19.0
260
6.04
017
6.42
453
5.85
254
18.5
472
18.5
567
522.3
378
28.5
277
24.0
125
24.3
908
22.7
765
8.40
367
8.50
183
8.16
553
18.6
167
18.8
810
215◦
114.5
709
21.5
261
16.9
448
16.2
410
14.9
048
1.09
833
1.05
464
1.03
971
14.4
567
14.4
500
214.5
881
21.6
674
17.0
018
16.3
805
14.9
487
1.71
266
1.70
278
1.67
007
14.5
267
14.5
651
316.4
698
23.0
782
18.7
185
18.3
505
16.8
709
3.55
957
3.52
774
3.41
199
15.1
623
15.1
651
416.9
153
23.4
311
18.7
862
18.4
918
17.0
601
3.57
925
3.68
648
3.48
758
16.6
180
16.7
398
519.3
833
26.1
952
21.3
274
21.6
336
19.7
653
5.53
400
5.50
461
5.38
513
16.8
430
16.8
847
Arbitrary boundary conditions 459
Fig. 5 Mode shapes for the open composite cylindrical shell with C–C–C–C boundary conditions
Fig. 6 Mode shapes for the composite open conical shell with C–C–C–C boundary conditions
θ0 = 150◦, the frequency parameter traces climb up and then decline, and reach their crests around ϑ = 45◦.Figure 8 gives the variations in the lowest three dimensionless frequencies Ω = ωL2
α
√ρh/D of a [0◦/ϑ]n
layered conical panel with C–C–C–C boundary conditions against the number of layers n (where n=1 meansthe single-layered scheme [0◦]; sn = 2 means the two-layered scheme [0◦/ϑ], and so forth). Four types of fiberorientations, i.e., ϑ = 30◦, 45◦, 60◦, 90◦, are considered. The layers of the conical shells are the same as the
460 T. Ye et al.
0° 15° 30° 45° 60° 75° 90°
40
50
60
70
80
Fiber orientation: ϑ
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd0° 15° 30° 45° 60° 75° 90°
40
50
60
70
80
90
100
110
Fiber orientation: ϑ
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd
(b) θ0=90°(a) θ0=60°
0° 15° 30° 45° 60° 75° 90°25
30
35
40
45
50
55
Fiber orientation: ϑ
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd0° 15° 30° 45° 60° 75° 90°
25
30
35
40
Fiber orientation: ϑ
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd
(c) θ0=120° (d) θ0=150°
Fig. 7 Relationships of frequency parameters Ω = ωL2α
√ρh/D with fiber orientations for a three-layered [0◦/ϑ/0◦] open
cylindrical shell with C–C–C–C boundary conditions
5 10 15 20 25 30
22
24
26
28
30
32
Layer number: n
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd5 10 15 20 25 30
25
30
35
Layer number: n
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd
(a) ϑ=30° (b) ϑ=45°
5 10 15 20 25 30
25
30
35
40
Layer number: n
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd
5 10 15 20 25 30
25
30
35
40
45
Layer number: n
Fre
quen
cy p
aram
eter
: Ω
1st 2nd 3rd
(c) ϑ=60° (d) ϑ=90°
Fig. 8 Relationships of frequency parametersΩ = ωL2α
√ρh/D with number of layers n for a [0/ϑ]n layered conical shell with
C–C–C–C restraints: a ϑ = 30◦; b ϑ = 45◦; c ϑ = 60◦; d ϑ = 90◦
aforementioned open conical shell used for Tables 8 and 9. The results demonstrate that the frequency parame-ters increase rapidly and may reach their crests around n = 8, and beyond this range, the frequency parametersremain unchanged. The fluctuations on the curves are due to the fact that the shells are symmetrically laminatedwhen n is an odd number, and the shells are unsymmetrically laminated when n equal to an even number.
Arbitrary boundary conditions 461
Tabl
e9
The
low
est
five
freq
uenc
ypa
ram
eter
sΩ
=ω
L2 α√ ρ
h/
Dof
atw
o-la
yere
d,cr
oss-
ply
[0◦/90
◦ ]op
enco
nica
lsh
ell
with
vari
ous
subt
ende
dan
gles
and
cone
angl
es(R
0=
1m,
L/
R0
=2,
h/
R0
=0.
1)
Con
ean
gle
(ϕc)
Mod
eθ 0
=90
◦θ 0
=13
5◦θ 0
=18
0◦
SD-S
D-
SD-S
DS–
S–S–
SC
–C–C
–CSD
-SD
-SD
-SD
S–S–
S–S
C–C
–C–C
SD-S
D-
SD-S
DS–
S–S–
SC
–C–C
–C
15◦
117
.167
624
.767
830
.438
815
.219
521
.887
823
.640
915
.413
519
.496
021
.220
72
19.6
945
34.4
967
42.9
727
19.7
142
22.4
416
25.8
260
17.1
750
20.5
506
22.6
686
332
.313
840
.331
546
.056
522
.388
432
.857
238
.692
319
.725
627
.435
728
.830
34
35.7
906
45.4
140
54.7
992
27.1
456
35.5
398
41.3
657
20.5
974
28.7
384
33.2
352
536
.909
360
.425
467
.077
730
.458
137
.922
442
.963
826
.666
834
.018
039
.516
530
◦1
15.9
104
21.1
735
25.0
914
13.8
939
18.5
233
21.1
854
13.6
790
17.3
619
18.9
734
216
.112
926
.509
733
.130
115
.925
819
.750
021
.571
315
.934
417
.555
519
.810
93
27.1
790
35.5
128
41.4
120
20.1
581
26.6
422
30.7
187
16.1
190
22.9
041
25.4
230
429
.331
738
.831
246
.995
222
.890
331
.318
737
.480
118
.203
024
.130
726
.459
65
32.0
561
44.9
850
52.4
736
24.9
739
32.8
993
38.4
122
20.9
119
30.1
536
34.0
252
45◦
113
.596
518
.185
421
.659
612
.248
315
.931
618
.666
212
.022
515
.055
516
.924
22
14.0
524
21.6
759
27.3
025
13.6
097
16.8
527
18.8
354
13.6
174
15.1
369
17.5
040
322
.146
931
.323
737
.806
716
.481
122
.590
026
.098
814
.057
819
.289
022
.121
44
26.9
242
34.1
011
39.3
996
18.8
779
25.8
526
31.0
656
15.8
363
20.3
512
22.5
772
527
.735
335
.543
441
.803
225
.141
328
.932
535
.211
917
.372
225
.403
428
.695
460
◦1
11.0
772
15.3
239
19.0
842
10.1
543
13.5
156
15.9
926
10.1
683
12.5
006
14.8
612
211
.871
717
.772
222
.429
411
.883
813
.787
616
.718
911
.082
912
.804
515
.404
33
19.3
125
26.1
572
30.1
185
12.3
946
19.2
306
23.0
050
11.8
912
16.4
014
18.5
954
423
.835
827
.845
934
.900
316
.516
120
.868
325
.094
412
.455
916
.466
819
.951
05
24.8
777
30.0
418
36.3
806
22.3
716
25.6
350
30.4
713
15.2
166
21.6
778
25.2
826
75◦
17.
8891
512
.509
415
.821
78.
0182
610
.424
413
.460
47.
8966
59.
8272
012
.871
82
10.6
232
12.7
665
17.2
753
8.27
453
10.8
437
14.4
250
8.40
074
9.94
901
12.9
913
317
.793
320
.495
025
.713
910
.635
114
.138
917
.428
88.
4314
512
.518
615
.904
04
20.5
913
24.8
100
30.7
926
15.1
784
16.8
320
21.1
273
10.6
423
12.7
082
15.9
064
523
.445
425
.139
632
.863
920
.620
422
.221
827
.182
213
.944
715
.951
819
.437
4
462 T. Ye et al.
3.2.2 Open spherical shells with arbitrary boundary conditions
In the following numerical examples, the present formulation is applied to the composite laminated openspherical shells with general restraints for which very limited amount of results are available in the litera-ture. Since there are no suitable comparison results in the literature, two illustrative examples are presentedfor the vibrations of composite laminated shallow spherical panels. As the first case, Table 10 shows thecomparison of the first nine non-dimensional frequencies Ω = ωL2
α/h√ρ/E1 of certain three-layered shal-
low spherical shells subjected to SD–SD–SD–SD boundary conditions, with results provided by Fazzolariand Carrera [12] by using the Carrera unified formulation. The shell parameters used in the comparisonare the following: E1 = 60.7 GPa , E2 = 24.8 GPa , μ12 = 0.23,G12 = G13 = G23 = 12 GPa , ρ =1700 kg/m 3, R = 2 m , h = 0.05 m , ϕ0 = 75.5◦, ϕ1 = 104.5◦, θ0 = 28.6◦. Four sets of lamination schemes,i.e., [0◦], [15◦/ − 15◦/15◦], [30◦/ − 30◦/30◦] and [45◦/ − 45◦/45◦], are performed in the comparison. It isclearly evident that the present solutions match well with the reference data, although a different shell theoryis used by Fazzolari and Carrera. The differences between these two results are attributed to different shelltheories used in the literature. It has been proven that the shallow shell theories will give inaccurate results forthe lower frequencies when applied to deep open shells [8]. In order to further verify the present formulation,in Table 11, the first six natural frequencies of a four-layered, angle-ply [45◦/− 45◦/45◦/− 45◦] open spher-ical shell with as many as eight combinations of classical restraints are presented, together with the modifiedFourier series solutions obtained by Ye et al.[28]. The shell parameters used in the comparison are as follows:E1 = 10GPa, E2 = 15 GPa , μ12 = 0.23,G12 = G13 = G23 = 0.5E2, ρ = 1700 kg/m 3, R = 10 m , h =0.01 m , ϕ0 = 87.1◦, ϕ1 = 92.9◦, θ0 = 11.5◦. Our results are in closed agreement with the existing ones.
The excellent agreement between the present results and the referential ones shown in Table 3 and Tables 10–11 indicates that the proposed Chebyshev–Ritz formulation is accurate. Having gained confidence in pres-ent formulation, some further vibration results for the composite laminated deep open spherical shells withvarious combinations of boundary conditions and shell parameters are presented. In the following exam-ples, unless otherwise stated, the geometric and material constants of the spherical shells are as follows:E2 = 10 GPa , E1/E2 = 15, μ12 = 0.25,G12 = G13 = 0.5E2,G23 = 0.2E2, ρ = 1,500 kg/m 3, R =1 m , h = 0.1 m , ϕ0 = 30◦, ϕ1 = 90◦. In Table 12, the first five frequency parameters Ω = ωL2
α/h√ρ/E1
of a three-layered, cross-ply [0◦/90◦/0◦] open spherical shell with various boundary conditions and subtendedangles are presented. It is obvious form the table that the increase of the subtended angle will result in anincrease in the natural frequencies of the shell. Meanwhile, we can see that an open spherical shell with greaterrestraining rigidity will have higher vibration frequencies. For any given frequency parameters, the correspond-ing mode shapes of the open spherical shell can be readily determined by Eq. (19) after solving the standardmatrix eigenproblem [see Eq. (22)]. For instance, the first four mode shapes for the spherical shell subjectedto C–C–C–C boundary conditions and with subtended angle θ0 = 45◦, 90◦ and 135◦ are plotted in Fig. 9.
4 Conclusions
The free vibration analysis of generally supported composite laminated deep open shells with various shellcurvatures including cylindrical, conical and spherical ones has been carried out by using the unified Cheby-shev–Ritz formulation. The first-order shear deformation shell theory based on the general shell equations isemployed in the present analysis. Under the current framework, each of displacements and rotations of the openshell, regardless of boundary conditions, is invariantly expressed as Chebyshev polynomials of first kind inboth directions. Thereby, all the Chebyshev expansion coefficients are treated equally and independently as thegeneralized coordinates and solved directly by using the Rayleigh–Ritz procedure. The convergence and accu-racy of the present formulation are checked by a considerable number of convergence tests and comparisons.A variety of numerical examples are presented for the free vibration of shallow and deep composite laminatedcylindrical, conical and spherical panels with various geometric dimensions and lamination schemes. Differ-ent combinations of classical boundary conditions (e.g., completely free, shear-diaphragm supported, simplysupported and clamped) and uniform elastic restraints are considered. These results may serve as benchmarksolutions for future researches. Parametric studies are also undertaken, giving insight into the effects of elasticrestraint parameter, fiber orientation, layer number, subtended angle as well as conical angle on the vibrationfrequencies of the composite open shells.
It should be stressed that the presented formulation is general since various boundary conditions, differentlamination schemes and geometric dimensions (shallow and deep), different numbers of layers (which may beisotropic or orthotropic) and forms of trial functions, such as the Chebyshev orthogonal polynomials of second
Arbitrary boundary conditions 463
Tabl
e10
Com
pari
son
offr
eque
ncy
para
met
ersΩ
=ω
L2 α/
h√ ρ
/E
1of
cert
ain
thre
e-la
yere
dop
ensp
heri
cals
hells
with
SD–S
D–S
D–S
Dbo
unda
ryco
nditi
ons(
R=
2m,
h=
0.05
m,ϕ
0=
75.5
◦ ,ϕ
1=
104.
5◦,θ 0
=28.6
◦ )
Lay
out
The
ory
Mod
enu
mbe
r
12
34
56
78
9
[0◦]
Pres
ent
8.14
234
12.3
699
13.9
511
18.1
169
20.0
452
25.8
037
26.2
185
27.4
026
28.2
361
Shal
low
[12]
8.15
995
12.4
546
14.0
268
18.3
345
20.1
208
26.0
524
15◦ /
−15
◦ /15
◦ ]Pr
esen
t8.
4670
412
.634
613
.906
018
.305
420
.444
125
.677
626
.149
929
.205
729
.999
4Sh
allo
w[1
2]8.
4908
312
.750
314
.008
318
.561
820
.570
325
.824
430
◦ /−
30◦ /
30◦ ]
Pres
ent
9.09
300
12.9
552
13.8
470
18.6
620
21.3
978
24.3
233
26.6
959
30.0
401
32.5
608
Shal
low
[12]
9.14
088
13.1
156
14.0
273
18.9
934
21.6
139
24.5
616
[45◦ /
−45
◦ /45
◦ ]Pr
esen
t9.
3961
412
.994
513
.846
318
.834
422
.425
223
.058
926
.881
830
.110
934
.187
7Sh
allo
w[1
2]9.
4812
513
.163
514
.077
219
.201
122
.669
523
.346
5
464 T. Ye et al.
Tabl
e11
Com
pari
son
offr
eque
ncy
para
met
ersΩ
=ω
L2 α/
h√ ρ
/E
1of
afo
ur-l
ayer
ed,a
ngle
-ply
[45◦ /
−45
◦ ] 2op
ensp
heri
cals
hell
with
vari
ous
rest
rain
ts(R
=10
m,
h=
0.01
m,ϕ
0=
87.1
◦ ,ϕ
1=
92.9
◦ ,θ 0
=11.5
◦ )
The
ory
Mod
eB
ound
ary
cond
ition
s
F–F–
F–F
F–S–
F–S
F–F–
F–C
F–C
–F–C
F–C
–C–C
S–S–
S–S
S–C
–S–C
C–C
–C–C
Shal
low
CST
[28]
12.
3732
7.68
860.
5360
10.1
1912.5
5031.7
9932.3
1035.8
982
5.81
3110.8
953.
0887
11.4
2915.2
2034.1
7834.5
9936.7
003
8.22
6113.3
973.
1118
13.4
2522.3
5134.9
4735.1
1340.6
454
12.1
4116.7
039.
0389
17.7
2926.4
4037.5
9438.4
8742.5
975
14.1
5217.8
309.
4796
20.3
8832.8
5742.4
2844.8
9848.4
696
19.3
6922.3
2816.0
3724.0
3435.3
9443.6
2545.8
5852.5
39Pr
esen
tFSD
T1
2.37
127.
6402
0.53
4710.0
3312.5
2931.7
2732.2
2935.8
272
5.78
5310.8
563.
0688
11.4
0215.1
2534.1
2734.5
4236.6
123
8.16
4613.3
783.
0946
13.4
0322.1
4734.8
8435.0
4340.4
884
12.0
7216.6
298.
9686
17.6
1026.3
3537.5
2138.3
9442.3
615
14.0
6317.7
519.
4149
20.2
2132.5
1942.3
0344.7
4148.1
836
19.2
2422.1
6115.9
1023.9
1135.2
1543.5
0045.6
6852.2
18
Arbitrary boundary conditions 465
Tabl
e12
The
low
estfi
vefr
eque
ncy
para
met
ersΩ
=ω
L2 α/
h√ ρ
/E
1of
ath
ree-
laye
red,
cros
s-pl
y[0◦/90
◦ /0◦
]ope
nsp
heri
cals
hell
with
vari
ous
boun
dary
cond
ition
san
dsu
bten
ded
angl
es(R
=1
m,
h/
R=
0.01,ϕ
0=
30◦ ,ϕ
1=
90◦ ,)
Subt
ende
dan
gle
(θ0)
Mod
eB
ound
ary
cond
ition
s
S–S–
S–S
C–C
–C–C
E1–
E1–
E1–
E1
E2–
E2–
E2–
E2
E3–
E3–
E3–
E3
F-E
1–F
–E
1F-
E2–F
–E
2F-
E3–F
–E
3SD
-E1–S
D–
E1
SD-E
2–S
D–
E2
45◦
19.
7199
110.6
160
7.75
043
9.97
924
7.53
871
4.84
501
6.10
886
4.83
664
6.99
260
6.90
766
210.4
420
11.0
496
8.47
375
10.6
548
8.40
413
6.20
637
7.00
090
5.96
863
7.35
915
9.52
070
310.5
030
11.9
162
9.28
904
11.0
468
9.00
117
7.05
572
7.34
667
6.80
885
9.66
062
9.79
653
412.3
306
13.6
860
10.2
981
12.8
290
9.48
521
7.15
364
9.51
640
7.06
314
9.77
147
11.6
611
514.7
358
15.5
032
11.0
335
14.9
427
10.8
029
8.33
472
10.4
855
8.24
406
10.7
090
12.5
133
90◦
17.
8795
48.
4745
76.
4939
68.
0607
86.
3505
32.
5075
52.
5391
32.
4010
54.
2894
64.
2734
72
8.61
756
9.23
281
6.53
775
8.76
147
6.43
753
2.92
650
3.30
916
2.92
217
5.06
022
5.15
764
38.
8602
39.
5547
07.
9947
19.
0546
87.
7893
14.
2096
64.
3612
04.
1779
66.
1819
06.
7287
74
9.19
082
9.60
484
8.25
598
9.29
629
7.87
940
4.24
592
4.86
417
4.20
653
7.90
007
8.47
822
510.1
869
10.7
409
8.81
178
10.3
5091
8.30
278
5.28
520
6.34
762
5.28
191
8 .54
197
8.69
632
135◦
17.
3429
17.
7322
86.
0025
27.
4461
25.
8907
41.
4068
21.
3856
11.
3435
73.
2342
43.
2322
32
7.87
965
8.41
371
6.03
433
8.01
320
5.93
277
1.64
706
1.78
815
1.62
269
4.34
648
4.32
489
38.
3475
28.
9437
06.
9418
78.
4835
46.
7900
32.
7117
92.
7667
02.
6742
04.
8954
34.
9804
74
8.64
326
9.11
783
7.32
004
8.76
981
7.18
292
3.07
157
3.21
054
3.02
063
5.95
350
6.23
238
58.
6815
59.
1473
17.
7453
18.
7737
37.
4041
03.
7105
73.
7390
73.
6502
46 .
4351
96.
4391
918
0◦1
6.95
809
7.22
125
5.81
021
7.02
677
5.70
471
0.84
237
0.82
639
0.80
409
2.72
744
2.72
822
27.
5868
28.
0518
35.
8127
47.
6944
05.
7111
00.
9718
61.
0290
60.
9460
63.
9994
63.
9897
03
7.88
846
8.41
494
6.48
874
8.01
283
6.35
717
1.76
551
1.84
907
1.73
317
4.35
630
4.35
580
48.
3160
48.
8519
86.
5488
88.
4498
66.
4487
92.
0230
42.
0703
91.
9738
24.
9818
55.
0224
15
8.45
045
8.89
242
7.26
833
8.52
388
6.99
451
2.80
979
2.93
758
2.77
912
5 .23
833
5.23
718
215◦
16.
6430
96.
8222
45.
7109
76.
6892
35.
6057
50.
5279
30.
5168
40.
5042
02.
4516
22.
4527
82
7.40
722
7.80
398
5.71
355
7.49
692
5.61
165
0.58
619
0.61
330
0.56
545
3.58
568
3.58
252
37.
6559
38.
1433
66.
1718
87.
7642
56.
0666
01.
1787
01.
2186
11.
1480
54.
1075
94.
0984
74
7.91
656
8.42
219
6.21
044
8.03
469
6.10
606
1.36
431
1.37
908
1.32
782
4.44
239
4.44
621
58.
2346
58.
7466
56.
7609
68.
3541
86.
6250
92.
0341
92.
0595
61.
9938
84 .
9207
64.
9181
5
466 T. Ye et al.
Fig. 9 Mode shapes for the composite open spherical shell with C–C–C–C boundary conditions
kind, the ordinary power polynomials and the Legendre orthogonal polynomials, can be easily accommodatedin the formulation. In addition, it can be readily applied to composite laminated deep open shells with more com-plex boundaries such as point supports, non-uniform elastic restraints, partial supports and their combinations.
Acknowledgments The authors would like to thank the reviewers for their Constructive comments. The authors gratefullyacknowledge the financial support from the National Natural Science Foundation of China (Nos. 51175098 and 51279035).
Appendix 1: Stiffness and mass matrices of laminated open cylindrical and conical shells
This appendix contains the detail expressions of the stiffness (K) and mass (M) matrices of a laminated openconical shell. By setting ϕc = 0, these expressions are applicable to the open cylindrical one. For the sake ofsimplicity, two indexes are predefined:
p = n + (N + 1)m + 1, p′ = n′ + (N + 1)m′ + 1 (A.1)
The elements in these two matrices are calculated according to the expressions given as follows:
{Kp,p′
u,u
}= [A11�
111 + A12s(�010 +�100)+ A22s2�00−1]�00 + A16�
010�10 + A16�100�01
+ A26s�00−1�10 + A26s�00−1�01 + A66�00−1�11 +
[R0ku
x0(−1)m+m′ + R1kux L
]�00
+[kuθ0(−1)n+n′ + ku
θL
]�001
{Kp,p′
u,v
}= A12�
010�10+ A16�111�00− A16s�00−1�10+ A22s�100�00+ A26s�100�00 − A26s2�00−1�00
+ A26�00−1�11 + A66�
100�01 − A66s�00−1�01
{Kp,p′
u,w
}= A12c�010�00 + A22sc�00−1�10 + A26c�00−1�01
Arbitrary boundary conditions 467
{Kp,p′
u,x
}= [B11�
111 + B12s(�010 +�100)+ B22s2�00−1]�00 + B16�
010�10 + B16�100�01
+ B26s�00−1�10 + B26s�00−1�01 + B66�00−1�11
{Kp,p′
u,θ
}= B12�
010�10+B16�111�00−B16s�00−1�10+B22s�100�00+B26s�100�00 − B26s2�00−1�00
+ B26�00−1�11 + B66�
100�01 − B66s�00−1�01
{Kp,p′v,v
}= A22�
00−1�11+ A26[(�100 − s�00−1)�01 + (�010 − s�00−1)]�10 + A44c2�00−1�00
+ A66�111�00 + (s�100 + s�010 + s2�00−1) A66�
00
+[
R0kvx0(−1)m+m′ + R1kvx L
]�00 + [kvθ0(−1)n+n′ + kvθL ]�001
{Kp,p′v,w
}= A22c�00−1�01 + A26c�010�00 − A26sc�00−1�00 − A44c�00−1�10 − A45c�100�00
{Kp,p′v,x
}= −A45c�000�00+B21�
100�01+B22s�00−1�01+B26�00−1�11 + B61�
111�00 − B61s�100�00
+ B62s�010�00 − B62s2�00−1�00 + B66�010�10
{Kp,p′v,θ
}= B22�
00−1�11 + B26[(�100 − s�00−1)�01 + (�010 − s�00−1)]�10 − A44c�000�00
+ B66�111�00 − B66s�00−1�10 + (s�100 + s�010 + s2�00−1) B66�
00
{Kp,p′w,w
}= A22c2�00−1�00 + A44�
00−1�11 + A45s�010�10 + A45s�100�01 + A55�111�00
+[
R0kwx0(−1)m+m′ + R1kwx L
]�00 +
[kwθ0(−1)n+n′ + kwθL
]�001
{Kp,p′w,x
}= A45�
000�01 + A55�011�11 + B21c�100�00 + B22sc�00−1�00 + B26c�00−1�10
{Kp,p′w,θ
}= A44�
000�01 + A45�011�00 + B22c�00−1�10 + B26c�100�00 − B26cs�00−1�00
{K p,p′
x,x
}= [D11�
111 + D12s(�010 +�100)+ D22s2�00−1]�00 + D16(�
010�10 +�100�01)
+ A55�001�00+D26s�00−1 (�10+�01)+ D66�
00−1�11 +[
R0 K xx0(−1)m+m′ + R1 K x
x L
]�00
+[Kxθ0(−1)n+n′ + K x
θL
]�001
{Kp,p′
x,θ
}= D12�
010�10 + D16�111�00 − D16s�00−1�10 + D22s�100�00+D26s�100�00
− D26s2�00−1�00 + D26�00−1�11 + D66�
100�01 − D66s�00−1�01 + A45�001�00
{Kp,p′θ,θ
}= D22�
00−1�11+D26[(�100 − s�00−1)�01+(�010−s�00−1)]�10+ A44�001�00+D66�
111�00
+ (s�100 + s�010 + s2�00−1)D66�00
+[
R0 K θx0(−1)m+m′ + R1 K θ
x L
]�00 +
[K θθ0(−1)n+n′ + K θ
θL
]�001
{Mp,p′
u,u
}={
Mp,p′v,v
}={
Mp,p′w,w
}= I0�
001�00;{
Mp,p′x,x
}={
Mp,p′θ,θ
}= I2�
001�00;{
Mp,p′u,x
}={
Mp,p′v,θ
}= 2I1�
001�00
468 T. Ye et al.
where
�abc =1∫
−1
[2a+b−1
La+b−1 λc daTm(x)
dxadbTm′(x)
dxb
]
dx, �e f =1∫
−1
[2e+ f −1
θe+ f −10
de Pn(θ)
dθe
d f Pn′(θ)
dθf
]
dθ
RL = R0 + sL , λ = sL
2(x + 1)+ R0, s = sin(ϕc), c = cos(ϕc)
Appendix 2: Mass and stiffness matrices of a laminated open spherical shell
This appendix contains the detail expressions of the stiffness (K) and mass (M) matrices for a laminated openspherical shell. The elements of these matrices are calculated by the expressions given as follows:
{Kp,p′
u,u
}= [A11�
11ms + A12
(�01
mc+�10mc
)+ A22�00ct
]�00+ A16�
01m0�
10+ A16�10m0�
01+ A26�00dt
(�10+�01)
+ A66�00ds�
11 + A55�00ms�
00 +[
Rskuϕ0(−1)m+m′ + RLku
ϕL
]�00 + [ku
θ0(−1)n+n′ + kuθL ]�00
ms{
Kp,p′u,v
}= A12�
01m0�
10 + A16�11ms�
00 − A16�01mc�
00 + A22�00dt�
10 + A26�10mc�
00 − A26�00ct �
00
+ A26�00ds�
11 + A66�10m0�
01 − A66�00dt�
01 + A45�00ms�
00
{Kp,p′
u,w
}= [A11�
01ms + A12�
01ms + A12�
00mc + A22�
00mc − A55�
10ms]�00 + A16�
00m0�
01 + A26�00m0�
01
− A45�00m0�
10
{Kp,p′
u,ϕ
}= [B11�
11ms + B12
(�01
mc +�10mc
)+ B22�00ct ]�00 + B16�
01m0�
10 + B16�10m0�
01
+ B26�00dt
(�10 +�01)+ B66�
00ds�
11 + B55�00ms�
00 − A55 R�00ms�
00
{Kp,p′
u,θ
}= B12�
01m0�
10 + B16�11ms�
00 − B16�01mc�
00 + B22�00dt�
10 + B26�10mc�
00 − B26�00ct �
00
+ B26�00ds�
11 + B66�10m0�
01 − B66�00dt�
01 + B45�00ms�
00 − A45 R�00ms�
00
{Kp,p′v,v
}= A22�
00ds�
11 + A26�10m0�
01 − A26�00dt�
01 + A26�01m0�
10 − A26�00dt�
10 + A66�11ms�
00
+ A66�10mc�
00 + A66�00 (�01
mc +�00)+ A44�00ms�
00 +[
Rskvϕ0(−1)m+m′ + RLkvϕL
]�00
+[kvθ0(−1)n+n′ + kvθL ]�00ms
{Kp,p′v,w
}= A12�
00m0�
01 + (A16�01ms − A16�
00mc + A26�
01ms − A26�
00mc − A45�
10ms
)�00 + A22�
00m0�
01
−A44�00m0�
10
{Kp,p′v,ϕ
}= B12�
10m0�
01 + B22�00dt�
01 + B26�00ds�
11 + B16�11ms�
00 − B16�10mc�
00 + B26�01mc�
00
− B26�00ct �
00 + B66�01m0�
10 − B66�00dt�
10 − A45 R�00ms�
00
{Kp,p′v,θ
}= B22�
00ds�
11 + B26�10m0�
01 − B26�00dt�
01 + B26�01m0�
10 − B26�00dt�
10 + B66�11ms�
00
+ B66�10mc�
00 + B66�00 (�01
mc +�00)− A44 R�00ms�
00
{Kp,p′w,w
}= A11�
00ms�
00 + A12�00ms�
00 + A12�00ms�
00 + A22�00ms�
00 + A44�00ds�
11 + A45�01ms�
10
+ A45�10m0�
01 + A55�11ms�
00 + [Rskwϕ0(−1)m+m′ + RLkwϕL ]�00 + [kwθ0(−1)n+n′ + kwθL ]�00ms
{Kp,p′w,ϕ
}= (B11�
10ms + B12�
00ms + B21�
10m0 + B22�
00ms)�
00 +�00mc�
10 (B16 + B26)+ A45 R�00m0�
01
+A55 R�01ms�
00
Arbitrary boundary conditions 469
{Kp,p′w,θ
}= (B16�
10ms − B16�
00mc + B26�
10ms − B26�
00mc
)�00 +�00
m0�10 (B12 + B22)+ A44 R�00
m0�01
+A45 R�01ms�
00
{Kp,p′ϕ,ϕ
}= [D11�
11ms + D12
(�01
mc +�10mc
)+ D22�00ct
]�00 + D16�
01m0�
10 + D16�10m0�
01
+ D26�00dt
(�10 +�01)+ D66�
00ds�
11 + A55 R2�00ms�
00 +[
Rs K ϕϕ0(−1)m+m′ + RL K ϕ
ϕL
]�00
+[Kϕθ0(−1)n+n′ + K ϕ
θL ]�00ms
{Kp,p′ϕ,θ
}= D12�
01m0�
10 + D16�11ms�
00 − D16�01mc�
00 + D22�00dt�
10 + D26�10mc�
00 − D26�00ct �
00
+ D26�00ds�
11 + D66�10m0�
01 − D66�00dt�
01 + A45 R2�00ms�
00
{Kp,p′θ,θ
}= D22�
00ds�
11 + D26(�10
m0 −�00dt
)�01 + D26
(�01
m0 −�00dt
)�10 + D66
(�11
ms +�10mc
)�00
+ A44 R2�00ms�
00 + D66�00 (�01
mc +�00)+[
Rs K ϕϕ0(−1)m+m′ + RL K ϕ
ϕL
]�00
+[
K ϕθ0(−1)n+n′ + K ϕ
θL
]�00
ms{
Mp,p′u,u
}={
Mp,p′v,v
}={
Mp,p′w,w
}= I0�
00ms�
00;{
Mp,p′ϕ,ϕ
}={
Mp,p′θ,θ
}= I2�
00ms�
00;{
Mp,p′u,ϕ
}={
Mp,p′v,θ
}= 2I1�
00ms�
00
where
�abη =
1∫
−1
[2a+b−1
�ϕa+b−1 ηdaTm(ϕ)
dϕadbTm′(ϕ)
dϕb
]
dϕ, �cd =1∫
−1
[2c+d−1
θc+d−10
dc Pn(θ)
dθc
dd Pn′(θ)
dθd
]
dθ
η = m0,ms,mc, ds, ct, dt; �ϕ = ϕ1-ϕ0; s = sin
(ϕ1 + ϕ0
2ϕ + �ϕ
2
)
; c = cos
(ϕ1 + ϕ0
2ϕ + �ϕ
2
)
t = tan
(ϕ1 + ϕ0
2ϕ + �ϕ
2
)
; m0 = 1, ms = s, mc = c, ds = 1
s, ct = c
t, dt = 1
t
References
1. Jin, G., Ye, T., Chen, Y., Su, Z., Yan, Y.: An exact solution for the free vibration analysis of laminated composite cylindricalshells with general elastic boundary conditions. Compos. Struct. 106, 114–127 (2013)
2. Jin, G., Ye, T., Ma, X., Chen, Y., Su, Z., Xie, X.: A unified approach for the vibration analysis of moderately thick compositelaminated cylindrical shells with arbitrary boundary conditions. Int. J. Mech. Sci. 75, 357–376 (2013)
3. Qu, Y., Long, X., Wu, S., Meng, G.: A unified formulation for vibration analysis of composite laminated shells of revolutionincluding shear deformation and rotary inertia. Compos. Struct. 98, 169–191 (2013)
4. Qatu, M.S., Sullivan, R.W., Wang, W.C.: Recent research advances on the dynamic analysis of composite shells:2000–2009. Compos. Struct. 93(1), 14–31 (2010)
5. Noor, A.K., Burton, W.S.: Assessment of computational models for multilayered composite shells. Appl. Mech.Rev. 43(4), 67–97 (1990)
6. Qatu, M.S.: Recent research advances in the dynamic behavior of shells: 1989–2000, part 1: laminated compositeshells. Appl. Mech. Rev. 55(4), 325–349 (2002)
7. Carrera, E.: Theories and finite elements for multilayered anisotropic, composite plates and shells. Arch. Comput. Meth.Eng. 9(2), 87–140 (2002)
8. Qatu, M.S.: Vibration of Laminated Shells and Plates, 1st edn. Elsevier Ltd, Amsterdam (2004)9. Carrera, E., Brischetto, S., Nali, P.: Plates and Shells for Smart Structures: Classical and Advanced Theories for Modeling
and Analysis, 1st edn. Wiley, UK (2011)10. Reddy, J.N.: Mechanics of Laminated Composite Plates and Shells: Theory and Analysis, 2nd edn. CRCpress, New
York (2004)11. Bardell, N.S., Dunsdon, J.M., Langley, R.S.: Free vibration of thin, isotropic, open, conical panels. J. Sound Vib. 217(2),
297–320 (1998)12. Fazzolari, F.A., Carrera, E.: Advances in the Ritz formulation for free vibration response of doubly-curved anisotropic
laminated composite shallow and deep shells. Compos. Struct. 101, 111–128 (2013)
470 T. Ye et al.
13. Ferreira, A.J.M., Carrera, E., Cinefra, M., Roque, C.M.C.: Analysis of laminated doubly-curved shells by a layerwise theoryand radial basis functions collocation, accounting for through-the-thickness deformations. Comput. Mech. 48, 13–25 (2011)
14. Ferreira, A.J.M., Carrera, E., Cinefra, M., Roque, C.M.C., Polit, O.: Analysis of laminated shells by a sinusoidal sheardeformation theory and radial basis functions collocation, accounting for through-the-thickness deformations. Compos.Part B: Eng. 42, 1276–1284 (2011)
15. Reddy, J.N., Asce, M.: Exact solutions of moderately thick laminated shells. J. Eng. Mech. 110, 794–809 (1984)16. Khdeir, A.A., Reddy, J.N.: Free and forced vibration of cross-ply laminated composite shallow arches. Int. J. Solids
Struct. 34(10), 1217–1234 (1997)17. Qatu, M.S.: Natural vibration of free, laminated composite triangular and trapezoidal shallow shells. Compos. Struct. 31, 9–
19 (1995)18. Soldatos, K.P., Shu, X.P.: On the stress analysis of cross-ply laminated plates and shallow shell panels. Compos.
Struct. 46, 333–344 (1999)19. Librescu, L., Khdeir, A.A., Frederick, D.: A shear deformable theory of laminated composite shallow shell-type panels and
their response analysis I: free vibration and buckling. Acta. Mech. 76, 1–33 (1989)20. Hosseini-Hashemi, Sh., Atashipour, S.R., Fadaee, M., Girhammar, U.A.: An exact closed-form procedure for free vibration
analysis of laminated spherical shell panels based on Sanders theory. Arch Appl. Mech. 82, 985–1002 (2012)21. Leissia, A.W., Chang, J.D.: Elastic deformation of thick, laminated composite shells. Compos. Struct. 35, 153–170 (1996)22. Oktem, A.S., Chaudhuri, R.A.: Fourier analysis of thick cross-ply Levy type clamped doubly-curved panels. Compos.
Struct. 80, 489–03 (2007)23. Viola, E., Tornabene, F., Fantuzzi, N.: General higher-order shear deformation theories for the free vibration analysis of
completely doubly-curved laminated shells and panels. Compos. Struct. 95, 639–666 (2013)24. Singh, A.V., Kumar, V.: Vibration of laminated shells on quadrangular boundary. J. Aerosp. Eng. 9, 52–57 (1996)25. Qatu, M.S.: Vibration analysis of cantilevered shallow shells with triangular trapezoidal planforms. J. Sound Vib. 191(2),
219–231 (1996)26. Qatu, M.S.: Effect of inplane edge constraints on natural frequencies of simply supported doubly curved shallow
shells. Thin-Walled Struct. 49, 797–803 (2011)27. Qatu, M.S.: Vibration studies on completely free shallow shells having triangular and trapezoidal planforms. Appl.
Acoust. 44, 215–231 (1995)28. Ye, T., Jin, G., Chen, Y., Ma, X., Su, Z.: Free vibration analysis of laminated composite shallow shells with general elastic
boundaries. Compos. Struct. 106, 470–490 (2013)29. Bardell, N.S., Langley, R.S., Dunsdon, J.M., Aglietti, G.S.: An h − p finite element vibration analysis of open conical
sandwich panels and conical sandwich frusta. J. Sound Vib. 226(2), 345–377 (1999)30. Chern, Y.C., Chao, C.C.: Comparison of natural frequencies of laminates by 3-D theory, part II: curved panels. J. Sound
Vib. 230(5), 1009–1030 (2000)31. Selmane, A., Lakis, A.A.: Dynamic analysis of anisotropic open cylindrical shells. Comput. Struct. 62(1), 1–12 (1997)32. Singh, B.N., Yadav, D., Iyengar, N.G.R.: Free vibration of laminated spherical panel with random material properties. J.
Sound Vib. 224(4), 321–338 (2001)33. Soldatos, K.P., Messina, A.: The influence of boundary conditions and transverse shear on the vibration of angle-ply laminated
plates, circular cylinders and cylindrical panels. Comput. Methods Appl. Mech. Eng. 190(18-19), 2385–2309 (2001)34. Lee, J.J., Yeom, C.H., Lee, I.: Vibration analysis of twisted cantilevered conical composite shells. J. Sound Vib. 225(5),
965–982 (2002)35. Hu, X.X., Sakiyama, T., Matsuda, H., Morita, C.: Vibration of twisted laminated composite conical shells. Int. J. Mech.
Sci. 44, 1521–1541 (2002)36. Lim, C.W., Liew, K.M., Kitipornchai, S.: Vibration of cantilevered laminated composite shallow conical shells. Int. J.
Solids Struct. 35(15), 1695–1707 (1998)37. Zhao, X., Liew, K.M., Ng, T.Y.: Vibration analysis of laminated composite cylindrical panels via a meshfree approach. Int.
J. Solids Struct. 40(1), 161–180 (2003)38. Zhao, X., Ng, T.Y., Liew, K.M.: Free vibration of two-side simply-supported laminated cylindrical panels via the mesh-free
kp-Ritz method. Int. J. Solids Struct. 46(1), 123–142 (2004)39. Bardell, N.S., Dunsdon, J.M., Langley, R.S.: Free and forced vibration analysis of thin, laminated, cylindrical curved
panels. Compos. Struct. 38, 453–462 (1997)40. Bercin, A.N.: Natural frequencies of cross-ply laminated singly curved panels. Mech. Res. Commun. 23(2), 165–170 (1996)41. Messina, A., Soldatos, K.P.: Influence of edge boundary conditions on the free vibrations of cross-ply laminated circular
cylindrical shell panels. J. Acoust. Soc. Am. 106(5), 2608–2620 (1999)42. Messina, A., Soldatos, K.P.: Vibration of completely free composite plates and cylindrical shell panels by a high-order
theory. Int. J. Mech. Sci. 41, 891–918 (1999)43. Selmane, A., Lakis, A.A.: Vibration analysis of anisotropic open cylindrical shells subjected to a flowing fluid. J. Fluid
Struct. 11, 111–134 (1997)44. Lee, S.J., Reddy, J.N.: Vibration suppression of laminated shell structures investigated using higher-order shear deformation
theory. Smart. Mater. Struct. 13, 1176–1194 (2004)45. Messina, A.: Free vibration of multilayered doubly curved shells based on a mixed variational approach and global
piecewise-smooth functions. Int. J. Solid Struct. 40, 3069–3088 (2003)46. Qatu, M.S., Leissa, A.W.: Free Vibrations of completely free doubly curved laminated composite shallow shells. J. Sound
Vib. 151(1), 9–29 (1991)47. Li, W.L.: Vibration analysis of rectangular plates with general elastic boundary supports. J. Sound Vib. 273, 619–35 (2004)48. Talebitooti, M.: Three-dimensional free vibration analysis of rotating laminated conical shells: layerwise differential
quadrature (LW-DQ) method. Arch. Appl. Mech. 83, 765–781 (2013)49. Warburton, T., Embree, M.: The role of the penalty in the local discontinuous Galerkin method for Maxwell’s eigenvalue
problem. Comput. Methods Appl. Mech. Eng. 195, 3205–3223 (2006)
Arbitrary boundary conditions 471
50. Sármány, D., Izsák, F., Vander Vegt, J.J.W.: Optimal penalty parameters for symmetric discontinuous Galerkin Discretisa-tions of the time-harmonic Maxwell equations. J. Sci. Comput. 44(3), 219–254 (2010)
51. Qu, Y., Yuan, G., Wu, S., Meng, G.: Three-dimensional elasticity solution for vibration analysis of composite rectangularparallelepipeds. Eur. J. Mech. A-Solid 42, 376–394 (2013)
52. Zhou, D., Cheung, Y.K., Lo, S.H., Au, F.T.K.: 3D vibration analysis of solid and hollow circular cylinders via Chebyshev–Ritzmethod. Comput. Methods Appl. Mech. Eng. 192, 1575–1589 (2003)