ARTICLE IN PRESS

Journal of Sound and Vibration 329 (2010) 733751

Contents lists available at ScienceDirect

Journal of Sound and Vibration

journal homepage: www.elsevier.com/locate/jsvi

Free vibrational characteristics of isotropic coupled

cylindricalconical shells
Mauro Caresta , Nicole J. Kessissoglou
School of Mechanical and Manufacturing Engineering, The University of New South Wales, Sydney NSW 2052, Australia

a r t i c l e in fo

abstract

Article history:
Received 11 March 2009
Received in revised form
29 September 2009
Accepted 4 October 2009
Handling Editor: L.G. Tham
Available online 28 October 2009

This paper presents the free vibrational characteristics of isotropic coupled conical
cylindrical shells. The equations of motion for the cylindrical and conical shells are
solved using two different methods. A wave solution is used to describe the
displacements of the cylindrical shell, while the displacements of the conical sections

are solved using a power series solution. Both DonnellMushtari and Flugge
equations
of motion are used and the limitations associated with each thin shell theory are
discussed. Natural frequencies are presented for different boundary conditions. The
effect of the boundary conditions and the inuence of the semi-vertex cone angle are
described. The results from the theoretical model presented here are compared with
those obtained by previous researchers and from a nite element model.
& 2009 Elsevier Ltd. All rights reserved.

1. Introduction
Cylindrical shells are widely reported in literature. Many researchers such as DonnellMushtari, Timoshenko, Reissner,

Flugge,
to name a few, have developed thin shell theory arising from different simplifying assumptions based on Loves
postulates. A range of general solutions for the cylindrical shell displacements and results for different boundary conditions
have been summarised by Leissa [1]. Conical shells have not been as widely reported in literature as in the case of
cylindrical shells. This is due to the increased mathematical complexity associated with the effect of the variation of the
radius along the length of the cone on the elastic waves. The approximate location of the natural frequencies for conical
shells has been found using the RayleighRitz method by several authors [15]. A transfer matrix approach was used by Irie
et al. [6] to solve the free vibration of conical shells. Tong [7] presented a procedure for the free vibration analysis of
isotropic and orthotropic conical shells in the form of a power series. Guo [8] studied the propagation and radiation
properties of elastic waves in conical shells.
Very little work can be found on the vibrations of coupled cylindricalconical shells, of which common applications are
submarine hulls, aircraft, missiles and autonomous underwater vehicles (AUVs). Early analytical and experimental work to
determine the natural frequencies and mode shapes of coupled conicalcylindrical shells used the nite element method
[9]. The classic bending theory was used by Kalnins [10] and Rose et al. [11] to examine rotationally symmetric shells.
Hu and Raney [12] examined the effects of discontinuities at the joint connecting the cone and cylinder. A transfer matrix
approach was used by Irie et al. [13] to solve the free vibration of coupled cylindricalconical shells. Efraim and Eisenberger
[14] applied a power series solution to calculate the natural frequencies of segmented axisymmetric shells. Patel et al. [15]
presented results for laminated composite joined conicalcylindrical shells using a nite element method.

 Corresponding author.

E-mail address: maurorestaca@yahoo.it (M. Caresta).

0022-460X/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jsv.2009.10.003

ARTICLE IN PRESS
734

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

In this work, the authors present a different approach to describe the free vibrational characteristics of coupled isotropic
cylindricalconical shells. The equations of motion for the cylindrical and conical shells are solved using two different
methods. The cylindrical shell equations are solved using a wave solution while the conical shell equations are solved using
a power series solution. These two methods are merged together for the rst time to describe the dynamic response of the
coupled shells. Two thin shell theories are used corresponding to DonnellMushtari and the higher order equations of

Flugge.
In the latter case it is shown that an approximation is required in order to apply the power series solution for the
conical shell. Results in terms of natural frequencies and mode shapes are compared with data available in literature. The
effect of the junction between the coupled shells and the boundary conditions is also investigated.
2. Equations of motion for thin shells
The equations of motion to describe the vibrations of cylindrical or conical shells can be derived according to a
particular thin shell theory using the standard derivation [1]. For completeness of the present study, the derivation of thin
shell theory is briey reviewed in Appendix A.
2.1. Equations of motion for a cylindrical shell
Using a cylindrical coordinate system (x, y), u, v and w are the orthogonal components of the shell displacement in the

axial, circumferential and radial directions, respectively, as shown in Fig. 1. According to Flugge
theory, the equations of
motion for a thin cylindrical shell are
2
3
3
@2 u 1  u
1 u @2 v
u @w
1 @2 u
2 @ u
2 @ w
2 1  u @ w

1 b 2

 2 2 0
b a 3 b
2
2
2
2a @x @y a @x
2a @x @y
@x
2a
@x
cL @t
@y

(1)

!
2
1 u @2 u
1  u @2 v 1 @2 v 1 @w
3  u @3 w
1 @2 v
2 31  u @ v
 2 2 0

b

2a @x @y
2 @x2 a2 @y2 a2 @y
2
2 @x2 @y
@x2
cL @t

(2)

!
@4 w
@4 w
1 @4 w
@3 u 1  u @3 u
3  u @3 v
2 @2 w


a

2
2a @x @y2
2 @x2 @y a2 @y2
@x4
a2 @y4
@x3
@x2 @y


2
u @u 1 @v
1@ w
2
w1 b 2 2 0
(3)

a @x a2 @y
cL @t
p
where a is the radius of the middle surface of the shell, b h= 12a is the thickness parameter, h is the shell thickness and
f @w=@x is the slope. cL E=r1  u2 1=2 is the longitudinal wave speed. For DonnellMushtari theory, the equations of
motion simplify to

b2 a 2

@2 u 1  u @2 u 1 u @2 v
u @w 1 @2 u

0

2
2
2
2a @x @y a @x cL2 @t 2
@x
2a @y

(4)

1 u @2 u
1  u @2 v 1 @2 v 1 @w 1 @2 v


0
2a @x @y
2 @x2 a2 @y2 a2 @y cL2 @t 2

(5)

b2 a 2

@4 w
@4 w
1 @4 w
2
2 4
4
2
@x
a @y
@x2 @y

u @u 1 @v w
1 @2 w
2 2 2 0
2
a @x a @y a
cL @t

x
v

w
u

Fig. 1. Coordinate system for a thin walled cylindrical shell.

(6)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

vc

wc
c

uc, xc
s

R0

R1

735

R2

Lc
Fig. 2. Coordinate system for a thin walled conical shell.

2.2. Equations of motion for a conical shell

For a conical shell, the coordinate system (xc , yc ) is dened in Fig. 2. The equations of motion are given in terms of uc and
vc that are the orthogonal components of the displacement in the xc and yc directions, respectively. wc is the displacement
normal to the shell surface. a is the semi-vertex angle of the cone. s is the coordinate used in the standard derivation
presented in Appendix A. R is the radius of the cone at location xc. R0 is the mean radius of the shell and corresponds to the
origin of the coordinate system. Lc is the length of the cone along its generator, and R1 , R2 are, respectively, the radii at the

smaller and larger ends of the cone. According to Flugge

theory, the equations of motion are
1 @2 uc
L11 L~ 11 uc L12 vc L13 L~ 13 wc  2
0
ccL @t 2

(7)

1 @2 vc
L21 uc L22 L~ 22 vc L23 L~ 23 wc  2
0
ccL @t 2

(8)

1 @2 wc
L31 L~ 31 uc L32 L~ 32 vc L33 L~ 33 wc  2
0
ccL @t2

(9)

ccL Ec =rc 1  u2c 1=2 is the longitudinal wave speed. Ec , rc and uc are, respectively, Youngs modulus, density and Poissons

equations of motion are more

ratio. Using DonnellMushtari theory, the differential operators L~ ij are zero. The Flugge
complicated due to the numerous higher order terms. The differential operators Lij and L~ ij are given in Appendix B.
3. Solutions to the equations of motion
The equations of motion for the cylindrical and conical shells are solved using two different methods, and are then
merged together to provide the complete response of the coupled conicalcylindrical structure. The cylindrical shell
equations are solved using a wave solution whilst the conical shell equations are solved using a power series method.

Furthermore, expressions for the conical shell displacements are obtained for both the DonnellMushtari and Flugge
theories.
3.1. General solutions for the cylindrical shell
General solutions to the equations of motion for a cylindrical shell can be assumed as [1]
ux; y; t Uejkn x cosnyejot

(10)

vx; y; t Vejkn x sinnyejot

(11)

wx; y; t Wejkn x cosnyejot

(12)

where kn is the axial wavenumber and n is the circumferential mode number. Substituting the general solutions given by

Eqs. (10)(12) into the Flugge

equations of motion given by Eqs. (1)(3) results in three linear equations in terms of U, V and
W. These linear equations can be arranged in matrix form as AU 0, where U U V WT contains the unknown wave
amplitudes. The elements of the matrix A are given in Appendix C. For a non-trivial solution, the determinant of the matrix
A must be zero. The expanded determinant results in an eighth order characteristic equation in kn . For each value of

ARTICLE IN PRESS
736

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

kn;i i 1 : 8, the axial and circumferential amplitude ratios can be obtained as Cn;i Un;i =Wn;i and Gn;i Vn;i =Wn;i ,
respectively. For harmonic motion, the complete solutions are given by
ux; y; t

1 X
8
X

Cn;i Wn;i ejkn;i x cosnyejot

(13)

Gn;i Wn;i ejkn;i x sinnyejot

(14)

n0 i1

vx; y; t

1 X
8
X
n0 i1

wx; y; t

1 X
8
X

Wn;i ejkn;i x cosnyejot

(15)

n0 i1

3.2. General solutions for the conical shell

The equations of motion for the conical shell are solved using the power series approach presented by Tong [7] for
shallow shell theory. This approach is applied here to the equations of motion for both the DonnellMushtari and the

Flugge
thin shell theories. General solutions to the equations of motion for a conical shell given by Eqs. (7)(9) can be
expressed as
uc xc ; yc ; t uc xc cosnyc ejot

(16)

vc xc ; yc ; t vc xc sinnyc ejot

(17)

wc xc ; yc ; t wc xc cosnyc ejot

(18)

where the xc-dependent component of the displacement can be expressed in terms of a power series by
uc xc

1
X

am xm
c

(19)

bm xm
c

(20)

cm xm
c

(21)

m0

vc xc

1
X
m0

wc xc

1
X
m0

Solutions for the conical shell displacements for the two thin shell theories are presented in what follows.
3.2.1. DonnellMushtari equations
Using the low order DonnellMushtari theory, the equations of motion given by Eqs. (7) and (8) (and where the
differential operators L~ ij are zero) are multiplied by R2 while Eq. (9) is multiplied by R4 [7], resulting in
R2 L11 uc R2 L12 vc R2 L13 wc 

R2 @2 uc
0
2 @t 2
ccL

(22)

R2 L21 uc R2 L22 vc R2 L23 wc 

R2 @2 vc
0
2 @t 2
ccL

(23)

R4 L31 uc R4 L32 vc R4 L33 wc 

R4 @2 wc
0
2 @t 2
ccL

(24)

Substituting Eqs. (16)(21) into Eqs. (22)(24) results in the following recurrence relations for m 0; 1; 2; . . .:
am2

4
X

Aa;i am3i

i1

bm2

2
X
i1

2
X

Ab;i am1i

i1

cm4

4
X

Ba;i bm1i

Ac;i am3i

i1

The coefcients in Eqs. (25)(27) are given in Appendix D.

2
X

Ca;i cm1i

(25)

i1
4
X

Bb;i bm3i Cb;1 cm

(26)

i1
3
X
i1

Bc;i bm3i

8
X
i1

Cc;i cm5i

(27)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

737

3.2.2. Fl
ugge equations

Using the Flugge

equations, the application of the power series solution is more complicated compared with the
DonnellMushtari theory due to the higher order terms. The coefcients of the operators Lij and L~ ij fi; j 1 : 3g include
terms of the form 1=Rk , k=1:4. Hence, the equations of motion given by Eqs. (17)(19) are multiplied by R4 in order to apply
the power series solution. Furthermore, the terms with h2 in the membrane force Ns given by Eq. (A.6), given in Appendix A,
are neglected, as in the DonnellMushtari theory. The approximation of Ns results in a new L~ 13 term, given by
h2 sin2 a cos3 a h2c sin2 a cos a
@ 1  uc h2c @3
3  u h2c sin a cos a @2



L~ 13  c
4
3
2
2
12
@xc 2R 12 @xc @y
2 12
R
12R
R4
@yc
c

(28)

Substituting Eqs. (16)(21) into Eqs. (7)(9) (multiplied by R4) results in the following recurrence relations for
m 0; 1; 2; . . .:
am2

6
X

A~ a;i am5i

i1

bm2

B~ a;i bm3i

i1

4
X

A~ b;i am3i

i1

cm4

4
X

6
X

A~ c;i am3i

i1

5
X

C~ a;i cm3i

(29)

C~ b;i cm3i

(30)

C~ c;i cm5i

(31)

i1

B~ b;i bm5i

i1

6
X

4
X

5
X
i1

B~ c;i bm3i

i1

8
X
i1

The recurrence coefcients in Eqs. (29)(31) are given in Appendix E. It is important to note that if the h2 term in Eq. (A.6)
is not neglected, the power series method cannot be applied since the recurrence relation given by Eq. (29) would
become

am2

6
X

A~ a;i am5i

i1

4
X

B~ a;i bm3i

i1

6
X

C~ a;i cm3i

(32)

i1

The new terms cm2 and cm3 are not compatible with the term am2 on the left side of the equation. It can be concluded

that the use of the power series solution with the Flugge
equations is only possible if the approximated membrane force Ns,
as in the DonnellMushtari theory, is used.
3.2.3. Conical shell displacements
Tong [7] showed that the xc-dependent part of the displacements can be expressed in terms of eight unknown
coefcients a0 ; a1 ; b0 ; b1 ; c0 ; c1 ; c2 ; c3 ; which can be determined from the boundary conditions at both ends on the conical
shell. In terms of the unknown coefcients, Eqs. (19)(21) can be written as follows:
uc xc u  x;

vc xc v  x;

wc xc w  x

(33)

where
u u1 xc    u8 xc 

(34)

v v1 xc    v8 xc 

(35)

w w1 xc    w8 xc 

(36)

x a0 a1 b0 b1 c0 c1 c2 c3 T

(37)

x is the vector of the eight unknown coefcients. In Eqs. (34)(36), ui xc , vi xc and wi xc are the base functions of uc xc ,
vc xc and wc xc , respectively. The convergence property of the series solutions uc xc , vc xc , wc xc given by Eqs. (19)(21)
has been previously discussed by Tong [7] and are maintained for the thin-shell theories presented here.
4. Boundary and continuity conditions
The two different methods corresponding to the wave solution and power series method both require the application of
four boundary conditions at each end of the shell to determine the unknown coefcients. Thus, the cylindrical and conical
shells can be coupled together by applying the required continuity and equilibrium conditions at the interface.
The remaining boundary conditions are applied at the ends of the coupled cylindricalconical shell. The forces, moments
and displacements at the junction and at the boundaries of the coupled shells are given in accordance with the sign

ARTICLE IN PRESS
738

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

Qx
Mx

Nx,

Nx

w
v
u

Nx,

wc
vc

Qx,c
Nx,c

uc

Mx,c

Mx

Nx

Nx,c

Qx

Nx,c
Nx,c

Mx,c
Qx,c

Fig. 3. Positive directions for the forces, moments and displacements of the cone and cylinder.

convention shown in Fig. 3. At the cylindercone junction, continuity of displacements, slope, forces and bending moment
are given by
u Uc

(38)

w Wc

(39)

v Vc

(40)

@w @wc

@xc
@x

(41)

N~ x;c  Nx 0

(42)


 

M
M
Nxy;c xy;c  Nxy xy 0
R2
a

(43)

Mx;c  Mx 0

(44)

V~ x;c  Vx 0

(45)

To take into account the change of curvature between the cylinder and the cone, the following notation was introduced:
Uc uc cos a  wc sin a

(46)

Wc uc sin a wc cos a

(47)

N~ x;c Nx;c cos a  Vx;c sin a

(48)

V~ x;c Vx;c cos a Nx;c sin a

(49)

The membrane forces Nx , Ny and Nxy , bending moments Mx , My and Mxy , transverse shearing Qx and the KelvinKirchhoff
shear force Vx can be derived for both shells. Different boundary conditions can be applied to the extremities of the coupled
shells. In this work, three different boundary conditions have been considered, corresponding to free, clamped and shear
diaphragm. The various boundary conditions, for example for the cylindrical shell, are


M
(50)
Free end : Nx Nxy xy Mx Vx 0
a
Clamped end : u w

@w
v0
@x

Shear-diaphragm SD end : Nx v Mx w 0

(51)
(52)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

739

a
R1

Fig. 4. Coupled cylindricalconical shell.

Table 1
Frequency parameters for the freeclamped cylindricalconical shell.
Mode order

Frequency parameter Oc

Irie et al. [13]

Efraim et al. [14]

Present (DonnellMushtari)

Present (Flugge)

1
T
2
3
4
5

0.5047

0.9312
0.9566
0.9718
1.0122

0.503779
0.609852
0.930942
0.956379
0.971634
1.012090

0.503752
0.609855
0.930916
0.956315
0.971596
1.011884

0.505354
0.609816
0.930904
0.956292
0.971538
1.011873

1
2
3
4
5
6

0.2930
0.6368
0.8116
0.9316
0.9528
0.9922

0.292875
0.635834
0.811454
0.931565
0.952178
0.992175

0.292908
0.635819
0.811446
0.931481
0.952189
0.991959

0.293357
0.636844
0.811434
0.931458
0.952120
0.991936

1
2
3
4
5
6

0.1010
0.5032
0.6916
0.8592
0.9164
0.9608

0.099968
0.502701
0.691305
0.859114
0.915870
0.960702

0.102034
0.502899
0.691479
0.859017
0.916072
0.960475

0.100087
0.502819
0.691353
0.858971
0.915877
0.960429

1
2
3
4
5
6

0.09076
0.3921
0.5148
0.7537
0.7970
0.9197

0.087603
0.391569
0.514478
0.753402
0.796590
0.919635

0.093771
0.392199
0.515184
0.753595
0.796983
0.919391

0.087330
0.391450
0.514424
0.753295
0.796557
0.919369

1
2
3
4
5
6

0.1477
0.3312
0.3965
0.6473
0.6932
0.8720

0.144619
0.330354
0.395649
0.646678
0.692805
0.871812

0.150574
0.331698
0.397604
0.647700
0.693197
0.871555

0.144478
0.330177
0.395495
0.646548
0.692690
0.871532

1
2
3
4
5
6

0.2021
0.2966
0.3730
0.5805
0.6138
0.8187

0.199546
0.296020
0.370901
0.579750
0.613363
0.817951

0.203896
0.296330
0.376227
0.581667
0.614222
0.819801

0.199540
0.295939
0.370707
0.579581
0.613231
0.818014

T denotes the purely torsional frequency.

ARTICLE IN PRESS
740

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

The four boundary conditions at each end of the coupled cylindricalconical shell together with the eight continuity
equations at the junction can be arranged in matrix form BX 0, where X is the vector of the 16 unknown coefcients
given by
X a0 a1 b0 b1 c0 c1 c2 c3 Wn;1    Wn;8 T
The vanishing of the determinant of matrix B gives the undamped natural frequencies of the joined shells.

n = 0; c = 0.6672

n =1; c = 0.4779

n =2; c = 0.3466

n =3; c = 0.2587

n = 4; c = 0.211

n = 5; c = 0.2097

Fig. 5. Lowest order mode shapes corresponding to n=0:5 SDSD case.

(53)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

741

5. Results
5.1. Natural frequencies
To conrm the validity of the presented method, results available in Refs. [13,14] are reproduced here. A coupled
cylindricalconical shell shown in Fig. 4 with free boundary conditions at the cone end and a clamped boundary for the

n = 0; c = 0.8371

n = 1; c = 0.5267

n = 2; c = 0.3771

n = 3; c = 0.2875

n = 4; c = 0.2362

n = 5; c = 0.2252

Fig. 6. Lowest order mode shapes corresponding to n=0:5 clampedclamped case.

ARTICLE IN PRESS
742

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

cylinder is examined, with the following data: L=a 1, h=a 0:01, R1 =a 0:4226, a 303 . The shells are of the same
material with Youngs modulus E 2:11  1011 N m2 , Poissons ratio u 0:3 and density r 7800 kg m3.
The dimensionless frequency parameter Oc oa=cL for the lowest six values of the circumferential mode number
n 0; . . . ; 5 are given in Table 1. The values of the frequency parameters agree well with those presented previously by Irie
et al. [13] and Efraim et al. [14]. A very small difference is observed between the two shell theories except at lower

frequencies, where the DonnellMushtari theory is not as accurate as the Flugge

theory [1]. When n=0, the equation of
motion for the circumferential displacement is uncoupled from the equations of motion for the axial and radial
displacements for both the conical and cylindrical shells, yielding a purely torsional mode [1]. The frequency value of
the mode with order n Z 0 T corresponds to the rst purely torsional mode. This frequency is omitted in the work of
Irie et al. [13] since they did not consider the torsional solution. The purely torsional frequency is reported in Efraim et al.

n = 0; c = 0.6618

n = 1; c = 0.7200

n = 2; c = 0.01001

n = 3; c = 0.02566

n = 4; c = 0.04649

n = 5; c = 0.07271

Fig. 7. Lowest order mode shapes corresponding to n=0:5 freefree case.

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

743

[14], but their corresponding mode shape appears to be in contrast with a purely torsional solution. To further validate the
analytical method presented in this work, a computational nite element model (FEM) was developed using Patran/
Nastran. Quadratic eight node (CQUAD8) elements were used for the 2D thin shell elements; the cylinder and the cone
were meshed with 20 elements in the axial direction and 30 in the circumferential direction. The Lanczos extraction
method was adopted in the analysis [16]. The lowest order mode shapes corresponding to n=0:5 for SDSD,
clampedclamped and freefree boundary conditions have been normalised and are, respectively, shown in Figs. 57.
The analytical results represented by the continuous line are practically indistinguishable to those obtained from the FE
model (represented by dots). Screenshots of the mode shapes from the FE model are also shown. The mode shapes for the
SDSD and clampedclamped cases are similar with a large deformation at the cylinder/cone junction for n=0. As the
circumferential mode number increases, a larger deformation of the cone with respect to the cylinder can be observed. For
the freefree case and n=0:1, the mode shapes show similar characteristics to the other boundary conditions while for nZ2,
a larger deformation for the cylindrical shell is observed compared to the displacement of the conical section.
5.2. Effect of the boundary conditions
The effect of boundary conditions on the free vibrational characteristics of a coupled conicalcylindrical shell is
examined. In Fig. 8, the lowest frequency parameter is plotted versus the circumferential mode number n. The frequencies

calculated using both the DonnellMushtari and Flugge

equations are compared with the results given by the FE model. A
logarithmic scale was used to emphasize the small differences in the results. It can be seen that the results given by the

Flugge
equations of motion match almost perfectly with the FE results. The results given by the DonnellMushtari
equations are affected by several issues for coupled cylindricalconical shells. Firstly, it can be observed that they perform

less well at low frequencies compared with the results given by both the Flugge
equations of motion and the nite element
model. Furthermore, they are in error for the freefree case. For this boundary condition, the equations for n=1 give two
incorrect frequencies of very low values that should be the zeroes associated with rigid body rotation. This is due to the
inconsistency of DonnellMushtari theory with free body motion, as reported by Kadi [17] and Kraus [18].
For clampedclamped and SDSD boundary conditions, the lowest frequency parameter decreases with n. For
freeclamped boundary conditions, the frequencies initially decrease and then increase after n=3. For the freefree case, a
very low frequency occurs at n=2.
5.3. Effect of the semi-vertex cone angle
Figs. 911 present the effect of the semi-vertex angle a of the conical shell on the frequency parameter Oc, for different
boundary conditions of the coupled shell. The following data for the coupled shells were used: L=a 1, h=a 0:01, Lc 1,
a 2 0; 903 . For extreme values of the semi-vertex angle corresponding to a=01 and 901, the conical shell degenerates to a
cylindrical shell and a circular plate, respectively. For the n=0 mode, the behaviour of the coupled shell is similar for all

100

Lowest frequency parameter c

ClampedClamped
SDSD
101
FreeClamped
FreeFree
FEM
D.Mushtari
Flgge

102

Wrong results from the D.Mushtari

equations (FreeFree)
103
0

Circumferential mode number

Fig. 8. Lowest frequency parameter Oc for different boundary conditions calculated using the DonnellMushtari and Flugge
equations and compared with
the results from an FE model.

ARTICLE IN PRESS
744

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

Frequency parameter c

1
n=0
n=1
n=2
n=3
n=4
n=5

0.8
0.6
0.4
0.2
0
0

10

20

30

40
50
60
70
Semivertex angle

80

90

Fig. 9. Lowest frequency parameter Oc for freefree boundary conditions.

Frequency parameter c

1
n=0
n=1
n=2
n=3
n=4
n=5

0.8
0.6
0.4
0.2
0
0

10

20

30

40
50
60
70
Semivertex angle

80

90

Fig. 10. Lowest frequency parameter Oc for freeclamped boundary conditions.

Frequency parameter c

1
n=0
n=1
n=2
n=3
n=4
n=5

0.8
0.6
0.4
0.2
0
0

10

20

30

40 50 60 70
Semivertex angle

80

90

Fig. 11. Lowest frequency parameter Oc for clampedclamped boundary conditions.

boundary conditions considered, resulting in a relatively constant value in the frequency parameter for increasing values of
a and then a mainly linear decrease in Oc . The corresponding motion is primarily axial. As the conical shell changes from a
cylindrical shell (at a=0) to a plate-like structure (at a=901), a decrease in axial stiffness occurs resulting in a decrease in the
frequency parameter. A similar behaviour to the n=0 mode is observed for the n=1 bending mode for a freefree coupled
shell. For nZ2, the frequency parameter is almost constant for all values of a for the freefree shell. For a coupled shell with
a free boundary at the conical shell end and clamped at the cylindrical shell end, for n=1 a slight increase of the frequency
parameter with a is observed, showing a small mass effect. A stiffening effect then dominates after a=651. In the
freeclamped shell, higher order circumferential modes result in a wavelike behaviour due to greater shape complexity.

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

745

6. Conclusions
A different approach to obtain the free vibrational characteristics of coupled cylindricalcoupled shells has been
introduced. Two different methods corresponding to a wave solution and power series method were used to obtain the
shell displacements. The shells were then coupled by means of continuity conditions at the cone/cylinder junction. Results
in terms of natural frequencies were compared for two different thin shell theories, corresponding to DonnellMushtari

and Flugge
equations, as well as with data presented previously in Refs. [13,14]. It was shown that in order to use the Flugge
equations of motion with the power series solution, an approximation of the shear force is required. The effect of four
classical boundary conditions at the ends of the coupled shells on the natural frequencies was investigated.

In general, little difference was observed between the results given by the two shell theories. The Flugge
equations were
shown to be in very close agreement with results from a nite element model, but the DonnellMushtari equations were
less accurate at low frequencies. Furthermore, for freefree boundary conditions of the coupled shells, the DonnellMushtari
equations generate errors in the values of the lowest frequency parameter for circumferential mode number n=1.
The method described in this work can also be applied to the coupled shells of different materials and thickness.
Appendix A. Equations of motion for thin isotropic shells

According to Flugge
theory, the equations of motion for a thin shell are given by
@BNs @ANys @A
@B
AB
@2 u

Ns y 
Qs  ABrh 2 0

Ny
@s
@y
@y
@s
Rs
@t

(A.1)

@ANy @BNsy @B
@A
AB
@2 v

Ns
Q  ABrh 2 0

N 
@y
@s
@s ys @y
Ry y
@t

(A.2)

AB
AB
@BQs @AQy
@2 w
Ns 
N
 ABrh 2 0

Rs
Ry y
@s
@y
@t

(A.3)

where u, v and w, respectively, denote the orthogonal component of the displacement. r is the density and h is the shell
thickness. The equations are given in terms of two independent coordinates s and y. The parameters A, B, Rs and Ry depend
on the type of shell. The forces and moments in Eqs. (A.1)(A.3) are given by [1]
Qs

1 @BMs @AMys @A
@B

M 

M
AB @s
@y
@y sy @s y

(A.4)

Qy

1 @AMy @BMsy @B
@A

Ms

M 
AB @y
@s
@s ys @y

(A.5)

Ns





Eh
h2 1
1
es
k
e

u
e



s
s
y
12 Rs Ry
Rs
1  u2

(A.6)

Ny





Eh
h2 1
1
e
ky  y
ey ues 

2
12 Ry Rs
Ry
1u

(A.7)

Ns y





Eh
h2 1
1
t esy

esy 

12 Rs Ry
2 Rs
21 u

(A.8)

Ny s





Eh
h2 1
1
t esy

esy 

12 Ry Rs
2 Ry
21 u

(A.9)

Ms



 
Eh3
1
1

es
ks uky 
2
Rs Ry
121  u

(A.10)

My



 
Eh3
1
1

uk


e
k
s
Ry Rs y
121  u2 y

(A.11)

Msy



Eh3
e
t  sy
241 u
Rs

(A.12)

Mys



Eh3
e
t  sy
241 u
Ry

(A.13)

V s Qs

1 @Mxy
B @y

(A.14)

ARTICLE IN PRESS
746

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

E is Youngs modulus and u is Poissons ratio of the material. Eq. (A.14) is the KelvinKirchhoff shearing force. The normal
strain es , ey , and shear strain esy of the middle surface and the rotations of the normal to the middle surface denoted by Ws
and Wy are given by

es

1 @u
v @A w

A @s AB @y Rs

(A.15)

ey

1 @v
u @B w

B @y AB @s Ry

(A.16)

A @u=A B @v=B

B @y
A @s

(A.17)

Ws

u 1 @w

Rs A @s

(A.18)

Wy

v
1 @w

Ry B @y

(A.19)

esy

The mid-surface changes in curvature ks , ky and twist t are given by

ks

1 @Ws Wy @A

AB @y
A @s

(A.20)

ky

1 @Wy Ws @B

AB @s
B @y

(A.21)





A @Ws =A B @Wy =B 1 1 @u
v @B
1 1 @v
u @A


B @y
A @s
Rs B @y AB @s
Ry A @s AB @y

(A.22)

According to the DonnellMushtari theory, the terms including Qs and Qy in Eqs. (A.1) and (A.2) are neglected
and Eqs. (A.6)(A.13), respectively, simplify to Ns Eh=1  u2 es uey , Ny Eh=1  u2 ey ues , Nsy Nys
Eh=21 uesy , Ms Eh3 =121  u2 ks uky , My Eh3 =121  u2 ky uks , Msy Mys Eh3 =241 ut. Furthermore,
the mid-surface changes in curvature ks , ky and twist t simplify to the following expressions:


1 @ 1 @w
1 @A @w
ks 
(A.23)
 2
A @s A @s
AB @y @y


1 @ 1 @w
1 @B @w
 2
B @y B @y
BA @s @s

(A.24)





B @ 1 @w
A @ 1 @w

A @s B2 @y
B @y A2 @s

(A.25)

ky 

t

For a cylindrical shell, the equations of motion for u, v and w can be derived from Eqs. (A.1)(A.3) using the following
parameters; A a, B a, Rs 1, Ry a and s x=a, where a is the mean radius of the shell and u, v, w and x are dened as
in Fig. 1. For a conical shell, the equations of motion for uc, vc and wc can be derived using A 1, B s sin a, Rs 1,
Ry s tan a, as well as using the change of coordinate given by s R=sin a and R R0 xc sin a. R, R0 , uc, vc, wc and xc are
dened in Fig. 2.
Appendix B. Differential operators for the conical shell
For the conical shell, omitting the subindex c, the differential operators are given by
sin2 a sin a @
@2
1  u @2

2
2
R @x @x
R
2R2 @y2

(B.1)

1  u h2 cos2 a @2
h2 cos2 a sin2 a

L~ 11
2
2
2
12
R4
2R 12 R
@y

(B.2)

L11 

L12

1 u @2
3  u sin a @

2R @x @y
2
R2 @y

L13 

sin a cos a u cos a @

R
R @x

(B.3)

(B.4)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

h2 sin2 a cos3 a h2 sin2 a cos a h2 1 @3

@ 1  u h2 @3
3  u h2 sin a cos a @2




L~ 13 
2
12
12 R @x3 @x 2R 12 @x @y2
2 12
R4
12R3
R4
@y
3  u sin a @
1 u @2

2
2R @x @y
R2 @y

L21

747

(B.5)

(B.6)

L22 

1  u sin a 1  u sin a @
1 @2 v 1  u @2

2 2
2
R @x R @y
2
2
2 @x2
R

h2 sin2 a cos2 a 3
h2 sin a cos2 a @
h2 cos2 a 3
@2
L~ 22
1  u 

1  u 2
4
3
2
12
12
2
@x 12 R
2
R
R
@x
L23

cos a @
R2 @y

(B.8)
(B.9)

h2 sin a cos a @
h2 cos a 3
@2
h2 cos a 3  u @3


1  u
L~ 23 
4
3
12
@x @y 12 R2
@y 12 R 2
2 @x2 @y
R
L31 

(B.7)

sin a cos a u cos a @


R @x
R2

(B.10)

(B.11)

h2 2 sin3 a cos a h2 sin2 a cos a @

h2 sin a cos3 a h2 cos a @3
h2 sin a cos a 1 u @2



L~ 31 
12
12
12 R @x3 12
@x 12
2 @y2
R4
R3
R4
R4


h2 cos a 1  u @3
12 R3
2 @x @y2

(B.12)

L32 

cos a @v
R2 @y

(B.13)

h2 sin a cos a 3 u @2
h2 sin2 a cos a 3 u @
h2 cos a 3  u @3

L~ 32 
12
2 @x @y 12
2 @y 12 R2
2 @x2 @y
R3
R4
L33 

cos2 a h2 4

r
12
R2

(B.15)

h2 cos2 a
@2
2 cos2 a 2 2
L~ 33 
4
12 R
@y

r 4 r 2 r 2 ; r2

(B.14)

@2
sin a @
1 @2

R @x R2 @y2
@x2

(B.16)

(B.17)

Appendix C. Elements of matrix A

The elements of matrix A for the cylindrical shell are given by
A11 O2  kn a2 

1  u 2
2
n 1 b
2

A12 jnkn a1 u=2

(C.1)
(C.2)

A13 jukn a jb kn a3  n2 kn a1  u=2

(C.3)

A21 A12

(C.4)
2

A22 O2  kn a2 1  u=21 3b  n2
2

(C.5)

A23 n  b nkn a2 3  u=2

(C.6)

A31 A13

(C.7)

A32 A23

(C.8)

(C.9)

A33 1  O2 b fkn a2 n2 2 1  2n2 g

ARTICLE IN PRESS
748

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

O oa=cL is the dimensionless frequency parameter. Using DonnellMushtari theory, Eqs. (C.1), (C.3), (C.5), (C.6) and (C.9),
respectively, reduce to A11 O2  kn a2 1  un2 =2, A13 jukn a, A22 O2  kn a2 1  u=2  n2 , A23 n and
2
A33 1  O2 b kn a2 n2 2 .
Appendix D. Recurrence terms for the DonnellMushtari equations
The recurrence terms for the DonnellMushtari equations are given by
Aa;1 rho2 sin2 a=Da

(D.1)

Aa;2 2rho2 R0 sin a=Da

(D.2)

Aa;3 Gm2  1sin2 a rho2 R20  Ehn2 =21 u=Da

(D.3)

Aa;4 GR0 sin am 12m 1=Da

(D.4)

Ba;1 Gn sin aum m  3 u=2Da

(D.5)

Ba;2 GR0 nm 1u 1=2Da

(D.6)

Ca;1 G sin a cos aum  1

(D.7)

Ca;2 GR0 sin a cos am 1

(D.8)

Ab;1 Gn sin a3 um m  u=2Db

(D.9)

Ab;2 GR0 nm 1u 1=2Db

(D.10)

Bb;1 rho2 sin2 a=Db

(D.11)

Bb;2 2rho2 R0 sin a=Db

(D.12)

Bb;3 Gum2 m2  1 usin2 a  Gn2 rho2 R20 =2Db

(D.13)

Bb;4 GR0 sin am 12m 1=2Db

(D.14)

Cb;1 Ehn cos a=1 uDb

(D.15)

Ac;1 G cos a sin3 a1 um  2u=Dc

(D.16)

Ac;2 GR0 cos a sin a2 3um  3u=Dc

(D.17)

Ac;3 GR20 cos a sin a1 3um=Dc

(D.18)

Ac;4 uGR30 cos am 1=Dc

(D.19)

Bc;1 Gn cos a sin2 a=Dc

(D.20)

Bc;2 2GR0 n cos a sin a=Dc

(D.21)

Bc;3 GR20 n cos a=Dc

(D.22)

Cc;1 rho2 sin4 a=Dc

(D.23)

Cc;2 4rho2 R0 sin3 a=Dc

(D.24)

Cc;3 sin2 aG cos2 a 6rho2 R20 =Dc

(D.25)

Cc;4 2R0 sin aG cos2 a 2rho2 R20 =Dc

(D.26)

Cc;5 D4m2 4m3  m4 sin4 a  GR20 cos2 a D2n2 m2 4n2  4n2 msin2 a  Dn4 rho2 R40 =Dc
Cc;6 DR0 sin a2 sin2 am2 2 sin2 am sin2 a 2n2 m 11 2m=Dc

(D.27)
(D.28)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

749

Cc;7 DR20 6 sin2 am2 sin2 a 2n2 m 2m 1=Dc

(D.29)

Cc;8 2DR30 sin a2m 1m 3m 2m 1=Dc

(D.30)

Da GR20 m 2m 1

(D.31)

Db GR20 m 2m 1u  1=2

(D.32)

Dc DR40 m 4m 3m 2m 1

(D.33)

G Eh=1  u2 ; D Eh3 =121  u2

(D.34)

where

Appendix E. Recurrence terms for the Flugge

equations

The recurrence terms for the Flugge

equations are given by
e a;1 rho2 s4 =D
ea
A

(E.1)

e a;2 4rho2 R0 s3 =D
ea
A

(E.2)

e a;3 fGm  22  1s4 Gn2 u=2 6rho2 R2  Gn2 =2s2 g=D

ea
A
0

(E.3)

e a;4 sR0 9Gs2 m 3Gs2 Gn2 u 4Gs2 m2  Gn2 4rho2 R2 =D

ea
A
0

(E.4)

e a;5 G6R2 m2  3mR2  R2  c2 h2 =2s2 Dn2 uc2 =2 rho2 R4 Gn2 uR2 =2  Dn2 c2 =2  Gn2 R2 =2=D
ea
A
0
0
0
0
0
0

(E.5)

e a;6 GR3 sm 11 4m=D

ea
A
0

(E.6)

e a;1 Gs3 num  u m  5=2D

ea
B

(E.7)

e a;2 GR0 s2 n3um  u 3m  9=2D

ea
B

(E.8)

e a;3 GR2 sn3um 3m  3 u=2D

ea
B
0

(E.9)

e a;4 GR3 nm 1u 1=2D

ea
B
0

(E.10)

e a;1 Gcs3 1 um  2u=D

ea
C

(E.11)

e a;2 GR0 cs2 2 3um  3u=D

ea
C

(E.12)

e a;3 Gcs2h2 n2 72umR2 h2 n2  2c2 h2 h2 n2 um  h2 n2 m  24R2  h2 n2 u  2h2 s2 m=24D

ea
C
0
0

(E.13)

e a;4 GR0 cm 124R2 u  2h2 s2 h2 n2 u  h2 n2 =24D

ea
C
0

(E.14)

e Gs3 num  3u m 1=2D

eb
A
b;1

(E.15)

e GR0 s2 n3um  5u 3m 3=2D

eb
A
b;2

(E.16)

e sn36c2 D 36GumR2  Guh2 c2 3Gh2 c2 36GmR2 36GR2  12GuR2 12c2 Du=24D

eb
A
b;3
0
0
0
0

(E.17)

e GR3 m 1u 1=2D
eb
A
b;4
0

(E.18)

eb
e b;1 rho2 s4 =D
B

(E.19)

eb
e b;2 4rho2 R0 s3 =D
B

(E.20)

eb
e b;3 G1 u  um  22 m  22 s4 =2 6rho2 R2  Gn2 s2 =D
B
0

(E.21)

eb
e b;4 R0 s8rho2 R2  3s2 G 4Gn2 4Gus2 m2  9Gus2 m 3Gus2  4s2 Gm2 9s2 Gm=2D
B
0

(E.22)

ARTICLE IN PRESS
750

M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

e b;5 3GuR2 m2  c2 Du=2 Dc2 c2 D=2  GR2 =2 GuR2 =2  Duc2  Gh2 mc2 =8 umR2 3G=2  3=2GmR2
B
0
0
0
0
0
1=8Gh2 umc2 3GR20 mq 1=24c2 Gh2 m2  1=24Gh2 uc2 m2  3=2c2 Dm c2 Dm2 3=2c2 Dum
eb
 c2 Dum2 s2  Gn2 R20 rho2 R40 =D

(E.23)

eb
e b;6 R0 sm 124GR2 m 24c2 Dm c2 Gh2 m 6GR2  c2 Gh2  6c2 Du  1=12D
B
0
0

(E.24)

e Gcs2 n=D
eb
C
b;1

(E.25)

e 2GR0 csn=D
eb
C
b;2

(E.26)

e cn24GR2 24Dm2 s2 4Gh2 ums2 2n2 Gh2 3Gh2 s2  24n2 D  Gh2 um2 s2  6Gh2 ms2 24c2 D
C
b;3
0
eb
 2c2 Gh2  3Gh2 us2 Gh2 m2 s2 =24D

(E.27)

e R0 csnm 148Dm  2Gh2 m 2Gh2 um  24D  3Gh2 u 5Gh2 =24D

eb
C
b;4

(E.28)

e DR2 cnm 2m 11 u=2D

eb
C
b;5
0

(E.29)

e c;1 Gcs3 1 um  2u=D

ec
A

(E.30)

e c;2 GR0 cs2 2 3um  3u=D

ec
A

(E.31)

e c;3 1=12cD24  36m  12m3 36m2 s3

A
ec
 1=12c12GR20  6Dn2 u c2 Gh2 6Dn2 um 36GuR20 m  Dn2  6Dn2 ms=D

(E.32)

e c;4 1=2R0 cm 1Dn2 6Dms2  2Ds2  6Dm2 s2 Dn2 u 2GuR2 =D

ec
A
0

(E.33)

e c;5 3DR2 csm 2m 1m=D

ec
A
0

(E.34)

e c;6 DR3 cm 3m 2m 1=D

ec
A
0

(E.35)

ec
e c;1 Gcs2 n=D
B

(E.36)

ec
e c;2 2GR0 csn=D
B

(E.37)

ec
e c;3 cnDum2 s2 6Dms2 2GR2  3Dm2 s2  3Ds2  Ds2 u=2D
B
0

(E.38)

ec
e c;4 Dcnm 1sR0 3 2um u  6m=2D
B

(E.39)

ec
e c;5 DR2 cnm  1m3 u=2D
B
0

(E.40)

e c;1 rho2 s4 =D
ec
C

(E.41)

e c;2 4rho2 R0 s3 =D
ec
C

(E.42)

e c;3 s2 c2 G  6rho2 R2 =D

ec
C
0

(E.43)

e c;4 2R0 sc2 G  2rho2 R2 =D

ec
C
0

(E.44)

e c;5 D4m3  m4  4m2 s4 Dmc2  c2 m  2c2  4n2 m 2n2 m2 4n2 s2 rho2 R4  c2 GR2 Dc2 n2
C
0
0
ec
Dn2 c2  c4 D  Dn4 =D

(E.45)

e c;6 DR0 sm 112s2  48s2 m3 72s2 m2  24n2  12c2 48n2 m 12c2 =12D
ec
C

(E.46)

e c;7 DR2 6s2 m2 s2 2n2 m 2m 1=D

ec
C
0

(E.47)

e c;8 2DR3 s2m 1m 3m 2m 1=D

ec
C
0

(E.48)

ARTICLE IN PRESS
M. Caresta, N.J. Kessissoglou / Journal of Sound and Vibration 329 (2010) 733751

751

where
e a GR4 m 2m 1
D
0

(E.49)

e b R2 12GR2 24Dc2 Gh2 c2 m 2m 1u  1=24

D
0
0

(E.50)

e c DR4 m 4m 3m 2m 1
D
0

(E.51)

c cos a;

s sin a

(E.52)

References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]

A.W. Leissa, in: Vibration of Shells, American Institute of Physics, New York, 1993.
H. Saunders, E.J. Paslay, P.R. Wisniewski, Vibrations of conical shells, J. Acoust. Soc. Am. 32 (1960) 765772.
H. Garnet, J. Kempner, Axisymmetric free vibration of conical shells, J. Appl. Mech. 31 (1964) 458466.
R.A. Newton, Free vibrations of rocket nozzles, Am. Inst. Aeronaut. Astronaut. J. 4 (1966) 13031305.
C.C. Siu, C.W. Bert, Free vibrational analysis of sandwich conical shells with free edges, J. Acoust. Soc. Am. 47 (1970) 943945.
T. Irie, G. Yamada, Y. Kaneko, Free vibration of a conical shell with variable thickness, J. Sound Vib. 82 (1982) 8394.
L. Tong, Free vibration of orthotropic conical shells, Int. J. Eng. Sci. 31 (1993) 719733.
Y.P. Guo, Normal mode propagation on conical shells, J. Acoust. Soc. Am. 96 (1994) 256264.
M. Lashkari, V.I. Weingarten, Vibrations of segmented shells, Exp. Mech. 13 (1973) 120125.
A. Kalnins, Free vibration of rotationally symmetric shells, J. Acoust. Soc. Am. 36 (1964) 13551365.
J.L. Rose, R.W. Mortimer, A. Blum, Elastic-wave propagation in a joined cylindricalconicalcylindrical shell, Exp. Mech. 13 (1973) 150156.
W.C.L. Hu, J.P. Raney, Experimental and analytical study of vibrations of joined shells, AIAA J. 5 (1967) 976980.
T. Irie, G. Yamada, Y. Muramoto, Free vibration of joined conicalcylindrical shells, J. Sound Vib. 95 (1984) 3139.
E. Efraim, M. Eisenberger, Exact vibration frequencies of segmented axisymmetric shells, Thin-Walled Struct. 44 (2006) 281289.
B.P. Patel, M. Ganapathi, S. Kamat, Free vibration characteristics of laminated composite joined conicalcylindrical shells, J. Sound Vib. 237 (2000)
920930.
[16] MSC/NASTRAN Basic Dynamic Analysis Users Guide, 1997.
[17] A.S. Kadi, A Study and Comparison of the Equations of Thin Shell Theories, Ph.D. Thesis, Ohio State University, Columbus, 1970.
[18] K. Kraus, in: Thin Elastic Shells, Wiley, New York, 1967.