You are on page 1of 12

Explicit analytical expressions for frequency equation and

mode shapes of composite beams


J.R. Banerjee
*
Department of Mechanical Engineering and Aeronautics, School of Engineering, City University, Northampton Square,
London EC1V OHB, UK
Received 18 August 1999; in revised form 21 March 2000
Abstract
A systematic procedure for the derivation of exact expressions for the frequency equation and mode shapes of
composite beams undergoing free vibration is presented by using the symbolic computing package REDUCE REDUCE. The eect
of material coupling between the bending and torsional modes of deformation, which usually exists in composite beams
due to ply orientation, is taken into account while developing the theory. The governing dierential equations of motion
of the bendingtorsion coupled composite beam are solved analytically for bending displacements and torsional ro-
tations in free vibration. For subsequent developments, the important case of the cantilever beam is chosen because of
its application to aircraft wings. The boundary conditions for displacements and forces for the cantilever are imposed
and the frequency equation is obtained, but rst in the form of a determinant, and then, in the form of an explicit
algebraic expression. The expressions for the mode shapes are also derived in explicit analytical form. The method is
demonstrated by an illustrative example of a composite beam for which some comparative results are available in the
literature. The future potential of this method, particularly in the context of aeroelastic optimisation of composite
wings, is considerable because it is very accurate, computationally ecient and importantly, free from ill-conditioning
problems usually associated with numerical matrix manipulation. 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Composite beam; Free vibration; Frequency Equation; Modal analysis
1. Introduction
With the advent of advanced composite materials, the free vibration analysis of composite beams which
are generally used as structural elements in the idealisation of high aspect ratio composite wings, has in-
spired continuing research interest, particularly amongst the aeroelasticians. This is because the free vi-
bration characteristics generally aect the aeroelastic properties, and importantly for composite wings, such
characteristics can be altered or changed favourably by suitable choice of ply orientations and stacking
sequence. Several investigators have studied the free vibration characteristics of the composite beams in
International Journal of Solids and Structures 38 (2001) 24152426
www.elsevier.com/locate/ijsolstr
*
Fax: +44-171-477-8566.
E-mail address: j.r.banerjee@city.ac.uk (J.R. Banerjee).
0020-7683/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S0020- 7683( 00) 00100- 1
recent years. Examples include Armanios and Badir (1995), Banerjee and Williams (1995), Dancila and
Armanios (1998), Hodges et al. (1991), Kapania and Raciti (1989), Marus and Kant (1996), Minguet
(1989), Minguet and Dugundji (1990), Shi and Lam (1999), Song and Librescu (1991), Weisshaar and Foist
(1985) and Wu and Sun, (1991). Some of these publications focus on cantilever beams and wings, and give
various procedures, which are by and large numerical, to arrive at the required natural frequencies and the
corresponding mode shapes. An attempt to derive the frequency equation and mode shapes of composite
beams in explicit analytical form seems to be a far cry, and apparently has not been made. Such attempts
have been made though, for metallic BernoulliEuler or Timoshenko beams (uncoupled from torsion) as
evident from the literature, see for example Tse et al. (1978), White and Heppler (1995) and Farghaly and
Shebl (1995). This is understandable for the reason that expressions for frequency equation and mode
shapes for composite beams are quite dicult to derive by traditional methods in sharp contrast to those of
metallic beams. The diculty would appear to arise mainly due to the bendingtorsion (material) coupling
eect which leads to the formulation of a higher order governing dierential equation (usually the sixth
instead of the fourth), whose solution leads to the frequency equations and mode shapes to be too com-
plicated or even unmanageable. However, if such expressions could be derived, they would be of great help,
particularly for cantilever composite beams and wings, because they could be used directly, as an accurate
and ecient base for aeroelastic analysis. This opens up further prospects for sensitivity and optimisation
studies in a computationally ecient manner. Indeed, the current investigation stems both from the need
for, and importance of, explicit analytical expressions for the frequency equation and mode shapes of
composite wings, as seen from an aeroelastic and dynamic response perspective. Thus, the central purpose
of this paper is to undertake an analytical formulation using symbolic computation, which leads to explicit,
but concise expressions for the frequency equation and mode shapes of cantilever composite beams ex-
hibiting (material) coupling between the bending and torsional modes of deformation. As an application of
this analytical development, an illustrative example conrming the correctness and accuracy of the ex-
pressions is provided.
The work described in this paper is partly motivated by a recent publication by Dancila and Armanios
(1998), who have used symbolic computation to deal with the coupled bendingtorsion vibration of
composite beams having circumferentially anti-symmetric conguration. However, the analytical devel-
opment of this paper is in marked contrast to that of Dancila and Armanios (1998) whose claim that a
closed form solution for coupled bendingtorsion vibration of a composite beam cannot be derived is
shown to be false. It is not the primary object of this paper to criticise and amend the work presented by
Dancila and Armanios (1998), but it happens that part of the research presented in this paper was carried
out at about the same time as that of the previous authors. It has been shown by the present author that
symbolic computing, coupled with an analytical penetration, oers an almost unlimited potential to
aeroelastic research workers, and no doubt, to others. Thus, some of the conclusions of Dancila and Ar-
manios (1998), which puts symbolic computing in unfavourable light are reversed.
2. Theory
Bendingtorsion materially coupled composite beams of both solid and thin-walled cross-sections have
been characterised in the literature by three main structural parameters related to the bending rigidity EI,
the torsional rigidity GJ and bending torsion material coupling rigidity K. A small sample of this literature
includes the work of Weisshaar and Foist (1985), Smith and Chopra (1991), Banerjee and Williams (1995),
Berdichevsky et al. (1992), and Reheld et al. (1990). However, for composite beams and wings, the
bending torsion material coupling rigidity K is of great signicance because it is dependent on the ply
orientation and stacking sequence which can be exploited to advantage, particularly for the purposes of
aeroelastic tailoring. There have been several attempts by, amongst others, Minguet and Dugundji (1990),
2416 J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426
Chandra et al. (1990), Reheld et al. (1990) and Berdichevsky et al. (1992) to obtain the theoretical and
experimental values for the rigidities EI, GJ, and K which are essential to the derivations that follow.
Fig. 1 shows a straight composite beam of length L with a solid rectangular cross-section and with a
symmetric, but unbalanced lay-up. Bendingtorsion coupling is well known to occur for such congura-
tions (Weisshaar and Foist, 1985; Kapania and Raciti, 1989; Minguet and Dugundji, 1990). Note that the
theory developed can be applied to composite beams of any general cross-section so long as the rigidities
EI, GJ and K are known (either by theory or by experiment), but the rectangular cross-section is shown here
only for convenience. In the right-handed axis system shown, the Y axis coincides with the (central) geo-
metric axis, which is the permitted bending displacement h(y; t) and torsional rotation w(y; t), where y is
measured from the origin shown and t is the time. Using the coupled bendingtorsional beam theory for
thin-walled composites with shear deformation, rotatory inertia and warping stiness neglected, the gov-
erning dierential equations of motion of the beam in free vibration are given by (Dancila and Armanios,
1998; Banerjee and Williams, 1995)
EIh
////
Kw
///
m

h = 0; (1)
GJw
//
Kh
///
I
a

w = 0; (2)
where m is the mass per unit length, I
a
is the polar mass moment of inertia per unit length about the Y axis,
and primes and dots denote dierentiation with respect to position y and t, respectively. Note that EI, GJ
and K above are, respectively, C
33
, C
22
, and C
23
, in the notation used by Dancila and Armanios (1998,
p. 3110).
If a harmonic (periodic) variation of h and w, with circular frequency x, is assumed, then
H(y; t) = H(y)e
ixt
;
w(y; t) = W(y)e
ixt
;
(3)
where H(y) and W(y) are the amplitudes of the harmonically varying bending displacement and torsional
rotation, respectively.
Substituting Eq. (3) into Eqs. (1) and (2) gives
EIH
////
KW
///
mx
2
H = 0; (4)
GJW
//
KH
///
I
a
x
2
W = 0: (5)
Eqs. (4) and (5) can be combined into one equation by eliminating either H or W to give
(D
6
aD
4
bD
2
abc)W = 0; (6)
where
W = H or W; (7)
Fig. 1. Co-ordinate system and sign convention for a laminated composite beam.
J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426 2417
D =
d
dn
;
n =
y
L
; (8)
a = a=c;
b =

b=c;
c = 1 K
2
=(EIGJ);
(9)
with
a = I
a
x
2
L
2
=GJ;

b = mx
2
L
4
=EI:
(10)
In Eq. (6), a, b and c are non-dimensional quantities and are all positive because it is known that
(Weisshaar and Foist, 1985)
0 < c < 1: (11)
The auxiliary (or characteristic) equation of the dierential Equation (6) can be reduced to a cubic
equation which can be proved to have three real roots of which one is positive and the other two are
negative (see Appendix A). A formal proof of this unique solution is stronger than the generalisation given
by Dancila and Armanios (1998) from a limited numerical study, see their Fig. 2 and the statement on pages
31123113: ``. . .the expression for d is intricate and no general statement regarding its sign seems possible''.
Contrary to this statement, a mathematical proof that d is always less than zero for all physical problems is
given in Appendix A of this paper.
The solution of the dierential Equation (6) shows that both H(n) and W(n) have the form (Appendix A)
W (n) = C
1
coshan C
2
sinhan C
3
cosbn C
4
sinbn C
5
coscn C
6
sincn; (12)
where W (n) = H(n) or W(n), C
1
C
6
are constants, and
a = [2(q=3)
1=2
cos(/=3) a=3[
1=2
;
b = [2(q=3)
1=2
cos(p /)=3 a=3[
1=2
;
c = [2(q=3)
1=2
cos(p /)=3 a=3[
1=2
;
(13)
with
q = b a
2
=3;
/ = cos
1
[(27abc 9ab 2a
3
)=2(a
2
3b)
3=2
[: (14)
Hence,
H(n) = A
1
coshan A
2
sinhan A
3
cosbn A
4
sinbn A
5
coscn A
6
sincn; (15)
W(n) = B
1
coshan B
2
sinhan B
3
cosbn B
4
sinbn B
5
coscn B
6
sincn; (16)
where A
1
A
6
and B
1
B
6
are two dierent sets of constants.
Substituting Eqs. (15) and (16) into Eq. (4) shows that the constants A
1
A
6
are related to the constants
B
1
B
6
by the following relationships:
2418 J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426
B
1
= (k
a
=L)A
2
; B
2
= (k
a
=L)A
1
;
B
3
= (k
b
=L)A
4
; B
4
= (k
b
=L)A
3
;
B
5
= (k
c
=L)A
6
; B
6
= (k
c
=L)A
5
;
(17)
where
k
a
= (

b a
4
)=

ka
3
; k
b
= (

b b
4
)=

kb
3
; k
c
= (

b c
4
)=

kc
3
; (18)
with

k = K=EI: (19)
The expressions for bending rotation H(n), the bending moment M(n), the shear force S(n), and the
torque T(n), can be obtained from Eqs. (15) and (16) as follows with prime, now, denoting dierentiation
with respect to n instead of y (Banerjee and Williams, 1995).
H(n) = H
/
(n)=L
= (1=L)A
1
asinhan A
2
acoshan A
3
bsinbn A
4
bcosbn A
5
csincn A
6
ccoscn; (20)
M(n) = (EI=L
2
)H
//
(n) (K=L)W
/
(n) = (EI=L
2
)H
//
(n)

kLW
/
(n)
= (EI=L
2
)A
1
acoshan A
2
asinhan A
3

bcosbn A
4

bsinbn A
5
ccoscn A
6
csincn; (21)
S(n) = (EI=L
3
)H
///
(n) (K=L
2
)W
//
(n) = (EI=L
3
)H
///
(n)

kLW
//
(n)
= (EI=L
3
)A
1
a asinhan A
2
a acoshan A
3
b

bsinbn A
4
b

bcosbn A
5
c csincn A
6
c ccoscn;
(22)
T(n) = (GJ=L)W
/
(n) (K=L
2
)H
//
(n) = (GJ=L)W
/
(n) (K=GJL)H
//
(n)
= (GJ=L
2
)A
1
g
a
coshan A
2
g
a
sinhan A
3
g
b
cosbn A
4
g
b
sinbn A
5
g
c
coscn A
6
g
c
sincn;
(23)
where
a =

b=a
2
;

b =

b=b
2
; c =

b=c
2
; (24)
g
a
= (

b ca
4
)=

ka
2
; g
b
= (

b cb
4
)=

kb
2
; g
c
= (

b cc
4
)=

kc
2
: (25)
The above expressions can be used to formulate the frequency equation and mode shapes of a composite
beam for any classical boundary conditions. However, like previous investigators (Weisshaar and Foist,
1985; Minguet and Dugundji, 1990; Dancila and Armanios, 1998), the demanding case of a cantilever as
applicable to an aircraft wing or a helicopter blade is taken up in detail, in the subsequent text.
3. Frequency equation
The end conditions for a cantilever beam are as follows:
at the built-in end(n = 0): H = 0; H = 0 and W = 0; (26)
at the free end(n = 1): S = 0; M = 0 and T = 0: (27)
Substituting Eq. (26) in Eqs. (15)(20), and Eq. (27) in Eqs. (21)(25) gives
J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426 2419
1 0 1 0 1 0
0 a 0 b 0 c
0 k
a
0 k
b
0 k
c
a aS
ha
a aC
ha
b

bS
b
b

bC
b
c cS
c
c cC
c
aC
ha
aS
ha

bC
b

bS
b
cC
c
cS
c
g
a
C
ha
g
a
S
ha
g
b
C
b
g
b
S
b
g
c
C
c
g
c
S
c
2
6
6
6
6
6
6
4
3
7
7
7
7
7
7
5
A
1
A
2
A
3
A
4
A
5
A
6
2
6
6
6
6
6
6
4
3
7
7
7
7
7
7
5
= 0; (28)
where
C
ha
= cosha; C
b
= cosb; C
c
= cosc;
S
ha
= sinha; S
b
= sinb; S
c
= sinc: (29)
Eq. (28) may be written in matrix form as
BA = 0: (30)
The necessary and sucient condition for non-zero elements in the column vector A of Eq. (30) is that
D = [B[ shall be zero. As a result, the vanishing of D determines the natural frequencies of the cantilever
beam in the usual way. Thus, the frequency equation which corresponds to the solution for the non-trivial
case is given by
D = [B[ = 0: (31)
Expanding the 6 6 determinant D of B algebraically is quite a formidable task and became possible
with the recent advances in symbolic computing. Thus, most of the work reported here, was carried out by
using the software REDUCE REDUCE (Rayna, 1986), particularly when expanding the determinant [B[, and more
importantly, when simplifying the expression for D. The nal expression obtained for D is given below
which is not necessarily in the shortest possible form, but is surprisingly concise.
D = f (x) = k
1
C
b
C
c
C
ha
k
2
C
b
S
c
S
ha
k
3
C
c
S
b
S
ha
k
4
S
b
S
c
C
ha
n
1
C
b
n
2
C
c
n
3
C
ha
; (32)
where
k
1
= 2a al
2
m
2
a cl
2
m
1
acl
1
m
2
2c cl
1
m
1
;
k
2
= a al
1
m
2
c cl
2
m
1
;
k
3
= a al
3
m
2
abl
2
m
2
b cl
2
m
1
;
k
4
= abl
1
m
2
b cl
1
m
1
c cl
3
m
1
;
(33)
n
1
= a al
2
m
1
c cl
1
m
2
; n
2
= a al
2
m
3
a

bl
2
m
2


bcl
1
m
2
; n
3
= a

bl
2
m
1


bcl
1
m
1
c cl
1
m
3
; (34)
with
l
1
= ak
b
bk
a
; l
2
= bk
c
ck
b
; l
3
= ck
a
ak
c
; (35)
and
m
1
= ag
b


bg
a
; m
2
=

bg
c
cg
b
; m
3
= cg
a
ag
c
(36)
with a; b; c and k
a
; k
b
; k
c
and g
a
; g
b
; g
c
and C
ha
; C
b
; C
c
; S
ha
; S
b
; S
c
already dened in Eqs. (13), (18), (25) and
(29), respectively. Note that it can be readily veried with the help of Eqs. (9)(18) that the value of the
determinant D = [B[ is zero when the frequency (x) is zero. This known value of D = [B[ =0 at x = 0
(which corresponds to a beam with no inertial loading so that the beam is at rest), can always be used to
avoid any numerical problems of overow at zero frequency when computing the value of D. Thus, for any
2420 J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426
other (non-trivial) value of x, the expression for D given by Eq. (32) can be used in locating the natural
frequencies by successively tracking the changes of its sign.
4. Mode shapes
Once the natural frequency x
n
is found from Eqs. (31) and (32), the modal vector A (in which one el-
ement may be xed arbitrarily) can be found in the usual way by deleting one row of the sixth order de-
terminant, and solving for the ve remaining constants in terms of the arbitrarily chosen one. (Note that
this choice is wholly arbitrary for the present problem.)
Thus, if A
1
is chosen to be the one in terms of which the remaining constant A
2
A
6
are to be expressed, as
in the present case, the matrix Equation (28), will take the following reduced order form. Note that the
terms relating to A
1
are taken to the right-hand side:
0 1 0 1 0
a 0 b 0 c
k
a
0 k
b
0 k
c
a aC
ha
b

bS
b
b

bC
b
c cS
c
c cC
c
aS
ha

bC
b

bS
b
cC
c
cS
c
2
6
6
6
6
4
3
7
7
7
7
5
A
2
A
3
A
4
A
5
A
6
2
6
6
6
6
4
3
7
7
7
7
5
=
1
0
0
a aS
ha
aC
ha
2
6
6
6
6
4
3
7
7
7
7
5
A
1
: (37)
The symbolic computing package REDUCE REDUCE (Rayna, 1986) was further used to solve the above system of
equations, yielding after extensive algebraic manipulation, the following mode shape coecients in terms of
A
1
:
A
2
= (U
1
=v)A
1
;
A
3
= (U
2
=v)A
1
;
A
4
= (U
3
=v)A
1
;
A
5
= (U
4
=v)A
1
;
A
6
= (U
5
=v)A
1
;
(38)
where
U
1
= b

bl
2
f
3
S
b
c cl
2
f
1
S
c
a al
2
f
2
S
ha
; (39)
U
2
= a

bd
3
C
ha
C
b
a cs
2
C
ha
C
c


bl
3
e
3
S
b
a cs
1
S
ha
S
c


b cd
3
C
b
C
c
s
3
; (40)
U
3
= a

bd
3
C
ha
S
b
c cl
3
f
1
S
c
a al
3
f
2
S
ha


b cd
3
S
b
C
c
; (41)
U
4
= a

b(d
3
al
2
)C
ha
C
b
c cl
1
f
1
C
c
a

bd
1
S
ha
S
b
cl
1
e
1
S
c
a a
2
l
2


b
2
d
3
; (42)
U
5
= b

bl
1
f
3
S
b
c cl
1
f
1
S
c
a al
1
f
2
S
ha
(43)
with
f
1
= aC
ha


bC
b
; f
2
=

bC
b
cC
c
; f
3
= cC
c
aC
ha
; (44)
e
1
= a aS
ha
b

bS
b
; e
2
= b

bS
b
c cS
c
; e
3
= c cS
c
a aS
ha
; (45)
d
1
= al
3
bl
2
; d
2
= bl
1
cl
3
; d
3
= cl
1
al
2
; (46)
J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426 2421
s
1
= al
1
cl
2
; s
2
= al
2
cl
1
; s
3
= a a
2
l
2
c c
2
l
1
; (47)
v = a al
2
f
2
C
ha
al
2
e
2
S
ha


b c(d
3
cl
1
)C
b
C
c


b cd
2
S
b
S
c


b
2
d
3
c c
2
l
1
; (48)
and l
1
; l
2
and l
3
have already been dened in Eq. (35). Note that a; b; c, and k
a
; k
b
; k
c
and a;

b; c, and
C
ha
; S
ha
; C
b
; S
b
; C
c
; and S
c
appearing in Eqs. (44)(48) are given by Eqs. (13), (18),(24) and (29), but must be
calculated for the particular natural frequency x
n
at which the mode shape is required.
Thus, the mode shape of the bending-torsion coupled composite beam is given in explicit form by re-
writing Eqs. (15) and (16) with the help of Eqs. (17) and (38) in the form:
H(n) = A
1
( coshan R
1
sinhan R
2
cosbn R
3
sinbn R
4
coscn R
5
sincn); (49)
W(n) = A
1
(k
a
coshan R
1
k
a
sinhan R
2
k
b
cosbn R
3
k
b
sinbn R
4
k
c
coscn R
5
k
c
sincn); (50)
where the ratios R
1
, R
2
, R
3
, R
4
and R
5
are respectively A
2
=A
1
; A
3
=A
1
; A
4
=A
1
; A
5
=A
1
and A
6
=A
1
, and follow
from Eq. (38).
5. Results
In order to validate and conrm the accuracy of the theory, the exact expressions for the frequency
equation and mode shapes given by Eqs. (32), (49) and (50) were programmed in Fortran to compute the
natural frequencies and mode shapes of a cantilever composite beam. The illustrative example is chosen
from the work of Minguet (1989) and Minguet and Dugundj (1990), which was also used by Banerjee and
Williams (1995). It is a at rectangular cross-section carbon-epoxy composite beam with stacking sequence
[45

=0

[
s
, length=0.56 m, width =0.03 m and thickness =0.00054 m. The rigidities and other properties are
taken from Minguet (1989) as EI =0.0143 Nm
2
, GJ =0.0195 Nm
2
, K=0.00632 Nm
2
, m=0.0238 kg/m, and
I
a
= 1:66 10
6
kgm. Using the frequency equation (32) and computing the roots of f (x) = 0, the rst ve
natural frequencies of the composite beam with cantilever end condition were computed. These are shown
in Table 1, alongside the iterative nite dierence results reported by Minguet (1989). The frequencies from
the present theory agree completely with the (exact) dynamic stiness results of Banerjee and Williams
(1995) and are within 1.1% of the results given by Minguet (1989) as can be seen. The dependency of f (x)
on x is shown in Fig. 2, identifying the rst two of the natural frequencies which satisfy the condition
f (x) =0. The author could have used the same illustrative example as that of Dancila and Armanios
(1998), but this was not possible because the composite beam data that were needed for the analysis were
not provided.
Table 1
Coupled bendingtorsional natural frequencies of a cantilever composite beam
Frequency number Natural frequency (rad/s)
Present theory Minguet (1989)
1 8.040 8.040
2 50.39 50.32
3 141.0 141.4
4 276.0 279.0
5 304.3 306.0
2422 J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426
The mode shapes of the above composite beam were computed using the explicit expressions for bending
displacement H(n), and torsional rotation W(n) given by Eqs. (49) and (50), respectively. The rst two
normal modes together with their corresponding natural frequencies are shown in Fig. 3. These results
agreed very well with the results reported by Minguet (1989), and complete agreement was found with the
dynamic stiness results of Banerjee and Williams (1995). Note that the frequencies shown in the paren-
theses are taken from the work of Minguet (1989).
Fig. 3. The rst two natural frequencies and mode shapes of the cantilever composite beam [45

=0

[
s
. (The natural frequencies shown
within parentheses are from Minguet, 1989.)
Fig. 2. Variation of f (x) against x.
J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426 2423
6. Conclusions
Analytical expressions for the frequency equation and mode shapes of a bendingtorsion (materially)
coupled composite beam with cantilever end condition have been derived in explicit form using the sym-
bolic computation package REDUCE REDUCE. The theory used is exact and there are no assumptions made en route
so that the natural frequencies and mode shapes can be found to any desired accuracy. The applicability of
the theory is demonstrated by numerical results, which show good agreement with published results. Al-
though the illustrative example given in this paper is that of a at composite beam, the theory can very well
be applied to a composite beam with an arbitrary cross-section provided the rigidity and other properties of
the beam are known. The proposed theory can be extended to other end conditions of the beam, and it
oers prospects for aeroelastic optimisation in a computationally ecient manner. The value judgement of
using a symbolic reduction package is at variance with a recently published paper whose shortcomings are
exposed.
Appendix A
Eq. (6) of the text is rewritten as
(D
6
aD
4
bD
2
abc)W = 0: (A:1)
Let W = e
pn
be the trial solution of (A.1). The auxiliary equation is then given by
p
6
ap
4
bp
2
abc = 0: (A:2)
Substituting
k = p
2
; (A:3)
eq. (A.2) can be reduced to a cubic equation to give
k
3
ak
2
bk abc = 0: (A:4)
This is of the form (Pipes and Harvill, 1971)
x
3
qx r = 0; (A:5)
where
x = k a=3;
q = b a
2
=3;
r = a(bc b=3 2a
2
=27):
(A:6)
Let
d = 27r
2
4q
3
: (A:7)
If d < 0, all the three roots of Eq. (A.5) are real with one of them positive and the other two negative
(Dancila and Armanios, 1998). For this case, the roots are given by (Pipes and Harvill, 1971)
x
1
= 2(q=3)
1=2
cos(/=3);
x
2
= 2(q=3)
1=2
cos(p /)=3;
x
3
= 2(q=3)
1=2
cos(p /)=3;
(A:8)
where
2424 J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426
cos/ = (3=q)
3=2
(r=2): (A:9)
Now, to satisfy the condition d < 0, the expressions for q and r from Eq. (A.6) are substituted into Eq.
(A.7) to obtain an expression for d which after simplication gives
d = 27r
2
4q
3
= 4b(a
2
c
2
b)
2
4a
4
bc(1 c
3
) a
2
b
2
(1 c
2
) 18a
2
b
2
c(1 c): (A:10)
Given the fact that a, b and c are always positive and c is always less than 1 (Eqs. (9)(11) of the main text),
it is quite clear from Eq. (A.10) that
d < 0: (A:11)
Simple substitution from Eq. (A.6) enables the three roots of k of Eq. (A.4) to be obtained by subtracting
respectively a/3 from each of the three roots of x given by Eq. (A.8). The square root of each of these three
roots (Eq. (A.3)) of k with a plus or a minus sign yields all the six roots of Eq. (A.2). So if
a; a; ib; ib; ic; ic are dened as the six roots with a; b and c being real, they are given by
a
2
= [2(q=3)
1=2
cos(/=3) a=3[;
b
2
= [2(q=3)
1=2
cos(p /)=3 a=3[;
c
2
= [2(q=3)
1=2
cos(p /)=3 a=3[;
(A:12)
where q and / have already been dened in Eqs. (A.6) and (A.9).
The solution of the dierential equation (A.1) is, thus,
W (n) = C
1
coshan C
2
sinhan C
3
cosbn C
4
sinbn C
5
coscn C
6
sincn (A:13)
with C
1
C
6
as constants.
References
Armanios, E.A., Badir, A.M., 1995. Free vibration analysis of anisotropic thin-walled closed-section beams. AIAA J. 33, 19051910.
Banerjee, J.R., Williams, F.W., 1995. Free vibration of composite beams-an exact method using symbolic computation. J. Aircraft 32,
636642.
Berdichevsky, V., Armanios, E., Badir, A., 1992. Theory of anisotropic thin-walled closed-cross-section beams. Compos. Engng. 2,
411432.
Chandra, R., Stemple, A.D., Chopra, I., 1990. Thin-walled composite beams under bending, torsional, and extensional loads.
J. Aircraft 27, 619626.
Dancila, D.S., Armanios, E.A., 1998. The inuence of coupling on the free vibration of anisotropic thin-walled closed-section beams.
Int. J. Solid Struct. 35, 31053119.
Farghaly, S.H., Shebl, M.G., 1995. Exact frequency and mode shape formulae for studying vibration and stability of Timoshenko
beam system. J. Sound Vibrat. 180, 205227.
Hodges, D.H., Atilgan, A.R., Fulton, M.V., Reheld, L.W., 1991. Free vibration analysis of composite beams. J. Amer. Helicop. Soc.
36, 3647.
Kapania, R.K., Raciti, S., 1989. Recent advances in analysis of laminated beams and plates, part I: shear eects and buckling, part II:
vibrations and wave propagation. AIAA J. 27, 923946.
Marus, S.R., Kant, T., 1996. Vibration analysis of bre reinforced composite beams using higher order theories and nite Element
Modelling. J. Sound. Vibrat. 194, 337351.
Minguet, P., 1989. Static and dynamic behaviour of composite helicopter blades under large deection. Ph.D Dissertation, Department
of Aeronautics and Astronautics, Massachusetts Institute of Technology, TELAC Report 89-7A.
Minguet, P., Dugundji, J., 1990. Experiments and analysis for composite blades under large deection, part I: static behaviour, part II:
dynamic behaviour. AIAA J. 28, 15731588.
Pipes, L.A., Harvill, L.R., 1971. Applied Mathematics for Engineers and Physicists. McGraw-Hill, New York.
Rayna, G., 1986. REDUCE software for algebraic computation. Springer-Verlag, New York.
J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426 2425
Reheld, L.W., Atilgan, A.R., Hodges, D.H., 1990. Non-classical behaviour of thin-walled composite beams with closed cross-
sections. J. Amer. Helicop. Soc. 35, 4250.
Shi, G., Lam, K.Y., 1999. Finite element vibration analysis of composite beams based on higher-order beam theory. J. Sound and
Vibrat. 219, 707721.
Smith, E.C., Chopra, I., 1991. Formulation and evaluation of an analytical model for composite box beams. J. Amer. Helicop. Soc. 36,
2335.
Song, O., Librescu, L., 1991. Free vibration and aeroelastic divergence of aircraft wings modelled as composite thin-walled beams.
Proceedings of the AIAA/ASME/ASCE/AHS/ASC 32nd Structures, Structural Dynamics, and Materials Conference, Baltimore,
MD, Paper no. AIAA-91-1187, 21282136.
Tse, F.S., Morse, I.E., Hinkle, R.T., 1978. Mechanical Vibrations: Theory and Applications, second ed. Allyn and Bacon, London.
Weisshaar, T.A., Foist, B.L., 1985. Vibration tailoring of advanced composite lifting surfaces. J. Aircraft 22, 141147.
White, M.W.D., Heppler, G.R., 1995. Vibration modes and frequencies of Timoshenko beams with attached rigid bodies. ASME
J. Appl. Mech. 62, 193199.
Wu, X.X., Sun, C.T., 1991. Vibration analysis of laminated composite thin-walled beams using nite elements. AIAA J. 29, 736742.
2426 J.R. Banerjee / International Journal of Solids and Structures 38 (2001) 24152426

You might also like