Arch Appl Mech (2011) 81: 263281

DOI 10.1007/s00419-010-0410-2
ORIGINAL
A. Ariaei S. Ziaei-Rad M. Ghayour
Transverse vibration of a multiple-Timoshenko beam
system with intermediate elastic connections due
to a moving load
Received: 25 April 2009 / Accepted: 13 January 2010 / Published online: 6 February 2010
Springer-Verlag 2010
Abstract Based on Timoshenko beam theory, the dynamic response of an elastically connected multiple-
beam system is investigated. The identical prismatic beams are assumed to be parallel and connected by a
nite number of springs. Assuming n parallel Timoshenko beams, the motion of the system is described by a
coupled set of 2n partial differential equations. The method involves a change of variables and modal analysis
to decouple and to solve the governing differential equations, respectively. A case study is solved in detail to
demonstrate the methodology and several plots of the midpoint deections of beams are given and investigated
for different values of moving load velocity and the stiffness of elastic connections. Fromthe numerical results
it is observed that the maximum deection of the multiple Timoshenko beam system is always smaller than
one of a single beam.
Keywords Multiple-beam system Timoshenko beams Moving load Transfer matrix method
1 Introduction
For more than a century, the analysis of continuous elastic systems subjected to moving loads has been the
subject of interest in many diverse elds such as civil and aerospace engineering [1]. In the past few decades,
much attention has been drawn to the vibration of single one-dimensional continuous systems such as beams
[14]. An important technological extension of the concept of the single beam is that of the double-beam
system such as double-beam cranes, double-beam spectrometers, double-beam interferometers, etc. Also,
the works in references [57] devoted to applying a double-beam system as a continuous dynamic vibration
absorber (CDVA) are especially interesting because of the great practical importance of CDVAs in many elds
of civil and mechanical engineering. As a complex continuous system consisting of two one-dimensional
solids joined by elastic medium, the elastically connected double-beam system has attracted great interest and
different aspects of its dynamics have been investigated by many researchers [519].
With arbitrary boundary conditions and forcing functions, the problem is difcult to solve. Under certain
conditions, though, the problembecomes tractable. Closed-formsolutions for the forced response of undamped
double-beamsystems can be obtained under specialized cases. The forced vibration of two beams connected at
two discrete points by two spring-dashpot subunits was considered by Dublin and Friedrich [14]. Subsequently,
Seelig and Hoppmann II [15] presented solution for differential equations of motion of an elastically connected
double-beam system subjected to an impulsive load. Rao [16] considered the free response of Timoshenko
beam systems. Vu et al. [17] presented an exact method for solving the vibration of a double-beam system
subjected to a harmonic excitation. The studied system consists of a main beam with an applied force, and
an auxiliary beam, with a distributed spring and dashpot in parallel between the two beams. Oniszczuk [18]
analyzed the undamped forced transverse vibration of an elastically connected simply supported double-beam
A. Ariaei S. Ziaei-Rad (B) M. Ghayour
Department of Mechanical Engineering, Isfahan University of Technology, 84156-83111 Isfahan, Iran
E-mail: szrad@cc.iut.ac.ir
264 A. Ariaei et al.
system. The modal expansion method is applied to ascertain dynamic responses of beams due to arbitrarily dis-
tributed continuous loads. Abu-Hilal [19] studied the dynamic response of a double-beam system traversed by
a constant moving load. The systemconsists of two elastic homogeneous isotropic Euler-Bernoulli beams. The
two simply supported prismatic beams are identical, parallel one upon the other, and connected continuously
by a viscoelastic layer.
This paper presents a unique yet method of obtaining the exact solution for the vibration of an undamped
multi-beam system due to a moving load. The method involves a change of variables to decouple the set
of multiple groups of two second-order differential equations, i.e., the Timoshenko beam theory. The three
restrictions of this method are (1) the beams must be identical, (2) the boundary conditions on the same side
of the system must be the same, though they can be arbitrary, and (3) the stiffness of intermediate springs
should have a special relation for decoupling the equation of motion that will be discussed later. All arti-
cles mentioned earlier considered a double-beam system with EulerBernoulli beam theory for the dynamic
analysis. The contribution of this article is to consider a multiple beam system subjected to a moving load
and is the extension of work carried on reference [20] which studied the free vibration of a single beam with
intermediate exible constrains. A further feature is using of Timoshenko beam theory. To the best of author
knowledge, the previous studies concerning with the double beam systems and all were formulated based on
the Euler-Bernoulli beam theory. Also, a novel method that shows the way to an eigenvalue problem is used
to dene new variables. The equations of motion are decoupled by using of these variables and therefore a set
of separate groups of two second-order differential equations are obtained and then the solution is obtained by
means of modal analysis. To demonstrate the technique in detail, a case study is chosen: the three beams are
simply supported and a load is moving on one of them. Several plots of the midpoint deections of the beams
are given and different values of moving load velocity and the stiffness of elastic connections are probed.
2 Formulation of the problem
2.1 Governing equations of a Timoshenko beam under a moving load
A multiple-Timoshenko beam system of length L for each beam and with m intermediate exible connections
is considered as in Fig. 1. It is assumed that these connections are located at points X
1
, X
2
, . . ., X
m
such
that 0 < X
1
< X
2
< < X
m
< L. The vibration amplitude of the transverse displacement of the i th
beam is denoted by Y
i j
(X, T) on the interval X
( j 1)
X X
j
; where the sub-index j represents the j th
segment and j = 1, 2, . . . , m + 1 (See Fig. 1). Each beam is now divided into m + 1 segments with lengths
L
1
, L
2
, . . . , L
m+1
, respectively, which are separated by m intermediate connections. By using Timoshenko
beam theory, the equation of motion for each segment of i th beam, assumed to have uniform cross section, is
A

2
Y
i j
T
2
AG
_

2
Y
i j
X
2


i j
X
_
= F
0
(X (T))
ir
(1a)
EI

2

i j
X
2
+ AG
_
Y
i j
X

i j
_
I

2

i j
T
2
= 0 X
( j 1)
< X < X
j
(1b)
where Y
i j
and
i j
are deection and rotation of segment j of beam i, respectively. Also, is the beam volu-
metric density, I is the cross-sectional moment of inertia, A is the cross sectional area, E is Youngs modulus
of elasticity, G is the shear modulus, and is the shear correction factor in Timoshenko beamtheory which is a
function of the cross-section and the Poisson ratio and (X (T)) denotes the Dirac delta function which
denes the moving load location. For the case of the moving load with the constant velocity of V, (T) is
equal to VT in which parameter T represents the time. All beams have the same characteristics. Also,
ir
is
dened as
_

ir
= 1 if i = r

ir
= 0 if i = r
(2)
where r is the number of beam that is subjected to a moving load. The boundary conditions are arbitrary but
similar for all beams. For example, for the simply supported case they are
Y(0, T) =

(0, T) = 0 (3a)
Y(L, T) =

(L, T) = 0 (3b)
Other types of boundary conditions can be treated easily as expressed in the Sect. 2.5 of the paper.
Transverse vibration of a multiple-Timoshenko beam system 265
Fig. 1 Multiple-Timoshenko beam system subjected to a moving load
2.2 Intermediate exible connections modeling
In this paper, each beam is divided into m+1 segments which are separated by m intermediate connections. It
leads to discontinuities at each spring location in shear force proportional to vertical displacement. Assuming
Y
0 j
as the base deection and the stiffness k
(n+1) j
equal to zero (see Fig. 1), the compatibility requirements
can be expressed as
EI

i ( j +1)
_
X
+
j
, T
_
= EI

i j
_
X

j
, T
_
(4a)
_
AG
_
Y

i ( j +1)
_
X
+
j
, T
_

i ( j +1)
_
X
+
j
, T
___

_
AG
_
Y

i j
_
X

j
, T
_

i j
_
X

j
, T
___
= k
i j
_
Y
i j
_
X

j
, T
_
Y
(i 1) j
_
X

j
, T
__
+k
(i +1) j
_
Y
i j
_
X

j
, T
_
Y
(i +1) j
_
X

j
, T
__
(4b)
Y
i ( j +1)
_
X
+
j
, T
_
= Y
i j
_
X

j
, T
_
(4c)

i ( j +1)
_
X
+
j
, T
_
=
i j
_
X

j
, T
_
i = 1, 2, . . . , n j = 1, 2, . . . , m (4d)
where X
+
j
and X

j
denote the locations immediately next and before the springs position X
j
, respectively. By
introducing the following quantities to the above equations
y
i j
=
Y
i j
L
, x =
X
L
, l
j
=
L
j
L
, x
j
=
X
j
L
, t =
T

L
, v =
V

L
, (t ) =
(T)
L
(5)
Thus, for each segment, Eqs. (1a, 1b) can then be expressed as
A

2
y
i j
t
2

AG
L
_

2
y
i j
x
2


i j
x
_
=
F
0
L
(x (t ))
ir
(6a)
EI
L
3

i j
x
2
+
AG
L
_
y
i j
x

i j
_

I
L
2

i j
t
2
= 0 x
( j 1)
< x < x
j
(6b)
266 A. Ariaei et al.
The non-dimensional compatibility conditions from Eqs. (4a4d) are now

i ( j +1)
_
x
+
j
, t
_
=

i j
_
x

j
, t
_
(7a)
_
y

i ( j +1)
_
x
+
j
, t
_

i ( j +1)
_
x
+
j
, t
__

_
y

i j
_
x

j
, t
_

i j
_
x

j
, t
__
= k
i j
L
AG
_
y
i j
_
x

j
, t
_
y
(i 1) j
_
x

j
, t
__
+k
(i +1) j
L
AG
_
y
i j
_
x

j
, t
_
y
(i +1) j
_
x

j
, t
__
(7b)
y
i ( j +1)
_
x
+
j
, t
_
= y
i j
_
x

j
, t
_
(7c)

i ( j +1)
_
x
+
j
, t
_
=
i j
_
x

j
, t
_
i = 1, 2, . . . , n j = 1, 2, . . . , m (7d)
2.3 Decoupling of equations
The system studied in this paper consists of n identical beams. In general, coupled partial equations are dif-
cult to solve. However, with a suitable manipulation in variables, the equations can be uncoupled and modal
analysis can be used to determine the solution. For this reason, new variables are dened as
w
p
(x, t ) =
n

i =1
c
pi
y
i
(x, t ) (8a)

p
(x, t ) =
n

i =1
c
pi

i
(x, t ) p = 1, 2, . . . , n (8b)
where the coefcients c
pi
should be determined in order to decouple the equations. Now, these coefcients
are exerted to Eqs. (6a, 6b) as below
n

i =1
c
pi
_
A

2
y
i j
t
2

AG
L
_

2
y
i j
x
2


i j
x
__
=
n

i =1
c
pi
F
0
L
(x (t ))
ir
(9a)
n

i =1
c
pi
_
EI
L
3

i j
x
2
+
AG
L
_
y
i j
x

i j
_

I
L
2

i j
t
2
_
= 0 x
( j 1)
< x < x
j
(9b)
Considering the Eqs. (8a, 8b), one can obtain
A

2
w
pj
t
2

AG
L
_

2
w
pj
x
2


pj
x
_
= c
pr
F
0
L
(x (t )) (10a)
EI
L
3

pj
x
2
+
AG
L
_
w
pj
x

pj
_

I
L
2

pj
t
2
= 0 x
( j 1)
< x < x
j
(10b)
These coefcients are also exerted to the compatibility Eqs. (7a7d) to get
n

i =1
c
pi

i ( j +1)
_
x
+
j
, t
_
=
n

i =1
c
pi

i j
_
x

j
, t
_
(11a)
n

i =1
c
pi
__
y

i ( j +1)
_
x
+
j
, t
_

i ( j +1)
_
x
+
j
, t
__

_
y

i j
_
x

j
, t
_

i j
_
x

j
, t
___
=
n

i =1
c
pi
_
k
i j
L
AG
_
y
i j
_
x

j
, t
_
y
(i 1) j
_
x

j
, t
__
+ k
(i +1) j
L
AG
_
y
i j
_
x

j
, t
_
y
(i +1) j
_
x

j
, t
__
_
(11b)
Transverse vibration of a multiple-Timoshenko beam system 267
n

i =1
c
pi
y
i ( j +1)
_
x
+
j
, t
_
=
n

i =1
c
pi
y
i j
_
x

j
, t
_
(11c)
n

i =1
c
pi

i ( j +1)
_
x
+
j
, t
_
=
n

i =1
c
pi

i j
_
x

j
, t
_
p = 1, 2, . . . , n j = 1, 2, . . . , m (11d)
Therefore, the new compatibility equations are in the following form:

p( j +1)
_
x
+
j
, t
_
=

pj
_
x

j
, t
_
(12a)
_
w

p( j +1)
_
x
+
j
, t
_

p( j +1)
_
x
+
j
, t
__

_
w

pj
_
x

j
, t
_

pj
_
x

j
, t
__
=
n

i =1
c
pi
_
k
i j
L
AG
_
y
i j
_
x

j
, t
_
y
(i 1) j
_
x

j
, t
__
+ k
(i +1) j
L
AG
_
y
i j
_
x

j
, t
_
y
(i +1) j
_
x

j
, t
__
_
(12b)
w
p( j +1)
_
x
+
j
, t
_
= w
pj
_
x

j
, t
_
(12c)

p( j +1)
_
x
+
j
, t
_
=
pj
_
x

j
, t
_
p = 1, 2, . . . , n j = 1, 2, . . . , m (12d)
The right-hand side of Eq. (12b) should be written in term of w
pj
. For this reason it is necessary that
n

i =1
c
pi
_
k
i j
L
AG
_
y
i j
_
x

j
, t
_
y
(i 1) j
_
x

j
, t
__
+k
(i +1) j
L
AG
_
y
i j
_
x

j
, t
_
y
(i +1) j
_
x

j
, t
__
_
=
pj
w
pj
_
x

j
, t
_
p = 1, 2, . . . , n j = 1, 2, . . . , m (13)
Substituting Eq. (8a) into Eq. (13) and after some rearrangements leads to
L
AG

k
1 j
+k
2 j
k
2 j
0 0 . . . 0
k
2 j
k
2 j
+k
3 j
k
3 j
0 . . . 0
0 k
3 j
k
3 j
+k
4 j
k
4 j
. . . 0
. . . . . . . . . . . . . . . . . .
. . . . . . 0 k
(n1) j
k
(n1) j
+k
nj
k
nj
0 . . . . . . 0 k
nj
k
nj

c
p1
c
p2
c
p3
. . .
c
p(n1)
c
pn

=
pj

c
p1
c
p2
c
p3
. . .
c
p(n1)
c
pn

p = 1, 2, . . . , n j = 1, 2, . . . , m (14)
One may rewrite Eq. (14) as

_
k
1 j
+k
2 j
_

pj
k
2 j
0 0 . . . 0
k
2 j
_
k
2 j
+k
3 j
_

pj
k
3 j
0 . . . 0
0 k
3 j
_
k
3 j
+k
4 j
_

pj
k
4 j
. . . 0
. . . . . . . . . . . . . . . . . .
. . . . . . 0 k
(n1) j
_
k
(n1) j
+k
nj
_

pj
k
nj
0 . . . . . . 0 k
nj
k
nj

pj

c
p1
c
p2
c
p3
. . .
c
p(n1)
c
pn

0
0
0
. . .
0
0

p = 1, 2, . . . , n j = 1, 2, . . . , m (15)
268 A. Ariaei et al.
where
p
is dened as

pj
=
AG
L

pj
(16)
Equation (15) is an eigenvalue problem. For nontrivial solution it is necessary that
det
_
K
j

pj
I
_
= 0 (17)
where K
j
is
K
j
=

_
k
1 j
+k
2 j
_
k
2 j
0 0 . . . 0
k
2 j
_
k
2 j
+k
3 j
_
k
3 j
0 . . . 0
0 k
3 j
_
k
3 j
+k
4 j
_
k
4 j
. . . 0
. . . . . . . . . . . . . . . . . .
. . . . . . 0 k
(n1) j
_
k
(n1) j
+k
nj
_
k
nj
0 . . . . . . 0 k
nj
k
nj

j = 1, 2, . . . , m (18)
The stiffness matrix K
j
obtained from each column of springs in Fig. 1 may have different eigenvalues, but
they should have the same normalized eigenvectors to decouple the compatibility equations for all segments
of n parallel beams. This concept should be examined before starting the solution.
When the n eigenvalues of
pj
were obtained from Eq. (17), the n related normalized eigenvectors can
be determined from Eq. (15). These eigenvectors are the coefcients of Eqs. (8a,8b) for determining new
variables to decouple the equations of motion. Using of Eq. (13) in Eq. (12b), the compatibility equations can
be rewritten as

p( j +1)
_
x
+
j
, t
_
=

pj
_
x

j
, t
_
(19a)
_
w

p( j +1)
_
x
+
j
, t
_

p( j +1)
_
x
+
j
, t
__

_
w

pj
_
x

j
, t
_

pj
_
x

j
, t
__
=
pj
w
pj
_
x

j
, t
_
(19b)
w
p( j +1)
_
x
+
j
, t
_
= w
pj
_
x

j
, t
_
(19c)

p( j +1)
_
x
+
j
, t
_
=
pj
_
x

j
, t
_
p = 1, 2, . . . , n j = 1, 2, . . . , m (19d)
Equations (10a,10b) with the compatibility Eqs. (19a19d) create a new system with n decoupled Timoshenko
beams. These equations have similar form and thus a unify method can be used to solve all of them. Figure 2
shows a schematic of this new system.
2.4 Eigenvalues and eigenfunctions of the multi-span Timoshenko beam
In previous section, by using a suitable manipulation of variables, n separated equations of motion in the
form of Eqs. (10a,10b) are obtained that each of them should be solved by considering related compatibility
Eqs. (19a19d). In the other words, solution of multiple Timoshenko beam system reduces to solution of n
new separated Timoshenko beam equations with intermediate exible constrains. The eigensolutions of the
multi-span beam are derived letting the forcing term to be zero. The solutions for the other boundary condi-
tions can also be obtained easily through a similar procedure. Using some rearrangement and by assuming
separable solutions w
pj
(x, t ) = u
pj
(x)e
i
p
t
and
pj
(x, t ) =
pj
(x)e
i
p
t
in Eqs. (10a, 10b) one can lead to
an associated eigenvalue problem for the pth set of equations [21],
u
i v
pj
(x) +(
p
+
p
)u

pj
(x) (
p

p

p
)u
pj
(x) = 0 (20a)

i v
pj
(x) +(
p
+
p
)

pj
(x) (
p

p

p
)
pj
(x) = 0 x
i 1
< x < x
i
(20b)
where

p
=
L
2
p
E
,
p
=
L
2
p
G
,
p
=
AL
3

2
p
EI
(21)
Transverse vibration of a multiple-Timoshenko beam system 269
Fig. 2 New system with n decoupled Timoshenko beams
A closed-form solution to this eigenvalue problem can be obtained by employing transfer matrix methods
[2123]. The general solutions of Eqs. (10a, 10b), for each segment, are [21]
u
pj
(x) = A
pj
cosh
1p
(x x
j 1
) + B
pj
sinh
1p
(x x
j 1
)
+C
pj
cos
2p
(x x
j 1
) + D
pj
sin
2p
(x x
j 1
) (22a)

pj
(x) = B
pj
q
1p
cosh
1p
(x x
j 1
) + A
pj
q
1p
sinh
1p
(x x
j 1
)
D
pj
q
2p
cos
2p
(x x
j 1
) +C
pj
q
2p
sin
2p
(x x
j 1
)
x
j 1
< x < x
j
(22b)
where

1p
=

_
_

p

p
2
_
2
+
p


p
+
p
2

1/2
,
2p
=

_
_

p

p
2
_
2
+
p
+

p
+
p
2

1/2
,

3p
=

p
(23)
q
1p
=
_

2
3p
+
2
1p
_

1p
, q
2p
=
_

2
3p

2
2p
_

2p
A
pj
, B
pj
, C
pj
and D
pj
are constants associated with the j th segment of pth set of equations. The con-
stants in the last segment (A
p(m+1)
, B
p(m+1)
, C
p(m+1)
and D
p(m+1)
) are related to those in the rst seg-
ment (A
p1
, B
p1
, C
p1
and D
p1
) through the compatibility requirements in Eqs. (19a19d) which can be
expressed as
270 A. Ariaei et al.

A
m+1
B
m+1
C
m+1
D
m+1

p
=

t
11
t
12
t
13
t
14
. . . . . .
. . . . . .
t
41
. . . . . . t
44

pm

A
m
B
m
C
m
D
m

p
=

t
11
t
12
t
13
t
14
. . . . . .
. . . . . .
t
41
. . . . . . t
44

pm

t
11
t
12
t
13
t
14
. . . . . .
. . . . . .
t
41
. . . . . . t
44

p(m1)
. . .

t
11
t
12
t
13
t
14
. . . . . .
. . . . . .
t
41
. . . . . . t
44

p1

A
1
B
1
C
1
D
1

p
=
_
T
44
_
pm
_
T
44
_
p(m1)

_
T
44
_
p1

A
1
B
1
C
1
D
1

p
=
_
T
44
_
p

A
1
B
1
C
1
D
1

p
(24)
where (T
44
)
pj
is a 4 4 transfer matrix which depends on the eigenvalue
p
, the elements of which are
derived as
(T
44
)
pj
=

cosh
_

1
l
j
_
sinh
_

1
l
j
_

j
q
2
cosh(
1
l
j )

1
q
2
+
2
q
1
+sinh
_

1
l
j
_

j
q
2
sinh(
1
l
j )

1
q
2
+
2
q
1
+cosh
_

1
l
j
_
0 0

j
q
1
cosh(
1
l
j )

1
q
2
+
2
q
1

j
q
1
sinh(
1
l
j )

1
q
2
+
2
q
1
0 0

j
q
2
cos (
2
l
j )

1
q
2
+
2
q
1

j
q
2
sin(
2
l
j )

1
q
2
+
2
q
1
cos
_

2
l
j
_
sin
_

2
l
j
_

j
q
1
cos (
2
l
j )

1
q
2
+
2
q
1
sin
_

2
l
j
_

j
q
1
sin(
2
l
j )

1
q
2
+
2
q
1
+cos
_

2
l
j
_

p
(25)
Through the application of Eq. (24), the four constants in the rst segment (A
p1
, B
p1
, C
p1
and D
p1
) can
be mapped onto those of the last segment, thereby reducing the number of independent constants to four. These
four remaining constants can be determined through the satisfaction of the boundary conditions.
For the case of a simply supported beam, the corresponding boundary conditions of Eqs. (3a, 3b) can thus
be expressed as
Y
i 1
(0, T) = 0 w
p1
(0, T) = 0 u
p1
(0) = 0 (26a)

i 1
(0, T) = 0

p1
(0, T) = 0

p1
(0) = 0 (26b)
Y
i (m+1)
(L, T) = 0 w
p(m+1)
(1, T) = 0 u
p(m+1)
(1) = 0 (26c)

i (m+1)
(L, T) = 0

p(m+1)
(1, T) = 0

p(m+1)
(1) = 0 (26d)
Satisfying the boundary conditions at the left support leads to
A
p1
= C
p1
= 0 (27)
By satisfying the boundary conditions at the right support, one can be obtained that
(S
24
)
p

A
m+1
B
m+1
C
m+1
D
m+1

p
=
_
0
0
_
(28a)
where
S
p
=
_
cosh
1p
l
m+1
sinh
1p
l
m+1
cos
2p
l
m+1
sin
2p
l
m+1
q
1p

1p
cosh
1p
l
m+1
q
1p

1p
sinh
1p
l
m+1
q
2p

2p
cos
2p
l
m+1
q
2p

2p
sin
2p
l
m+1
_
(28b)
Transverse vibration of a multiple-Timoshenko beam system 271
Substitution of Eq. (24) into Eq. (28a) and use of Eq. (27) leads to
S
p

A
m+1
B
m+1
C
m+1
D
m+1

p
= S
p
T
p

A
1
B
1
C
1
D
1

p
= (R
24
)
p

0
B
1
0
D
1

p
=
_
0
0
_
(29a)
where
R
p
= S
p
T
p
=
_
r
11
r
12
r
13
r
14
r
21
r
22
r
23
r
24
_
p
(29b)
Therefore, the existence of non-trivial solutions requires
det

r
12
r
14
r
22
r
24

p
= 0 (30)
This determinant provides the single equation for the solution of eigenvalue
p
. This is a matrix of only
2 2 dimensions; therefore, it is possible to obtain the corresponding characteristic equation easily. The
eigenfunctions, u
pj
(x) and
pj
(x), are obtained by back substitution of the obtained natural frequency into
Eqs. (29), (24) and (22).
2.5 Boundary conditions
For the cases of usually used boundary conditions, through the similar procedure, the following relations can
be obtained:
Left boundary condition: The existence of non-trivial solutions for the constants A
1p
, B
1p
, C
1p
and D
1p
requires
(1) Free boundary condition:
det

r
11

q
1

1
q
2

2
r
13
r
12
+
q
1

1
q
2
+
2
r
14
r
21

q
1

1
q
2

2
r
23
r
22
+
q
1

1
q
2
+
2
r
24

p
= 0 (31)
(2) Simply-Supported boundary condition: The existence of non-trivial solutions is the same as Eq. (30).
(3) Clamped boundary condition:
det

r
11
r
13
r
12
+
q
1
q
2
r
14
r
21
r
23
r
22
+
q
1
q
2
r
24

p
= 0 (32)
Right boundary condition: in this case, the matrix [S
24
]
p
changes and will be expressed as follows
(1) Free boundary condition:
S
p
=
_
(
1
q
1
) sinh
1
l
m+1
(
1
q
1
) cosh
1
l
m+1
(
2
+q
2
) sin
2
l
m+1
(
2
+q
2
) cos
2
l
m+1
q
1

1
cosh
1
l
m+1
q
1

1
sinh
1
l
m+1
q
2

2
cos
2
l
m+1
q
2

2
sin
2
l
m+1
_
p
(33)
(2) Simply-Supported boundary condition: The matrix [S
24
]
p
is the same as Eq. (28b).
(3) Clamped boundary condition:
S
p
=
_
cosh
1
l
m+1
sinh
1
l
m+1
cos
2
l
m+1
sin
2
l
m+1
q
1
sinh
1
l
m+1
q
1
cosh
1
l
m+1
q
2
sin
2
l
m+1
q
2
cos
2
l
m+1
_
p
(34)
272 A. Ariaei et al.
2.6 Forced response
The equation of motion, i.e. Eqs. (10a, 10b), is rewritten again here:
A

2
w
pj
t
2

AG
L
_

2
w
pj
x
2


pj
x
_
= c
pr
F
0
L
(x (t )) (35a)
EI
L
3

pj
x
2
+
AG
L
_
w
pj
x

i j
_

I
L
2

pj
t
2
= 0 x
( j 1)
< x < x
j
(35b)
Using the modal expansion theory, the forced response w
p
(x, t ) and
p
(x, t ) for pth set of equations of new
system can be expressed as
w
p
(x, t ) =
N

k=1
u
kp
(x) p
kp
(t ) (36a)

p
(x, t ) =
N

k=1

kp
(x) p
kp
(t ) (36b)
where p
kp
(t )are the generalized coordinates (time functions) for the elastic deection and orientation of the
beam element. Functions u
kp
(x) and
kp
(x)are the respective transverse and rotational eigenfunctions (modal
shapes) of pth equations. Equations (36a, 36b) are written for through the beam and therefore the index j,
related to j th segment of the beam, is neglected. The use of same time functions, p
kp
(t ), in the Galerkin
approximation is a common practice and a standard assumption which has been utilized by many researchers
in other areas of mechanics. In reference [24], two different time functions were used for w
p
(x, t ) and
p
(x, t ),
respectively. The results show that the difference in obtained solutions is small and concluded that assuming
the same time functions for variables w
p
(x, t ) and
p
(x, t ) is reasonable. Substituting Eqs. (36a, 36b) into
Eqs. (35a, 35b), yields
N

k=1
_
Au
kp
(x) p
kp
(t )
AG
L
_
u

kp

kp
_
p
kp
(t )
_
= c
pr
F
0
L
(x (t )) (37a)
N

k=1
_
EI
L
3

kp
(x) p
k
(t ) +
AG
L
_
u

kp

kp
_
p
kp
(t )
I
L
2

kp
(x) p
kp
(t )
_
= 0 (37b)
On the other hand, from the free vibration analysis we have

AG
L
_
u

kp

kp
_
= A
2
kp
u
kp
(x) (38a)
EI
L
3

kp
(x) +
AG
L
_
u

kp

kp
_
=
I
L
2

2
kp

k
(x) (38b)
where
kp
is the kth natural frequency of the pth set of equations. Substituting Eqs. (38a, 38b) into Eqs. (37a,
37b) results in
A
N

k=1
u
kp
(x)
_
p
kp
(t ) +
2
kp
p
kp
(t )
_
= c
pr
F
0
L
(x (t )) (39a)
I
N

k=1

kp
(x)
_
p
kp
(t ) +
2
kp
p
kp
(t )
_
= 0 (39b)
In order to solve for p
kp
(t ) from Eqs. (39a, 39b), the orthogonality conditions [24] of
1
_
0
_
u
i p
(x)u
j p
(x) +
I
AL
2

i p
(x)
j p
(x)
_
dx =
i j
(40)
Transverse vibration of a multiple-Timoshenko beam system 273
are utilized, where i, j = 1, 2, . . . , N and
i j
is the Kronecker delta. Multiplying Eq. (39a) by u
qp
(x) and
Eq. (39b) by
qp
(x), adding each side together, integrating over the entire length of the beam, and with the
use of the orthogonality relationship (40), we get
p
qp
(t ) +
2
qp
p
qp
(t ) = c
pr
F
0
AL
1
_
0
u
qp
(x) (x (t )) dx = c
pr
F
0
AL
u
qp
((t )) = Q
qp
(t ) (41)
The generalized coordinates p
qp
(t ) are solved from Eq. (41) by the convolution integration method and can
be written as
p
qp
(t ) = p
qp
(0) cos
qp
t +
p
qp
(0)

qp
sin
qp
t +
1

qp
t
_
0
sin
qp
(t ) Q
qp
()d (42)
where
p
qp
(0) =
1
_
0
_
w
0p
(x)u
qp
(x) +
I
AL
2

0p
(x)
qp
(x)
_
dx (43a)
p
qp
(0) =
_
1
0
_
w
0p
(x)u
qp
(x) +
I
AL
2

0p
(x)
qp
(x)
_
dx (43b)
and w
0p
(x),
0p
(x), w
0p
(x),

0p
(x) are initial conditions of the new system and obtained from Eqs. (8a, 8b)
as below
w
0p
(x) =
n

i =1
c
pi
y
0i
(x) w
0p
(x) =
n

i =1
c
pi
y
0i
(x) (44a)

0p
(x) =
n

i =1
c
pi

0i
(x)

0p
(x) =
n

i =1
c
pi

0i
(x) p = 1, 2, . . . , n (44b)
where y
0i
(x),
0i
(x), y
0i
(x),

0i
(x) are initial conditions of the original system.
The eigenfunctions u
qp
(x) and
qp
(x) used in Eqs. (43a, 43b) are from Eqs. (22a, 22b) and which can
be divided into m + 1 different functions separated by the intermediate spring position x
j
(= X
j
/L), j =
1, 2, . . . m. Thus, the eigenfunction u
qp
(x) and
qp
(x) can be expressed as [21, 22]
u
qp
(x) = f
qp( j )
= A
qp( j )
cosh
1qp
(x x
j 1
) + B
qp( j )
sinh
1qp
(x x
j 1
)
+C
qp( j )
cos
2qp
(x x
j 1
) + D
qp( j )
sin
2qp
(x x
j 1
)
(45a)

qp
(x) = g
qp( j )
= B
qp( j )
q
1qp
cosh
1qp
(x x
j 1
) + A
qp( j )
q
1qp
sinh
1qp
(x x
j 1
)
D
qp( j )
q
2qp
cos
2qp
(x x
j 1
) +C
qp( j )
q
2qp
sin
2qp
(x x
j 1
)
x
j 1
< x < x
j
(45b)
Then, the generalized forcing term Q
qp
(t ) in Eq. (41) can be written as
Q
qp
(t ) = c
pr
F
0
AL
u
qp
((t )) = c
pr
F
0
AL
f
qp( j )
((t )) x
j 1
< (t ) < x
j
(46a)
The term
1

qp
_
t
0
sin
qp
(t ) Q
qp
()d in Eq. (42) can thus be expressed
1

qp
t
_
0
sin
qp
(t ) Q
qp
()d =
c
pr
F
0
AL
qp
t
_
0
sin
qp
(t ) f
qp( j )
[()] d
x
j 1
< () < x
j
(46b)
274 A. Ariaei et al.
Fig. 3 Three-Timoshenko beam system subjected to a moving load
Fig. 4 The rst two natural frequencies of the decoupled beams versus the stiffness of springs: lled circle the rst frequency of
beam 1, small open circle the rst frequency of beam 2, large open circle the rst frequency of beam 3, lled square the second
frequency of beam 1, small open square the second frequency of beam 2, large open square the second frequency of beam 3
The generalized coordinates p
qp
(t ) can be obtained from Eq. (42), and thus, the forced response solutions
w
p
(x, t ) and
p
(x, t ) can then be reconstructed from Eqs. (36a, 36b). Afterwards, by using of Eqs. (8a, 8b),
the response of original system is obtained. These equations are rewritten here in the matrix form as
W = CY (47a)
= C (47b)
where
W =
_
w
1
w
2
. . . w
n
_
T
, Y =
_
y
1
y
2
. . . y
n
_
T
=
_

1

2
. . .
n
_
T
, =
_

1

2
. . .
n
_
T
(48)
C =

c
11
c
12
. . . c
1n
c
21
c
22
. . . c
2n
. . . . . . . . . . . .
c
n1
c
n2
. . . c
nn

Transverse vibration of a multiple-Timoshenko beam system 275

Fig. 5 Mode shapes of the original system for k = 20 10
6
N m
1
related to rst natural frequencies of the decoupled beams:
solid line beam 1, dotted line beam 2, dashed line beam 3. (a
i
) Transverse mode shapes, (b
i
) angular mode shapes
Therefore, the original variables y
i
(x, t ) and
i
(x, t ) are obtained from Eqs. (49a, 49b) as below
Y = C
1
W (49a)
= C
1
(49b)
3 Results and discussion
Multiple similar simply supported beams (Fig. 3) with the rectangular cross sections are considered. Unless
stated otherwise, all numerical results presented in this section are based on the following numerical data:
L = 50 m, E = 2.1 10
11
N m
2
, = 0.3, = 0.85, m = 3930 kg m
1
, height = 1m, width = 0.5 m,
F
0
= 385.5 kN,L
1
/L = 0.3, L
2
/L = 0.4, k
11
= k, k
21
= 2k, k
31
= 3k, k
12
= 2k, k
22
= 4k, k
32
= 6k.
First, it should be examined if the normalized eigenvectors of matrices K
1
and K
2
written for two columns
of springs are similar. From Eq. (18), they can be obtained as
K
2
= 2K
1
= 2k

3 2 0
2 5 3
0 3 3

(50)
It is clear that the normalized eigenvectors of matrices K
1
and K
2
are the same and therefore, the method
described here can be used for this problem.
276 A. Ariaei et al.
Fig. 6 Mode shapes of the original system for k = 2010
6
N m
1
related to second natural frequencies of the decoupled beams:
solid line beam 1, dotted line beam 2, dashed line beam 3. (a
i
) Transverse mode shapes, (b
i
) angular mode shapes
Each decoupled beam has its special natural frequencies; however, all of them are the natural frequencies
of the original system. It is expected that by increasing the stiffness of the springs, the natural frequencies of
the system are increased. The results of the calculations are presented in Fig. 4. In this gure, the effect of the
stiffness parameter k on the natural frequencies of the systemis investigated. The natural frequencies indicated
by circle are the rst and those by square are the second natural frequencies of the three beams. It is notable
that all these frequencies are the natural frequencies of the original system. The values of the frequencies
increase by increasing the stiffness parameter k. For k = 0, there is no elastic connections between segments
and therefore both original and newsystems are similar. As it can be seen in Fig. 4, the rst and also the second
natural frequencies of all three beams coincide.
One mode shape is related to each natural frequency. By using of Eqs. (49a, 49b) and the mode shapes of
decoupled beams, the mode shapes of original system are extracted. Figures 5 and 6 show the mode shapes of
the system depicted in Fig. 3 by considering k = 20 10
6
N m
1
. The left column shows the transverse mode
shapes while the right column illustrates the rotational mode shapes. It can be seen that existence of elastic
connections has changed the characteristics of mode shapes such as the number and the location of nodes in
mode shapes.
Figures 5a
2
b
2
and 6a
2
b
2
show that the second beam remains at when the system vibrates with the
frequencies of 4.809 and 7.230Hz. Also, in these gures the ratio between the mode shape of the rst beam
to the mode shape of the third one at the corresponding frequencies is constant and equal to 1.5. This can be
claried by making a note of the stiffness of springs in Fig. 3. In this model, the ratio k
31
/k
21
= k
32
/k
22
= 1.5
and therefore, the pure force exerted on the second beam will vanish when the system vibrates in this mode
and thus, this beam remains rigid during the vibration.
Transverse vibration of a multiple-Timoshenko beam system 277
Fig. 7 Dimensionless mid-point deection versus the position of the moving load for V = V
cr
/4: solid line beam 1, dotted
line beam 2, dashed line beam 3. (a) k = 0.0, (b) k = 2.5 10
6
N m
1
, (c) k = 5, 10
6
N m
1
, (d) k = 10 10
6
Nm
1
,
(e) k = 100 10
6
N m
1
In Figs. 7, 8 and 9, x-axis represents the normalized position of the moving load and y-axis represents the
deection response, which has been normalized to the maximumstatic deection of an intact simply supported
Euler-Bernoulli beam that is dened as
y =
F
0
L
2
48EI
(51)
Also in these gures, V
cr
is the critical velocity and dened as [3]
V
cr
=

1
L

(52)
278 A. Ariaei et al.
Fig. 8 Dimensionless mid-point deection versus the position of the moving load for V = V
cr
/2: solid line beam 1, dotted line
beam 2, dashed line beam 3. (a) k = 0.0, (b) k = 2.5 10
6
N m
1
, (c) k = 5 10
6
N m
1
, (d) k = 10 10
6
N m
1
,
(e) k = 100 10
6
N m
1
where
1
is

1
=
_
EI
m
_

L
_
2
(53)
Figure 7 shows the dimensionless midpoint deections for beams in the original system when V = V
cr
/4.
The parameter r in this gure indicates the beam number, on which the load is traveling. From Fig. 7a it is
observable that when k = 0, the deections of central points of beams that the load is moving on are identical.
That is reasonable because all beams have the same characteristics. In this case, other beams remain rigid
because of the zero value of k. Also, this gure shows that increasing the values of stiffness parameter k
Transverse vibration of a multiple-Timoshenko beam system 279
Fig. 9 Dimensionless mid-point deection versus the position of the moving load for V = V
cr
: solid line beam 1, dotted line
beam 2, dashed line beam 3. (a) k = 0.0, (b) k = 2.5 10
6
N m
1
, (c) k = 5 10
6
N m
1
, (d) k = 10 10
6
N m
1
,
(e) k = 100 10
6
N m
1
decreases the central point deection of the beam on which the load is traveling. However, other beams have
a different scenario. The maximum midpoint deection of these beams increases by increasing the stiffness
parameter k, but for high values of k, it decreases to almost zero due to strong-coupling between beams. This
may justify by the fact that increasing the value of parameter k leads to an increase in the coupling between the
beams and the ground. Therefore, a very high value for k leads to a strong-coupling between beams. Further-
more, it is observed that the absolute maximum deection of the system increases by increasing the parameter
r. In other words, the absolute maximum deection of the system reduces in this case by the traveling of the
load on the beam which is closer to the ground.
Figure 8 shows the dimensionless deections for beams when the moving load travels at a speed of
V = V
cr
/ 2 and different values of stiffness parameter k. This gure shows that, in general, the system behaves
280 A. Ariaei et al.
as in the case for V = V
cr
/4. But from Fig. 8 it is observable that the absolute maximum values of the system
occur at the time when the load traversed about 60% of the beams length for relative small values of k, i.e.,
weak elastic coupling. However, it shifts a little to the left by increasing the stiffness parameter k. Similar
to Fig. 7, this gure shows that increasing the value of stiffness parameter k decreases the deection of the
beam on which the load is traveling. Furthermore, the absolute maximum deection of the system increases
by increasing the parameter r for a specic value of k. By a comparison between Fig. 8a and Fig. 8be, it is
also observed that the maximum deection of the multiple Timoshenko beam system is smaller than one of a
single beam.
Figure 9 shows the dimensionless deections for beams when V = V
cr
and different values of stiffness
parameter k. This gure shows that for very small (decoupled) values of the stiffness parameter k, the absolute
maximum deection of the beams is reached after the load departs from the beam. For the case of relative
high values of k, it shifts to the left and occurs short before the load left the beam. That is due to the fact that
a change in the stiffness parameter k leads to a change in the natural frequencies of the system. Again, this
gure shows that increasing the values of stiffness k decreases the deection of the beam on which the load is
traveling.
4 Conclusions
A closed form solution is developed for analyzing the vibration problem of an elastically connected multiple
Timoshenko beam system subjected to a moving load. A proper change in variables is applied to decouple the
equations of motion. Then, by using of the modal expansion method, the dynamic response of beams caused
by a moving load is received. From the numerical results it is observed that existence of elastic connections
has changed the characteristics of mode shapes such as the number of nodes. Also, it is seen that the maximum
deection of the multiple Timoshenko beam system is always smaller than one of a single beam.
From the obtained results the following characteristics can be noted:
1. The absolute maximum deection of the system reduces in this case when the load is traveling on the
closer beams to the ground. However, it may be depended to the stiffness of elastic connections.
2. The maximum deection of the multiple Timoshenko beam system is always smaller than one of a single
beam.
3. Increasing the values of the stiffness parameter k decreases the deection of the beam which is subjected
to the moving load. But for other beams, increasing the stiffness k leads to increase the maximummidpoint
deection for relative small values of k and decrease it to almost zero for strong-coupling of the beams.
References
1. Stokes, GG.: Discussion of a differential equation relating to the breaking of railway bridges. Trans. Cambridge Philos. Soc.
8 (1849), 707735; reprinted in Math. Phys. Papers, 2, 178220 (1883)
2. Timoshenko, S.: Vibration Problems in Engineering. Wiley, New York (1974)
3. Fryba, L.: Vibration of Solids and Structures under Moving Loads. Academia, Prague (1972)
4. Ziaei-Rad, S., Ariaei, A., Imregun, M.: Vibration analysis of Timoshenko beams under uniform partially distributed moving
masses. J. Multi-Body Dyn. 221, 551566 (2007)
5. Chen, Y.H., Lin, C.Y.: Structural analysis and optimal design of a dynamic absorbing beam. J. Sound Vib. 212, 759769
(1998)
6. Szczesniak, W.: Vibration of elastic sandwich and elastically connected double-beam system under moving loads. Scientic
Works of Warsaw University of Technology, Civil Engineering, vol. 132, pp. 111151 (1998)
7. Kawazoe, K., Kono, I., Aida, T., Aso, T., Ebisuda, K.: Beam-type dynamic vibration absorber comprised of free-free beam.
J. Eng. Mech. 124, 248258 (1998)
8. Seelig, J.M., Hoppmann, W.H. II.: Normal mode vibrations of systems of elastically connected parallel bars. J. Acoust. Soc.
Am. 36, 9399 (1964)
9. Kessel, P.G., Raske, T.F.: Damped response of an elastically connected double-beam system due to a cyclic moving load.
J. Acoust. Soc. Am. 49, 371373 (1971)
10. Chonan, S.: Dynamical behaviours of elastically connected double beam system subjected to an impulsive load. Trans. Jpn.
Soc. Mech. Eng. 41, 28152824 (1975)
11. Yamaguchi, H., Saito, H.: Vibrations of beams with an absorber consisting of a viscoelastic solid and a beam. Earthquake
Eng. Structur. Dyn. 12, 467479 (1984)
12. Kukla, S.: Free vibration of the system of two beams connected by many translational springs. J. Sound Vib. 172, 130135
(1994)
13. Oniszczuk, Z.: Free transverse vibrations of elastically connected simply supported double-beam complex system. J. Sound
Vib. 232(2), 387403 (2000)
Transverse vibration of a multiple-Timoshenko beam system 281
14. Dublin, M., Friedrich, H.R.: Forced responses of two elastic beams interconnected by spring-damper systems. J. Aeronaut.
Sci. 887, 824829 (1956)
15. Seelig, J.M., Hoppmann, W.H. II.: Impact on an elastically connected double-beam system. ASME J. Appl. Mech. 31,
621626 (1964)
16. Rao, S.S.: Natural vibrations of systems of elastically connected Timoshenko beams. J. Acoust. Soc. Am. 55, 12321237
(1974)
17. Vu, H.V., Ordonez, A.M., Karnopp, B.H.: Vibration of a double-beam system. J. Sound Vib. 229(4), 807822 (2000)
18. Oniszczuk, Z.: Forced transverse vibrations of an elastically connected complex simply supported double-beam system.
J. Sound Vib. 264, 273286 (2003)
19. Abu-Hilal, M.: Dynamic response of a double Euler-Bernoulli beam due to a moving constant load. J. Sound Vib. 297, 477
491 (2006)
20. Lin, H.P., Chang, S.C.: Free vibration analysis of multi-span beams with intermediate exible constraints. J. Sound
Vib. 281, 155169 (2005)
21. Lin, H.P.: Direct and inverse methods on free vibration analysis of simply supported beams with a crack. Eng. Struct. 26,
427436 (2004)
22. Loya, J.A., Rubio, L., Fernndez-Sez, J.: Natural frequencies for bending vibrations of Timoshenko cracked beams.
J. Sound Vib. 290, 640653 (2006)
23. Lin, H.P., Chang, S.C.: Forced responses of cracked cantilever beams subjected to a concentrated moving load. Int. J. Mech.
Sci. 48, 14561463 (2006)
24. Dadfarnia, M., Jalili, N., Esmailzadeh, E.: A comparative study of the Galerkin approximation utilized in the Timoshenko
beam theory. J. Sound Vib. 280, 11321142 (2005)