Acta Mech Sin (2010) 26:721733

DOI 10.1007/s10409-010-0365-0

RESEARCH PAPER

Assessing dynamic response of multispan viscoelastic thin beams

under a moving mass via generalized moving least square method
Keivan Kiani Ali Nikkhoo Bahman Mehri

Received: 12 June 2009 / Revised: 23 December 2009 / Accepted: 24 December 2009 / Published online: 8 July 2010
The Chinese Society of Theoretical and Applied Mechanics and Springer-Verlag GmbH 2010

Abstract Dynamic response of multispan viscoelastic thin

beams subjected to a moving mass is studied by an efficient
numerical method in some detail. To this end, the unknown
parameters of the problem are discretized in spatial domain
using generalized moving least square method (GMLSM)
and then, discrete equations of motion based on Lagranges
equation are obtained. Maximum deflection and bending
moments are considered as the important design parameters. The design parameter spectra in terms of mass weight
and velocity of the moving mass are presented for
multispan viscoelastic beams as well as various values of
relaxation rate and beam span number. A reasonable good
agreement is achieved between the results of the proposed
solution and those obtained by other researchers. The results
indicate that, although the load inertia effects in beams with
higher span number would be intensified for higher levels
of moving mass velocity, the maximum values of design
parameters would increase either. Moreover, the possibility
of mass separation is shown to be more critical as the span
number of the beam increases. This fact also violates the linear relation between the mass weight of the moving load and
the associated design parameters, especially for high moving
mass velocities. However, as the relaxation rate of the beam

K. Kiani A. Nikkhoo (B)

Department of Civil Engineering, Sharif University of Technology,
P.O. Box 11365-9313, Tehran, Iran
e-mail: nikkhoo@civil.sharif.edu
K. Kiani
e-mail: k_kiani@civil.sharif.edu
B. Mehri
Department of Mathematical Sciences,
Sharif University of Technology, Tehran, Iran
e-mail: mehri@sharif.edu

material increases, the load inertia effects as well as the

possibility of moving mass separation reduces.
Keywords Moving massbeam interaction Multispan
viscoelastic beam EulerBernoulli beam Generalized
moving least square method (GMLSM)

1 Introduction
Evaluating dynamic response of beam structures traversed by
moving loads (masses) has been remarkably paid attention
to by the scientific communities for more than a century. The
significance of this problem intensifies as the inertia effects
of moving loads are taken into account, especially for high
values of load magnitude and velocity. Usually, the overall
structural behaviour of bridge structures excited by moving
loads (masses) could be simulated as single or multispan
beams under such loading patterns.
Regarding the problem of solids subjected to moving
loads, a large quantity of research was provided by Frba
[1], specifically for single and multispan beam structures.
However, the proposed analytical solutions to the aforementioned mechanical problem are mostly restricted to the moving load cases. The existence of a moving mass makes the
problem very difficult because the interaction force between
the moving mass and the base beam varies in time and space.
Exploring a single span beam problem under a moving mass
has resulted in many publications using miscellaneous methods. In this regard, Akin and Mofid [2] studied the dynamic
response of an EulerBernoulli beam excited by a moving
mass considering various boundary conditions of the beam
utilizing discrete element technique. They stated that the
effect of moving load inertia would be more important on
dynamic response of beams for high moving mass velocities.

123

722

Rao [3] explored the behaviour of a simply supported beam

under the effect of moving load including its inertial effects
in the analysis, employing mode superposition and multiple
scale method. He showed that the effect of moving mass on
the transient response of the beam is notable. Nikkhoo et al.
[4] carried out an extensive parametric study on the role of
moving mass weight and velocity for a simply supported
beam via eigenfunction expansion method. They reported
that the effects of convective acceleration terms in moving
mass formulation could not be ignored for masses traveling with velocities greater than a specific one, named as the
critical velocity. Another comprehensive parametric study
on the evaluation of design parameters including maximum
deflection and bending moment of beams was scrutinized
by Kiani et al. [5] via reproducing kernel particle method.
They introduced some spectra in terms of beam slenderness
and boundary conditions as well as moving mass weight and
velocity. It was also mentioned that according to the slenderness and boundary conditions of the beam, the appropriate
beam theory (EulerBernoulli, Timoshenko, or higher order
beam theory) should be selected for proper assessment of the
dynamic behaviour of the beam structures.
Concerning the problem of multispan beams excited by
moving loads, Wu and Dai [6] studied the responses of a multispan nonuniform beam subjected to a series of loads moving with varying speed in identical and opposite directions
considering the effects of the moving load speed and acceleration. The transverse vibration of a beam with intermediate
point constraints subjected to a moving load was analyzed
by Lee [7] using the assumed mode method. In the proposed
model, the intermediate constraints were modeled by linear
springs with sufficient large stiffness. As a more applicable
analytical model, Chaterjee et al. [8] explored the dynamic
behaviour of a multispan continuous bridge under a moving
vehicular load by considering the interaction effect between
the vehicle and the bridge structure. For this purpose, they
modeled the vehicle as a single unsprung or sprung mass.
Furthermore, the torsion in the bridge caused by eccentrically applied load of vehicles and randomness of the surface
irregularity of the pavement were also included. Henchi et al.
[9] evaluated the dynamic response of multispan nonuniform
beam structures excited by multiple moving loads utilizing
spectral element method. Some similar studies in the analysis
as well as identification of the moving loads acted upon multispan continuous bridge structures could be found in Refs.
[1012].
In contrast to other investigations, Ichikawa et al. [13]
entered the inertia effects of a moving load in problem formulation of the continuous beams. The governing equations of
motion were solved employing the eigenfunction expansion
method, satisfying the continuity conditions at the intermediate supports. Their results were indicatory of inertia importance on assessing dynamic response of such beam structures,

123

K. Kiani et al.

especially for high moving mass weights and

velocities. Besides, the possibility of moving mass separation was detected by monitoring the sign of contact force
magnitude during the course of excitation. This phenomenon has also been cited by some other researchers [14,15].
In this article, some parametric studies on vibration of
multispan viscoelastic EulerBernoulli beams traversed by
a moving mass are carried out. In this regard, generalized
moving least square method (GMLSM) is utilized for spatial discretization of the unknown parameters of the problem. This numerical method was developed by Atluri et al.
[16] considering the first derivative of the unknown field as
an additional independent variable, through the modification
of local approximation in the moving least square method.
Static analysis of thin beams with different boundary conditions using GMLSM shows considerable results for both
deflection and bending moment fields of the beam [16]. In the
remainder, discrete equations of motion based on Lagranges
equation are derived. Variation of important design criteria
such as maximum deflection and bending moments of the
beams in terms of moving mass weight and velocity as well
as the span number and beams material relaxation rate are
explored. The results are indicatory of moving load inertia
importance as the span number increases. However, the role
of inertia effects for the higher beam span number comes
into play for higher levels of moving mass velocity. Besides,
the velocities associated with the maximum values of design
parameters increase as the span number increases. A linear
relation could be seen between the mass weight of the moving
load and the design parameters for a wide range of moving
mass velocities. In fact, this linear relation is violated for high
levels of moving mass velocities, specifically for the higher
span number. Equally important, is the possibility of moving
mass separation from the base beam that could be monitored by checking the sign of the contact force. The obtained
results show that this phenomenon would occur in lower levels of moving mass weight for the higher beam span number.
Furthermore, higher values of beams material relaxation rate
result in a decrease of moving load inertia effects as well as
the possibility of the mass separation from the base beam for
all ranges of the considered span number and moving mass
weight and velocity.

2 Problem assumptions
The under study system is a finite N S-span beam with length
L in which constrained at support locations by axial linear
springs with constant K z as illustrated in Fig. 1. The length
of the ith span of the beam is li , and for the case of equal
span lengths, li = L/N S; i = 1, 2, . . . , N S. In order to
prevent the axial movement of the beam, it is axially fixed in
one end. The components of elastic fields of the system are

Assessing dynamic response of multispan viscoelastic thin beams under a moving mass via generalized moving least square method

723

Fig. 1 Topology of a multispan

viscoelastic beam subjected to a
moving mass

outlined in the Cartesian coordinate system, with the x-axis

coincident with the neutral axis of the undeformed beam, and
the z-axis perpendicular to the beam neutral axis toward the
applied gravitational acceleration g. In the modeling of the
problem, the following assumptions are made: (1) The material of the beam is linear viscoelastic isotropic homogeneous
with elastic modulus of E b , and viscosity of x in the x direction. The material behavior of the beam obeys KelvinVoigt
model with the rate of relaxation of x = x /E b , in which
assumed to be age independent. (2) The cross section area of
the beam, Ab , and the beam density, b , are uniform along
the beam. (3) At time t = 0, the moving mass of mass M
enters the left hand end of the beam with constant velocity v,
and the only applied load is due to the normal contact force
of the moving mass on the beam. Furthermore, the moving
mass would be in contact with the beam at all times. (4) The
only component of acceleration for the mass over the beam is
u z M = (u z + 2v u z,x + v 2 u z,x x )x=x M , in which u z = u z (x, t)
denotes the transverse displacement component of the beam.

3 The numerical solution via GMLSM

In recent years, a huge effort has been devoted to meshless numerical methods, due to their powerful capabilities
in solving problems involving discontinuity, moving boundaries, large deformations or severe shooting. In contrast to the
finite element method (FEM), the meshless methods utilize
shape functions with higher continuity. Therefore, the reproduced values of the fields under study and their derivatives
are more accurate than those of FEM in most of the boundary value problems (BVPs). No dependence of these methods
on meshing the problem domain for construction the shape
functions is another major advantage, especially in solving
the problems that the domain geometry or their boundaries
are changing continuously. However, most of these methods utilize element approach for integrating the appropriate
weak form of governing partial differential equations of the

problem. In the present work, GMLSM is employed to solve

the governing IBVP of the multispan beam. A brief description of GMLSM is presented in Sect. 3.1 Afterwards, the
application of GMLSM to viscoelastic multispan Euler
Bernoulli beams subjected to a moving mass is provided in
the following section.
3.1 An introduction to GMLSM
This so-called meshless method was developed by Atluri
et al. [16] to incorporate the derivative data of the field variable into the interpolation scheme. To this end, the conventional moving least square interpolation scheme was
generalized by introducing the slope as an additional independent variable. Consider a continuous function u(x)
defined on a one-dimensional domain , where the nodal values and its derivative at the distinct points x I (1 I N P;
(0)
(1)
N P = number of particles) are given as u I and u I , respec(0)
(1)
tively; i.e., u I = u(x I ), u I = du(x I )/dx. For each point
x , one may consider a local approximation u x (x) in a
proper small neighborhood of x = x as
u(x) u x (x) = p T (x x )bb (x ),

(1)

in which
is a polynomial bases vector and b (x ) is the vector of unknown coefficients. For instance, the mth order polynomial in one dimension is expressed as p T (x) =
[1, x, x 2 , . . . , x m ]. The first derivative of the local approximation u x (x) is obtained as

d  T
du x (x)
u ,x (x)
=
p (x x )bb (x)
,
(2)
dx
dx
the coefficient vector b (x ) is determined to minimize the
following [16]

NP



(0) 2
Jx (bb ) =
w (0)
I ( x ) u x (x) u I
pT

I =1
(1)
+w I (x

du x (x)
(1)
u I
)
dx

2 
,

(3)

123

724

K. Kiani et al.

substituting Eqs. (1) and (2) into Eq. (3), leads to


NP



(0)
(0) 2
w I (x ) p T (x I x )bb u I
Jx (bb ) =
I =1
(1)

+w I (x )

d[pp T (x I x )bb ]
(1)
u I
dx

and
(n)

I,x x =

(n)

or in a matrix form
T



Jx (bb ) = Pb u (0) W (0) (x ) Pb u (0)

T


+ P ,x b u (1) W (1) (x ) P ,x b u (1) ,

1
(n)
+2 A 1
,x jk P ,x Lk W L I + 2 A ,x jk P Lk W ,x L I

(4)

(n)
1
(n)
A 1
+A
,x L I
jk P ,x x Lk W L I + 2 A jk P ,x Lk W

1
(n)
A jk P Lk W ,x x L I .
+A

(5)

P = [pp (x1 x ), p (x2 x ), . . . , p (x N Px )]T ,



p (x1 x ) p (x2 x )
p (x N P x ) T
P ,x =
,
,
,...,
x
x
x
(6)
(n)

T
(n)
u = u (x1 ), u (n) (x2 ), , u (n) (x N P ) ; n = 0, 1,

(n)
(n)
W (x ) I J = w I (x ) I J ; I, J = 1, 2, . . . , N P,

in which I J is the Kroneker delta, and there is no summation

on I . Requiring stationary condition of Jx (bb ) with respect to
the vector b yields
(7)

Assume u z = w(x, t) to be correspond to the transverse displacement of the beam from the equilibrium state. According
to the EulerBernoulli beam theory, the small rotation of the
transverse plane of the beam about the y-axis is equal to
w,x . Therefore, the longitudinal displacement is expressed
as u x = zw,x . In the case of small transverse displacement, according to the KelvinVoigt mechanical model, the
only nonzero components of strain and stress fields are provided as
x x = zw,x x and x x = z(E b w,x x + x w ,x x ),
correspondingly. Subsequently, the bending moment of the
beam is given by
Mb = E b Ib (w,x x + x w ,x x ).

L = T (U + V ),

A (x ) = P T W (0) P + P T,x W (1) P,x ,

C (n) = P T W (n) ; n = 0, 1.

(8)

Solving for b from Eq. (7), and substituting it into Eq. (1)
lead to
u(x )

1


(k) (x )uu (n) =



n=0

1 
NP


(n)

(n)

I (x )u I ,

(9)

n=0 I =1

1
T =
2

(10)

U =

1
2

or



2
b Ib w ,x
+ Ab w 2 dx,

L
2
E b Ib w,x
x dx +
0

(n)
A 1
p Tj (0)A
jk P Lk W L I ,

L
(11)

k=1 L=1 j=1

(n)

I,x =

m 
NP 
m



(n)
p Tj (0) A 1
,x jk P Lk W L I

k=1 L=1 j=1


(n)
1
(n)
A 1
+A
jk P ,x Lk W L I + A jk P Lk W ,x L I ,

123

V =

1
2


K z w 2 d,
b

M g w + 2v w ,x + v 2 w,x x

(16)

where I is the (n + 1)th kind of the GMLSM shape function associated with the I th particle. Furthermore, the first
and second derivatives of the GMLSM shape functions are
readily derived as follows
(n)

L
0

(n) (x ) = p T (0)A
A 1 P T W (n) ; n = 0, 1,


m 
NP 
m


(15)

in which T is the kinetic energy of the beam, U is the elastic

strain energy of the beam and springs, and V is the potential
energy of the beam under a moving mass loading. These
parameters are expressed as in the following

in which

(14)

The Lagrangian functional is defined as

where

(n)

(13)

3.2 Problem formulation by employing GMLSM

in which


(n)
p Tj (0) A 1
,x x jk P Lk W L I

k=1 L=1 j=1

2 

A (x )bb = C (0)u (0) + C (1)u (1) ,

m 
NP 
m


(12)

w(x x M ) H (L x M )dx,
in which b is the boundary of the beam domain, Ib is the
second moment inertia of the beam, is the Dirac delta function, and H is the Heaviside step function. The term M[g
(w + 2v w ,x + v 2 w,x x )] in Eq. (16) denotes the contact force
between the rigid moving mass and the elastic beam. Through
this definition, the third and fourth assumptions in Sect. 2 are
systematically satisfied. As it will be explained in Sect. 4.2,
the contact force could be considered as a key parameter

Assessing dynamic response of multispan viscoelastic thin beams under a moving mass via generalized moving least square method

725

Table 1 Comparison study of the first six dimensionless frequency parameters of an elastic continuous EulerBernoulli beam for different span
numbers
1

Ichikawa et al. [13]

3.141 592 6

3.926 602 3

6.283 185 3

7.068 582 7

9.424 778 0

10.210 176 1

Present method

3.139 839 9

3.926 452 9

6.282 978 2

7.070 911 9

9.439 178 8

10.208 452 3

Ichikawa et al. [13]

3.141 592 6

3.556 908 5

4.297 529 7

6.283 185 3

6.707 595 6

7.429 541 3

Present method

3.139 658 9

3.555 195 7

4.298 319 5

6.285 069 9

6.710 117 9

7.440 465 8

Ichikawa et al. [13]

3.141 592 6

3.393 231 3

3.926 602 3

4.463 324 4

6.283 185 3

6.545 413 8

Present method

3.139 559 2

3.391 651 0

3.925 947 8

4.464 886 6

6.286 111 5

6.548 347 5

Method

NS = 2

NS = 3

NS = 4

to investigate the possibility of the moving mass separation

from the base of multispan beams.
The dissipation function of the beam due to existing damping within the beam structure material is provided by
1
R=
2

2
x Ib w ,x
x dx.

(17)

 J (x);
w(x, t) =  TI (x) w I (t) = w TJ (t)
I, J = 1, 2, . . . , N P,

(18)



(1)
 TI (x) = (0)
I (x), I (x) ,


(0)
(1)
w TI (t) = w I (t), w I (t) .
Substituting Eq. (18) into Eqs. (16) and (17) leads to
L
b




Ib  TI,x w I  TJ,x w J




+Ab  TI w I  TJ w J dx,
1
2

+v 2 TJ,x x w J

L
0

1
+
2

1
R=
2


(x x M ) H (L x M )dx,




x Ib  TI,x x w I  TJ,x x w J dx,

(20)




E b Ib  TI,x x w I  TJ,x x w J dx




K z  TI w I  TJ w J d,

and then using the Lagranges equations,

R
(n)
w I

L
(n)
w I

d L
dt w (n)
I

= 0, I = 1, 2, . . . , N P; n = 0, 1, the discretized

equations of motion for a multispan viscoelastic thin beam

under a moving mass could be derived as
x + C x + K x = f ,
M

(21)

in which
M I J =



b Ab I  TJ + Ib I,x  TJ,x dx

 I (x M )
TJ (x M ) H (L x M ),
+M

in which

U =




TJ,x w J
M  TI w I g  TJ w J + 2v

Based on spatial discretization utilizing GMLSM, the only

unknown parameter of the problem, w(x, t), could be dis(0) (0)
(1) (1)
cretized as w = I w I + I w I ; I = 1, 2, . . . , N P,
(0)
where I is a free index, N P is the number of particles, I
(1)
and I are the first and second kind GMLSM shape func(0)
(1)
tions associated with the I th particle, and w I and w I are
the nodal parameter values associated with the I th particle.
For more convenient in calculations, w(x, t) could be written
in the matrix form as

1
2

V =

L
0

T =

(19)

C I J =

L
x Ib I,x x  TJ,x x dx
0

 I (x M )
TJ,x (x M ) H (L x M ),
+2Mv
K I J =

L
E b Ib I,x x  TJ,x x dx +
0

+Mv

(22)

K z  I  TJ d

b
T
 J,x x (x M )H (L
 I (x M )

x M ),

 I (x M )H (L x M ),
f I = Mg


(0)
(1)
(0)
(1)
(0)
(1) T
x = w 1 , w1 , w 2 , w2 , . . . , w N P , w N P .
It should be noted that because of the inertia effects of the
), the damping matrix (C ),
moving mass, the mass matrix (M

and the stiffness matrix (K ) are time dependent as long as the

123

726

K. Kiani et al.

Fig. 2 Comparison of the time

histories of normalized
deflection at the midpoint of the
first span via GMLSM and
eigenfunction expansion method
[13] for: a = 0.5, moving
load; b = 0.5, MN = 0.4;
c = 1.2, moving load;
d = 1.2, MN = 0.4

+
Table 2 The values of max{Wst }, max{Mb,st
}, and max{Mb,st
} for
elastic multispan beams under statically applied loads

NS

W st
M st

0.020 77

M st+

0.244 18

0.015 06

0.014 59

0.014 56

0.097 82

0.099 82

0.100 19

0.185 73

0.178 60

0.178 35

Note: W st = max{Wst }/(Mgl13 /E b Ib ), M st = max{Mb,st

}/(Mgl1 ),
and M st+ = max{M + }/(Mgl1 )
b,st

moving mass has not left the beam. Furthermore, the damping and stiffness matrices are not symmetric. Therefore, a
suitable approach should be adopted for solving the governing equations of motion in the time domain. The generalized
Newmark- method [5] is employed for required calculations in each time step. In all of analyses throughout this
paper, it is assumed that the presumed beam is originally at
rest; i.e., x (0) = 0 and x (0) = 0.
4 Numerical examples
4.1 Comparison of the GMLSM and other researchers
results
In order to check the capability of GMLSM in predicting the
natural frequencies of multispan beams, the obtained results
by this method are compared with those of eigenfunction
expansion method [13]. For this case study, the parameters
of the problem are as follows: l1 = 10 m, E b Ib = 3.367 9

123

108 Nm2 , b Ab = 450.333 kg/m. Besides, it is assumed that

K z = 106 E bl1 and n = (b Ab n2 /E b Ib )0.25l1 in which
n is defined as dimensionless frequency corresponding to
the nth natural frequency, n . For computing the natural frequencies of the structure, one may consider x (t) = x 0 eit .
Substituting the recent relation into Eq. (21) leads to the following eigenvalue equation in the case of free vibration with
no damping


+ K x 0 = 0.
(23)
2 M
By solving this set of eigenvalue equations via a suitable
method, one can obtain eigenvalues and corresponding eigenvectors. The eigenvalues provide the natural frequencies of
the structure with free vibration, and the eigen vectors give
the free vibration modes associated with the natural frequencies. For the numerical study in this part using GMLSM,
4 N S+1 uniformly distributed particles, five Gaussian points
in each computational cell, the third order base function,
and exponential weight function with influence domain of
3L/(N P 1) are considered. The obtained results of the
first six dimensionless frequencies for different span numbers
are presented in Table 1. Furthermore, the obtained results by
Ichikawa et al. [13] using eigenfunction expansion method
are given for more comparison. It is observed from Table 1
that there is a reasonable good agreement between the computed results of GMLSM and those of eigenfunction expansion method in all the cases.
In order to examine the correctness of the proposed numerical procedure, the time histories of midspan deflection in the
first span of the beam are compared with those of Ichikawa

Assessing dynamic response of multispan viscoelastic thin beams under a moving mass via generalized moving least square method

727

Fig. 3 Variation of normalized

design parameters due to
changes of moving mass weight
and velocity for a single span
beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001;
d x = 0.01

Fig. 4 Variation of normalized design parameters due to changes of moving mass weight and velocity for a two-span beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001; d x = 0.01

et al. [13]. To this end, consider a prismatic elastic multispan beam having four uniform spans, and it is assumed that
a moving mass starts to move at the left-hand end of the
first span. The multispan beam has the following properties:
E b Ib = 1.96 109 Nm2 , b Ab = 103 kg/m, l1 = 20 m.
For more convenience in verification of the results, one may
consider the normalized parameters = v/( l1 ), MN =
M/(b Abl1 ),break and w = E b Ib w/(Mgl13 ) in which =

E b Ib /b Ab /l12 . In the numerical analysis using GMLSM,

17 uniformly distributed particles, 5 Gaussian points in each
computational cell, third order base function, and the cubic
weight function with influence domain of 3L/(N P 1) have
been used. Moreover, the magnitude of the time steps is set
equal to L/(100v) during traveling of the moving mass on
the beam. The diagrams of midspan deflection in the first
span have been depicted in Fig. 2 for both GMLSM and

123

728

K. Kiani et al.

Fig. 5 Variation of normalized design parameters due to changes of moving mass weight and velocity for a three-span beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001; d x = 0.01

Fig. 6 Variation of normalized design parameters due to changes of moving mass weight and velocity for a four-span beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001; d x = 0.01

eigenfunction expansion method [13], including two different values of the parameters and M N . As the graphs show,
in general a good agreement is achieved between the results
of two methods in all the cases.

123

4.2 Parametric study of the problem

For elaborating parametric studies, some normalized parameters are defined. In this regard, the maximum dynamic

Assessing dynamic response of multispan viscoelastic thin beams under a moving mass via generalized moving least square method

729

Fig. 7 Normalized design

spectra in term of moving mass
(load) velocity for a single span
beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001;
d x = 0.01

Fig. 8 Normalized design

spectra in term of moving mass
(load) velocity for a two-span
beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001;
d x = 0.01

deflection, the maximum negative as well as the maximum

positive dynamic bending moment (design parameters) are
normalized with respect to the appropriate maximum
parameter values associated with the equivalent statically

=
applied loads, i.e., Wmax,N = max{w}/ max{Wst }, Mmax,N

+
+
max{Mb }/ max{Mb,st }, and Mmax,N = max{Mb }/ max
+
}. Table 2 demonstrates the calculated values of max
{Mb,st

+
{Wst }, max{Mb,st
} and max{Mb,st
} by GMLSM up to four
spans beams. Moreover, VN = / , according to Nikkhoo
et al. [4].

In Figs. 3, 4, 5 and 6, the variation of design parameters

as a function of moving mass weight and velocity is plotted
for different span numbers and material relaxation rates. As
Fig. 3 shows, the values of design parameters in a single span
beam are proportional to the mass weight of the moving load
for different velocities. This fact was stated by Nikkhoo et al.
[4] and Kiani et al. [5] for a simply supported elastic beam.
Moreover, as the magnitude of the material relaxation rate
increases, the same behavior could be observed in the elastic beam. In contrast to the single span beam, in multispan

123

730

K. Kiani et al.

Fig. 9 Normalized design

spectra in term of moving mass
(load) velocity for a three-span
beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001;
d x = 0.01

Fig. 10 Normalized design

spectra in term of moving mass
(load) velocity for a four-span
beam: a x = 0.000 01;
b x = 0.000 1; c x = 0.001;
d x = 0.01

ones, the mentioned linear relation would be violated for

high moving mass velocities, specially for even medium values of material relaxation rate (i.e., VN = 1.0 for two and
three spans beam, and VN 0.75 for four spans beam). The
main reason of this fact is the tendency of the moving mass
to separate from the base beam during the course of vibration. This phenomenon would be discussed in some detail
later. For high values of material relaxation rate (x = 0.01),
which implies a highly viscous beam, the same linear relation appears between the design parameter values and the
weight of moving mass. Equally important is the variation
of design parameters as a function of span number as well as
material relaxation rates. As shown in Figs. 3a, 4a, 5a and 6a,

123

for a certain value of material relaxation rate, increasing the

span number would result in higher values of design parameters. However, as the material relaxation rate increases, the
design parameters values would lessen for a specific beam
span number.
Alternatively, the design parameters spectra as a function of moving mass velocity and material relaxation rate
for a certain moving mass weight could be scrutinized for
both moving mass and moving load cases. Figures 7, 8, 9
and 10 deal with these spectra. According to these figures,
the moving load inertia effects are completely obvious, specifically for high moving mass velocities. As Fig. 7 shows,
for VN > 0.2 the effects of load inertia would be highlighted

Assessing dynamic response of multispan viscoelastic thin beams under a moving mass via generalized moving least square method

731

Fig. 11 Minimum and

maximum normalized values of
the contact force as a function of
moving mass weight and
velocity for a single span beam:
a x = 0.000 01; b
x = 0.000 1;
c x = 0.001; d x = 0.01

Fig. 12 Minimum and

maximum normalized values of
the contact force as a function of
moving mass weight and
velocity for a two-span beam:
a x = 0.000 01; b
x = 0.000 1;
c x = 0.001; d x = 0.01

regarding most of the considered values of material relaxation rate. Furthermore, for low values of x in a single span
beam (Fig. 7a, b), the maximum values of design parameters happen for VN 0.75 which is proposed here as the
critical velocity. This critical velocity as well as the value
of each design parameter decreases as the material relaxation rate increases. According to the Figs. 8, 9 and 10,
the threshold velocity in which the inertia effects would be
intensified, takes higher values as the span number of the
beam increases. For example, for two-, three-, and four-span

beams, this threshold velocity takes approximately the values

of 0.4V  , 0.5V  , and 0.6V  , respectively. Similarly, the critical velocity also increases as the span number of the beam
increases. As a general point of view, for different studied
beams, the fluctuations in the design spectra, especially for
low moving mass velocities, lessen as the magnitude of x
increases and it results in more smooth design spectra.
On the other hand, study of the possibility of mass separation phenomenon during the course of loading is of great
importance. The possibility of this fact could be checked by

123

732

K. Kiani et al.

Fig. 13 Minimum and

maximum normalized values of
the contact force as a function of
moving mass weight and
velocity for a three-span beam:
a x = 0.000 01;
b x = 0.000 1;
c x = 0.001; d x = 0.01

Fig. 14 Minimum and

maximum normalized values of
the contact force as a function of
moving mass weight and
velocity for a four-span beam:
a x = 0.000 01;
b x = 0.000 1;
c x = 0.001; d x = 0.01

the sign of the contact force between the moving mass and
the base beam, defined as follows

F(t) = Mg M w(x
M , t) + 2v w ,x (x M , t)

+ v 2 w,x x (x M , t) ,
(24)
the separation of the moving mass from the base beam would
be possible when the sign of F(t) changes from positive
to negative. In order to explore the possibility of the afore-

123

mentioned phenomenon, the variations of normalized minimum and maximum contact forces as a function of moving
mass weight and velocity for different span numbers and
x are depicted through the Figs. 11, 12, 13 and 14. The
associated contact force is normalized by Mg. In these figures, the changes of maximum and minimum contact forces
due to change of moving mass weight for certain velocities
of the moving load are depicted. Besides, the effects of the
span number and x are shown in these plots. It is worth

Assessing dynamic response of multispan viscoelastic thin beams under a moving mass via generalized moving least square method

mentioning that according to the third assumption in Sect. 2,

as long as the separation has not been occurred, the minimum value of the contact force would be zero, otherwise, it
would take negative values. It means that the lower the value
of the minimum contact force, the higher the possibility of
the moving mass separation from the base beam. Based on
Figs. 11a, 12a, 13a and 14a, the onset of mass separation
happens for lower moving mass weights as the span number
of the beam increases. For example, while the separation of
the moving mass in a single span beam occurs for MN > 0.2
in an assumed range of moving mass velocities, in the fourspan beam this phenomenon takes place for MN > 0.05. It
implies that the higher span number, the higher possibility
of mass separation. Expectedly, for viscous beams this phenomenon would be of less importance in which for highly
viscous beams, there would be no mass separation for any
values of moving mass weights and velocities (Figs. 11d,
12d, 13d and 14d). On the other hand, regardless of the possible negative values of the contact force, the linear relation
between the values of design parameters and the mass weight
of the moving load could be discovered through the same linear relation existing between the maximum observed contact
force and the moving mass weight. This fact could be seen
for all ranges of the considered span number and x .

5 Conclusions
In this work, design parameters of multispan viscoelastic
thin beams subjected to a moving mass were investigated
by a recently developed numerical scheme. To this end, the
unknown parameters of the problem were discretized in the
spatial domain using generalized moving least square method
and then, discrete equations of motion are gained based on
Lagranges equation. The design parameter spectra in terms
of mass weight and velocity of the moving mass were illustrated for multispan viscoelastic beams. The accuracy of the
proposed solution was confirmed by comparing the obtained
results with those of other researchers, and a reasonable good
agreement was achieved. The results showed that, although
the load inertia effects in the beams with a higher span number would be intensified for higher levels of moving mass
velocity, the maximum values of design parameters would
increase either. Moreover, the velocity associated with the
maximum values of design parameters increases as the span
number of the beam increases. A linear relation was ascertained between the mass weight of the moving load and the
design parameters for a wide range of moving mass velocities. Actually, this linear relation could be violated for high

733

levels of moving mass velocity, specifically for a beam with

a higher span number. On the other hand, the possibility of
moving mass separation from the base beam was examined
by checking the sign of the contact force. The obtained results
revealed that the tendency of moving mass separation from
the base beam would occur in lower levels of moving mass
weight for a higher beam span number. Furthermore, higher
values of beams material relaxation rate would result in a
more decrease of inertia effects of the moving load as well
as the possibility of the mass separation, irrespective of span
number, weight and velocity of the moving mass.

References
1. Frba, L.: Vibration of Solids and Structures under Moving
Loads. Thomas Telford, London (1999)
2. Akin, J.E., Mofid, M.: Numerical solution for response of beams
with moving mass. ASCE J. Struct. Eng 115(1), 120131 (1989)
3. Rao, G.V.: Linear dynamics of an elastic beam under moving
loads. ASME J. Vib. Acoust. 122, 281289 (2000)
4. Nikkhoo, A., Rofooei, F.R., Shadnam, M.R.: Dynamic behavior and modal control of beams under moving mass. J. Sound
Vib. 306(3-5), 712724 (2007)
5. Kiani, K., Nikkhoo, A., Mehri, B.: Prediction capabilities of classical and shear deformable beam models excited by a moving mass.
J. Sound Vib. 320(3), 632648 (2009)
6. Wu, J.S., Dai, C.W.: Dynamic responses of multispan nonuniform
beam due to moving loads. ASCE J. Struct. Eng. 113(3), 458
474 (1987)
7. Lee, H.P.: Dynamic response of a beam with intermediate point
constraints subjected to a moving load. J. Sound Vib. 171(3), 361
368 (1994)
8. Chatterjee, P.K., Datta, T.K., Surana, C.S.: Vibration of continuous bridges under moving vehicles. J. Sound Vib. 169(2), 619
632 (1994)
9. Henchi, K., Fafard, M., Dhatt, G., et al. : Dynamic behaviour of
multispan beams under moving loads. J. Sound Vib. 199(1), 3350
(1997)
10. Zhu, X.Q., Law, S.S.: Orthogonal function in moving loads identification on a multispan bridge. J. Sound Vib. 245(2), 329345 (2001)
11. Zhu, X.Q., Law, S.S.: Percise timestep integration for the dynamic
response of a continuous beam under moving loads. J. Sound
Vib. 240(5), 962970 (2001)
12. Dugush, Y.A., Eisenberger, M.: Vibrations of non-uniform continuous beams under moving loads. J. Sound Vib. 254, 911926 (2002)
13. Ichikawa, M., Miyakawa, Y., Matsuda, A.: Vibration analysis
of the continuous beam subjected to a moving mass. J. Sound
Vib. 230(3), 493506 (2000)
14. Lee, H.P.: The dynamic response of a Timoshenko beam subjected
to a moving mass. J. Sound Vib. 198(2), 249256 (1996)
15. Lee, U.: Revisiting the moving mass problem: onset of separation
between the mass and beam. ASME J. Vib. Acoust. 118(3), 516
521 (1996)
16. Atluri, S.N., Cho, J.Y., Kim, H.G.: Analysis of thin beams using the
meshless local Petrov-Galerkin method with generalized moving
least squares interpolations. Comput. Mech. 24, 334347 (1999)

123