Sei sulla pagina 1di 9

Linear Dynamics of an Elastic

Beam Under Moving Loads


G. Visweswara Rao
Sr. Technical Staff Member,
Engineering Mechanics Research Corp. Ltd.,
Bangalore 560 001, India

The dynamic response of an Euler-Bernoulli beam under moving loads is studied by mode
superposition. The inertial effects of the moving load are included in the analysis. The
time-dependent equations of motion in modal space are solved by the method of multiple
scales. Instability regions of parametric resonance are identified and the moving mass
effect is shown to significantly affect the transient response of the beam. Importance of
modal interaction arising out of the possible internal resonance is highlighted. While the
external resonance is due to the gravity effects of the moving load, the parametric and
internal resonance solely depends on the load mass parameterratio of the moving load
mass to the beam mass. Numerical results show the influence of the load inertia terms on
the beam response under either a single moving load or a series of moving loads.
S0739-37170001703-7

Introduction
Moving load and structure interaction phenomenon has been a
subject of wide interest and investigation 1,2. It is well recognized 3 that the transverse inertia effects of a moving load are
not negligible in the analysis of flexible structures subjected to
high-speed vehicles. Olsson has investigated the coupled bridgevehicle response analysis in truncated modal space using a direct
integration method. The transverse inertia effects have been a
topic of interest not only in bridge dynamics and in the design of
railway tracks, guide-way systems but also in other engineering
applications such as modern high-speed precision machinery processes 4. Katz et al. studied the aspect of parametric instability
that arises from a sequence of moving loads. Akin and Mofid 5
presented a numerical-analytical method to study the response of
a beam carrying a moving mass and having different boundary
conditions. The results highlighted the importance of moving
mass model compared to the moving force model particularly in
the cases of large moving mass and high velocity. In fact, with
inclusion of inertia effects the critical velocity tends to a lower
value than that in the straightforward case of concentrated force
model 6. In this context the aspect of dynamic response amplification assumes additional significance. Here the dynamic response amplification is to be regarded with respect to the maximum static response of the structure. The strong influence on the
dynamic amplification factor of the vehicle weight to that of the
structure and the vehicle speed is highlighted by Paultre et al. 7
in a series of experimental studies on highway bridges in service.
Gbadeyan and Oni 8 presented a general approach to study the
dynamic response of beams as well as plates under an arbitrary
number of concentrated moving masses. The results of Gbadeyan
and Oni and also of Lee 9 showed that the inertial effects of the
moving mass are not negligible even if the velocity of the moving
mass is small.
The moving mass model renders the structural system a parametrically excited one and hence causes instability 10,11. For
finite length beams a moving load persists for limited time duration and as such the loss of stability and consequent growth of
unbounded response with time may be insignificant. However the
response amplification is of vital interest particularly when a series of moving loads traverse the beam. In this paper, the moving
mass model is adopted and beam response is studied via modal
superposition utilizing the orthonormalized beam eigenfunctions.
Due to the presence of transverse inertia terms the equations in

modal space form a parametrically excited system of linear differential equations. These equations are studied by method of multiple scales 12,13. The method of multiple scales enables one to
identify possible resonances, both external and internal. The beam
response over a range of moving velocities and mass ratio parameter is examined in detail. The effect of series of moving loads
and the associated response amplification is also studied.

Beam Under Moving Load


The equation of motion governing the linear dynamics of a
uniform beam Fig. 1 of length L, Youngs modulus E, mass per
unit length m, and cross-sectional moment of inertia I, subjected
to a moving load of mass M is
EI 4 y/ x 4 m 2 y/ t 2 c y/ tM x v t 2 y/ t 2
2 v 2 y/ x t v 2 2 y/ x 2
F t x v t

In Eq. 1 y(x,t) is the vertical deflection of the beam in terms of


the spatial coordinate x and the time variable t. F(t) is the force
acting on the beam due to the moving load and is equal to Mg
where g is the acceleration due to gravity. v is the velocity of the
moving load to be defined in appropriate units. c is the damping
constant. stands for Diracs delta function.
It is assumed in writing Eq. 1 that the moving load has constant velocity and is in continuous contact with the beam. The
fourth term in the equation corresponds to the Coriolis acceleration associated with the moving load as it traverses along the
vibrating beam. In the absence of this term Eq. 1 reduces to the
familiar case of a beam under a moving concentrated force.
It is convenient to introduce the following nondimensional
quantities:

Contributed by the Technical Committee on Vibration and Sound for publication


in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received Nov. 1998;
revised March 2000. Associate Technical Editor: B. Yang.

Journal of Vibration and Acoustics

(1)

Copyright 2000 by ASME

Xx/L,

v t/L,

Y y/L

Fig. 1 Beam with moving load

JULY 2000, Vol. 122 281

Table 1 Modal interactionpossible cases of internal resonance

Furthermore, assuming simply supported boundary conditions at


the beam ends, an approximate solution is selected in terms of a
finite number of mode shape functions, n (X) as
Y X,

Y X
n

(2)

Here n (X)Sin n X. Y n (t) are coordinates in modal space.


Now, the equations of motion of the modal coordinates Y j (t) in
terms of the non-dimensional quantities are described by
Y j 2 j Y j 2j Y j 2

Y n Sin n Sin j

2n Y n Cos n Sin j n 2 2 Y n Sin n Sin j


2 P Sin j .

j
j / v /L with
j EI/m j /L 2
(4)

The coefficient in the damping term of Eq. 3 follows from the


familiar transformation:

The parameter mainly distinguishes the moving force model


from a moving mass one. As can be seen from Eq. 3, if is
small enough to be ignored representing the case of the moving
force it is only the external resonance with j j , that may be
of interest. In case one wishes to study the possible interaction
between the beam and the moving mass, it is obvious that a general solution technique is necessary to be developed. To this end,
a perturbation approach based on the method of multiple scales
MMS 12 is adopted in the present study.
The solution to the modal coordinates in Eq. 3 is assumed in
the form
Y j , Y j0 0 , 1 , . . . Y j1 0 , 1 , . . .
2 Y j2 0 , 1 , . . . . . . . . .

(6)

where m represents different independent time scales given by

m m ,

m0,1, . . .

(7)

Substituting Eq. 6 in Eq. 3 and noting that


d/d D 0 D 1 2 D 2 . . . ,
d 2 /d 2 D 20 2D 0 D 1 2 2D 0 D 2 D 21 . . .
where D 0 d/d 0 ,

(8)

D 1 d/d 1 , . . .

the following set of linear ordinary differential equations result:

0 :D 20 Y j0 2j Y j0 0

(9)

1 :D 20 Y j1 2j Y j1 2 j D 0 Y j0 c.c. 2 D 0 D 1 Y j0 c.c.

1/2

(5)

Perturbation on Modal Coordinates

(3)

In Eq. 3 dots stand for derivatives with respect to the nondimensional time. Also PgL/ v 2 .
is the nondimensional parameter describing the ratio of the
moving load mass M to the mass of the beam given by mL. j is
the jth beam natural frequency nondimensionalized with respect to
frequency of moving load. Thus
M /mL,

2 j c/m / v /L

DY
n

2
0

n0 exp i n j 0 exp i n j 0 c.c.

2n D 0 Y n0 i exp i n j 0 iexp i n j 0 c.c.


n 2 2 Y n0 exp i n j 0 exp i n j 0 c.c.

i P exp i j 0 c.c. ]

(10)

While deriving the above Eqs. 910, the damping is taken as .


Further, in these equations c.c. stands for complex conjugate and i(1) 1/2. D j d/d j and D 2j d 2 /d 2j ( j0,1,2, . . . ). Solution
of Eq. 9 is now taken as
Y j0 A j exp i 0 c.c.

(11)

where A j is a complex amplitude yet to be determined and which is a function of time scales m , m1,2 . . . .
282 Vol. 122, JULY 2000

Transactions of the ASME

Substitution of Y j0 in Eq. 10 results in


D 20 Y j1 2j Y j1 2i 2j A j exp i j 0 c.c. 2i j D 1 A j exp i j 0 c.c.
1/2

n A
n

2
n

exp i n 0 c.c. exp i n j 0 exp i n j 0 c.c.

2n i n A n exp i n 0 c.c. i exp i n j 0 i exp i n j 0 c.c.

i P exp i j 0 c.c.

It is obvious from the terms on the right-hand side of Eq. 12 that


the moving load inertial effect is manifested in the form of parametric resonance when j approaches j . This is in fact the same
external resonance that can occur under a moving concentrated
force P alone. Distinction is however to be made between the two
resonances in that the former is a case of dynamic instability
leading to unbounded response growth with time while the external resonance causes only a response magnification. It can also be
noticed from Eq. 12 that internal resonance also may occur and
that in turn causes modal interaction among different modes (n
j). The cases of possible internal resonance are detailed in
Table 1.

(12)

Parametric Resonance with j j and with no Internal Resonance


It is worthwhile to examine first the case where the modal interaction is absent and each mode attains parametric resonance
condition independent of the other. From Eq. 12, it can be seen
that as j approaches j , secular and small-divisor terms develop. The complex amplitude A j is now obtained from the condition that these terms are to be eliminated from Eq. 12.
Introducing at this stage a detuning parameter j to quantify the
deviation of each j from j in the form

Fig. 2 Maximum beam transient response plot for time 1. 0.3. No internal resonance.

Journal of Vibration and Acoustics

JULY 2000, Vol. 122 283

Fig. 3 Maximum steady state beam response for v in the range of 10100
msec. 0.3.

j j j

(13)

and equating the secular and small-divisor terms to zero leads to


the condition on A j :
2i j D 1 A j 2i 2j A j 1/2

j j 2 A j exp 2i j 0

2 2j j 2 2 A j

i Pexp i j 0 c.c.

0.5 2j j 2 2 a j
P Sin j

(19)

(14)

Equations 18 and 19 are autonomous differential equations in


a j and transformed phase angle j . This is made possible by
redefining the phase angle j in terms of a new parameter j in
the form

(15)

j j j

Noting that A j is independent of 0 it follows that


dA j /d dA j /d 1 D 1 A j

j a j j j a j j 0.25 j j 2 a j Cos 2 j

(20)

If now A j is assumed to be of the form


A j 1/2 a j exp i j

(16)

with amplitude a j and phase j , it follows that


dA j /d 1/2 i j a j j a j j exp i j

(17)

and Eq. 14 results in following sets of two first order differential


equations for each amplitude and phase with j1,2, . . . n.

j a j 2j a j 0.25 j j 2 a j Sin 2 j P Cos j


(18)
284 Vol. 122, JULY 2000

Equations 18 and 19 can now be directly integrated to obtain


a j and j for each modal response in the transient period. The
transient solution of each mode valid up to the order of is now
obtained by combining Eqs. 6, 11 and the solution Y j1 from
Eq. 12. It is given by
Y j , a j Cos j j
b j j a j Cos 3 j j /8j c

(21)

Transactions of the ASME

Fig. 4 Stable and unstable beam responses in 20 cycles of integration time.


0.3.

By substituting Eq. 21 in Eq. 2 the beam response can be


obtained.

Parametric Resonance with j j and with Internal


Resonance
Noting that n n and examining Eq. 12 for possibility of
(n j) n j and n (n j) j for n j, one can obtain corrections to the first order Eqs. 18 and 19 in the following form:

j a j 2j a j 0.25 j j 2 a j Sin 2 j P Cos j


0.25

a n Sin Sin

n j

Solution to Eqs. 22 and 23 are utilized to finally obtain the


transient response of the beam.

Series of Moving Loads with UniformNonuniform


Time Lag
In the analysis presented so far, the case of a single load traversing the beam is considered. However, it may be more critical
for the supporting beam when it is traversed by a series of moving
loads P k , each associated with a uniform/nonuniform time lag
k . In this case Eq. 3 takes the form
Y j 2 j Y j 2j Y j 2

U k

Y n Sin n

k Sin j k 2n Y n Cos n k Sin j k

(22)
n 2 2 Y n Sin n k Sin j k

j a j j j a j j 0.25 j j 2 a j Cos 2 j
0.5 2j j 2 2 a j P Sin j
0.25

a n Cos

n j

Cos j n
Journal of Vibration and Acoustics

U P
k

Sin j k

(24)

(23)

and the first order differential equations for the amplitude a j and
j of each modal response Y j can be obtained as
JULY 2000, Vol. 122 285

Fig. 5 Maximum beam response in time 1 with Internal Resonance.


0.3.

j a j 2j a j 0.25

U
k

j j a j Sin 2 j jk
2

U k P
k

j a j j j a j j 0.25

Cos j jk

(25)

U j a
k

Cos 2 j jk 0.5 2j j 2 2 a j

U P
k

Sin j jk

(26)

Here jk j k .
The beam response can now be obtained following the same
steps as in the case of a single moving load.

Results and Conclusions


The main interest being the transient response of the beam over
the travel time of the moving load, numerical results are obtained
for tL/ v i.e. 1 unless stated specifically otherwise. As such,
in the case of a single moving load, Eqs. 18 and 19 are integrated for 1. For more than one load the integration time is
increased by the respective time lag associated with each load.
The data used in the numerical simulation are as follows: L
10 m, E2.01011 N/m2 , v 1 100 m/sec., 0.10.5,
0.01. Further, 7850 kg/m3 and Abeam cross-sectional area
0.01 m2 so that mass per unit length, m A. The number of
modes considered in the analysis that is found to be adequate in
modal summation is three. Figure 2 shows the locus of the beam
maximum response for each moving load velocity in the range
1100 m/sec. The ordinates in this figure are normalized with
respect to static midspan deflection of the beam under a concen286 Vol. 122, JULY 2000

trated central load. Modal interaction due to internal resonance is


ignored in obtaining this result. The influence of the load inertia
term is apparent from this figure wherein the result for the concentrated force model also is shown. It can be observed that the
inertia effect is predominant at lower moving load velocities. Thus
for the chosen value of the mass parameter 0.3, the maximum
response plot corresponding to the moving mass model lies above
the corresponding plot of the straightforward concentrated force
model for velocities less than 45 m/sec. Above this velocity the
concentrated force model yields higher response. In other words
for the system under consideration the moving load inertia effects
are negligible beyond this velocity. The qualitative nature of the
response is the same at both xL/2 i.e. X0.5 and xL/4 i.e.
X0.25. Lowering of the critical velocity by the moving load
inertia terms is best-illustrated in Fig. 3. The results in this figure
are obtained for a longer period of time integration with 20.
The result may not be relevant in the case of structures with the
moving load present only for 1 (tL/ v ). The result in Fig. 3
is to be interpreted with the assumption that the load periodically
appears again on the beam at the end of each travel time equal to
L/ v as is the case familiar with machine tools. It is pertinent here
to note that the moving load inertia terms render each modal Eq.
12 parametrically excited. This parametric resonance is a case of
dynamic instability and hence leads to unbounded response
growth with time Fig. 4. On the other hand the external resonance associated with the moving force model is a simple case of
response magnification.
In addition to parametric resonance, the moving load inertia
causes internal resonance as indicated in Table 1. The resulting
modal interaction also contributes to higher deflection as shown in
Fig. 5. Comparing the result with that of Fig. 2 moving mass
model, the increase in beam response is about 25 percent implying that the modal interaction and hence the internal resonance
phenomenon in moving load problems need special attention.
The strong influence of the mass parameter is noticeable in
Fig. 6. The values of the parameter are marked on the correTransactions of the ASME

Fig. 6 Effect of mass parameter on maximum beam response in time 1 at a


X 0.5 and b X 0.25

sponding maximum response curves in the figure. The effect of


is significant on two counts: 1 there is considerable increase in
the beam deflection as the parameter value increases and 2 the
response peaks occur at lower moving load velocities as the parameter, increases. In any case the influence of the moving load
inertia ceases to exist at higher velocities as per the linear perturbation analysis presented in this paper.
From practical considerations it is required to analyze the beam
response under the action of arbitrary number of moving loads
with random magnitudes and velocities. While this is out of the
scope of this paper, it is the case of two moving loads that is
considered here and the results are presented in Fig. 7. The figure
shows the maximum response of the beam subjected to two moving loads with same velocity and same mass parameter, 0.3,
but having a time lag of L 2 1 . The time lag parameter L
between the two moving loads is marked on each response plot in
the figure. In obtaining the results in the figure the total integration
interval takes care of the fact that the leading load leaves the beam
for some part of it. The results in this figure amply illustrate the
fact that transient response of the beam gets magnified with each
moving load, particularly when it follows the preceding load
within one time interval equal 1. The magnification reaches
Journal of Vibration and Acoustics

maximum as the time lag approaches zero. Further the maximum


response acquires two peaks if the second load considerably lags
behind the front one. For example, in Fig. 7 the beam maximum
response shows two predominant peaks for L 0.5 which of
course indicate independent action of the moving loads.
Hence, from the analysis and results presented in the paper, the
following conclusions are drawn:
The method of multiple scales analysis presented in this paper is able to capture the parametric and internal as well as the
external resonance phenomena that are possible in moving load
problems.
The influence of the moving load inertia is manifested in the
modal equations in the form of the mass ratio parameter . This
parameter is used in the perturbation analysis to bring out the
essential characteristics of the beam response.
Both the parametric and internal resonances cause significant
response magnification for either a single load or a series of moving loads during their travel time.
For more than one single load the time lag associated with
each moving load considerably influences the beam maximum
response.
JULY 2000, Vol. 122 287

Fig. 7 Two moving loads. Effect of time lag 2 1 on maximum response


at a X 0.5 and b X 0.25. 0.3.

Nomenclature
aj
c
j
m
n
t
v
y(x,t)
A
Aj
EI
F(t)
L
M
P
X
Y j (t)
,

jth modal amplitude parameter


beam damping constant
modal number
beam mass per unit length
modal number
time
constant moving load velocity
beam deflection at coordinate x and time t
beam cross-sectional area
complex modal amplitude
beam bending stiffness
moving force Mg
beam length
moving load mass
moving force parameter, gL/ v 2
nondimensionalized coordinate, x/L
modal coordinate
Dirac delta function
moving mass parameter, M /mL
jth beam natural frequency, (EI/m)( j /L) 2
nondimensionalized jth beam natural frequency with
respect to v /L

288 Vol. 122, JULY 2000

m
j
j , j
k
jk

nondimensionalized time, v t/L


jth mode shape function
damping parameter(c/m)/2 j v /L)
damping parameter /
independent time scales, m
detuning parameter of the jth mode
phase parameters related by j j j
time lag of kth load
phase lag parameter of kth moving load in jth,
modej k

References
1 Fryba, L., 1972, Vibration of Solids and Structures Under Moving Loads,
Noordhoff, The Netherlands.
2 Ting, E. C., and Yener, M., 1983, Vehicle-Structure Interactions in Bridge
Dynamics, Shock Vibr. Dig., 15, No. 12, pp. 39.
3 Olsson, M., 1985, Finite Element, Modal Co-ordinate Analysis of Structures
subjected to Moving Loads, J. Sound Vib., 99, No. 1, pp. 112.
4 Katz, R., Lee, C. W., Ulsoy, A. G., and Scott, R. A., 1987, Dynamic Stability
and Response of a Beam Subject to a Deflection Dependent Moving Load,
Trans. ASME, J. Vib., Acoust., Stress, Reliab. Des., 109, pp. 361365.
5 Akin, J. E., and Mofid, M., 1989, Numerical Solution for Response of Beams
with Moving Mass, J. Struct. Eng., 115, No. 1, pp. 120131.
6 Duffy, D. G., 1990, The Response of an Infinite Railroad Track to a Moving,
Vibrating Mass, ASME J. Appl. Mech., 57, pp. 6673.

Transactions of the ASME

7 Paultre, P., Proulx, J., and Talbot, M., 1995, Dynamic Testing Procedures for
Highway Bridges Using Traffic Loads, J. Struct. Eng., 121, No. 2, pp. 362
375.
8 Gbadeyan, J. A., and Oni, S. T., 1995, Dynamic Behavior of Beams and
Rectangular Plates under Moving Loads, J. Sound Vib., 182, No. 5, pp.
677695.
9 Lee, H. P., 1996, Dynamic Response of a Beam with a Moving Mass, J.
Sound Vib., 191, No. 2, pp. 289294.

Journal of Vibration and Acoustics

10 Bolotin, V. V., 1964, The Dynamic Stability of Elastic Systems, Holden Day,
San Francisco.
11 Nelson, H. D., and Conover, R. A., 1971, Dynamic Stability of a Beam
Carrying Moving Masses, ASME J. Appl. Mech., 38, pp. 10031006.
12 Nayfeh, A. H., and Mook, D. T., 1979, Nonlinear Oscillations, Wiley, New
York.
13 Rao, G. V., and Iyengar, R. N., 1991, Internal Resonance and Non-Linear
Response of a Cable under Periodic Excitation, J. Sound Vib., 149, No. 1,
pp. 2541.

JULY 2000, Vol. 122 289

Potrebbero piacerti anche