On the Response of Beams to an ArbitrarJy Number

of Concentrated Moving Masses

by MILOMIR M. STANISIC: and JAY c. HARDIN

School of Aeronautics, Astronautics and Engineering Sciences

Purdue University, Lafayette,
. _ Indiana

ABSTRACT: A theory describing the response of a beam under an arbitrary number of

moving mus8e.s is developed. The theory is baaed on the Fourier technique and shows that,
for a simply supported beam, the resonance frequency is lower with no corresponding
decrease in maximum amplitude when the inertia is considered.

I. Introduction

The problem of a beam carrying moving masses has been of interest for
several years. Particularly in the construction of railway bridges, it is of
fundamental importance. Recently, the advent of long highway bridges,
together with the increased velocity and mass of automobiles, has forced a
renewal of consideration of this problem.
The problem was first attacked by Jeffcott (1) in 1929. He was closely
followed by Steuding (2) and Odman (3). In these papers, the solution was
presented in approximate form involving rather laborious permutation
techniques. In 1951 Pestel (a), by means of Rayleigh-Ritz technique,
established a system of coupled differential equations governing the equili-
brium of a curved bar subjected to a single moving mass, in addition to
external loads. The usual iteration technique for solution of these coupled
differential equations was used but no numerical results were presented.
The object of this paper is (a) to present a simple technique concerning the
response of a beam to an arbitrary number of moving masses with various
velocities in the presence of the external load, and (b) to develop a solution
in a series form, but “closed” in the engineering sense, so that a practising
engineer can evaluate the response of the beam to the moving masses in a
reasonable period of time. The external load is considered to be a result of the
weight of the moving masses only, but no difficulty arises to include other
deterministic loads, such as impulses, etc.
The method developed in this paper is based on Fourier analysis and leads
to an approximate rapidly converging solution readily amenable to design
analysis and calculation. This method is not connected with any previously
developed techniques in this subject. The usual assumptions in the Navier
beam theory of constant sections are adopted. Straight beams of slender
proportions are considered, i.e. the shearing and rotating inertia effects are
neglected. A numerical technique for solution, involving higher order approxi-
mations of the entire problem, is also discussed. The same method was

115
Milomir M. Stanis’ and Jay C. Hardin

applied to the problem of response of a plate to moving masses with arbitrary

velocities, without any difficulties of a pragmatical nature (5).

II. Formulation of the Problem

Consider the vibration of a uniform simply supported beam carrying an

arbitrary number of discrete masses M,, M,, . . . , MN. The mass Md is assumed
to strike the beam at t = 0 and travel across it with velocity vi. The equation
of motion, with damping neglected, may be written

where
EI = flexural rigidity of the beam: lbin2;
Y(x, t) = transverse deflection of the beam : in. ;
p = mass density of the beam material : lb in-4 sec2;
A = cross-sectional area of the beam: in2;
g = acceleration of gravity : in set-2

and 6(x-vi t) is the Dirac delta function defined to be zero everywhere

except at x = vi t, i.e.
s(z-vit) = 0, x#qt,
and, in addition,
L

0 I
s(2-vit)dx = 1, (2)

where L is the length of the beam, see Fig. 1.

The boundary conditions are

Y(0, t) = Y(L, t) = 0,
Y,,(O, t) = Y,,(L, t) = 0. (3)

Applying the Fourier finite sine transform, i.e.

to Eq. 1, we have
Z(m, t) =
s0
L
Y(x, t) sin y&z (4)

.%(m, t) + 43(m, t)

(5)

where

(6)

is the eigenvalue of the problem and T[ ] stands for the transform of the
quantity in the brackets. Now, note that Z(lc, t) is just the coefficient of

116 Journal
of The Franklin Institute
 Response of Beams to an Arbitrary Number of Moving Masses

Y(x, t) expanded in a Fourier sine series. Thus

2 * h-X
Y(x, t) = z kzlZ(k, t) sin z (7)
and

(8)
Also, choosing to express 6(x - vuit) as an even function, we may expand it
in a Fourier cosine series, namely,
12” n7Tvit
6(x-vit) = L+L~~lcos-l--cos~. (9)

Thus, using the relation cos 01sin/3 = *[sin (CX

+/3) -sin (CX
- p)] with Eqs. 8
and 9, we obtain the expression
2 * krx
S(x - vi t) 6(x, t) = ~2 EIG(k, t) sin 7

1
2 a, 00
nrrvi t sin (n+ k)nx_ sin (n
(10)
x1 kyll(k t) cm37 TX .
+L’i L L

Finally, multiplying this expression by sinmrx/L and integrating over the

length of the beam, we find the desired transform

T 5 MiS(x-vit)I&,(x,t)
[ i=l I

= k g Mi kt,(rn, t) + 2kglZll(k, t) sin F sin F] . (11)

il

Equations 5 and 11 lead to

Z&h t) + w&Z@, t) + 1 z Mi Z,,(m, t) + 2 E Z,(k, t) sin F sin y]

PAL i-1 k=l

= s glMisinT, (12)
i
which is the transformed equation for our problem.

III. Solutions of the Transformed Equation

For the purpose of solution we consider only one mass M traveling with
velocity v, Fig. 1. Solutions for greater numbers of masses may be obtained
in the same manner. Evidently, the following special cases from Eq. 12
follow :
(a) Moving Force. If we neglect the inertia term, we have the classical case
of a moving force. Under this assumption, Eq. 12 becomes
mnvt
Z,,(m, t) + wtZ(m, t) = P sin ~ (13)
L ’

Vol.287,No.2,February196Q 117
Milomir M. Xtani&ic’ and Jay C. Hardin

where
p_!Q (14)
PA .
The general solution of Eq. 13 is
. mrrvt
Z(m,t) = A,cosw,t+B,sinw,t+ sin---. (15)
rnrrv 2 L
~%[I-;i j
L%
Now, if we assume that the beam is initially at rest, we have the initial
conditions :
Z(m, 0) = Z,(m, 0) = 0. (16)

I-IV
M-

yLd
FIG. 1. Geometry of the moving mass.

Applying Eq. 16 to Eq. 15, we find

& j[
_..z mrrvt
sin--- mi7v
.Z(m, t) = sin w, t (17)
m77v L -hn I
-hl

and the solution Y(x, t) is given by Eq. 7, i.e.

. m37vt m3i-v .

Y(x, t) = y $ [ sln~e~~~~t
m_-l I sin??. (18)

We note that this agrees with the classical solution given by Timoshenko (6).
(b) Moving Mass-First Approximation. If we consider only the linear
inertia term, Eq. 12 becomes
2
P . mwut
-%(m, t) + (lW;a) z(m, t) = (19)
(l+R)SmL’
where
R= M
pAL’ (20)
The general solution is directly analogous to Eq. 15. Applying the conditions
of Eq. 16, we find
P . mnvt rnrvJ(1 t-R) . w,t
Z(m,t) = sin- . (21)
sm-- L L% $O+R) I

118 Journal of The Franklin Institute

 Response of Beams to an Arbitrary Number of Moving Masses
Therefore, by Eq. 7, we have

. mrut mrvJ( 1 + R) . CO,,

sin----
L L% slnJ(l+R)t . m5rx
I sin-. (22)
m~vJ(l+R)
___~ 2 L
L% I)

Z(m,t)

14oc
FIG. 2. Convergence of coefficients for moving mass solution.

Evidently, the series given by Eq. 22 converges rapidly. To show this, we

must demonstrate that the coefficients, Z(m, t), of the spatial sine term in
Eq. 22 are convergent. Note that we may write

P 1 +mnvJ(l +R)
z(m, t) < (23)
mnvJ(l+R) 2 LWnz
kn 11
Now, using Eq. 6, we have

P(l+$f)A;;R)]*) 1
.Wbt)< 7T 4 EI
Lv 2 m4 (24)
z pA-(l+R) n
O[ ( )I’ -

Therefore, the coefficients Z(m, t) and the solution Y(x, t) converge as m--4.
The first three coefficients are graphed in Fig. 2, from which the convergence
can be seen. Note also that when R = 0, Eq. 22 reduces to the moving force
solution. Since the equation is linear, we might add that the solution for an
arbitrary number of masses may be obtained by superposition of the
individual solutions.

Vol. 287, No. 2. February 1969 119

_Milomir M. Stank% and Juy C. Hardin

(c) Moving Mass-Second Approximation. If we consider the first term of

the series of nonlinear terms, Eq. 12 becomes
rrvt mnvt mxvt
(I+ R) .&(m, t) + wkZ(m, t) + 2RZ,,( 1, t) sin - sin __ = PsinT. (251
L L
This equation holds for all m. Set m = 1. Thus

4
ml> t) + 1 +R+211sin2d] zP,t)= [1 +R+2~siqsin~-
(26)
[
While this equation is resistant to analytic techniques, it yields readily to
numerical procedures. Once Z( 1, t) is obtained, it may be substituted into
Eq. 25 for calculation of the other Z(m, t). The solutions for m = 1,2,3 are
tabulated in Table I and plotted in Fig. 3.

TABLE I
Moving Mass-Second Approximation,
for R = 0.25,L/v = 1*50sec, w1 = 2.44 rad see-j

t al, t) w, t) w3, t)

0.0 0.000 o-000 o*ooo

0.1 0.139 0.306 0.382
0.2 1.078 1.726 1.323
0.3 3.479 4.581 1.242
0.4 7.785 8.025 0.273
0.5 14.209 9.009 0.129
0.6 22.766 6.847 - 1.422
0.7 33.311 1.686 - 1.467
0.8 45.559 - 3.321 - 1.150
0.9 59.089 - 7.369 - 0.318
1.0 73.327 - 8.066 - 0,224
1.1 88.897 - 6.0193 0.669
1.2 100.693 - 3.644 1.477
1.3 111.685 - 1.850 0.915
1.4 119.183 - 1.500 0.390
1.5 121.905 - 1,479 0.307

Obviously higher approximations are possible by considering more terms

of the series and following the same procedure. However, considering the
rate of convergence of the lower-order solutions, it should not be necessary to
continue this process.

IV. Remarks on the Solution

In a problem such as this, one is interested in the maximum amplitude of

vibration and the conditions under which it can occur. For the classical
solution, Eq. 17, it can be shown that under certain values of the velocity the

120 Journal of The Franklin Institute

 Response of Beams to an Arbitrary Number of &Coving &lasses

condition of resonance occurs. In this case, the amplitude of vibration

becomes a linear function of time, Fig. 4. The maximum value of the ampli-
tude, which occurs when the mass is at the end of the beam, e.g.

(27)

40- R= !Lq

60-
w, = 2.44 rad/sec.

80-

100 -
Moving Moss 2nd Approx.

120-
Moving Moss 1st Approx.

140-

t
+Y(bqt)

FIG. 3. Comparison of approximate solutions.

FIG. 4. Amplitude growth under resonant conditions.

Vol. 257, X0. 2, February 1969 121

Milomir M. Stan&% and Jay C. Hardin

is, for the first natural frequency, about 50 per cent greater than the maxi-
mum static deflection
6 = MgL3 (28)
Tim’
However, in general, the velocities are small enough and the beam natural
frequencies large enough that the condition does not occur.
Now, consider the solution given by Eq. 22. The condition of resonance
requires that

where CO, is any one of the natural frequencies of the beam. However, the
frequency at which resonance occurs is lowered by the factor (1 + R)-1.
Moreover, since
lim -Wn, t)
=+J,
O-+mnzl(l+R)

is in determinant we may apply 1’Hospital’s rule. Therefore

*+msi(lfR)
lim
=&Ii
.Z(m,t)=g m dCl:R)cos~-$sin~
[
~
1 . (30)

Thus, t,he maximum amplitude, which occurs at t = L/v, is approximately

(31)

Therefore, we conclude that the maximum amplitude caused by the moving

mass is equal to that given by the moving force solution.
It should be noted that for higher approximations, the solution resulting
from Eq. 12 cannot be obtained in closed form, since for any higher approxi-
mation Eq. 12 represents a set of coupled equations of motion. In this case,
the resonance condition can be examined only by numerical procedures.

V. Conclusion

A theory is presented for the response of a beam under an arbitrary

number of moving masses. The theory is simple enough to be used in com-
putation for design considerations. The equation of motion is given in terms
of S-Dirac functions and is solved through the use of Fourier finite sine
transforms. An analytic first approximation is obtained and compared with
the solution for a moving force. It is found that, for a simply supported beam,
the resonant frequency is lower with no corresponding decrease in maximum
amplitude when the inertia is considered. For any higher approximation, the
solution can be obtained by means of numerical techniques. The first of these
is presented. The convergence of the solution has been established.

122 Journal of The Franklin Institute

 Response of Beams to an Arbitrary Number of Moving Masses

References

(1) H. H. Jeffcott, “On the Vibration of Beams under the Action of Moving Loads”,
Phil. Mug., Ser. 7, Vol. 8, p. 66, 1929.
(2) H. Steuding, “Die Schwingung von Triigern bei bewegten Lasten”, Ingen.-Arch.,
Vol. 5, p. 275, 1934.
(3) S. T. A. Odman, “Differential Equation for Calculation of Vibrations Produced in
Load-bearing Structures by Moving Loads”, Preliminary Publications. Inter-
national Assn. for Bridge and Structural Engineering, 3rd Congress, LiBge, 1948.
(4) E. Pestel, “Tragwerksauslenkung unter bewegter Last”, Ingem-Arch., Vol. 19,
p. 378, 1951.
(5) M. M. Stan%%, J. C. Hardin and Y. C. Lou, “On the Response of the Plate to a
Multi-masses Moving System”, Acta Mech., Vol. 5, p. 37, 1968.
(6) S. Timoshenko, “Vibration Problems in Engineering”, 3rd ed., Princeton, N.J.,
D. Van Nostrand Company, p. 352, 1955.

Vol.057,X0. 2,February 1969 123