You are on page 1of 21

Journal of Sound and Vibration (1996) 198(2), 149169

DYNAMIC ANALYSIS OF BEAMS ON AN ELASTIC


FOUNDATION SUBJECTED TO MOVING LOADS
D. Tn:xnir:1N:x :Nb Y. Znicr
School of Civil Engineering, Queensland University of Technology, G.P.O. Box 2434,
Brisbane, Queensland 4001, Australia
(Received 18 November 1995, and in nal form 2 April 1996)
A simple procedure based on the nite element method has been developed for treating
the dynamic analysis of beams on an elastic foundation subjected to moving point loads,
where the foundation has been modelled by springs of variable stiffness. The effect of the
speed of the moving load, the foundation stiffness and the length of the beam on the
response of the beam have been studied and dynamic amplications of deections and
stresses have been evaluated. The technique is extended to the analysis of railway track
structures, where the effect of the spring stiffness of the moving load is also incorporated.
The entire analysis has been programmed to run on a microcomputer and gives fast and
accurate results. Several numerical examples are presented. The technique and the ndings
will be useful in railway track design.
7 1996 Academic Press Limited
1. INTRODUCTION
Investigation of the response of beams on an elastic foundation subjected to static or
moving loads has attracted engineers and researchers for many decades. In his classic
monograph, Hetenyi [1] has presented a closed form solution for an innitely long beam
on an elastic foundation under static loads and series solutions for the cases of nite beams.
The outcome of Hetenyis work and the subsequent work of others has been mainly applied
to the analysis and design of railway tracks, with most of the research pertaining to static
analysis [2]. However, it is well known that when a structure is subjected to moving loads,
there will be amplications in the deections and stresses in comparison to those obtained
from a static analysis of the structure subjected to the same loads. Although more
complicated, the signicance of dynamic analysis is thus evident [3]. Moreover, dynamic
analysis is an important aspect in any complete structural investigation. A signicant
feature in the analysis of beams on elastic foundation is the rapid damping of the response
away from the load [1, 2].
Timoshenko et al. [4] solved the governing differential equation for the dynamic analysis
of a simply supported beam subjected to moving loads by mode superposition. They found
that the maximum dynamic deection was 15 times the static deection when the travel
time was half of the fundamental period of the structure. Warburton [5] analytically
investigated the same problem and found that the maximum dynamic amplication in
deection was 1743, and that this occurred when the travel time was 081 times the
fundamental period of the structure. This nding was later conrmed by nite element
analysis. Dynamic response of multiple-span beams has been studied by Ayre et al. [6] and
Honda et al. [7], where the effects of span length and the number of spans on the dynamic
response were examined. Many researchers have used the nite element method to
149
0022460X/96/470149 +21 $25.00/0 7 1996 Academic Press Limited
b. 1n:xnir:1N:x :Nb .. znicr 150
investigate the dynamic response of beams under moving loads [3, 8, 9]. The Wilson u
method or the Newmark method was used in the numerical integration of the governing
equations of motion.
Dynamic analysis of beams on an elastic foundation (BEF) under moving loads has
received less attention, even though the results could be readily applied to the analysis and
design of railway tracks. Timoshenko et al. [4] analytically solved for the free vibration
of beams on elastic foundations. Ono and Yamada [10] presented a classic method for free
and forced vibration analysis of BEF. Trochanis et al. [11] presented a method for
analyzing beams on elastic foundations under moving loads and applied the results to the
analysis of railway tracks. They used the fast Fourier transform (FFT) technique in their
analysis. The beam on an elastic foundation under a moving load has also been treated
by Kenny [12], Fryba [13] and Fryba et al. [14].
Thambiratnam and Zhuge [2, 15] have developed a simple nite element model to
analyze simply supported beams on an elastic foundation (BEF) of any length. At rst this
model was employed in the static analysis of BEF and the results were applied to railway
track structures, and then the model was used in free vibration analysis of BEF. In the
present paper, the technique is extended to the analysis of BEF subjected to moving loads.
This extension has resulted in one simple procedure being now available for the static, free
vibration and dynamic analyses of beams on an elastic foundation. Axial effects and
variations in foundation properties along the length of the beam can be easily
accommodated in the method, which will yet have a signicant reduction in complexity
in comparison to any of the presently available methods.
In the nite element model, the elastic foundation is represented by springs with known
stiffness. The moving concentrated load is assumed to travel along the beam with constant
velocity. Newmarks method is used in the numerical integration of the equations of
motion. The length of the simply supported beam, the speed of the moving load and the
magnitude of the foundation stiffness are the main parameters in the study. Time histories
of deections and stresses can be obtained, from which the resulting dynamic
amplications can be calculated. The inuence of the suspension stiffness on the dynamic
amplications is also investigated.
The effect of the length of the beam on the response was particularly studied in order
to extend the application of the method to railway tracks. Finally, the effects of two
moving loads travelling on a beam were studied. It is evident that by choosing an
appropriate value for the length of the beam, and the number of elements to model the
beam, a simple but efcient method can be obtained for the dynamic analysis of beams
on an elastic foundation or of railway tracks.
2. FORMULATION OF THE METHOD
2.1. rirxrN1 s1irrNrss x:1rix
Consider an element ij of length L of a beam on an elastic foundation as shown in Figure
1, having a uniform width b and a linearly varying thickness h(x). It will be a simple matter
to consider an element having a linearly varying width if the need arises. Neglecting axial
deformations, this beam on an elastic foundation element has two-degrees-of-freedom per
node; a lateral translation and a rotation about an axis normal to the plane of the paper,
and thus possesses a total of four degrees of freedom. The (4 4) stiffness matrix k of
the element is obtained by adding the (4 4) stiffness matrices k
B
and k
F
pertaining to the
usual beam bending and foundation stiffness respectively. Since there are four end
b.N:xic :N:i.sis or nr:xs 151
displacements (or degrees of freedom), a cubic variation in displacement is assumed, in
the form
v =Aa, (1)
where A=(1 x x
2
x
3
) and a
T
=(a
1
a
2
a
3
a
4
). The four degrees of freedom
corresponding to the displacements v
1
, v
3
and the rotations v
2
, v
4
at the longitudinal nodes
are given by
q =Ca, (2)
where q
T
=(v
1
v
2
v
3
v
4
) and C is the connectivity matrix for an element ij between x =0
and x =L (Figure 1).
From equations (1) and (2),
v =AC
1
q. (3)
If E is the Youngs modulus, and I =bh(x)
3
/12 is the second moment of area of the beam
cross-section about an axis normal to the plane of the paper, the bending moment M in
the element is given by
M=D d
2
v/dx
2
=DBC
1
q, (4)
where D=EI(x) and B=d
2
A/dx
2
=(0 0 2 6x).
The potential energy U
B
due to bending is
U
B
=
1
2
g
l
0
d
2
v
dx
2
M dx (5)
which, upon using equations (3) and (4), becomes
U
B
=
1
2
q
T
(C
1
)
T
6g
l
0
B
T
DB dx
7
C
1
q. (6)
Upon using Lagranges equations, the (4 4) element stiffness matrix k
B
is obtained from
the potential energy of the element as
k
B
=(C
1
)
T
k
B
C
1
, (7)
where
k
b
=
g
l
0
B
T
DB dx. (8)
Figure 1. A beam on an elastic foundation element.
b. 1n:xnir:1N:x :Nb .. znicr 152
The above expression can be evaluated explicitly. If the element has linearly varying
thickness, this can be taken into consideration by using the appropriate expression for I.
The potential energy U
F
due to the foundation stiffness is given by
U
F
=
1
2
g
l
0
v
T
k
f
v dx, (9)
where k
f
is the stiffness of the foundation. Use of equation (3) in equation (9) will yield
U
F
=
1
2
q
T
(C
1
)
T
6g
l
0
A
T
k
f
A dx
7
C
1
q. (10)
Using Lagranges equations, the (4 4) stiffness matrix k
F
pertaining to the foundation
stiffness is given by
k
F
=(C
1
)
T
k
F
C
1
, (11)
where
k
F
=
g
l
0
A
T
k
f
A dx. (12)
The above expression can also be evaluated explicitly and, nally, the complete stiffness
matrix for the element is
k =k
B
+k
F
. (13)
2.2. rirxrN1 x:ss x:1rix
For dynamic analysis, it is also necessary to derive the element mass matrix. The element
mass matrix is a matrix of equivalent nodal masses that dynamically represent the actual
distributed mass of the element. In this investigation, the mass matrix is derived by
considering the kinetic energy due to lateral velocity. This is consistent with the derivation
of the stiffness matrix where axial effects were ignored.
The kinetic energy of the element shown in Figure 1 is given by
T=
1
2
g
l
0
(v )
T
r dVv , (14)
where the lateral velocity v is given by the time derivative of the displacement v and r is
the mass density. Further simplication gives
T=
r
2
(q )
T
(C
1
)
T
6g
l
0
A
T
h(x)A dx
7
C
1
q . (15)
Using Lagranges equations, on the kinetic energy term above, the mass matrix m is given
by
m=(C
1
)
T
m C
1
, (16)
where
m =r
g
l
0
A
T
h(x)A dx. (17)
b.N:xic :N:i.sis or nr:xs 153
Figure 2. A moving load on a simply supported beam on an elastic foundation.
2.3. : nr:x oN :N ri:s1ic roiNb:1ioN sinrc1rb 1o : xoviNc coNcrN1r:1rb io:b
Consider a beam on an elastic foundation as shown in Figure 2, with a moving
concentrated force travelling along the beam. The beam has been discretized into a number
of nite elements. Following the usual procedure for stiffness analysis of structures, the
governing equation of motion for the beam can be represented as
[M]{q} +[C]{q } +[K]{q} ={f } =[N]
T
f
0
, (18)
where [M] is the structure mass matrix, [C] is the damping matrix, [K] is the structure
stiffness matrix, [N]
T
is the transpose of the shape functions for the beam element [6] which
are evaluated at the position of the force, f
0
is the magnitude of the concentrated force,
and {q}, {q } and {q} denote the displacement, velocity and acceleration vectors
respectively.
In equation (18), damping effects are neglected and the shape functions can be
represented as
[N] =[0 0 0 . . . N
1i
N
2i
N
3i
N
4i
0 0 0 . . . ], (19)
where
N
1i
=1 3
0
x
l
1
2
+2
0
x
l
1
3
, N
2i
=x
0
x
l
1
1
2
,
N
3i
=3
0
x
l
1
2
2
0
x
l
1
3
, N
4i
=x
$0
x
l
1
2

x
l
1%
, (20)
in which i is the number of the element on which the load is acting.
[N]
T
is a vector with zero entries except for those corresponding to the nodes of the
element on which the load is acting [3]. For a beam element in this study, the number of
non-zero entries within the n 1 vector will be four. This 4 1 sub-vector is time
dependent as the load moves from one position to another within an element. As the load
moves to the next element, this sub-vector will shift in position corresponding to the
degrees of freedom of the element on which the load is then positioned.
The Newmark method of direct step-by-step integration is employed in the present study
to solve the governing equation (18) and the numerical procedure is implemented in a
FORTRAN program. The program has been run on a mainframe computer by using the
CONVEX-220 system.
3. NUMERICAL EXAMPLES AND DISCUSSION
In this section numerical examples are treated to illustrate the procedure, and the effects
of some parameters are investigated.
b. 1n:xnir:1N:x :Nb .. znicr 154
3.1. b.N:xic :xiiiric:1ioNs iN brrirc1ioNs
3.1.1. A simply supported beam without foundation stiffness
To validate the method, a simply supported beam without an elastic foundation
subjected to a concentrated force moving with constant velocity, is analyzed and the results
are compared with those from the existing analytical solution of Warburton [5] and nite
element analysis of Lin and Trethewey [3], where the set of second order differential
equations have been solved by the RungeKutta numerical integration scheme.
Results for the dynamic amplication factors f
D
, dened as the ratio of the maximum
dynamic and static deections at the center of the beam, are computed and compared in
Table 1, for different values of t/t, where t denotes the travelling time of the force moving
from the left end of the beam to the right end, while t denotes the time after the moving
load enters the beam from the left end.
It can be seen that the present results, obtained with only four elements modelling the
beam, compare quite well with the results of the others and thereby conrm the validity
of the proposed numerical procedure. The maximum value of the dynamic amplication
is about 17 and occurs at t/t =1234. The maximum discrepancy between the present
results and those of others is less than 5%, which is acceptable, especially as the results
in reference [3], obtained from a numerical model, and those in reference [5], obtained by
considering only the rst mode, may not be exact. With a ner mesh, results obtained from
the present procedure, match more closely the results from the numerical approach in
reference [3]. This problem is not directly relevant to the scope of the present study, but
it was treated as a test case to validate the numerical procedure. The technique is extended
to beams on an elastic foundation in the next section.
3.1.2. The effect of foundation stiffness
A simply supported beam of length 10 m resting on a uniform elastic foundation, as
shown in Figure 2, is considered. The elastic modulus of the beam, E=205e11 N/m
2
, the
Poisson ratio n =03 and the second moment of area I =1844e4 m
4
. The moving load
has a constant velocity V=167 m/s (60 km/h), and the foundation stiffness k
f
is varied
from 0 to 114e8 N/m
2
, and includes values typical for railway tracks. Results for the
dynamic amplication of mid-span deections are shown in Figure 3. It can be seen that
the dynamic amplications f
D
initially increase as k
f
increases. However, when k
f
is
increased beyond the value of 114e6 N/m
2
, the dynamic amplications f
D
decrease. The
dynamic amplication is largest for k
f
=114 e6 N/m
2
, and this value is about three times
the value for k
f
=0.
T:nir 1
Dynamic amplications, f
D
fD
ZXXXXXXXXXXXXXXXCXXXXXXXXXXXXXXXV
t/t This study Warburton [5] Lin and Trethewey [3]
01 1040 1040 1053
05 1330 1250 1252
10 1710 1710 1705
1234 1723 1740 1730
15 1630 1710 1704
20 1500 1550 1550
b.N:xic :N:i.sis or nr:xs 155
Figure 3. The effect of foundation stiffness on the dynamic amplication in mid-span deection: V=60 km/h.
When the velocity is changed to V=81 m/s (29 km/h), similar results are obtained, as
shown in Figure 4, where the dynamic amplications at the beam center are plotted
with respect to the foundation stiffness. These results indicate that when a concentrated
force travels along a beam, the dynamic effects are greatly inuenced by the foundation
stiffness.
Figure 4. The variation of the peak value of the dynamic amplication in deection with foundation stiffness:
V=81 m/s; length =10 m.
b. 1n:xnir:1N:x :Nb .. znicr 156
Figure 5. The effect of travelling speed on the dynamic amplication in mid-span deections V=15, 30, 60
and 100 km/h; kf =114e7.
3.1.3. The effect of travelling speed
The same beam as treated in section 3.1.2 is considered, but with the beam resting on
a foundation having a stiffness k
f
=114e7 N/m
2
. The velocity of the moving load is varied
from 41 m/s (15 km/h) to 278 m/s (100 km/h). The results for the dynamic amplications
in mid-span deections are shown in Figure 5. It is interesting to note that the dynamic
amplications are reasonably constant for various travelling speeds, within the range
considered. However, at smaller values of the foundation stiffness, the dynamic
amplication increases slightly with travelling speed, as shown in Figure 6, which also
shows a slight decrease in the dynamic amplication for the largest value of the foundation
stiffness treated. Hence, when the foundation stiffness is relatively large, as in the case of
foundations of railway tracks, the inuence of the travelling speed, in the range
15100 km/h, is quite insignicant as amply demonstrated by the results in Figures 5 and 6.
3.1.4. The effect of the span length of the beam
In order to study the inuence of the length of beam on its dynamic response, a simply
supported beam with a constant foundation stiffness k
f
=114e7 N/m
2
is considered. The
moving load has a velocity V=81 m/s (29 km/h). The length of the beam is varied from
5 m to 40 m and, for each case, the dynamic amplications on the mid-span deections
are shown in Figure 7. It can be seen that when the span length Le10 m, these dynamic
amplications on the mid-span deection remain constant.
However, at lower values of k
f
this is not the case, as shown in Figure 8. When the
foundation stiffness k
f
is reduced to 114e2 N/m
2
, the peak dynamic amplication increases
with the span, as shown in the gure, and no convergence is observed for the range of
L considered here. Practical values of k
f
are greater than 114e5 N/m
2
, and for this range
the dynamic amplications converge (to a value of approximately 36) for Le20 m, and
the effect of speed is not signicant for such cases.
b.N:xic :N:i.sis or nr:xs 157
3.2. b.N:xic :xiiiric:1ioN iN s1rrssrs
In the design of beams on an elastic foundation or of railway tracks, the most important
considerations are the allowable bending stress and allowable vertical deection. In the
previous section, dynamic amplications in deections of beams on an elastic foundation
subjected to moving loads were treated and the inuence of certain parameters was studied.
In this section a similar treatment is presented for the dynamic amplications in bending
stresses. For this purpose, the dynamic amplications in stress f
S
can be dened as the
ratio of the maximum dynamic stress to the maximum static stress at the center of the
beam. The maximum bending stress s
S
in a beam may be calculated using the simple
applied mechanics formula in equation (21):
s
S
=M
m
/Z
0
, (21)
where M
m
is maximum bending moment (kN m) and Z
0
is the section modulus of the beam
(m
3
).
3.2.1. The effects of the span length of beam
A simply supported beam resting on a foundation with a stiffness k
f
=114e7 N/m
2
is
considered. A concentrated load moves on this beam with a velocity V=80 km/h. The
length of the beam is varied from 5 m to 40 m and the resulting dynamic amplications
in stresses f
S
are shown in Figure 9. It can be seen that the values of the dynamic
amplications of the mid-span stress are reasonably constant for different span lengths.
However, as shown in Figure 10, when k
f
was reduced to 114e2 N/m
2
, these dynamic
amplications increase with the span length of the beam and no convergence is noticed
for the range of span lengths considered. When k
f
=114e5 N/m
2
, the dynamic
amplications sensibly converge when Le10 m, as shown in this gure.
Figure 6. Variation of peak values of dynamic amplication in deection with travelling speed. +, kf =114e2;
r, kf =114e4; , kf =114e5; e, kf =114e7.
b. 1n:xnir:1N:x :Nb .. znicr 158
Figure 7. The effect of the span length of beam on the dynamic amplication in mid-span deections.
Figure 8. The variation of peak values of dynamic amplication in deection with span length. r, kf =114e2;
, kf =114e5; e, kf =114e7.
3.2.2. The effects of foundation stiffness
A simply supported beam of length L=10 m is considered, on which the load travels
with a velocity of V=167 m/s (60 km/h). The stiffness of the foundation is varied from
0 to 114e8 N/m
2
. This range covers the typical foundations of railway tracks. For each
case the dynamic amplications f
S
were calculated and are presented in Figure 11. It can
be seen that initially, the dynamic amplication increase with k
f
. When k
f
is increased
b.N:xic :N:i.sis or nr:xs 159
Figure 9. The effect of the span length on the dynamic amplication in mid-span stress.
beyond the value of 114e5 N/m
2
, the dynamic amplications decrease with k
f
. These
results are similar to those obtained for deections. However, it is evident that the dynamic
amplications in the stresses are less than those in the deections. The maximum values
of the dynamic amplications are about 36 for deection, and less than 2 for stress. This
Figure 10. The variation of peak values of dynamic amplication in stress with span length. r, kf =114e2;
, kf =114e5; e, kf =114e7.
b. 1n:xnir:1N:x :Nb .. znicr 160
Figure 11. The effect of foundation stiffness on the dynamic amplication in mid-span stress. V=167 m/s.
trend (viz., a smaller dynamic amplication for stress) has also been observed in the study
of dynamic amplications in bridges due to moving loads [21], and could be explained as
follows. Stresses considered in these studies are those due to bending effects and are
obtained from the curvatures or second derivatives of the vertical deections. It is not
necessary for a function (deection) and its second derivative (stress) to have the same
dynamic amplication and in the present case amplications in the second derivatives of
the deection are smaller than those in the deection.
Beams on an elastic foundation with other support (boundary) conditions can be treated
with equal ease by the numerical procedure established in this paper. In order to illustrate
the versatility of the present model, the problem treated above is re-analyzed with the beam
having xed supports. Dynamic amplications in the bending stresses f
S
are shown in
Figure 12 for various values of foundation stiffness. It can be seen that the trends are
similar to those for a simply supported beam, but with the maximum value occurring for
k
f
=114e6 kN/m
2
.
3.2.3. The effect of travelling speed
A simply supported beam of span L=10 m and resting on a foundation having a
stiffness k
f
=114e7 N/m
2
is considered. The velocity V of the moving load is varied from
60 km/h to 120 km/h. The results for the dynamic amplications in stresses f
S
are shown
in Figure 13. The values of the dynamic amplications in the mid-span stress remains more
or less constant at the different speeds. It is evident that these results are similar to those
for the deections shown in Figure 5. The maximum dynamic amplication can be
observed at a speed of 100 km/h. Again, the dynamic effects on the stresses are smaller
than those on the deections, the average peak dynamic amplication being about 16 for
stresses and 37 for deections.
b.N:xic :N:i.sis or nr:xs 161
In Figure 14 is shown the effect of the travelling speed on the dynamic amplications
in the mid-span stress for different values of foundation stiffness. It can be seen that when
k
f
is reduced to 114e2 N/m
2
, dynamic amplications increase with travelling speed, up to
about V=60 km/h, a trend observed earlier with dynamic amplication in deections.
This gure also shows that convergence characteristics depend on the foundation stiffness.
4. MODELLING INFINITELY LONG BEAMS ON AN ELASTIC FOUNDATION:
APPLICATION TO RAILWAY TRACKS
4.1. n:cicroiNb
Despite over 100 years of operating experience, the design of railway tracks usually
depends to a large extent upon the engineering experience of the designer [17, 18]. The
available design expressions are at best empirical. This is because the methods of analyzing
railway tracks are complicated, as the tracks are innitely long and supported on beds the
properties of which can vary along the track length. Moreover, the dynamic response
characteristics of the tracks are not well enough understood to form a rational design
method.
Although research in seeking simpler and/or improved methods of track analysis has
been going on, rail selection procedures and design have remained relatively static for over
100 years [19]. This is exemplied by the continuous common adoption of the quasi-static
design approach together with the beam on elastic foundation analysis of tracks. However,
various safety factors have been introduced to keep pace with the gradual increase in
severity of operating conditions. Tew et al. [19] have given a comprehensive coverage of
some of the more important quasi-static methods. Kerr [20] has presented an interesting
compilation of papers on various aspects pertaining to railway tracks.
Figure 12. The effect of foundation stiffness on the dynamic amplication in mid-span stress: xed beam.
V=60 km/h.
b. 1n:xnir:1N:x :Nb .. znicr 162
Figure 13. The effect of travelling speed on the dynamic amplication in mid-span stresses.
The current practice for designing the railway track structure is based upon satisfying
several criteria for strength of the individual components [18, 19]. The important criteria
for the railway track are the allowable bending stress and the allowable vertical deection.
For the purpose of analysis, a railway track is treated as a beam on an elastic foundation
Figure 14. The variation of the peak values of dynamic amplication in stress with travelling speed.
r, kf =114e2; , kf =114e4; e, kf =114e5; E, kf =114e7.
b.N:xic :N:i.sis or nr:xs 163
Figure 15. The effective length of BEF for modelling railway tracks. Q, k/kf =0; q, kf/kf =01; R, k/kj =02.
(BEF)a model rst proposed by Winkler in 1867 and used by Zimmermann in 1888, over
100 years ago. Using the BEF model, Hetenyi [1] analyzed railway track under static loads
by solving the governing differential equation. Hetenyis classic solution is still used, often
in the form of a computer program, to analyze and design railway tracks. The approach
is quasi-static where the design load is obtained by multiplying the static wheel load by
one of many available impact factors [17, 19] to account for dynamic effects. There are
several such formulae, each with its merits and/or limitations. For want of a more
comprehensive dynamic analysis procedure, this is the method used at present.
There are, however, some shortcomings in the present method of analyzing and
designing railway tracks: the analysis is tedious and not well understood; and the effects
of moving loads cannot be fully accounted for by using impact factors in a quasi-static
approach and variations in track and/or foundation properties along the length cannot be
accounted for.
In an earlier paper by Thambiratnam and Zhuge [2], a special technique was proposed
to handle innitely long beams (or tracks) by using an equivalent nite beam for static
analysis. The length of this beam was chosen so that when it is subjected to a concentrated
load the response curves for the deection, the bending moment and the shear force are
rapidly damped away from the load [1]. This equivalent beam of nite length has been
tested to give converging results. The stiffness of the spring represents the stiffness of the
rail foundation, which comprises the sleepers, the ballast and the subgrade.
4.2. 1nr rrrrc1ivr irNc1n or 1nr nr:x
To extend the present procedure to the analysis of railway tracks, it is necessary to
identify the required span length of the beam on an elastic foundation. The rapid damping
of the beam response away from the load enables the track to be modelled as a nite beam
on an elastic foundation. The point load moving at a constant speed is representative of
the usual wheel load, currently used in the analysis and design of railway tracks.
Convergence studies for beam deections and stresses carried out in the earlier sections
of the paper can be used to obtain this nite span length, with the appropriate value for
the foundation stiffness k
f
. For rail foundations, the range of values for this stiffness is
from 52e6 N/m
2
to 354e7 N/m
2
. From the results in the earlier sections, it can be seen
that, for this range of k
f
, dynamic amplications in both deections and stresses converge
for span lengths Le10 m. The corresponding value of the dynamic amplication in stress
is about 16, in the absence of damping.
b. 1n:xnir:1N:x :Nb .. znicr 164
4.3. xobriiiNc 1nr xoviNc io:b :s : siriNc x:ss
It is necessary to address the applicability of the simple model developed herein to
railway track analysis. Railways track which have been modelled and analyzed as beams
on an elastic foundation have stood the test of time. However, the dynamic effects on the
track response have not been fully understood, as mentioned earlier, and in this paper an
attempt is made to investigate the dynamic amplications, in order to provide a more
realistic design. As with other track analysis models, the model proposed in this paper also
relies on the same beam on elastic foundation concept. Real loads on railway tracks are
sprung masses, although they have been simplied up to now as moving point loads. In
this section the effect of modelling the moving load as a sprung mass will be investigated.
Following a procedure similar to that used by Lin and Trethewey [3] and ignoring damping
and the mass of the wheel, the governing equations can be derived as
$
[M]
0
[N]
T
m
1
m
1 %6
q
y
7
+
$
[K]
k[N]
0
k
%6
q
y
7
=
6
[N]
T
m
1
g
0
7
, (22)
where m
1
is the sprung mass, k is the spring stiffness and y is its deection, measured from
the static equilibrium position before the moving load enters the beam. Double dots above
quantities denote second time derivatives of those quantities. The other notations are as
before. A convergence study was carried out to determine the effective length of the simply
supported beam on an elastic foundation to be used in modelling railway tracks. In Figure
15 it is shown that when Le10 m, dynamic amplication in the mid-span bending stress
converges for stiffness ratios in the range 0 Qk/k
f
Q020, where k is the stiffness of the
sprung mass. Analogous results were obtained for f
D
, the dynamic amplication in
deection. Therefore, to illustrate the effect of the sprung mass on the dynamic
Figure 16. Dynamic amplications in mid-span deections: the sprung mass case.
b.N:xic :N:i.sis or nr:xs 165
Figure 17. Dynamic amplications in mid-span stresses: the sprung mass case.
amplications in the railway track response, a 10 m long simply supported beam resting
on an elastic foundation with k
f
=114e7 N/m
2
is considered, on which the mass moves
with a velocity of 166 m/s (60 km/h). Dynamic amplications in the mid-span deections
f
D
and the mid-span stresses f
S
are shown in Figures 16 and 17 for different values of
the stiffness ratio k/k
f
. It can be seen that the effect of the spring stiffness is a reduction
in the dynamic amplications. When k/k
f
=02, there are roughly 21% and 26%
reductions in the deection and stress amplications respectively.
4.4. iN1rr:c1ioN or :b:crN1 vnrri io:bs
To determine the worst effects at a point on the railway track, it may be necessary to
superpose bending moments and deections caused by adjacent wheel loads, as shown in
Figure 18. To investigate the effect of adjacent wheel loads, a beam with length L=10 m,
Figure 18. Typical wheel loads on railway tracks.
b. 1n:xnir:1N:x :Nb .. znicr 166
Figure 19. Time histories of maximum deections due to one- and two-wheel loads, v =120 km/h. L=10 m;
x =32; V=120 km/h.
resting on a foundation having a stiffness k
f
=114e7 N/m
2
, is considered. The beam is
subjected to two wheel loads, P
0
and P
1
, at a constant distance apart of X
1
=32 m and
moving with a velocity V=120 km/h.
The results for the maximum vertical deections w and maximum bending stresses s
S
in the beam, which occur at mid-span, are shown in Figures 19 and 20 respectively,
together with those obtained for a single wheel load, travelling with the same speed. It can
be seen that the peak values of both deections and stresses are more or less the same with
Figure 20. Time histories of maximum stresses due to one- and two-wheel loads, v =120 km/h. L=10 m;
x =32; V=120 km/h.
b.N:xic :N:i.sis or nr:xs 167
T:nir 2
Impact factors, f
India f=1 +
V
5814(kf )
05
Germany (for speeds up to 100 km/h) f=1 +
V
2
3 10
4
South Africa f=1 +492
V
D
Clarke f=1 +
1065V
D(kf )
05
AREA f=1 +521
V
D
WMATA f=(1 +386 10
5
V
2
)
2/3
either one or two wheel loads. However, the duration of the peak response, especially the
deection, is increased with two wheel loads acting on the beam. Therefore, in a real
railway track structure, the duration of the peak response will depend on the speed of the
train and its length, which in turn will determine the number of axles contributing to the
maximum response. Similar results were obtained when the travelling speed was reduced
to V=60 km/h.
4.5. ixi:c1 r:c1or ror r:iiv:. 1r:ci brsicN
Dynamic loading on railway tracks has been subject of extensive investigation by railway
authorities throughout the world. The main method adopted to cater for dynamic effects
in the design is to apply an impact factor to the static wheel load; i.e.,
P=f P
S
, (23)
where P is the design wheel load (kN), P
S
is the static wheel load (kN) and f is a
dimensionless impact factor (q1). The expressions used for the calculation of the impact
factor have been determined empirically and are always expressed in terms of train speed.
To develop expressions for the impact factor, the major factors have been [19] the train
speed, the wheel diameter, the vehicle unsprung mass, the track condition (including track
stiffness, geometry and joint condition), the track irregularities, the track construction, the
static wheel load and the vehicle condition.
Seven impact factor formulae are shown in Table 2, where D is the diameter of the wheel.
Values of impact factors have been calculated from these formulae, by using the speeds
and foundation stiffnesses used in the examples treated in this paper. In all of the above
formulae, the travelling speed V has been considered to be the most important factor,
whereas the effect of foundation stiffness is considered in only two formulae. In the present
study on dynamic analysis, it has been found that the foundation stiffness k
f
has a
signicant effect on the response of the beam. Impact factors calculated from the various
formulae give an average value of about 15. This value is close to the maximum dynamic
amplication in stress f
S
obtained in this paper. However, for vertical deection, the
dynamic analysis indicated a higher amplication, at about 3. When damping effects are
considered, the amplications can be expected to diminish. Moreover, as shown earlier,
there will be a further reduction in the dynamic amplications when the effect of spring
stiffness of the wheel load is included in the analysis.
b. 1n:xnir:1N:x :Nb .. znicr 168
5. CONCLUSIONS
A simple nite element method for the dynamic analysis of beams on an elastic
foundation, subjected to a concentrated moving load, has been presented in this paper.
This technique will be attractive for treating beams on an elastic foundation under moving
loads in general and, as shown in the paper, can be easily extended to treat railway track
structures. The effects of some important parameters, such as the foundation stiffness, the
travelling speed, the length of beam and the stiffness of the sprung mass, have been studied.
Dynamic amplications in stresses and deections have been found to be about 16 and
3 respectively, when the moving load was modelled without any spring stiffness. These
amplications diminish with the spring stiffness. Impact factors calculated from several
formulae gave an average value of about 15 for the parameters used in the present study.
Damping is not included in the present formulation but its effect will be a reduction in
the amplitudes of the responses. The entire analysis has been conveniently programmed
and, with few elements, gives converging results which, for a limiting case, compare well
with those from existing solutions.
REFERENCES
1. M. Hr1rN.i 1961 Beams on Elastic Foundation. Ann Arbor, MI: The University of Michigan
Press.
2. D. P. Tn:xnir:1N:x and Y. Znicr 1993 Journal of Microcomputers in Civil Engineering 8,
467476. Finite element analysis of track structures.
3. Y. H. LiN and M. W. Trr1nrvr. 1990 Journal of Sound and Vibration 136, 323342. Finite
element analysis of elastic beams subjected to moving dynamic loads.
4. S. TixosnrNio, D. H. YoiNc and W. Wr:vrr 1974 Vibration Problems in Engineering. New
York: John Wiley; fourth edition.
5. G. B. W:rnir1oN 1976 The Dynamic Behaviour of Structures. Oxford: Pergamon Press.
6. R. S. A.rr, F. Forb and L. S. J:consrN 1950 American Society of Mechanical Engineers,
Transactions of the Journal of Applied Mechanics 17, 112. Transverse vibration of a two-span
beam under the action of a moving constant force.
7. H. HoNb:, T. Konori and H. Y:x:n: 1986 International Association of Bridge and Structure
Engineering 98, 5775. Dynamic factors of highway steel girder bridges.
8. F. V. Hiino 1978 Shock and Vibration Digest 10, 2735. Finite element analysis of structures
under moving loads.
9. J. HiNo, T. Yosnixir: and K. KoNisni 1986 Journal of Sound and Vibration 96, 4553. A nite
element method prediction of the vibration of a bridge subjected to a moving vehicle load.
10. K. ONo and M. Y:x:b: 1989 Journal of Sound and Vibration 130, 269297. Analysis of railway
track vibration.
11. A. M. Trocn:Nis, R. Cnriii:n and J. Biri:i 1987 Journal of Geotechnical Engineering 113,
879895. Unied approach for beams on elastic foundations under moving load.
12. J. T. KrNNr. 1954 Journal of Applied Mechanics 21, 359364; 1955, 22, 436. Steady-state
vibrations of beams on elastic foundation for moving load.
13. L. Fr.n: 1972 Vibration of Solids and Structures under Moving Loads. Groningen: Noorhoff.
14. L. Fr.n:, S. N:i:ciri and N. Yosnii:v: 1993 Journal of Sound and Vibration 163, 3135.
Stochastic nite elements for a random foundation with uncertain damping under a moving
force.
15. D. P. Tn:xnir:1N:x and Y. Znicr 1993 in Proceedings of Second Asia-Pacic Conference on
Computational Mechanics, Sydney, Australia, 231235. A simple nite element method for free
vibration of beams on elastic foundation.
16. K. C. Rocir., H. R. Ev:Ns, D. W. Grirri1ns and D. A. Nr1nrrco1 1983 The Finite Element
Method. Oxford: BSP Professional Books.
17. C. Esvrib 1989 Modern Railway Track. Duisburg, Germany: MRT Productions.
18. N. F. Do.ir 1980 Railway Track Design: a Review of Current Practice. Canberra, ACT:
Australian Government Publishing Serivce.
b.N:xic :N:i.sis or nr:xs 169
19. G. P. Trv, S. M:ricn and P. J. Mi11oN 1991 A Review of Track Design Procedure. Melbourne,
Victoria: BHP Research Laboratories for Railways; Volume 1, Rails.
20. A. D. Krrr 1978 in Proceedings of a Symposium held at Princeton University, April 1975. Oxford:
Pergamon Press. Railroad track mechanics and technology.
21. J. SrN1niiv:s:N, D. P. Tn:xnir:1N:x and G. H. Br:xrib 1996 Research Report, Physical
Infrastructure Centre, Queensland University of Technology, Brisbane, Australia. Dynamic load
testing of a curved bridge.

You might also like