Anda di halaman 1dari 10

Compurers & .Strurtwes Vol. 53. No. 6, pp. 1427-1436.

i994
Copyright 8 1994 Elstvier Science Ltd
Pergamon ~5-794~~)EO2~W Printed in Great Britain. All rights reserved
0045.7949/94 $7.00 + 0.00

NATURAL VIBRATION ANALYSIS OF RAIL TRACK AS


A SYSTEM OF ELASTICALLY COUPLED BEAM
STRUCTURES ON WINKLER FOUNDATION
Z. Cai,t G. P. Raymond$$ and R. J. Bathurst?
TDepartment of Civil Engineering, Royal Military College, Kingston, Ontario, Canada, K7K 5LO
JDepartment of Civil Engineering, Queen’s University, Kingston, Ontario, Canada, K7L 3N6

(Received 9 May 1993)

Abstract-An analytical model for analyzing the vertical free vibration of a rail track is presented. The
track structure is represented as a system of eiasticahy coupled beam structures resting on a Winkler
foundation. The rail and the tie beams are described by any combination of the two existing beam theories,
the Bernoulli-Euler type, and the Timoshenko type, while the rail is assumed to be periodically supported
at discrete points on cross-track tie beams. A generalized track element, which consists of a rail span (beam
segment), two adjacent ties, and the coupling spring stiffnesses, is established to discretize the track system
into identical units. A concept of an equivalent frequency-dependent spring coefficient for the rail support
system is introduced to formulate the dynamic stiffness matrix of the track element. Solutions are provided
for the natural frequencies of the track and the associated mode shapes of the rail and the ties under
transversely (cross-track) symmetric vibration. The free vibration results are used to obtain the dynamic
receptance response of a typical field track and to compare them with an existing model and field
experimental data.

NOTATION subscript r, t rail, tie, respectively


area subscript i count of tie number, or rail/tie intersection
integration coefficients point
integration coefficients subscript m, n mode number
rail pad damping (point damper)
track bed damping (distributed damper)
length from tie end to rail seat INTRODUCTION
bending rigidity
force vector of track element The classical analysis of a rail track treats the rail as
shear modulus an infinitely long Bernoulli-Euler type beam resting
equivalent frequency-dependent spring co- on a Winkler foundation which represents the com-
efficient
bined stiffness of the rail support system: the rail
rail/tie contact (pad) stiffness (point spring)
track bed stiffness (distributed spring) pads, the ties and the track bed [l-3]. Solutions for
dynamic stiffness matrix of track element this class of track model are usually in closed forms
global dynamic stiffness matrix of track limited on determination of the maximum travelling
tie spacing speed of a moving load at which instability of the
generalized mass of tie as individual beam
on elastic foundation track takes place. Recent studies have encompassed
unit mass per length (of rail or tie) a much broader range of topics related to rail track
total number of ties included in the model vibrations. Many authors, however, have followed
axial force in rail the traditional approach and represented the track as
radius of gyration (of rail or tie)
a continuous rail beam supported uniformly on an
rail n th mode shape
tie mode shape as indivjduai beam elastic foundation [4-l 11. The discrete nature of the
tie mode shape in nth mode of track rail support system and the bending effects of the ties
natural frequency of entire track have been investigated by only a limited number of
tie natural frequency authors [12-141. In most cases, the rail is described by
tie rotation angle as individual beam
tie rotation angle in n th mode the Bernoulli-Euler theory (or simple beam theory)
rail rotation angle in nth mode of track that considers only the flexure and vertical inertia of
globai nodal disp~a~ment vector of rail the beam. The Timoshenko beam theory, which takes
displacement vector of track element into account the effects of the shear distortion and the
shear distortion rigidity
rotatory inertia of the beam, has only previously been
Timoshenko shear coefficient
used to describe the rail when the rail support
components are modelled as a continuous Winkler
8 To whom correspondence should be addressed foundation [15, 161. To the authors’ knowledge, the

1427
1428 Z. Cai et al.

Timoshenko beam theory has not been applied to elements representing the resilience of the rail pads
modelling both the rail and the tie as flexible beams and rail-fastening mechanisms. As a simplification,
while the rail is discretely supported on the ties. The the ties can also be considered as rigid masses. To
use of the Timoshenko theory in such a discretely consider concrete ties that have deeper shoulder
supported track model should represent the most sections, the tie beam can be non-uniform as shown
recent approach in modelling rail track vibration in Fig. l(b). The distributed spring constants beneath
problems. each tie represent the resilience of the track foun-
The track vibration mode1 presented herein con- dation (ballast and subgrade). A similar discrete
siders a ballasted tie track where both the rail and the support track model, but where only the
tie may be described by any combination of the two Bernoulli-Euler beam theory can be used, was devel-
beam theories, while the rail is discretely supported oped by Clark et al. [13] and Newton and Clark [15].
on the ties. The tie can also be simplified as a discrete
mass element. In addition, an axial force in the rail FORMULATION OF GENERALIZED TRACK ELEMENT
is included to simulate thermal forces existing in the
rail. The solution procedure used to obtain the One distinct structural feature of a rail track is its
characteristics of the track natural vibration employs periodic chain-type configuration. To obtain sol-
a track discretization technique that involves the utions to the fundamental vibration characteristics of
formulation of a generalized track element and the the track mode1 shown in Fig. 1, a generalized track
derivation of an equivalent frequency-dependent element, as illustrated in Fig. 2(a), is devised. It is an
spring coefficient to represent the rail support com- assembly of one rail span and the sub-structure
ponents. The resilient, flexural and inertial character- system that includes the rail/tie contact stiffness el-
istics of the track components underlying the rail, ements (rail pads + rail fastening units), two adjacent
namely, the resilient rail/tie contact springs (rail pads tie beams, and the two arrays of track foundation
and/or rail fastening devices), the tie beams, and the springs beneath the ties. To reduce the degrees of
ultimate foundation, are completely preserved. The freedom involved in the element, the resilient, flex-
exact dynamic stiffness matrices of the beam members Ural, and mass inertia effects of the rail support
involved in the generalized track element are used to components in the sub-structure system are rep-
formulate the vibration problem so that accuracy is resented by an equivalent spring stiffness at each end
maintained for the solution of related high mode of the rail span, as seen in Fig. 2(b). Thus. the
eigenvalues. Only the natural vibration solution generalized track element is simplified to a uniform
methodology for the track system is presented herein. rail beam segment supported at the two ends by two
The application of the natural vibration model has identical general spring coefficients &. The equival-
formed the basis for solving forced vibrations of the ent spring coefficient depends on the inertia1 and the
track system subjected to any arbitrary loading func- elastic characteristics of the rail support components,
tions, such as impulsive forces [ 171 or dynamic forces as well as the vibration frequency, R, of the track. Its
resulting from train vehicle masses, suspension expression is given as follows:
springs, and irregular wheel/rail interactions [ 18. 191.

MODEL DESCRWHON

The idealization of the rail track is illustrated in


Fig. 1. Both the rails and the ties are described as where k, represents the rail/tie contact stiffness, z,, is
elastic beams, either of the Bernoulli-Euler type or the n th mode of the tie as an individual beam on an
the Timoshenko type. The rail is supported at discrete elastic foundation, d, is the distance from the rail seat
points on the ties through the rail/tie contact spring to the end of the tie, w, is the natural frequency of

(a) Longitudinal View

(b) Transverse (Cross-Track) View

Fig. 1. Idealized vertical rail track vibration model.


Natural vibration analysis of rail track 1429

(c) SingleSpan Element

Fig. 2. Formulation of generalized track element.

the individual free tie and IV,, is the corresponding denote the modal deflection and rotation angle of the
generalized tie mass. The summation term in the rail, respectively. Thus, the entire track structure is
denominator of the above relationship is for the tie idealized to a continuous assembly of the generalized
beam and is in terms of either only symmetrical track element and the adjacent single rail span el-
modes, or only asymmetrical modes. Equation (1) is ement (Fig. 2c) in a periodic manner.
derived (the procedure is omitted here) by applying
the method of modal analysis to the rail support DETERMINATION OF NATURAL FREQUENCIES
sub-structure system (tie beam on an elastic foun- AND NODAL DISPLACEMENTS
dation). This expression applies whether the tie is
The natural frequencies and associated vibrational
described by the simple Bernoulli-Euler theory or by
modes of the track are determined by formulating the
the Timoshenko theory. The use of this equivalent
overall dynamic stiffness matrix of the track. The
spring stiffness greatly simplifies the analysis, as is
resulting set of simultaneous equations is:
demonstrated below, and yet the resilient, flexural
and inertial characteristics of the underlying rail
support components are completely preserved. Note [~R@)I~AR
I= 0, (4)
that system damping is ignored for the natural vi-
bration analysis. Combining the above-defined spring where [I&(Q)] represents the global dynamic
coefficient with the exact dynamic stiffness matrix stiffness matrix, which is an assembly of the dynamic
(EDSM) of the rail span, the dynamic stiffness matrix stiffness matrices of the generalized track elements
of the generalized track element is formulated: and the single beam elements, and {A,} = [W,, 8,)
W,, 02, , W,, QNlT represents the harmonic dis-
placements of the rail at all the nodal points (or rail
r k,, + ke k,, k,, k,, 1
seats) of the track elements. Here N is the number of
iK,l = k*’
31
k,, k,3k;k
:;I e
3 t2)
ties included in the track model. Obviously, only a
limited extension of the track can be taken when
1 4, k,, ka k, j formulating the global dynamic stiffness matrix
[I&(R)]. A numerical example is given in a later
where k, is the element of EDSM for the uniform rail section to illustrate how the number of ties considered
span. It should be noted that each element k,, is a in the track model affects the solutions.
transcendental function of the frequency R. The The determinant of the overall dynamic stiffness
expressions for k, are provided in Appendix A for matrix needs to be rendered zero in order to obtain
both the Bernoulli-Euler beam theory and the Timo- non-trivial solutions of {AR} in eqn (4). that is:
shenko beam theory. Knowing the dynamic stiffness
matrix of the track element, the relationship between I&(WI = 0. (5)
the nodal forces and the corresponding displacements
shown in Fig. 2(b) becomes: This forms the characteristic frequency function of
the track system. The algorithm used to search the
eigenvalue R in the above equation is as follows: the
matrix [&(R)] is first triangulated at two successive
where {F} =V’i,Mi,F,+,,Mi+,lT, and (6,) = R values, Ri, R,, , . The number of negative diagonal
[wi*ei9wi+l*“j+llT7in which subscript i refers to the elements determines the sign of each determinant. If
ith node (rail/tie connection point), F, M denote the the signs are the same, no solution for R exists in this
nodal force and moment, respectively; and W, 0 frequency interval. If they are opposite, a root exists,
1430 Z. Cai et al.

as illustrated in Fig. 3(a). The Reguiu Falsi procedure FORTRAN 77 with double precision on an IBM
is then applied to search the root. To avoid false RISC 6000 workstation.
solutions, or ‘poles’, as illustrated in Fig. 3(b), the
determinant A0 of &(Q)] at the mid-interval, i.e., NATURAL MODES OF THE RAIL
ff, = (Q, + Q, ,)/2, is always caiculated for each fre-
Once the nodal displacements of the ith rail
quency interval. If the determinant A0 has the same
span, or the ith subvector of {Ann), i.e.,
sign as A, (or A,, I 1 and I& I < 141 (or l4l < I4 +, I)
14n7ein3 W,+l.n~ei+~,nlT~ are determined, the mode
then a root exists (Fig. 3a). Otherwise a pole exists
shape function of the ith rail span is obtained by
(Fig. 3b) and the next frequency increment is
solving the following ordinary differential equations
searched. In order to avoid numerical difficulties such
that govern the natural vibration of a uniform rail
as overflow during the calculation of the determinant,
the loga~thmic values of all the diagonal elements are beam, with {Ah).! = IB’+ @,, wi, I.“, Bi+Jr applied
as the boundary condition at the two ends.
used. The frequency increment ($+ , - Qi) needs to
If the rail is described by the Bernoulli-Euler beam
be sufficiently small so as not to miss any solutions.
theory:
As a double check, a method developed by Wittrick
and Williams [20] to determine the number of eigen-
values of a periodic elastic structural system is
adopted to check the total number of eigenfrequen-
ties existing within the current frequency increment.
If the rail is described by the Timoshenko beam
According to [ZO], the number of eigenfrequencies,
theory:
J(Q), of the track/beam system exceeded by a given
Q is determined by J(Q) = J,(Q) + s{&(R)}. Here
s{&(R)) is the number of negative diagonal elements W,(x) - O,(x) = 0 @a)
in the triangulated dynamic stiffness matrix f&(Q)].
J,(Q) is the number of eigenfrequencies exceeded by m,r2Qi-E.AGr + Pa 8,(x )
R if all the nodal displacements of the track/beam e;(x) +
r
system are thought of as being constrained to zero.
Thus, J,,(Q) is equal to (N + 1) times (where N is the
number of ties considered in the track model) the
total number of eigenfrequencies of a rail span ex-
ceeded by 8, plus N times the total number of where W,,(X), e,(x) are the modal deflection and
eigenfrequencies of the sub-structure, or the under- rotation angle of the rail of the n th mode, EI, is the
lying rail support unit consisting of a rail/tie contact flexural rigidity, KAG, is the shear rigidity where ti is
stiffness, a tie beam (or mass). and an array of the Timoshenko shear coefficient, m, is the mass per
distributed spring stiffness (track foundation). unit length of rail, r is the radius of gyration and P,
The nth solution, a,. of eqn (5) is now substituted is the axial force existing in the rail. The solutions of
into eqn (4), yielding the singular equation: the modal shape functions from the above equations
are presented in Appendix B.
tfGt(Q,,N&,i1 = 0 (6)
NATURAL MODES OF THE TIE
from which the eigenvector of the nth mode,
(AR,,)=IR’jn.@lnt W~nr~2,,i...,~~,,1B~~lf is ob- Having determined the nodal displacement of the
tained by using the Gaussian elimination method rail, lVie, atop the ith tie, the corresponding rail/tie
(with maximum pivoting) with the assignment of contact force (or rail seat force), ,fb,(t), may be
unity rotation angle to any arbitrary node (except the expressed as:
two end nodes and the center node). The solution
procedure for eqns (5) and (6) is implemented using f;,(t) = %,, W,,,e’*x’, (9)

where &, is the equivalent frequency-dependent


spring coefficient corresponding to the n th mode of
the track. This force is then applied on the sub-struc-
ture, that is a tie model as an elastic beam on a
Winkler foundation. The resulting tie mode shape
function is obtained as follows:

Root Pole
Fig. 1. Illustration of solution to the characteristic matrix
equation of track system.
Natural vibration analysis of rail track 1431

Table 1. Track parameters


Rail and other track parameters Tie parameters
Elastic modulus E_= 207 GN/m2 Elastic modulus E, = 70 GN/m*
Poisson’s ratio = d.28 Poisson’s ratio = 0.30
Timoshenko shear coeff. = 0.34 Timoshenko shear coeff. = 0.833
Cross-sectional area = 7.17 x IO-’ m2 Tie spacing = 0.7 m
Second moment of area = 23.5 x 10e6 m4 Tie length = 2.50 m
Radius of gyration = 57.2 x lo-’ m Tie width = 0.25 m
Bending rigidity EI, = 4.86 MNm2 Depth of mid section = 0.14 m
Shear rigidity RAG, = 197.2 MN Depth of shoulder section = 0.21 m
Unit mass m, = 56.3 kg/m Rail gauge length = 1.50m
Rail pad stiffness k,, = 280 MN/m Tie end to rail seat = 0.50 m
Track bed stiffness k. = 144 MN/m/m

where Z,(y), $,,(y) are the modal deflection and modes, which are of more importance for obtaining
rotation angle of the ith tie in the nth track mode, forced vibration solutions. When the number of ties
z,(y), /I,,,(y) are the deflection and rotation angle of reaches 20, the variations become insignificant. All
the m th mode of the tie as an individual free beam further calculations are then performed using 25 ties.
on an elastic foundation, M,,, is the generalized mass Cases 1 and 3 are used as illustrative examples
of the m th free tie mode and w, is the circular herein. Below a frequency limit of 1100 Hz, eqn (5)
frequency of the mth free tie mode. Equation (10) yields 93 natural frequencies for case 1, and 96 for
applies to both the Bernoulli-Euler beam theory and case 3. The frequency limit of 1100 Hz is chosen
the Timoshenko beam theory. based on a frequency domain analysis of the same
track [21] which showed the significant vibration fre-
NUMERICAL EXAMPLES OF TRACK NATURAL quency range to be below 1000 Hz (a 10% increase is
VIBRATION added for the analysis). A total of five modes
(rigid + four symmetric elastic modes) of the tie as an
As a numerical example to illustrate the use of the
individual beam on a Winkler foundation are taken
track vibration model, eqns (5) and (6) are solved for
into account to obtain the equivalent frequency-
the natural frequencies and nodal displacements, and
dependent spring coefficient expressed in eqn (1).
eqns (7), (8) and (10) for associated mode shapes, of
Typical rail and tie mode shapes and the correspond-
a typical concrete-tie rail track. The parameters used
ing natural frequencies of the track for some selected
for the track are given in Table 1. The proper track
modes are shown in Fig. 4(a) (case l), and Fig. 4(b)
length to obtain a representative solution is first
(case 3). For the rail, note the similarities between the
examined by using a different number of ties with a
two cases for the first three selected modes, and the
tie spacing of 0.7 m to formulate the overall dynamic
clear differences for the rest of the selected modes.
stiffness matrix of the track. The procedure is re-
The effect of the ties as discrete supports on the rail
peated using three different combinations of the rail
mode shape is clearly demonstrated by the last two
and tie beam theories. These include:
selected modes. This is especially clear for case 3
1. Bernoulli-Euler rail and Bernoulli-Euler tie; (Fig. 4b) where the Timoshenko beam theory is used
2. Timoshenko rail and Bernoulli-Euler tie; to describe both the rail and the tie beams. The tie
3. Timoshenko rail and Timoshenko tie. mode shapes for the two cases, however, are entirely
similar for all the modes shown. For the first two
The frequencies obtained for case 3 with a different
modes, the rigid mode of the tie as a mass dominates
number of ties are given in Table 2 for a variety of
the tie’s vertical motion. Examination of all the 93
selected vibration modes. It is seen that for all the
modes for case 1, and all the 96 modes for case 3
modes considered, the variation in the natural fre-
revealed that when the natural frequency of the track
quency diminishes as the number of ties, or the track
approaches and exceeds the tie’s first elastic mode, the
length, increases. This is especially true for the lower
bending flexibility of the tie beam prevails in govern-
ing the tie’s vibration modes. It is also seen that for
Table 2. Frequencies (Hz) obtained for different number of all the modes of the track shown, the tie does not
ties and mode numbers for the Timoshenko case vibrate beyond its third elastic mode shape as an
Mode no. individual beam. This is because the highest fre-
No. of ties I IO 30 50 70 quency considered herein, 10.55Hz for case 1, and
950 Hz for case 3, of the entire track is far from
5 161.3 250.8 767.3 reaching that of the fourth elastic mode of the
10 158.6 222.4 551.4 1100
15 158.4 186.5 442.0 865.1 1155 tie, which was calculated to be 1430 Hz for a
18 158.4 182.9 260.2 534.6 902.2 Bernoulli-Euler type tie beam (case l), and 1243 Hz
20 158.3 176.1 251.5 446.8 789.6 for a Timoshenko type tie beam (case 3).
25 158.3 176.0 250.0 443.4 783.2 The natural frequencies of the track with the
30 158.3 176.0 249.2 440.3 780.5
Timoshenko rail and tie beams (case 3) are generally
1432 Z. Cai el al.

Moda Mode Shape of Rail Mode Shape of 12th Tie


12th Ti Cantor Tie 1 Rail Soat Rail Saat
K-’ 0
-1
(158.3 Hz)

18.2 m 2.5 m
(a) Bernoulli-Euler Rail and Tie

Mode Mode Shape of Rail Mode Shape of 12th Tie


No: ’ Center Tie 1 Rail Seat Rail Seat
(1) a ’ 0
-1 -1
(158.3 HZ)
I

(30) 0
I rx,A,n &A,,, AIll
-1 0? ::
“(250.0 Hz) ”

(50) : _E - I I I 11 I I , I
-1 (443.4 HZ) v

(70) : A A& I* *_ Irn, _I “4b/\’ r\, A

-I (783.2 HZ)

(80) : n,
-1 vq _G-=v=d
(863.6 g)
I

" 0

-1
18.2 m 2.5 m
(b) Tlmoshenko Rail and Tie

Fig. 4. Example rail and tie mode shapes and track natural frequencies.

lower than with the Bernoulli-Euler rail and tie It should be pointed out that the mode shapes of
beams (case 1). The difference in the natural frequen- the track model thus obtained may not accurately
cies becomes more obvious as the mode number represent those of a much longer track. However,
(or frequency) increases. This clearly indicates the they are accurate solutions (within numerical toler-
increased effects of the rotatory inertia and shear ance) for the particular track length considered. The
distortion considered in the Timoshenko beam use of a finite extension track model to solve both rail
theory for higher frequency vibrations. Similar find- track vibration and static loading problems has also
ings have been reported for other types of struc- been adopted by others [13,15,24-261. The appli-
tures [22,23]. cation of the presented free vibration model for
Natural vibration analysis of rail track 1433

(a) MID-SPAN EXCITATION

1 0.--....--- BERNOULLI-EULER RAIL & DISCRETE TIE MASS


2 v---.- TlMOSHENKO RAIL & DISCRETE TIE MASS
3 O----- BERNOULLI-EULER RAIL & BERNOULLI-EULER TIE
- 4 a- TIMOSHENKO RAIL % BERNOULLI-EULER TIE
5 0----- TIMOSHENKO RAIL & TIMOSHENKO TIE
-- MODEL BY GRASSIE (6)
X FIELD TEST DATA BY GRASSIE (6)

10 100
FREQUENCY (Hz)

(b) ABOVE-TIE EXCITATION

. . . . . . . . . . BERNOULLI-EULER RAIL & DISCRETE TIE MASS


-.-.- TIMOSHENKO RAIL & DISCRETE TIE MASS
----- BERNOULLI-EULER RAIL & BERNOULLI-EULER TIE
- TIMOSHENKO RAIL & BERNOULLI-EULER TIE
.---- TIMOSHENKO RAIL & TIMOSHENKO TIE
- MODEL BY GRASSIE (6)
FIELD TEST DATA BY GRASSIE (6)

I
I 1 I ,

10 100 1000
FREQUENCY (Hz)

Fig. 5. Dynamic receptance of track-present model versus field data and existing model.

solving forced vibrations of the track under impulsive Table 1). The method of modal analysis and the
loading and dynamic wheel/rail interactions have technique of Fourier transform are used to formulate
been addressed elsewhere [17-191. the solutions. Details of the solution procedure are
published elsewhere [21]. The theoretical results, with
EXAMPLE APPLICATION OF MODEL
five different options of beam theories for the rail and
the tie, are compared with field experimental data and
To briefly illustrate the applicability of the track an existing discrete model, both due to Grassie [16],
model for simulating field rail track vibration prob- of the same track. This is shown in Fig. S(a) for
lems, the natural vibration results presented above mid-span excitation, and Fig. 5(b) for above-tie exci-
are used to obtain the dynamic receptance character- tation. For both excitation scenarios, the first peak
istics of the field track (with the parameters given in coincides closely with the natural frequency of the
1434 Z. Cai er al.

first mode of the track system. The more pronounced dominates the tie’s response for low modes of track
peak at approximately 750 Hz under the mid-span vibration. The bending flexibility of the tie prevails
excitation is due to the resonance of the rail span as when the vibrational frequency of the track exceeds
a beam pinned at both ends [16,27]. In Fig. 5(a) it is that of the tie’s first elastic mode.
seen that all five options give results close to the field The reasonably close comparison between the
measurement data at frequencies below approxi- theoretical dynamic receptance results obtained using
mately 250 Hz (no experimental data are available the track vibration model and the field experimental
below 70 Hz). Above this frequency, it is obvious that data of the same track indicates that the use of the
the theoretical results with the Timoshenko rail beam presented track model provides a reasonable rep-
option (5,4,2) are closer to the field data than those resentation of the field track under vibration. This is
with the Bernoulli-Euler rail beam option (1,3). The especially the case when the Timoshenko beam the-
differences between option 5, 4 and the field data are ory is used to describe the rail.
very insignificant at frequencies above 300Hz and
their accuracy is much more favorable than Grassie’s Acknowledgemenr-Acknowledged is funding to conduct
model. This indicates the importance of considering this study from a National Scientific and Engineering Re-
search Council (Canada) grant awarded to the authors.
the effects of the rotary inertia and the shear distor-
tion in the rail subjected to high frequency vibrations
REFERENCES
as well as the discrete nature of the tie supports.
These effects in the tie are not obvious as the differ- 1. J. T. Kenney Jr, Steady state vibrations of beam on
ence between option 5 and 4 is minimal. elastic foundation for moving load. J. appl. Mech.,
The difference between the results of the five ASME 76. 359-364 (1954).
2. P. M. Maihews, Vidration of a beam on elastic foun-
options is less when the excitation is above a tie dation (I). Z. angew. Math. Mech. 38, 105-I 15 (1958).
(Fig. 5b), than when the excitation is at the mid-span 3. P. M. Mathews. Vibration of a beam on elastic foun-
(Fig. 5a). This is because the response of the track is dation (II). Z. anger. Math. Mech. 39, 13-19 (1959).
dominated more by the rail span as a deep beam 4. A. D. Kerr, The continuously supported rail subjected
to an axial force and a moving load. Int. J. Mech. Sci.
spanning two adjacent ties when the excitation is
14, 71-78 (1972).
at the mid-span, and is dominated more by the 5. M. A. Dokainish and W. H. Elmaraghy, Steady-state
rail-pad-tie and track foundation as an integral vibrations of rail on an elastic damped foundation
system when the excitation is above a tie. The effects subjected to an axial force and moving load. ASME
Paper N75-RT-3, IEEE-ASME Joint Railroad Conf.,
of the rotatory inertia and the shear distortion of the
San Francisco (1975).
rail are obviously less significant in the latter case 6. J. J. Labra, An axially stressed railroad track on an
than in the former. elastic continuum subjected to a moving load. .4c/a
Mech. 22, 113-129 (1975).
7. R. A. Mair, Natural frequency of rail track and its
CONCLUSIONS relationship to rail corrugation. The Cir. Engng Trans.,
Institution of Engineers, Australia, 6-l I (1977).
A theoretical track/beam system model with differ- 8. B. Prasad and V. K. Garg. Dynamic models of a
ent theoretical options for describing the rail and tie railroad track system, Appl. Math. Modelling 3,359%366
beams has been presented. The formulation of the (1979).
9. S. P. Patil, Natural frequencies of a railroad track.
generalized track element and the equivalent fre-
J. appl. Mech.. ASME 54, 299-304 (1987).
quency-dependent spring coefficient representing the 10. S. P. Patil, Response of infinite railroad track to vibrat-
resilient, flexural and inertial effects of the underlying ing mass. J. Engng Mech.. ASCE 114, 688.-703 (1988).
rail support components have largely simplified the 11. D. G. Duffy. The response of an infinite railroad track
natural vibration analysis and yet preserved the real to a moving, vibrating mass. J. uppl. Mech., ASME 57,
66-73 (1990).
structural features of the track/beam system. The
12. K. Ono and M. Yamada. Analysis of railway track
complex eigenvalue problem of the interconnected vibration. J. Sound Vibr. 130, 269-297 (1989).
frame-like beam system is replaced with that of a 13. R. A. Clark, P. A. Dean, J. A. Elkins and S. G. Newton,
system of uniform beam elements on discrete sup- An investigation into the dynamic effects of railway
vehicles running on corrugated rails. J. Mech. Engng
ports. Through a numerical example, it has been
Sci. 24, 65-76 (1982).
demonstrated that the shear distortion (SD) and 14. S. L. Grassie and S. J. Cox, The dynamic response of
rotatory inertia (RI) of the rail beam are two import- railway track with flexible sleepers to high frequency
ant factors affecting the natural vibration character- vertical excitation. Proc. Inst. Mech. Engrs 7(198(D)),
istics of the track. Consideration of these two factors 117-124 (1984).
15. S. G. Newton and R. A. Clark, An investigation into the
reduces the natural frequency of the track and signifi- dynamic effects on the track of wheelflats on railway
cantly alters the mode shapes of the rail described by vehicles. J. Mech. Engng Sci. 21, 287-297 (1979).
the Bernoulli-Euler beam theory, especially for 16. S. L. Grassie. Dynamic modeling of rail track and
higher mode vibrations. The discrete support of the wheelsets. Proc. Ad Inr. Conf. Recent Advances in
Structural Dvnamics, University of Southampton.
track provided by the rail/tie contact stiffnesses, the
pp. 681498 (1984).
ties, and the track foundation system is evidently 17. Z. Cai and G. P. Raymond, Response of railway track
important for high frequency vibrations of the to dynamic and static loading. Proc. 5th Int. HeaLy Haul
track/beam system. The rigid movement of the tie Railway ConJ. June. Beijing. China, pp. 362-368 (1993).
Natural vibration analysis of rail track 1435

18. 2. Cai and G. P. Raymond, Theoretical model for k*, = Eu,u*[(a,u* - L,u,)(l -CC) +(A,u, + I,u,)sSl
dynamic wheel/rail and track interaction. Proc. fowl Inr.
Wheelset Congress, Sept., Sydney, Australia, pp. k, = - B(i,u, + d,u,)(u,cS - u*sC)
127-131 (1992).
19. G. P. Raymond and Z. Cai, Dynamic track support k,, = - Bu, u,(l,u, + a,u,)(u,s + ups)
loading from heav~erifaster tram -sets. Trans. Bes. -Bee.,
No. 1381. National Research Council, Washin~on, k,,=Btc,u,(a,u,Ca~ff2)(C-Cf
D.C. pp. 53-59 (1993).
20. W. H. Wittrick and F. W. Williams, A general algor- 4, = k,,
ithm for computing natural frequencies of elastic struc-
tures. Quart J. Mech. appl. Marh. 24(3), 263-284 (1971). h, = -6,
21. 2. Cai and G. P. Raymond, Dynamic modelling of
parameters controlling railway track vibration., Proc. k4s= B&u, +A&(u,S -u+)
Ist ht. Cortf ~orion and Vibration Control, Sept.,
Yokohama, Japan, pp. 976-981 (1992). k43 = - kz
22. F. Y. Chene. Vibration of Timoshenko beams and
frame works-j. Strucr. Dir., ASCE %, 551-571 (1970). k,=k,,, (A4)
23. T. M. Wang and T. A. Kinsman, Vibration of frame
structures according to the Timoshenko beam theory. where
J. Sound Vibr. 14, 215-227 (1971).
24. J. C. 0. Nielsen and T. J. S. Abrahamsson, Coupling of s = sin& Ir), c = cos(l, i,), S = sinh(~~~~~,
physical and modal components for analysis of moving
non-linear dynamic systems on general beam structures.
Inc. J. Numer. Meth. Engng 33, 1843-1859 (1992).
C = cosh(l,l,) (A5)
25. H. A. El-Ghazaly, A. N. Sherbourne and F. Arbarbi,
EI
Strength and stability of railway tracks--II, determinis- B= (‘46)
tic, finite element stability analysis. Comput. Struct. 39, (uf - u$sS + 2u,u,(cC - I) I
2345.
26. G. P. Raymond, Analysis of railway track structures- A, and I, are given by eqn (B7), and u, and u, by eqn (BlO),
ARTS. Camput. Snwt. 41, 1403-1409. Appendix B, with subscript n excluded.
27. 2. Cai, Modelling of rail track dynamics and wheel/rail When a>@,
interaction. Ph.D. thesis, Department of Civil Engin-
eering, Queen’s University, Kingston. Ontario, Canada k,,=Bu,u,(l,u,-I,u,)(-u,sC+u2cS)
(19927. -
k,,= -Bu,u,[(A,u,+i,~~)(l -CC)-(l,u2+dzu,)sS]
APPENDIX A
k,, = B(1, u, - A&(u, cS - u,sC)
For the Bernoulli--Euler type beam:
k,, = Bu,u,(l,u, -12u2)(u,s -u,S)
k,, = B&&(1: + i;)(&sC + I,cS)
k,, = - Bu, u,(l, u, - izuz)(c - C)
k,, = B~,~~[(A~-~~)(l -cC)+21,l,sS]
k,,=k,,
kz2 = Bfl: + L~)(i,sC - I,cS)
k.,, = k,,
k,, = - BA,i.,(l,: + A;)(,+ + I$)
k,, = B&u, - A,u,)(u,s -u, S)
klz = Bl,A,(d; + d;)Cc

k,, = k,, k,, = k,,

k*=k,,. (A7)
k,, = -k,,

k,, = B(1: + A$)(?., S - &s) where

k,, = - kz, s = sin@, I,), c = cos(l, Ir), S = sin&l,),

ka = kz2. (Al) c = cos(l, /,) (‘48)


EI
where B= (A9)
(ui + u$)sS + 2u, u&C -3 ’
c = cos(A,/,), s = sin(l, $,), C = cosh(d&,
I, and 1, are given by eqn (Bl2), and u, and ur by eqn (Bl4),
S = sinh(& Ir,) (A2) Appendix B, with subscript n excluded.

EI
5= APPENDIX B
643)
21,&(1 -CC)-(l:-I:)sS
Bernoulli-Euler beam theory
A2and i;, are given by eqn (B2), Appendix B, with subscript
The mode shape function of the rail, solution of eqn (7).
n excluded.
is:
For the Timoshenko type beam:
_
when Q 6 Q, where S& is given by eqn (B5), Appendix B, W,(x) = A,cos A,,x, + B,, sm I,,x,

k,, = Bu,u,(l,u, + Izu,)(u,sC + u,cS) + C,” cash &,j, + D, sinh 1,,1,, (BI)
1436 2. Cai el al.

where .$ = x - x, is the local coordinate within the ith span, in which


see Fig. 2, and A,, and &, are:

where a = P,/EI,, and b, = (m,R~)/EI,. in which P, is the


axial force in the rail, EI, is the flexural rigidity of the rail
and 111,is the unit mass of the rail.
The nodal displacements ( W,, . fl,,, IV,+,,“, 0,+ ,,“) are sub-
stituted into eqn (Bl) to solve for the integration coefficients The integration coefficients, A,, to D,,, and A,, to D,,, are
A,, to &,: related by the following expressions:

I,,= -u,,,A,,, B,n=r+nB,,, c,n=~rnC,“.

D,, = ~2n Din > (B9)

where

where the matrix [L,,] is identical for each span and is %=%,“(I
+). u2.=lb(l+~) @IQ
expressed below:
and can be determined in the same way as for the
Bernoulli-Euler case. The form of the results is identical to
eqns (B3) and (B4). The only changes required are to
substitute the values of A,, and I,, from eqn (B7) instead of
eqn (B2).
For R, > R,, the rail shape functions of the nth mode are:

where W’,(x) = A,,cos I,,f, + B,,sin I,,,.?, + C,,cosA,.\-,


c=cosI,,/,,s=sini,,I ,C=coshj.,,l,,S=sinhLz,l,, in
which 1, is the length o Pthe rail span. + D,, sin L,,.f, (BI la)

Timoshenko beam theory H,(x) =A;,sini,,.i-,+B,,cosL,,.~,+ C,,sinI,,I,


The solution of eqn (8) for the mode shape function of +~,,,ccos~,,,.?:,. (Bllb)
the rail is expressed in different forms for the Timoshenko
beam theory, depending on the magnitude of the natural where
frequency with respect to the reference frequency given
by: (B12)

Qc=(y)‘2, (B5) in which LX,,/I, are determined from eqn (B8).


The relationships between A,, to D,, and x,, to 8,,, are:

where h-AC, is the rigidity of shear distortion of the rail, and A;,,= -qnA,,. &=u,,B,,. c,;,= -u:,C,,.
r, is the radius of gyration of the rail cross-section.
For R, < Q, the n th mode shape of the rail can be written D,n = uznD,,, . (Bl3)
as:
where u,” and Us-,are
W,(X) = A,, cos A,,.?, + B,, sin A,,.$ + C,, cash 1r. .f,

+ D,nsinh izn.fc (B6a)

6,(.x) = A;, sin I.,,.?, + & cos &,.I=,+ r,, sinh & The matrix [L,] is:

+ B,,jcash L,.i-,, (B6b)

where 0,(.x) is the angular mode shape of the rail and A,,. AZ.
are determined from:

(B7) where c = cos n,,/,, s = sin A,,/,. c = cos L,, lP ,


S=sin1,1,.

Anda mungkin juga menyukai