This may be the author’s version of a work that was submitted/accepted

for publication in the following source:

Mandal, Nirmal, Dhanasekar, Manicka, & Sun, Yan Quan

(2016)
Impact forces at dipped rail joints.
Proceedings of the Institution of Mechanical Engineers, Part F: Journal of
Rail and Rapid Transit, 230(1), pp. 271-282.

This file was downloaded from: https://eprints.qut.edu.au/75124/

c Consult author(s) regarding copyright matters

This work is covered by copyright. Unless the document is being made available under a
Creative Commons Licence, you must assume that re-use is limited to personal use and
that permission from the copyright owner must be obtained for all other uses. If the docu-
ment is available under a Creative Commons License (or other specified license) then refer
to the Licence for details of permitted re-use. It is a condition of access that users recog-
nise and abide by the legal requirements associated with these rights. If you believe that
this work infringes copyright please provide details by email to [email protected]

Notice: Please note that this document may not be the Version of Record
(i.e. published version) of the work. Author manuscript versions (as Sub-
mitted for peer review or as Accepted for publication after peer review) can
be identified by an absence of publisher branding and/or typeset appear-
ance. If there is any doubt, please refer to the published source.

https://doi.org/10.1177/0954409714537816
 Journal of Rail and Rapid Transit

Fo
Impact forces at dipped rail joints
r

Journal: Part F: Journal of Rail and Rapid Transit

Pe

Manuscript ID: JRRT-13-0200.R1

Manuscript Type: Article

Date Submitted by the Author: n/a

er

Complete List of Authors: Mandal, Nirmal; CQUniversity, School of Engineering and Built Environment
Dhanasekar, M; Queensland University of Technology
Sun, Yan; Central Queensland University,
Re

railway track, Railway track design, Rail wheel interaction, Signalling, static
Keywords:
wheel load, structure parameters, Wheel/Rail Interaction

Impact forces develop at the wheel/rail interface due to the presence of

vi

defects in the running surface of the wheel and/or the railhead. This paper
reports on wheel impacts caused due to permanently dipped rail joints
characterised by high frequency impact forces of high amplifications of
ew

static load that occur for a very short duration (P1 forces), followed by
relatively low frequency, lower amplitude forces (P2 forces) that occur for a
longer duration. These impact forces are affected by the design of
components adjacent to the wheel and the rail, namely the bogie primary
Abstract:
suspension and rail seat pads; the influences of stiffness and damping
characteristics of these components are investigated. A modified 3D
wagon-track system dynamics simulation model that accounts for defects
in track is used to obtain the impact force time series which is converted to
impact force factors and compared with a set of field measured data
reported in the literature. A simplified equation for the determination of
impact force factors due to dipped rail joints is also proposed and
validated.

http://mc.manuscriptcentral.com/JRRT
Page 1 of 32 Journal of Rail and Rapid Transit

1
1
2
3
4
5
Impact forces at dipped rail joints
6
7 Nirmal K. Mandal1#, M. Dhanasekar2 and Yan Quan Sun1
8
9 1
Central Queensland University, Centre for Railway Engineering, Rockhampton 4702, Australia
10 2
11 School of Civil Engineering & Built Environment, Queensland University of Technology, Brisbane,
12 Australia
13
14
Submitted to: Journal of Rail and Rapid Transit, January 2014
15
16 Abstract
17
18
Impact forces develop at the wheel/rail interface due to the presence of defects in the running surface of
Fo
19
20
21 the wheel and/or the railhead. This paper reports on wheel impacts caused due to permanently dipped rail
22
r
23 joints characterised by high frequency impact forces of high amplifications of static load that occur for a
24
Pe

25 very short duration (P1 forces), followed by relatively low frequency, lower amplitude forces (P2 forces)
26
27 that occur for a longer duration. These impact forces are affected by the design of components adjacent to
28
er

29 the wheel and the rail, namely the bogie primary suspension and rail seat pads; the influences of stiffness
30
31
and damping characteristics of these components are investigated. A modified 3D wagon-track system
32
Re

33
34 dynamics simulation model that accounts for defects in track is used to obtain the impact force time series
35
36 which is converted to impact force factors and compared with a set of field measured data reported in the
vi

37
38 literature. A simplified equation for the determination of impact force factors due to dipped rail joints is
39
ew

40 also proposed and validated.

41
42 Keywords: wheel/rail contact; impact force; wagon speed; dipped rail joint; static wheel load.
43
44
45
46
47 # Corresponding author: (Dr Nirmal K Mandal)
48
49 E-mail address: [email protected]; Tel: +617 4923 2064; Fax: + 617 4930 6984
50
51
52
53
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 2 of 32

2
1
2
3
4 Nomenclature
5
6 C damping matrix
7 CH Hertz contact coefficient
8
9 Dwr spring displacement
10 Drn rail displacement at contact normal direction
11
12 Dwn wheel displacement at contact normal direction
13 d displacement vector
14
15 F force vector
16 Fn force at contact normal direction
17
18 h maximum depth of the dipped rail joint
Fo
19 me, mt, kt, ct equivalent track system parameters defined in Jenkins et al. (1974)
20
21 K stiffness matrix
22 KH linearised wheel/rail Hertzian contact stiffness
r
23
24 Kr rail stiffness matrix
Pe

25 Kt track stiffness matrix

26
Kw wagon stiffness matrix
27
28 M mass matrix
er

29
mu unsprung mass of bogie
30
31 P0 static wheel load of wagon
32
P1 high frequency short peak force
Re

33
34 P2 low frequency delayed peak force
35
Rw wheel radius
36
vi

37 v0 vertical input excitation

38 V0 operating speed
39
ew

40 wr, ww, wt vertical displacements (static) of rail, wheel and track respectively
41 X dimensionless parameter
42
43 2α angle in radians of dipped rail joint
44
λd wavelength of the dipped rail joint
45
46 µ(x) function representing the profile of the dipped rail joint
47
48
49
50
51
52
53
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 3 of 32 Journal of Rail and Rapid Transit

3
1
2
3
1. Introduction operational and track parameters as a
4
5
6 dimensionless number (X). Dynamics of
7
8 The term ‘dipped rail joint’ is used in this freight wagons crossing dipped rail joints of
9
10 paper to define a permanent geometric
11
varying size at different speeds have been
12
13 deviation or permanent dip from the original simulated for the purpose of the analysis. A
14
15 running top rail surface. As such, dipped rail single bogie model obtained by modifying
16
17
joints (Fig.1) characterised by the the detailed 3D model developed by Sun and
18
Fo
19
20 wavelength ( λd ) and defect height (h) are a Dhanasekar [1] has been used for the
21
22 examination of the wheel impacts due to
source of significant wheel impact. The
r
23
24
dipped rail joints.
Pe

25 objectives of this paper are to quantify wheel

26
27
impact forces at dipped rail joints using a
28
er

29
half-wagon model to modify a 3D wagon- 1.2 Impact forces
30
31
32 track dynamics simulation model and For trains travelling at high speed when
Re

33
34 approaching rail joints containing a dip (Fig.
35
propose new simplified equations for the
36
1), wheels momentarily lose contact with the
vi

37 impact forces.
38
39 railhead in the dipped zone of rail 1 and land
ew

40
41 on the connected rail 2, generating an impact
42 1.1 Scope
43
44 The P1 and P2 forces are modelled and a high frequency vibration of the joint
45
46 considering various design and operational (Fig. 2). Because of dynamic action, wheels
47
48 jump and land on the railhead, causing P1
49 parameters, resulting in complex equations.
50
51 This paper presents new simple models to and P2 forces.
52
53 Two types of contact impact forces evolve
predict P1 and P2 impact forces due to
54
55
56 dipped rail joints, considering many because of wheel/rail interactions at dipped
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 4 of 32

4
1
2
3
rail joints. They are P1 and P2 forces. The presented in Fig. 3 (a) based on the
4
5
6 time history of wheel impact force exhibits a geometric mean of the force time history [2].
7
8 characteristically high frequency (≅200Hz – Another method of determining P2 is to
9
10
11 1000Hz) force immediately after the impact locate the highest peak just after the high
12
13 (known as the short peak or P1 force) frequency shocks are damped out [3] as in
14
15 Fig. 3 (b). Values of P2 as reported in this
16
followed by a low frequency (≅50Hz –
17
18 200Hz) peak of reduced magnitude (known paper have been determined using the latter
Fo
19
20 as the delayed peak or P2 force) as shown in approach.
21
22 Appropriate processes for the design of new
Fig. 2. The P1 force remains only for a very
r
23
24
tracks and the examination of existing tracks
Pe

25 short duration whilst the P2 force acts for a

26
27 relatively longer duration. As such, the P1 must account for the dynamics due to
28
er

29
force potentially damages the local contact wheel/rail contact. It is generally accepted
30
31
32 region of the wheel tread and railhead whilst that the wheel/rail contact forces generated
Re

33
34 by wagons containing good suspension
35
the P2 force has the potential to damage
36
most wagon and track components located systems and wheel profiles running on
vi

37
38
39 in the vicinity of the contact zone. Both defect-free track remain within ± 10% of the
ew

40
41 static load [4]. The size of the permanent
42 types of peak force are therefore of
43
44 significance. dip affects the frequency and magnitude of
45
46 It is easy to locate high frequency P1 forces the dynamic forces; defects with larger
47
48
49 from the force time responses. There are two depth (h) and smaller wavelength ( λd )
50
51 concepts to calculate P2 forces. Fig. 3 generally induce larger impact forces. It is
52
53
illustrates their calculation methods. The therefore necessary to understand the levels
54
55
56 first method of calculating P2 force is of wheel impact forces that are found to
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 5 of 32 Journal of Rail and Rapid Transit

5
1
2
3
occur in actual train operations to enable the the other hand, the dynamic impact factor
4
5
6 provision of adequate safety measures or the (DIF) described in this paper defines the
7
8 initiation of interventions to protect the track level of impact developed due to localised,
9
10
11
structure. small sized geometric deviations. The DLF
12
13 and DIF should therefore be regarded as
14
15 1.3 Literature on P1 and P2 forces serving different purposes; whilst the DLF
16
17
18 In continuous rails with defect-free surfaces, accounts for the entire track structure, the
Fo
19
20 various studies [5-7] identified that the DIF accounts for a limited section of the
21
22 dynamic load factor (DLF) was affected by a track structure close to the geometrically
r
23
24
large number of parameters including the dipped location.
Pe

25
26
27 operating speed, the wheel diameter and the Shen [2] modelled the wheel/rail forces for
28
er

29
track condition parameter. Rail tracks are rail freight vehicles using the software
30
31
32 generally designed using an equivalent package TUTSIM to analyse joint related
Re

33
34 dynamic load calculated by multiplying the track excitation and estimated the P2 force as
35
36
static wheel load by the DLF. There are the geometric mean force of the force time
vi

37
38
39 many formulae that define the DLF as a history.
ew

40
41 function of the track condition, operating Steenbergen and Esveld [11] and
42
43
44 speed and wheel load [8-10]. These Steenbergen [12] elaborated on the
45
46 formulae have typically been developed to generation of dynamic wheel/rail contact P1
47
48 account for the large geometric deviations and P2 forces due to rail weld irregularities.
49
50
51 that the track would develop over its lifetime These irregularities were found to cause
52
53 from the original design position; typically localised dynamic forces which penetrated
54
55
56
the DLF varies from 1.05 to 2.45 [4]. On
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 6 of 32

6
1
2
3
into the railhead rather than loading the rail system and the track was modelled as a layer
4
5
6 as bending moments. system consisting of two Timoshenko beams
7
8 A discrete support model was used by (rails), 88 sleepers, 176 sleeper pads and
9
10
11
Dukkipati and Dong [13] to investigate the ballast blocks resting on subgrade formation.
12
13 characteristics of the dynamic forces due to Short wavelength (~0.1m) defects of 0.5mm
14
15 a dipped rail joint. They used the available dip were reported to generate DIFs of
16
17
18 experimental data of the dynamic forces at approximately 5.0 for a passenger car speed
Fo
19
20 dipped joints for validation. The rail gap at of 300km/h. Lower operating speeds of the
21
22 the joint was considered as a rail harmonic higher axle load wagons examined in the
r
23
24
irregularity function, and a constant value of paper were reported to generate much lower
Pe

25
26
27 dip angle of 0.01radian was used for dynamic impact forces.
28
er

29
simulation. The effects of different Sun and Dhanasekar [1] developed a 3D
30
31
32 wavelengths of the dipped joint, axle loads, wagon-track system dynamics model to
Re

33
34 stiffness of ballast and pad on the dynamic examine different aspects of short
35
36
force were investigated. Other than P1 and wavelength defects of rail and considered
vi

37
38
39 P2 forces, the concept of the P12 force was wheel-flats, dipped joints, engine burns and
ew

40
41 proposed in this research. railhead corrugation using a multiple contact
42
43
44 point model for the simulation of wheel/rail
45
46 1.4 Modelling strategies contact. A detailed track subsystem was an
47
48 Wen et al. [14] reported a detailed time important feature of this model. In the case
49
50
51 domain dynamic model similar to that of dipped joints, the authors used the impact
52
53 reported by Sun and Dhanasekar [1]; the velocity as an input excitation to the rail
54
55
56
passenger car was modelled as a 35 DOF joint similar to the concept presented by
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 7 of 32 Journal of Rail and Rapid Transit

7
1
2
3
Zhai and Sun [15]. A simpler method of 1.5 Field measurements
4
5
6 describing the rail joint excitation had been Dhanasekar and Bayissa [18] presented a
7
8 presented by Lu and Gill [16] in which the comparative study of the structural
9
10
11
defect was represented by sinusoidal profiles responses of two types of IRJ: square cut
12
13 that describe the instantaneous bending of and inclined cut. Vertical and shear strains
14
15 the joint in terms of its wavelength and were considered in the time, frequency and
16
17
18 maximum depth of the defect. wavelet domains. For inclined cut IRJs, the
Fo
19
20 Relating to the Rayleigh wave speed of sub- vertical impact strain was lower and shear
21
22 grade of the railway track, Yang, et al. [17] strain was higher respectively when
r
23
24
carried out a dynamic stress analysis of a compared to those of the square cut IRJs.
Pe

25
26
27 ballasted railway track bed using a two- Mutton and Alvarez [19] examined the
28
er

29
dimensional dynamic finite element performances of dipped and peaked
30
31
32 analysis. The effects of geometric variation aluminothermic welds in a very heavily
Re

33
34 in rail head level, train speed, loaded iron ore railway line primarily
35
36
acceleration/braking etc. were considered. It through field measurements. Impact force
vi

37
38
39 was observed that the dynamic effects of factors were presented in response to
ew

40
41 track were very sensitive to those wagons running at different speeds over
42
43
44 parameters, and the stresses due to dynamic peaked and dipped weld joints.
45
46 effects increased dramatically when the train
47
48 speed was greater than the Rayleigh It is observed that many research studies
49
50
51 velocity. have been conducted to estimate P1 and P2
52
53 forces considering empirical formulae,
54
55
56
simulations, field testing etc. Measurement
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 8 of 32

8
1
2
3
4
techniques are sometimes complex. An P1 = P0 + 2α V0 K H me / (1 + me / mu )
5 mu 1/ 2
6 obvious question arises – is there any simple P2 = P0 + 2α V0 ( ) {1 −
7 mu + mt
8 method to estimate P1 and P2 forces ctπ
9 } kt mu
10 kt (mu + mt )
11
considering both track design and
12
13 operational parameters. This paper addresses (1)
14
15 this question. The simulation results of this where P0 is static wheel load, mu is
16
17
18 paper for P1 and P2 forces correlate well unsprung mass, KH is linearised Hertzian
Fo
19
20 with previous predictions [1, 6, 9, 19, 20], wheel/rail contact stiffness, V0 is the vehicle
21
22 giving a lower bound of prediction of P1 and speed, α is half the total dip angle, and me,
r
23
24
P2 impact forces at dipped rail joints. This mt, kt and ct are equivalent track system
Pe

25
26
27 can help improve the design process of track parameters for half the track as defined in
28
er

29
structures. [9]. In the literature, the total dip angle is
30
31
32 represented by 2α as in [9] which is the
Re

33
34 2. Theory convention used in this paper, and by α in
35
36
other publications such as [4].
vi

37 2.1 Empirical relations

38
39 In pioneering research on rail joints, Jenkins The British Rail Joint Model formula [4] for
ew

40
41 a dipped rail joint is given in Eq. (2).
42 et al. [9] provided empirical relations for the
43
44 evaluation of the P1 and P2 impact forces
45
46 imposed on a single rail as follows: K t mu 0.5
47 P1 = P0 + 2α V0 ( ) (2)
48
g
49
50 where Kt is track stiffness at the joint (kN/m)
51
52
and g is the gravitational constant (m/s2).
53
54
55 Jeong [8] modified the British Rail model and
56
57 provided a formula for P2 as shown in Eq. (3).
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 9 of 32 Journal of Rail and Rapid Transit

9
1
2
3
representing the profile of the short
4
5
wavelength defects (dipped rail).
6 muV0 1 + (ω 2τ ) − 1 + (ω 2τ )
2 2 2
7 P2 = P0 + (3)
8 Rw 1 + (ω 2τ ) 2 For modelling multiple contact phenomena,
9
10
multiple springs at the interface are used.
11 Kr 2αRw
12 where ω 2 = , τ = , Rw is the
13
mu V0 The displacement of the springs is
14
15 wheel radius and Kr is the rail stiffness. determined similarly to the method of Sun
16
17
18 and Dhanasekar [1]:
Fo
19
20 2.2 Theoretical aspects of simulations
21 ∆Dwr = Drn − Dwn − µ (x) (5)
22 Away from empirical relationships of P1 and
r
23
24
P2 forces, dynamic analyses are carried out By multiplying the summation of the
Pe

25
26
27 using the lumped parameter model, the beam displacements of all springs with the spring
28
er

29 constant (KH), the vertical contact force is

on distributed support model and the beam
30
31
32 on discrete support model [13] to estimate determined as shown in Eq. (6):
Re

33
34 P1 and P2 forces. These models typically use
35 Fn = K H ∑ ∆Dwr (6)
36
Hertzian contact theory and couple the
vi

37
38
39 wagon and track models, and the normal The dipped joint encountered in the passage
ew

40
41 contact force (Fn) is determined [1, 21] as
42 of the wheel is modelled as a discontinuous
43
44 shown in Eq. (4): function shown in Fig. 4. For a similar dip,
45
46 the function shown in Eq. (7) was adopted
47
48
49
Fn = C H {wr − ww − µ ( x )}3 / 2 (4) by Lu and Gill [16]; this paper also adopts
50
51 this formulation:
52 where CH = Hertz contact coefficient, wr, ww
53
54 = vertical displacements (static) of rail and
55
56 wheel respectively and µ(x) = function
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 10 of 32

10
1
2
3
µ (x ) = through the secondary suspension. One
4
5  x − x0
6 h[1 − cos (2π λd
)] x0 ≤ x < ( x0 + λd / 4) bolster (two wheelsets) with primary
7 
8  x − x0 suspension is considered, and a sinusoidal
9 h [1 + cos (2π λd
)] x0 + λd / 4 ≤ x
10 
11  ≤ ( x0 + λd / 2) dipped rail joint profile (similar to that
12 
13 0 x > ( x0 + λd / 2) adopted by Lu and Gill [16]) is incorporated
14
15 (7) on both rails at the same location along the
16
17
18 in which h represents the maximum depth of track. The number of wagon subsystem
Fo
19
20 the dipped joint profile and 2 α is the total degrees of freedom (DOF) considered is 19
21
22 dip angle. which includes a half-wagon body with
r
23
24
three linear displacements in the x, y and z
Pe

25
26
27 3. Numerical simulations, results directions, one bogie frame with one angular
28
er

29
rotation and one parallelogram rotation
30 and validations
31
32 about the z axis, two side frames each with
In the numerical calculations, a half-wagon
Re

33
34 two linear displacements in the y and z
35 (more precisely a single bogie) model is
36
directions and two angular rotations about
vi

37 adopted as schematically shown in Fig. 5.

38
39 the x and y directions, and two wheelsets
ew

40 This vehicle-track interaction model is

41 each with two linear displacements in the y
42 established using coupling of the wagon
43
44 and z directions and one angular rotation
subsystem and the track subsystem through
45
46 about the z direction.
47 the wheel/rail interface subsystem as
48 Each DOF represents a particular linear
49 described by Sun and Dhanasekar in [1].
50
51 displacement and angular rotation about the
52
The bogie frame mass is connected to the
53 Cartesian coordinate system (x, y, z) fixed to
54 wheelset unsprung mass through the primary
55
56
the centre of each principal mass. The total
suspension and to the wagon body mass
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 11 of 32 Journal of Rail and Rapid Transit

11
1
2
3
number of mode shapes of rail (Timoshenko over double dots represent first and second
4
5
6 beam) is 110 and the number of sleepers and time derivatives respectively.
7
8 positions of pairs of ballast blocks (under the The modified model has been shown to be
9
10
11
rail seats of each sleeper) are both 120 relatively well correlated with the results
12
13 respectively. The individual DOFs are five available in the literature [1, 20] from
14
15 for each rail, three for each sleeper and one authors who modelled the dipped joint as an
16
17
18 for each ballast block. Consequently, the input vertical excitation ( v0 ) related to
Fo
19
20 total number of degrees of freedom used in longitudinal operating speed V0 as shown in
21
22 the analysis are 1719 (total DOFs = 19 + Eq. (9):
r
23
24
110×2×5 + 120×3 + 120×2×1).
Pe

25 v 0 = 2α V0 (9)
26
27 The following equations of motion of the
28 In this paper, however, the sinusoidal
er

29
wagon and track subsystems are solved by
30
31 representation of the dip (Eq. 7) is used [16].
32 the Newmark - β (implicit) method in the
Re

33 All other track and operational parameters

34 time domain to effectively account for the
35
are kept the same as those used by Sun and
36
nonlinearities such as the nonlinear Hertz
vi

37
38 Dhanasekar [1] and Zhai [20]. The speed is
39 contact and creepages:
ew

40 kept as 52km/h to allow direct comparison

41
42 M w d&&w + Cw d&w + K w d w = Fw
43
(8) with results available in the literature [1, 20].
M d&& + C d& + K d = F
t t t t t t t
44
45 Vertical dynamic forces due to the dipped
46 in which subscripts w and t stand for wagon
47 joint evaluated by the present study and that
48
and track respectively; M, C, and K
49
50 of Sun and Dhanasekar [1] and Zhai [20] are
51 represent mass, damping and stiffness
52 shown in Fig. 6.
53 matrices, and d and F are displacement and
54 A small time lag of the occurrence of P1
55
56
force vectors respectively; the over dots and
57 peak force is provided intentionally to
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 12 of 32

12
1
2
3
facilitate clear graphical visualisation of the the current model (Fig. 7). Field practice is
4
5
6 peak for each case. It is seen that the vertical that the depths of dipped joints are typically
7
8 dynamic force predicted by the current measured using a 1m long metal
9
10
11
model compares well with other results in straightedge (ruler) placed on top of the
12
13 spite of the differences in the methods of railhead, which translates into a quarter
14
15 representing the defect. It is therefore wavelength of 500mm. The field measured
16
17
18 considered that the model is reliable for the P1 forces corresponding to various train
Fo
19
20 prediction of the impact force time history speeds from 40km/h to 80km/h are plotted in
21
22 due to dipped rail joints. It is observed from Fig. 7 with the results obtained from four
r
23
24
this time history (Fig. 6) that Zhai’s simulations carried out using the current
Pe

25
26
27 numerical simulation predicts a P1 force of model. These four simulations correspond
28
er

29
about 215kN, the present study about to the Set 1 defect size (500mm quarter
30
31
32 201kN, Zhai’s experiment about 200kN and wavelength & 0.4mm depth) shown in Table
Re

33
34 Sun and Dhanasekar [1] about 200kN. The 1 used for four operating speeds (25km/h,
35
36
current study yields an error of about 6.4% 50km/h, 75km/h and 100km/h). In spite of
vi

37
38
39 relative to the result of Zhai’s numerical the differences in the source of these data
ew

40
41 simulation, whilst the errors compared with sets, the exceptionally good correlation
42
43
44 the values of other cases are negligibly small between the P1 forces determined is
45
46 (≈0.5%). remarkable. Fig. 7 also presents data from
47
48 P1 forces measured via a wayside other classical formulae and a new formula
49
50
51 monitoring system at a dipped welded joint proposed in this paper. These results are
52
53 as reported by Mutton & Alvarez [19] are discussed later in Section 6 of this paper.
54
55
56
used to demonstrate a good correlation with Based on the results presented in that
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 13 of 32 Journal of Rail and Rapid Transit

13
1
2
3
section, one could have confidence in the prominent at low speeds (see 50km/h results
4
5
6 results predicted by the current model. in Fig. 8 for example). At higher speeds,
7
8 this phenomenon increases leading to the
9
10 4. Impact force factors
11
value of P/P0 to vanish, a situation known as
12
13 “jump” which subsequently leads to impacts
14 Table 1 shows three sets of defect size; for
15 when the wheel contact re-occurs. This jump
16 this study, four operating speeds (25km/h,
17
18 appears to occur just after the occurrence of
50km/h, 75km/h and 100km/h) coupled with
Fo
19
20 the first high frequency impact force. Time
21 each defect size are considered.
22 of jump (loss of contact) remains very small.
r
23 This results in a total of twelve cases. Time
24
The fluctuation of dynamic effects reduces
Pe

25
26 histories for the vertical impacts considering
27 gradually once the jump has occurred. The
28 the Set 2 quarter wavelength defect with
er

29
damping of this fluctuation is higher for
30
100km/h and 50km/h velocity are presented
31
32 lower speeds. Fig. 8 shows the sensitivity of
Re

33 in Fig. 8.
34 this dynamic force to the speed for a
35 Characteristics of wheel/rail contact at the
36
particular defect size.
vi

37
38
starting point and at the apex of the dipped
39 Fig. 9 presents the variation of P1 and P2
ew

40 joint provide interesting observations. It is

41 forces due to speeds and defect sizes. It is
42 seen from Fig. 8 that the vertical dynamic
43
44 observed that both P1 and P2 forces increase
45 force remains stable before approaching the
46 with increases in both speed and defect size.
47 dipped joint, maintaining a mostly constant
48 These vertical impact forces co-exist with
49 magnitude (P0, static wheel load). However,
50
51 their corresponding horizontal component
52 during the contact at the edge of the defect
53 counterparts due to rail cant. These
54 (point x, Fig. 8) the contact ceases
55
56
horizontal impact forces are different to the
momentarily. Such behaviour is not
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 14 of 32

14
1
2
3
lateral forces due to wheel flanging contact rapidly with the increase in pad stiffness. In
4
5
6 (which have not occurred due to dipped terms of dynamic stability of contact force
7
8 joints in the current analysis). These (P1), it is therefore suggested that rail seat
9
10
11
horizontal impact forces remain around 0.05 pad stiffness should be kept low at the
12
13 in the steady state reflecting a 1:20 cant. design stage.
14
15 Similar to vertical impacts, the horizontal There are three different popular types of
16
17
18 impacts also increase with speed and defect pads used in simulations based on stiffness
Fo
19
20 size; however, relative to the vertical categorised as soft, medium and hard. Their
21
22 impacts, these values remain smaller (about values are for soft (stiffness, Kpad =100x106
r
23
24
5% of the vertical impact). Hence the lateral N/m, Cpad = 43x103 Ns/m), medium (Kpad
Pe

25
26
27 force factors determined from each analysis =280x106 N/m, Cpad = 63x103 Ns/m) and
28
er

29
are not presented here. hard (Kpad =800x106 N/m, Cpad = 93x103
30
31
32 Ns/m) [22]. The soft values of pad stiffness
Re

33
34 5. Sensitivity of some design are considered in this simulation to observe
35
36
parameters the influences on P1 and P2 forces. A
vi

37
38
39 The dependency of P1 and P2 forces to rail separate report will be developed analysing
ew

40
41 the other two cases (medium and hard)
42 seat pad damping and unsprung mass are
43
44 also investigated. However, a higher order magnitude of pad
45
46 Fig. 10 shows the sensitivity of DIF to the stiffness is usually used in railway track
47
48 design (Kpad = 450MN/m, Cpad = 65 kNs/m)
49 rail seat pad stiffness. The influence of pad
50
51 stiffness up to 106 N/m on dynamic force and these are the values used in the
52
53 simulations of this study except for the
54
factors is negligible. However, beyond this
55
56 stiffness level, vertical force increases sensitivity of pad stiffness analysis.
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 15 of 32 Journal of Rail and Rapid Transit

15
1
2
3
The unsprung mass of an axle has an damping on DIF have been examined
4
5
6 interesting effect on P2. It is noticed from independently of each other. To understand
7
8 Fig. 11 that P1 remains approximately the the combined effect of these parameters, a
9
10
11
same, but the P2 force continues to increase dimensionless factor (X) incorporating
12
13 with the increase in the unsprung mass of various design and operational vehicle and
14
15 the axle. It could therefore be inferred that track parameters has been defined through
16
17
18 high unsprung mass would significantly dimensional analysis as:
Fo
19
20 increase track deflections and possibly
21 h M V0 K pad
22 X = 4 (10)
increase the rate of track degradation. It is λ d C pad 4 E rail I rail
r
23
24
important to note that P2 force becomes
Pe

25 where h and λ d are depth and wavelength of

26
27 greater than P1 for an unsprung mass larger
28 the defect respectively in metres, V0 is the
er

29
than 3 tonnes; inclusion of primary
30 operating velocity in m/s, M is the mass of
31
32 suspension to minimise the unsprung mass is
wagon in kg, Cpad and Kpad are the damping
Re

33
34 clearly an advantage. As the second peak
35 and stiffness parameters of the pad in Ns/m
36
magnitude exceeds that of the first peak,
vi

37 and N/m2 respectively, and Erail and Irail are

38
39 further research is warranted to clarify this
ew

40 the Young’s Modulus and mass moment of

41 phenomenon for the fundamental
42 inertia of the rail in N/m2 and m4
43
44 determination of P1 and P2 forces.
respectively. There are 53 design and
45
46
47 operational parameters used in the
48 6. Non-dimensional analysis
49 simulations. Table 2 presents those
50
51 In the previous sections, the effects of speed,
52
parameters. The parameter ‘X’ is not a
53 defect size and design parameters such as
54 unique control variable on P1 and P2 forces.
55
56
primary suspensions and rail seat pad
A few important parameters out of many are
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 16 of 32

16
1
2
3
used to undertake a limited dimensional up maintaining a similar trend. The safety
4
5
6 analysis. This is effectively a parametric margin of 1.05 is included because of
7
8 study to show the combined effects of a few uncertainty in dynamic action. For example,
9
10
11
operational and design parameters on impact static linear spring stiffness is employed in
12
13 forces. the simulations and, during impact, the
14
15 P1 and P2 obtained from the twelve analyses stiffness increases due to dynamic stiffening.
16
17
18 are plotted against the dimensionless A higher margin greater than 1.05 in the
Fo
19
20 parameter X in Fig. 12. Two linear equation would give a situation diverging
21
22 relationships, one for P1 and one for P2, in from actual simulation results.
r
23
24
terms of the dimensionless parameter X are Due to the simplicity and high correlation
Pe

25
26
27 obtained with very good coefficients of with available field data, the expressions
28
er

29
correlation. It has been found that the shown in Fig. 12 for P1 and P2 may be used
30
31
32 magnitudes of P1 and P2 increase with the for the evaluation of impact forces due to the
Re

33
34 increase in the value of X. passage of wheels across dipped rail joints
35
36
It can be noted from the current models of known size. Alternatively, if the impact
vi

37
38
39 (Figs 7 and 12) that the dynamic impact forces were measured using onboard
ew

40
41 factor passes through the vertical axis at a devices, the simple equation may be used to
42
43
44 DIF of unity. However, the proposed models determine the geometric defect size.
45
46 do not. This is because an additional safety However, for the purpose of structural
47
48 margin was intentionally put into Eqs (11) design of track systems, a more conservative
49
50
51 and (12). As a result, the intercept of these prediction of the impact forces would be
52
53 equations on the vertical axis changes from necessary. With a view to understanding
54
55
56
1.0 to 1.05 and the proposed models move how well the existing classical methods
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 17 of 32 Journal of Rail and Rapid Transit

17
1
2
3
calculate the impact forces, Jenkins’ formula lower than the classical formula predicted
4
5
6 [9] and the British Rail Joint Model formula values.
7
8 [4] calculated results have been plotted in To be consistent with Eq. (11), the equation
9
10
11
Fig. 7 for varying operating speeds; for impact force P2 has also been modified
12
13 generally the classical formulae have been as shown below:
14
15 found to predict the impact force P1 with a
16 P2 = 6.0 X + 1.05 (12)
17
18 conservative margin of up to 12% on
Eqs (11) and (12) are plotted in Fig. 12 (and
Fo
19
20 average. Fig. 7 also shows the plot (referred
21
referred to in that figure as “Proposed P1”
22 to in that figure as the “Proposed Equation”)
r
23 and “Proposed P2” respectively).
24
of the predicted impact force P1 using the
Pe

25 The benefit of the above equations could be

26
27 following equation:
28 appreciated if one considers higher
er

29
30 P1 = 13.0 X + 1.05 (11) operational velocities (most rail operators
31
32 Eq. (11) has been modified by increasing the aim to increase operational speeds to
Re

33
34
35 constant from 1.00 to 1.05 to impart a improve efficiency and profitability). The
36
vi

37 conservative prediction. Fig. 7 shows that P1 and P2 impact factors predicted by Eqs
38
39
the above equation, although much simpler, (11) and (12) for the three defect sizes under
ew

40
41
42 could predict the impact force P1 with values the four operating speeds are presented in
43
44 similar to the other two classical equations Table 3. It can be seen there that, for h =
45
46
47
for speeds up to 35km/h. For speeds higher 1.2mm, an impact factor of 1.66 is reported
48
49 than 35km/h, Eq. (11) generally appears to by the P1 equation. It is interesting to note
50
51 have optimised the impact force P1 at values that a DIF of similar order was reported by
52
53
54 higher than the field measured data and Zakeri et al. [23] using a dynamic model for
55
56 simulation of the dynamic responses of a
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 18 of 32

18
1
2
3
train-track interaction system subjected to a Joint Model formula [4]. One of the reasons
4
5
6 single V-shape rail irregularity. For a single for the conservative nature of these classical
7
8 V-shaped irregularity with 1.2mm depth at formulae could be related to the earlier
9
10
11
the centre of the 1000mm rail span and a technologies of track maintenance which
12
13 train speed of 160km/hr, a maximum DIF of have improved significantly in recent times.
14
15 about 2.0 was reported; given that the coal The new formulae provide opportunities for
16
17
18 trains do not travel at speeds above 80km/hr, railway engineers to decide between the
Fo
19
20 the DIF of 1.66 predicted by Eq. (11) classical upper limit predictors and the new
21
22 appears in order and conservative. lower limit predictors in their design. The
r
23
24
Impact force factors P1 and P2 evaluated values predicted by the current equations
Pe

25
26
27 using the Jenkins’ formula and the P1 impact differed from the other classical formulae
28
er

29
force evaluated by the British Rail Joint predicted values only by a margin of 5% for
30
31
32 Model formula [4] are also presented in defect size h=0.4mm and operating speed of
Re

33
34 Table 3. The design and operational 25km/h. The higher the operating speeds
35
36
parameters used in the current simulation are and defect depths, the higher are the
vi

37
38
39 input into the empirical formulae of Jenkins difference margins, with the highest level of
ew

40
41 and the British Rail Joint Model in the variation of P1 being 45% and of P2 being
42
43
44 calculation of the P1 and P2 force factors. 75%. It therefore appears that the simple
45
46 Data for defect size h=0.4mm are presented non-dimensional Eqs (11) and (12) could
47
48 in Fig. 7. Based on the data in Table 3, it provide a cost effective, lower bound
49
50
51 could be stated that it is less conservative to measure of prediction of P1 and P2 impact
52
53 use Eqs (11) and (12) rather than the factors.
54
55
56
Jenkins’ formula [9] and the British Rail
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 19 of 32 Journal of Rail and Rapid Transit

19
1
2
3
4
5 Table 1: Defect sizes
6
7 Item Quarter wavelength, λ d (mm) Depth, h (mm)
8
9
10 Set 1 500 0.4
11
12
13 Set 2 500 0.8
14
15
16 Set 3 500 1.2
17
18
Fo
19
20
21
22
r
23 Table 2: Parameters of typical track components (* McClanachan [24], ** means assumed,
24
Pe

25
26 *** calculated)
27
28
Notation Parameter Value
er

29
30
31
mR Rail mass per meter 68kg/m
32
7.77x10-3 m2
Re

33 AR Rail cross-section area**

34
35 E Elastic modulus of rail* 2.07x1011 N/m2
36
vi

37 G Shear modulus of rail* 8.10x1010 N/m2

38
39 Iraily Rail section moment of area about Y axis* 2.94x10-5 m4
ew

40
41 Irailz Rail section moment of area about Z axis* 4.90x10-6 m4
42
43
44
Cpad Pad damping** 65 kNs/m
45
46 Kpad Pas stiffness** 450 MN/m
47
48 ms Sleeper mass** 270 kg
49
50 Isx Mass moment of inertia of sleeper about X axis** 90 kg m2
51
52 Hs1 Height between the mass centre of sleeper and its top surface** 0.115m
53
54 Hs2 Height between the mass centre of sleeper and its bottom surface** 0.115m
55
56
57 Ls Sleeper spacing* 0.685 m
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 20 of 32

20
1
2
3 Lb Effective length of support area of the sleeper at a rail seat** 1.075m
4
5
6
Bb Effective width of support area of the sleeper at a rail seat** 0.24m
7
8 Hb Height of ballast** 0.30 m
9
10 Eb Elastic modulus of ballast** 130x106N/m2
11
12 Es Subballast modulus** 50x106 N/m2
13
14 ρb Density of ballast** 2600 kg/m3
15
16
MEC Wagon car body mass (empty)* 8100kg
17
18
Fo
19 MC Wagon car body (loaded)* 66100kg
20
21 JCx Mass moment of inertia of wagon car body about X axis* 85576kgm2
22
r
23
24
JCy Mass moment of inertia of wagon car body about Y axis* 647182kgm2
Pe

25
26 JCz Mass moment of inertia of wagon car body about Z axis* 652982kgm2
27
28 mBl1 Mass of the front bolster* 465kg
er

29
30
31 mBl2 Mass of the front bolster* 365kg
32
Re

33 JBl1x Mass moment of inertia of front bolster about X axis* 175kgm2

34
35 JBl1y Mass moment of inertia of front bolster about Y axis* 9.7kgm2
36
vi

37
38 JBl1z Mass moment of inertia of front bolster about Z axis* 176kgm2
39
ew

40 JBl2x Mass moment of inertia of rear bolster about X axis* 137kgm2

41
42
JBl2y Mass moment of inertia of rear bolster about Y axis* 17.6kgm2
43
44
45 JBl2z Mass moment of inertia of rear bolster about Z axis* 138kgm2
46
47 msf Mass of sideframe* 447.5kg
48
49
50 Jsfx Mass moment of inertia of sideframe about X axis* 100.4kgm2
51
52 Jsfy Mass moment of inertia of sideframe about Y axis* 115.6kgm2
53
54 Jsfz Mass moment of inertia of sideframe about Z axis* 115.6kgm2
55
56
57 mw Wheelset mass* 1120kg
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 21 of 32 Journal of Rail and Rapid Transit

21
1
2
3
Jwx Mass moment of inertia of wheelset about X axis* 420.1kgm2
4
5
6 Jwy Mass moment of inertia of wheelset about Y axis* 100.4kgm2
7
8 Jwz Mass moment of inertia of wheelset about Z axis* 420.1kgm2
9
10
11
KSy Stiffness coefficient of secondary suspension about Y axis*** 0.85x106N/m
12
13 KSz Stiffness coefficient of secondary suspension about Z axis* 2.55x106N/m
14
15 KPa Parallelogram stiffness of bogie** 5.15x107N/m
16
17
18 CPa Parallelogram damping of bogie** 1.79x105Ns/m
Fo
19
20 LC1 Longitudinal distance from the mass centre of wagon car body
21
22 to the mass centre of the front bolster* 5.18m
r
23
24
LC2 Longitudinal distance from the mass centre of wagon car body
Pe

25
26
27 to the mass centre of the rear bolster* 5.18m
28
er

29
BSlr Semi lateral distance between the left and right secondary
30
31
32 suspension in a bogie* 0.8001m
Re

33
34 HCBe Height between the mass centre of the wagon car body
35
36
and the bolster (empty)** 0.49m
vi

37
38
39 HCBl Height between the mass centre of the wagon car body
ew

40
41 and the bolster (loaded)** 1.2m
42
43
44 HSfSb Height between the mass centre of the sideframe and the
45
46 bottom of the secondary suspension** 0.308m
47
48 HwSf Height between the mass centre of the wheelset and the sideframe* 0.05m
49
50
51 Lw Semi longitudinal distance between wheelsets in a bogie* 0.838m
52
53 rw Wheel radius* 0.425m
54 ______________________________________________________________________________
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 22 of 32

22
1
2
3
Table 3: Comparisons of P1 and P2 force factors obtained from different studies
4
5
Defect Depth h = 0.4mm h = 0.8mm h = 1.2mm
6
7
8 ( α = 0.0008rad) ( α = 0.0016rad) ( α = 0.0024rad)
9
10
Speed (km/h) 25 50 75 100 25 50 75 100 25 50 75 100

Fo
11
12
13 Equation (11) P1 1.05 1.11 1.17 1.22 1.11 1.22 1.33 1.43 1.17 1.33 1.50 1.66

rP
14
15
16 Jenkins’ Formula, P1 1.11 1.22 1.34 1.45 1.22 1.45 1.68 1.91 1.34 1.68 2.03 2.38
17
18
19 ee
British Rail Joint Model Formula, P1 1.10 1.20 1.29 1.40 1.20 1.40 1.59 1.79 1.30 1.59 1.88 2.17

rR
20
21 Equation (12), P2 1.03 1.05 1.08 1.10 1.05 1.10 1.15 1.20 1.08 1.15 1.22 1.30
22
23
24
25
Jenkins’ Formula, P2
ev
1.10 1.21 1.31 1.42 1.21 1.42 1.63 1.84 1.31 1.63 1.95 2.26

iew
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46 http://mc.manuscriptcentral.com/JRRT
47
48
Page 23 of 32 Journal of Rail and Rapid Transit

23
1
2
3
7. Conclusions phenomenon warrants further
4
5
6 research.
7
8 An analysis of wheel impact forces due to 3. The proposed equations for P1 and
9
10
11
dipped rail joints has been presented. Dipped P2 are simple to use and result in
12
13 joints were idealised as having a sinusoidal economic predictions of impact
14
15 profile in the model. The dependencies of forces due to dipped rail joints. It is
16
17
18 the dynamic impact force factors to speed, recommended that these equations be
Fo
19
20 primary suspension and rail seat pad used as lower limit predictors, whilst
21
22 damping along with unsprung mass have the classical equations [4, 9] are used
r
23
24
been presented. Simplified non-dimensional as upper limit predictors.
Pe

25
26
27 equations have been proposed for the
28
er

29
determination of the P1 and P2 impact Acknowledgements
30
31
32 factors. The authors would like to thank Rail CRC
Re

33
34 The following conclusions have emerged
35 Project 75 which partly funded the work
36
from this study:
vi

37 reported in this paper and Tim McSweeney,

38
39 1. Both the P1 and P2 forces increase Centre for Railway Engineering, Central
ew

40
41 with the increase in speed and defect
42 Queensland University, Australia for his
43
44 depth as specified. proof reading of this paper.
45
46 2. With the increase of unsprung mass,
47
48 P2 shows an increasing trend, while
49
50
51 P1 peaks at 4 tonnes/axle and then
52
53 decreases for heavier axle loads; this
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 24 of 32

24
1
2
3
References [6] Doyle, NF. Railway track design: a
4
5
6 [1] Sun, YQ. and Dhanasekar, M. A review of current practice: Occasional paper
7
8 35. Bureau of Transport Economics,
9 dynamic model for the vertical interaction
10
11 of the rail track and wagon system. Commonwealth of Australian, Canberra,
12
13 International Journal of Solids and 1980.
14
15 [7] Broadley, JR. Johnston, GD. and Pond,
16 Structures 2002; 39: 1337–1359.
17
18 [2] Shen, Z. On the principles and methods to B. The dynamic impact factor. Railway
Fo
19
20 reduce the wheel/rail forces for rail freight Engineering Conference, Institution of
21
22 Engineers Australia, Sydney, 1981, pp. 87–
r
23 vehicles. Vehicle System Dynamics Supplement
24
91.
Pe

25 1 1992; 20: 584–595.

26
27 [8] Jeong, DY. Preliminary engineering
28
[3] Iwnicki, S. Simulation of wheel-rail contact
er

29
forces. Fatigue Fracture of Engineering analysis of joint bar fatigue life, Volpe
30
31
32 Materials and Structures 2003; 26: 887–900. Center Report to Federal Railroad
Re

33
34 Administration, Cambridge, Massachusetts,
35 [4] Jeffs T. and Tew GP. A review of track
36
USA 2002.
vi

37 design procedures: Volume 2 - Sleepers and

38
39 ballast, BHP Research – Melbourne [9] Jenkins, HH. Stephenson, JE. Clayton,
ew

40
41 GA. Morland, GW. Lyon, D. The effect of
42 Laboratories, Railways of Australia 1991.
43
44 [5] Office for Research and Experiments. track and vehicle parameters on wheel/rail
45
46 Question D 71, Stresses in the rails, the ballast vertical dynamic force. Railway
47
48 Engineering Journal 1974; 3: 2–16.
49 and in the formation resulting from traffic
50
51 loads, Interim report No. 1. International Union [10] Talamini, B. Jeong, DY. and Gordon, J.
52
53 Estimation of the fatigue life of railroad joint
54
of Railways, Utrecht, Netherlands, 1965.
55
56
bars, Proceedings of the ASME/IEEE Joint
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 25 of 32 Journal of Rail and Rapid Transit

25
1
2
3
Rail Conference & Internal Combustion Engine [15] Zhai, W. and Sun, X. A detailed model
4
5
6 Spring Technical Conference, Pueblo, for investigating vertical interaction
7
8 Colorado, USA, 2007. between railway vehicle and track. Vehicle
9
10
11
[11] Steenbergen, MJMM. and Esveld, C. System Dynamics Supplement 1 1994; 23:
12
13 Relation between the geometry of rail welds 603–615.
14
15 and the dynamic wheel-rail response: [16] Lu, G. and Gill, KF. Track-
16
17
18 Numerical simulations for measured welds. transmission system dynamic analysis.
Fo
19
20 Journal of Rail and Rapid Transit 2006; Journal of Rail and Rapid Transit 1993;
21
22 220: 409–423. 207: 99–113.
r
23
24
[12] Steenbergen, MJMM. Quantification of [17] Yang, LA. Powrie, W. and Priest, J A.
Pe

25
26
27 dynamic wheel-rail contact forces at short rail Dynamic stress analysis of a ballasted
28
er

29
irregularities and application to measured rail railway track bed during train passage.
30
31
32 welds. Journal of Sound and Vibration 2008; Journal of Geotechnical and
Re

33
34 312: 606-629. Geoenvironmental Engineering 2009:
35
36
[13] Dukkipati, RV. and Dong, R. The 135(5): 680 – 689.
vi

37
38
39 dynamic effects of conventional freight car [18] Dhanasekar, M. and Bayissa, W.
ew

40
41 running over a dipped-joint. Vehicle System Performance of square and inclined
42
43
44 Dynamics 1999; 31: 95–111. insulated rail joints based on field strain
45
46 [14] Wen, Z. Xiao, G. Xiao, X. Jin, X. Zhu, M. measurements. Journal of Rail and Rapid
47
48 Dynamic vehicle–track interaction and plastic Transit 2012; 226: 140–154.
49
50
51 deformation of rail at rail welds. Engineering [19] Mutton, PJ. and Alvarez, EF. Failure
52
53 Failure Analysis 2009; 16: 1221–1237. modes in aluminothermic rail welds under high
54
55
56
axle load conditions. Engineering Failure
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 26 of 32

26
1
2
3
Analysis 2004; 11: 151–166. 24: 72-85.
4
5
6 [20] Zhai, W. Locomotive and track system [23] Zakeri, JA. Xia, H. and Fan, JJ.
7
8 coupling dynamics and its application to the Dynamic responses of train-track system to
9
10
11
study of locomotive performance. Journal of single rail irregularity. Latin American
12
13 China Railway Science (in Chinese) 1996; Journal of Solids and Structures 2009; 6:
14
15 17(2): 58–73. 89–104.
16
17
18 [21] Wu, TX. and Thompson, DJ. Theoretical [24] McClanachan, MJ. Investigation of
Fo
19
20 investigation of wheel/rail non-linear extreme wagon dynamics in Central
21
22 interaction due to roughness excitation. Vehicle Queensland coal trains, Master Thesis,
r
23
24
System Dynamics 2000; 34: 261–282. Central Queensland University, Australia,
Pe

25
26
27 [22] Ripke, B. and Knothe, K. Simulation of 1999.
28
er

29
high frequency vehicle track interaction,
30
31
32 Vehicle System Dynamics Supplement 1995;
Re

33
34
35
36
vi

37
38
39
ew

40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 27 of 32 Journal of Rail and Rapid Transit

27
1
2
3
4
5
6
Direction of Travel
7
8 Wheel
Rail 2
9 Contact Point
10 Rail 1
11
12
13
14
15
16
17
18 Figure 1: A dipped rail joint
Fo
19
20
21
22
r
23
24
Pe

25
26 P1 time series
27 P1
28
P2 time series
er

29 Wheel P2
30 Load
31
32 Po
Re

33
34
35 Time
36
vi

37
38
39 Figure 2: Wheel/rail contact at dipped joint: impact forces
ew

40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 28 of 32

28
1
2
3
4
5
6
P1
7 Impact
8 force
9
10
11 (a)
12
P2
13
14
15 Time (s)
16
17
18
P1
Fo
19
20 P2
21
22 Impact
r
23 force
24 (b)
Pe

25
26
27
28 Time (s)
er

29
30
31 Figure 3: Calculations of P1 and P2 forces
32
Re

33
34
35
36
vi

37
38
39
ew

40 λd / 4 λd / 4
41
42
43
44 α α h
45
46 µ (x)
47
48
49
50
51
52 Figure 4: Modelling of dipped joint profile
53
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 29 of 32 Journal of Rail and Rapid Transit

29
1
2
3
4
5 y
Start Finish z
6
7
8
9 x
10
11 Bogie
12
13
14
15
16
17
18
Fo
19 Rail
20 Sleepers Dipped Joint
21
22
r
23 Figure 5: Dynamic wheel/rail interaction of a half-wagon model (one bogie) on the symmetrical
24
Pe

25
26 dipped joint
27
28
er

29
30
31 220
32 Sun et al. [1] Zhai [20] Simulation
Re

33 200
34
35 Zhai [20]
180
Vertical force (kN)

36 Experimental
vi

37
38 160
39
Current Model
ew

40 140
41
42 120
43
44
45 100
46
47 80
48
49 60
50
51
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
52 Time (s)
53
54
55 Figure 6: Comparison of current model with Sun and Dhanasekar [1] and Zhai [19]
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 30 of 32

30
1
2
3 1.50
4
5 1.45
6 Jenkins et al. [9]
7 1.40
8
9 1.35
10 Proposed Equation
Dynamic Impact Factor

11
1.30
British Rail Formula [4]
12 1.25
13
14 1.20
15
16 1.15
17
18 1.10
Fo
19 Field Experimental Data,
20 1.05 Mutton & Alvarez [19]
Current Model
21
1.00
22
20 30 40 50 60 70 80 90 100
r
23
24
Pe

25 Speed (km/h)
26
27
28
er

29 Figure 7: Correlation of impact factor data of the current model with other data
30
31
32
Re

33
34
35
36
vi

37
38
39
ew

40
41
42
43
44
45
46
47
48
49
50
51 Figure 8: Time history of vertical dynamic impact factor showing a jump at x, quarter
52
53
54 wavelength of 500mm, depth of defect 0.8mm and speeds of 100km/h and 50km/h
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
Page 31 of 32 Journal of Rail and Rapid Transit

31
1
2
3 1.70
4 P1 Factor for h=0.4 mm
5
6 1.60 P1 Factor for h=0.8 mm
7 P1 Factor for h=1.2 mm
8 1.50
P2 Factor for h=0.4 mm
9
Dynamic Impact Factor

10 1.40 P2 Factor for h=0.8 mm

11 P2 Factor for h=1.2 mm
12 1.30
13
14
15 1.20
16
17 1.10
18
Fo
19 1.00
20
21 0.90
22
r
23
24 0.80
10 20 30 40 50 60 70 80 90 100 110
Pe

25
26 Speed (km/h)
27
28
Figure 9: Effect of speed and dip depth on P1 and P2
er

29
30
31
32
Re

33
34
35
36
vi

37 P1
38
39
ew

40
41 P2
42
43
44
45
46
47
48
49
50
51
52
53 Figure 10: Effect of rail seat pad stiffness on P1 and P2
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT
 Journal of Rail and Rapid Transit Page 32 of 32

32
1
2
3
4
5 P2
6
7
8
9
10
P1
11
12
13
14
15
16
17
18
Fo
19
20
21
22
Figure 11: Effect of unsprung mass on P1 and P2
r
23
24 1.80
Pe

25
26 1.70
27 P1 = 13.0x + 1.0
28 1.60 R2 = 0.9954
er

29 Proposed P1
Dynamic Impact Factor

30 1.50
31
32 1.40
Re

33
34 1.30
35
36 1.20
vi

37
38 1.10
P2 = 6.0x + 1.0
39
R2 = 0.9757
ew

40 1.00
Proposed P2
41
42 0.90
43
44
0.80
45 0 0.01 0.02 0.03 0.04 0.05 0.06
46 X (dimensionless factor)
47
48
49 Figure 12: Non-dimensional formulae for P1 and P2
50
51
52
53
54
55
56
57
58
59
60
http://mc.manuscriptcentral.com/JRRT