Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
To cite this article: Xinggao Shu & Nicholas Wilson (2008) Simulation of dynamic gauge
widening and rail roll: effects on derailment and rolling contact fatigue, Vehicle System
Dynamics: International Journal of Vehicle Mechanics and Mobility, 46:S1, 981-994, DOI:
10.1080/00423110802037214
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Vehicle System Dynamics
Vol. 46, Supplement, 2008, 981–994
This paper compares the different effects of rail rotations versus rail lateral translations on wheel/rail
(W/R) contact geometry and vehicle dynamic performance. This study, conducted by the Transporta-
tion Technology Center, Inc. (TTCI), shows that: rail rotation has significant effects on the W/R
contact geometry, especially on contact positions and contact areas, while rail translation effects on
wheel contact positions and contact areas are negligible; a W/R contact model with assumptions of
rigid track or elastic track with only lateral and vertical rail translations cannot reflect the W/R contact
geometry variations caused by rail rotations; the contact geometry variations caused by rail rotations
strongly depend on the specific wheel and rail profiles under evaluation; proper rail rotation flexibilities
within the safety limits could improve wheelset steering performance and decrease W/R contact stress
and wear.
Keywords: wheel/rail contact; track models; rail rotation; gauge spreading; NUCARS®
1. Introduction
Railway vehicles operate on flexible tracks. Track flexibility is generated by the elastic rails,
ties, fastening components, ballast, and foundations. Rail movement and resultant track geom-
etry have to be carefully controlled to secure vehicle safety and running quality. The simulation
of railway vehicle dynamics coupled with rail movements and track structure flexibility has
been needed to resolve problems for both vehicle and track as well as to better understand the
complicated mechanism of the vehicle/track coupling dynamic system.
Three kinds of track models are commonly used by vehicle dynamics simulation pro-
grammes: (a) the rigid track model, which assumes the rail is fixed to ground with infinite
stiffness, (b) the elastic support track model, which assumes the rail is rigid but can move in
only vertical and lateral directions with stiffness and damping constraints between the rail and
the ground, and (c) the flexible track model, which assumes flexible rail and tie bodies with
connections between the rails and ties and between the ties and the ground.
These three types of track models have been implemented in the NUCARS® vehicle/track
interaction multi-body simulation programme [1,2].1 The multi-body flexible track model in
NUCARS allows the simulation of a complete track structure under a vehicle, including the
effects of the flexibility of the rails and track structure. Its features have been described in
previous publications [1,2]. One of the most important features is the capability of modelling
rail movement which includes rail vertical and lateral translation and rail rotation, which
cannot be modelled with a rigid track model or an elastic support track model. Rail and cross
tie flexibilities are simulated with assumptions of Euler–Bernoulli beams.
Clearly, the introduction of the flexible track model considerably complicates the wheel/rail
(W/R) interaction simulation. There have been questions about the simulation of track
structures.
• Why do we need to simulate rail rotation since the elastic support track model already allows
the rail to move laterally and vertically?
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
• Are the simulation results significantly different without including rail rotations?
• To what extent can rail rotation affect the vehicle and track structure dynamic performance?
This paper intends to answer part of these questions through case studies and simulation
comparisons including a benchmark of simulation prediction against test results.
2. Different effects of rail rotation versus rail lateral translation on W/R contact
geometry
W/R contact geometry and contact creep forces, which result in the unique W/R dynamics
phenomena, are two fundamental aspects of railway vehicle dynamics study. Changes in W/R
contact geometry lead to changes in the wheelset’s dynamic behaviour. W/R contact geometry
is determined by the wheel and rail profiles and wheelset and rail movements. To simplify
the W/R contact geometry analysis, the track is usually assumed to be rigid without any
rail movement. The rigid track assumption is convenient for static geometry analysis, but
clearly not accurate for dynamic analysis because W/R contact geometry changes with the
rail movements. To take into account the effect of rail movements on W/R contact geometry,
the NUCARS elastic support track model introduced elastic supports beneath the rail to allow
for lateral and vertical rail translations. The gauge clearance variation caused by the two
rail lateral displacements was then transformed to an equivalent wheelset shift, which was
used to interpolate parameters in the W/R contact geometry table generated with rigid track
assumptions. Clearly, the rail vertical displacement and the rail rotation effects on W/R contact
geometry cannot be taken into account by using the elastic rail support assumption.
Under most conditions, rail head lateral displacement comes from rail rotation, as the test
results in the following sections demonstrate. In many previous studies it appears that the
rail rotation effect has been taken into account by modelling the rail lateral translation with
equivalent displacements. The key point of this study is to show the significant difference
between the effects of rail rotation and rail lateral translation on W/R contact geometry and
their effects on dynamic performance, which could not be treated equivalently.
Figures 1 and 2 show the typical new and worn AAR1B wheel/AREMA 136 pound/yard
rail profiles used in this study. The rail lateral displacement effect on W/R contact geometry
was investigated using the CFIT-WRCON Programme included in the NUCARS software
package by varying the wheelset back-to-back distance, which is equivalent to the gauge
variation caused by rail lateral displacement.
Figure 3a shows that the left and right wheel rolling radius differences change with the
variation of W/R clearances caused by the variation in the wheel back-to-back distance. The
wheel back-to-back distance values for these three cases are 1.347 m (53.047 in., label L0),
Vehicle System Dynamics 983
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Figure 2. Worn rail profiles. (a) Left rail; (b) Right rail.
1.335 m (52.547 in., labelled L1), and 1.322 m (52.047 in., labelled L2), respectively. Figure 3b
shows that the right wheel contact positions and contact patch areas are identical for these
three cases, which means the wheel contact positions and contact patch areas for the new
wheel and rail combination are not changed by the rail lateral translation. Since the wheel
contact positions are not changed by rail lateral translation, the W/R contact angles are not
changed either. The wheel contact position used in this paper is relative to the wheel coordinate
frame of reference, as Figure 1 shows. The contact patch areas were calculated using a 100 kN
(22.5 kips) vertical load on the axle centre.
Figures 4 and 5 also show that the wheel contact positions and contact patch areas for the new
and worn wheel with worn rail combination are not changed by rail lateral movement. However,
the W/R clearance is clearly changed due to rail lateral movement; the equivalent conicity on
the wheel tread also changes for the worn wheel/worn rail combination, as Figure 5a shows.
The effect of rail rotation on W/R contact geometry was also investigated by using the
CFIT-WRCON Programme by varying the rail cant angle. However, the rail is usually rotated
towards the track centre with a cant angle; the rail rotation angle for this study is towards the
field side of the rail.
Figure 6a shows that both the wheel rolling radius differences and W/R clearances change
with the rail rotations for the new wheel/new rail combination. The rail rotation angle for these
984 X. Shu and N. Wilson
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Figure 3. Effect of rail lateral displacement on contact geometry (new wheel, new rail). (a) Rolling radius difference;
(b) Wheel contact position and contact area.
Figure 4. Effect of rail lateral translation on contact geometry (new wheel, worn rail). (a) Rolling radius difference;
(b) Wheel contact position and contact area.
Figure 5. Effect of rail lateral displacement on contact geometry (worn wheel, worn rail). (a) Rolling radius
difference; (b) wheel contact position and contact area.
five cases are 0.00 (original 1:40 cant, labelled with R0), 1.43 (R1), 2.86 (R2), 3.80 (R3), and
5.00 (R4) degrees, respectively.
Figure 6b shows that the right wheel contact angles on the flange root are dramatically
changed due to the rail rotation. Figure 7 shows that the right wheel contact positions and
contact patch areas are significantly changed, even with the rail rotated only 1.43 degrees,
Vehicle System Dynamics 985
which could easily occur during vehicle curving. This is clearly different from the effect
of the rail lateral translation shown in Figure 3b. The same conclusion can also be drawn
from the W/R contact geometry analyses for new wheel/worn rail and worn wheel/worn rail
combinations, as Figures 8, 9, and 10 show.
Figure 11 illustrates how the W/R contact positions on the tread are moved towards the
flange root due to the rail rotation for the worn W/R combination. Figure 11a shows the two-
point contact position when assuming the rail cannot rotate. The right wheel flange contacts
the rail at the bottom edge of the worn gauge face. The wheel will never have a chance to
contact the rail at the contact positions shown in Figures 11b through 11d if the rail only moves
laterally without rotation.
These case studies clearly show the following.
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
• Rail rotation has a significant effect on W/R contact geometry, especially for contact posi-
tions and contact areas, while rail lateral translation effects on wheel contact positions and
contact areas are negligible.
Figure 6. Effect of rail rotation on contact geometry (new wheel, new rail). (a) Rolling radius difference; (b) wheel
contact angle.
Figure 7. Effect of rail rotation on contact position and contact area (new wheel, new rail).
986 X. Shu and N. Wilson
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Figure 8. Effect of rail rotation on contact geometry (new wheel, worn rail). (a) Rolling radius difference; (b) wheel
contact position and contact area.
Figure 9. Effect of rail rotation on contact geometry (worn wheel, worn rail). (a) Rolling radius difference;
(b) wheel contact angle.
Figure 10. Effect of rail rotation on contact position and contact area (worn wheel, worn rail).
Vehicle System Dynamics 987
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Figure 11. Effect of rail rotation on contact position (worn wheel, worn rail), with rail rotation angle 0.00(a),
1.43(b), 2.86(c), 3.80(d) degrees, respectively.
• A W/R contact model with assumptions of rigid track or elastic support track with lateral and
vertical rail translation cannot properly account for the W/R contact geometry variations
caused by rail rotations
• The contact geometry variations caused by rail rotations strongly depend on the specific
wheel and rail profiles being evaluated
These analyses show only a few cases with different rail rotation angles. For W/R dynamic sim-
ulations, the rail rotation effect could be simulated through the interpolation of pre-calculated
contact geometry tables generated using a series of rail rotation angles. The NUCARS track
model does not use pre-calculated tables; instead, the W/R contact geometry is calculated
online based on the W/R movement including rail lateral and vertical translations and rota-
tion. The railway vehicle and track system dynamic simulation is thus improved through use
of a more realistic W/R contact model with the capability of modelling rail rotation.
A gauge widening test was conducted at the Transportation Technology Center, Pueblo,
Colorado, at the Facility for Accelerated Service Testing (FAST) track to evaluate the gauge
widening strength of track with different types of ties and fastening components under vertical
and lateral loads. The vertical and lateral loads were applied to the rail heads through a con-
trolled hydraulic system installed on the track loading vehicle (TLV). First, a 178 kN (40 kips)
vertical load was applied to each of the two rail heads; the lateral gauge widening load was
then applied to the two rail heads using a 22 kN (5 kips) step increasing to the maximum 89 kN
(20 kips) load. The static rail base and rail head displacements were measured and recorded
along with the vertical and lateral loads.
A one-layer flexible track model was used in NUCARS to simulate the rail lateral and
vertical displacement and rotation under the action of the continuous widening loads. The
track model included two 35.05 m (1380 in.) long flexible rails, with 68 lateral and vertical
bending modes and 68 torsion modes around the rail centre axis in the longitudinal direction.
The rail was connected to the ground with 69 parallel connections having 20-inch spacing,
988 X. Shu and N. Wilson
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
and each connection used nonlinear stiffness and damping parameters in the vertical, lateral,
and roll directions. The dynamic simulation predictions were compared with the test results to
adjust the rail support stiffness. Figure 12 shows that the simulation predictions closely match
the test results with a maximum 1.2 degrees rail rotation angle. The NUCARS track model
simulation predictions shown in Figure 13 demonstrate that the rail head lateral displacement
is greater than the rail base as would be expected for rail rotation. As expected, displacements
are largest at the centre of the group of 15 ties surrounding the load point. These results show
that the model can be used for the evaluation of rail flexibility and benchmark purposes.
The rail rotation effects on W/R contact stress and wear were investigated through case studies
simulating a loaded 100-tonne hopper car running on a flexible track with and without rail
rotations. The three-piece truck hopper car runs through a 6-degree (291 m radius), left-hand
curve with 127 mm super elevation. The transition curve is 15 m long; the curve body is 60 m
long to ensure the vehicle reaches an equilibrium position in the curve. The running speed of
the vehicle is 64 km/h. The W/R friction coefficients are 0.5. The AAR1B new wheel and worn
rail profiles shown in Figure 2, and the same 35.05 m long, one-layer track model described
Vehicle System Dynamics 989
Table 1. Simulation cases for comparisons of the effect of rail rotations and rail lateral translations.
Case A Case B
Run Axle shift (mm) Rail roll (degree) Axle shift (mm) Rail roll (degree)
in Section 3 were used for these case studies. The implementation of the track element in
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
creepages (γx , γy , and ωz ) at the contact patch. A higher wear index can induce either rolling
contact fatigue or higher rates of wear.
Figures 18 and 19 show the Cases A and B contact wear index distribution maps on the
leading axle right wheel, respectively corresponding to the same runs of Figures 16 and 17.
Not only the contact positions, but also the wear index maximum values are significantly
different for Cases A and B, with the Case A wear index maximum values on the flange only
half of the values of Case B.
Figure 20 shows that the Case A flange contact wear index on the leading axle right wheel
at equilibrium positions linearly decreases with the decrease in the strength of the rail rotation
stiffness, while the Case B contact wear index stays relatively constant regardless of the
strength of rail lateral stiffness. This phenomenon can also be explained by the different
effects of rail rotation and rail lateral translation on W/R contact geometry. Figure 21 shows
Vehicle System Dynamics 991
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Figure 22. Rolling radius difference (run 4). (a) Case A; (b) Case B.
Vehicle System Dynamics 993
that Case A has better steering performance than Case B. The reason is that the contact points in
the flange root of Case A, shown in Figure 22, provide higher wheel rolling radius differences
than that of Case B, which results in better steering performance.
Gauge widening and rail rollover derailments occur due to the interaction of W/R forces and
the dynamic motion of the rails and track structure [3,4]. For example, on 1 July 1986, the
test train operating at FAST derailed. The accident investigation and ensuing tests for this
derailment concluded that gauge widening and rail rotation, as affected by locomotive tractive
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
effort, and differential lubrication between the high and low rails were significant contributors
to derailment.
Subsequent research work and investigations into other gauge widening and rail rollover
derailments of freight cars showed that truck warp and W/R contact geometry can also be
fundamental contributors [5–7]. The rail rotation and lubrication effects on gauge widening
derailment were simulated by using the same vehicle and track model described in Section
4. The friction coefficients for the simulation of the differential lubrication cases were 0.1 on
the high rail gauge face, 0.5 for other parts of the two rails; the friction coefficients for the dry
rail cases were 0.5 throughout the rails.
Figure 23 shows that the truck side lateral/vertical force ratios, and rail head lateral displace-
ments for the differential lubrication cases are larger than that of the dry rail cases. A gauge
spreading derailment due to rail rollover could occur as the rail rotation stiffness weakens,
as Figure 24 shows. Tests and simulations show that lubrication of the head of the low rail
can significantly decrease the gauge spreading forces and rail rotations caused by differential
lubrications.
Figure 23. NUCARS simulation of truck side lateral/vertical (L/V) ratios and rail head displacements.
6. Conclusions
The questions raised in the beginning of this paper can be answered based on the
aforementioned case studies and analyses.
• Rail rotation has a significant effect on W/R contact geometry, especially on contact posi-
tions and contact areas, while rail translation effects on wheel contact positions and contact
areas are negligible.
• A W/R contact model with assumptions of rigid track or elastic support track with lateral and
vertical rail translations cannot properly account for the W/R contact geometry variations
caused by rail rotation.
• The contact geometry variations caused by rail rotation strongly depends on the specific
Downloaded by [Moskow State Univ Bibliote] at 23:11 16 February 2014
Acknowledgements
The authors would like to thank Mr. John Elkins, formerly of TTCI, who contributed significantly to the NUCARS
track model development and the understanding of rail rollover derailments, and the AAR for allowing publication
of the test results.
Note
References
[1] J.A. Elkins, B. Brickle, N.G. Wilson, S. Singh, and H. Wu, Track structure modeling with NUCARS™ and its
validation, Proceedings of the 17th IAVSD Symposium on the Dynamics of Vehicles on Roads and Track, Lyngby,
Denmark, August 2001.
[2] X. Shu, N. Wilson, C. Sasaoka, and J. Elkins, Development of a real-time W/R contact model in NUCARS® and
application to diamond crossing and turnout design simulations, Proceedings of the 19th IAVSD Symposium on
the Dynamics of Vehicles on Roads and Track, Milan, Italy, August 2005.
[3] Heiss, Daniels, Rownd, Wilson, Laine, and Schwartz, Derailment at FAST, July 1, 1986, FRA report, FRA/ORD-
88/01, Washington, DC, January 1988.
[4] K.J. Laine and N.G. Wilson, Effect of track lubrication on gage spreading forces and deflections, Association of
American Railroads, AAR Report (R712), Washington, DC, August 1989.
[5] J.A. Elkins and N.G. Wilson, Rail rollover derailments caused by large truck turning moments on improperly
lubricated track, ASME Winter Annual Meeting, San Francisco, CA, December 1989.
[6] S.E. Mace, Evaluation of rail rollover derailment study, FRA report, FRA/ORD-93/12, Washington, DC, May
1993.
[7] D. DiBrito, S. Mace, and N. Wilson, Effects of W/R contact geometry on wheel set steering forces, 4th International
Conference on Contact Mechanics and Wear of Rail/Wheel Systems, Vancouver, Canada, July 1994.