Wear 265 (2008) 14601464

Contents lists available at ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

Wheel/rail contact analysis of tramways and LRVs against derailment

Yasuhiro Sato a, , Akira Matsumoto a , Hiroyuki Ohno a , Yasushi Oka b , Hiromichi Ogawa b
a
b

National Trafc Safety and Environment Laboratory (NTSEL), 7-42-27 Jindaiji-higashimachi, Chofu, Tokyo 182-0012 Japan
Sumitomo Metal Technology, Inc., 1-8-11 Harumi, Chuo-ku, Tokyo 104-6109 Japan

a r t i c l e

i n f o

Article history:
Accepted 6 December 2007
Available online 6 June 2008
Keywords:
Wheel/rail contact
LRV
Derailment
Contact simulation
Tramway

a b s t r a c t
Through the development of measurement tool of rail and wheel proles, it is much easier to acquire the
precision data of the proles. Although most of the digital data are used independently for the purposes
of rail or wheel prole management at present, such data make it possible to analyze the contact status
especially to identify the location of the contact point between rail and wheel tread. The authors developed
a simulation tool which can not only simulate the normal wheel/rail contact condition, but can also
simulate the situations such as back-wheel ange contact in switching section and the critical contact
condition of derailment phenomena. In this paper, the data processing method and the applications to
the analysis of derailment process of LRV vehicles are introduced.
2008 Yasuhiro Sato. Published by Elsevier B.V. All rights reserved.

1. Introduction
In recent years, some cities in Japan introduced LRVs to the conventional tramlines as LRV is environmental friendly and passenger
kindly, and some other cities are going to introduce LRT systems.
Most of the LRVs adopted in Japan have the latest low oor technology from Europe as European countries have long experience in
using LRT systems. In order to realize the low oor conguration,
most of the LRVs have independent rotating wheel-set system. This
construction makes the different running characteristics of wheels
to negotiate on rail from those of solid wheel-set systems.
On the other hand, the tracks and switches in most tramway
lines are not upgraded when LRVs are introduced. The conventional
tracks and switches have been constructed for traditional tram
vehicles and the top of rails and switches may be worn to a contour
to t with the prole of conventional wheels. In the extreme case,
LRVs are put into through-operation both on tramway system and
heavy railway system, although both systems have different regulations, for example, back wheel ange clearance, wheel ange
height, wheel diameter and so on.
For those reasons, it is very important to investigate the contact
property between wheel and rail, especially in switch section, in
order to prevent the risk of derailment before the LRV is put into
operation. Unfortunately, the analysis using digital prole data has
never been done for such purpose until now. One of the reasons

Corresponding author. Tel.: +81 422 41 3210; fax: +81 422 76 8602.
E-mail address: [email protected] (Y. Sato).

for this situation is, most of the conventional tools for wheel/rail
contact analysis are developed for the purpose of new wheel prole
design [1]. They consider mainly on the prole of wheel tread but
not of wheel ange, especially of the back ange of wheel which
will contact with guard rail in switch section or in curved section
of tramway lines.
In order to deal with this problem, a wheel/rail contact simulation tool is developed. This tool can deal with not only the
conventional contact problem of wheel tread contact including two
point contact (wheel tread and wheel ange), but with the back
ange contact problem including three points contact (wheel tread,
wheel ange and back of wheel ange which may happen in switch
section). This tool can also deal with wheel-set roll over situation
which may happen in the lateral turn over derailment.
In this paper, the details of the newly developed wheel/rail contact analyzing tool and its application to the investigation of LRV
derailment is reported, and the proposal to evaluate the derailment
safety by concerning the contact location on wheel is explained.

2. Contact analyzing tool

2.1. Prole data
The digital prole data of rail and wheel are measured by using
MiniProf developed by GREENWOOD Engineering [2]. As the proles of wheel and rail are measured respectively, they are combined
and deployed into the same plane coordinate by considering the
back ange distance of wheel-set and the geometry of track. The
coordinate is dened by the top level of rails and the center axis of

0043-1648/$ see front matter 2008 Yasuhiro Sato. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2007.12.019

Y. Sato et al. / Wear 265 (2008) 14601464

1461

Fig. 1. Combined matrix of wheel and rail coordinate.

Fig. 4. Denition of contact angle.

Fig. 2. Contact point analysis model.

wheel-set. The cant elevation of track is neglected. A given vertical distance is added to the wheel prole matrixes. Fig. 1 shows
the typical combined matrixes of wheels and rails in the plane
coordinate.
2.2. Contact points analyzing process
Contact points analysis is done for given lateral displacement
of wheel-set. The yawing angle of wheel-set is neglected because
3D contact analysis is necessary for the yawing movement of
wheel-set, and the analysis will be much more complicated. Elastic
deformation and displacement are also neglected.
At rst, the shortest vertical distance between wheel and rail of
each pair is calculated as shown in the model of Fig. 2. The shortest vertical distance of left wheel/rail pair may not be the same as
that of right wheel/rail pair. Fig. 3 shows the case where the left
wheel/rail pair may still remains a certain distance to get in contact when the wheel-set moves vertically down through the same
distance as the shortest distance of right wheel/rail pair. In this
case wheel-set is rotated around the wheel-set center by considering the distance difference between left side and right side pairs.
Coordinates of wheel prole matrix are recalculated according to

Fig. 3. Vertical distance difference.

the wheel-set rotation angle. The shortest distances between wheel

and rail pair are analyzed again. This process is repeated again and
again until the distance difference is small enough to an acceptable
value. At this state, the coordinates which have the shortest vertical
distance are considered as the contact points.
Contact angle is obtained after the contact point is xed. Fig. 4
shows the denition of contact angle between rail and wheel. This
gure shows the enlargement of the digital rail prole coordinates
including the contact point. The inclination angle of the approximate line of 5 continuous coordinates around the contact point is
dened as the contact angle.
For high rail side, the vertical distance between wheel ange
and rail is also analyzed when wheel-set lateral displacement is
large enough that the wheel ange may get in contact. With the
increase of wheel-set lateral displacement, the contact point on
wheel may jump from tread section to wheel ange. The transfer
state is dened as two points contact.
For low rail side, the vertical distance between wheel back ange
and guard rail is also calculated if guard rail exists. Sometimes the
wheel tread of low rail side is going to be elevated from rail once
the wheel back ange get in contact with the guard rail. The contact
analysis of low rail will be done just between the wheel back ange
and the guard rail.
Wheel rolling radius and altitude from top of ange to the contact point can be calculated once the contact point coordinate is
obtained.
The analyzing tool is extended to analyze the state of the wheelset of train overturning process. In the state of train overturning
process, just one side of wheel/rail pair engaged in the contact
and the analysis is done for one side. As shown in Fig. 5, at
rst, the contact point for the shortest vertical distance between
wheel and rail is detected, and after that wheel-set is rotated little by little around this contact point until another contact point
appeared. The angle of the wheel-set at this state is the overturning
angle.

Fig. 5. Wheel-set rolling over analysis model.

1462

Y. Sato et al. / Wear 265 (2008) 14601464

Fig. 7. Contact with wheel-set gauge.

Fig. 6. The derailed switch section.

3. Application in derailment investigations

3.1. LRV derailment investigation
3.1.1. Derailment site investigation
This analyzing tool was applied to the derailment accident investigation of new LRV systems. The line where new LRVs were put into
operation is basically a tramway line, but the vehicles were used for
through-operation to a conventional railway line. For this reason,
the conventional vehicles are constructed under the regulation of
railway line. On the other hand, LRVs are designed under the regulation of tramway. Before the rst LRV eet was put into service,
many running tests were done in depot and in the tramway section.
The wheel ange of the rst eet had been worn in this period and
after that it was put into service operation. Not so many test run
was carried out for the second eet because the performances of
LRVs were considered equal to the results of test runs of the rst
eet.

The accident happened in curved section of railway line just one

day after the new LRV of second eet was put into service, where
the rst eet had operated for months. After the derailment Guard
Rail was installed to the derailed section in order to prevent the
wheel from derailment. Derailment happened again in a switch of
tramway section just one trip after the re-operation. Fig. 6 shows
the section of the switch where the derailment happened. From the
gure it can be found, this section is tramway and the switch has
only one tongue rail in it.
Top of rail proles in the derailed section are measured. Wheel
proles of LRV eets and the typical conventional eet are also measured. At the same time, a wheel-set gauge of LRV is manufactured
and taken to site for contact state identication. Fig. 7 shows one
of the contact states by using the wheel-set gauge.
3.1.2. Contact simulation of derailment process
Wheel/rail contact simulations between those measured data
and designed proles are carried out. The simulation results show
perfect agreement with the phenomena observed at the derailment site. Figs. 8 and 9 show the simulation results in conventional
railway section and Fig. 10 in tramway line.
From these simulation results it can be found, the contact characteristics between rst eet and second eet are quite different,
especially in the wheel ange contact region. For the rst eet, the
contact angle reaches to the maximum value of about 70 once the
wheel ange enters into contact region. For the second eet, wheel
ange enters into contact region by much less lateral displacement

Fig. 8. Contact simulation in railway section (leading wheel-set of the rst LRV eet).

Y. Sato et al. / Wear 265 (2008) 14601464

1463

Fig. 9. Contact simulation in railway section (leading wheel-set of the second LRV eet).

Fig. 10. Contact simulation in tramway section.

of the wheel-set and the contact angle is much smaller at the rst
stage of ange contact. According to the Nadals equation [3], the
critical value of derailment coefcient become smaller for small
wheel ange contact angle, and it means easier to derail.
Another fact found from the simulation results is: the vertical
distance between the contact point at wheel ange and the top of
rail is much small for the second eet compared with that of the
rst eet. Shallow contact position means less friction energy is
needed for the contact point transporting to the top of rail (derail).
As the thickness of ange and the back ange distance of the
two eets are different, the low rail side wheel of the rst eet is
much easier to be guided by the guard rail in tramway compared
with the second eet shown in Fig. 10. In Fig. 10, the back ange
of low rail side wheel has not been guided by guard rail even the
ange of high rail side wheel has contacted with rail. This means,
some unguarded blank exists between the back ange and guard
rail for the wheel-set of the second eet, and wheel-set cannot be
guided by the guard rail even in sharp curve section of tramway.
According to the simulation and other analysis results, solutions
to prevent derailment had been proposed to the railway company
and the vehicle manufacture. The proposals were accepted and
alternation of wheel-set design was carried out. Now the second
eet is in operation on tramway and conventional railway line without problem.

3.2. Rolling over derailment simulation

In case of over speed operation at curve section or in strong
lateral wind blowing, vehicle may be put into the danger of turn
over derailment. In this situation, wheel-set may roll over on the
track. Sometimes, the contact traces on top of rail can be found in
the derailment section (Fig. 11). It is not easy to identify whether
or not the trace is printed by the derailed wheel.
This analyzing tool can help to verify the practice. Fig. 12 shows
one of the simulation results to identify the contact trace printed
by the rolling over wheel. Zero point in this gure is the point to
dene the track gauge and is used to calculate the lateral position
of contact point.

4. Derailment measuring concept

Nadals limit is dependent on wheel/rail proles and the coefcient of friction between them, which is generally believed to be
conservative. Measured lateral-to-vertical load values or recorded
value during test [4] often exceeded Nadals limit without any
apparent danger of wheel climbing. In recent years, safety criteria
were examined at a eld test in turnout in Japan [5]. They investigate the behavior of the wheel-set and the relationship between

1464

Y. Sato et al. / Wear 265 (2008) 14601464

Fig. 13. Contact simulation of derailment process.

Fig. 11. Contact trace on top of rail.

of wheel climbing, the movement of contact point involves the friction energy. Total energy from the state of rst wheel ange or back
ange contact to the critical point on rail or guard rail will be evaluated by this parameter. The detail of the evaluation methodology
is still under verication.
5. Conclusions

Fig. 12. Rolling over and contact trace.

some kinds of parameters and the wheel rise. In another research,

the relationship between the friction coefcient and wheel climb
derailment is particularly discussed [6]. In both of the papers, the
wheel rise similar to vertical displacement of the wheel is investigated. It is needless to say that the vertical displacement of the
wheel is directly connected with the risk of wheel climb derailment.
We look at the wheel rise in a new light. The wheel rise is related to
the quantity of the shift of contact point between wheel and rail. So
the location of contact point on wheel ange or back ange has great
inuence on the derailment crisis. Apart of the derailment coefcient (lateral force/wheel load), which is commonly used in Japan
to evaluate the safety of vehicle in curve section, a new evaluation
parameter which concerns with contact point location is proposed.
This parameter is named derailment energy. This parameter is
the function of the altitude of contact point to the critical point
on rail gauge corner, wheel load, lateral contact force, friction
coefcient and so on. Fig. 13 shows one of the examples of movement of contact point on wheel and on top of rail. In the derailment

In order to analyze the derailment process of LRV in tramway and

in conventional railway, a simulation tool which can not only simulate the normal wheel/rail contact condition, but can also simulate
the situations such as back-wheel ange contact in switching section or the contact at critical condition of derailment is developed.
This tool is applied to the derailment investigation of LRV. Simulation results agree well with the practical measurements. From the
analysis by using this tool, the reasons of derailment are illuminated. Solutions were proposed to the railway company and the
manufacture. After the improvement according to our proposal,
derailment risk of the LRV is greatly reduced.
From the process of derailment investigation, a new parameter
we call derailment energy, that can evaluate the derailment risk
is proposed.
References
[1]
[2]
[3]
[4]

I.Y. Shevtsov, V.L. Markine, C. Esveld, Wear 258 (2005) 10221030.

http://www.railway-technology.com/contractors/track/greenwood/.
M.J. Nadal, Ann. Mines 10 (1896) 232288.
W.C. Shust, J.A. Elkins, S. Kalay, M. ElSibaie, Wheel Climb Derailment Test Using
AARs Track Loading Vehicle, AAR report R-910, Pueblo, CO, 1997.
[5] H. Ishida, K. Ueki, Y. Yamashita, K. Asano, H. Kaneko, Proc. Jpn. Soc. Mech. Eng.
(9536) (1995) 391393 (in Japanese).
[6] K. Nagase, Y. Wakabayashi, H. Sakahara, Proc. Inst. Mech. Eng. 216 (2002) 237247
(Part F).