Full Scale Derailment Testes On Freight Wagons
Full Scale Derailment Testes On Freight Wagons
To cite this article: G. Diana, E. Sabbioni, C. Somaschini, D. Tarsitano, P. Cavicchi, M. Di Mario &
L. Labbadia (2021): Full-scale derailment tests on freight wagons, Vehicle System Dynamics, DOI:
10.1080/00423114.2021.1877745
Rail, Italy; c Ricerca e Sviluppo, RFI-Rete Ferroviaria Italiana, Roma, Italy; d Direzione Tecnica, Trenitalia S.p.A.,
Italy
1. Introduction
Derailment of freight trains may cause serious damages to the railway infrastructure, the
rolling stock and the environment, which yield to economic losses, service disruptions,
casualties and other undesirable consequences. Therefore, improving running safety of
freight trains it has been, and still is, a high priority for railway industries and governments.
Being most of derailments due to infrastructure or rolling stock failures [1–4], accurate
maintenance standards were defined to reduce the derailment risk. Moreover, a wide range
of systems, to detect derailment and/or mitigate the outcome of a derailment, were investi-
gated [5–8], both wayside (hot axle/bearing detectors [9–11], acoustic bearing/wheel defect
detectors [12,13], and wheel impact load detectors [6,12]) and on-board vehicle (based on
acoustic emissions [14] or, more often, on acceleration measurements [15,16]).
The present paper focuses on this latter solution. Wayside detection systems must, in
fact, be permanently placed at strategic locations along railroad tracks to monitor passing
trains and to investigate specific safety-related symptoms. However, in the case of freight
trains, there sometimes occur situations when one or several wheelsets derail and continue
their motion over the sleeper cage. Train motion resistance in such cases does not signif-
icantly change and the driver might not notice the derailed wheelset. Then the derailed
wheelsets, continuing their motion, may seriously damage the track and, when a curve or
a switch is reached, may cause the overturning of the wagon.
To avoid these conditions, which obviously threaten safety and integrity of the infras-
tructure, derailment detectors were introduced and installed on freight wagons [16].
Derailment Detectors (DD) are fundamentally based on measuring impacts as one/more
wheelsets derail. As a triggering threshold is reached, the derailment detector stops the
train by activating automatic application of the brake. Although reliable in most of con-
ditions, some cases of false tripping or miss-detections led the European Railway Agency
(ERA) to promote further investigations before deciding on imposing by law the use of
derailment detectors on freight trains [4,17].
A possible solution for a more robust derailment detection would be represented by a
device, equipping each wagon, which includes the sensors, an autonomous power supply
system (such as those presented in [18]) and a processor that analyses the signals and sends
a warning to the locomotive. On the locomotive, the information about which wagon/axle
has derailed is provided to the driver. It is the driver who decides whether to stop the train
or not and which braking procedure to use.
The aim of this study is to analyse which sensor signals, data processing algorithm and
criteria can be used to make derailment detection more reliable. On this purpose, full-scale
derailment experimental tests were carried out with an instrumented freight wagon and the
results were systematically analysed.
Owing to the complexity of the experiment, only a few data are available in the scientific
literature concerned with derailment tests and which can be profitably used to design the
detection system. Most of the tests are carried out in laboratory using scale models [8,19]
or roller rigs for studying the derailment dynamics of a single axle [20,21]. As an example,
Braghin et al. [20] propose some derailment criteria less conservative than the one by Nadal
[22] on the basis of the results of wheel-climb tests carried out with a single solid axle placed
on a roller rig. However, these criteria require the measurement of the longitudinal contact
force component and/or of the angle of attack.
In [19], tests with a 1/10 scale model were carried out to investigate the dynamics of
a wagon during low-speed wheel-climb derailment, using an automotive IMU (Inertial
Measurement Unit). It was found that the vertical acceleration of the axle-box and the
pitch and roll rates of the carbody may provide useful information to detect the early signs
of derailment.
In [23] the results of full-scale experimental tests carried out with a three-wagon train,
where the last truck was equipped with a specialised wheelset, are presented. During the
tests, a device causes the derailment of the specialised wheelset. Analysing the acceleration
measured at the centre of the carbody, it was found that, after the derailment of the
specialised wheelset, the sleeper frequency can be identified.
In [24] low-speed full-scale derailment tests were performed to study post-derailment
safety measures. Tests were carried out using both a trailer and a motor vehicle. Two kinds
VEHICLE SYSTEM DYNAMICS 3
of track (CRT-I slab ballastless track and CRT-II bi-block sleeper ballastless track) were
adopted to investigate the effect of the track type on the derailment process. Test results
indicated that post-derailment movements of the vehicle include running and vertical
bounce. Reducing the derailment speed and increasing the weight of the head of the train
allow reducing the possibility of derailment. Moreover, it was found that CRTS-II ballast-
less track contributes to constrain the lateral movement of the derailed vehicle and reduce
the post-derailment distance.
The present paper presents a further investigation of wagon dynamics both during
wheel-climb derailment and during the motion with one axle derailed. Two different full-
scale derailment tests were, in fact, carried out with an instrumented wagon equipped with
a commercial derailment detector. During the first test, the freight wagon was towed by the
locomotive with one wheelset derailed. During the second test (similar to the experiments
reported in [24]), the freight wagon was launched by the locomotive on a derailer (specifi-
cally designed for the test) to assess its behaviour during the derailment phases. Tests were
performed within the JRC (Joint Research Centre) Transportation at the Railway test cen-
tre of Bologna San Donato using two wagons, one empty and one loaded with 50 tons.
During the tests, accelerations were measured at the bogie and at the carbody. Moreover,
the intervention of the DD was monitored. The performed tests allowed to
• assess the behaviour of a freight wagon during the derailment and during motion with
only one wheelset derailed;
• analyse the behaviour of a commercial pneumatic derailment detector;
• evaluate the wagon acceleration during the derailment phases.
Moreover, tests highlighted the operating thresholds for railway detectors and possible
causes of false tripping and/or miss-detections. As mentioned earlier, this information is
intended to be used for the development/improvement of derailment detection algorithms.
2. Tested vehicle
Derailment tests were carried out using two instrumented 4-axle freight wagons, (series
Rgmms R1F, Figure 1(a)), equipped with Y25 bogies. The main characteristics of the
vehicle are reported in Table 1.
In order to investigate different loading conditions, two different tests were carried out:
one with an empty wagon (tare weight 20.5 ton) and the other using a laden wagon (global
weight equal to 68 ton). A diesel locomotive D345 was used to accelerate the wagons.
The selected locomotive is capable to accelerate the wagon (both empty and laden) from
standstill to about 30 km/h over a distance of 250 m.
During the tests, a commercial pneumatic Derailment Detector (DD) was mounted on
the buffer beam at one end of each wagon (Figure 1(b)). The considered derailment detec-
tor consists of an emergency brake valve, a spring-mass valve and an indicator device. In
case of derailment, the spring-mass valve detects the increase in the vertical acceleration
and immediately activates the emergency brake valve, which opens the main brake pipe
and releases the emergency brake. Specifically, the spring-mass valve is triggered when
the vertical acceleration gets into the range 6.5–11.5 g. To avoid braking the vehicle when
derailment conditions are detected and thus keep the wagon running, during the tests the
4 G. DIANA ET AL.
Figure 1. (a) Freight wagon Rgmms; (b) Air reservoir connected to the derailment detector.
Figure 2. Test track line (left): details of sleepers and fasteners (right).
derailment detector was connected to a compressed air reservoir instead of the main brake
pipe, as shown in Figure 1(b) (see Section 4 for details).
3. Performed tests
Tests were performed at the Railway test centre of Bologna San Donato. The railway track
is standard railway line (six-track line), consisting of flat-bottom steel rails supported by
concrete sleepers, which are themselves laid on crushed stone ballast [25]. The gauge is
1.435 m and the sleepers are equally spaced of 0.7 m (see Figure 2). The two central rails,
which run straight for about 400 m, were used during the tests.
Two different tests were carried out (make reference to Figure 3), which are described
in the following:
Figure 3. Performed tests: (a) full derailment tests; (b) single-derailed-axle tests.
Tests were repeated considering two loading conditions (freight wagon empty and
laden).
Figure 4. (a) Details of the right ramp of the derailer. (b) Derailer placed on the rail and approach of the
wheels to the wedges.
during wheel-climbing and sideways shove, wedge profiles of the designed derailer were
properly smoothed (see Figure 4(a)).
The performed test aims at identifying possible causes of miss-detections as the wagon
speed increases.
4. Experimental set-up
As mentioned earlier, aim of the tests is measuring the freight wagon accelerations during
both wheel-climbing derailment and motion with a single derailed axle, to identify
(a) the conditions triggering the derailment detector and the activation threshold;
(b) relevant information for developing a derailment detection algorithm.
On this purpose, the freight wagon was instrumented with several accelerometers.
The position of the sensors, installed on the freight wagon for Test 1 and Test 2, is respec-
tively schematised in Figure 6(a,b), together with the reference system. The measurement
axis of the accelerometers is indicated with an arrow.
As it can be seen, accelerometers are placed in correspondence of
• the four corners of the bogie to measure vertical (z-axis) and lateral (y-axis) accelera-
tions;
• the derailment detector (DD) to measure the vertical acceleration sensed by the device.
Figure 7. Installed sensors: (a) accelerometer in correspondence of the DD; (b) pressure transducer; (c)
piezo-accelerometer at one corner of a bogie; (d) encoder.
VEHICLE SYSTEM DYNAMICS 9
Figure 8. Scheme of the connection between the DD and the air reservoir.
Signals were sampled at 10,250 Hz and low pass-filtered at 100 Hz with a fourth-order
Butterworth filter.
The choice of the filter cut-off frequency was driven by these criteria:
• highlight the low-frequency vehicle dynamics during the derailment process, in order
to clearly identify the phases of the derailment (see the next section for details). As men-
tioned in the introduction, the aim of the research is to analyse the derailment process
in order to define an algorithm able to detect derailment, which is not only based on the
crossing of an acceleration threshold;
• filter out spikes due to deformable modes of vibration of the frame of the wagon, excited
by impacts (not necessarily due to derailment) and which may erroneously trigger the
intervention of a derailment detector;
• analyse the behaviour of a commercial pneumatic DD, which, as mentioned earlier,
is triggered by a spring-mass valve. As a matter of fact, the considered DD acts like
a mechanic low-pass filter as its bandwidth is limited by the dynamics of spring-mass
valve. Therefore, including high-frequency components in the acceleration signal would
make difficult to study the behaviour and the operating threshold of the DD.
As an example, Figure 9 shows the effect of the filter cut-off frequency on the vertical
acceleration measured at the DD during the full derailment test carried out with the laden
wagon. As it can be seen, when the filter cut-off frequency is set at 2500 or 1000 Hz, accel-
eration peak values are more than doubled, reaching maximum values of 35 g.1 Moreover,
when high-frequency acceleration components are not filtered out, the phases (see the next
session for details) of the derailment are hardly recognisable.
Figure 9. Time history of the vertical acceleration at the DD during the full derailment test with the
laden wagon: effect of the filter cut-off frequency.
• the derailment of the first axle does not trigger the intervention of the DD until the
impact of the axle-box with the rail occurs (point 2 in Figure 11 and Figure 12-2);
• the acceleration peak during the collision between the axle-box of the first wheelset and
the rail is comparable for the empty and the laden wagon (about 10 g);
VEHICLE SYSTEM DYNAMICS 11
Figure 10. Time history of the speed of the wagon, of the pressure in the air reservoir (upper part) and of
the vertical acceleration at the DD (lower part); the green dashed-dotted line represents the intervention
of the DD. (a) Laden wagon; (b) empty wagon.
• in the laden-wagon test, derailment of the third and fourth wheelsets (points 5 and 6
in the upper part of Figure 11) yields to peak accelerations comparable with one of the
first wheelsets. Vice versa, when the wagon is empty, peak acceleration due to the derail-
ment of the third and fourth wheelsets (points 5 and 6 in the lower part of Figure 11) is
12 G. DIANA ET AL.
Figure 11. Vertical acceleration in correspondence of the DD: laden wagon (top), empty wagon (bot-
tom). Red dashed line identify the intervention of the DD. Points: (1) first axle approaching the derailer;
(2) first axle-rail impacts; (3) second axle-rail impacts; (4) third axle approaching the derailer; (5) third
axle-rail impacts; (6) fourth axle-rail impacts; (7) wagon stands still.
Figure 12. Derailment of the laden wagon3 (the yellow line highlights the position of the derailer).
significantly lower than the one due to the derailment of the first axle. When the wagon
is empty, motions of front and rear bogies are less coupled, since the carbody is free
to pitch. Therefore, when the front bogie is derailed, the rear one bounces on the rail.
The motion of the vehicle during derailment includes running and vertical bounce [24].
Therefore, when the wagon is laden, after the collisions it does not bounce vertically but
mainly runs on the track bed and sleepers;
• the laden wagon stops almost immediately after the derailment of the fourth axle (point
7 in Figure 11): the vehicle travels for less than 1 m (about 1 s) after the derailment of
all the axles. Vice versa, the empty wagon continues to travel for more than 2 s after the
derailment of all the axles (about 5.5 m). As previously mentioned, when the wagon is
loaded, its motion is mainly running. This and the additional mass, due to the presence
of the load, increase the resistance to the longitudinal motion of the derailed vehicle. As
a consequence, the post-derailment distance is reduced when the vehicle is loaded [24];
VEHICLE SYSTEM DYNAMICS 13
Figure 13. Vertical acceleration at the DD and at the front left/right corners of the bogie. Laden wagon.
Low-pass filter with cut-off frequency set to 40 Hz.
• in the laden-wagon case, after the derailment of the front bogie and before the rear bogie
arrives at the derailer (between 0.75 and 1.5 s, the upper part of Figure 11), a frequency,
related to the passage of the derailed wheelsets (i.e. the axles of the front bogie) over the
sleepers, can be recognised. Vice versa, in the empty wagon case, the accelerations, due
to the passage of the derailed wheelsets over the sleepers, are clearly visible only after
all the wheelsets have derailed (between 2.25 and 3.2 s, the lower part of Figure 11).
Since there is no load, the wagon bounces on track and thus the sleeper frequency is
less visible.
As mentioned earlier, the impact of the axle-box of the first derailed axle with the rail
(Figure 12-2) triggers the intervention of the DD, when the wagon is empty or laden.
During the impact, the acceleration at the DD exceeds 10 g in both cases.
To deepen the analysis during the derailment of the first axle, the laden wagon test is
analysed in more in detail.2 Figure 13 focuses on the derailment of the first axle (first 1s of
the test). The vertical accelerations measured at the DD (blue line) and at the front corners
of the front bogie are shown. The black solid line highlights the intervention of the DD.
Measurements are low-pass filtered at 40 Hz to better highlight bogie and wagon dynamics
during the derailment. Note that the derailer provides a shove to the right according to the
running direction and thus the left axle-box is the one colliding with the rail.
The following events can be clearly identified in the accelerations at the bogie corners
and at the DD: approach of the first axle at the derailer, derailment of the first axle, impacts
of the axle-box of the first and the second wheelset with the rail. It can be noticed that the
peak acceleration at the front left corner of the bogie, when the axle-box collides with the
rail, is about five times larger than the one at the DD. The acceleration peak at the right
corner of the bogie is instead almost the double of the one at the DD.
14 G. DIANA ET AL.
Figure 14. Vertical acceleration at the DD and at the back-tail of the wagon. Laden wagon.
It can also be observed that the first axle derails less than 0.3 s after its arrival at the
derailer (at a speed of 25 km/h). Moreover, acceleration at the DD is delayed with respect
to acceleration at the bogie front corners of about 0.025 s. As already noticed by other
researchers [19], early detection of derailment would be in theory possible (e.g. measuring
the bogie/frame pitch rate). However, considering the dynamics of a pneumatic braking
system, this information cannot be effectively used for the implementation of a system for
derailment prevention. Vice versa, this information may be profitable used by a derailment
detection algorithm.
While during the derailment of the front bogie, the wagon dynamics is similar for the
laden and the empty case, it is significantly different in the following. As mentioned earlier,
when the wagon is empty, the carbody is freer to pitch. As a consequence, accelerations,
induced by the derailment of the rear bogie on the front part of the wagon, are small com-
pared to the laden case. To further analyse the laden case, Figure 14 compares the vertical
acceleration at the front (in correspondence of the DD, red line) and at the rear of the wagon
(blue line). The green dashed line represents the intervention of the DD. As it can be seen,
derailment of the first axle leads to accelerations lower than 2.5 g at the rear of the wagon.
These accelerations would not trigger the intervention of a DD placed at the other end
of the wagon with respect to the derailing axle (the activation threshold is, in fact, above
6.5 g). This clearly highlights the need of installing a DD at each end of the wagon, as it is
already done in standard working conditions.
Attention is focused on the empty wagon test. Figure 15(a) shows the acceleration mea-
sured in correspondence of the DD (blue curve), which is placed at the back of the wagon
(i.e. near to the derailed axle) and the wagon speed (red curve) vs. the travelled distance.
The magenta dashed line identifies the intervention of the DD.
The motion over the sleepers and the rail fastening system (screw and rail clamp) of
the derailed axle excites vibrations with the sleeper frequency (which are spaced of 0.7 m
along the considered track). Amplitude of vibrations shows a dependence on vehicle speed
[23]. Specifically, amplitude of vibrations increases from 0 to 15 km/h, as it can be seen
in Figure 15(c) where a zoom of the vertical acceleration at the DD is reported (region 1
in Figure 15(a)). Until speed is lower than 5 km/h, peak acceleration remains much lower
than the DD activation threshold (i.e. 6.5 g). Thus, if the derailment occurs at very low
speeds (e.g. during operations in depots), the DD may miss-detect it and the wagon would
continue its motion with one or more axles derailed damaging the infrastructure. In the
speed range 5–15 km/h, vibration amplitude reaches its maximum value. Within this range,
in fact, the sleeper frequency lies in the range of natural frequencies of the wagon (about
3–4 Hz for the empty vehicle). When the bounce motion of the wheelset is excited in res-
onance (speed is in the range 8–10 km/h), the axle-box of the derailed axle collides with
Figure 15. Single-derailed-axle test with empty wagon: (a) vertical acceleration in correspondence of
the DD and wagon speed vs. travelled distance; (b) impact between the axle box of the derailed wheelset
with the rail; (c) zoom of the vertical acceleration in the speed range 0–15 km/h (region 1) and (d) in the
speed range 22–25 km/h (region 2).
16 G. DIANA ET AL.
the rail (see Figure 15(b)), triggering the intervention of the derailment detector, which is
highlighted by the magenta dashed line in Figure 15(c). Therefore, similar to the wheel-
climbing derailment, the DD is activated by the impact of the axle-box of the derailed
axle with the rail, which yields to a peak acceleration exceeding the minimum activation
threshold of 6.5 g.
As the speed increases, amplitude of vibrations, induced by the passage over the sleep-
ers, reduces since modes of vibration of the vehicle are excited in seismic zone. Figure 15(d)
shows a zoom (region 2 in Figure 15(a)) of the acceleration in correspondence of the DD
when the vehicle speed is in the range 22–25 km/h. The sleeper frequency is still recognis-
able, but the amplitude remains always in the DD no-activation zone (vertical acceleration
lower than 6.5 g). This suggests that if one or more axles derail (due to a broken rail, other
rail/axle defects or bad vehicle handling [2]) within this speed range, a miss-detection may
occur, since the triggering threshold may not be reached. An algorithm able to process and
analyse the measured signals would instead be able to detect the derailment anyway.
6. Concluding remarks
This paper presented the results of experimental full-scale derailment tests on an instru-
mented freight wagon. Two different tests were carried out: derailment of the full wagon
(achieved using an ad hoc designed derailer) and wagon running with one derailed axle.
During the tests, a commercial derailment detector was mounted on the vehicle and exper-
iments were repeated with the vehicle empty and laden. The main achievements can be
summarised as follows:
• the activation threshold of the derailment detector is about 7 g (consistently with its
datasheet);
• during all the performed tests, the activation of the derailment detector is triggered by
the impact of the axle-box of the derailed wheelset with the rail. If no impacts between
the axle-box and the rail occur, the activation threshold of the derailment detector might
not be exceeded. This could be the case of a derailment at low speeds due to rail/axle
defects or bad vehicle handling;
• tests with a single derailed axle suggest that if one or more axles in the speed range
15–25 km/h derail, a miss-detection may occur since the vibrations, induced by the
passage of the derailed axles over the sleepers, might not trigger the activation of the
derailment detector if the wagon is empty. Within this speed range, in fact, the sleeper
frequency (generally spaced of 0.65–0.76 m [28]) is likely to excite the wagon in seismic
zone;
• performed tests confirm the need of installing a derailment detector at each end of a
wagon, as already prescribed for ordinary working conditions.
Performed tests highlighted that devices detecting derailment, based on the crossing
of a threshold, may lead to some miss-detections in some specific conditions. To address
these issues a derailment detection algorithm, able to process and analyse the acceleration
signals, could be developed based on the data collected within the present study and
included into an electric control unit having computing capabilities.
VEHICLE SYSTEM DYNAMICS 17
Notes
1. Note that this is not the peak value measured during the tests. This result justifies the choice
of the measurement range of the accelerometers: a lower measurement range would lead to a
possible saturation of the signal; this compromises the measurement.
2. Similar considerations can be drawn considering the empty wagon test.
3. Readers are invited to view the videos supplied in the electronic version of the paper to have a
clearer illustration of the derailment process.
Acknowledgements
The authors kindly acknowledge the JRC (Joint Research Centre) Transportation for having funded
this research project and all the other partners, especially the ones involved in the task.
Disclosure statement
No potential conflict of interest was reported by the author(s).
ORCID
E. Sabbioni http://orcid.org/0000-0002-4356-8814
References
[1] Ulianov C, Defossez F, Franklin G, et al. Overview of freight train derailments in the EU: causes,
impacts, prevention and mitigation measures. Traffic Safety. 2016;4:317–336.
[2] Liu X, Saat MR, Barkan CPL. Analysis of causes of major train derailment and their effect on
accident rates. Transp Res Rec: J Transp Res Board. 2012;2289(1):154–163.
[3] Cherkashin UM, Zakharov SM, Semechkin AE. An overview of rolling stock and track moni-
toring systems and guidelines to provide safety of heavy and long train operation in the Russian
Railways. Proc IMechE, Part F J Rail Rapid Transit. 2009;223(2):199–208.
[4] Veritas DN. Assessment of freight train derailment risk reduction measures: Part A Final
Report. Report No: BA000777/01; 2011.
[5] Brabie D, Andersson E. Dynamic simulation of derailments and its consequences. Veh Syst
Dyn. 2006;44(S1):652–662.
[6] Wasiwitono U, Zheng D, Chiu WK. How useful is track acceleration for monitoring impact
loads generated by wheel defects? Proceedings of 5th Australasian Congress on Applied
Mechanics; Brisbane, Australia, 2007, 10–12 December.
[7] Wu X, Chi M, Gao H, et al. Post-derailment dynamic behaviour of railway vehicles travel-
ling on a railway bridge during an earthquake. Proc IMechE Part F: J Rail and Rapid Transit.
2014;0(0):1–22.
[8] Wu X, Chi M, Gao H, et al. The study of post-derailment measures to limit the extent of a
derailment. Proc Inst Mech Eng, Part F: J Rail Rapid Transit. 2016;230(1):64–76.
[9] Tarawneh C, Aranda JA, Hernandez VV, et al. An analysis of the efficacy of wayside hot-box
detector data. Proceedings of Joint rail Conference; Pittsburgh, PA, USA, 2018, April 18–20.
[10] Garip G, Ustoglu I, Mumcu TV, et al. Hot box detection system design for railway vehicle safety.
Recent Advances in Financial Planning and Product Development. 2014;17:31–36.
[11] Lunys O, Dailydka S, Bureika G. Investigation on features and tendencies of axle-box heating.
Transp Probl. 2017;10(1):105–114.
[12] Asset Management Consulting Limited (AMCL). Monitoring of wheel impact loads. London:
Rail Safety and Standards Board Ltd.; 2012.
[13] Liu Y, Qian Q, Liu F, et al. Wayside acoustic fault diagnosis of train wheel bearing based
on Doppler effect correction and fault-relevant information enhancement. Adv Mech Eng.
2017;9(11). doi:10.1177/1687814017732676.
18 G. DIANA ET AL.
[14] Amini A, Entezami M, Papaelias M. Onboard detection of railway axle bearing defects using
envelope analysis of high frequency acoustic emission signals. Case Stud Nondestr Test Eval.
2016;6:8–16.
[15] Macucci M, Di Pascoli S, Marconcini P, et al. Derailment detection and data collection in freight
trains, based on a wireless sensor network. IEEE Trans Instrum Meas. 2016;65(9):1977–1987.
[16] Paddison JE, Bračković J, Deuter M. Derailment detection and emergency brake systems for
freight wagons. Proc. of the third international conference on railway technology: research,
development and maintenance; Cagliari, Italy, 2016, April 5-8.
[17] European Railway Agency. Recommendation on the provision proposed by the RID committee
of Experts requiring the use of Derailment Detection Devices. ERA/REC/01-2009/SAF, 2009.
[18] Cii S, Sabbioni E, Tarsitano D, et al. Wireless sensor nodes for diagnostics and monitoring of
freight vehicles. Proceedings of Railways; Sitges, Spain, 2018, September 3–7.
[19] Hung C, Suda Y, Aki M, et al. Study on detection of the early signs of derailment for railway
vehicles. Veh Syst Dyn. 2010;48(S1):451–466.
[20] Braghin F, Bruni S, Diana G. Experimental and numerical investigation on the derailment of a
railway wheelset with solid axle. Veh Syst Dyn. 2006;44(4):305–325.
[21] Sweet LM, Karmel A. Evaluation of time-duration dependent wheel load criteria for wheel-
climb derailment. J Dyn Syst Meas Control. 1981;103:219–227.
[22] Nadal MJ. Theorie de la stabilite des locomotives – Part II. Movement de Lacet. Annales des
Mines. 1896;10:232–255.
[23] Boronenko Y, Orlova A, Iofan A, et al. Effects that appear during the derailment of one wheelset
in the freight wagon: simulation and testing. Veh Syst Dyn. 2006;44(S1):663–668.
[24] Guo L, Wang K, Lin J, et al. Study of the post-derailment safety measures on low-speed
derailment tests. Veh Syst Dyn. 2016;54(7):943–962.
[25] Connor P. Track basics. The Rail Technical Website-Track basics; 2017. http://www.railway-
technical.com/archive/
[26] Specialised Force. Aldon 4014 series SaberTooth portable derails DataSheet. https://www.
specialisedforce.com.au.
[27] Specialised Force. Aldon 4014 series SideKick derail booster DataSheet. https://www.
specialisedforce.com.au.
[28] Specialised Force. Aldon 4114 series wheel shovers DataSheet. https://www.
specialisedforce.
com.au.