1.4A. Francois Burger
1.4A. Francois Burger
1.4A. Francois Burger
Broken rails are considered the most common cause of derailments by some [Lui et al. 2008].
Figure 1. Frequency and severity of class 1 main line freight train derailments, 2001 - 2010 (Lui et
al 2008)
• Track circuits – Track circuits provide a varying degree of broken rail detection
depending on the configuration employed. In cases where only one rail is used as a
“Signal Rail”, broken rail detection is not guaranteed on the other rail especially where
this is employed as an electric traction return conductor.
These traditional methods to detect rail breaks are very costly, require expensive equipment
and manpower and are not always reliable. Most of these methods are intermittent resulting
in extended periods during which a rail break can occur and be undetected.
In India a New Delhi newspaper article mentioned a few years ago that as many as 400 rail
inspectors are killed annually during visual inspections at night.
In South Africa a high incidence of rail breaks on heavy haul tracks led to the initiation of
ultrasonic broken rail detection technology development in 1996. The first large scale roll out
of the system was completed in June 2014 on the Transnet Freight Rail ore line (OREX) and
has since detected 7 breaks, potentially contributing significantly towards Transnet Freight
Rail cost savings. It also facilitated the identification of numerous large rail flaws which
could then be repaired timeously before it resulted in breaks. The system is reported to be
very stable requiring only a few maintainers over the 842 kilometre installation. This system
operates by transmitting guided waves between permanently installed transmit and receive
transducers spaced approximately 900 meters apart. The version 4 system has been proven to
reliably detect rail breaks.
During 2016, two test installations were installed on Indian Railways track sections. After
initial teething trouble, and constant disruptions due to theft and vandalism, the installation
on the 60kg rail section is now operating very stable and false alarm free.
In a quest to improve system performance, thereby reducing cost per kilometre, research and
development effort to improve the system is underway. During this development, sound
propagation research resulted in an optimised ultrasonic transducer. This together with the
introduction of advanced signal processing techniques resulted in a large improvement of
system sensitivity, and drastically improved operating turn-around speed making the system
compatible with high traffic density metro systems. Sound propagation research also
confirmed the feasibility of using reflectometry techniques to identify and monitor all larger
rail flaws over a distance of at least 1 kilometre both sides of one unit. It can also pinpoint the
position of a crack or break with high accuracy. The reflectometry technique paves the way to
rail condition monitoring.
2. SYSTEM DESCRIPTION
2.1. General
The Ultrasonic Broken Rail Detector system operates on a simple pitch–catch principle, see
figure 2. The rail is excited with an ultrasonic wave at one point and monitored some distance
away for the arrival of this wave. Should the expected wave (signal) not arrive in a certain
time period, or not conform to certain criteria, an alarm will be issued.
The system makes use of permanently installed piezoelectric transducers at alternate transmit
and receive stations along the length of the rail. The transmit stations operate autonomously
and transmit bursts in both directions along the rail. The burst length and repetition rate is
used by a receiver to distinguish the signals received from the left or right rail, and the up or
down side.
Development of the system started in 1997 and the layout has remained unchanged while
numerous improvements to components have been required to achieve reliable operation.
Long range transmission is achieved by using a mode of propagation that has the most of the
energy confined to the head of the rail, and very little motion in the foot of the rail. See the
dispersion curves below where the operating point is marked by the red dot. The desired
mode of propagation is achieved by selecting a specific frequency and transducer mounting
position and orientation.
Figure 4. Wave number - frequency dispersion curves for UIC60 rail
Ultrasonic transducers are clamped to the rail using heavy duty spring clamps. The spring
loading keeps the transducer firmly in place and properly located in the rail profile under the
head and against the rail web. The design of the spring clamp evolved to the point where it is
now fully reliable and maintains the required contact force between the transducer and rail
permanently without any requirement for adjustment.
Filtering and a number of logic operators are applied to ensure that only valid signals are
recognized and to detect the presence of trains in the vicinity of the receiver. This is required
to ensure false-alarm free operation during the passage of trains, sometimes kilometres long,
which mask the interrogation burst sequences for long periods.
Although the principle and design of the system is quite simple, achieving reliable operation
under the traction current- and environmental conditions prevailing along the track was quite
a feat. Challenges faced with the Ore Line Ultrasonic Broken Rail Detector system are as
follows:
Safe-guarding of GPRS modules for being branded “bad client” by the network and
then being ignored.
Lightning and surge protection; The Ore Line track runs through notorious lightning
prone areas in the north western parts of the country. Protection of sensitive electronics
from lightning induced surges proved to be quite a feat. Whilst previous similar
installations were completely decimated by lightning, designs and measures taken were
extremely effective with not a single report of surge damage during the operational life
of the system at the Ore Line.
Transducer cable design and testing requirements to ensure that cables are reliable
under high voltage operation (1300Vp transmitter drive burst).
System design to facilitate diagnostics; equipment failures must not manifest as broken
rail alarms.
Electronics design reliability; electronics have to survive under severe vibration and
extreme temperature conditions. Design analysis to prevent overstressing of
components and environmental stress screening tests (ESS) were performed.
The Version 4 Ultrasonic Broken Rail Detector system installation on the Ore Line in South
Africa was completed during June 2014. This is an 841km long single track rail with passing
loops every 40 kilometres. 931 Ultrasonic Broken Rail Detector units were installed spaced
on average at 890 meters. GPRS technology is used to relay alarms to the Centralized Control
Centre (CTC) at Saldanha Bay on the west coast of South Africa. Roughly 50% of receiver
units are located in “dark areas” with no cellular phone network coverage. At these locations
digital radios are used and communicate alarms to a number of “radio high sites” from where
alarm messages are relayed to the CTC.
Figure 6. Electronic module (left) and solarpanel mounting on mast pole (right)
In the absence of mains power along the OREX track, solar power supplies are used. Much
effort was spent on designs to facilitate installation of solar panels and antennas high on the
mast under live conditions, and to prevent the theft of components. Wayside enclosure design
with cross flow air vents and shade roofs keep the internal temperature low. This is essential
to preserve battery life.
Performance
The Ore Line Ultrasonic Broken Rail Detector installation has reported quite a number of
breaks and major flaws. During 2016 and 2017 alone, 14 breaks were detected. This is
potentially a very big cost saving considering that derailment cost may be as much as R80
million per event, not including the loss of income due to halted operations.
As flaws develop and grow, signal propagation is gradually diminished to the point where the
received signal level drops below the detection threshold. Such a case will manifest as a
broken rail alarm but will be intermittent in most cases. Ultrasonic and visual inspection of
the affected section between a transmitter and receiver unit has in almost all the so called
“false alarm” cases revealed a major, critically sized crack or flaw. Such action further
enhances the utility of the system since these cracks are identified and repaired before it
results in a break with potential catastrophic consequences should a derailment arise from it.
System stability in terms of not issuing intermittent false alarms is now reported as
being excellent.
The latest availability figure is approximately 99%. This is disregarding removed and
stolen equipment.
The ultrasonic broken rail detector system was installed in India on the NCR close to
Allahabad on UIC60 rail between mast pole 835-11 and 859-23. First commissioning was
completed in August 2016 but numerous problems caused poor performance initially. These
were as follows:
These problems were all resolved by end of February 2017 and the cause of false alarms at
these units positively identified and repaired during the Feb maintenance visit. These were all
relative small issues for example:
Figure 10. NCR UBRD unit (left) and alarm terminal display (right)
The Allahabad installation has been operating 100% stable without false alarms since end of
February. A second visit during Feb 2018 revealed quite a number of theft and vandalism
related issues and steps to further harden the system are now being considered.Personnel of
Pioneer Global Enterprises Private Limited, the India UBRD agent, have been trained and are
now proficient in performing maintaining of the system. Success of the Allahabad installation
can also be attributed to extensive measures of the agent to maintain and guarding the system
and to replace damaged and stolen equipment.
The Allahabad installation has proven that the ultrasonic broken rail detector system is
compatible with India rail conditions and can be successfully rolled out on a large scale. It
was proven that the system can operate fast enough to limit alarm notification delays to less
than 10 minutes. Integration of the alarm system with India cellular networks and fixed line
internet was executed successfully, and the UBRD system at Allahabad is now fully
operational under India local conditions and with India communication infrastructure.
This exercise was also instrumental in the identification of design changes to customise the
system to India Railways specific requirements and to make components even more robust
and theft resistant.
2.2.3. Northern Rail Roorkee Test Section
The result was that even after substantially reducing unit separation distances in troublesome
areas, the system was still not able to perform stable under low temperature conditions since
signal propagation through these rail sections was found to be insufficient. Rail Propagation
modelling was performed and shows that a higher frequency is required for 52kg rail. The
current system is therefore not suitable for deployment on 52kg rail.
Notwithstanding the signal strength issue at the Roorkee test installation, 6 actual rail breaks
and two maintenance breaks were successfully reported.
False alarms previously encountered were proven to be due to specific faults which were
caused by vandalism and theft. It is clear that the UBRD system is suitable for deployment in
Indian. Further modifications to harden it against theft and vandalism are planned and should
result in the system being very stable and effective.
3. RECENT UPGRADES
After securing government funding from the Department of Science and Technology, several
upgrades were done, mainly to improve operating distance and speed (interrogation interval).
The resulting version 5 system uses modern digital signal processing (DSP) technology.
Modern DSP integrated circuit current consumption has been reduced to a point where it can
be used in solar powered systems.
The signal processing gain achieved with the digital signal processing technique, together
with improved ultrasonic transducer performance enables operation at twice the distance
previously possible. A single rail interrogation operation which previously took 20 seconds to
complete has now been reduced to less than 1 second.
The incorporation of radio modules in the transmitter and receiver circuits now enables
synchronised operation. Previously, false alarms could stem from situations where a number
of successive transmitter burst sequences by chance overlap with passing trains, preventing
receivers from detecting the interrogation pulses. Version 5 receivers request burst sequences
from adjacent transmitters once trains clear the area, ensuring that the rails are interrogated
during the quiet period immediately after trains have passed. This drastically increases
detection probability and further speeds up operating speed. This feature makes the version 5
system suitable for high traffic density locations such as metro systems.
The design of the transducer was based on research into the propagation and transduction of
guided wave ultrasound in rails. Measurements of propagation of ultrasound in rails were
performed using a scanning laser vibrometer and a technique was developed to estimate the
attenuation of individual modes of propagation and the ability of transducers to excite these
modes (Loveday & Long 2015).
Figure 11. Measured ultrasonic propagation 400 meters from the transducer
A numerical modelling technique was developed which combines three - dimensional finite
element modelling (of the piezoelectric transducer) with semi-analytical finite element
modelling of the wave propagation in the rail (Loveday 2008).
Figure 12. Small size of version 5 transducer (left) compared to the version 4 transducer
The modelling and measurement tools were used to optimize a small transducer, which
targets a selected mode of propagation and produces an order of magnitude greater excitation
while being more sensitive on receiving than the transducer used in the version 4 system. A
procedure to bond the version 5 ultrasonic transducer to the rail using epoxy was developed
and has now been proven to be reliable. Test installation transducers fitted with epoxy are
still intact and fully operational after 18 months.
While UBRD Version 5 offers attractive improvements over the Version 4 system it remains
primarily a broken rail detector. Research is currently being performed to detect defects such
as cracks prior to complete breakage of the rail. If successful, this research could result in a
system that detects, locates and possibly monitors the growth of cracks and thereby enables
condition based maintenance. It is hoped that in future the Version 6 system will be able to
detect cracks and will be a broken rail prevention system.
For such a system to be economically feasible it is necessary that defects can be detected at a
reasonable range and that different types of defects in different parts of the rail cross-section
can be detected without the use of an excessive number of transducers at each measurement
station. As the system would monitor the rail in near real time it is not necessary that very
small defects be detected a number of months before they would lead to a broken rail.
Pulse – echo measurements have been performed using the transducers developed for the
Version 5 system, which are very effective for exciting and sensing a mode of propagation
with energy confined in the head of the rail. These measurements showed that it is possible
to detect reflections from aluminothermic welds at a distance of 1000m from the transducer.
A numerical modelling technique (Benmeddour et al. 2011) has been implemented and used
to predict the reflections that would occur from transverse defects in the rail head and from
aluminothermic weld caps. Based on this comparison it is expected that transverse defects
will be detectable at long range prior to these defects resulting in broken rails.
Detection of defects occurring in the rail web and foot requires the use of different modes of
wave propagation. A mode that is confined to the web has been identified and a transducer
has been optimised to excite this mode of propagation. Achieving wave propagation and
defect detection at a useful range in the foot of the rail is significantly more difficult. Energy
propagating in the rail foot is quickly absorbed by the polymer pads placed between the rail
and the concrete sleepers and could also be reflected by the rail clips, which tie the rail to the
sleepers. Developing a method capable of detecting even relatively large cracks in the foot
of the rail at a useful range remains a challenging but important task.
5. CONCLUSIONS
Ultrasonic broken rail detector technology has been rolled out successfully on South African
heavy haul tracks and has proven its ability to reliably operate under harsh traction current
and lightning prone conditions. The number of broken rails successfully detected and
reported by the version 4 system installed on the Ore Line potentially saved the operator a
huge sum of money, keeping in mind that the cost of the installation was less than the direct
cost of one derailment. Maintainability of the system was also proven over operation of the
system over 2 years and 8 months.
The Allahabad installation in India has proven that the ultrasonic broken rail detector system
is compatible with India rail conditions, has been fully integrated with India infrastructure,
and hence can be successfully rolled out on a large scale.
Excellent advances were made during research on pulse-echo techniques, paving the way for
rail condition monitoring and the management of rail maintenance using automated real time
crack and flaw inspection techniques.
6. REFERENCES
Liu, X. Saat. M. & Barkan. P. 2008. Analysis of Causes of Major Train Derailment and Their
Effect on Accident Rates
Loveday, P.W. & Long, C.S. 2015. Laser vibrometer measurement of guided wave modes in
rail track. Ultrasonics 57: 209-217.