Ultrasonic Broken Rail Detector

Technical background

1 Scope

This document presents technical information on the Ultrasonic broken rail detector. The
information covers the basic operation, the technology employed, the application, installation and
maintenance of the system.

2 Background

The Sishen-Saldanha iron ore export railway line suffers from unpredictable rail breaks causing a
great loss annually to Spoornet. Rail breaks are however a common phenomena in all railways over
the world. This can cause very costly derailments if not detected before the passage of a train. Small
amounts of impurities trapped in the metal structure of the rail are potential points of rail defect
development. The development of these defects is associated with metal fatigue, which is a function
of the loading applied to the rail. Many of the minute irregularities of metal structures in the rail are
a given factor of rail manufacturing and will most probably never cause an actual rail break. These
irregularities are practically undetectable when the rail is installed and therefore the only effective
method to prevent breaks from occurring is to monitor the rail for growth of rail defects.

The most common defect is a crack that develops in the crown of the rail, at an angle of
approximately 70° relative to the horizontal axis. It is known as a kidney defect, because of its
distinctive shape. Various methods have been developed to detect the defects and the most
commonly used is an ultrasonic system, which transmits an acoustic signal into the rail and
measures the reflection of the waves from the defect. This is a very effective method, but requires
scanning of the complete line on a regular basis and is thus very labour intensive. Motorised
vehicles, equipped with sophisticated ultrasonic measurement equipment, are employed to increase
the frequency and efficiency of defect detection.

Although the design of some signalling systems are such that it often detects clean rail breaks this is
not a guaranteed meganism in the signalling configuration employed in South Africa and other part
of the world. With the move to transmission based signalling in the future, this advantage will
however also disappear.

Because of these difficulties and the unavailability of a suitable solution, Spoornet initiated the
development of an acoustic broken rail detector, which has the capability of detecting a broken rail
RailSonic Page 1
almost continuously without human intervention. This is the first successful development of its kind
in the world and has been installed on a trail basis on the Orex line, the CoalLink heavy haul line
and the TTCI test centre in North America. The technical evaluation of the system on the Spoornet
lines was completed in November 2002. The development was supported by the UIC and formed
part of the Joint Research Project No. 2.

3 Basic Principle of Operation

The system basically consists of a transmitter unit wired to a transducer mounted on the rail to
introduce an ultrasonic signal into the rail at a given point (see Figure 1). Each receiver will
monitor a 3.5 km stretch of rails for breaks. A receiver station thus consists of two transmitters on
either side of the receiver installed at a maximum distance of 1.75 km. Signal insertion (Transmit
stations) and monitoring (Receive stations) equipment are interleaved on the rail at a distance of
1.75 km apart, and operates on both rails. A high-energy ultrasonic pulse is bi-directionally inserted
into both rails via ultrasonic transducers at a programmable interval by the transmit electronics.

R ail
~

R eceiver T ran sm itter R eceiver

Figure 1. Block diagram of the Ultrasonic Broken Rail Detector System (UBRD)

These signals propagate along the rail and are detected at the receive stations. Adjacent transmit
stations insert the same transmit frequency into the rail, but at different pulse rates to enable
receivers to establish the direction from which the signal originates. The receive station will monitor
both rails and will send the equipment status as well as alarm conditions to the central monitor
station via a digital interface. Five digital output bits are available to communicate equipment status
as well as alarm conditions from the local receiver to a master station. Due to the modular design it
is possible to monitor only portions of the line. The main components of the acoustic rail break
detection system are the ultrasonic transducers, the transmitter, the receiver and the digital interface.

RailSonic Page 2
3.1 Components of the acoustic rail break detector system

3.1.1 The ultrasonic transducer

The ultrasonic transducer is used to send and receive ultrasonic signals to and from the rail at an
optimised frequency. The head of the transducer could be shaped to fit the rail profile and a suitable
attachment method is incorporated. The head of the transducer is cast from marine grade aluminum
and the piezoceramic stack and back mass was housed inside a Ni-Cd plated canister sealed to the
head. A stainless steel connector with hermetic sealing is mounted in the housing so that no
moisture can penetrate the housing and reach the piezoceramics. The electrical isolation break
down voltage between the transducer and the rail is 2.5 kV.

3.1.2 The transmitter station

The transmitter consists of a power supply, signal generator, a power amplifier and a impedance
matching circuit. The signal generator generate a signal consisting of
five pulses at the transmitter frequency. The time spacing between
the pulses are settable to a to suite the application and the available
power. When operated from a solar supply, such as on the Orex line
one transmitter is set to 1 second while the second transmitter is set
to 1.5 seconds. This will enable the receiver to determine from which
side the signal arrived. The frequency at which the train of five
pulses is send is configurable. On the Orex line the default
Figure 3: Amplified signals
at the receiver. configuration of three pulse trains per 10-minute period is used. The
signal produced by the signal generator is amplified to a high voltage
by the power amplifier. This amplified signal is then applied to the acoustic transducer via an
impedance matching circuit.

3.1.3 The receiver station

The receiver station consists of an impedance matching circuit, a pre-amplifier, decoding

electronics and a power supply. A very small signal (approximately 5 uV) is received at the
ultrasonic transducer is send to the preamplifier via the impedance matching circuit. The
preamplifier amplifies the signal with an gain settable between
92 and 106 dB. The amplified signal is then send to the
decoding electronics to decode the signal and check its
validity. See figure 5 for signals at the receiver. If three
consecutive pulses with the correct time discrimination are
received system will indicate a no broken rail. If no valid
signal were received within a given period (ten minutes for the
Orex line) the system will indicate a broken rail. The status of
the four pieces of rail monitored by the receiver will be
indicated on the digital interface via opto-isolated bits. A firth
bit indicate the presence of a train within the section. No
broken rail detection is available while a train is present in a
monitored section. Figure 2: Experimental transducer.

RailSonic Page 3
4 Tests conducted

4.1 Trail installation on the Orex line

Approximately 190km of the OREX iron export line was equipped with the Ultrasonic Broken Rail
Break Detector (UBRD) system. This installation was to investigate detection reliability and the
false alarm, and other parameter which could influence the performance of the system. The
following paragraphs detail some of the investigation embarked on.

4.1.1 Methods of coupling to the rail

Initially the transducers were mounted on the rail by means of a casting purpose made for the UIC
60 rail profile. It was bolted through the web on the neutral axis at an angle to ensure good physical
contact with the crown of the rail. Epoxy glue is used between the casting and the rail to improve
contact. Tests were conducted to determine the consistency of acoustical signal transfer and to
investigate other method of coupling to the rail. This resulted in the current method of clamping the
transducer to the rail eliminating the need to drill the rail.

Figure 4. Investigations undertook on the Orex line.

4.1.2 Influence of rail temperature variation

It was observed that detection efficiency varied between night and daytime and therefore it was
expected that temperature had an influence in the attenuation of the signal. Tests were conducted
starting early in the morning and continued well into the day to obtain a relatively wide rail
temperature range.

4.1.3 Influence of section distance

Tests were conducted on different section lengths in order to quantify the attenuation as a function
section length. During these tests the receiver transducer was at a fixed position and the transmitter
was placed at varying distances making use of a repeatable clamping method.

4.1.4 Attenuation characteristics of geographically different sites

Since the system was installed at locations distributed over the total length of 800km is was possible
to compare received signal strength between different sections of the same length. The sections

RailSonic Page 4
were all scanned for rail defects making use of ultrasonic scanning to ensure that significant rail
defects were not present. All defects greater than 15mm were removed in order to compare signal
strengths between different sections. The results gave an indication of the variation in attenuation
between different sections of normal serviceable rail.

4.1.5 Variation of equipment characteristics (Transducers and transmitters)

The transfer of electrical to acoustic energy was investigated for randomly selected transducers to
determine the spread in equipment characteristics. This was performed with fixed transmitter and
receiver positions. The transducers were exchanged making use of a repeatable rail clamping
method. The results were used to quantify the variation between transducers.

4.1.6 Position of the receive transducer relative to the acoustic wave length
Since the wavelength of the acoustic signal is within the mounting resolution and given the fact that
more than one mode propagate through the rail, the possibility that received signal amplitude could
be influenced by the positioning of the receiver transducer relative to a signal wavelength was
investigated.

4.1.7 Detection algorithms

Different detection algorithms were investigated to determine the robustness in the presence of
electrical corona in the 50kV AC traction system, and the effects caused by rain storms . The error
probability of the detection algorithms was also simulated in the presence of noise and compared to
more sophisticated algorithms.

4.1.8 Influence of rail defects on the system

Tests were conducted on rail sections, which contained defects of varying size to determine the
effect. Once such a defect was identified the acoustic signal strength was measured before and after
the removal of the defect. Effects of rail temperature and rail stress were also investigated.

4.2 Final evaluation of the acoustic broken rail technology by Spoornet

The investigations undertook on the Orex line enabled the system design to be adapted to improve
performance and supported the definition of application limits of the system. With this completed
final evaluation of the system was performed at various test sites.

4.2.1 Spoornet evaluation on the OREX line

4.2.1.1 Evaluation setup

A standard configuration was proposed for the 4 selected test sites along the Orex Iron-Ore line.
The configuration included 2 transmitter sites, 1 repeater site and 1 receiver site. Figure 5 show the
configuration used. The test sites was installed at km’s 64, 205, 565 and 665 from Saldanha. The
rail status was monitored at the receivers for a period of two months. The changes of all the status
bits was recorded at the receiver sites.

RailSonic Page 5
Figure 5: Acoustic rail break detector evaluation setup

4.2.1.2 Evaluation process

The process was started by identifying 4 sites on the Orex line which represented worst case
conditions. The status changes of the sites as well as any physical activity at any of the sites were
reported to Spoornet Engineering and IMT. The logs were downloaded from the equipment during
site visits at the end of the evaluation period. Using the gathered information the analysis of the
results were completed and presented to Orex management. The evaluation process is shown in
Figure 6.

Figure 6: Evaluation process for the Acoustic rail break detector

RailSonic Page 6
4.2.1.3 Evaluation Results
The system was evaluated from May 2001 – June 2001. Minor problems were found with the
repeater setup. Alterations was made to make provision for the problems experienced with the
repeater. The system was then re-evaluated for a second period after alterations were made from
Aug 2001 – October 2001. The results from the evaluation were acceptable to the management of
Spoornet and the system accepted for rollout on the complete line.

4.2.2 UIC evaluation at TTCI in Pueblo, North America

An evaluation set was installed at TTCI, the AAR test facility in Pueblo, North America. The
evaluation ran for a period of six months from October 2001 to April 2002.

Figure 6 show the test setup at TTCI in Pueblo. Due to the short distance available the system was
installed with two transmitters transmitting from one side to the receiver at the signal house in order
to simulate the normal installation conditions. With this test the system was tested under conditions
not present in South Africa and the results was published in a report on the acoustic broken rail
detector drawn up by TTCI for the UIC. The test results from TTCI were very positive. The only
concern was that the system did work across temporary rail joints and is the subject of further
research.

Figure 7: Acoustic rail break detector setup used at TTCI

RailSonic Page 7
4.3 Evaluation of the system in a tunnel and double line sections

The system was also installed on the coal export line (CoalLink) in a tunnel and on a portion of
double railway line. The purpose of this installation was to evaluate the effect of concrete rail slabs
used in tunnels on the attenuation of the signal in the rails. Furthermore the amount of cross-talk
between two adjacent lines were investigated. The detail of these installations are shown in Figure
8.

5 Current configuration

5.1 Standard system configuration

The standard configuration employed in South Africa is shown in Figure 4. The receiver detect the
known signal at a distance of up to 1750 m (Configuration used by Spoornet) and indicate via the
rail status bits that the rail is in tact. As seen in figure 8, the acoustic signal is different for each
section of rail monitored, and indicated separately by the receiver. Each monitored rail has its own
status bit indicating the status of that rail (Ok or Broken). The fifth status bit indicates rail activity,
which occurs when a huge amount acoustic noise is detected at the receiver. Trains and people
working on the rail close to the receiver mainly cause this noise. By monitoring the rail status bits in
conjunction with the status bits of neighbouring receivers an algorithm is implemented to detect
component failures in the system and thus increase the integrity of the system. For example, if rails
3 and 4 fail at the same time, it is probably the receive transducer on that rail leg which has failed.

Figure 8: The acoustic rail break detector standard configuration

5.2 System with repeaters

In cases where communications is not available and the section of rail between a transmitter and
receive needs to be extended the receiver and transmitter may be combined to form a repeater unit.
This is a built in function of the units and it is automatically selected when a transmitter and
receiver is plugged into the same backboard. When a the signal from the transmitter (far right in
figure 9) is received by the receiver at the repeater site, the transmitter retransmits the same code

RailSonic Page 8
into the rail. This signal is then finally received at the receiver, shown on the left hand side of
Figure 9.

Rail

~ ~

Receiver Repeater Transmitter

Figure 9. Repeater configuration

This configuration has the advantage of protection on long sections but have the disadvantage of the
whole section being in-active when a train enter the section. If the system configuration is not
designed properly it can lead to false alarms.

RailSonic Page 9