With the ever-present focus on putting passengers first, train reliability has to be high up the priority list to achieve this. Modern rolling stock is being equipped with many electronic systems that bring their own maintenance and reliability challenges.
One fundamental element is nonetheless the train wheel and the axle journal bearing upon which it rotates. Should this bearing catastrophically fail, then the train is likely to be declared a failure and to be taken out of service for an expensive repair. At worst case, the train could derail, with the risk of this being a major accident.
The monitoring of axle journal bearings is therefore an important safety precaution. Equipment to detect hot axle boxes (HABD) has been available since the 1960s and has been refined in subsequent decades. This technology, whilst effective, has reliability issues and offers only reactive alarms when a failure has occurred. HABD systems are not compatible with third-rail electrification infrastructure and mainly target the major freight routes, where derailment is known to be a higher risk.
Alternative acoustic technology, originating from Australia in the late 1990s and trialled in the UK in 2007, is now universally available. RailBAM (Rail Bearing Acoustic Monitor), produced by Track IQ, a Wabtec company, is well established in this technology, as well as providing other trackside monitoring systems in the UK. At a recent meeting of the London & SE section of the IRSE, the RailBAM system was explained, just proving that signal engineers are eager to learn about technology in all rail disciplines!
An introductory article on the RailBAM system appeared in issue 151 (May 2017), written by Stuart Marsh. This explained how the technology was developed by the Australian company Trackside Intelligence, which subsequently became branded Track IQ and was taken over by Wabtec in 2015, enabling a larger penetration into worldwide rail markets.
All bearings emit noise and that noise will vary according to the condition of the rolling surfaces within it. The RailBAM system captures the sound emitted from every axle journal passing a trackside monitoring site to check for defects developing on the rolling surfaces. In the event of a defect detection, an alert is sent to the rolling-stock maintainer, thus allowing remedial work to be planned without impacting on the day-to-day train service.
Such predictive advice allows defects to be fixed long before anything more serious happens whilst in traffic.
The principal units are two acoustic cabinets mounted trackside and a third enclosure away from the track that houses all the necessary electronics, communications and power equipment for operation of the system. Wabtec’s FleetOne web-based server allows the rolling-stock maintainer to access its own fleet data and create email alarms and alerts to suit its business.
The acoustic cabinets incorporate a motorised shutter that opens upon detecting the approach of a train, thus exposing the array of microphones to the side of the train. Ideally, a cabinet will be placed on either side of the track so as to collect sound from bearings on both sides of an axle in one pass, and this is typically the arrangement on single-track lines. On double or multiple-track railways, there is limited space between the running lines, so the system is split across adjacent tracks, allowing complete coverage to be achieved over a return trip.
Identifying the train is vitally important, which is achieved by fitting the train with RFID (Radio Frequency Identification) tags. Just how many tags are required for each train or vehicle needs a pragmatic consideration. A fixed-formation diesel or electric multiple unit will probably only require two tags, one associated with each cab end. On a passenger train, where coaches may be swapped around, a tag is needed on every vehicle. This includes the high-speed train (HST) fleet. On freight trains, a tag is needed on the locomotive(s) and every wagon (or fixed wagon pair or triple) if robust bearing monitoring is to be achieved. Tags are passive units programmed with the identity of that particular piece of rolling stock and are powered by the trackside reader as the train passes.
The RailBAM system has been successfully operating for over 20 years. Its flexible and robust design enables installations in the extreme heat of the UAE and the icy cold of North America. Power can be taken from the local electricity provider or solar battery supplies. Connection to the internet is either via wired networks (often the railway internal network), cellular radio, or even satellite communications in extremely remote locations.
The principle of operation is that, as a train passes the acoustics sensors, the sound profile of each axle journal bearing is captured and stored. The acoustic reading is then linked to the train tag data, the pass-by date and the time. Knowing the train identity allows the FleetOne system to compile the train consist and associate each dataset with the correct wheel position and axle journal bearing. Repeat pass-bys are recorded to provide updated data for each bearing, thus allowing the data to be used to observe any worsening condition and trigger alerts while also enabling analysis of behaviour trends and patterns.
With the acoustic system being sensitive enough to detect even small defects in the bearing, it is then typically over 100,000km before the defect would mature into a service-affecting failure. This means that, generally, no action need be taken until the next routine maintenance visit.
There are some constraints as to where and how the trackside sensor units should be positioned. For optimum results, they should be on straight track which is in good condition – if any ‘sleeper pumping’ is happening, this can cause noise and confuse the readings. The site also needs to be where normal train speeds are between 13 and 80mph. This does not present a challenge on high-speed lines, as even the fastest trains have to slow as they approach stations or junctions, which are usually ideal locations for the trackside readers.
The types of defect that are detected include:
- Rolling surface defects, including wheel flats and wheel roughness;
- Roller bearing wear, including for an individual roller;
- Cup/cone spalls (flakes of metal broken off from the bearing components);
- Multiple spalls;
- Extended spalls.
Since the RailBAM system is aimed at discovering developing defects rather than catastrophically damaged components, a root-cause history can be investigated and built up, which in turn will lead to the implementation of bearing improvements and a reduction in the bearing attrition rate.
As would be expected, the RailBAM system is continually evolving and improving. The data from over 100 RailBAM systems worldwide is hosted on Track IQ’s own data server in Adelaide (Australia), allowing the performance to be monitored daily and improvements deployed globally.
UK performance and statistics
Whilst the majority of systems have been installed overseas, the first UK application was a limited trial in 2007 on Southern at Three Bridges, targeting the Class 377 fleet. Although successful, the trial was not expanded.
In 2009, Siemens installed the first RailBAM system in the UK and the first globally to target only passenger stock. This was on the South West Trains network and monitored Class 444 and 450 units with the trackside unit installed at Swaythling, between Southampton and Eastleigh. The system captures data for the whole Siemens fleet as well as other trains using the route.
A second trackside unit was later installed at Mortlake with the fitting of tags to more of the SWT fleets, thus enabling all units on what is now South Western Railway to be monitored.
Noting the results, First Great Western installed a trackside unit at Kensal Green, near to North Pole depot, and began equipping its HST fleet with tags. This was then expanded to take in other classes of FGW trains.
As one would expect, the number of axles monitored by the three RailBAM installations is counted in the millions as the statistics show:
- More than 265 million bearing data sets reads in total;
- Equates to an average of 35 million bearing data sets/year;
- 8,600 collective days of operation across the three sites;
- Monitoring of approximately 5,000 tagged assets;
- Successful identification of several hundred defective bearings.
The overall system has achieved 99.99 per cent reliability and more fleets are being tagged as franchises are renewed.
The business case
The whole justification for this technology is to allow predictive rather than reactive maintenance. This yields huge savings, in the order of millions of pounds when taking into account the material optimisation as well as the penalties imposed when rolling stock failures occur.
The main beneficiary is the train maintenance provider, be this a train operator (TOC) or a leasing company (ROSCO). In most cases, the upfront investment is borne by the maintaining organisation, which seeks the approval of Network Rail to allow Track IQ to install the lineside units. With a typical system installation costing less than £300,000, the benefit for the rolling stock maintainer and the infrastructure provider is extremely strong, whilst also improving safety.
One question is whether the RailBAM system can replace HABDs, of which there are 228 on Network Rail. Whilst the acoustic equipment can detect defects well in advance of hot axle box systems, thus redefining what is ALARP (as low as reasonably possible) for bearing monitoring, it is primarily designed for trains with the more common roller bearings. The effectiveness of acoustic monitoring on shell bearings, of which there are a number remaining in service, is still being evaluated.
So, for the present, there are no plans to remove the HABDs. However, since the detection of axle journal bearing problems is made much earlier and more reliably, it is estimated by the RSSB and Network Rail that as few as 35 trackside acoustic sites are needed to provide national coverage.
With all elements of rolling stock monitoring now becoming the norm, a holistic approach to link RailBAM into other systems, such as FleetOne, is under consideration and is all part of the goal to improve overall train and infrastructure reliability.
Thanks to Nicholas Kay of Track IQ and Paul Baker of Bakerail Services for taking time to explain the system to a bunch of signal engineers!
This article first appeared in Issue 177 of Rail Engineer, Aug/Sep 2019.