Being a magazine writer can be a noisy occupation sometimes, and ear defenders don’t seem to help a great deal. As a far greater author once said, it’s the sound of those deadlines whooshing past!
But then, railway engineering isn’t exactly known for being quiet either. That’s not to say, of course, that noise can’t be useful. On the contrary, the sound of working machinery can be highly complex and subtleties hidden within it can provide a great deal of useful information.
An Australian company, Track IQ, part of Wabtec Control Systems Pty Ltd (formally Trackside Intelligence Pty Ltd), has applied acoustic technology, together with some rather clever computer algorithms, to create a system that is delivering great benefits to the global rail industry.
Wheelsets and their axle journal bearings are expensive ‘consumables’ on rail vehicles. Employing preventative maintenance regimes that make use of time or mileage drivers for the replacement for items that aren’t necessarily broken is wasteful and expensive. The objective, of course, is to avoid failures in service, which can be highly disruptive and costly through performance penalties or major incident repairs.
Axle journal bearings might typically have an L10 [L subscript 10] life expectancy (the life that 90 per cent of the bearings can be expected to reach or exceed) of 1.5-1.8 million miles, but the default maintenance measure is to replace them at half that mileage. The graph of failure vs. mileage is not linear so, at half L10, a failure rate of 3.6 per cent might be expected.
While early swap-out improves availability, it does not eliminate failures entirely. Bearings have a hard life and external factors such as wheel faults (flats, out-of-round, tread spalling), and damper failures can shorten their lives in unpredictable ways.
Moving towards bogie maintenance interventions based upon condition monitoring and trend monitoring makes much greater economic sense. However, the ability and confidence to safely extend maintenance and overhaul periodicities requires the gathering of accurate empirical information on wear rates and fault progression.
Analysing the acoustics produced by bearings enables the very early detection of defects and the trending of deterioration. Acoustic monitoring can be used without having to take the vehicles out of service and it allows bearing faults to be detected and monitored months, or even years, in advance of intervention.
Track IQ’s system, known as Rail Bearing Acoustic Monitoring (RailBAM®), measures the acoustic signature of wheel bearings as they pass trackside (wayside-mounted) microphone arrays. The acoustic signatures can be attributed to specific wheelsets by means of Automatic Vehicle Identification, making use of ISO/IEC 18000 (RFID – radio-frequency identification) tagging or similar such systems.
From the moment the system is operational, bearing data and defects are reported. Over a short period of time, the acoustic characteristics of every bearing operating on a route can be trended and reported upon where RFID is employed. The nature of the acoustic signature is dependent upon the bearing type and size and varies with the speed of the vehicle and the direction in which it is running.
Defects, such as roller or ring surface spalls, water etching and brinelling, cause structural responses of the bearing components, which radiate sound in a uniquely characteristic manner. Frequencies of distress are derived for the various bearings passing the sites and are stored in the system’s database.
The detected bearing signatures are compared to the fault frequencies and are then trended over multiple pass-bys. Even minute bearing defects can be detected and, because each bearing failure mode produces a different tell-tale acoustic signature, the exact nature of those faults can be determined. Customers receive this information daily via an email alert or they can access the system real-time to view live train data.
As each wheel passes the RailBAM equipment, its diameter and speed of travel is determined using optical technology. Complex computer processing of the acoustic signature takes account of the wheel rotational speed. Even changes in rpm due to braking or acceleration can be accounted for. The result is that defects can be accurately identified and assessed at any train speed between 20mph and 85mph.
Predicting the life expectancies of each and every bogie in a fleet can allow maintenance intervention to be precisely tailored. Because frequent readings are taken, the detection of a bearing/wheelset fault that is running away to failure will raise an alert, allowing the defect to be remedied during the next scheduled maintenance exam, without the need for reactive measures. A truly predictive maintenance regime allows facilities, material and resources to be pre-allocated in order to rectify defects without impacting on asset availability or reliability during revenue service.
Initially trialled at Earlswood on the Brighton main line in 2007, the first permanent European RailBAM site was installed two years later at Swaythling near Southampton. This site is still in use and provides coverage, principally, of the Siemens Desiro fleet of Class 444 and Class 450, operating with South West Trains, as well as the Bombardier Voyager and Super Voyage Class 220 and Class 221 operating on the CrossCountry route.
The deployment of RailBAM on Wessex was later expanded to include a site in Mortlake. Installed and operated directly by South West Trains, the Mortlake site monitors its Class 458 and Class 455 fleets.
The use of RailBAM has been instrumental in a programme undertaken by GWR to extend the mileage of its HST Mk3 coach wheel bearings. GWR operates a fleet of 119 Class 43 locomotives and 464 Mk3 coaches. Coach bogie overhaul periodicity was 750,000 miles, whereas the wheelsets were limited to 600,000 miles because of axle journal bearing life. Whilst technical changes were made to the bearing configuration to improve bearing life, the 600,000 mile limitation remained.
The RailBAM system installed at Kensal Green in 2013 monitors traffic into and out of Paddington Station and has been used both to monitor current bearing performance and to mitigate the risk of Mk3 coach bearing failure in service. It has enabled a risk-based assessment of detected faults to challenge the 600,000 mile wheelset overhaul periodicity and align it with the bogie overhauls.
In 2016, the RailBAM system correctly identified the very early onset of defects in nineteen bearings that were removed from service for analysis. Since the introduction of RailBAM, the fleet has had no positive hot axle box detection (HABD) alerts. As well as significantly reducing maintenance costs, this has reduced service delays and it means that train staff no longer have to be placed in a position of danger to diagnose hot bearings track-side.
There’s a similar story in Ireland. The Iarnród Éireann Hyundai Rotem 22000 Class ‘Intercity Railcar’ DMU fleet had an expected axle bearing life of three million kilometres (1.9 million miles), but was suffering premature failures after about one million kilometres.
Apart from the commercial impact of much lower than expected bearing life, this had also raised a significant fleet safety concern. These deteriorating components were not detected by the lineside HABD network although, during 2015, it was reported that a total of 48 HABD alarms were generated for all train fleets on the network. All of these proved to be false alarms and were highly disruptive to operations.
Three RailBAM systems were installed during 2015, together with three Wheel Condition Monitoring (WCM) systems (of which more below) and these provide coverage of the entire operational rolling-stock fleet. Since November 2015, there have been six axle-bearing alarms generated by the RailBAM system for the Hyundai Rotem fleet, which prompted the removal from service and replacement of the identified axle bearings. In all cases, the bearings were found to be defective and had been detected weeks or months ahead of the bearing progressing to complete failure.
IE now has much greater confidence in its fleet operations as a result, and the system provides welcome additional safety containment while the root cause of the bearing failures is addressed. WCM is now being used to plan wheel turning and consideration is being given to drive bogie overhaul plans using RailBAM data.
Track IQ was acquired by Wabtec Corporation in October 2015. This has been a highly positive move for both companies, giving Track IQ greater access to new markets and Wabtec a diversification into the data services sector and an important segment of the wayside condition monitoring market.
In recent years, the RailBAM system has undergone significant technical improvements. Miniaturisation and full digitisation of the processing equipment means that the size and weight of the trackside equipment has been greatly reduced. Importantly, an ‘Inboard’ version has been developed, which has the microphone arrays clamped to the sleepers within the four-foot. This application is designed to monitor inboard axle journal bearings, final drive units and even gearboxes. It has been made possible by the use of digital microphones that can withstand diverse environments and can cope with a much greater amplitude range than previously. The new microphone arrays are smaller, more rugged and can withstand the high noise levels generated by engines and slipstream turbulence, whilst still being able to detect that all-important subtle signature of a bearing in distress.
Often deployed alongside RailBAM is Track IQ’s Wheel Condition Monitor (WCM®) system. Solid-state sensors are clamped to the rails and are used to measure the rail deflection and vibration as wheelsets pass. Defects such as wheel flats, spalling and out-of-roundness are automatically assessed.
The WCM is a hybrid system using both accelerometers and strain gauges to give 100 per cent wheel surface coverage and to resolve multiple defects on a wheel. This system also measures the loading of wheelsets and alerts can be raised for vehicle end-to-end load imbalances, side-to-side imbalances and even diagonal imbalances across vehicles and bogies. Such problems might occur because of unequal loading, or suspension failures. All measurements are taken at line speed.
RailBAM has so far been deployed in the UK at just three locations: Swaythling, Mortlake and Kensal Green since 2009 – Swaythling being the first permanent RailBAM site in Europe and the first dedicated to passenger rolling stock. WCM systems were later installed alongside RailBAM at Swaythling and Mortlake in 2012.
The safety and economic benefits are readily apparent, so this begs a question as to why has there been such a slow uptake in the UK. It’s a different story in Ireland and elsewhere in Europe, where RailBAM and WCM have been adopted more widely and quickly.
In the UK, we to talk a lot about the need to adopt new technology within the rail industry, but we don’t seem very good at actually doing it. Perhaps there should be a better means of incentivising the adoption of new equipment. In the UK, a Network Rail sponsor is required in order to introduce new technology onto the infrastructure and gain Product Acceptance. This applies even if the technology has proven usage elsewhere. At the same time, Group Standards leave little room for innovation. In the UK, such legislation is written around the use of HABD and onboard monitoring.
Some European countries have a more open approach. In France or Germany, for example, if new equipment doesn’t introduce new risks and cannot fail in a way that will adversely affect the existing infrastructure, such as the integrity of the track or signalling, it is deemed to be ‘off rail’ equipment and is not subject to the full suite of legislation that would otherwise apply.
Nicholas Kay is business development manager, Europe for Track IQ and he’s passionate about safety and maintenance cost reduction. He asks a rhetorical question: “Has the benchmark of ALARP (as low as reasonably practicable) been redefined by using acoustic wayside monitoring?” He expands on this by saying: “The rail industry is, to some degree, complacent, accepting of a traditional bearing maintenance regime at half-L10 life. The introduction of predictive monitoring systems should allow us to revisit past assumptions.” In other words, he is advocating a move away from prescriptive maintenance towards safer and more economical predictive maintenance techniques, working smarter, not harder.
Nicholas is highly upbeat about the take-up of RailBAM, irrespective of the potential UK market. So he should be, with strong uptake in France and Norway in recent years, plus strong growth in the domestic Heavy Haul market in Australia, as well as volume sales (over 40 systems) in China and (over 20 systems) in India. The future is looking positive for Track IQ.
The RailBAM site at Kensal Green sees over 200 trains per day into and out of Paddington. It is the busiest RailBAM system Track IQ support across the 14 countries on six continents where systems currently reside. The three UK sites are monitoring collectively over 74,000 tagged bearings. Other bearings pass by the systems and are monitored but are not on tagged vehicles. This means they are not reported to owners, operators or maintainers, which could be considered wasted data, or potentially even a safety risk. To date, the three UK sites have seen well over 46,000,000 bearing pass-bys since starting operation.
After a slow start, it seems highly likely that acoustic bearing monitoring will become more widely deployed within the UK and Europe. Certainly, as we have seen, it is justified in terms of cost benefits and increased safety. Clearly, this is something that the UK rail industry should be making more of a noise about.
This article was written by Stuart Marsh.