Some of the most expensive consumables on a rail vehicle are its wheelsets and axle bearings which make up a significant part of an operator’s maintenance budget. Scrutiny of railway costs, triggered by the McNulty report, means that successful railway companies must achieve exceptional wheelset operational performance and at the same time maintain very high levels of safety performance.

In the rail engineer issue 93 (July 2012) we reported on the recent wheelset seminars held at the Institution of Mechanical Engineers (IMechE). Topics reviewed included the management of safety, an overview of standards, testing processes and an insight into current research. This is indeed vital and impressive work, but out on the real railway, practical problem solving can be just as important.

New profile

Designed, built and maintained by Alstom, the class 390 Pendolino sets were introduced in 2002. This hard-working 57-strong fleet (574 vehicles) clocks up 17 million train miles per year. The operator, Virgin Trains, works closely with Alstom on engineering issues and together they have worked collaboratively with Network Rail in solving wheel/rail interface problems.

Dr Mark Burstow, principal vehicle track dynamics engineer, Network Rail, has recently presented a paper at the IMechE on this very subject. He described how, after reports of class 390 rough riding in the Hilmorton area near Rugby, a joint investigation by Network Rail, Virgin Trains and Alstom was undertaken. Analysis of track geometry, rail and wheel profiles revealed no special problems at Hilmorton, but the collected data was used to calculate wheel conicity.

High conicity may be attributed to worn rails/wheels or tight track gauge and it can result in a greater tendency for wheelsets to hunt. Calculations by Dr Burstow confirmed that hunting might be expected at the Hilmorton site, but only from worn P8 profile wheels. Rail grinding was undertaken in order to reduce gauge corner contact and this resulted in the conicity for worn wheels being reduced enough for stability to be restored. The drop was not significant however, and to avoid a return to instability a longer term solution was required.

As an alternative to changing the rail profile the wheel profile can be altered. P12 was the chosen profile, as it has a lower initial conicity than the commonly used P8. Increased flange wear had resulted when P12 was tried on the class 450 outer suburban EMU fleet, but Dr Burstow anticipated that the curvature of the WCML would reduce this problem. P12 was originally developed in order to reduce rolling contact fatigue (RCF) without a significant reduction in steering ability.

Virgin and Alstom decided to trial P12 initially on one trainset, with some P8 profiles retained as a control. The P12 trial trainset ran to 522,000 miles without major problems and flange height grew at the same rate as for P8. Flange wear for P12 was found to be less than for P8 between 50,000 and 200,000 miles. Some P12 conicity values approached instability at as much as 500,000 miles – almost twice that typically achieved by P8 profiles.

To provide a larger sample size and to see how well the P12 performed on smaller diameter wheels, rather than the new full-size wheels used in the original trial trainset, the scope of the trial was extended to include five further trainsets. So successful has this trial been that a decision has now been taken to extend the P12 wheel profile to the majority of the Pendolino fleet, with a small number of trainsets remaining with P8 wheels as a control sample. The lower conicity and reduced RCF damage should allow significant extension of wheel life.


Class 91 locomotives were introduced as part of the East Coast Main Line modernisation and electrification programme of the late 1980s. These 4.7MW 140mph machines were built by GEC Alstom with BREL as a sub-contractor. They feature a low unsprung mass, having the traction motors mounted on the main frame rather than within the bogies. The target bogie overhaul period was set at 750,000 miles, but the complex transmission system and unconventional bogie design led to service problems.

Paul Sutherland, asset manager with Eversholt Rail, Tony Brown, head of engineering with East Coast, and John Simpson, Principal Engineer with Interfleet, outlined the problem-solving exercise – a story that spans 20-years, albeit with many years in hiatus.

On the class 91, a virtual traction centre is formed by the use of a Watt’s linkage, although the primary suspension system is conventional. Coil springs carry the vertical load and rolling rubber ring units provide lateral and yaw guidance. Timken 150mm roller bearing units with lithium grease lubrication (standard on BR at the time) had given excellent service in HSTs, Class 90s and freight applications, but this was not a dedicated high speed bearing. Failures started during 1994 after roughly 500,000 miles service, with blackening/oxidation of the grease, signs of overheating and fatigue spalling of the races.

The use of synthetic greases offered a way forward, giving lower base oil viscosity, lower operating temperature and better resistance to oxidation. Unfortunately the results were only partially successful, so new SKF bearings with Shell Nerita 2858 grease were then trialled. This exercise resulted in 12 failures within 10 months, so during 1994 a new Timken bearing was trialled along with Alvania 2760B grease. These bearings were proven on TGV trains and passed extensive rig testing, but in service on Class 91 the grease proved unstable and ran out of the bearings.


In 1997, Timken application engineers were consulted on a better solution. They offered the NG2 bearing which has fewer rollers (18 vs 23) and more steeply angled raceways. Used on ICE and TGV very high speed trains, these bearings would, if anything, be slightly over-engineered for the class 91 application. As a result, Timken NG2 bearings with Shell Nerita 2858 grease were trialled on three class 91 locos. The result of this trial was fatigue spalling after 170,000 miles, so they too were withdrawn from service.

Because there was no theoretical reason why the Timken bearings hadn’t given the desired results, it was at this time that possible bogie geometry issues began to be considered. An investigation revealed that the bogie frame pads were not flat and that there were significant variations in component tolerances. Ultimately, half of the class 91 bogie frames required re-machining. Bearings were meanwhile being changed out at half life (375,000 miles) with resultant high labour and materials costs and adverse effects on loco availability.

Lateral float

Timken NG2 bearings with Shell Nerita 2858 grease were giving excellent performance on ICE trains, with no fatigue failure or grease degradation even after one million kilometres. ICE bogies have coil spring primary suspension but a lateral stiffness of just 9MN/m as against 22MN/m on Class 91. Crucially, on ICE the bearings are free to float laterally in the axlebox and are not clamped. Accordingly, a Class 91 was fitted with four different wheelset configurations, all having floating axleboxes.

The trial was then extended to four locos with NG2 and Nerita 2858 during 2008. They ran successfully to 450,000km (two locos) and 550,000km (two locos). This showed that the problem had been with the bogie, not the Timken bearings. From early 2010 Timken NG2 bearings have been fitted at scheduled loco overhauls. Heavy maintenance periodicity remains at 425,000km, but a plan to restructure this has been initiated. Rather than being based on bearings (450,000km max) this will now be based around wheel life. The benefits are greater availability, simplicity and reduced costs.

Coordinated approach

Iain Nairne is fleet overhaul manager at Southern. His fleet comprises over 300 units made up of more than 1,100 vehicles. Each weekday 2,300 diagrams are operated, adding up to 33 million train miles per year, and yet Southern maintains a combined fleet availability of over 92%. Southern has one Atlas tandem wheel lathe, located at its Selhurst overhaul facility, with all fleets receiving wheel turning based on condition.

During his recent IMechE presentation, Iain described how the workload is optimised by turning between peak periods which avoids any units having to wait in a queue. Turning is mainly undertaken to remove wheel tread cavities and wheelflats, although it can also correct flange wear, tread wear and rollover.

Southern uses the WheelChex system to detect wheelset problems. WheelChex is a brand name for a Wheel Impact Load Detector (WILD) system developed by DeltaRail. The system has been used effectively by Southern and many other operators within the UK and overseas. At the WheelChex site at Salfords on the Brighton main line, about 175,000 Southern trains pass each year, making up 74% of the total fleet. Southern trains account for only 9% of the level 1 warnings generated, the balance being overwhelmingly due to freight operations.

In his presentation, Iain Nairne outlined how he has undertaken a case study on Southern’s class 377 Electrostar fleet which has highlighted the fact that a small but significant number of wheel bearings require premature exchange due to defects. Class 377 was the last fleet to utilise 120mm bearings, the new standard being 130mm. Iain has established that lateral load is the primary cause of defects, although electrical damage (traction return current) is also a factor. To detect this, train riders have travelled on every class 377 unit on a 28 day cycle, but using human perception to detect wheel bearing faults has proved difficult. It was even harder to pinpoint them to specific wheelsets, with half of the wheelsets removed having no defects.

Iain went on to describe how a complementary system to WheelChex, known as RailBAM®, can be used to detect wheel bearing defects and pinpoint them to specific wheelsets within a train consist. Australian company Trackside Intelligence Pty Ltd (Track IQ) developed the RailBAM (Rail Bearing Acoustic Monitor) system and is now working with Siemens to lead sales in the UK and Continental Europe. Using acoustic principles it is able to detect bearing defects up to 100,000 miles in advance of final bearing failure. RFID (radio-frequency identification) tags allow trains to be identified and defects to be monitored over time.

RailBAM was successfully trialled on Southern over a 5-month period in 2007, during which 24 bearing defects were reported, with just one false report. The mature system has since been successfully deployed at Swaythling (2009) and Mortlake (2011) on the Wessex Route to monitor the whole of the South West Trains fleet, as well as trains from other TOCs and FOCs passing the sites.

Weibull analysis

Chris Tait, fleet projects and contracts manager First ScotRail, has made use of a concept that might not be familiar to many of us – that of Weibull Analysis. Sometimes termed reliability life data analysis, it attempts to make predictions about the life of all products in a population. It does this by fitting a statistical distribution to the “life data” gained from a representative sample of units. Chris has successfully utilised the Weibull analysis technique to improve wheelset management on the First ScotRail class 170 DMU fleet.

ScotRail has four maintenance depots but just one wheel lathe, located at Shields depot, Glasgow. The class 170 fleet comprises 59 3-car sets – 41 Express units and 18 Suburban units. The class 170 express units average 178,000 miles a year while the suburban units cover about 132,000 miles. Both have a planned wheelset renewal periodicity of 715,000 miles and a planned re-profile cut every 140,000 miles. The wheels are designed to the standard P8 profile.

Wheel lathe data from Shields depot was used to determine the average number of days from known good wheel to damaged wheel. This was then converted into average mileage. Ten samples from each vehicle wheel position were used in the analysis. An average wear rate per millimetre was determined for each wheel position, allowing a wheel wear predictor to be established.

The Weibull characteristic life result for class 170 shows a clear distinction between driving vehicles and centre vehicles.

The results were compared to WheelChex wheel impact alerts from January 2010 to August 2011. These supported the findings of the Weibull analysis, demonstrating that the middle vehicle wheels exhibit fewer wheel impacts and therefore require less tyre turning than the outer vehicle’s wheels.

Chris Tait says, “Weibull and WheelChex data analysis has identified that a staggered tyre turning regime is more applicable for the class 170 fleet. The review of tyre turning periodicity using Weibull analysis has identified the current 140,000 miles is no longer optimal. A staggered periodicity is more suitable, at 150,000 miles for driving vehicles and 185,000 miles for middle vehicles.”

Implementation of the staggered tyre-turning regime will result in each unit requiring seven wheel lathe operations from bogie overhaul to overhaul. There are currently five operations with the 140,000 mile periodicity and it is thought that the increase in visits to the Shields Depot wheel lathe will be compensated by quicker turnaround times and flexibility within depot work patterns.


Wheelsets are of course important drivers for vehicle availability and therefore for customer service. They can also be key components in causing damage and cost to infrastructure. Safely prolonging the life of these expensive items has already paid dividends, but this brief insight belies the extent of the ongoing work.

Clever engineering, pragmatic problem solving and the use of new techniques and technology will continue to optimise the balance between wheelset safety, performance and cost, underlining what we’ve known for a long time – that engineering excellence needs to lie at the heart of any modern, efficient railway system.