Rolling stock maintenance and remote condition monitoring

This is the fourth time that Rail Engineer has reported on London Business Conferences’ symposium on rolling stock maintenance, featuring Remote Condition Monitoring (issues 136, 148 and 171). This year was the eighth such conference and more than 120 speakers and delegates assembled in west London over two days in early December. 

Whilst the main topic was Remote Condition Monitoring (RCM) enabled by on-train sensors, wireless communications, cloud storage and analysis, there were a number of other interesting topics.

Back in 2015, enthusiasts for RCM were clearly seeing themselves as evangelists – follow us and it will all be good, they said! Roll on four years and it is evident they were right. Many presentations showed the benefit of the RCM approach and industry leaders, including Thierry Fort, executive director of rolling stock engineering with SNCF Mobilities, and Johannes Emmelheinz, CEO of Siemens Mobility Customer Services, demonstrated the clear benefits to their organisations and to their customers. 

New trains come with this facility as a matter of course, but, increasingly, there is a demonstrable business case for retrofitting existing vehicles, as speakers from Latvia, India, Belgium and Saudi Arabia testified. 

There was also an increasing trend to recognise that “throwing a bunch of sensors” at a train is not necessarily the right answer. A clever data scientist, working with a rolling stock expert, ought to be able to extract similar information from a single sensor that a less-expert team might gain from several dozen. To an extent this is true of the wheel bearing sensors described below. They were installed to monitor bearing performance, then clever people found they also provided information about the wheels and the track. 

Many speakers emphasised the importance of leadership, management and taking the people along on the journey, making the point that these are business change projects not IT projects. In 2015, this emphasis ranged from restrained to non-existent, so progress has been made. 

This report will examine some of the case studies, including managing wheels and axle bearings. It will also discuss some of the people issues and, as this conference always has some “and now for something completely different” moments, a selection of these will be described. 

Wheels and axle bearings

The condition of wheels and axle bearings can be assessed by the use of wayside equipment or by the use of axlebox sensors. Siemens, Voestalpine, TrackIQ and Talgo are suppliers of the former whilst Perpetuum and SKF supply the latter. There has been much debate about the merits of the two techniques, and what follows are some conclusions from drawing on the points made by the above-mentioned suppliers, together with input from Chiltern Railways’ Simon Jarrett (pictured below) and Huddersfield University’s Professor Adam Bevan.

Many factors determine when maintenance interventions are required on wheelsets, such as hollow profile, thin flange, flats, rolling contact fatigue and, sometimes, cavities or out-of-roundness. In addition, most bearing manufacturers specify a mileage limit between overhauls. 

The traditional method of checking wheels is by measuring profiles during inspection in depots; something that is labour intensive and not especially accurate. Automated systems have been available for over 20 years. Bearings, generally, were overhauled at the specified interval and occasional failures were detected using hot axlebox detectors (HABD). A hot wheel bearing is, at best, hours from failure, so a train with a detected bearing is almost always taken out of service immediately. A much better method is to use vibration monitoring, which can give several months warning of failure. 

Simon Jarrett reported on one of the earliest uses of GOTCHA trackside wheel condition and acoustic axle bearing monitoring, which Chiltern introduced at Wembley in 2013. Simon described the maintenance intervals for bearings and the nature of wheel wear defects before monitoring and the intervals achieved after monitoring was introduced. He said that wheel maintenance can now be planned and that there is no planned axle bearing maintenance, as shown below: 

Over the last four years, the monitoring system identified 46 bearings that needed to be changed. Simon said that, in the early days of monitoring, fitters would strip any removed bearing to confirm the defect, a practice that has now ceased due to his confidence in the system. He cautioned, however, that the trackside monitoring system is unlikely to catch a bearing that has no lubrication or severely contaminated grease and they have had three overheated bearings in the last four years. For these, the traditional HABD is the last line of defence.

Rail Engineer’s previous reviews of this conference have reported on similar results on Southeastern, which has equipped its Electrostar fleet with Perpetuum self-powered axlebox accelerometers. This year, Perpetuum has a competitor in SKF’s Insight Rail product, which is a small, battery-powered sensor that transmits data directly by 2G/3G mobile signals. 

The key debate seems to be whether it is better to have track-side or on-vehicle monitoring. Both provide information about the state of wheel bearings and wheels, but vehicle-mounted sensors can also provide information about the track and should detect a hot axlebox, although a much shorter response time will be required. Then there is the question of how many trackside systems might be needed to cover an operator’s whole network – for a large train operator with an extensive network, a single trackside installation would probably be insufficient. 

These are challenges to be worked out between potential suppliers and operators, with account being taken of the value to the infrastructure manager of the information that vehicle-mounted sensors might generate. 

In his talk, Adam Bevan provided insight into the question of the optimum frequency of wheel turning – either “little and often” or “leave it until absolutely necessary”. There is one truism, that a wheel with flanges at the thin-flange limit needs an awful lot of metal to be turned off to restore the profile. This is a condition to be avoided. That aside, other factors – loss of diameter, tread hollowing and growth of rolling contact fatigue cracks – are the key reasons for wheel turning. 

Adam described the work his team had done analysing data from monitoring systems and wheel-turning records to identify the optimum wheel reprofiling interval, which proved to be just before RCF crack growth accelerates. This is based on knowledge about crack growth. Following initiation, the RCF cracks grow slowly to start with, but then they reach a point after which they start to grow rapidly. 

The analysis indicated that leading axles tend to suffer more crack damage than trailing axles. For the particular fleet in the study, this work led to the following strategy; that wheels would be turned every 200 days instead of every 241 days, that the cut depth would be 4 to 6mm instead of 13mm, and the wheelset monitoring system would look out for any unusual occurrences such as wheel flats. This approach was expected to increase wheel life by between 50 and 70 per cent and deliver approximately a seven per cent reduction in costs.

A view from Paris

Thierry Fort (pictured above) presented a case study illustrating the results of using data in condition-based maintenance. He talked about the modern train fleets in and around Paris that have been using this approach over the last six years. The process was not unusual – connected train, big data, cloud storage, analysis by data scientists and rolling-stock experts, up-to-date maintenance centre, processes which involve data/information at the heart of all activities and having the information available before the train arrives back at depot. The results were extremely positive: failures in service were reduced by 40 to 50 per cent (reliability rate doubled), labour cost was reduced by 20 per cent, train availability was improved by 20 to 30 per cent and there was a reduction of 30 per cent in the number of work orders generated.

Thierry concluded with a look forward to a future in which artificial intelligence could be used for much more than train maintenance. Focussing on video surveillance, or “Video Protection” that would involve on-train analysis of the on-board CCTV, he noted that such a process could alert control to the signs of trouble on board a train, for example a brawl, and allow the human operators to check and, if necessary, intervene.

Now for something completely different!

One of the challenges faced by maintainers and owners of older fleets is sourcing the parts required for maintenance. This is particularly true of parts that are not routinely needed. Richard Flint of Chiltern Railways highlighted some of the challenges with a fleet which, bar a few vehicles, is over 20 years old. 

He said that Chiltern has 12,000 different parts in its stores, yet this is just a small subset of parts contained in the OEM’s Illustrated Parts Lists (IPL); for example, the Class 168 IPL extends to eight thick volumes. All too often, the supplier of the original part – the OEM – is no longer in existence and it is hugely time consuming to identify an alternative supplier. Even if the OEM is still around, it is often impossible to obtain small quantities of parts quickly. What supplier of plastic parts wants to make a rush order of, say, ten armrests? 

This situation led Chiltern and Angel Trains, the owner of the Class 165 and 168 fleets, to investigate additive manufacture, better known as 3D printing. The process starts by 3D scanning the original part, adjusting the scan to correct any deficiencies in the scanning or, indeed, in the original part. Then the item is printed on an industrial 3D printer. Three items were chosen for trial, seat back handles, seat armrests and a complex piece of trim around the driver’s power-brake controller. 

These trials have proved to be successful and, for similar items, the process presents many advantages. These include the ability to produce economically small quantities of parts otherwise unobtainable, with shorter lead times. This delivers a reduction in the time taken to renew parts on the fleets, a reduction in management time for the management of vehicle defects, an increase in vehicle availability and, above all, an increase in customer satisfaction.

People issues

At each of these conferences, the importance of managing the people issues as an integral part of the RCM projects has been emphasised. This time, Neil Robertson, chief executive of the National Skills Academy for Rail, gave a forthright challenge to the UK rolling stock leaders, along the lines of “it’s not enough to take the people with you, you’ve got to find them first!”

Neil’s talk, “Professionalising the Workforce”, described the many challenges facing the rolling stock engineering and maintenance organisations. He illustrated the skills shortage, that 30 per cent of the employees are over 50, that it is likely to get worse before it gets better, and that a great deal more has to be done to encourage girls into engineering in general and railways in particular. 

Moving on to discuss the challenge that fleet maintainers have when a new fleet is introduced, Neil compared it with the automotive sector, where cars have got gradually more complex and auto mechanics have been converted to auto technicians over roughly a 10-year period.

Contrast that with a new fleet introduction and the fleet manager often has little more than 10 months to do a similar conversion for his or her team. Some of them might not adapt to the task of using laptops and data to diagnose faults, which uses quite different parts of the brain than is typical in engineering. Neil particularly highlighted the skills shortage of good data scientists, stating that there is likely to be a “shortfall of circa 60 heads across the sector per year for the next 20 years”, although there is some light at the end of the tunnel as “England has seen a 129-fold increase in apprentices starting digital or data-related apprenticeships since 2014”. 

He also highlighted that level 4 data technician and level 6 data scientist shared apprenticeships which are available and he recommended a visit to the NSAR website to find out more. He said that companies experienced in using data tend to gravitate to the level 4 qualification with the aim of improving data cleansing whereas those with least experience tend to look to the level 6 qualification. 

Finally, for those looking at how to encourage youngsters to join the rail industry, Neil recommended looking out for the “Routes into Rail” web site which will be coming soon. 

Closing down

This conference also included a first for your writer – a paper on closing down a maintenance operation! Stephen Head and Aoife Considine from Heathrow Express (HEx) described their work on “Managing an End of Life Fleet and Depot”. In brief, Heathrow Express currently owns and operates Class 332 CAF/Siemens trains. The trains are maintained by Siemens in a dedicated depot in Old Oak Common (pictured below) that has to be demolished to make way for HS2. 

This depot also maintains the Class 360 sets that used to operate the Heathrow Connect service and which should have been replaced by Crossrail/TfLRail/Elizabeth line Class 345 trains in mid-2018. 

It is intended that, in future, HEx will be operated by Great Western Railway using reconfigured Class 387 EMUs which are being fitted with ETCS for operation in the Heathrow Express-owned tunnels. This changeover was due to happen in December 2019, but is also running late. 

The challenge for Stephen, Aoife and Siemens management was to continue to deliver safe, reliable rolling stock, and to motivate the staff to do so in the certain knowledge that their jobs would end at a date that keeps shifting. 

Another relevant factor is that the Bombardier depot which maintains Elizabeth line Class 345 trains and the Hitachi depot that maintains GWR’s Inter-City Express trains are right next door to the HEx/Siemens depot. Both of these depots need experienced train-maintenance staff.

All this meant that the management had to deal with three issues. Firstly, motivation and retention of staff; secondly, planning and re-planning the maintenance activities on fleets where the date for end of operations keeps slipping, and, thirdly, disposal of the Hex-owned Class 332 trains and safely closing down the site. All this was being done whilst working in a depot environment that increasingly resembles a building site. 

In terms of managing the operation as a going concern, Aoife stressed the importance of open and honest communication with the team to engender pride in the quality of service that HEx delivers. However, they had also recognised that, with the adjacent depots, the competitive pressures were severe, so had introduced retention payments and, where staff had left, contract staff had been recruited to fill gaps. 

Maintenance planning was carried out using what was described as a simple scenario planning technique: “What have we got to do if the trains have to keep going for 6, 12 or 18 months beyond the original forecast end date?” Aoife also said that all the planning is tempered by a healthy dose of flexibility and pragmatism.

Maturity

In attending four of these conferences over five years, Rail Engineer has noticed a certain maturity evolving over the whole business of RCM. In 2015, those attending came over as pioneers, or even mavericks. Today, they are seen as respectable members of the team who can genuinely deliver improved performance and reduced costs. 

The influx of new trains (in the UK at least) is helping, as they include data collection facilities as standard, but the case for retrofit is increasingly being made and this trend will continue.