From scientific research to the impact of… The rail doctors

ProRail, Delft University of Technology and NWO consolidated their knowledge to make the rail network as resilient as possible to future disruptions. New inspection methods ensure better maintenance planning, earlier interventions and less inconvenience for train passengers, according to Sytse Bisschop and Paul van der Voort from ProRail.


Text: Marion de Boo, photography: Getty

‘We are taking various approaches to reduce delays’, says Van der Voort. He is the programme manager of the data lab in Utrecht, which is located in the listed building De Inktpot, the largest brick building in the Netherlands. The long corridors are reminiscent of train tunnels, full of decorated brick arches, with beautiful inlaid patterns of green and yellow enamelled precious stones. The eye-catcher in the data lab is a large model railway track. ‘This is where we test programs for automatic train control without drivers. Improved predictions of the duration of delays, lower maintenance costs, less noise pollution …’, Van der Voort is brimming with plans. ‘Working on trains is a fantastic job. The train is for everybody and also sustainable.’ Bisschop nods in agreement. ‘I trained as a mathematician, but now I can explain what I do to my neighbour, and he benefits from it too.’

Healthy rails vibrate differently

Last August, ProRail and NS (Dutch train operator) fitted two passenger trains, an intercity and a local train, with special measuring equipment to discover initial defects on the track. Van der Voort: ‘A “healthy” rail vibrates differently from a defective piece with an initial fracture invisible to the naked eye.’ The idea of listening to vibrations in the track arose during the scientific research programme ExploRail [see box]. Fourteen sensors have been mounted beneath the “measuring trains”, and the most important eight have been screwed onto the wheel dust covers. Bisschop: ‘They very accurately measure the movements of the wheel axis relative to the rails and measure the vibrations in three directions. We call those “axle box acceleration measurements”. Other sensors measure the movements of the train, temperature and relative humidity, the position of the train (via GPS) and the box acceleration. Two microphones measure the train noise.’

Not quite the right type

Installing all this was far from easy. Van der Voort: ‘It was great to see “our” double-decker train run into the workplace in Eindhoven. Unfortunately, it was not quite the right type, because the conduit our cables had to run through was missing in this model. Our colleagues from NS then rapidly charted another double-decker that was on the way to Maastricht. However, we subsequently discovered that the bolt with sensors that we wanted to screw onto the axle box cover as an exact replacement for the original bolt was the wrong size due to a less than clear working drawing. Fortunately, a local company could quickly supply replacement bolts. When we modified our local train the next day, something went wrong with the cables. We did not receive other cables from Germany until four in the afternoon, and it was almost midnight by the time we finished. The next morning, we opened our laptops in Utrecht and saw the real-time measurements streaming. That was a fantastic moment.’

Filtering the noise from the data

That’s when the real work started. How were we going to sieve useful information from this mass of data? How do we precisely determine the location of the train? How do we establish whether a defect is serious or not? How do we separate the noise from the data? Does a high outcome really mean a large defect on the track, or is it perhaps just a defective sensor? Van der Voort: ‘Data developers removed the noise from the data, interpreted the signals and combined these with data from the traffic control post into daily and weekly reports. These reveal where the biggest problems on the track are located, whether those problematic spots have been measured before and whether the problems have since increased. Loose bolts, rail deflection, worn heads, subsiding tracks, damage to the steel or at the end points of switches: in half of the cases, the measurements have led to faster repairs.’

14 measuring trains

Bisschop: ‘From previous research in collaboration with Delft University of Technology, we knew that the principle could work. But tackling and applying such a scientific proof-of-concept in practice is often a considerable challenge. Another problem had to be solved. We will now roll out the idea further. With 14 measuring trains, we can monitor the entire rail network each week without disrupting the timetable. That means maintenance is better planned, interventions are realised earlier and at lower costs, and rail passengers experience less inconvenience.’ The sound measurements are important in view of the trend of switching from theoretical calculations of noise pollution to direct measurements. Van der Voort: ‘Sometimes, a spot turns out to be far noisier than was calculated. In those cases, we want to know what is causing that and what we can do about it.’ He would also like to equip the measuring trains with cameras on the roof to inspect overhead cables. A little while later at the station, that proves to be a good idea: a defective overhead cable stops all train traffic towards The Hague for the remainder of the day.­­­­­­­­­

What is ExploRail?

The Dutch rail network is the busiest in Europe and is becoming busier still. To make the network more resilient to disruptions and to optimise collaboration in the rail world, the research programme ExploRail was realised from 2011 to 2017. In this programme, ProRail collaborated with NWO Domain Applied and Engineering Sciences (AES, which at the start of the programme was still Technology Foundation STW), NWO Domain Social Sciences and Humanities (SSH) and Delft University of Technology. In the programme component DrTrack, a monitoring system was developed to very precisely detect small fractures in the train rails that might not yet even be visible to the naked eye. A rail grinder can then grind the rails at night, without causing any inconvenience for train passengers.

We have to monitor, look ahead and realise preventative maintenance

‘The rail world is conservative by nature’, says programme initiator Rolf Dollevoet, research coach at ProRail and a professor at Delft University of Technology. ‘Innovations are associated with risks and investments are made for the next decades. Rails have a lifetime of 50 years and bridges up to 100 years. At the same time, we need to keep ahead of the growth in passenger traffic. Over the next ten years, we expect thirty to fifty percent more train passengers. The ministry wants us to run trains 24 hours per day, but that would leave no time to perform maintenance at night. And during the day, inspectors can no longer walk along the rails because it is too dangerous. We will therefore need a radically different approach. We have to monitor, look ahead and realise preventative maintenance to ensure that the entire system does not derail.’

NWO magazine on science

Onderzoek is NWO's new magazine and the successor to Hypothese. NWO develops science policy in dialogue with the field and we want to show that. In this magazine we discuss the developments, wishes and other voices in the scientific field. Onderzoek is sent free of charge to researchers, policymakers, journalists and other relations of NWO.