Vehicle Track Interface Seminar 2025

Rail Engineer reported extensively on the Vehicle/Track Systems Interface Committee (V/TSIC) and Adhesion Research (ADHERE) seminar in 2024. The 2025 event, held in York in early March, provided the opportunity for over 100 delegates to hear presentations on a mix of original research, updates, and practical experience of applying learning. This article will cover the five V/TSIC presentations. Another article will cover ADHERE presentations.

Rail management/infrastructure modelling

Brian Whitney, Network Rail’s engineering expert for track and switches & crossings (S&C) is a regular speaker and, unusually, spoke about the process of monitoring track rather than his usual subject – the number and nature of track defects. He described the current Network Rail fleet of monitoring trains. Most of them are fitted with extremely high-tech equipment, some of it ‘cutting edge’, but there is an issue with the vehicles which are mostly life-expired with an average age of 54 years. Examples are the New Measurement Train, introduced in 2003, a 1970s High Speed Train (Class 43 locomotives and Mark 3 coaches), and the even older Structure Gauge coaches, based on 1950s Mark 1 coaches.

While the aim is for all track to be covered by monitoring trains, there is still plenty of inspection provided by maintenance workers on the track. All S&C and some plain line is inspected this way. Road-rail vehicles, trolleys, or hand propelled ‘walking sticks’ are also sometimes used on sections of track not inspected by train. The aim is to replace ‘boots on ballast’ inspections with trainborne inspection of track geometry, ultrasonic testing of rails, overhead line (OLE), rail wear, ground penetrating radar (looking at ballast layers and fouling), and structural gauging, and the use of drones for site surveys for renewals/ enhancement and incident response.

Brian added that three visits to a site are often required to inspect the track, scope the work after a defect has been identified, and then to carry out the repair. That said, wider aspects such as combination effects and root cause, such as track geometry, support conditions, and embankment conditions, are not always considered. This is a significant amount of activity given that 3,000 to 3,500 rail defects and around 2,000 new and 3,000 to 5,000 repeat track geometry faults are identified every month. In this context, Brian reported that the number of rail defects has increased slightly between 2024-2025 compared with the previous year whereas the rail break total is down a little. The ambition is to be able to gain accurate information about and the precise location of faults, with sufficient detail to allow the fix to be designed so that the only ‘boots on track’ activity is to carry out the repair.

Brian showed the progress being made with Network Rail’s Infrastructure Monitoring Programme in pursuing this ambition. For example, Machines with Vision’s RailLOC system has been fitted to an Ultrasonic Test Unit and the Class 153 Video Inspection Units to improve locational accuracy to within 30mm – a huge improvement. A Fault Navigator app has been developed for iPhones and iPads to guide verifiers directly to the suspect fault and Network Rail is working on a temporary variation to standards to enable this new way of working to be proved and adopted.

High quality video and machine learning developments are providing better reports of faults than manual methods leading to a review of the rail visual inspection standard.

As a result of the trials, validation reports are coming thick and fast, Brian said. For example, the first instance of removing a cyclic top speed restriction using Unattended Geometry Monitoring System (UGMS) data from a passenger train on Wales & West region was achieved in September 2024. Brian is now working with Network Rail’s Technical Authority on safety cases, together with process and standards changes to enable UGMS to be used as an alternative to attended measurements or dedicated track recordings using an infrastructure monitoring train.

All this has included vehicle trials and formal commissioning, trade union engagement, building ‘digital’ patrol route plans, training end users, and making sure that video inspection paths are put in the Working Timetable. The first timetabled runs happened in early September 2024 to supplement manual S&C inspections and work towards the manual inspections being turned off.

“What’s next?”, Brian asked. In 2025-2026, further vehicles (e.g., more Class 153s) will be built and/or commissioned so that trials can be rolled out as a national business-as-usual service, delivering the benefits of boots off ballast, improved identification of potential service impacting faults, accessible, exploitable data, and a foundation for further machine learning applications.

Impact of freight trains on bridges

Geoff Watson from Southampton University described work on RSSB research project T1300 Heavy Axle Weight Assessment for Underline Bridges.

The maximum axle weight permitted on secondary routes in Great Britain can vary. Under certain circumstances, Network Rail issues a dispensation to operate wagons with a heavier axle weight, up to a maximum of 25.1 tonnes (RA 9) and 25.4 tonnes (RA10) on RA8 routes normally limited to 22.8 tonnes. Sometimes, there is a speed restriction imposed. The dispensation could be withdrawn at any time introducing uncertainty for freight operators and risks being unable to operate economically.

The aim of the project is to understand the impact of Heavier Axle Weights (HAW) on underline bridges and provide a tool to assess future route accessibility based on condition for two asset types – masonry and metal bridges – taking into account future HAW traffic. The work involved laboratory testing, field measurements, and modelling.

Geoff presented a case study of Briscoes Bridge near Reading on the Berks and Hants line. It is a brick-built flat arch originally designed by Brunel (who would never have imagined the loads it might be required to carry). It was reported in a 2019 report that its “arch barrel is very noticeably live under load of passing high speed and freight trains where the arch shows a ripple/wave effect in the direction of train travel … (estimated 10mm movement)”. It was upgraded in 2020 using the Masonry Arch Repair and Strengthening systems, but subsequent inspections reported significant movement post remediation. Unlike metallic structures, masonry is not really flexible.

Geoff’s team used a variety of sensors to measure deflections at different locations on the bridge as different types of train passed over. These were used as inputs to a finite element model and the effect of passing these loads across the bridge were demonstrated. Laboratory testing was carried out on typical masonry samples. The conclusion was that the bridge is responding well to the loads imposed on it and the remediation is performing very well.

Geoff also outlined laboratory tests carried out on ‘early steel’ from a very old bridge over a canal in Deptford and on wrought iron from a bridge in Audenshaw. Tests included accelerated corrosion and assessing the strength of sections that have either thinned or significantly pitted through corrosion.

The ongoing work is expected to create a framework for modelling other asset types, such as earthworks, in the future.

Guidance of derailed trains

Dr. David Griffin from RSSB and Dr. Julian Stow from University of Huddersfield described progress made on research project T1316 ‘Design features than can provide guidance to trains when derailed’ since the last presentation in March 2024.

Having developed a dynamic simulation model that could explore the behaviour of a derailed train interacting with ballast and sleepers, Julian described how it had been used to evaluate three methods of containment of a train following derailment: guard rails, a bogie mounted stopper, and duo block sleepers as shown in the diagram above.

The simulation cases included high and low cant curves from 200-metre to 20,000-metre radius, and speeds from 25km/h to 200km/h.

The results from the simulations were then used as a basis for developing the risk model. Guard rails were found to be effective for derailment speeds up to 50mph, while bogie-mounted stoppers were only effective for derailment speeds up to 30mph and where the curve radius was more than 1,000 metres. Duo-block sleepers were also effective for derailment speeds up to 50mph with a curve radius more than 400 metres.

David then explained both how the risk model has developed over the last year and how the derailment containment solutions might be most effective in reducing risk. The complex multi-factor risk model aims to calculate for each 25-metre section of the GB rail network:

How likely is it that the train might derail ‘here’? This takes about 40 independent causes of derailment, grouped into eight generic derailment types, into account and is dependent on the assets present at the location and train type.

Where do the derailed vehicles end up? For each of the eight derailment types a derailment cone is calculated.

This is the likely path of the derailed train for each derailment type – based on speed, curvature and switches and crossings. Derailment mitigation will impact the trajectory. Note the calculation does not consider the effect collision with structures, earthworks etc, has on the path of derailed train.

Based on where the derailed vehicles end up, what possible escalations occur? The model calculates this based on 11 possible escalations. The probability of each escalation is based on: the assets that are present in the cone; the train type(s) involved; and the speed of the derailed train.

For each escalation -what are the potential injuries/loss? The model calculates the safety loss for the base derailment plus any escalations.

Derailment speed is often lower than line speed due to presence of stations, speed restrictions etc., and historic documents suggest that derailment typically occurs at about 75% of line speed and is consistent with a quadratic distribution. Thus, on a 100mph line, the probability that derailment speed is less than 50mph is 0.125, implying that guard rails would mitigate 12.5% of derailments.

The conclusions for each of the three mitigations measures were:

Guard Rails: To fit guard rails, sleepers need to be replaced. Therefore, retrofitting guard rails at an existing location is expensive as it requires track replacement (nominally £2.3 million per km). If guard rails are fitted at the same time as a track renewal, the additional cost is about £600,000 per km. Exact cost would depend on the location and length of the guard rail section. Provisional results from the model suggest that there are some locations where guard rails could be reasonably practicable to fit during renewal.

Duo-block sleepers: Duo-block sleepers are similar to guard rails at mitigating the consequences of derailment. They are unlikely to be viable as retro-fitment (similar to guard rails). More work is needed to understand the wider advantages (e.g., reduced concrete – embedded carbon) and dis-advantages of duo-block sleepers (e.g., corrosion resistance) to enable cost-benefit analysis.

Bogie mounted stoppers: Bogie-mounted stoppers appear to be the least effective of the three derailment mitigations. Unlike guard rails and duo-block sleepers, they have general application and could potentially mitigate all derailment risk (i.e. not only installed at a particular high-risk location) but could be very costly. Design and approval are estimated at approximately £125,000 per bogie type (up to 50% more if the bogies need strengthening). Installation cost could be approximately £5,000 per bogie based on a simple attachment method (up to 100% more if the bogie needs strengthening). A rough Rail Engineer estimate suggests there would be little or no change out of £20 million for just the Thameslink Class 700 fleet. Retro-fitting to existing fleets may be difficult depending on bogie design.

Next steps include developing a policy on mitigation measures which might range from nothing to a blended approach of more than one solution. If measures are recommended then high-level designs for the mitigation measures will be created, and an implementation plan developed.

RCF, squats, and studs

Professor Mark Burstow, Network Rail’s principal vehicle track dynamics engineer, presented the V/T SIC Permanent Project Group (PPG) update. Mark described progress on managing hunting – a vehicle/track system issue where the latest work was covered in Issue 212, Jan/Feb 2025. He also described progress on understanding and hence treating rail surface damage including Rolling Contact Fatigue (RFC), squats, and studs.

He presented the promising results of work to understand these faults better using data obtained from both train-borne ultrasonic inspection and eddy current testing. Four sites had been identified for detailed investigation of squats, all of which were on straight or shallow curved track and not near signals or station stops, allowing the elimination of traction/braking forces as a dominant factor.

AI Assisted Video Recording (AIVR) data had been used to augment site visit data. Eddy current measurements and vehicle dynamics simulations had all provided useful information, but it is still a work in progress.

Modelling rail defects

Philomenah Holladay and Professor David Fletcher from the University of Sheffield presented a groundbreaking approach to modelling rail defects.

David said that a great deal of research has been carried out into the properties of premium steels, but their installed behaviour is sometimes different from that expected, something that only becomes apparent sometime after installation. He said that rail microstructure might be able to be optimised if it could be modelled to assess its response to load.

David briefed the group on the microstructure of most rail steel which takes the form of grains of pearlite. Pearlite consists of layers of soft ferrite (Fe) and hard cementite (Fe3C). Ferrite is ductile and tough, cementite is highly wear resistant. The spacing between these layers usually defines the rail steel’s wear resistance and response to load.

He outlined rail steel’s response to load where plastic flow causes strain hardening which increases wear resistance. However, plastic flow cannot continue indefinitely, hardness cannot rise infinitely, and wear debris and cracks form as the limit of ductility is reached. This process, known as microstructural damage, underpins most rail failures. Typically, it is assessed on rail sections or laboratory simulated samples using microscopy. More recently, a process called nano-indentation has provided better results.

Modelling has, since the 1990s, used finite element analysis. This process struggles when trying to assess more than a few hundred wheel passages (stress cycles). An alternative hybrid approach was developed: the ‘brick models’ which combines stress analysis with experimentally determined material response. The properties of ‘bricks’ determined from experimental data is validated against wear rates based on incremental accumulation of permanent deformation to the rail steel with each wheel passage (ratchetting strain accumulation). This was very promising but employed bespoke code that was hard to maintain and reuse.

Philomenah described a new approach – FLAME GPU accelerated computing. The Computer Science team in Sheffield has developed a framework for parallel computing on PC graphics cards giving thousands of computing cores, whereas the classic microprocessor only has four or eight cores. This has allowed the ‘brick model’ to be re-implemented and validated against the old versions, delivering a modelling environment that is easier to handle. The faster computing is enabling deeper microstructural modelling with ‘bricks’ down to micron size. Multiple millions of bricks are possible, even on entry level hardware, allowing new ways to explore the mechanisms by which rail failure happens.

Animations showed modelled crack formation.

Philomenah explained how the technique is already modelling much experimental behaviour, though there are discrepancies at small wear depth/low contact cycles which she is now addressing through modelling an additional microstructural failure mechanism.

Future work includes modelling what happens when rail is ground; developing and assessing novel materials; and gaining a better understanding of the brittle white etched layer which can form on the rail surface due to wheel slide or grinding heat input.

The five presentations all showed how research and development is contributing to better and safer railways. It was fascinating to hear about original work modelling scenarios hitherto seen as impossible, and to hear from industry professionals who both support and sponsor such activity and then work hard to apply the results to their discipline. 

 With thanks to the presenters and RSSB’s Paul Gray for their assistance.

Image credit: Network Rail