Over the last few years, several serious accidents investigated by RAIB have involved freight train braking systems. Accidents are costly, but significant costs are also incurred even when no accident occurs, for example when wagons are withdrawn from service due to wheel flats. In its reports, RAIB recommended better management of adhesions risks, reviewing wagon maintenance, and better understanding of compressive forces along freight trains.
One of several industry groups recently set up to explore freight train safety and performance improvement (see below) is the Wagon Condition Programme Board, chaired by Tim Shakerley. Its RSSB-hosted sub-group, Freight Braking and Adhesion Research Group (FBARG), chaired by Neil Ovenden, presented its latest work at a seminar in Derby in December 2025.
The group exists to coordinate RSSB and Network Rail funded research into freight train braking and adhesion performance. Its work includes: (i) exploring performance and compatibility of freight wagon brakes and their combined impact on freight train braking, formation of freight wagon wheel flats, and potential derailment risks; (ii) considering the impact of low adhesion in a freight context; and (iii) assessing the risks of longitudinal compressive forces under braking or during shunting.
The results will be used to evaluate potential mitigations and inform standards, operational controls, vehicle design, and maintenance. What follows is based on the fundamental principles of steel-on-steel railways described in box at the bottom of page 18.
Existing knowledge
Mott MacDonald’s Robert Morley described factors that lead to freight wagon wheel flats. This included reviewing existing knowledge of the prevalence and variability of wheel flats across the GB freight fleet. Relevant standards and operational techniques including train preparation, marshalling, and driving practices were also reviewed. The work identified several issues including:
Standards: Brake performance is specified and validated in terms of maximum performance, typically the emergency brake. Usually, with defensive driving, and especially in poor adhesion conditions, lower braking rates are demanded and there are no known specifications or performance testing requirements for these lower brake demands. This means that different wagon types in a train might have different responses at lower brake demands leading to some wagons over-braking and some under-braking.
Detection: There is limited knowledge of where wheel flats occur. Wheel flats are often detected by Network Rail’s Wheel Impact Load Detection equipment, but the slide event will have occurred somewhere else and at some other time. Innovations such as VTG’s iWagon are beginning to be used to improve understanding of the times and locations of wheel slide events, as well as the brake force applied and the wagons’ reaction to wheel slide events.
Brake System Design: Many modern wagon designs utilise bogie-mounted brake gear which is compact and provides a more direct application of brake force to the wheel tread, without the need for the brake rigging associated with body-mounted brake actuators. Operators have provided evidence that some wagons fitted with bogie-mounted brake gear may be more susceptible to wheel slide than otherwise identical wagons fitted with body-mounted brake rigging. Further work was recommended.
In addition, inadvertently forgetting to release handbrakes continues to cause wheel flats. Wider adoption of handbrake interlock systems is improving the position, although these systems bring their own risks and may not be a practical solution in all cases.
Overcharge: This is where the brake pipe pressure is temporarily raised over the normal value to ensure that, in principle, the brake distributors’ control reservoirs along a train are uniformly charged to improve consistency of brake control along a train. It is not consistently documented when and how overcharge should be used and there is some evidence it can be detrimental to brake system performance in certain circumstances. While different specifications have been used over the years, nothing is currently mandated in standards or the Rule Book. An ongoing RSSB standards project, based on further industry research and stakeholder input, is proposing to update the relevant documents to address this.
The report identified operational improvements that could be readily implemented:
- Timely reporting and effective action for low-adhesion events and adhesion black-spots.
- Allow greater flexibility in train pathing requirements during periods of low adhesion.
- Trial the effect of operating all or part of the train to run in goods instead of passenger timing settings during periods of low adhesion.
- More rigorous control of load distribution on loaded wagons.
- Alignment of rail head treatment trains to treat ‘at risk’ rails before freight use.
Finally, wagons’ routine static brake testing could be improved including: (i) development of a standardised, comprehensive brake test regime, including load sensing systems and sensitivity to extremes of brake pressure tolerances; (ii) methods for automation to provide greater detail, consistency, accuracy and traceability, and (iii) cost effective means to measure block/pad loads at the brake interface to determine brake performance. The business case benefits of such changes against the costs of implementation will be evaluated before any change is implemented.
Braking performance and adhesion
Dr Julian Stow from the University of Huddersfield attempted, among other things, to answer the question, prompted by the Petteril Bridge derailment investigation: “Why did a locked, flatted wheel apparently fail to start turning after the brakes were released?”
Julian presented analytical models and discussed the various factors that might impede wheel motion. The University has a full-scale wheel-rail rig, HAROLD, which was used in some experiments and a wheel flat was accidentally created during the test, yielding much useful information.
In summary, if brake torque exceeds the torque available between wheel and rail because of insufficient adhesion, the wheel will slow and lock, followed by harsh wear caused by sliding, creating a flat. The flat size depends on axle load. For example, a 25-tonne axle load at 60mph sliding for 20 seconds (approximately 500 metres) is likely to cause a flat between 70mm and 130mm long.
A self-sustaining flat might be created if an initial large wheel flat is created and there is not enough adhesion to restart rotation. The wheel may stay locked even after brake release, sliding on the flat and enlarging it. Alternatively, if the wheel speed drops sufficiently, an existing wheel flat can re-lock the wheel, causing it to slide on the flat and grow larger.
When a wheel flat forms and the wheel resumes rotation, the edges of the flat tend to wear out, increasing its length while its depth remains fixed. A worn flat causes lower resistance torque and impact load compared to a new flat, even if it is longer. Therefore, length alone is not a reliable measure, but it implies that new flats are more prone to self-sustaining or re-locking.
Daniel Jones from Serco presented his work carried out with Adam Twigg, which investigated factors impacting braking in low adhesion, collating knowledge and experience gained over the last 40 years together with analysis.
Examples included findings by British Rail Research from at least 35 years ago that had showed that: (i) dry sand typically restores the coefficient of friction from about 0.05 to 0.10 – 0.12 within one wheel rotation; ii) adhesion gels such as sandite persist on the rail for longer than sand but deliver a smaller improvement and, once compacted, can act as a lubricant; and (iii) freight train length has a minimal impact of brake application time, but a very significant impact on brake release time.
From discussions with freight operating representatives, Serco learned that freight drivers are trained to drive defensively. Each train is ‘assessed’ for the feel of its brake performance and driving style is modified accordingly, and, as this is integral to the training of drivers, there is little feedback on train braking variability.
They added that simulating conditions in post incident review has historically been very difficult. Brake performance requirements have also changed over the years. The standard in 1993 specified maximum and minimum stopping distances for individual wagons, whereas the current version omits the minimum value. Moreover, although there are clauses requiring that a new wagon design should be assessed for compatibility with those to which it might couple, no particular values are mandated. Design differences – e.g., cast iron or composition brake shoes, body or bogie brake cylinders – can lead to different responses for partial brake applications. These differences can lead to some wagons being over-braked and others under-braked compared with the desired brake demand, a possible issue in poor adhesion conditions.
The current standards require meeting an end goal with proof of compliance demonstrated by the manufacturer and Entities in Charge of Maintenance. It was reported, however, that there are sometimes inconsistent methods of calculating brake force between manufacturers, and subsequent maintenance data is limited to measuring brake cylinder pressure, not block forces. These factors are vitally important when considering ETCS.
Rail Engineer has already referred to the challenges of inputting freight brake force data from a database called R2 into ETCS (Issue 216, Sep-Oct 2025), but here it was shown that the data may be incorrect, with the statement: “On ETCS it is essential that Drivers do not rely on the brake curve with which they are presented.”
From this work, several interim recommendations were made including carrying out a suite of brake tests to validate R2 values and review the impact of possible standards for initial brake application and overcharge.
Longitudinal compressive forces
The derailment at London Gateway in 2021 (Issue 206, Jan-Feb 2024) highlighted an issue with the longitudinal dynamic behaviour of the train. Dynamic longitudinal compressive forces (LCF) within the train were sufficient to unload and derail the wheels of a short unladen wagon in the middle of the train. Paul Molyneux-Berry from the University of Huddersfield described work to characterise LCF and define safe operational limits. This addresses one of RAIB’s recommendations.
A review of LCF derailments has identified four main causes:
- Slow propelling movements over crossovers or sharp curves. Most common in GB and Europe. Applying power before train brakes are fully released is a factor in some incidents.
- Dynamic LCF in a rapid transition from traction to braking. This occurs in all countries, on main line or in yards, often at low speed and is frequently the result of an emergency brake application or a collision.
- Dynamic LCF caused by ‘run-in’ of slack on undulating gradients. Most common with very long trains in North America and Australia. Excess slack in the train, for example couplings not tightened, can be a contributory factor.
- Descending steep gradients with braking by leading locos. This occurs with very long trains in North America and Australia, especially when the locos are using dynamic braking.
Paul described the factors which, combined with high LCF, might lead to a derailment. In Europe, North America, and Australia, standards specify the Endurable LCF (ELCF) for vehicles, train formations, and brake set up to keep LCF below the ELCF threshold. EN15839 is the primary European LCF testing standard for vehicles. GB has no mandatory requirement to assess ELCF in wagons which means that there is the potential to import LCF derailment risk on to the railway. There would be a great deal of work to demonstrate GB compliance with EN15839, but this standard does not cover all GB situations. With higher speeds and loads than is typical in Europe, the LCF environment in GB freight trains may be more demanding.
Paul summarised that:
- In GB, LCF derailment incidents generally occur at low speeds <10mph, usually in protected moves.
- High-speed LCF derailments are rare, both in theory (from the simulations) and from European and GB LCF derailment history.
- Light/short wagons with bar couplers, marshalled between loaded wagons, may be highest risk and are becoming more numerous to meet demand.
- For GB scenarios, causes of high-LCF events within a train, which could escalate into a derailment, are both well understood qualitatively, but not quantitatively.
- GB has no mandatory assessment of ELCF, nor controls on formation and brake regime to limit LCF in trains.
- The modelling techniques and software developed for this project are effective for modelling LCF and ELCF for GB freight train scenarios.
- EN15839 may NOT be effective for assessing vehicle propensity for dynamic LCF derailment.
The seminar was well attended by the wider rail freight community with lively discussion. Rail Engineer’s takeaway was the enthusiasm for improvement, demonstrated by the large audience.
Rail freight operates in a commercially challenging and tight margin environment, and the focus of the seminar and the further work remains on establishing cost effective mitigation measures for improving the performance of freight train braking and adhesion – including operational controls, improving inter-wagon brake performance consistency, and continuing to explore the merits of on-board technology such as i-Wagon.
There are ambitions to be able to run longer and faster freight services to support freight growth and better integrate with passenger services on busy sections of mainlines. Having better understanding of freight vehicle/train braking is a key aspect of developing these capabilities.
Further information on the research can be found by can be found at: https://www.rssb.co.uk/research-catalogue by searching T1350, T1351, and T1352.
Wheel/Rail braking and adhesion in context
The wheel/rail contact patch is approximately the size of a 5p coin. A significant advantage of this small contact patch is low rolling resistance, enabling energy efficiency, but even small amounts of contamination can result in low adhesion. The wheel/rail interface is impacted by a wide range of external factors. Changes to adhesion can be highly transient and localised.
What is adhesion and why is it important?
Adhesion is the measure of friction between the wheel and the rail and limits the ability of trains to accelerate and brake. Adhesion is usually described using the coefficient of friction (μ), expressed as a decimal fraction or a percentage. Achievable braking rate is often expressed as a percentage of acceleration due to gravity (%g); achievable rail braking rate in %g is limited by and approximately equal to the local μ, expressed as a percentage. All main line train braking in the UK is dependent on available adhesion.
As an illustration of the problem, dry rail μ in the range 0.3-0.4 and wet rail μ in the range 0.1-0.15 is usually adequate for general rail work. But in locations where heavy contamination such as damp and compressed leaf film have formed, μ can fall to values as low as 0.05- 0.01. This level of friction is lower than is achieved in automobile engines using good quality synthetic oil.
A modern disc-braked passenger train requires an adhesion of at least 9% (μ >0.09) to deliver a nominal full-service braking rate of around 9%g to prevent wheel slide (based on all axles braking their own weight) with higher brake rates for emergency braking. A typical freight wagon can deliver a nominal braking rate of around 7%g (μ >0.07) for a full-service brake application on level, dry track.
A locomotive requires more adhesion to start a heavy freight train without spinning its wheels, requiring an adhesion >20% (μ > 0.2).
If more adhesion is demanded than is locally available, this will result in wheel spin or wheel slide. The majority of GB passenger trains are fitted with Wheel Slide Protection (WSP) systems. Most freight locomotives are fitted with Wheel Slip Control, mainly for traction purposes. GB freight wagons are not fitted with WSP, although some wagons have been and are being delivered with wheel flat prevention equipment.
If there is insufficient adhesion, a rotating wheel locks and slides along the railhead. The resultant heating generates a flat spot on the running circumference of the wheel tread. The longer the period of the slide, the longer the length of wheel flat. Extensive wheel flats can cause infrastructure damage and derailment. Low adhesion can also cause other safety incidents such as signals passed at danger, platform over-runs, and collisions.
Image credit: iStockphoto.com