In Rail Engineer 217 (Dec-Nov 2025) we reported on a September 2025 IMechE event which featured several case studies on rolling stock cracks and fractures. In that issue, we examined the standards, tools, and techniques involved in managing cracks and fractures on trains.
In essence, if standards are complied with, knowledge of the infrastructure is incorporated in designs, and suitable monitoring is carried out, the risks of cracks forming or developing can be managed.
However, cracks still happen and industry can learn from how others dealt with them. In this article, we present four case studies discussed at the event: three concerning UK rolling stock and a fourth concerning a wheel-rail interface issue elsewhere.
Case study 1: GWR classes 800 and 802
In April 2021, cracks were discovered in the anti-roll bar and yaw damper bracket on the carbodies of Great Western Railway (GWR) Class 800 and 802 trains. These were being investigated and managed when, on 8 May 2021, cracks were detected at the other end of the bolster around the lifting pad. These were in parent material (not a weld) in an area with limited operational loadings.
A decision was made by Hitachi, the Entity in Charge of Maintenance (ECM), to withdraw the affected fleet. The initial investigation and return of the trains to service was described in Rail Engineer May/June 2022, ending at the point that the inspection regime was being brought under control and permanent repairs were being evaluated.
Andrew Skinner, head of engineering at GWR, praised the team working between the various interested parties (GWR, the ROSCOs Agility Trains and Eversholt, and Hitachi, with various others including DfT, ORR, and independent consultancies).
Investigations and studies had arrived at the following conclusions:
Weld repairs to the root of the body/bolster welds would be difficult as it would be hard to achieve the correct weld fusion.
In-service-testing with strain gauges over representative routes showed that there were occasions where the loads on both the yaw damper bracket and Anti-Roll Bar (ARB) droplink were higher than allowed for in standards, which adversely affected fatigue life.
Daily checks for cracks were unsustainable.
Further Finite Element Analysis (FEA) work demonstrated that with specific welds intact there was no risk of lifting pad detachment allowing checks to be moved to 36 days with repair in the following 36 days.
Hitachi concluded its technical investigations, developed repair procedures, and designed modifications. In parallel, a repair programme was developed named project ACER. Initially, this was based at Arlington Fleet Services at Eastleigh, one of the few sites capable of accepting nine-car trains which are more than 230 metres long.
The strategy was as follows: (i) all vehicles to be modified to the same level while maintaining fleet availability AND ensuring fleet safety; (ii) delivery was to be from Eastleigh supported by the Hitachi Manufacturing team at Newton Aycliffe with the objective of; (iii) minimising unit downtime from passenger service while optimising modification timeframe using standard repair procedures; (iv) while trains were awaiting modification the situation would be contained by ongoing analysis of Fleet Check results and crack propagation rates; and (v) risk management of any emerging issues.
The scope of Project ACER was modified and now contains two workstreams to allow prioritisation of units. One was interim repairs with a shorter downtime carried out on individual cars within units to manage availability. Some of this work will be carried out at London North Pole depot. The permanent modifications require a longer downtime and are carried out on full sets.
Andrew explained the scope of the permanent modification which involve extensive work.
The full modification is on a leading vehicle and involves the following activities:
- Removal and replacement of the ARB block and weld line (WL) 1 (the longitudinal weld connecting the bolster to the car body).
- Application of stiffener plates.
- Application of new design of stiffener bracket.
- Removal of 7XXX aluminium ARB blocks replacing with 5XXX aluminium.
The modification redistributes forces through the new stiffener bracket to ensure that the carbody extrusion, bolster, and WL 1 are capable of withstanding normal operational inputs for at least the design life of the vehicle.
The work is extensive and early units were out of service for a long time. As a result, there was further collaboration leading to Hitachi’s design team implementing several changes to the programme to accelerate modifications which has reduced the work time in 2025. Even so a unit might still be out of service for over 40 days.
In summing up this project, Andrew made the understatement that managing cracks is not easy! He added that while working together might promote many views, challenge and support are key to successful outcomes. The strategy must suit all stakeholders and must be arrived at through consensus.
It is important to model and test the as-is situation and potential solutions, and it is equally important to search for alternative potential solutions to optimise time, cost, and/or quality.
Although this section refers to the work on GWR’s trains (93 units, 605 cars), as far as Rail Engineer is aware, this issue affects all the Class 8XX trains built so far (Classes 800, 801, 802, 803, 805, 807, and 810 – approximately 240 units, 1,500 cars). That’s a massive repair programme.
Case study 2: Yaw damper bracket failures on Turbostars
One of the challenges when dealing with cracks and/or fractures is to understand the root cause so that the repair/redesign is more resistant to a repeat failure than the original design. Sometimes the root cause is quite unusual as Carl Woolley, managing director of Design and Analysis Ltd (D&A) and Mark Gay, fleet strategy manager at Porterbrook, outlined.
A review of the carbody design during a life extension project highlighted that the welds attaching the aluminium yaw damper bracket to the aluminium carbody extrusion had a rather optimistic weld classification and should be subject to sample inspection. The inspection revealed cracks in approximately 3% of the suspect locations and more than 90% of the cracks were at the wheel 3-4 location. The original FEA model was refined to current standards and the impact of a) the worst crack found and b) possible future crack propagations were modelled and analysed working to EN12663. The conclusion was that there was little concern for structural integrity.
Repair options were discussed following inspection of a unit with a crack while undergoing a bogie change. Weld repair options – described as “dig out crack and reweld” or “replace a section of solebar C slot” – were undesirable, so it was decided to explore whether the area around the crack could be cut away. Carl Woolley described next steps which involved assessing the cut-out area against the full carbody FEA, the results of which were promising. The modelling was validated by strain gauge tests.
The aim was to: (i) calculate the predicted life from strain-time histories at the key features to support repair proposals; (ii) investigate the influence of yaw damper loads; and (iii) to identify any influencing factors that contribute to the failure of the yaw damper weld end. CrossCountry Trains kindly agreed to support the trial using a 170/1 unit. D&A designed & prepared the test spec, and Alstom was selected as the test house. Testing was to be done in service on the Nottingham – Cardiff route. The installation included:
- 18 Strain Gauges.
- 13 Accelerometers.
- 5 Thermocouples.
- 2 Displacement gauges.
- 1 Air pressure sensor.
- 1 speed signal radar.
- One GPS locater.
It was concluded that the strain gauge runs validated the modelling, but no mechanical strain cases were found to explain why there was a failure. There was, however, significant thermal strain associated with wheel 3-4 where the majority of the cracks had been found. The repair could proceed while the thermal strain issue was further investigated.
A likely cause became clear. The engine exhaust runs alongside the solebar/damper bracket and further analysis of the strain gauge results showed clearly that stress in the affected parts was directly linked to temperature. This was confirmed with static trials on the same unit at position 3-4. This showed that the strain increased when the coach was stationary with the engine idling which can only be caused by thermal loading.
As a result, in addition to cutting out around the cracked area, insulation was applied to the exhaust in the critical areas.
As all this was being finalised, a new crack at the bolster edge was identified. This crack had initiated in parent material, not at a weld toe and is above the bolster. The unit was put on a seven-day check for crack growth, but none was ever reported. No other cracks of this nature were found. Any repair for this crack would require the removal of the bogie, bolster, longitude and inter-end welded fabrications. D&A concluded the crack should be managed in the same manner as the original cracks, but the cutout repair is not possible for this vehicle. D&A carried out a fracture mechanics study which concluded that the risk of the crack growing and causing structural issue was negligible.
It was concluded in the end that all the cracks would be protected from corrosion but otherwise left as is, but subject to inspection during heavy maintenance. The only modification was to reduce the thermal loading into the bolster.
Case study 3: London Underground Bakerloo Line trains
Three engineers from Transport for London – Matthew Brown, David Lewis, and Steven Morris – described how cracks were managed and repaired on the 1972 tube stock which has passed its 50th birthday. The bogies have been prone to issues throughout the fleet’s life and many parts have been replaced. The bogies are visually inspected during three-yearly bogie overhauls. During one inspection, a large crack was found on the top channel, welded to the bogie side frame.
It had not previously been noted and was bigger than the length allowed on the fracture map. A check of 10% of the 36-train fleet showed that approximately 6% of those checked were cracked outside the fracture map criteria.
As a result, non-destructive testing of the bogie frame top channels is now required both during overhaul and through a one-off inspection while the bogies remain in service, even though access is challenging.
Repeat NDT was carried out at frequencies dictated by the crack length on the previous test. If a crack was discovered that was longer than allowed during the once-round inspection, the train unit was stopped until a weld repair could be carried out. A test run with strain gauges provided results that enabled allowable crack lengths in the fracture map to be relaxed, and new re-inspection criteria set. At present, a weld repair process has been approved with a prototype shown in the photos below.
Case study 4: Matching the two sides of the wheel rail interface
When building a new freight railway or completely rebuilding an old one to modern standards, you will agree in advance the standards that you are going to work to, and you will comply with them. Then you can open the railway and all will be well. Or maybe not, as Professors Bridget Eickhoff and Felix Schmid explained.
The railway, which shall remain nameless, is approximately 300km long and is a new build, largely single track, but laid on an existing formation with minimum curve radii of 300 metres. The ballast is new, so are the concrete sleepers, and the CEN 54 E1 rails. The rolling stock is formed of new 120 tonne Co-Co locomotives and 100 tonne gross weight wagons with Y25 bogies. The trains run loaded in one direction and empty in the other. Trial running was successful, but the wagons’ wheelsets were showing severe flange wear after less than 30,000km in service.
Initial information highlighted the following:
- Severe flange wear on the wagons: S1002 wheel profiles with a 1 in 40 base cone angle.
- No issues on locos: ‘a local wheel profile’ with a 1 in 20 base cone angle.
- Gauge face wear: on some curves. Rails installed with a 1 in 20 rail inclination.
- Switches and crossings: no major wear issues but some strange wear patterns.
- Lubrication underspecified: no routine lubrication on wheels or rails.
Based on information received, Bridget and Felix thought that there was a possible mismatch between the wagon wheel profiles and the rail profile/inclination leading to a lack of steering and hence severe wear on ~300-metre radius curves. They were told that an initial application of grease, by hand, to key curves was having some positive effect. Clearly, more detailed investigation and face-to-face discussions were required.
During the initial site visit everything seemed well built and maintained, except for the severe flange wear on the wagons albeit slowed by the hand lubrication mentioned above. Further assessment and detailed discussions with senior stakeholders led to a number of conclusions.
Different organisations were responsible for different sub-systems of the railway, namely, infrastructure and its maintenance, rolling stock and its maintenance, signalling, and control. All individual sub-systems had been designed and built to latest technical standards and were of good quality with no obvious flaws. However, there was no clear responsibility for system interfaces or integration, or indeed, for the consistency of the applied standards. There was an apparent mismatch between wagon wheel profile and track design, resulting in pretty much zero equivalent (effective) conicity. There was also a possible mismatch between the wagon wheel profile and S&C design. The stakeholders had understood that the use of S1002 1 in 40 wheel profiles on 1 in 20 inclination rails was consistent with the approach of the French national network.
Further background information suggested that the railway’s specifiers had proposed lubrication twice a year. This is clearly not adequate in any climate. Lubrication introduced on external advice had bought time for a more robust solution(s) to be developed which might include different wheel profiles and investigation of automatic lineside and/or train-based flange lubrication.
In the presentation, Bridget discussed the results of the EU funded DynoTrain review of European wheel-rail interfaces, which had shown that: (i) Germany (DB) installs the rails at 1 in 40 and generally uses wheels (e.g., S1002) that are based on a 1 in 40 cone; (ii) Britain (NR) installs the rails at 1 in 20 and generally uses wheels (e.g., P8) based on a 1 in 20 cone; and (iii) France (SNCF) installs the rails at 1 in 20 but mostly uses wheels with a 1 in 40 cone – a mixed system.
However, it was clear from the DynoTrain results that the in-service condition in France (1 in 40 wheels, 1 in 20 rails) is much more similar to the situation in Germany (1 in 40 system) than to Britain (1 in 20 system). In other words, the French system appears to behave as though it has 1 in 40 rails. The difference between the designed 1 in 20 rail installation and how the in-service rails appear is a result of the wear of the rails in service, Bridget explained.
Wheel profiles are typically turned back to as-new every few hundred thousand kilometres (every two or three years), whereas rails are generally reprofiled infrequently, and less than 2% of rails are replaced each year. Thus, after a few years, the in-service rail profiles reflect the shape of the wheels. They are thus very similar to those installed at 1 in 40, resulting in appropriate values of equivalent conicity.
Applying the French approach without fully understanding the likely impact was probably the root cause. Following the French model and allowing the railway to bed in over 40 to 50 years was obviously not a practical option, since the wheelsets on the wagons would not have survived the first six months.
Using video cameras mounted under a wagon clearly showed that wheels were running hard into flange contact on many curves, thereby confirming the diagnosis.
Actions taken included continuing with hand lubrication of relevant curves and research into the installation of fixed lubricator installations. Drivers were briefed to note any locations where flange squeal is heard, so that lubrication can be targeted. The drivers were delighted to do this as it made them feel part of the team!
Other actions included procuring a detailed study of the wheel-rail interface and wagon behaviour for this railway, considering options for an alternative wheel profile more suited to the new infrastructure, and undertaking a controlled trial of EN 13715 EPS wheel profiles on a wagon to check wheel wear and ride performance. Felix and Bridget said that all these steps were underway and the wear problem is under control, though not yet resolved.
Some important lessons were learned. First is the importance of managing the wheel-rail system as a whole. All parts can be designed, built, and maintained to the latest standards, but this does not guarantee compatibility and good performance. Apparently small differences can have a large impact because of the forces and number of interaction cycles. When a 1 in 20 wheel profile is overlaid on a 1 in 40 wheel profile, the difference is often so small that the image must be enlarged so that the difference is clearly visible. Wheel-rail friction must be managed with lubrication and observation.
Involving key stakeholders in discussions of the issue enabled the team to move from ‘who is to blame?’ to ‘how do we solve this?’ Probably the most important lesson is that all parties need to be part of the solution – there is no single change that will fix everything at an interface!
Summing up
These studies illustrated that what engineers plan in their designs is not always realised in practice. Loads might be higher than specified, operating conditions might change, materials might be unsuitable for the application, weld classifications might be optimistic, or standards in common use elsewhere might, for one reason or another, be incompatible on a new railway.
However, the cost of remedying the failures can be expensive, especially on large fleets. Indeed, Rail Engineer believes that the cost of rectifying the Class 8XX cracks across the approximately 1,500 vehicles so far built could exceed £100 million.
Image credit: