In February 2022, some 70 delegates and speakers met at the Institution of Mechanical Engineers (IMechE) headquarters in Westminster to hear about managing fractures on the railway. Over the last three years there have been three issues with rolling stock cracks and fractures with at least one instance of significant service impact. This was the first opportunity to meet and discuss lessons learned. It was also the first face-to-face seminar at IMechE’s HQ since the Covid restrictions started in March 2020. By coincidence, the meeting took place on the day that the majority of restrictions ended.
The event covered principles, standards, case studies, and experience from aviation and track, as well as techniques for managing cracks. There was so much material that this article concentrates on an introduction to the issue of designing and managing rolling stock structures, and case studies highlighting what can go wrong. In the presentations there is a common theme of designs not matching their working environments because those environments were not fully understood; something that is very difficult to achieve. Two of the case studies are included here and the third, concerning Great Western Railway’s class 80X Inter City Express Trains, will be in the next issue.
David Crawley, managing director at Xanta Limited, explained the key factors in managing cracks and fractures based on his over 40 years’ experience in aerospace and rail. He made it clear that cracks in mechanical structures are to be expected. They arise from many factors including poor design, poor selection of materials, inappropriate specification, and operating conditions that vary from those assumed. General design criteria include appropriate strength and stiffness to provide the required operational duty, fatigue resistance to avoid initiation of failures, and toughness to avoid catastrophic failure. The designer needs to know the loading of the structure. The principal sources of rolling stock loads include tare and laden conditions, buff and draw loads, track induced loading, and crashworthiness.
David explained that these loads can sometimes be difficult to quantify, an issue that crops up in the case studies later in the article. He illustrated this with reference to London’s historic Hammersmith Bridge which opened in 1887 and which has been closed to motor traffic since April 2019. David wryly observed that Sir Joseph Bazelgette, the bridge’s designer, might reasonably not have designed for either today’s motorised traffic nor global warming in his original requirement set, and that he might have expected maintenance and modifications to be undertaken to cope with changing conditions. Designers tend to make conservative assumptions, but they are often not conservative enough and there can be obscure sources of loading such as harmonics and resonances.
Other things that can go wrong include poor materials selection or substandard material supply, misunderstood or incomplete application of standards, misunderstood load measurements, analysis errors, inadequate maintenance, and build technique creating residual stresses. The latter is hard to detect.
David said that we should expect cracks because often there are unknown and/unknowable factors that challenge design assumptions. The numbers of rolling stock designs (together with aircraft and bridges) are so small that each is essentially a bespoke design requiring new learning, and acquired learning is often lost in a roughly 20-year cycle. Another factor is that the longer the asset life, the more likely there will be changes affecting loading that invalidate original design assumptions.
So, cracks are common, but engineers and asset managers need to know which ones matter because failure to react could be unsafe, and managing them can lead to loss of service. It is important therefore only to deal with cracks that matter. The potentially dangerous ones on trains are those in the primary load paths; those where there is no alternative load path and where failure might lead to collisions, gauge infringements, and/or derailment. Structures include car bodies, bogies, and items mounted on them including doors.
David also highlighted some of the myths encountered when dealing with cracks such as ‘permanent’ weld repairs in parent metal (only suitable as a temporary repair) and drilling a hole at the root of the crack to stop the crack (rarely, if ever, successful). He stressed that a crack generally demonstrates that a design and its operating environment are not, or never were, compatible. One or both must change. In practice this means that the design must change to accommodate the environment e.g., different materials, more load paths, or different fastenings. Occasionally, allowing failure and repeated like-for-like replacement can be an economic solution. This involves either establishing the root cause of the problem or carrying out a redesign to render the issue impossible.
Finally, David said that he is often asked: “whose fault is it?” He made four points: (i) suppliers can’t be accountable for clients’ operating environments; (ii) clients may not know or have the competence to specify the operating environment; (iii) use of standards won’t capture reality; and (iv) condition tends to degrade with time. He said that collaborative testing, analysis, and commitment to maintain condition standards is a practical solution; true systems thinking! He also recommended reading the HSE document ‘Reducing Risks, Protecting People’, commonly known as R2P2. Echoes of David’s talk including residual stress, environment different to that assumed, and issues with design/manufacture all feature in the case studies which follow.
The Jubilee line 1996 Tube Stock was introduced in 1997 as a fleet of 59 trains composed of two three-car units. It was designed to allow fitment of an additional trailer car. In 2007 a 7th car, known as a special trailer car (SpT) was added to all trains and four additional seven-car trains were built. In 2011 the fleet was upgraded to Automatic Train Operation. The line operates between Stratford and Stanmore, with a service of up to 30 trains per hour and, pre-covid, carried over 200 million passenger journeys per year. Each train travels 175,000 km per year, and there are 441 cars in the fleet.
Case Study 1
Jubilee Line 1996 Tube Stock Inner Longitude/Drawgear Cracks.
Discussed by Steve Whysall former Principal Engineer, Transport for London (TfL); Loucas Papaloucas then Principles Engineer TfL; and Will Marshall, Jubilee Line Fleet engineering manager.
At 13:00 on 17 October 2019, a large crack was found on the inner longitude, the main structure that supports the drawgear, on a special trailer car (SpT) – see panel – during an underframe examination. This was immediately escalated to London Underground’s (LU) fleet engineering team. By 14:00, three more SpTs had been inspected and two more cracks had been found. This was a potential fleet-wide safety concern. By 14:30 the issue had been reported to the professional head of rolling stock engineering and, by 15:00, the issue had been escalated to senior management recommending that there be a controlled withdrawal of the fleet after the evening peak to allow a visual inspection and only trains meeting set criteria should return to service the next day.
By 17:00, inspection criteria had been generated for use by the train technicians who would carry out the inspection and, by 18:00, a Case for Continued Safe Operation had been produced; a document that argues why the approach is safe and what actions will be taken to control and reduce risk.
Steve Whysall and Loucas Papaloucas explained that the aim of the instruction was to remove all potentially unsafe vehicles quickly from service whilst minimising the overcrowding risk at stations, itself, a safety risk. For example, that evening most of the 15,000 audience at the O2 Arena adjacent to North Greenwich station would have been stranded with little alternative public transport options if the fleet had been withdrawn. It was also desired not to swamp the depot with trains which would have made moving trains around for inspection slower, but they wanted to provide confidence that trains offered for service the following day would be safe. The next day 10 of the 63 trains were stopped from service due to cracks.
The inspection process set out the types of cracks and lengths allowed. Inspectors were asked to identify the fractures by type, size, and severity, with data collected in a way that would allow the fleet team to monitor crack growth.
Will Marshall took up the theme of data collection. There were some 252 damaged positions which required frequent non-destructive inspections with the rest of the 1512 locations being inspected on a three-monthly basis. To minimise the number of stopped trains, a complex matrix of crack types, assessment criteria, action levels, and reinspection intervals taxed depot management. Managing 63 trains for a once-round inspection is one thing but managing the varying criteria made it much more complex. As a minimum, some 7,500 inspections per year were required. Will reported that he had to develop the record keeping system over three iterations before he was confident that he was fully able to track the various inspections and defects.
The immediate objective was to design repairs to keep trains in service whilst the root cause was sought. The most severe fractures were not permitted back into service until a temporary welded repair could be executed. Slightly less severe ones were allowed back in following fitment of a 5mm steel spreader plate between the drawgear bracket and the longitude, while some of the smaller, less critical cracks remained in service without intervention. A programme of work was set up to implement a temporary welded repair on all positions with cracks present, recognising that the low fatigue life would remain and cracks might return in a few years.
While all cars were affected, it became clear that the majority of cracks were at both ends of the SpT and the adjacent uncoupling non driving motor (UNDM). As part of the investigation into why the cracks/fractures were happening at all, the engineers investigated why the SpT and adjoining ends were more prone to failure than other locations.
The key factors included: no load cases/fatigue assessment for inter-car forces as a result of buff and draw, and poor interface design and management (stiff bracket, mounting surfaces not flat, fixing hole positioning) resulting in a deficient fatigue life (even for the former 6-car formation). For the 7th car project, the assumption had been made that the design was already proven. In fact, the introduction of the 7th car significantly reduced the fatigue life and the introduction of the higher performance associated with ATO further reduced fatigue life although to a smaller degree.
Another issue was the small number of occasions per trip where very high longitudinal forces were seen as the trains passed gaps in the current rails (it is LU practice not to have power bus lines along the train so each of the four power cars loses traction as it passes a gap delivering a compressive or tensile pulse to the couplings). Based on the measurements and assessments, the SpT to UNDM had a calculated fatigue life (years to crack initiation in 2.5% of the population) of 2.25 years, SpT to trailer, 8.7 years, compared with the similar, but six-car, 1995 tube stock of 11 years.
The repair work is ongoing.
Case study 2
Detached Yaw Damper Bracket, classes 195, 331. Discussed by Graham Taylor, Project Director CAF UK.
On Saturday 3 April 2021 a yaw damper bracket was found detached from the carbody of a power car of unit 195 121, a class 195/1 Northern DMU. At one end of the bracket the double C-section equipment mounting rail (C-rail) had fractured while at the other end the C-rail was intact, but the mounting bolts had fractured. The immediate reaction was to initiate an inspection of all the other class 195 trains and inform all other operators who might be affected. Graham said that this design is the same or similar on all the trains that CAF has recently supplied to the UK (Caledonian Sleeper mk 5, TPE mk5a coaches, Northern Class 331 EMU, TPE class 397 EMU, West Midlands class 196 and Transport for Wales class 197 DMU) and those trains were inspected too.
These inspections, which continue to the present time, show that only one bracket had become detached but a number of cracks in the outer section of the C-track were reported. As a temporary solution, a spreader plate was proposed. This increased the size of the load path from the yaw damper bracket into the C-rail and was secured to the C track with 12 bolts and much larger backing plates. This effectively made the mounting to the C-rail much stronger and bypassed the damaged section on those that had already cracked.
The existing yaw damper bracket was, in turn, bolted to the spreader plate. This made the yaw damper bracket 20mm lower than before, which was only a problem in the rare circumstance of a crush load with the air springs deflated. It was proposed to the customer on 15 May 2021, and the three affected trains were back in service by 31 May 2021, with further spreader plates available for use where required. Attention then turned to designing a bracket that restored the correct height.
Root Cause Analysis
It was clear that the loads being experienced by the C-rail were outside its ability to withstand them. A detailed assessment of the design and its environment has taken place. To inform the re-design, a detailed finite element sub-model of the yaw damper and anti-roll bar area was created, and lab tests were carried out to validate the model. An instrumented train was operated to assess the actual loads recorded in operation and these were found to be consistent with the design assumptions. Material specifications were checked, and manufacturing and assembly tolerances were also considered as were damper specifications. Finally, the various types of bogie – i.e. power (diesel), power (electric), and trailer – were considered.
Graham reported that only one crack has been identified on class 331/1 – four-car units and that cracks are concentrated in class 195 units (both two and three-car) and on class 331/0 – three-car units. Class 331/1 units run on different routes from the class 195 and class 331/0 units. This has led to the investigation exploring, so far without success, whether there is a critical point on one of the routes or depots seen only by Class 195 and Class 331/0 units that can produce this failure, such as a very tight curve that cause the damper to be fully opened or closed thus generating forces beyond the design proof loads.
A new design concept has been produced to cope with this condition, if indeed it is the root cause, which has not been proven. It is, effectively, a new yaw damper bracket integral with the spreader plate and is some three times longer than the original yaw damper bracket with the vehicle lifting pocket integral with the yaw damper bracket. The design has been conceived to redistribute the load introduced by the damper between four different ‘legs’, and 16 bolts connecting it to the C-rail. Load per bolt is reduced by 3.5 times compared to the original design. This reduces the stress to 1/5th of its former value giving an increase in forecast fatigue life in the C-rail from a nominal 55,000 miles to 170,000,000 miles. The new design is, however, somewhat heavier and Graham said that manipulators are also being designed to make installation of the new bracket easier.
This final solution is applicable to CAF UK Civity type EMUs and DMUs and the detailed design/change request was presented on 19 November 2021. At the time of writing, a design review was under way.
It is evident that these three incidents caused considerable disruption to either customers, routine maintenance operations or both. The requirements of railway legislation and standards encourage extensive assessment of the risks and likely consequences of structural failure during design, and there are well practiced arrangements for inspection and testing for cracks during maintenance. But could more be done? Should there be a periodic review of the environmental conditions that deliver most of the loads that train structures should withstand?
Such a review might have identified the Jubilee line cracks and at least the yaw damper/anti roll bar aspect of the class 8XX. But could a routine review have warned of the stress corrosion cracking issue or the problem with the CAF yaw damper bracket believed to be caused by over extension or compression of the damper? If we cannot catch all issues, perhaps reducing the frequency of surprises is a prize worth seeking. Rail Engineer will surely return to this topic.