ADHERE & V/T SIC: Wheel/Rail Interface and Adhesion

Rail Engineer has been reporting on the activities of the Vehicle/Track System Interface Committee (V/T SIC) and the Adhesion Research Group (ARG) for many years. V/T SIC is a cross-industry body under the umbrella of the RSSB which aims to improve knowledge of, and recommend actions to improve, this vital interface. It was created following the derailment at Hatfield in October 2000 when a rail broke into many pieces caused by rolling contact fatigue.

The railway industry had realised it needed to understand this and other issues caused by train wheels running on the track. At the time, rail breaks were running at a rate of over 1,000 per year. Since then, great strides have been made but, as illustrated in this and accompanying articles, work remains and issues that were thought to be resolved, have re-emerged.

A particular wheel rail interface issue – adhesion – has been an issue since the earliest days of steel wheels on steel rails. The added factor of leaves on lines, subject to much ridicule in the popular press, but a safety risk nonetheless, seems to have become more serious since the end of steam and the introduction of mostly disc braked trains. Additionally, the objective of achieving a generally good ride seems to have made the situation worse. In the 1990s the Adhesion Working Group was established, and one of its successors, ARG, manages the ADHERE research programme and is a sub-group of V/T SIC.

Each year, their research and other investigations are presented to the industry. In recent years the seminars have been virtual or hybrid, but for 2024 the seminar was in person only, a move that was welcomed by both speakers and the 100 or so people who attended the day at RSSB’s offices in London. Product demonstrations and networking were much in evidence.

Presentations were a good mixture of research and updates on current issues, some covered here and others in their own articles.

Trailing arm bushes

Professor Jason Zheng Jiang from the University of Bristol and Dr. Heiko Müller from Gummi-Metall-Technik GmbH (GMT), presented their work on producing an improved trailing arm bush (see Trailing Arm Suspension above). This is a four-way collaboration between the University of Cambridge, University of Bristol, University of Huddersfield, and GMT, a supplier of railway trailing arm bushes.
The goal of the project is to construct a novel trailing arm hydro bush device and experimentally demonstrate its performance advantages over the current hydro bush for the Mark 4 coach. Although the work focusses on the Mark 4 coach, the developed technology will be widely applicable.

The prototype, the design details of which are confidential, has shown some very promising results during laboratory tests. It has delivered a 20% reduction in longitudinal static stiffness compared with GMT’s current design and 28% compared with the Mark 4 current design, delivering a modelled reduction in simulated track damage for larger radius curves (e.g., 1,200 metres) without harming forecast passenger comfort and stability, compared with the standard GMT bush. Lab-based accelerated life testing has also demonstrated adequate fatigue life of the bush comparable with the products already in use. Currently, preparations are being made to fit the new design to a Mark 4 bogie which has arrived at Huddersfield so that it can be tested on the University’s full-scale HAROLD rig.

A further design iteration is under way making use of the inerter principle originally developed at the University of Cambridge. An inerter is a two-terminal device in which the forces applied at the terminals are equal, opposite, and proportional to relative acceleration between the nodes unlike a conventional hydraulic damper where the forces are proportional to relative velocity. It is intended that the bush will incorporate inertance in the longitudinal and lateral planes.

The candidate design is forecast to deliver a 90% reduction in longitudinal static stiffness compared with the current design and is predicted to deliver much lower damage in 200-metre-radius curves and even lower damage on 1,200-metre curves than mentioned above. Passenger comfort is maintained at all speeds operated by Mark 4. The next step is to realise the outline design in hardware.

Axle bearings

A 10-car train will have at least 80 axle bearings, more if it has axle mounted final drives. Each one represents a single point critical safety failure risk. If the bearings are competently designed, lubricated, and kept free of foreign material, they are usually trouble free. That said, a single failed bearing can cause a major safety disaster. LNER’s Keith Mack who is chair of the RSSB sponsored Wheelset Management Group, and also chair of RAE/3/-/2 UK bearing group, presented its latest work. The group has representation from bearing manufacturers and overhaulers, lubricant OEMs, RSSB, passenger and freight operators, and Network Rail.

Keith explained that axle bearing failures in the UK remain low, but he emphasised the importance of proper investigation if one does fail. He referred to a failure on a class 390 in February 2021, and he commended Alstom’s investigation as best practice. Sometimes the damage caused by a failed bearing is so extensive that it is impossible to determine the root cause. In the 2021 incident, the root cause was contamination between the axle guide and the bearing on the outboard side during heavy overhaul.

To share this best practice, RSSB will be publishing Technical Note TN2315 dealing with evidence conservation of failed axle bearings in service, filling a gap that there is no industry recognised way of collecting evidence to learn from bearing failures, especially as priority is usually given to restoring the vehicle to operational service. It will set out a methodical approach to collecting data at the point of failure and preserve the failed bearings for further analysis to help prevent similar failures and it will support RIS-2709-RST ‘Rail Industry Standard for the Identification of Roller Bearings Defects’ which contains criteria for assessing bearing condition.

Finally, Keith stressed the importance of storing bearings correctly so that they remain fully lubricated and outlined some changes to improve the clarity of RIS-2704-RST ‘Rail Industry Standard for Wheelsets Handling and Storage’ which applies to handling and storage of wheelsets and bearings.

Wrong Side Track Circuit Failures

Autumn leaf fall is known to cause contamination of the railhead. A mixture of leaves, moisture, and rolling from wheels results in a black leaf layer. When the leaf film is wet, it is very slippery but, as it becomes dryer, it is an effective insulator leading to Wrong Side Track Circuit Failures (WSTCF). This is a significant safety risk and, before an assessment of comparative risk in the 2010s, the use of sand as an adhesion improver was restricted because of the perceived risk of sand insulating track circuits and causing WSTCFs. The comparative risk assessment showed that not stopping in time was a much higher risk than a WSTCF.

That said, the latter risk cannot be ignored and Dr Will Skipper from the University of Sheffield reported on ‘Understanding how Railhead and Wheel Contamination affects Track Circuit Performance’ (RSSB project T1222). Ten years of Network Rail WSTCF data was analysed showing where the incident happened, the type of track circuit and the cause (confirmed or inferred). Four main contamination causes were identified: leaf, rust, sand, and grease, with leaf the majority cause, bigger than the other categories added together. Incidents in October and November on DC track circuits were by far the largest individual entries.

Further analysis showed that the overwhelming majority of track circuits saw no failures, and very few saw more than an average of one WSTCF every two years.

Laboratory tests were carried out to understand the effect of three different pressures of 8 tonnes, 12 tonnes, and 20 tonnes, using leaf, rust, and sand contaminants. The results for leaf contaminant showed that, for the lower loads, the test rig surfaces remained isolated for the majority of the tests, whereas there was some conductivity for the 20-tonne equivalent load. For sand, the 12-tonne and 20-tonne tests demonstrated conductivity whereas the 8-tonne load showed a range of voltages illustrating that the electrical resistance in the contact area changed through the test.

Track tests were carried out on the Wensleydale Heritage Railway using a class 142 Pacer train. This is a two-car, two axles per car, circa 12.5 tonne axle load tare, i.e., the middle load from the laboratory tests. Genuine DC track circuit equipment was used. Tests were carried out using clean rail, a leaf layer, sand on clean rail, and sand on the leaf layer. For the leaf test there was a spike in voltage when the first wheelset entered the contaminated area and a smaller spike when both axles of the leading axle were on the contamination (leaf powder mixed with water, spread on the rail, and rolled in with two train passes).

For sand, there was a large voltage spike during the first pass but much less persistent on the second pass. The combination of sand and leaf resulted in longer and larger voltage spikes than either of the previous tests, the worst of both worlds.

‘Spikey’ traces

Further analysis explained some of the ‘spikey’ traces. In reality, the electricity feeding DC track circuit is actually a rectified AC voltage for DC track circuits, unless operated by battery. This is only apparent when measuring with a high sample rate. These AC voltages are further affected by the inductance in the coil of the relay, creating negative sections to the signal.

These behaviours were key for analysing the behaviour of the circuit when contamination was applied to the railhead. As the train enters the track circuit and the sander is activated, the electrical behaviour is different to a leaf contaminated contact. When viewed at high speed, as the sand is applied the ‘DC’ waveform is suddenly insulated leading sometimes to large peaks in voltage at the return voltage. This on/off type behaviour and the relay coil inductance in the relay causes the large increases in voltage – something that does not appear to be common knowledge within the signalling community.

As part of the track trials, images were taken by a drone operated by Network Rail at a height of over 50 metres above the track. These were of good enough quality to allow for the assessment of railhead contamination.

The study also examined WSTCF records on sample lines. The researchers sought to evaluate whether the results from the lab tests were reflected in service, especially as wheel-rail contact pressure was a very important factor in the laboratory. When analysing operations, load is easy to quantify but other factors, i.e., relating to contact area and position on rail, are harder to assess as this depends on wheel/rail profiles and running band. The link between train weight and WSTCF was apparent in headcode data.

Using traffic analysis and WSTCF results, Sheffield generated ‘heat maps’ illustrating the contribution of various factors on WSTCF risk: Traffic (frequency, type, mix}; track circuit type; rail head treatment train runs; and tree proximity.

For example, on the Bletchley to Bedford line (BBM), there was a mix of WSTCF causes that would need to be addressed, whereas on the line to Hull (HUL1) the analysis points to the dominant factor of track circuit vulnerability.

Comprehensive information about this work is available on the RSSB’s website.

Enzyme leaf fall treatment

An article in Rail Engineer 202 (May/June 2023) described work by Dr Leonardo Gomez, University of York, and Professor Roger Lewis, University of Sheffield, exploring the use of enzymes as a natural way of dealing with leaf contamination. There was a reminder that enzymes are biological molecules made of protein and they are powerful catalysts that work under ambient conditions (no high temperatures and pressures).

Because each enzyme is specific to its task, the potential for unintended consequences is minimal, they are water soluble, and enzymes have been developed and are produced at a large scale for industrial applications (e.g., food manufacture) and can be highly cost-effective and robust. Tests in the laboratory continue to determine the most effective enzyme and to understand the effect on the leaf layers. Further climate chamber tests are planned to assess different enzymes in different conditions of humidity and temperature. Full scale rig tests are also planned to assess how enzymes promote leaf layer removal by the passage of wheels. This research is still in its early stages.

Machine learning and images

Last year’s article also reported on the work of Morinoye O. Folorunso from University of Sheffield to use on board forward and downward facing camera images together with machine learning to estimate rail adhesion. The principle was demonstrated last year and, since then, work has proceeded with implementing a web-based tool that can present adhesion risk based on users’ input images, learning about unusual contamination that the machine learning needs to be trained for (for example, mud on the track at a crossing) and further full scale tests.

The capture unit was further tested on a Scotrail class 158 train which had a different lamp bracket mounting point to that used previously, giving the team a slight issue with alignment. Despite that, over 6,500 data points were collected. Further data was collected at the Wensleydale Heritage Railway. Tests included applying leaves to the track which the system detected.

The estimated friction levels were validated using a rail mounted tribometer with good correlation. Professor Roger Lewis explained how the camera box approach can be used in conjunction with new (or existing) railhead cleaning technology. This might enable a reduction in the amount of cleaning consumables required. He also outlined how this technology could lead to near real time adhesion mapping. He outlined a vision of the near future where there is integration between meteorological and environmental models and adhesion estimating using data from service trains to inform drivers, and longer term planning by operators and maintainers.

With thanks to RSSB’s Ben Altman and speakers at the seminar for their assistance with this article.

Lead image credit: Malcolm Dobell