Poor ride a cure for hunting?

In Rail Engineer 208 (May-June 2024), we reported that Dr Mark Burstow, Vehicle Track Dynamics Engineer at Network Rail, had identified hunting as one of the causes of poor ride on some fleets resulting from high equivalent conicity between wheels and rails – a system problem. Although not all poor ride issues are the result of hunting, having identified this root cause, the next challenge was to do something about it.

This article covers what is being done on the Great Western main line which can potentially be rolled out to other fleets, as well as work by a Network Rail engineer exploring methods of evaluating track features contributing to high equivalent conicity.

Recap

Hunting is usually triggered where the equivalent conicity is high, and this tends to be at locations where the gauge is tight, but it would be wrong to think that this is solely a track issue. There are two factors contributing to ‘tight gauge’.

Standards define the gauge and wear limits for wheels and rails, but hunting can occur when both wheels and track are within these limits. There are some locations where the gauge is a nominal 1432mm (within allowed limits) rather than the normal 1435mm, a hangover from a standard that was superseded in the 1980s. But wheels can and do contribute to tight gauge.

When running on straight track there is necessarily clearance between the rail gauge faces and the wheel flanges. On older trains, as wheels wear rail to flange clearance might be maintained or, as the flange wears, might effectively slightly widen the gauge. Modern bogies can run for hundreds of thousands of miles with their wheelsets remaining within allowed wear tolerances.

But, as the wheel tread wears, the flange can become effectively wider (illustrated left), reducing the clearance between the flange and rail gauge face, effectively needing to run on track with a gauge of 1436mm or 1437mm to maintain the clearance that a new wheel would see on nominal 1435mm track.

Moreover, this sort of tread wear tends to lead to a steeper transition from tread to flange, leading to higher equivalent conicity and hunting, sometimes even on nominal track. That said, as Dr Burstow reported, wheels that have run for over 300,000 miles and which are still within flange height limits, width limits, and are free from RCF or other surface defects, are a tribute to the bogie designer. Also, note how quite small dimensional changes can have a significant impact.

Accelerometers

Rail Engineer’s interest in a further article was prompted by news that Hitachi is fitting HMAX accelerometers (formerly Perpetuum Onboard) to all of the axleboxes on Great Western’s Inter City Express Trains (IET) and this will be extended to other IET fleets. The purpose is to help optimise axle bearing and wheel maintenance (see Panel 1), but Rail Engineer wondered whether the lateral accelerometers on the HMAX product could be used to detect the onset of hunting before it becomes apparent to the customers. We learned that is exactly what is being done.

Clive Burrows, group engineering director at FirstGroup, took the lead. He said: “Excellent collaboration, sharing of data and trust have been the essential elements of bringing together the knowledge, experience, and insight we now have from these impressive data collection and analysis systems.  It has been a real pleasure for me to coordinate and steer the work of this cross-industry team that includes Hitachi Rail, Omnicon Balfour Beatty, Network Rail Wales & West, Network Rail Technical Authority, RSSB, GWR, MTR Elizabeth Line, and FirstGroup.

“I am confident this work will enable us to improve the way we manage our railway assets in a more efficient and effective way.  After all, the wheel-rail interface and its associated system must be one aspect of engineering that is almost unique to railways.”

Clive added that it is the Hitachi Rail system and Omnicom Balfour Beatty Unattended Geometry Measurement System (UGMS) (see panel 2) being used together that provides the most useful information.

For example, UGMS outputs enable track gauge to be compared with the HMAX outputs to identify precise locations where narrow gauge is causing lateral ride instability. The investigation team has been supported by experts including Network Rail’s Dr Burstow, RSSB’s Professor Bridget Eickhoff, and Hitachi Rail’s Dr David Vincent.

When Rail Engineer discussed this with David Vincent, it was apparent that the work is not yet a routine activity but is showing that:

Analysis of data showed that the accelerometers picked up signs of the onset of hunting when wheels have run around 230,000 miles.

While it might be thought that turning wheels at a lower mileage is a disadvantage compared with the >300,000 miles routinely achieved, Hitachi is already demonstrating that turning at 230,000 miles whilst still well within limits improves wheel life because less metal is removed on each turn.

When reports are received of ‘rough ride’, Hitachi is quickly able to show if it is a train issue (high conicity needing wheel re-profiling), helping Network Rail to manage the event more efficiently.  Currently, when rough riding is reported by staff or customers the standard response is to ‘stop and caution’ until the track has been inspected.  With monitoring, wheels are re-profiled before there are any reports of poor customer comfort related to high equivalent conicity.

Equally, if the system shows many trains experiencing “rough ride” (of which hunting is one example) at the same location, this points to a possible track issue.

The system can detect the issue before the customer notices.

David added that much of the data analysed so far uses the body-mounted accelerometer which is not fitted to all cars, but even earlier detection is likely to be possible if the axlebox lateral accelerometer signal is used.

Next steps include setting all this out in systems, process, limits, and training so it becomes business as usual, extending fitment to all the other IET fleets and working with other Train Operating Companies and other regions of Network Rail to embed the approach being piloted on GWR.

Optical measurement

Clearly, hunting can be detected if the fleet has axlebox accelerometers, but what of fleets that are, perhaps somewhat older and are not so equipped? As stated earlier, reports of hunting and rough riding might lead Network Rail to impose a speed restriction and also send resource to site to inspect and measure profiles and gauge. Clearly this requires a line block and people proficient in taking very accurate measurements (often in the dark) and assessing the results. But what if this could just be data/information obtained from one of the vehicles that Network Rail uses routinely to measures its track?

An approach to evaluating equivalent conicity from train-borne rail measurements and reference wheel profiles was explored by Network Rail’s Chris Fuller in his MSc thesis at the University of Birmingham. His work compared on-site measurements made using Miniprof instruments against profiles measured by one of Network Rail’s Ultrasonic Testing Units (UTUs).

Assessment of equivalent conicity also requires track gauge which is measured by Network Rail’s Track Recording Vehicles (TRV).  The work showed that the UTU profiles are of sufficient quality to be used to assess equivalent conicity, albeit some smoothing was required to remove the slight variability (or glitches) in the optical measuring method. The UTU provides a rail profile every two metres – much shorter intervals than is practicable to do manually. It is important to have accurate gauge results from the TRVs and one of Chris’ recommendations is to validate the geometry channels from its TRV fleets against a known track datum. Another issue is that the calculation of equivalent conicity is mathematically complex.

Chris also investigated the ‘quick conicity’ (see Panel 3) assessment method to determine whether particular characteristics of rail profiles or track geometry could be identified that were likely to contribute to high equivalent conicity. He was able to make recommendations which might lead to a ‘conicity indicator’ channel on the UTU’s recordings, assuming that reliable track gauge data can be obtained (e.g. adding it to the UTU). Much more work is required but the promise is that Network Rail and the train operators will be able to set maintenance intervention limits (e.g., rail shape/gauge and wheel mileage) that virtually eliminate hunting.

Conclusion

Hunting is a system issue which requires joint understanding and action, this article showing the benefits to be obtained from such cooperation. The deployment of both on-train technology and newly developed assessment techniques offers much promise and is welcome. As well as improved ride for passengers, there is a potential reduction in maintenance costs for both trains and track which could be applied across the UK network, hopefully relatively quickly and with minimal disruption. There is more work to be done in fully scaling all this including ensuring that data from multiple sources can be integrated into useful information that operators and maintainers can respond to effectively.

Image credit:

Panel 1: Origins of fleet fit axlebox accelerometers

Axle bearings tend to have a good long life but can occasionally fail prematurely. They are single point failure, safety critical components. Bearing and rolling stock manufacturers generally specify a conservative life/maintenance regime, yet most bearings cleaned and examined at this life show no signs of distress. Indeed, your writer has experience of fleets where the overwhelming majority of the bearings last the life of the fleet. If a bearing is soon to fail, it generally gets hot, and all over the UK main line, Hot Axlebox Detectors (HABD) are fitted to identify this serious imminent failure. If one of these is triggered, the train must be stopped and examined immediately.

Detecting imminent failure is not good enough in a number of industries; much more notice is required. Engineers thought this principle could be applied to rail and that likely failure could be identified before such a fault becomes service affecting. Roller bearings exhibit vibration patterns related to factors such as the number of rollers and the bearing speed. If these vibrations are monitored, changes in the vibration pattern from, for example, pits in rollers or spalling in the races, will indicate deterioration, giving plenty of warning before the bearing finally fails.

This approach was adopted by SouthEastern Trains over 10 years ago when it fitted Hitachi Rail’s Perpetuum self-powered tri-axial accelerometers/RF transmitters to its 170-strong Electrostar fleet (other axlebox vibration monitoring systems are available). This involved fitting eight axlebox accelerometers and a body mounted radio receiver/data concentrator (some with their own accelerometer, giving the ability to monitor ride comfort) to each of its 680 cars.

The self-powered/radio feature minimised the amount of wiring. The aim was to extend the intervals between bogie overhauls and eliminate the risk of premature bearing failure. As wheels on rails are forms of rolling element bearings, albeit rather large ones(!), it was soon realised that wheel flats and other wheel defects could be monitored by the system. Rolling forward to 2024, SouthEastern Trains:

Plans wheel turning based on reported condition from the Perpetuum system, identifying wheels that need reprofiling even though many defects are invisible to the naked eye.

Identifies wheel flats as soon as they occur providing time to plan their rectification.

Avoids trying to ‘roll out’ flats which just stores up greater problems later on, sometimes avoiding the need for a ‘balancing turn’ on other wheelsets on the same car.

Achieves longer wheel life. With reprofiling being carried out when the system identifies the need, less metal is removed (5mm compared with 10-15mm for wheels managed visually). In general, when presented for reprofiling as part of planned routine maintenance based on vibration analysis, the wheels look in good condition and are well within the flange thickness/flange height limits allowed.

Has moved largely to planned wheel lathe work enabling a significant saving on avoiding having to purchase one new wheel lathe – a saving that more than paid for the installation of the sensors.

Panel 2: Unattended Geometry Measurement System

A number of Intercity Express trains have been fitted with Omnicon Balfour Beatty’s Unattended Geometry Measurement System (UGMS) system. It does the same job as the track geometry measurement system on Network Rail’s New Measurement Train (NMT), but a UGMS train might run over a section of track perhaps a couple of times a day whereas the NMT only covers each section once every eight weeks. Apart from the benefit of providing an accurate gauge measurement when hunting is detected on the train, UGMS provides Network Rail engineers immediate validation that repairs have achieved the required geometry improvements and can provide information about the rate of geometry degradation. Each train has had to be individually calibrated to demonstrate to Network Rail’s Technical Authority that its data/information is accurate.

UGMS comprises the following key elements:

• A Main Processing Unit housed in a 19-inch rack located behind the drivers cab of the leading vehicle, including means to transmit data wirelessly.
• A pair of optical/inertial units attached to the trailing bogie of the leading vehicle the unit. These house the cameras, lasers and six-axis inertial transducers.
• A speed probe, independent of the train’s probes, fitted to the axle box housing and connected to one of the optical/inertial units.
• An independent Audible Warning System detector mounted onto the bogie frame on the track centre line and connected to one of the optical/inertial units.
• A tri-band (GSM/GPS/Wi-Fi) antenna.

Panel 3: Quick Conicity: the Burstow method

In research published in 2016, Dr Mark Burstow (Network Rail) and Andreas Haigermoser (Siemens) proposed and evaluated a ‘quick conicity’ measurement by measuring gauge between the two rails at different points across the rail head.

Whilst tight gauge contributes towards an increase in equivalent conicity, they concluded there was a poor correlation when using gauge values measured 14mm down the gauge face of rail, the usual measuring point for track gauge. This is because it provides no indication of the rail shoulder height. Instead, they proposed taking track gauge measurements 3mm below the crown of the rail. These measurements identified rail profiles with a high gauge face shoulder caused by uneven wear, a feature which can contribute to a high equivalent conicity and dynamic wheelset instability, particularly when encountered by wheel profiles with a high wear resulting in a thicker flange root area.

Image credit: Hitachi Rail