Each year, the Institution of Mechanical Engineers’ (IMechE) Railway Division holds a competition of technical presentations called Future of Rail. The competition is open to all young railway engineers including apprentices, undergraduates, and those in the early stages of their career. Entrants present to a panel of judges and other attendees in a competition format at the home of the IMechE in the heart of Westminster, with the winner taking home £1,000.
This year’s winner was Mujadded Al Rabbani Alif, a research associate (computer vision) at the Institute of Railway Research, University of Huddersfield. His topic, Automated Wheelset Inspection, involved using mobile phones for freight wagon maintenance. Rail Engineer has previously reported on the rail freight industry’s desire to improve wagon maintenance in cost effective ways, and this was one potential example.
Wheel condition is critical for safe operation and wheel wear can affect stability, ride quality, and, in extremis, derailment. Regular inspection is required but currently is often carried out using manual methods in sidings, sometimes in the dark by using handheld gauges which must be placed on the wheels. This generally involves working under the wagon with the maintainer laying on the ballast. These results are then recorded on paper.
So how could this be improved? Almost everyone carries a powerful computer, with a high-resolution camera, that can record its geo-location and has a torch. It’s called a mobile phone. Could these be used to measure wheels? Could image analysis replace manual gauging for routine inspection? And, if so, could it be done accurately in real depot and siding conditions?
The answer to all of these questions is “yes” and Mujadded described how.
Four step process
The proposed process involves four steps: (i) the operator captures a short video of the wheel using a mobile phone; (ii) this image is sent to a server which automatically identifies the wheel in the video and (iii) digitally calculates the key wheel dimensions; finally, (iv) the dimensions are sent back to the mobile phone and are stored and tracked over time.
The process was tested on both loose wheelsets in depot and those under wagons with over 100 wheels being inspected for evaluation. Mujadded demonstrated a variety of video capture types including vertical and horizontal camera movement so that the wheel shape was shown.
Comparisons were made using the industry standard Miniprof tool; a time-consuming process compared with the mobile phone. Analysis of the camera and Miniprof measurement showed differences of less than one millimetre.
Video capture and analysis take place in about 10 seconds (subject to a decent 4/5G or WiFi connection). It was clear from the videos shown that there is a certain technique in capturing decent video. It is not onerous but worth talking about (see later). Moreover, the system rapidly responds if the analysis fails or gives a nonsense result, so the measurement can be repeated.
This process enables faster and simpler wheel inspections which can be carried out more frequently, and improved tracking of wheel wear over time supports condition-based maintenance decisions. If applied to rail freight there is an opportunity to extend wheel life from better monitoring (reprofile wheels at the right time to minimise the amount of metal removed during turning). This would help reduce casualty wheel turning and help reduce the rail freight industry’s maintenance spend of approximately £60 million per annum.
Mujadded said that the learning points from the project so far include coping with real depot conditions which introduce lighting and perspective challenges to reliable wheel detection. This is essential before measurement and data collected in operational environments can be used in routine maintenance and monitoring.
Future progress
The project is showing that the technique has great promise. The next phase will focus on expanding testing across a wider range of wheel types, lighting conditions, and operational railway environments to further improve system robustness and reliability. Future development will also include an image and video capture guidance system designed to help users record the wheel correctly by providing guidance on camera positioning, viewing angle, and movement during capture. This is intended to improve measurement consistency and make the system easier to use in real maintenance conditions.
Additional work will explore integration with freight maintenance and asset-management systems to support digital tracking of wheel wear and condition-based maintenance strategies. The project team has also applied for further funding to progress the technology towards Technology Readiness Level (TRL) 6. TRL 6 is the stage where a concept transitions from a laboratory validation to a fully functional, high-fidelity prototype. At this level, the technology is rigorously tested in a ‘relevant’ or simulated environment to demonstrate its engineering feasibility and readiness to support larger-scale industry testing in operational railway environments.
A key objective throughout the project has been simplicity and accessibility. The intention is not to replace specialist inspection systems entirely, but to provide a low-cost and scalable solution that enables more frequent inspections using mobile devices already widely available to maintenance teams.
Highly commended
Second place went to Danny Evans, overhead-line asset engineer (maintenance support) at Network Rail (North East), for his presentation describing how drone-based photogrammetry and 3D point cloud data can be used to drive overhead line asset management and improve maintenance capabilities to deliver safety and performance improvements.
Third place went to James Fitzmaurice, Class 707 Fleet Engineer at Angel Trains, for his presentation about the role of an owner’s fleet engineer, including retrofitting toilets to this nearly new fleet.
The judging panel was chaired by Alice Callaghan who is chair of the Railway Division Young Members’ Committee and whose day job is with TransPennine Trains. The event was organised and compered by Ellie Smith from Transport for London.