Project Summary
This project will, in collaboration with Freightliner and Network Rail, develop a prototype of a system that evaluates the braking capability and emissions of freight trains in real-time using signal-system data and stores the results. The system will also show the results of currently running trains – as well as those in the past- of any part of the Great British rail network on a web-based dashboard.?Each freight train has a unique weight and a mix of wagons, so its braking capability (and emissions) greatly varies. Traffic control and signal systems use a fixed minimum assumption, thereby forcing freight trains to sometimes run unnecessarily slowly, increasing the number of stops (which leads to more emissions), and losing line capacity. Building on the AI-based technique developed by UCL, the project targets a 4% reduction in the required headway for freight trains, which leads to an increase in the capacity of busy railway lines, thereby promoting modal shift to rail and decarbonisation.
Project Achievements
Optimising rail freight is a complex challenge, but this collaboration between UCL and Freightliner provides a data-driven solution. By leveraging an AI-based prototype, the project produced a prototype of a machine learning-based system that calculates braking Force and emission from the train operation records, moving away from “fixed minimum assumptions”—generic braking standards that currently force freight trains to run slower than necessary—toward a more precise, train-specific approach. Our approach, which combined physics models and data-driven approaches, ensures the system remains transparent and safe for rail operations while providing more accurate real-time performance estimates.
Conclusions
We have developed the system in collaboration with Freightliner. In the end, collaboration went very well and we are hoping to continue this collaboration, while aiming other freight rail companies, rail infrastructure managers, signal system suppliers, and policy makers as potential customers for this product. The protype has been tested and is, we believe, at TRL 6 now.
Next Steps
We are working to set up a company to commercialise this product. We also have other ideas that derive from this project but addresses the same theme: how to extract, from passively collected data, business critical information that helps the railway companies improve their operation and increase revenues. In parallel, we want to increase the TRL of the developed product. Using the prototype, we will engage with a wide range of railway stakeholders.
