Project Summary
UK freight planners often rely on manual spreadsheet processes. They can spend weeks each quarter converting commercial demand into operational train plans while managing hundreds of variables. These include locomotive types, terminal operating windows, wagon compatibility, and customer cut-off times – a slow, labour-intensive approach that leaves capacity and revenue untapped. This project will develop an automated train planning system combining two core modules: a constraint satisfaction solver (a systematic algorithm that finds solutions meeting all operational requirements simultaneously) that rapidly generates feasible train plans, and an AI-based optimiser that fine-tunes these plans to maximise asset utilisation while minimising driver hours and operational costs. By automating this critical planning process and reducing planning time from weeks to hours, the tool will make rail freight more competitive, enabling better capacity utilisation and supporting the modal shift toward cleaner, higher capacity rail.
Project Achievements
The project developed a working prototype to automate key parts of quarterly rail freight planning, replacing slow manual spreadsheet processes with evidence-based optimisation. Using Freightliner’s real Q4 2025 operational data, the team built and validated a wagon circulation model at operational scale, covering 466 services across 21 terminals and producing results in under a minute. The model supports both feasibility checking and optimisation, including scenario testing and trade-offs between fleet size and leave-over wagons. It identified a potential reduction of 183 wagons, equivalent to 9.2% of the fleet requirement. A second joint model was also developed to route commercial demand across multi-leg services with time windows, and was successfully verified on progressively more complex test cases. The project established a strong academic-industry collaboration, created a solid foundation for future locomotive integration and full-scale validation, and demonstrated clear potential to improve efficiency, decision-making and decarbonisation in rail freight planning.
Conclusions
This project shows that rail freight planning can be approached in a more systematic, transparent and responsive way than manual spreadsheet workflows alone. Beyond delivering a prototype, the work clarified how complex operational knowledge can be translated into formal planning logic while still reflecting the realities of day-to-day practice. Close engagement with Freightliner was particularly important in ensuring that the research stayed relevant to operational needs and highlighted the importance of flexibility, scenario testing and practical usability. The project also established a credible pathway for further development. With additional commercial demand and depot data, the modelling framework can be extended to full joint validation and progressively integrated into industry workflows. More broadly, the work demonstrates the value of combining academic optimisation expertise with industry knowledge to address real transport challenges, creating a foundation for future deployment, continued collaboration and wider application across freight and other complex planning domains.
Next Steps
The focus of the next steps will be not only on technical development, but also on building a long-term collaboration. A key priority is to continue working with Freightliner to secure the remaining commercial demand and locomotive depot data needed for full joint validation against historical quarterly plans. This would provide a stronger basis for testing the tool in a form closer to operational use. Alongside this, the project provides a strong platform for follow-on funding and a future collaborative grant application. Potential next steps include a larger demonstrator or pilot project to support further model development, system integration and industry testing. This would enable features such as what-if scenario analysis, dynamic terminal rules and maintenance routing, while supporting a phased pathway into practice. More broadly, the project has opened up a promising route for continued academic-industry collaboration, with scope to extend the work through future research partnerships.
