Project overview
Building on the OCCASION project, DITTO continued the process of developing optimisation formulations, algorithms and processes that make better use of existing capacity without compromising service reliability. It was part of an industry-wide initiative called FuTRO (Future Traffic Regulation Optimisation) and was related to the development of in-cab signalling and the adoption of the European Rail Traffic Management System (ERTMS).
The project had the following four key components:
- Development of optimisation tools that maintain safe operating conditions and do not exceed theoretical capacity limits.
- Quantification of the trade-offs between the provision of additional train services and the maintenance of service quality so as to develop working timetables that optimise capacity utilisation without compromising service reliability.
- Combination of dynamic data on the status of individual trains to produce an optimal system-wide outcome in real time.
- Use of Artificial Intelligence to examine tractable solutions to real-time traffic control.
The project involved a consortium of three Universities:
- Southampton
- Swansea
- Leeds
with industrial support from Arup, Siemens Rail Automation and Tracsis.
The work at Southampton focussed primarily on computer modelling. Analytical methods were developed to calculate capacity utilisation indices and relate these to the propagation of delays, with encouraging results. A stochastic version of the job shop scheduling algorithm was developed in parallel, to optimise train timetables by identifying and adding additional potential train paths. A dynamic simulation model, Trackula, developed by the University of Leeds and based on their car following model, Dracula, was used to adjust train running speeds in real time. This micro-simulation was linked to a macro-assessment of the network, originating with solutions to the Multi-Commodity Network Design Problem.
The outputs of these tools were integrated, and some of the results were demonstrated in public domain software called OnTrack developed by Swansea University, the primary purpose of which is to undertake safety analyses. The potential benefits of Artificial Intelligence and machine learning applications in the domain of railway traffic management were assessed. For road traffic, such expert controllers often outperform existing algorithms. In such cases, machine learning tools can be used to produce new algorithms which can outperform human controllers over an extended period.
This work is part of the University’s Rail Research portfolio
