Train scheduling with application to the UK rail network

Abstract

Nowadays, transforming the railway industry for better performance and making the best usage of the current capacity are the key issues in many countries. Operational research methods and in particular scheduling techniques have a substantial potential to offer algorithmic solutions to improve railway operation and control. This thesis looks at train scheduling and rescheduling problems in a microscopic level with regard to the track topology. All of the timetable components are fixed and we aim to minimize delay by considering a tardiness objective function and only allowing changes to the order and to the starting times of trains on blocks. Various operational and safety constraints should be considered. We have achieved further developments in the field including generalizations to the existing models in order to obtain a generic model that includes important additional constraints. We make use of the analogy between the train scheduling problem and job shop scheduling problem. The model is customized to the UK railway network and signaling system. Introduced solution methods are inspired by the successful results of the shifting bottleneck to solve the job shop scheduling problems. Several solution methods such as mathematical programming and different variants of the shifting bottleneck are investigated. The proposed methods are implemented on a real-world case study based on London Bridge area in the South East of the UK. It is a dense network of interconnected lines and complicated with regard to stations and junctions structure. Computational experiments show the efficiency and limitations of the mathematical programming model and one variant of the proposed shifting bottleneck algorithms. This study also addresses train routing and rerouting problems in a mesoscopic level regarding relaxing some of the detailed constraints. The aim is to make the best usage of routing options in the network to minimize delay propagation. In addition to train routes, train entry times and orders on track segment are defined. Hence, the routing and scheduling decisions are combined in the solutions arising from this problem. Train routing and rerouting problems are formulated as modified job shop problems to include the main safety and operational constraints. Novel shifting bottleneck algorithms are provided to solve the problem. Computational results are reported on the same case study based on London Bridge area and the results show the efficiency of one variant of the developed shifting bottleneck algorithms in terms of solution quality and runtime.

More information

Identifiers

Catalogue record

Export record