Improving the Efficiency of Heavily Used Railway Networks through Integrated Real-Time Rescheduling

Demand for rail travel has grown strongly within the last few years and growth is projected to continue in coming years. While central parts of the European rail network are operated punctually at its capacity limits, building new tracks is very cost intensive and thus only possible in exceptional cases. Adding new services to satisfy increasing demand is often achieved by reducing buffer times between consecutive trains. Buffer times are used to reduce the impact of train delays on overall reliability. While reducing buffer times increases capacity, it also means that small delays to a single train may propagate quickly through the network causing knockon delays to trains impacted by the delayed train. To handle these challenges, new railway operation principles are required. The thesis introduces new railway operation principles implying a new framework that combines rail traffic management by rescheduling in real-time with advice tools supporting drivers and guards in order to provide information allowing them to follow precisely the dynamically changeable schedules. Designed as a superimposition of two feedback control loops, the framework aims to increase capacity and stability in heavily used mixed rail traffic networks. The rescheduling part thereby assures that deviations or events causing conflicts are detected and solved automatically within real-time based on predefined optimisation criterions. These updated schedules are transmitted and visualised to drivers and guards who adjust their behaviour accordingly. Recommendations ensure that specified trajectories are followed accurately. The components to improve or develop as part of the new integrated real-time rescheduling framework are presented in this thesis. Possible designs and their consequences are pointed out. Specifications and requirements for components of sub-processes are also described in order to achieve an efficient and effective framework. With this new framework, rail traffic flow is improved and thus unnecessary time and energy consuming signal stops can be avoided. Consequently, delays and their network-wide propagation are significantly reduced. The thesis points out that enhancements of conventional railway operations must be done in three areas. Firstly, an uplink transmitting the train’s state, position, delay and event information regularly and more frequently than today must be set up. This is used to identify a deviation or event quickly as well as to predict the future behaviour accurately and is the basis for rescheduling. Secondly, generating feasible schedules within real-time for larger networks and high traffic density is another challenge. Although promising results have already been obtained, developing algorithms to generate new schedules in real-time will still need research. Thirdly, advice tools visualising the new schedules and the connection from the traffic management system to the train providing this information in real-time is required in order to operate trains accurately accordingly to new recommendations by drivers. A new network decomposition strategy where buffer times are minimised in capacity bottleneck areas and running time supplements are used in compensation areas in-between is proposed. The combination of an integrated real-time rescheduling framework with a new network decomposition strategy allows scheduled buffer times to be reduced in capacity bottleneck areas. As a result, infrastructure utilisation can be improved without having more delays than with conventional railway operation principles. A new tolerance band principle is also described. Using this, rescheduling as well as planning are more formalised. A train exceeding the given tolerance directly initiates new rescheduling, resulting in fast identification and a clear process for all actors. For scheduling, tolerance bands are added to minimal headways instead of allocating general, imprecisely planned buffer times. Also, open questions regarding inherent conflicts between rescheduling performance, nervousness, duration, and feasibility are exemplified. Finally, the additional benefits and limits of the new framework are derived and illustrated by simulations and case scenarios in the areas of Lucerne, Bern and Winterthur. It is shown that the knock-on delay of a single train applying the new framework may be reduced by up to several minutes if an unnecessary signal stop is avoided. For bottleneck areas where recorded delay patterns were applied, the thesis also shows that the overall delay may be significantly reduced with the integrated real-time rescheduling framework. However, precise operations following the reference trajectory is required. Test runs have shown that the robustness and reliability of the driver’s display tool are necessary prerequisites but difficult to achieve. Test trials with the new display tool proved that following a reference trajectory is possible. Nevertheless, the aspired level of accuracy only seems achievable when the reference speed is displayed. In particular, references with abrupt speed changes, which may occur to avoid conflicts as part of the rescheduling, will not be manageable without speed information. To summarise, the thesis demonstrates that with the combination of real-time rescheduling and precise train operations, the benefits of new traffic management systems are utilised and also significantly improved. Altogether, the integrated real-time rescheduling framework developed is a promising approach to improve the performance of the railway system both efficiently as well as effectively.