Railway crew rescheduling for disruption management

Yuan, Jie (2023). Railway crew rescheduling for disruption management. University of Birmingham. Ph.D.

Preview

Yuan2023PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (6MB) | Preview

Abstract

Unforeseen events from external influences such as major weather events, and internal causes such as infrastructure failures disrupt daily dense train operations. Such disruptions can quickly spread over the network and cause planned crew schedules to become infeasible to follow. Being one of the important steps in recovery of the railway service following a disruption, if crew rescheduling is not properly considered, it can jeopardise the return to stable service. This thesis mainly focuses on railway crew rescheduling for disruption management.

This thesis studies real-time railway crew rescheduling in theory and practice. By carefully examining the current literature on railway crew rescheduling, this thesis presents a detailed analysis and comparison of the current methods. This thesis provides models and methods for the crew rescheduling problems caused by two distinct types of disruptions: minor disruptions and significant disruptions. Sensitivity tests are conducted on several parameters to explore the impact on solutions. Meanwhile, this thesis considers that optimisation tools for solving the railway crew rescheduling problem cannot be a standalone optimisation tool for controllers to use. If no solution and no further information is given by an optimisation tool, time will be wasted. If no feedback is given by an optimisation tool, there will be no resolution. When such situations occur, controllers usually do not know what has happened inside an optimisation tool and how to get potential solutions. A feedback mechanism is proposed to output the reasons for not producing solutions and to adjust parameter values used in the crew rescheduling problems to give a good chance of generating results with revised values.

A timetable rescheduling model is proposed to model the impact on train services of a disruption and predict the recovery period. The recovery period measures how quickly a timetable can return to its normal level. A disruption neighbourhood is introduced as an idea, which is used to identify the drivers that should be considered in the crew rescheduling model for significant disruptions. It is characterised by the drivers who are included in the model and the recovery period. Algorithms are proposed to find the drivers that should be considered in a disruption neighbourhood to obtain good solutions. Several mathematical techniques and methods are proposed to speed up the solution time for the crew rescheduling problem for significant disruptions.

Further, the integrated rolling stock and crew rescheduling problem is still an immature research area. This thesis presents detailed formulations to model the problem and explores this problem with retiming possibilities. It can provide mutually feasible rescheduling solutions between rolling stock rescheduling and crew rescheduling. Several goals that relate to rolling stock and crew during disruption management are considered, analysed and further grouped into different objectives. Two kinds of multicriteria decision making (MCDM) methods are used to produce a set of optimal solutions for the integrated problem.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):