A Comparison of Two Exact Methods for Passenger Railway Rolling Stock (Re)Scheduling

In passenger railway optimization one of the main goals is to make efficient plans that can accommodate all passengers or, if that is not possible, that minimize the seat shortages. In the planning process, a railway operator tries to match the demand by first selecting an appropriate timetable followed by a matching rolling stock schedule. We focus on the second stage where a rolling stock schedule must be found given the timetable and passenger demand in this research. Not only the planning process, in which a long time ahead a rolling stock schedule must be determined, is considered, but also the real time construction of rolling stock schedules during disruptions. If during operations an unexpected event causes the timetable, the rolling stock, and crew schedule to become infeasible, then these schedules need to be rescheduled. To ensure that the operations are not deadlocked, new feasible schedules must be found promptly. For a general overview of models for railway rescheduling during disruptions and disturbances we refer the reader to Cacchiani et al. [7], and for an overview on disruption management processes in general we refer to Jespersen-Groth et al. [16]. The major difference between rolling stock scheduling in the planning phase and rolling stock rescheduling during disruptions is the time available to come up with a solution. Next to that, during the rescheduling phase there is less flexibility since the trains are already running, and choices made before the disruption occurred cannot be reversed. A natural consequence is that it may not be possible to assign rolling stock to all train services, i.e., some train services may need to be cancelled. Therefore in the rescheduling case it is highly undesirable, but considered feasible, to assign no rolling stock to some services. There is a body of literature tackling the problem of assigning rolling stock to passenger train services. These papers focus either on the planning process or on the rescheduling process. The majority of the approaches existing in literature considers one specific (and national) network, and does not benchmark against other approaches. In this paper we want to make a start with comparing