Practice Oriented Algorithmic Disruption Management in Passenger Railways

How to deal with a disruption is a question railway companies face on a daily basis. This thesis focusses on the subject how to handle a disruption such that the passenger service is upheld as much as possible. The current mathematical models for disruption management can not yet be applied in practice, because several important practical considerations are not taken into account. In this thesis several models are presented which take important practical details into account:
(1) Creating a macroscopic global feasible solution for all three resource schedules, instead of focussing on one individual resource schedule.
(2) Scheduled maintenance appointments required by certain rolling stock units are included while rescheduling.
(3) Dead-heading trips to transfer rolling stock units from stations with a surplus of inventory to stations with a shortage of inventory.
(4) Adjusted passenger demand, the passenger demand is not static, but depends on the capacity appointed to the previous trips. Finally,
(5) Checking whether a rolling stock circulation is feasible with respect to the available depot tracks (the shunting yard) within a station.
We make use of different techniques to solve the models, for instance, mixed integer linear programming, column generation, constraint programming, and heuristic models are used in this thesis. The results demonstrate that these five practical considerations can be taken into account in the disruption management models.