PhD-candidate: Danny Zhu
Start: Fall 2022
With millions of passenger train trips each day, railway networks play a big role in the mobility systems of many countries. To operate these trips, railway operators are tasked with solving several complex planning and replanning problems. One of these problems pertains to the planning of individual train units, or rolling stock. A rolling stock schedule specifies which rolling stock units operate which trip in the timetable. It is necessary that the rolling stock is efficiently operated and utilized to maintain a good balance between the service to passengers and the incurred operational costs.
On the day of operation, disruptions that range from a missing rolling stock unit in a train trip to a complete blockage of a railway track as a result of an accident or a train breakdown can occur. Consequently, it is possible that rolling stock is unable to follow the originally planned route, which implies that it cannot complete the remaining trips that it was scheduled for. Unavailability of rolling stock influences passenger satisfaction, as passengers possibly have to stand or even wait for the next train in case not enough units are available to facilitate the service.
Currently, the rescheduling of rolling stock after disruptions is mostly done by hand. Railway dispatchers manually try to find an updated rolling stock allocation without the use of any algorithmic support. This can largely be attributed to the difficulty of the problem; whilst various papers have formulated and solved the rolling stock (re)scheduling problem as a mathematical model, several assumptions are made to find solutions in reasonable running times.
In this research project, we aim to reduce the gap between theory and practice regarding the current methods for rolling stock rescheduling. In particular, we intend to add some aspects which are currently omitted in the literature, such as the availability of railway infrastructure and dynamic and uncertain information flows, to more closely resemble disruptions as they occur in reality. By doing so, we hope to contribute to the development of decision support tools that can be used by rolling stock dispatchers.
