A mixed integer linear programming model for rolling stock rebalancing

Federico Farina, Dennis Huisman, Roberto Roberti, S. S. Azadeh

    Research output: Contribution to conferenceConference abstract for conferenceResearchpeer-review


    This work deals with the Rolling Stock Rebalancing Problem (RSRP) with deadhead trips. In the rolling stock circulation problem, given a set of timetables and the rolling stock material available, the various rolling stock units are assigned to each timetable; this plan also defines in which station’s inventory the different rolling stock units will be stored when not used. This daily plan is usually repeated each day of the week (with modifications during weekends or holidays). If the plan is repeated cyclically, then at every station the stored units at the end of the day need to match the required units at the beginning of the following day. Because of disruptions in the daily operations, planned maintenance in the network or others reasons, the timetables can deviate from what was originally planned therefore the balancing condition may be violated and there can be a number of off-balanced stations (with a surplus or a deficit of units). To rebalance the rolling stock in the network new deadhead trips, trips that run empty (starting from a surplus station A to a final station B where there is a deficit of those units), have to be scheduled in between the passenger trains by the railway operators. The novel contribution of our work is a new mathematical formulation that allows to route and schedule new deadhead trips in the network in order to solve the off-balanced stations. We will show that it can solve realistic instances on both the Dutch and Danish railway networks.
    Original languageEnglish
    Publication date2018
    Publication statusPublished - 2018
    EventEURO 2018 conference on Operational Research - Valencia, Spain
    Duration: 9 Jul 201811 Jul 2018


    ConferenceEURO 2018 conference on Operational Research


    Dive into the research topics of 'A mixed integer linear programming model for rolling stock rebalancing'. Together they form a unique fingerprint.

    Cite this