Rescheduling of light rail trains during disruption : an optimization model for Bybanen in Bergen

When a disruption occurs in an urban rail system, it usually results in significant disturbances due to limited operational flexibility. In this thesis, we develop an optimization model that efficiently reschedules trains during partial blockage on a double-tracked light rail line. The rescheduled timetable is obtained by a mixed-integer linear programming model that minimizes the sum of delay at all stations by rescheduling trains through the opposite track using crossovers. The numerical analyses are performed on three case studies based on real-world data from Bybanen light rail system in the city of Bergen. Our findings suggest that the proposed optimization model can safely reschedule train operations through crossovers located at their actual position in the network. Our findings also indicate that when minimizing delay at all stations instead of at the final stations, it contributes to more evenly distribution of passenger delay. This is demonstrated by comparing two different objective functions. The results furthermore imply that by increasing frequencies, a crossover strategy will be harder to implement following larger density of trains. Changing from manual to automatic crossovers seems to have little effect on rescheduling of train operations. When expanding to double-tracked crossovers, however, the results indicate that punctuality and train operations are significantly improved. Finally, as the optimization model solves the most comprehensive case study in six seconds, the model can be applied by dispatchers in real-time decisions