Integrated timetable rescheduling and rolling stock reassignment for an urban railtransit system during large-scale disruptions

Railway transport systems are crucial for the competitiveness of the national or regional economyand the mobility of people and goods. These systems play an increasingly significant role in urbanpublic transportation networks because of their high capacity, strict punctuality, and low energyconsumption. In actual operations, unavoidable disruption often happens and delays the originaltrain timetable, which creates difficulties in achieving punctuality goals. When trains are delayedfrom the regular operation, train dispatchers adjust the impacted train timetables fromperturbations (using train dispatching activities, e.g., re-timing, re-ordering, and re-routing). In thispaper, we studied real-time traffic management using a mixed-integer linear programming (MILP)model. Flexible stopping is innovatively integrated with delaying, canceling, and re-ordering toreschedule a timetable during railway disruptions. In the model presented in this paper, theassignment of rolling stock to courses is also considered. Flexible stopping means that theoriginally scheduled stops could be skipped for each train. In contrast, extra stops could be added,considering that during disruptions, a skipped stop could reduce the delays of passengers at theirexpected destinations. In contrast, an added stop could provide passengers with more alternativepaths for re-routing.A real-world instance of London's new Elizabeth line, UK, was used to test the proposed modelbased on several disruption scenarios. The Elizabeth line (formally known as Crossrail) will stretchmore than ۶۰ miles from Reading and Heathrow in the west through central tunnels across toShenfield and Abbey Wood in the east. The RAS۲۰۲۲ competition provides these probleminstances. We solved the model with the Gurobi solver. The computational results from the Gurobisolver show that the performance of the proposed model is appropriate for solving the problem.The model can solve all three evaluation problem instances with good quality within the time limit.Our experimental results found that delay is preferable to cancel a course in our model