Real-time train routing and scheduling through mixed integer linear programming: Heuristic approach

In railway traffic management, when an unexpected event perturbs the system, finding an effective train routing and scheduling in real-time is a key issues. Making the right routing and scheduling decisions may have a great impact on the efficiency of the system in terms of delay propagation. However, the time available for making these decisions is quite short: in few minutes a viable set of routes and schedules must be delivered to the dispatching system. In this paper, we assess the performance of a mixed integer linear programming (MILP) formulation exploited as a heuristic approach: we seek for the best feasible solution given a limited and predefined computation time. We run an experimental analysis on instances representing traffic in the Lille Flandres station, France. The results show that the approach tested is very promising, often finding the optimal solution to the instances tackled. Moreover, we show how the performance can be improved by tuning the parameters of the MILP solver.