Intelligent decision support for scheduling and rerouting trains on an Italian railway line

This paper deals with an operational problem faced by railway traffic controllers. During daily operations, trains need to be scheduled on a bidirectional single-track line with multiple stations to cope with various sources of disturbance that alter the existing timetable. We propose a decision support tool based on a detailed mathematical formulation of the single-line train scheduling problem. State-of-the-art scheduling and rerouting optimization algorithms are adopted to compute good quality solutions in a short computation time. Practical data from RFI, train delays and disrupted tracks are investigated in order to quantify the gap, in terms of delay minimization, between the solutions provided by optimization and by the current FIFO-based dispatching practice. The computational results show that advanced scheduling algorithms are useful when dealing with any kind of disturbed traffic situation, since train ordering decisions have to be taken at each station in both traffic directions. Differently, advanced rerouting algorithms are more suitable when mixed traffic is scheduled, since overtaking between trains travelling in the same direction is only favourable when trains travel at different speeds.