Ant colony optimization for train routing selection: Operational vs tactical application

Railway traffic is often perturbed by unexpected events. To effectively cope with these events, the real-time railway traffic management problem (rtRTMP) seeks for train routing and scheduling methods which minimize delay propagation. The size of rtRTMP instances is strongly affected by the number of routing alternatives available to each train. Performing an initial selection on which routings to use during the solution process is a common practice to simplify the problem. The train routing selection problem (TRSP) reduces the number of routings available for each train to be used in the rtRTMP. This paper describes an Ant Colony Optimization (ACO) algorithm for the TRSP, and analyses its application in two different contexts: at tactical level, based on historical data and with abundant computation time, or at operational level, based on the specific traffic state and with a limited computation time. Promising results are obtained on the instances of the Lille terminal station area, in France, based on realistic traffic disturbance scenarios.