Traffic disruptions: a dynamic heuristic approach

Abstract

A heuristic solution to rescheduling railway traffic, when an already planned schedule cannot be adhered to because of disruptions is presented in this paper. A mixed integer programming (MIP) formulation is developed to model disruptions that lead to conflicts on a single track bi-directional scheduled traffic with over-taking and follow-up capacities. The proposed objective is to minimise the total delay of all trains in the operational domain at their respective destinations. We have developed a rescheduling heuristic that re-sequences the conflict trains using a look-ahead-based dynamic ranking (LADR) approach. LADR procedure ranks the conflict trains on their static priority, myopic delay, critical ratio, number of future conflicts and total future delay. The heuristic procedure takes as input a conflict free schedule and is action-initiated by any disrupt to this schedule. The conflicting trains are identified and on the basis of their sorted LADRs departure times are assigned. Each conflicting train is rescheduled one at a time till all trains are rescheduled to their destinations. Our results are promising when evaluated against three other known procedures on a medium-sized real world data.