Optimized Restoration Schedule for Disrupted Railroad Network

Abstract

A method is proposed for optimizing the restoration sequence of damaged components in a disrupted rail freight network. Given the network’s demand matrix and capacity settings, a mixed-integer linear program (MILP) model is developed for assigning capacitated freight flows and minimizing total hourly costs. The cumulative cost increment (excess) is computed for each examined restoration sequence in a disruption scenario, using the duration of each restoration step and minimized hourly costs in intermediate network states from the MILP. A simple genetic algorithm (GA) is applied for finding the restoration sequence that minimizes the excess after a disruption event. A numerical case with a small network and a disruption scenario is synthesized to demonstrate this method. The optimized restoration sequence and schedule found by the GA are proved globally optimal in this case by exhaustive enumeration. The GA’s effectiveness is further verified with three additional disruption scenarios. A sensitivity analysis shows that the minimized excess is more sensitive to capacities of damaged components when capacity levels of components or upper limits of travel time are lower. The minimized excess is highly sensitive to the demand level when freight flows are moderately undersaturated. It is also found that restorations of damaged nodes should have higher priorities at lower capacities of damaged nodes, higher capacities of damaged links, and higher unit costs of alternate shipments by trucks.