Recovery Scheduling of Road Networks Considering Day-to-Day Flow Evolution

Natural disasters can lead to substantial disruptions in road networks, making many critical links unusable. It is important to timely repair the damaged links as they allow transportation of emergency services, relief materials, and so forth, after disasters. Many existing studies that focus on optimal recovery of damaged links after disasters assume that only a single agency is available for repair. Moreover, most of the existing studies do not consider travel time on links to be a function of traffic flow passing through the links and assume that the traffic flow gets distributed based on user equilibrium each time a link is repaired. However, such a traffic distribution is unrealistic as it assumes that the traffic flow remains the same across all the days for which a link is repaired and the traffic distribution gets suddenly modified whenever a link is fully repaired. The goal of this paper is to address these gaps in the literature of disaster recovery. We study the problem of determining the optimal repair scheduling of damaged links to minimize the sum of the total system travel time over the repair duration given that multiple repair agencies are available for recovery. Also, we consider a day-to-day traffic flow evolution where the route choices of travelers depend on the travel conditions of the previous day. We formulate this problem as a mixed-integer non-linear program. We proposed two solution methodologies to solve the problem: a genetic algorithm and a greedy algorithm. We tested these methodologies under different settings.