An adaptive large neighbourhood search algorithm for the post-disaster repair crew scheduling and routing problem with logistics support

Post-disaster road restoration is a race against time: crews must be routed to critical sites while construction materials arrive exactly when repairs can begin. Yet most studies plan routing and logistics separately, overlooking that repairs cannot proceed without on-site materials. This paper studies the post-disaster repair crew scheduling and routing problem with logistics support (PRCSRP-LS), which jointly plans crew routes and repair-material deliveries under binding site-level timing requirements. The objective is to minimise weighted inaccessibility time, prioritising early reconnection of isolated communities rather than merely shortening total repair time. We formulate a mixed-integer linear programming (MILP) model that captures selective repair, repair-material delivery timing, and evolving network connectivity. To handle realistic large-scale instances, we design an adaptive large neighbourhood search (ALNS) algorithm with a three-phase controller that tunes destruction intensity via Beta distributions and a perturbation–restart scheme to escape local optima. We also develop three problem-specific lower bounds to certify solution quality. Computational results show that the proposed ALNS produces executable restoration plans that tightly track the lower bounds, improving accessibility while maintaining logistical feasibility.