Airline integrated robust scheduling with a variable neighborhood search based heuristic

This study investigates an airline integrated robust scheduling problem, subsuming three steps: schedule design, fleet assignment, and aircraft routing, while considering the effects of propagated delays. We propose a mixed integer programming model incorporating propagated delays and flight re-timing decisions; the model is extended towards a second one in terms of passenger spill/recapture effects, optional flights, and passenger miss-connections. In order to solve our models for large-scale airline instances, we propose a column generation procedure as well as a sequential variable neighborhood search (VNS) heuristic; the integration of which leads to significantly improved convergence. Based on a real airline instance, from a Chinese legacy airline, we report an extensive set of numerical experiments to validate the effectiveness and efficiency of our models and solution techniques. Our results show that the VNS-based algorithms are significantly faster than the widely applied column generation algorithm for large instances. Sensitivity analysis further illustrates the benefits of the two integrated models in terms of revenue and propagated delays.