Models and local-search algorithms for integrated airline scheduling, fleeting and pricing Seminar

We discuss about models and algorithms for an integrated airline scheduling, fleeting and pricing problem. The integrated model simultaneously optimizes the decisions of schedule design, fleet assignment, seat allocation, pricing and considers passengers' spill and recapture. The resulting problem is a mixed integer non-convex problem due to the explicit representation of a demand model integrated with a supply planning model.

We present a local search heuristic to tackle the complexity of the problem based on a decomposition of the problem. The first model is a fleet assignment model where the pricing decision is fixed. The fleet assignment sub-model is a mixed integer linear problem. The second model is a revenue management model where the fleet assignment decision, i.e., the transportation capacity, is fixed. This revenue sub-model is a continuous nonlinear problem. These sub-models are solved iteratively with ad-hoc local search mechanisms. The local search heuristic is presented in comparison to two other heuristic approaches: a heuristic procedure provided by an open-source generic MINLP solver and a sequential approach which mimics the current practice of airlines. The three approaches are tested on a set of experiments with different problem sizes. The local search heuristic outperforms the two other approaches in terms of the quality of the solution and computational time.