Two-level decomposition-based matheuristic for airline crew rostering problems with fair working time

Abstract

We propose a two-level decomposition-based matheuristic algorithm to solve a practical airline crew rostering problem with fair working time. The goal is to find an optimal assignment of pairings to individual crew members that satisfies hard constraints reflecting, for instance, international flights, rest days and regulatory requirements. The objective is to achieve a fair working time of crew members. We propose a two-level decomposition algorithm applying partial optimization under special intensification conditions (POPMUSIC). The method decomposes the original problem into a master problem and a subproblem. The master problem determines an assignment of pairings and rest days. The subproblem checks the feasibility of the original problem when the solution of the master problem is fixed. These problems are iteratively solved by embedding cuts into the master problem. A new method for solving the master problem by a generalized set partitioning reformulation is proposed. The effectiveness of the proposed method for real-world data is shown via computational experiments.