Creating Initial Solutions for the Tail Assignment Problem

Abstract

One of many optimization problems in the airline industry is the tail assignment problem, i.e. to decide which aircraft operate which flight. Initial solutions can be used to warm-start optimization algorithms. In this thesis, the optimization algorithm uses a time window heuristic together with column generation. This thesis investigates different methods to create initial solutions for the tail assignment problem. The chosen methods consist of greedy algorithms and other simple heuristics. The goal was that the methods should assign at least 95% of the flights and this is achieved by most methods and test cases. Also, when using the produced initial solutions as input to the optimization algorithm, the value of the objective function is improved for some of the test cases.