Airline planning is a complex and difficult process. The biggest airlines in the world plan for and operate fleets of over 700 aircraft using tens of thousands of crew members. As such, small percentages in savings translate to millions of dollars. In this thesis, we study the pairing and duty generation problem in the context of airline crew scheduling, and propose approaches to improve the computational speed and the solution quality. We propose several enumeration algorithms to generate all possible duty periods of a given schedule to improve on the time required to generate duty periods; and present a set of column generation models to improve on the solution quality. When tested on a real test case study, the proposed approaches are found to improve the computational times from 142 seconds down to less than one second, and the cost savings of 13.7%.

Airline planning is a complex and difficult process. The biggest airlines in the world plan for and operate fleets of over 700 aircraft using tens of thousands of crew members. As such, small percentages in savings translate to millions of dollars. In this thesis, we study the pairing and duty generation problem in the context of airline crew scheduling, and propose approaches to improve the computational speed and the solution quality. We propose several enumeration algorithms to generate all possible duty periods of a given schedule to improve on the time required to generate duty periods; and present a set of column generation models to improve on the solution quality. When tested on a real test case study, the proposed approaches are found to improve the computational times from 142 seconds down to less than one second, and the cost savings of 13.7%.