A robust optimization approach to reserve crew manpower planning in airlines

Airlines are continually faced with the challenge of efficient planning and utilization of their cockpit crew resources. Crew costs account for a major portion of airline operating expenses and easily exceed several million dollars annually for large air lines. Airlines invest significant resources in developing minimal cost monthly work schedules for regular crews to cover flying requirements. But in reality, a large portion of operational crew utilization depends on the planning process sequence and operational variability. The inherent variability and uncertainty in airline operations, combined with the lack of integration across planning modules, forces significant hedging through the use of reserve crew staff; in some cases airlines carry up to 30% of available personnel as reserves. In spite of maintaining reserve crews, airlines also depend on overtime flying (both voluntary and involuntary) by regular crews to cover flying during daily operations. Such large and costly reserve commitments, combined with overtime flying costs, reflect the lack of information available to planners due to the combined uncertainty of daily operations and complexity associated with incorporating such random variations in traditional optimization methods.; To address the entire planning problem, we propose an integrated optimization strategy for efficient utilization and minimal staffing of reserve crews. The primary objective is to reduce short term operating costs and long range staffing costs. Typically, the two main reasons that reserves are needed are: irregular operations when regular crews are not available to pick up disrupted trips; and conflicts caused by crew training and vacation leaving trips uncovered. The latter accounts for more than two-thirds of the total reserve demand. We propose a long range planning model that incorporates operational costs and utilization to estimate reserve staffing by building reserve utility functions. We then propose a mixed-integer optimization model to plan operational reserves and crew recovery strategies to control reserve utilization. Finally, we propose a new mixed-integer recurrent training optimization model which minimizes conflicting trips and maximize reserve crew utilization in-conjuction with the operational reserve planning model. Computational results are provided for all these models and the effectiveness of these strategies is discussed.