Airline Integrated Recovery

Air transportation delays have become commonplace in the aviation industry. While this is not surprising considering that the growth in demand of air transportation has outpaced that of capacity, it is of considerable cost. It is estimated that in 2007 alone delays have incurred an estimated cost airlines $40 billion. While the need to design an intelligent approach under irregularity is obvious, it is an extraordinarily difficult problem to solve given the complexity of operations. There are principally four components that comprise a recovery operation: repairing the schedule, rerouting individual aircraft, repairing crew schedules, and reassigning affected passengers to new itineraries.

Each of these four components of airline recovery is difficult to solve considering their size and complexity. Finding a feasible recovery plan alone is often a difficult task, let alone an intelligent one. The problem is exacerbated by the fact that airlines are bounded in the solution time as they would like to find a good solution as close to real time as possible. It is naturally ambitious to apply advanced optimization methods to solve the airline recovery problem. While there has been some work done in solving the given problem, the solution methodology is sequential, and thus suboptimal. Therefore there is an obvious desire to improve the existing solution by developing one integrated framework, referred to as integrated recovery. As each component of the problem is difficult to solve alone, naturally the integrated recovery model is a considerable challenge. While there has been some work in the realm of integrated recovery, to our knowledge there is not a single study that provides an attempt to computationally solve the problem.

The Air Transportation Laboratory (ATL) at Georgia Tech seeks to develop a comprehensive framework to use optimization in solving the airline integrated recovery problem. The ATL seeks to be the first to show computational results of this large-scale integer programming problem that minimizes aggregate passenger delay with crew considerations. We employ both column generation and cutting plane methods in a simultaneous fashion.

About Evan McClain