Rolling Horizon Approach for Aircraft Scheduling in the Terminal Control Area of Busy Airports

Abstract

This paper addresses the real-time problem of scheduling aircraft in a Terminal Control Area (TCA). At a congested airport, airborne decisions need to be taken regarding take-off and landing operations in given time horizons of traffic prediction. In this work, aircraft are rescheduled at holding circles, air segments and runways during disturbances in order to minimize the propagation of aircraft delays. We formulate this problem via an alternative graph formulation, i.e., a detailed job shop scheduling model of the aircraft scheduling problem. A rolling horizon framework is introduced to manage busy traffic situations with a large number of delayed aircraft. As scheduling algorithms, we compare a Branch and Bound (BB) algorithm with a First Come First Served (FCFS) rule that we use as a surrogate for the dispatchers behaviour. The solution algorithms are evaluated on practical size instances from Roma Fiumicino and Milano Malpensa airports, in Italy. Traffic disturbances are simulated for the assessment of the solution procedures. We distinguish between entrance aircraft delays and small random variations on the aircraft entry times in the TCA. Rolling horizon parameters are evaluated such as the extension of the current traffic flow forecast and the time shift between the start of successive traffic predictions. Experimental results demonstrate that BB solutions are of better quality compared to FCFS solutions. BB computes feasible schedules that better minimizes aircraft delay and travel time minimization. Also, BB requires less frequent changes of aircraft scheduling decisions between consecutive traffic predictions. The performance of BB is less affected by random variations of entry times.