Optimal Aircraft Traffic Flow Management at a Terminal Control Area during Disturbances

This work addresses the real-time problem of managing take-off and landing operations in presence of traffic disturbances at a busy Terminal Control Area (TCA). The possible aircraft conflict detection and resolution actions are aircraft timing and routing decisions. An important objective of traffic controllers is the minimization of delay propagation, which may reduce the aircraft travel times and their energy consumption. To improve the effectiveness of air traffic monitoring and control in a busy TCA, we present an optimization-based decision support system based on a rolling horizon framework. The problem is modelled via an alternative graph formulation, i.e. a detailed model of air traffic flows in the TCA, and solved by aircraft rescheduling and rerouting algorithms. We compare a truncated branch and bound algorithm for aircraft rescheduling with fixed routes, a tabu search scheme for combined aircraft rescheduling and rerouting, and the first in first out (FIFO) rule that we use as a surrogate for the dispatchers behaviour. Computational experiments are based on practical size instances from the Milan Malpensa airport, in Italy. Disturbed traffic situations are generated by simulating various sets of delayed landing/departing aircraft and a temporarily blocked runway. We evaluate different parameters of the rolling horizon framework, such as the frequency of aircraft retiming and rerouting and the time horizon of prediction, i.e., the extension of the current traffic flow forecast, including roll and look-ahead periods. The roll period is the time shift between the start of successive traffic predictions. A detailed analysis of the experimental results demonstrate that the solutions produced by the optimization algorithms are of better quality compared to FIFO, in terms of delay and travel time minimization. However, the optimization approaches require frequent re-timing and re-routing in consecutive time horizons