Analysis and uncertainty of airport pushback rate control policies

Abstract

This thesis analyzes the effects of two algorithms that control the departure of aircraft at congested airports, with an emphasis on the uncertainty of the underlying processes. These algorithms, N-control and dynamic programming, belong to a broader class of control policies called Pushback Rate Control (PRC) policies that calculate a pushback rate for departing aircraft based on the state of the airport surface congestion. During times of congestion, these algorithms limit the amount of aircraft on the airport surface while maintaining departure throughput. This reduces the taxi-out time of aircraft, resulting in reduced fuel burn and emissions. This thesis introduces the policies and simulates their performance at LaGuardia Airport while varying two policy parameters, the length of the prediction interval and the number of prediction intervals, under several types of uncertainty, including the departure schedule and arrival rate. As will be shown, each policy results in significant taxi-out time reductions, saving airlines at least 60,000 minutes of taxiing over a 2-month period with the traditional 15-minute time window simulations. However, when accounting for the uncertainty in the algorithm inputs or the variation of policy parameters, the performance of both PRC policies degrades. By accounting for the variation of policy parameters and the different sources of uncertainty that affect airport surface management, the main contribution of this thesis provides a realistic analysis of PRC policies.