Queuing Model for Taxi-Out Time Estimation

Abstract

Flights incur a large percentage of their delays on the ground during the departure process between their scheduled departure from the gate and takeoff. Because of the large uncertainties associated with them, these delays are difficult to predict and account for, hindering the ability to effectively manage the Air Traffic Control (ATC) system. This paper presents an effort to improve the accuracy of estimating the taxi-out time, which is the duration between pushback and takeoff. The method was to identify the main factors that affect the taxi-out time and build an estimation model that takes the most important ones into account. An analysis conducted at Boston Logan International Airport identified the runway configuration, the airline/terminal, the downstream restrictions and the takeoff queue size as the main causal factors that affect the taxi-out time. Of these factors the takeoff queue size was the most important one, where the queue size that an aircraft experienced was measured as the number of takeoffs that took place between its pushback time and its takeoff time. Consequently, a queuing model was built to estimate the taxi-out time at Logan Airport based on queue size estimation. For each aircraft, the queuing model assumes knowledge of the number of departure aircraft present on the airport surface at its pushback time and estimates its takeoff queue size by predicting the amount of passing that it may experience on the airport surface during its taxi out. The prediction performance of the queuing model was compared at Logan Airport with a running average model, which represents the baseline used currently in the Enhanced Traffic Management System (ETMS). The running average model uses a fourteen-day average as the estimate of the taxi-out time. The queuing model improved the mean absolute error in the taxi-out time estimation by approximately twenty percent and the accuracy rate by approximately ten percent, over the fourteen-day running average model.