Evaluation of improved pushback forecasts derived from airline ground operations data

Accurate and timely predictions of airline pushbacks can potentially lead to improved management of airport surface traffic, including reductions in the variability and average duration of costly airline delays. One factor which affects the realization of these benefits is the level of uncertainty inherent in the turn processes. Novel analyses of the minimum inherent uncertainty yield a significant lower bound on the predictability of airline pushbacks under the best possible conditions. These analyses are based on a large and detailed dataset of approximately 104 real-world turn operations obtained through collaboration with Deutsche Lufthansa AG. Three techniques are developed for predicting time-to-go until pushback as a function of available ground time; elapsed ground time; and the status (not started, in progress, or completed) of individual sub-processes in the turn such as catering, fueling, etc. This lower bound result shows that airport surface traffic management must incorporate robust mechanisms for coping with pushback demand stochasticity. These results also characterize the forecast horizon over which pushback predictions are accurate. Copyright © 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.