Prediction of Off-Block Time Distribution for Departure Metering

Abstract

The uncertainties related to target off-block time (TOBT), the pushback-ready time predicted by aircraft operators, affect greatly airport operations. The accuracy of TOBT is, in general, difficult to be improved, because there are many uncertain factors in the departure process, e.g., delays in the passengers’ boarding. A better understanding of TOBT uncertainties, however, may help to improve airport surface operations. Currently, TOBT is estimated as a single point in time and updated as necessary by aircraft operators. Instead, the author proposes that TOBT is estimated as a distribution with a Johnson-SU distribution. The distribution parameters are estimated with time by neural networks using the history of TOBT updates. The main benefit of the proposed method is found in assigning the better pushback approval time of each departure aircraft for more efficient surface operations, which is demonstrated clearly by the simulation results. Using the proposed method, the aircraft operators can save fuel from improved ground operations via a probabilistic approach at the cost of reporting TOBT as a single point.