Skip to main content

Probabilistic Prediction of Aircraft Turnaround Time and Target Off-Block Time

Paper ID

SIDs-2023-26

Conference

SESAR Innovation Days

Year

2023

Theme

High performing airport operations

Project Name

Keywords:

Machine learning, Target Off-Block Time, Turnaround

Authors

Paolino De Falco, Jan Kubat, Vladimir Kuran, José Rodriguez Varela, Salvatore Plutino and Alessandro Leonardi

DOI

https://doi.org/10.61009/SID.2023.1.16

Project Number

Abstract

Collaborative decision making airports are extremely dependent on the precision of the target off-block time (TOBT) which is the target time set by an airline or ground handler agents for the off-block departure. This value reflects any delays that can be attributed to the aircraft operator or to the ground handling operations and must be updated by +/- 5 minutes when differing from the previous released value. Lastminute changes of the TOBT are undesired as they might alter the pre-departure sequencing resulting in very late air traffic flow management departure slots. Therefore, accurate predictions of turnaround times and last TOBT values are essential for better planning and tactical management of stands. This paper presents a set of probabilistic machine learning models to predict turnaround time and last TOBT values in nominal operational conditions at Prague, Geneve, Arlanda and Fiumicino airports. The turnaround models exhibit mean absolute errors ranging from 9 to 7 minutes during the strategic/pre-tactical planning phase, and from 6 to 4 minutes during the tactical planning phase. A validation exercise using ground handlers’ data shows potential benefits for airport operations. Finally, a model trained on all the available data from the four airports demonstrates the potential to generalise the approach without compromising the quality of predictions. Prague and Fiumicino airports will deploy the model in the operational environment during the first quarter 2024.