Forecasting of taxi times: The case of Barcelona-El Prat airport

Citation

Lordan, Oriol; Sallan, Jose M; Valenzuela-Arroyo, Marta (2016). Forecasting of taxi times: The case of Barcelona-El Prat airport. Journal of Air Transport Management. 56(B), 118-122.

doi: http://dx.doi.org/10.1016/j.jairtraman.2016.04.015

Abstract

One of the challenges that air transport management is facing is to develop predictive tools for ground operations of aircraft, in particular of estimation of taxi times. The aim of this paper is to define a forecasting model for taxi times for a specific airport: Barcelona-El Prat. This model uses log-linear regression analysis to estimate taxi times with variables that can be computed before operation to account for route- and interaction-specific factors influencing taxi time. The resulting model has a strong predictive validity, but requires a sample size covering an extensive time of airport operations. The model results show that route-specific factors are useful to estimate taxi times, and the combination of stand and rapid exit variables (for landings) and runway (for take offs) accounts for a great part of the variability of taxi times.

Study of predictors of taxi times at BCN Barcelona-El Prat Airport

Citation

Valenzuela-Arroyo, M., Lordan, O., Sallan, J. M. (2014). Study of predictors of taxi times at BCN Barcelona-El Prat Airport. 2014 ATRS World Conference. KEDGE Business School, Bordeaux, France.

Abstract

PURPOSE
This study develops a model to predict taxi in and out times at Barcelona-El Prat Airport. The model includes a set of variables that take into account the aspects that can affect taxi times, such as control areas, operation hours, type of aircraft, and others.

METHODOLOGY
The study has been carried out using airport’s daily operational data for the months of June, July and August 2013. After filtering the information, several linear regression models have been built which provide a predicted value of taxi times as a function of the predictors taken into account.

ANTICIPATED RESULTS
The area predictors combined with the hour of operation explain more than the 75% of variance of taxi times. Also, predictor models show that taxi times are greatly dependent on the type of operation.
The taxi in model has a stronger predictive power than the taxi-out model. Taxi-in accounts for 76% of variance and the mean error of residuals is 0.48 min, while the same values for taxi-out model are 45% and 2.16 min, respectively. The reason to be of these differences is the uncertainty produced by waiting times at runway headers in taxi out models, a variable that cannot be controlled.

KEYWORDS
Taxi Times, Linear Regression Model, Predictor, Barcelona-El Prat Airport