Taxi time analysis and prediction with ADS-B data. A case study in Barcelona-El Prat airport.
Document typeBachelor thesis
Rights accessOpen Access
This document contains a study about taxitime analysis in Barcelona-El Prat airport using ADS-B data. Section 1 shows how to decode ADS-B data and what useful information can be recovered to perform the taxitime analysis. Section 2 shows how to model the airport, including runways, taxiways and stands based on AIP data, satellite images and official maps. Section 3 shows how the positions obtained from ADS-B and the modeled airport can be related to be able to unequivocally define the trajectory that an airplane has followed through this modeled airport. Section 4 shows from all the information compiled in the previous sections, how can be determined the factors that affect taxitime and how to create a model that allows estimating them. In previous studies taxitime calculation had been restricted to very specific situations or locations, but in this document, will be tried to relative all parameters (working with speeds, relative queues and differentiation for operations) to be able to extend this calculation to all airport operations. The results show that this goal has been achieved with an accuracy of 2 minutes (A-CDM requirement) of 73$\%$ in departures and 97$\%$ in arrivals. The proposed model is not only characterized by high accuracy in static conditions but also shows a good adaptation to changing conditions. Although it's true that the model doesn't work well when the training and evaluating data sets have different conditions, the model has proven to be valid under new conditions with a very small set of training data in the new conditions. Traditional models are based on calculation of point-to-point histories, which need a very large period of data to extract conclusions. With this methods it's difficult to calculate taxitime just after a condition change. With the proposed model, thanks to calculations in velocities and relativities, model is able to extract information from all the available data and create predictions with good accuracy even if the conditions have recently changed. This document also presents a real case of calculation in extraordinary conditions where adaptation capacity of the model can be seen in Barcelona-El Prat airport during a one-month runway closure due to maintenance.