Latest revision as of 18:42, 28 January 2021

Abstract

This thesis deals with traffic forecasts of Airspace Users for Air Navigation Service Providers. Currently there is a high amount of uncertainty in the traffic forecast. Air Traffic Flow Managers anticipate for unforeseen traffic by increasing the forecasted maximum capacity threshold by more than 10%. The European ATM research program SESAR aims ultimately to reduce the difference between the forecast and real maximum capacity threshold by less than 3%. Prediction uncertainty results in sector over-load and sub-optimal traffic flow in the air transportation system. The objective of this thesis is to investigate sector demand predictability by quantification of the difference between real and forecast traffic, and evaluate the predictability improvement by improvement of departure time predictions. Statistical analysis is performed by plotting time & count uncertainty against look-ahead time to sector entry. On a busy day, for a look-ahead time of 2 hours and longer, flights have a higher probability to be delayed than to be earlier as planned. A comparison is made between the forecast and real number of flights entering a sector for a given time of day window. Looking at the forecast time period, it can be seen that some forecasted flights did not enter in the window anymore (out), and some additional flights entered in the window that were initially not forecasted (in). `In' and `out' flights can be explained due to flights being earlier or delayed, or flights deviating from the planned route. Looking at 10 minutes before entry, about 30% to 40% are in/out flights, which is a large amount. In general, for a look-ahead of 0 to 3 hours, there are more `out' than `in' flights resulting in an over-prediction. Over-prediction means that there are more flights anticipated than really entered for a given time window. In order to reduce over-prediction, it is suggested, taking safety into account, to reduce the number of `out' flights that deviate from the planned route. For a high capacity Maastricht Upper Area Control sector on a normal day, a 5% decrease of these `out' flights, reduces the over-prediction by 10%. Furthermore, flight phases that are major causes of uncertainty are descent, taxi and the slot allocation process. A departure time prediction improvement of 50% results in 20% arrival time error reduction, and 30% mean sector entry time error reduction, for a 6 hour look-ahead time. The used sensitivity method does not yield realistic sector occupancy count because the effect of changed ATC procedures due to improved predictability is not incorporated.

Original document

The different versions of the original document can be found in:

• http://pdfs.semanticscholar.org/6965/90fcbb60ce00ada0735aa232a4e409eca88e.pdf

• http://arc.aiaa.org/doi/pdf/10.2514/6.2015-3331,

http://dx.doi.org/10.2514/6.2015-3331

• https://repository.tudelft.nl/islandora/object/uuid%3A0db87565-1d65-4dfd-8a15-ac20482372f0,

https://repository.tudelft.nl/islandora/object/uuid%3A0db87565-1d65-4dfd-8a15-ac20482372f0/datastream/OBJ/download,

https://arc.aiaa.org/doi/10.2514/6.2017-3602,

https://academic.microsoft.com/#/detail/199385616