Air traffic management : on spotting the origin of deviations causing over-deliveries in the European sky

Details

Supervisors

Van Vyve, Mathieu

Faculty

Louvain School of Management

Degree label

Master [120] en Ingénieur de gestion

Abstract

This paper will look at some crucial concerns about safety requirements, and more precisely about the case of over-deliveries. The over-delivery is a concept stating that an airspace is undergoing more traffic than its established capacity. The reason why this can bring safety issues is obvious: the risk of a mid-air collision if each flight is not correctly monitored is non-negligible. The purpose of this paper is to provide a consistent analysis of the origin where the flights start deviating from their initial flight plan, with this deviation causing an intrusion in a regulated sector. The development of an algorithm as well as some possibilities that would be noticed during the analysis to improve the way Over-deliveries are managed will be developed too. The first part of this master thesis will try to explain the context in which this analysis is needed, the current methods settled for preventing Over-deliveries to occur and the basic notions required to understand the whole concept. Air traffic management is a very specific topic, and the vocabulary linked to this sector is precise. We will then have a look at the processes necessary for managing a flight, as well as the role of the Operations Centre in the decision of limiting the traffic in an airspace during a given period. We will also explain the original project that was planned, and the reasons why it could not reach the objectives that had been defined. The second part will explain the basic algorithm developed for finding the origin of the deviations leading to the intrusion of regulated airspaces. It will bring the materials that will be used all along the evolution of this algorithm, and actually represents the “core” of the algorithm. In the third part, we will develop the different flight patterns that were found and for which the algorithm does not find the correct origin. We will see the reasons why it fails to do so, and the improvements added to the algorithm to try to solve this. It is important to note that some of the improvements are not entirely accurate, as some did not succeed in fixing the problem with a 100% rate. The next part will discuss these improvements outputs and provide quantitative results on the chosen sample. We will describe step by step the refinement of the algorithm for each of the improvements previously explained, and see the contribution of these to the final accuracy of the algorithm. The results will then be analyzed, and we will see how they can be treated to find usable clusters and potential links between some factors of the flights. The way to use the information provided by the analysis will be developed in the last part, as well as some suggestions that can be made to reduce the over-deliveries figures and still refining the algorithm.