The development of decision support models for European air traffic flow management

Congestion severely affects air traffic in the US and Europe. To protect air traffic controllers from overloads a planning activity, Air Traffic Flow Management (ATFM), emerged during the 1970s. ATFM control actions range from departure delays to the re-routing of flights. This research explores how models can be used to support decision-making in European ATFM. To date, most research into this subject has been directed at ATFM in the US, which differs from European ATFM both in terms of decision-making and time scales. Fieldwork was carried out at the EUROCONTROL Central Flow Management Unit, the organisation that manages traffic flows in most of the European airspace. The fieldwork was an OR intervention aimed at identifying suitable decision support models for re-routing flights. The research described here contributes by: 1) describing the European ATFM field and identifying decision support needs; 2) structuring the problems involved in re-routing flights in Europe; 3) providing a framework for the development of re-routing decision support systems (DSS) and 4) assessing the usefulness of optimisation approaches to re-routing flights. A demonstrator is developed to illustrate different re-routing decision support possibilities to the users. This leads to conclusions on the feasibility of various decision support functions including an identification of models and algorithms which can be used for each of the functions. Conclusions on levels of automation and complexity for re-routing DSS are also taken. Three integer models for re-routing flows are presented. They differ in the way congestion is represented. The models are tested on data of traffic crossing the whole French upper airspace. The test reveals that the models can be of use in re-routing flows and can provide significant savings in delays. It also shows that an 'intelligent' component to define the scope of the optimisation problem and a component to process all the data for the models, are needed in a re-routing DSS. The models are compared in terms of impact on congestion, size and execution time and conclusions on their feasibility taken. Extensions to the models are suggested.