Feedback Control of the National Airspace System

Abstract

DOI: 10.2514/1.51203 This paper proposes a general modeling framework adapted to the feedback control of traffic flows in Eulerian modelsof the NationalAirspace System. It is shown thatthe problems of scheduling androuting aircraft flowsin the NationalAirspaceSystemcanbeposedasthecontrolofanetworkofqueueswithload-dependentservicerates.Focus can then shift to developing techniques to ensure that the aircraft queues in each airspace sector, which are an indicator of the air traffic controller workloads, are kept small. This paper uses the proposed framework to develop control laws that help prepare the National Airspace System for fast recovery from a weather event, given a probabilistic forecast of capacities. In particular, the model includes the management of airport arrivals and departures subject to runway capacity constraints, which are highly sensitive to weather disruptions. I. Introduction T HE frequent occurrence of air traffic delays in the National Airspace System (NAS), along with the projected increase in demand, motivate the scheduling of flight operations to better use available system resources. The process of planning operations to balancetheavailablecapacityandthedemandforresourcesisknown asTrafficFlowManagement(TFM).Thistaskiscurrentlyconducted manually by air traffic controllers (ATC), and contributes significantlytotheirworkload.Tomeettheincreasingtrafficdemand,there is a desire to introduce a greater level of automation and decision support for air traffic management. Research on the TFM problem has traditionally focused on developing open-loop policies for scheduling aircraft operations. Because of the typical travel times of cross-country flights, openloop traffic flow management policies need to be determined 5– 6 hours ahead of the time of operations. Such policies prescribe the positionofeachaircraftinthesystemateachinstant,andareobtained by solving large-scale integer programs [1,2]. This approach is difficult to scale to the scheduling of approximately 40,000 flights a day, and typically does not address the many sources of uncertainty present in the system. Weather, in particular, is a major source of disruption that requires constant adjustment of the schedules. For instance, 66% of all NAS delays in 2009 were attributed toweather. ‡ Moreover, open-loop traffic flow management algorithms require precise weather forecasts several hours ahead of time, which are arguably beyond the limits of even state-of-the-art weather forecasting tools [3]. The disturbance attenuation properties of feedback control make closed-loop control policies for the NAS very attractive. Attempts have been made to introduce some feedback in the decision algorithmswhilestilltryingtooptimizeeachaircrafttrajectory[4,5]. More recently, researchers have started developing new models that are more tractable for the purpose of control, which only record aircraft counts in specific control volumes of airspace rather than follow individual aircraft. These aggregate flow models, called