Surface Congestion Management Hamsa Balakrishnan Massachusetts Institute of Technology TAM Symposium 2013
Motivation 2
Surface Congestion Management Objective: Improve efficiency of airport surface operations Decrease taxi times, decrease fuel burn, improve/maintain airport throughput Multiple interconnected, constrained resources: Gates, aprons, taxiways, runways, departure routes, etc. Departure Metering Departure Route Assurance Taxi Routing Surface Congestion Management Tool Suite Sequencing & Scheduling Airport Configuration Runway Assignment 3 [Idris 2000]
Role of Departure Metering Departure metering just one element of required surface management toolset Departure metering regulates pushbacks during congested periods Decreased engines-on time, fuel burn & emissions Principle can work at any congested airport, but implementation details will vary e.g., ATC facility vs. airline ramp tower Demand on Surface Taxi Routing Max efficiency limit 1 Departure Metering Airport Configuration Surface Congestion Management Tool Suite Excess congestion 2 3 4 Time Interval Departure Route Assurance Sequencing & Scheduling Runway Assignment Excess flights held until later time intervals when they can be more efficiently accommodated [A. Nakahara, 2012] 4
Challenges and Opportunities Uncertainty Weather, runway configuration, demand (pushback/arrival times), operational variability, human factors, Level of certainty varies depending on information source, type, and time frame What is capacity? Level of effort vs. expected benefit Aggregate queue-based control vs. RTA-based trajectory control Information requirements Ease of adaptability to different airports and operating environments Availability of (diverse) operational data 5
Operating Environments: Runway Configuration BOS Runway Configuration Usage; 6/1/11-8/31/11 33L 27 22L, 22R 15R 27 33L LGA Runway Configuration Usage; 6/1/11-8/31/11 4 4 31 31 4 13 4 31 13 4, 13 12% 31 4 22L, 27 22R, 22L 47% 47% 4R, 4L 9, 4R 37% 26% 17% 22 13 PHL Runway Configuration Usage; 6/1/11-8/31/11 27L 27L 27R 27R 9R 9R 9L 9L 22 22 22 31 9R 9L 17% 77% 27R 27L 6
Operating Environments: Airline Mix BOS Aircraft Operations by Airline American Continental Delta 12% 8% 23% Other LGA Aircraft Operations by Airline Other American 8% 19% 8% Shuttle America Spirit Continental Chautauqua AirTran JetBlue 16% 13% Cape Air Delta 18% Air Canada 23% Southwest United 12% USAirways AirTran Air Canada JetBlue Southwest United PHL Aircraft Operations by Airline Southwest Chautauqua Delta American Other 10% UPS USAirways United 8% 68% USAirways 7 All data from 6/1/11-31/8/11
40 35 Operating Environments: Demand Variations BOS Surface Metrics (22L,27 22R,22L); 6/1/11-8/31/11 Number of Active Departures Queue Size Taxi Time 40 35 LGA Surface Metrics (22 13); 6/1/11-8/31/11 Number of Active Departures Queue Size Taxi Time Number of Aircraft/Time (mins) 30 25 20 15 10 Number of Aircraft/Time (mins) 30 25 20 15 10 5 5 0 6 8 10 12 14 16 18 20 22 24 Local Time (hrs) 40 35 PHL Surface Metrics (27R 27L); 6/1/11-8/31/11 Number of Active Departures Queue Size Taxi Time 0 6 8 10 12 14 16 18 20 22 24 Local Time (hrs) Number of Aircraft/Time (mins) 30 25 20 15 10 5 8 0 6 8 10 12 14 16 18 20 22 24 Local Time (hrs)
Airport Operational Efficiency Metrics Daily operational efficiency reports to BOS Tower (since Aug 2011) Compare inter-departure separations with target values Demand level (combination of departure queue length and number of taxiing departures) for each 15-min interval Khadilkar and Balakrishnan, Air Traffic Control Quarterly, 2013 9
Some other projects: Efficient & Equitable Arrival/Departure Scheduling Given a set of flights with estimated arrival times at the airport, the aircraft need to be sequenced into the landing (takeoff) order, and the landing (takeoff) times need to be determined Need minimum (wt. class dependent) wake vortex separation (Safety) Currently FCFS; resequencing could increase throughput (Efficiency) Fair resequencing: Constrained Position Shifting (CPS) [Dear 1976] We show that scheduling under constrained position shifting can be solved in (pseudo-)polynomial time as shortest-path problems on variations of this network Balakrishnan and Chandran, AIAA 2006, ATM R&D Seminar 2007, Operations Research 2010 Chandran and Balakrishnan, ACC 2007 Lee and Balakrishnan, ACC 2008, Proceedings of the IEEE 2008 10
Some Other Projects: Prediction of air traffic delays Predict departure delay on a link by considering: Current delay state of the network Interdependencies between network elements Time-of-day and day-of-the-week Delays at origin, destination, and on link Delay state of the National Airspace System (NAS) Type of delay day in the NAS Delay states obtained by k-means clustering of delays NY, Chicago and Atlanta emerge as main delay centers Avg. link delay in min Centroids of NAS delay states. Color represents avg. link departure delay over 2- hr time-window 11 Rebollo and Balakrishnan, ICRAT 2012
Other Research Network modeling and congestion control of airport surface operations [Khadilkar and Balakrishnan, AIAA Journal of Guidance, Control and Dynamics 2013] Mechanisms for resource allocation and reallocation [Balakrishnan, Conference on Decision and Control 2007; Ramanujam PhD thesis 2011] Discrete-choice models of configuration selection processes [Ramanujam and Balakrishnan, American Control Conference 2010] Factors influencing pilot penetration of weather [Lin and Balakrishnan, Transportation Research Record 2014] Distributed feedback control of the National Airspace System [Le Ny and Balakrishnan, AIAA Journal of Guidance, Control and Dynamics 2011] Models of engine performance from flight recorder data [Khadilkar and Balakrishnan, Transportation Research Part D 2012] Integration of control and communication algorithms for NextGen [Park et al., IEEE Transactions on Intelligent Transportation Systems 2013] 12