Predictor of Airport Runway Capacity (PARC)

The Predictor of Airport Runway Capacity (PARC) is a decision support tool for air traffic managers to estimate the near-term capacity of the individual runways of an airport for traffic planning and control. To estimate runway capacity, PARC analyzes historical data describing an airport’s traffic movements, operating conditions, operating procedures, and geospatial data to determine the time intervals between successive aircraft using an airport’s runway and variables that influence it. Variables may include facility, aircraft, flight plan and weather conditions. Second, PARC uses the data to construct Bayesian Network (BN) statistical models of the joint probability of inter-aircraft time spacing and the variables for each airport’s runway. Third, PARC performs Monte-Carlo simulations of the traffic planned to use each runway, sampling the BN models to estimate the spacing of successive takeoff, landing and taxi crossing aircraft, to obtain a distribution of possible runway capacities. Fourth, PARC selects a a target runway capacity from the distribution for airport traffic management. The advantages of PARC are adapting to the characteristics of the airport and accounting for the anticipated operating conditions to provide accurate estimates of runway capacity. Phase I demonstrates processing of FAA System Wide Information Management (SWIM) traceable data sources for modeling for 1-year of data for Atlanta-Hartsfield International Airport (KATL). Phase I uses the data to constructs BN models of inter-aircraft time spacing and to validate the models. Phase I demonstrates the greater accuracy of the BN models in representing and predicting inter-aircraft time spacing than a simpler single-event probability model. Phase I demonstrates sampling of the BN models for different airport, aircraft and flight plan conditions to obtain inter-aircraft spacing values to be used in Monte-Carlo simulations for capacity prediction.