Aggregate Airspace Congestion Modeling David Lovell Department of Civil and Environmental Engineering & Institute for Systems Research University of Maryland College Park, MD joint work with Michael Ball, Yufeng Tu, and Bala Chandran 1 Evolving from Monitor Alert • Improvements • Queueing effects in the NAS caused by capacity restrictions • Stochastic departure times of aircraft •Applications • Collaborative decision-making and schedule refinement • Congestion alerts over the course of the day • Outputs • Probability of congestion in sector x at time t • Distribution of delays to flights • Airspace capacity estimation Issues related to model scale • Microscopic simulation models are OK when time is plentiful • Our goal: rougher estimates in a single pass • Limited network topology • Fluid modeling of “probability flows” – continuum diffusion equations operating on tandem queues • Convergence, or “laws of large numbers” Propagation of “probability packets” Probability 1 min 0.1 0.2 .14 0.3 .14 .23 0.1 .06 0.5 0.3 .07 .36 .36 .14 .06 .03 .14 .07 Time Aircraft “A” Aircraft “B” Example of interaction effects 1 a ir c ra f t 2 1 a ir c ra f t 0 .8 0 .9 0 .7 0 .8 probability of occupancy probability of occupancy 0 .6 0 .7 0 .6 0 .5 0 .4 0 .3 0 .5 0 .4 0 .3 0 .2 0 .2 0 .1 0 .1 0 0 1 0 0 ti m e • P{cancel} = 0 • tA ~ uniform(0,100) 2 00 0 0 1 00 2 00 3 00 ti m e For aircraft A and B: • P{cancel} = 0.5 • tA, tB ~ indep. uniform (0,100) • server capacity = 1 Preliminary results Bala attempted to categorize interactions according to some degree of severity: “strong” and “weak” and developed heuristic methods to handle these situations: Sector Count Sector Count 25 25 20 20 15 15 10 10 5 5 0 0 No of aircraft No of aircraft Model Model Simulation Simulation 1 1 39 39 77 77 115 115 153 153 191 191 229 229 267 267 305 305 343 343 381 381 419 419 457 457 495 495 • Time period Time period Predicted Actual Travel Time Travel Time 119 119.555 116.199 340 111.755 114.744 954 356.743 364.500 Flight No