A dynamic stochastic model for converging inbound air traffic

Jun Chen, Purdue University

Abstract

Weather accounts for the majority of congestion in the National Airspace System which highlights the importance of addressing weather uncertainty to mitigate delays, and this paper presents an effort in this direction. Firstly, a new dynamic stochastic 0-1 Integer Programming (IP) model is proposed, which models the Single Airport Ground Holding Problem (SAGHP) with respect to uncertainty in the separation between flights instead of Airport Acceptance Rate (AAR) or landing capacity. Uncertainty in separation according to different weather conditions is represented through the scenario tree by using stochastic linear programming. Considering time separation constraints instead of AAR constraints, our model is able to schedule a more accurate plan for the individual flight in minutes. Secondly, a converging inbound air traffic model is formulated based on our dynamic stochastic IP model. We address a problem involving two paths inbound air traffic merging into a single airport in which uncertainty in separation from Minute-In-Trail restrictions is considered. Although "First Come, First Serve" policy is still obeyed by flights on the same path, the experimentation has shown that, allowing flights on different paths to switch arrival orders can help reduce the total delays. Finally, in order to tackle the running time problem faced by the disaggregate integer model we built, we introduce dual decomposition method into the model to improve the computing efficiency. The original problem is decomposed scenario by scenario into several sub-problems based on the dual decomposition method; then a parallel computing algorithm is developed to handle these sub-problems. Such combination increases the model's computational efficiency.

Degree

M.S.A.A.

Advisors

Sun, Purdue University.

Subject Area

Aerospace engineering|Industrial engineering|Transportation planning|Operations research

COinS