Predicting Airspace Congestion using Approximate Queueing Models

Abstract

We present an approximate method for the analysis of queueingdelays in highly dynamic networks with schedule-based stochasticarrivals and time-varying service times. We also develop anintuitively appealing network flow model representation of theproblem and compare the performance of both models to a much moredetailed simulation on several sample networks. The two approachesare applied to the problem of estimating queueing delays in theairspace, which is modeled as a node-capacitated network withtime-varying capacity constraints and aircraft departure-timeuncertainty. We demonstrate the use of these models in airspacecongestion prediction and airline schedule evaluation.