Article ID: | iaor19982198 |
Country: | United States |
Volume: | 43 |
Issue: | 6 |
Start Page Number: | 995 |
End Page Number: | 1011 |
Publication Date: | Nov 1995 |
Journal: | Operations Research |
Authors: | Bertsimas Dimitris J., Odoni Amedeo R., Peterson Michael D. |
Keywords: | probability, transportation: air, networks: flow |
A previous paper (1992) by the same authors studied the phenomenon of transient congestion in landings at an airport and developed a recursive approach for computing moments of queue lengths and waiting times. This paper extends our approach to a network, developing two approximations based on the prior method. Both approaches work by using delay information estimated at one location to update arrival schedules at other points in the network. We present computational results for a simple 2-node network, comparing the performance of the approximations with an alternative simulation approach. The methods give similar results in light to moderate traffic but show a growing disparity under heavier traffic, where the algorithms underestimate the true magnitude of delay propagation relative to simulation. Finally, to illustrate the usefulness of the modeling, we show how the results may be used to explore the issue of interaction between airports. Although this particular application motivated development of the model, the method is, in principle, applicable to other multiclass queueing networks where service capacity at a station may be modeled as a Markov or semi-Markov process. The model represents a new approach for analyzing transient congestion phenomena in such networks.