National airspace model: Stochastic optimization of flight frequencies after airport losses

Brian Levine, Cornell University

Photo of Brian Levine

Substantial effort has been expended to develop optimization models for airplane flight frequency planning. However, these models were designed to pursue only the economic goals of individual airlines. In the world we live in today, we must be prepared to overcome the loss of an airport(s) or part of our air traffic control system through terrorism or natural disasters. Therefore it is important to develop models focused on recovering from a substantial loss of air system infrastructure. This will represent a significant departure from current models, including (1) the inclusion of equity among the individual airlines and the government's role in maintaining system efficiency, (2) the incorporation of the economic capabilities of carriers, (3) uncertainty in the recovery of passenger and cargo demands over time (including how this would differ for various types of disruptions), and (4) the incorporation of uncertainty in the flow of aircraft throughout the airspace.

Currently we are looking at how to model this problem as a multi-stage stochastic mixed-integer optimization, with the goal being to optimize flight frequencies by carrier and type of equipment between pairs of airports, so as to accommodate as much demand as is feasible and appropriate. In order to do this, we must determine which of the above factors are the most important and how to collect data for the input.

Abstract Author(s): Brian Levine