Air Traffic Flow Management

TL;DR

This paper investigates the properties of the Traffic Flow Management Problem's LP relaxation, finding that it often yields integral solutions, which simplifies solving the problem efficiently in air traffic management.

Contribution

The study provides computational evidence that the LP relaxation of TFMP frequently produces integral solutions, supporting its practical use in air traffic flow management.

Findings

LP relaxation often yields integral solutions

Integral solutions occur even with conflicting schedules

LP relaxation can be solved efficiently on low-resource machines

Abstract

Air Traffic Flow Management is the regulation of air traffic in order to avoid exceeding airport or flight sector capacity in handling traffic, and to ensure that available capacity is used efficiently. We have tried to explore the logic behind the claims by Bertsimas et.al about integral solutions to the LP relaxation of the Traffic Flow Management Problem(TFMP). Polyhedral theory only indicates that the stronger TFMP formulation of Bertsimas et.al might lead to integral solutions in some cases. Our computations indicate that the encouraging results reported by Bertsimas et.al are not merely fortuitous or due to their specific data set. Indeed, we found that the TFMP had integral solutions even in case of artificial data sets generated to include severe conflicts in the flight schedules. In our limited tests with 4-5 scenarios, we obtained non-integral solutions only once. This is of significant practical importance because, the LP relaxation can be solved even on small machines with low memory and processor speed.