Uncertainty Feature Optimization for the Airline Scheduling Problem

In this paper, we present an application to the Airline Scheduling Problem (ASP) of the Uncertainty Feature Optimization (UFO) framework which combines both a proactive scheduling algorithm and a reactive recovery algorithm used for re-optimization when disruptions occur. We show that re-timing some flights of the original schedule allows for more delay absorption. This means the solution is robust against some delays. Additionally, in case of severe disruption requiring reoptimization, the retiming increases the performance of the recovery algorithm: the number of disrupted passengers, and thus associated compensation costs, are reduced. We provide computational results for the public data of an European airline provided for the ROADEF Challenge 2009. (http://challenge.roadef.org 2009/index.en.htm)