Dynamic Disruption Management in Airline Networks Under Airport Operating Uncertainty

Air traffic disruptions result in flight delays, cancellations, passenger misconnections, and ultimately high costs to aviation stakeholders. This paper proposes a jointly reactive and proactive approach to airline disruption management, which optimizes recovery decisions in response to realized disruptions and in anticipation of future disruptions. The approach forecasts future disruptions partially and probabilistically by estimating systemic delays at hub airports (and the uncertainty thereof) and ignoring other contingent disruptions. It formulates a dynamic stochastic integer programming framework to minimize network-wide expected disruption recovery costs. Specifically, our Stochastic Reactive and Proactive Disruption Management (SRPDM) model combines a stochastic queuing model of airport congestion, a flight planning tool from Boeing/Jeppesen and an integer programming model of airline disruption recovery. We develop a solution procedure based on look-ahead approximation and sample average approximation, which enables the model's implementation in short computational times. Experimental results show that leveraging even partial and probabilistic estimates of future disruptions can reduce expected recovery costs by 1%-2%, as compared with a myopic baseline approach based on realized disruptions alone. These benefits are mainly driven by the deliberate introduction of departure holds to reduce expected fuel costs, flight cancellations, and aircraft swaps.