A Genetic Algorithm Approach for Ground Delay Program Management: The Airlines’ Side of the Problem

In this paper, we present a model for slot allocation for flight landings during Ground Delay Programs (GDPs). The model efficiently assigns inbound flights affected by GDP to available landing slots such that the overall downline impact resulting from delaying these inbound flights is minimized. In this model, a Genetic Algorithm (GA) is integrated with a flight simulation model. The GA searches for the optimal landing-slot allocation pattern. The flight simulation model guides the search by evaluating the overall airline performance for each generated slot-allocation pattern. It captures the schedule interaction of the different resources (aircraft, pilot, and flight attendants), analyzes the downline impact of any selected slot-allocation pattern, and describes it quantitatively. This impact is represented in terms of delay of flights, misconnects of aircraft, crew, and passengers, as well as crew work rules violations (illegalities). Several experiments with hypothetical GDPs and actual airline schedules are presented. Results show fast convergence of the algorithm with an average improvement in the system performance of about 23% compared with the do-nothing scenario.