Reallocating arrival slots during a ground delay program

Abstract

This paper presents a new model and solution methodology for the arrival slot reallocation problem faced by airlines when responding to a ground delay program (GDP). The objective is to reassign the flights in the GDP to time slots made available by the Federal Aviation Administration (FAA) such that flight delay and passenger missed connection costs are minimized. The problem is formulated as a dynamic program and solved with the help of branch and bound. Using data provided by American Airlines, initial tests showed that while the results were good for relatively small instances, as more flights were included, computation times grew exponentially. Given that the problem needs to be solved quickly in practice, the methodology was incorporated in a rolling horizon framework where larger problems are split into smaller subproblems and solved sequentially. This led to some degradation in solution quality but there was still considerable cost savings compared to the initial slot assignments proposed by the FAA. Computational experiments with both real and randomly generated data confirmed that problems of practical size could be solved within 5 min.