Solving the Integrated Schedule Generation and Fleet Assignment Problem: an ACOBased Metaheuristic Approach

• Autor(a) principal: Caetano,Daniel J.
• Data de Publicação: 2015
• Outros Autores: Gualda,Nicolau D. F.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Journal of Transport Literature
• Texto Completo: http://old.scielo.br/scielo.php?script=sci_arttext&pid=S2238-10312015000300030
• topic: air transportation; schedule generation; fleet assignment; metaheuristic; ant system

Abstract Traditionally, the initial steps on airline planning – Schedule Generation and Fleet Assignment problems – are solved separately. This traditional approach usually leads to suboptimal solutions, since flight profitability – the decision criteria to schedule a flight – depends on what aircraft type will be used on that flight. On the other hand, the type of aircraft assigned to a flight will be different accordingly to the available scheduled flights. Because of this interdependence, airlines avoid complete redesign of their flight network, adopting a conservative approach and slightly improving existing suboptimal schedules over time. The integrated solution for both problems, albeit desirable, leads to large-scale models of the NP-Hard class. Some of the original linear constraints may become non-linear in the integrated problem, bringing further complexity to the solution process. This article presents a linear programming formulation of this integrated problem along with a heuristic approach, called MAGS, based on the ACO metaheuristic. Both the exact solution and the one provided by MAGS are obtained and compared for the case of a Brazilian airline. The results show the applicability of MAGS to real world cases, presenting solutions with objective function values distant no more than 6% of the optimum and much lower processing times than LP model on more complex configurations.