Solving the integrated airline recovery problem using column-and-row generation

Abstract

Airline recovery presents very large and difficult problems requiring high quality solutions within very short time limits. To improve computational performance, various solution approaches have been employed, including decomposition methods and approximation techniques. There has been increasing interest in the development of efficient and accurate solution techniques to solve an integrated airline recovery problem. In this paper, an integrated airline recovery problem is developed, integrating the schedule, crew and aircraft recovery stages, and is solved using column-and-row generation. A general framework for column-and-row generation is presented as an extension of current generic methods. This extension considers multiple secondary variables and linking constraints and is proposed as an alternative solution approach to Benders’ decomposition. The application of column-and-row generation to the integrated recovery problem demonstrates the improvement in the solution runtimes and quality compared to a standard column generation approach. Columnand-row generation improves solution runtimes by reducing the problem size and thereby achieving faster execution of each LP solve. As a result of this evaluation, a number of general enhancement techniques are identified to further reduce the runtimes of column-and-row generation. This paper also details the integration of the row generation procedure with branch-and-price, which is used to identify integral optimal solutions.