​

Flight scheduling, fleet assignment, and aircraft routing are the three most prominent decisions in airline planning as they contribute towards a majority of the costs and revenues of an airline company. These decisions have to be made 10-12 weeks prior to the flight date as mandated by labor unions in order to accommodate cabin crew scheduling requirements. In this study, we develop a two-stage stochastic programming model for the integrated flight scheduling, fleet assignment, and aircraft routing problem. The model helps in tackling propagated delay, which is a serious matter to consider in airline planning because it results in huge costs and inefficient utilization of aircraft and crew. Deadhead flights are another important matter considered in this work; deadhead flights involve sending empty aircraft from one airport to another in order to serve a highly profitable flight leg. Additionally, alliances in the form of codeshare agreements were considered to test the effect of expanding the airline's outreach network while retaining low costs. Sample average approximation (SAA) algorithm is used to tackle the uncertainty in the demand while column generation is used to solve the resulting highly complex problem. Computational experiments conducted on a real-life airline company's flight network show that modeling the stochastic problem with 100 scenarios is sufficient to capture the effect of demand and fare uncertainty and to provide a solution with an optimality gap less than 1% within a reasonable computational time. Furthermore, the results show that column generation can solve the model in a fraction of the time a commercial solver takes. A sensitivity analysis on different parameters of the model was carried out and points out the applicability of the proposed model and solution in practice.​