Automatic Generation and Integration of Maintenance Feasibility Feedback Into the Fleet Assignment Model

Kurzfassung

An airline’s planning process usually spans a long period of time and consists of a sequence of subproblems that are solved one after another. A common problem of this sequential decision-making process is that decisions made in one step render finding a feasible solution for a later step impossible. For example, airlines first assign the fleet type to the flights of a week. However, this assignment does not necessarily ensure regular maintenance being possible for each aircraft, which is noticed later during rotation planning. In practice, this problem is solved through human feedback loops between the airline’s planning departments. In this paper, a digital representation of this process is presented as part of the Digital Airline Twin (DAT), which is currently under development at the Institute of Air Transport of the German Aerospace Center (DLR). Two different sources of maintenance infeasibility are addressed through iterative modification of the fleet assignment model until the subsequent aircraft maintenance routing model can find a feasible solution. Firstly, based on connected component analysis constraints are added that enable aircraft to return to maintenance hubs. Secondly, overuse of the most efficient fleet types is penalized to leave capacity for regular maintenance. The novelty of this approach lies mainly in the modular, feedback-based link between both planning processes, in contrast to a fully integrated optimization model. Experiments were executed demonstrating that the algorithm can successfully resolve the encountered infeasibilities. Hence, it is a suitable tool for automating as well as better understanding fleet assignment replanning based on maintenance feasibility feedback.