Overhaul-Tasks Scheduling Model: A Structured Approach for Solving Combinatorial Problems on Aircraft Maintenance Events

Abstract

When an aircraft maintenance event occurs, the overhaul tasks management process requires the execution of all the tasks to perform and has to guarantee the on-time aircraft delivery and the respect of the daily flight schedule. The scheduling process has to take into account several constraints (as deadlines — respect of ground time and intermediate deadlines —, resources availability — material, equipment, infrastructures and human resources —, precedence relations — among activities or groups of activities) and has to realise a set of objectives (grounding reduction and efficient resources allocation). For this purpose an optimisation model has been developed: it supports overhaul departments and their operative management in scheduling set of activities during maintenance events. The model is able to get the optimal tasks scheduling for each aircraft overhaul, building efficient sequences and assigning the best activities starting time. It takes into account the process constraints and a multi-objective function has been applied, in order to assure the minimum makespan (grounding) — for a cost decreasing policy — and to get the earliest “activities starting” — for a manpower release as soon as possible. The problem has been approached as a typical Resources-Constrained Project Scheduling Problem — the most general class of scheduling problems — and it belongs to the class of NP-complete combinatorial problems. The model has been mathematically formalised as an integer linear problem and a time-windows approach has been followed, using variables referred to activity starting time instead of to precedence relations, because of the restricted Pert. Interesting the resulting structure of the problem, obtained by applying some resolution method techniques for reducing computational complexity: phases-based scheduling, destructive improvement, stepping improvement, lower bound evaluation techniques. The Multi-Mode Case has been adopted as feature of the model, introducing flexibility in the way to perform activities. The model has been developed using CPLEX and computational results are very appreciable, considering the plethora of variables to manage: about two thousand activities are performed on heaviest events in an elapsed time of forty or fifty days, for a final total amount of one million seven hundred thousand variables.