Scheduling in aerospace composite manufacturing systems: a two-stage hybrid flow shop problem

This research investigates a real-world complex two-stage hybrid flow shop scheduling problem which is faced during the manufacturing of composite aerospace components. There are a number of new constraints to be taken into account in this special hybrid flow shop, in particular limited physical capacity of the intermediate buffer, limited waiting time between processing stages, and limited tools/molds used in both stages in each production cycle. We propose a discrete-time mixed integer linear programming model with an underlying branch and bound algorithm, to solve small- and medium-size problems (up to 100 jobs). To solve the large instances of the problem (up to 300 jobs), a genetic algorithm with a novel crossover operator is developed. A new heuristic method is introduced to generate the initial population of the genetic algorithm. The results show the high level of computational efficiency and accuracy of the proposed genetic algorithm when compared to the optimal solutions obtained from the mathematical model. The results also show that the proposed genetic algorithm outperforms the conventional dispatching rules (i.e., shortest processing time, earliest dues date and longest processing time) when applied to large-size problems. A real case study undertaken at one of the leading aerospace companies in Canada is used to formulate the model, collect data for the parameters of the model, and analyze the results.