A genetic algorithm for order acceptance and scheduling in additive manufacturing

Abstract

We consider the problem of order acceptance and scheduling faced by an additive manufacturing facility consisting of multiple build chambers and postprocessing operations for support removal and surface finishing. We model each build chamber as a batch processing machine with processing times determined by the nesting and orientation of parts within the chamber. Due to the difficulty of developing an explicit functional relation between part batching, batch processing time, and postprocessing requirements we develop random-keys based genetic algorithms to select orders for complete or partial acceptance and produce a high-quality schedule satisfying all technological constraints, including part orientation and rotation within the build chamber. Extensive computational experiments show that the proposed approaches yield significant improvements in profit over the situation where all orders must be accepted, and produce solutions that compare favourably to statistically estimated bounds.