Powder bed fusion factory productivity increases using discrete event simulation and genetic algorithm

Abstract

Powder bed fusion is importance is growing with uses across industries in both polymer and metallic components, particularly in mass individualization. However, due to the relatively slow mass deposition speed compared to conventional methods, scheduling and production planning play a crucial role in scaling up additive manufacturing productivity to higher volumes. This paper introduces a framework combining discrete event simulation and a genetic algorithm showing makespan improvement opportunities for multiple powder bed fusion factories varying workers, jobs and available equipment. The results show that bottlenecks move among workstations based on worker and capital equipment availability, which depend on the size of the facility indicating a resource-driven constraint for makespan. A makespan reduction of 78% is achieved in the simulation. This shows the trade-off of worker and capital equipment to achieve makespan improvements. The addition of personnel or equipment increases production with further gains achieved by scheduling optimization. Two levels of job demands are analyzed showing productivity gains of 45% makespan improvement when adding the first worker and additional savings with scheduling optimization using a genetic algorithm up to 11%. Most research on additive manufacturing production has focused on the quality of produced parts and printing technology rather than factory level management. This is the first application of this methodology to varying sizes of these potential factories. The method developed here will help decision-makers to determine the appropriate number of resources to meet their customer demand on time, additionally, finding the optimal route for jobs before starting the production process.