Workload control with shifting bottlenecks: norms optimisation through design of experiments

Abstract

Workload control is a production planning and control system designed to overcome the trade-off between high throughput and short and stable lead time. Specifically, work-in-process is continuously monitored, and new jobs are not admitted in the shop floor until work-in-process drops below predefined threshold values or norms. To exploit performance, norms should be fine-tuned to minimise queues, without generating starvation at the bottleneck machines. The optimisation process is straightforward for a perfectly balanced system, but much harder in case of shifting bottlenecks. The paper focuses on this issue and presents an innovative procedure, based on the response surfaces method, which allows one to optimise the norms in a precise way, keeping unaltered the maximal or desired throughput of the manufacturing system. A comprehensive simulation analysis demonstrated the quality of the proposed approach and showed the importance of using different norms to boost the overall performance of the manufacturing system.