Parameters and sensitivity in workload control

Both practitioners and scientists have recognized the importance of workload control (WLC) for job shop practices. WLC principles have been integrated into a comprehensive concept, which receives a lot of attention in scientific literature in recent years. Though meant as a robust concept for dynamic environments, the WLC concept needs a large number of parameters to be specified. The key decision of order release requires the specification of workload norms, planned throughput times, a release frequency, and a time limit.

Past studies have shown the delicacy of parameter setting. Still, research providing sound theoretical foundations is limited. This paper aims at improving the basis for setting parameters, showing the impact of parameters on performance and analysing sensitivity. It provides an in-depth analysis of the mechanisms within WLC that affect the relevant performance dimensions. It exposes the critical elements, which result in a fragile balance between control of average throughput times and variance of due date deviations. The influence of each parameter on this balance is assessed in a simulation study.

The results provide a number of new insights. Due date performance appears to depend on only one of the intended pillars within WLC. The intended due date orientation of the order release sequence hardly contributes to performance, but is partly compensated by the control of shop floor throughput times. Paradoxically, the influence of the planned throughput time values and workload norms is shown to be limited, while the choice of the workload calculation method, release frequency and time limit appears critical. Either throughput time performance or due date performance is highly sensitive to each parameter. The results have important consequences for implementations of WLC concepts in practice and give rise to a change of focus in parameter setting. (c) 2005 Elsevier B.V. All rights reserved.