Drum-buffer-rope and workload control in high-variety flow and job shops with bottlenecks : an assessment by simulation

Thurer, Matthias and Stevenson, Mark and Silva, Cristovao and Qu, Ting (2017) Drum-buffer-rope and workload control in high-variety flow and job shops with bottlenecks : an assessment by simulation. International Journal of Production Economics, 188. pp. 116-127. ISSN 0925-5273

Preview

PDF (Thurer-et-al_IJPE_2017)
Thurer_et_al_IJPE_2017.pdf - Accepted Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (1MB)

Abstract

Two key concepts in the production planning and control literature that incorporate an order release function are the Theory of Constraints, with its drum-buffer-rope release method, and Workload Control, with its load-based release methods. When order release is applied, jobs are not directly released to the shop floor – release is controlled to realize certain performance measures. The performance impacts of drum-buffer-rope and Workload Control order release have been assessed separately, but the two approaches have not been directly compared in one study. This is a major shortcoming that leaves practitioners without guidance on which release method to select. This study assesses the performance of drum-buffer-rope and Workload Control release in a pure job shop and a general flow shop with varying levels of bottleneck severity. Both bottleneck oriented and non-bottleneck oriented Workload Control release methods are included. Simulation results show that Workload Control release methods lead to better performance than drum-buffer-rope if bottleneck severity is low. But Workload Control, including its bottleneck oriented release methods, is outperformed by drum-buffer-rope if a strong (or severe) bottleneck exists. Workload Control gains an advantage in balanced shops due to its unique load balancing function, which attempts to evenly distribute workloads across resources. But this becomes functionless when there is a strong bottleneck. Our sensitivity analysis suggests that the performance differences between release methods are not affected by routing characteristics or the proportion of jobs that visit the bottleneck.