Bottleneck-oriented order release with shifting bottlenecks: An assessment by simulation

Bottleneck shiftiness is an important managerial problem that has received significant research attention. The extant literature has shown, for example, that protective capacity reduces the likelihood of the bottleneck shifting. Yet the actual performance impact of a bottleneck shift has been widely neglected. We posit that there are at least two interrelated effects that may impact shop performance: (i) the direct effect of the change in bottleneck position; and, (ii) the indirect effect of the order release method incorrectly identifying the bottleneck (i.e. assuming the bottleneck is Station X when it is actually Station Y). The latter is particularly acute in the context of bottleneck-oriented order release methods such as Drum-Buffer-Rope (DBR) as these release methods use feedback from the (assumed) bottleneck to control release. Using controlled simulation experiments we demonstrate that a bottleneck shift to a station upstream of the assumed bottleneck has a negligible effect on DBR performance while a downstream shift is detrimental to performance. Meanwhile, the distance, i.e. the number of stations between the actual and assumed bottleneck, has a negligible performance impact. These results have important managerial and research implications for DBR and other release methods.