Abstract

This study uses simulation to assess the performance of alternative methods for detecting momentary bottlenecks in high-variety contexts that produce on a to-order basis. The results suggest that using the utilisation level of a station to detect bottlenecks leads to the best performance, but that this method suffers from high nervousness. Using the active period of a station appears to be a better overall choice for practice given its good performance and low nervousness. Meanwhile, methods that focus on the workload at a station are a viable alternative, but they may become dysfunctional in shops with directed routings and a limit on the queue. This negative effect is even stronger if the corrected workload measure is used, as recently suggested in the literature on short term capacity adjustments. Finally, using the inter-departure time detection method leads to the worst performance since: (i) it counterintuitively detects non-bottlenecks instead of bottlenecks; and, (ii) it is based on historical data, leading to a response delay.

摘要

本研究使用模拟来评估检测瞬时的替代方法的性能在按订单生产的多样化环境中的瓶颈。结果表明，使用站点的利用率水平来检测瓶颈会导致最佳性能，但这种方法存在高度紧张的问题。鉴于其良好的性能和低紧张性，使用站点的活跃期似乎是更好的整体练习选择。同时，专注于车站工作量的方法是一种可行的替代方案，但在具有定向路线和队列限制的商店中，它们可能会变得功能失调。如果使用校正后的工作量衡量标准，这种负面影响会更加强烈，正如最近在有关短期容量调整的文献中所建议的那样。最后，使用出发时间检测方法会导致最差的性能，因为：(i) 它违反直觉地检测非瓶颈而不是瓶颈；(ii) 它基于历史数据，导致响应延迟。