A quantitative comparison of bottleneck detection methods in manufacturing systems with particular consideration for shifting bottlenecks

Abstract

There are numerous different bottleneck detection methods both in academia and in industry. However, most do not take the shifting of bottlenecks into account. This paper compares a number of methods, namely methods based on cycle times and utilizations, waiting times and queue lengths, the arrow method, the turning point method, the active period method, and the bottleneck walk. All methods are tested against two different manufacturing systems, a pseudo static system consisting of a static system that changes once halfway through the production run and is designed to have shifting bottlenecks, and a dynamic system that includes variations for a multitude of different processes.