Mean station reliabilities cause throughput overestimates in production system design

In production system design, we typically lack actual station reliability data for throughput analysis using a model, because these stations do not yet exist. Hence, we either use the mean values from similar existing equipment or mean reliability estimates provided by equipment manufacturers. The real stations may have better or worse reliabilities compared to the means. Hence, when the system is built, the real system throughput may be acceptable or poor depending on actual station reliabilities. This paper compares predicted model throughput using station mean reliabilities with real system throughput. We find that the model often overestimates system throughput. We develop an upper bound and the maximum probability of overestimation when there is an infinite buffer size after each station. We also provide the ranges of overestimation for systems with limited buffers. These results may be used as a “rule of thumb” to adjust system throughput estimation. Monte Carlo simulation is discussed as an approach to analyze real system performance.