An Improved Aggregation Method for Performance Analysis of Bernoulli Serial Production Lines

Aggregation method has been widely utilized to evaluate the performance measures of production lines. However, traditional aggregation method (TAM) has low prediction accuracy in production lines with multiple bottlenecks, such as “inverted bowl” lines and “oscillatory” lines. Therefore, the root causes of the low prediction accuracy are first investigated. Extensive numerical studies indicate that A-units are one of the major causes, where a A-unit is defined as the subsystem between two consecutive bottlenecks. Then, an improved aggregation method (IAM) is proposed to improve the prediction accuracy of TAM. IAM is established by extending the traditional two-machine aggregation building blocks to general multimachine aggregation building blocks in A-units. Specifically, for a small-scale A-unit, an aggregation building block is established for all the machines and buffers in the A-unit. For a large-scale A-unit, a heuristic rule is proposed to divide the A-unit into several small-scale production line segments, where an aggregation building block is established for each segment. Numerical studies indicate that IAM can effectively improve the estimation accuracy of the aggregation method while maintaining a reasonable computational efficiency.