EMPIRICAL EVALUATION OF EXPONENTIAL AND INDEPENDENCE ASSUMPTIONS IN QUEUEING MODELS OF MANUFACTURING SYSTEMS

Robert R. Inman

Production and Operations Management · 1999 · 150 citations · 35 references

Concepts

TL;DR

The study uses real data from two automotive body welding lines to examine the common queueing network assumptions of independence and exponentiality, which are often adopted for mathematical tractability. The study aims to (1) present realistic randomness data from two manufacturing systems to guide research and (2) evaluate the validity of the independence and exponentiality assumptions in automotive manufacturing. The authors analyze real processing, interarrival, failure, and repair time data from two welding lines and test the independence and exponentiality assumptions against these empirical distributions.

Abstract

This paper presents actual data (processing times, interarrival times, cycles‐between‐failures, and time‐to‐repair) from two automotive body welding lines. The purpose is twofold. First, to help researchers focus their work on realistic problems, we exhibit the nature of randomness actually found in two industrial manufacturing systems and provide a data source for realistic probability distributions. Second, we assess the validity of two common assumptions regarding this randomness in automotive manufacturing. Many queueing network models assume that certain random variables are independent and exponentially distributed. Though often reasonable, the primary motivation for the independence and exponentiality assumptions is mathematical tractability.

References

35