Scaling-up process characterization

Abstract

Most of the literature on statistical process monitoring (SPM) concerns a single process or a few of them. Yet, in a modern manufacturing plant there may be thousands of process measurements worth monitoring. When scaling up, supporting software should provide three important capabilities. First, it should sort by appropriate summary measures to allow the engineer to easily identify processes of most concern. Second, it should show process health of all processes in one graph. Third, it should meet the challenges of big data for test multiplicity, robustness, and computational speed. This article describes the summaries and graphs that can be especially useful when monitoring thousands of process measures that are to be analyzed retrospectively. This article's goal is to propose these for routine reports on processes and invite discussion as to whether these features belong in the mainstream.