Research on Scientific Derivation of Control Limits in Control Charts

Abstract

Control Charts (CC) are the means to “manage the process behaviour” by analysing subsequent samples at regular intervals of time.: good decisions depend on Scientific analysis of data. Often, the data are considered Normally distributed; this is not completely right; data must be analysed according to their distribution: decisions are different with different distributions, because the Control Limits of the CC depend on the distribution. We compare our findings with Shewhart findings; later we extend the ideas to deal with “rare events”, with data not Normally distributed; we compare our results, found by RIT, for various cases in the literature: there is a big difference between the Shewhart CC and the Time Between Events CC; considering that, future decisions of Decision Makers will be both sounder and cheaper, when data are not normally distributed. ARL depends on the data distribution, not only on the “false alarm rate”. The novelty of the paper is due to the Scientific Way of Computing the Control Limits, both for the mean and for the variance.