5. Normality analysis

In the first step, we saw the legitimate reasons why we can initially assume that the dispersion of manufacturing processes can be modeled by the normal distribution. However, this assumption is not always verified. Some product characteristics are "bounded", such as freezing temperature, or the geometric characteristics of mechanical parts (flatness, straightness, eccentricity, etc.). In these cases, normality is not necessarily the rule. Performance in describing the "non-conformity" rate based on Cp and Cpk indicators can then be impacted. For this reason, we need to think about the position of the manufacturing process distribution in relation to the required tolerances, so that we can consider other capability indicators, the aim always being to determine the rate of non-conformity generated by the process.

Numerous statistical tools can be used to determine...