EXPANDED MEASUREMENT UNCERTAINTY IN 100% INSPECTIONS

How to apply the second edition of VDA Volume 5 in practice

19 June 2019: Morteza Farmani

Customer requirements are easy (to understand): zero defects in all parts. Since a 100% inspection is usually not economical, companies frequently apply statistical process control (SPC) in industrial production. Even though, realistically viewed, this method does not guarantee that 100% of the parts are free from defects, it leads to a proportion of defects within a range that can only be measured in parts per million (ppm).

Being a major industry, automotive manufacturers demand a minimum process capability Cp/Cpk of 1.67 from their external suppliers; this equals a theoretical maximum share of rejects of 0.6 ppm. In case they meet this requirement, quality control charts control the process based on subgroups. When they are not able to satisfy this requirement or a sample indicates a process failure, the parts produced (since the last subgroup o.k.) become subject to a 100% inspection before they can be supplied to the customer. This is why it is not recommended to choose a subgroup size that is too large. Especially in series production, a subgroup n.o.k. quickly causes a 100% inspection of several thousand parts.

This article describes how measurement uncertainty affects 100% inspections. It provides you with background information and explains how to consider the expanded measurement uncertainty at the specification limits in practice.


Similar articles