Management of measurement uncertainty for effective statistical process control

Abstract

In the context of quality assurance strategies, statistical process control techniques and conformance testing are necessary to perform a correct quality auditing of process outcomes. However, data collection is based on measurements and every measurement is intrinsically affected by uncertainty. Even if adopted instruments are in a condition of metrological confirmation, random and systematic measurement errors can not be completely eliminated. As a consequence, wrong measurement-based decisions can seriously decrease companies profits because of larger costs of repairing and shipping, as well as for the loss of reputation associated with customers dissatisfaction. This paper deals with the problem of the growth in false acceptance risk due to both random and systematic errors. In particular, it is shown that a systematic contribution, that often is the prevailing kind of uncertainty associated with the use of an electronic measuring equipment can seriously increase the worst-case risk to take wrong decisions when performing various statistical process control procedures. Nevertheless, such risks can be greatly decreased by selling suitably the Test Uncertainty Ratio associated with the measuring equipment.