The development of a customer-centred, response surface methodology-based robust parameter design optimisation model under a moderately skewed production process

Abstract

The philosophy of quality by design (QbD) has been developed to facilitate a proactive improvement by incorporating quality into production processes. The robust parameter design (RPD), one of the most effective QbD methods, identifies optimum operating conditions that achieve a target value with minimum variance. However, the vast majority of RPD models have been developed with a manufacturer's point of view based on the assumption that the process is normally distributed. This paper contributes to the current body of knowledge by reconsidering traditional RPD concepts in two ways. First, it is known that quality should be viewed from the customer point of view. Final products are often subjected to screening inspection and only conforming products are distributed to the customer, while rejected products are scrapped or reworked. This inspection of the products results in a truncated distribution. Second, real-world process distributions for smaller-the better and larger-the-better quality characteristics are often skewed. Accordingly, this paper develops response surface methodology-based RPD models by introducing a skew normal distribution and integrating the customer's perceived truncated statistics into RPD. Comparative studies are also presented. Finally, numerical examples show that the proposed optimisation schemes are superior over the traditional counterparts.