An improved approach for constructing lower confidence bound on process yield

Process yield, the percentage of processed product units passing inspection, is a standard numerical measure of process performance in manufacturing industry. Based on the expression of process yield, Boyles (1994) presented a yield-measure index, S<SUB align="right"><SMALL>pk</SMALL></SUB>, for normally distributed processes. In order to compute the index value, sample data must be collected and a great degree of uncertainty may be introduced into the yield assessment due to sampling errors. To remedy for this, several existing techniques have been applied to construct the confidence bounds for S<SUB align="right"><SMALL>pk</SMALL></SUB>. In this article, an alternative approach is proposed to construct the lower confidence bound for S<SUB align="right"><SMALL>pk</SMALL></SUB>. To examine and compare the performances of the proposed generalised confidence intervals (GCIs), a series of simulations is conducted. The results show that the proposed GCIs approach is superior to the standard bootstrap in terms of coverage rate. Therefore, this article recommends GCIs approach for assessing the process yield in real applications. [Received 4 November 2010; Revised 5 January 2011; Accepted 7 January 2011]