A conditional decision procedure for high yield processes

Abstract

A control chart based on the cumulative count of conforming items between two successive nonconforming ones has been shown to be useful in manufacturing industries, particularly for high-quality processes. However, as the decision is based on a single count value, it is relatively insensitive to process shifts. In this paper a conditional procedure is proposed whereby the sensitivity is improved when the process shift is moderate to large in either direction. In addition, optimal limits are defined in such a way that the average run length becomes maximum when the process average is at the nominal level. The performance of the conditional chart and its optimal limits are investigated and compared with the traditional case. The idea of a conditional procedure is to utilize some of the previous runs when a count value exceeds the limits. This procedure is similar to the supplementary run rules, but the conditional procedure is used when the process is out side of the control limits whereas the supplementary run rules focuses on the in-control situation. The conditional procedure has tighter limits than its original chart. The tighter limits together with the conditionally take care of the process without sacrificing its original in-control probability whilst improving the sensitivity.