Developing variables two-plan sampling scheme with consideration of process loss for lot sentencing

Abstract

This paper considers a well-known two-plan sampling system, Tightened-Normal-Tightened (TNT) sampling scheme, which includes two single sampling plans (SSPs) with regulations for transitioning between tightened and normal inspections. Tightened inspection is conducted when there is a noticeable decline in quality, and normal inspection is employed when the quality is satisfactory. Most acceptance sampling plans evaluate product quality in terms of process yield, which does not differentiate between products that fall within specification limits. To address this issue, a new performance measure called the process loss index Le was developed, which accounts for the quality loss function. This index is determined by dividing the “expected quadratic loss” by the “square of half the specification width.” Using this index, two primary types of variables TNT sampling scheme were constructed and evaluated in this study. Their efficacy and features were investigated, discussed, and compared. The findings reveal that the TNT sampling strategy delivers the required quality assurance with a reduced sample size compared to the conventional SSP.