Track: Quality Control and Quality Management

Abstract

Acceptance sampling plan is a statistical tool of quality control. It is a well-considered decision system that meets both consumers and producers need for product quality. The purpose of acceptance sampling plan is to provide a decision-making guideline, whether or not the lot is likely to be accepted. Many efforts were directed to make sure that we are meeting the requirement in the most cost-efficient way. A common practice is single acceptance sampling plan, because of its simplistic. However, with its simplistic comes with a downside that a larger sample size is usually required to avoid making a judgment mistake, because there is only a chance to sentence the lot. In this paper, for the expectation of reducing the required sample size, we propose a two-stage sampling plan to give a questionable lot another chance of inspecting samples before deciding to accept or reject the lot. The proposed plan is designed on the basis of one-sided capability index with the assumption of the quality characteristic follows a normal distribution. The concept of two-stage sampling is that decisions in the second-stage sampling inspection is independent to the first-stage sampling inspection, having the desired to operate on variables inspection. The optimal parameters are determined by optimization of the design model, which minimizes the average sample number (ASN) and satisfies the producer's and the consumer's risks at the specified quality levels. The utility of the propose sampling plan is shown through comparison with the single sampling plan in terms of sample size required for inspections under various combinations of quality levels and required risks. The result indicates that two-stage sampling plan involves smaller required sample than that corresponding to the single sampling plan. Thus, the propose plan proceeds towards to help practitioners efficiently sentence the submitted lots.