View article

Acceptance sampling plan for attributes is widely used for product quality inspection at the in-line (semi-finished products) and end-line (finished products) manufacturing process. Acceptance sampling plan for attributes, adopted by manufacturing industries generally follow the international standards of ISO 2859-1. However, some literatures suggested that procedures outlined by the ISO 2859-1 standard especially the “Acceptance Quality Limit”(AQL) which are used as the parameter to select a sampling scheme is often confusing. This is with regards to its definition which is found to be inconsistent throughout many sections of the standard. In one section, AQL is defined as quality level of the worst tolerable process averages that AQL should not be confused with the value of process average in the other sections. The concern is that the standard procedure based on AQL does not employ a well-defined process parameter that can reflect the actual risk of implementing sampling plan. Due to this, the suppliers lack relevant information to design the appropriate sampling plan to shift from 100% in-line process to sampling inspection. This study employs the acceptance sampling inspection plan for attributes in a semiconductor company: Company A by incorporating probability distribution of its in-line 100% inspection historical data. The result shows that by applying the actual process parameter to probability distribution, the in-line process average provides more insight into designing a sampling inspection plan. This proposed methodology is appropriate for the suppliers who plan to shift from the 100% internal in-line process to sampling inspection.