Use of Statistical Modeling in Product Inspection Plan Development

• Activity Number: 67
• Type: Topic Contributed
• Date/Time: Monday, August 12, 2002 : 8:30 AM to 10:20 AM
• Sponsor: Section on Physical & Engineering Sciences*
• Author(s): Yasuo Amemiya*+
• Affiliation(s): IBM T. J. Watson Research Center
• Keywords: mixed effect model analysis ; cluster sampling ; simulation-based probability estimation

Abstract

The development of hardware products, such as components in server and storage systems, is facing new challenges today. The reliability and performance of a hardware component depend heavily on the operating conditions determined by the customer usage pattern. Since the customer/user pattern varies considerably, simulating such conditions in product testing may be impossible. This testing inadequacy, combined with pressure for achieving technological limits in a shorter product development cycle, can lead to reliability problems in the field. Thus, it is often desirable to improve the product quality based on field data for the purpose of eliminating or alleviating the reliability problems. Since the design and manufacturing process can not be altered easily or quickly, one practical way to deal with this problem is to change the product inspection sampling strategy. This inspection is to be performed, not for process control, but for guaranteeing that the number of unacceptable component products being shipped to the field is very small. In this paper, an approach for developing such an inspection plan based on statistical modeling is proposed and illustrated.