Determination of Prototype Test Requirements to Meet Performance Specification and Reliability Target

Usually it is desirable to predict the reliability of a new product while it is still undergoing design and development. Additionally, It is also usually desirable to determine the number of prototypes that need to be tested, and the minimum number of cycles they should pass, without any failures, under specified loading conditions, to ensure that the final product will pass the specified life cycle requirement under specified loading and have the required reliability at the required confidence level. Generally, such reliability prediction information is required before any prototypes are available. This paper attempts to answer these questions by means of Weibull analysis, and by utilizing past reliability data of products that closely approximate the new product being developed. In addition to demonstrating the use of Larson's Binomial Distribution Nomograph in establishing sample size requirements for reliability tests, this paper looks at accelerated testing with a view to reduce sample size requirements by testing longer or harder than the design specification. Past reliability data on power mirror switches are utilized here to arrive at recommended test requirements for future development programs.