Sample size for failure detection and reliability demonstration

Motor vehicle reliability is commonly measured by nonconforming conditions per hundred (C/100). Two approaches based on the Weibull distribution were developed to determine the test quantity necessary for (1) detection of any desired C/100 rate or (2) reliability demonstration. Tables are presented summarizing results that allow the test designer readily to establish test quantity to meet specific reliability program requirements. Trade-off effects are also illustrated, i.e., as test length is reduced, test quantities must be increased to meet the same objective. Although the methodology was developed for automotive application, it can be applied to any product at the system or component level