Sample size determination of bogey tests without failures

Highly reliable products that will last for a long time can result in the impracticality of conducting life testing at the designed normal usage conditions. When accelerated life tests can be applied to represent the normal product operating conditions with a shorter testing time, engineers usually terminate the accelerated life test upon successfully running to a multiple of a pre-specified bogey without any failure. This testing philosophy is called ‘bogey test’ or ‘extended test’ in the automotive industry and is a subset of type I censored tests in standard reliability literature. This paper focuses on the determination of sample size when planning or assessing the applicability of such a bogey test. The underlying life characteristic is assumed to follow a Weibull distribution with a given shape parameter. In addition, the required reliability and confidence levels at the bogey to be demonstrated are pre-determined.