Planning Life Tests in Which Units Are Inspected for Failure

In some life tests, exact failure times cannot be observed. Instead failures are detected upon inspection. Equally spaced inspection times can be statistically inefficient, especially when a product has a decreasing failure rate. In such cases, equal spacing in log time provides an attractive alternative. This paper gives guidelines for choosing statistically efficient inspection times and the approximate sample size that achieves a specified degree of precision for estimating a particular quantile of a Weibull time-to-failure distribution. This information can be used to plan more efficient life tests.