Consideration of Uncertainty in Reliability Demonstration Test Planning

The risk of a failed test (type-II statistical error) is rarely considered in both End of Life and Success Run Tests (Zero Failure Tests) in terms of reliability demonstration testing. In order to make the right decision under uncertainty, the remaining risk can be calculated for the considered sample size using numerical or approximative-analytical approaches. As with any other hypothesis test, prior knowledge is necessary to estimate the distribution of the alternative hypothesis. If this prior knowledge is stemming from life tests with very small sample sizes, the information is subjected to uncertainty, that has to be considered when planning the test. This paper presents an approach to consider the uncertainty in reliability demonstration test planning within a numerical-bootstrap and an approximative-analytical approach. The implications are illustrated using some exemplary results.

Keywords: Uncertainty, Probability of test success, Reliability demonstration, Test planning.

Download PDF