Estimating the Duration of Failure-Based Reliability Demonstration Tests

It is well known that failure-based reliability demonstration tests (RDTs) are generally much more efficient than failure-free RDTs. The major disadvantage of failure-based tests, however, is that planning them properly is more cumbersome than for their failure-free counterparts. While approaches to estimate the test result and the test cost have already been presented in literature, this paper shows how the test duration can be estimated. As an obvious approach, a simulation using the Monte Carlo method is suggested. However, as simulations are lengthy, different approximation methods are derived as well. A simulation study is used as benchmark to evaluate and compare the approximation approaches. It is shown that one of them gives highly accurate estimates for the duration of RDTs in most relevant cases. For more individual and complex test situations, a region of competence is defined and the expected error is given. With these recommendations, a simulation can be avoided in most cases, drastically simplifying and accelerating the test planning process.