Uses for model-based probability of detection curves

Evaluating the reliability and capabilities of nondestructive evaluation (NDE) methods has normally been done empirically with many costly round-robin test block exercises. These results are necessarily problem specific and limited in scope. Here, an alternative approach is described where model-based simulations are used to predict the probability of detection versus flaw size curves for components which can then be validated experimentally with a few samples. It is demonstrated how knowledge of these curves can be used to evaluate NDE inspection performance, to determine the influence of NDE tests on component reliability, and to design components with built-in inspectability criteria.