A methodology for dealing with uncertainties on NDE data when used as inputs to fracture mechanics analyses

This study is directed at two main objectives, firstly to carry out a review of the different methods of presentation of NDE performance and reliability data and secondly to define a methodology of combining uncertainties in NDE results with Fracture Mechanics analyses. The literature review has revealed that in past experimental trials there have been several different, albeit interrelated, methods for presentation of NDE performance and reliability data. The three most common are ``Probability of detection``, ``Mean sizing error and standard deviation`` and ``Response versus size`` data. In the context of NDE and Fracture Mechanics, it is concluded that the measured response versus actual flaw size data presentation provides the most useful quantitative information. It is also recommended that, where possible, a re-analysis of past NDE data is carried out in the form of response versus size graphs and that future trials should ideally include data presented in this format.