Evaluating the reliability of ultrasonic testing of safety relevant components by model-assisted determination of the POD

The manufacturing quality and aging of components and materials of safety relevant constructions like components in nuclear power plants are of crucial importance. Material defects have to be detected and evaluated correctly at an early stage. Austenitic and dissimilar welds, for instance the circumferential pipe welds in nuclear power plants, still bear uncertainties for the inspection with ultrasonic techniques. Beyond early detection, the correct evaluation of defects in view of their location, size and geometry is highly important for the safe operation of such installations. In a common research project, MPA Stuttgart and Fraunhofer IZFP have addressed the evaluation of the influences of different factors on the reliability of ultrasonic testing procedures. Focusing on austenitic components, several representative test blocks with model defects as well as artificially grown, realistic defects have been examined. Based on the acquired datasets an analysis has been performed to determine respective POD-curves, quantitatively characterizing the scenario-specific Probability of Detection. To enlarge the statistical basis, simulation techniques have been applied to generate additional datasets to be included in the analysis. Here, a variation of relevant defect parameters such as depth and orientation has been performed to generate a broader basis for the respective probability distributions. We will briefly introduce the applied simulation technique and our approach to use synthetic data for the determination of the POD in an â-versus-a analysis. We will report on representative results of this Model-Assisted determination of the POD using a combination of experimental and simulated data. To overall objective of the mentioned research project is to use the combined results from a statistical evaluation of material properties and the NDT results presented here to develop a methodology for a probabilistic fracture mechanics assessment of component reliability.