Abstract

Estimations of probability of detection (POD) for examinations performed in the nuclear power industry are derived from empirical studies performed in laboratory-like environments on test blocks that simulate relevant component geometries in the field. The data generated by these studies provide technical bases to inform regulatory decision making. However, due to the physical size of many components, the resources required to manufacture test blocks and to perform a study can be extensive. As a result, sample sizes are often determined by feasibility and confidence bounds for resulting POD estimates can be large. Further, because the data are collected under well-controlled laboratory conditions, it is possible that these studies provide non-conservative estimations of POD relative to examinations performed in the field. Model-Assisted POD (MAPOD) concepts can potentially enhance existing POD estimations without requiring additional empirical testing. This work provides an overview of MAPOD in the context of nuclear power applications and presents existing POD estimations for dissimilar weld components that have been obtained by empirical testing. Finally, discussion is provided regarding how MAPOD concepts may enhance the existing POD estimates for reducing uncertainty or to more accurately reflect field POD.

How to Cite:

Meyer, R. M., Holmes, A. E. & Lin, B. ., (2019) “Probability of detection estimations for dissimilar metal welds”, Review of Progress in Quantitative Nondestructive Evaluation .