Probabilistic failure assessment of nuclear piping components
This paper presents results of a research project aiming at improving probabilistic analysis methods and developing computational tools for estimating failure probabilities of ductile piping components with crack-like defects. The main focus of the study is an extension of the failure assessment diagram (FAD) approach to accomplish robust and reliable calculations of small failure probabilities, as usually required for safety relevant nuclear components. Specifically, the first- and second-order reliability methods (FORM, SORM) as well as different sampling algorithms are investigated as to their performance and numerical stability. In particular, problems resulting from the presence of multiple design points, e.g. within the transition range from ductile fracture to plastic collapse, are shown to be potential sources of numerical inaccuracies, and an appropriate solution is derived for such cases. Subsequent benchmark calculations reveal a high efficiency of the improved numerical algorithms within the whole range of failure probabilities. Another key aspect considered in this study concerns the statistical description of material data, e.g. crack resistance curves, involved in probabilistic analyses. Finally, the methods and computer tools developed within the framework of this research are applied for evaluating some pipe burst tests described in the literature.