FURTHER COMMENTS ON VALIDATION APPROACHES FOR WELD RESIDUAL STRESS SIMULATION

Tensile weld residual stress (WRS) in the presence of primary water has been identified as a main driver for stress corrosion cracking in dissimilar metal welds found in the cooling circuit of pressurized water reactors. Thus, WRS data are needed to support plant management decisions, and so are of interest to a broad range of stakeholders in pressure vessel technology. In an effort to validate predictive WRS models and quantify modeling uncertainty, the U.S. Nuclear Regulatory Commission (NRC) and the Electric Power Research Institute (EPRI) have cooperatively organized a research program on WRS. This paper is a follow-up to earlier work, which described a series of WRS data analysis methods that provide a range of figures of merit, from simple (e.g., root mean square difference from a benchmark) to complex (e.g., comparison of predicted crack growth behavior), that can be used to judge WRS data quality. The present work applies those methods to assess model outputs developed during the second NRC/EPRI international round robin study (Phase 2b) on WRS modeling. The results of exercising the data analysis methods are presented, and compared to results obtained from an analysis of data from the previous round robin (Phase 2a).