VALIDATION APPROACHES FOR COMPUTATIONAL WELD RESIDUAL STRESS MODELING

摘要

In pressurized-water reactors, weld residual stress can drive primary water stress corrosion cracking (PWSCC). Weld process models, based on non-linear finite element techniques, can provide estimates of weld residual stress that are useful in assessing risk of PWSCC, and designing plant maintenance programs (e.g., scheduling of inspection and repair). A weld model validation approach, comprising comparison of weld model output against an established benchmark, would enable an assessment of model output precision and bias; however, the nuclear industry lacks a consensus weld model validation approach. The paper provides technical detail for a range of example data analysis approaches that provide potential validation metrics for assessment of model output relative to a benchmark, where relevant benchmarks are derived from either model or measurement data. The potential validation metrics range from simple (e.g., root mean square difference) to complex (e.g., predicted crack growth behavior). We further apply the data analysis approaches to two sets of weld model outputs, each developed in the context of an industry round robin exercise organized by the U.S. Nuclear Regulatory Commission and the Electric Power Research Institute. The analysis results provide data for consideration within the larger technical community, to support development of a consensus approach to weld residual stress model validation.