DISSIMILAR METAL WELD PWSCC INITIATION MODEL REFINEMENT FOR XLPR PART Ⅱ: A STATISTICAL FRAMEWORK FOR THE INTEGRATION OF FIELD AND LABORATORY DATA

摘要

This paper-the second paper in a two part series-presents a generalized statistical framework for the integration of laboratory and field data in the development of a probabilistic model for the prediction of the PWSCC initiation time. The fidelity of such a model is critical to the probabilistic prediction of leakage and instability risks associated with PWSCC susceptible components. The statistical framework presented in this paper applies a laboratory data set for refinement of underlying model dependencies and a field data set for regression of the aggregate initiation (failure) time model. This statistical framework addresses several challenges that arise in practice including treatment of: a) heterogeneous conditions, b) time-varying conditions, c) suspended item data, d) the distinction between detection versus initiation, e) multiple flaw initiation and f) missing or poorly quantified data.