USING LATIN HYPERCUBE SAMPLING EXPERT SYSTEMS AS BASIS FOR BEST-ESTIMATE PLUS UNCERTAINTY METHODOLOGY IN RELAP5-3D ADVANCED THERMAL-HYDRAULIC CODE

摘要

The Code Scaling, Application, and Uncertainty (CSAU) process has been used to calculate the total uncertainty of parameters that limit plant operations for online commercial power plants. However, in its current form CSAU is unwieldy, requires considerable resources to evaluate important parameter uncertainties, and is performed in large measure using expert committees and response surface functions. A new, innovative approach designed to eliminate these shortcomings and to be automated when using an advanced thermal-hydraulic code such as RELAP5-3D consists of the following four parts. (i) Incorporation of the contribution of expert committees into expert systems. (ii) Replacement of response surface functions with Latin hypercube sampling (LHS). (iii) Automation of a significant portion of the process. (iv) Proof-of-principle application of the methodology assessed based on the existing CSAU data base. Each part is summarized and its inherent qualities outlined.