Global Sensitivity and Registration Strategy for Temperature Profile of Reflood Experiment Simulations

摘要

This paper illustrates the capability of a global sensitivity analysis (GSA) framework applied to the TRACE thermal-hydraulics (TH) system code in the context of selected flooding experiments with blocked arrays reflood experiments. The proposed GSA framework deals with functional outputs (temperature profiles) and aims at quantifying the sensitivity of a specific feature of the reflood curve (its amplitude) to the physical parameters of TRACE. The framework uses a registration strategy based on the Square Root Slope Function (SRSF) transform to separate the amplitude and phase of the temperature profile. The registration is followed by a dimension reduction on principal component basis and the estimation of Sobol' sensitivity indices. This paper compares the SRSF registration to the more traditional landmark registration and shows its excellent properties. Given the simple nature of the reflood curve, the Sobol' indices obtained on the amplitude of the reflood curve also compare well with those obtained on the scalar maximum temperature of the curve. This suggests the framework to be of interest for deriving the sensitivity of the amplitude of more complex TH transients to the physical parameters of the code.