Application of Gaussian process model to generate a success criteria map with estimate of safety margin and uncertainty analysis

Abstract

A multi-dimensional success criteria map for a Level 1 PSA model is generated for the injection phase of a LBLOCA in a typical PWR that integrates the safety function of three systems: safety injection tanks (SIT), high-pressure safety injection, and low-pressure safety injection, using response surfaces derived from a Gaussian process model regression methodology. The safety margin is quantified for multiple configurations of the safety systems and as a function of the emergency diesel generator start, warm-up, and loading sequence timing representing the dominate sequence timing parameter relevant to the event tree. The maximum allowable warm-up time is identified that represents a bifurcation point in the thermal hydraulic response of the plant to the LBLOCA following SIT injection and the start of active safety injection systems. The demonstrated methodology retains high fidelity treatment of many input uncertainties during best estimate simulation of the transient, but allows the analyst to focus on a regression analysis on the most important application specific parameters thereby overcoming the curse of dimensionality inherent to the analysis of complex systems.