Uncertainty analysis for containment response of U.S. EPR TM reactor to large break loss-of-coolant accidents

This paper presents an uncertainty analysis applying the GOTHIC containment analysis code to simulate the first 24-hours following a large-break loss-of-coolant accident (LBLOCA) in AREVA's U.S. EPR TM plant. The uncertainty method is modeled after a study performed by the Gesellschaft fur Anlagen und Reaktorsicherheit (GRS) using data from the Heifidampf-Reaktor (HDR) Test T31.5. The analysis method incorporates an assessment of phenomenological importance, identifying the dominant contributors that influence the principle analysis metric, containment pressure. As with the GRS approach, this study employs non-parametric statistics. This analysis illustrates U.S. EPR containment response sensitivity to realistic variation in a set of important model parameters influencing containment conditions during LBLOCA. In considering a set of model uncertainty parameters, a number of GOTHIC variation calculations were performed (59 calculations) to effect a best estimate plus uncertainty result at 95/95 coverage/confidence level for the key metric, containment pressure. The results of the importance analysis showed condensation phenomena on the surface of the containment structures to be important during the passive cooling period, which occurred prior to the start of HL (hot leg) injection of SI (safety injection). In this study, hot leg injection was assumed to initiate at 1.5 hours. Condensation phenomena faded in importance after 1.5 hours due to the hot leg injection of SI suppressing steaming. Structure conduction, especially, the physical properties of concrete, retained importance throughout the transient. (authors)