Uncertainty evaluation of best-estimate multi-dimensional calculation during LBLOCA for APR1400 plant

Demand on Best-estimate (BE) calculation with uncertainty quantification for the emergency core cooling system (ECCS) performance analysis during Loss of Coolant Accident (LOCA) is continuously increasing in nuclear industries and regulations. A BE methodology, KINS-Realistic Evaluation Model (KINS-REM), has been developed to support the independent audit calculation for the regulation of Korean nuclear power plants, and the code accuracy and the related statistical method have been improved for a BE code, RELAP5/MOD3.3 thermalhydraulic system code. The MARS (Multi-dimensional Analysis of Reactor Safety) code having a multi-dimensional calculation capability was adopted to replace the RELAP5 code and the uncertainty evaluation scheme was modified to treat uncertain parameters appropriately using the nonparametric statistical method based on the third order Wilks formula in KINS-REM. The advanced power reactors (APR1400) to be evaluated has the ECCS design features such as four separated high pressure injection systems with direct vessel injection and fluidic device in safety injection tanks. Recent research results with MARS code showed that multidimensional (multi-D) reactor vessel model predicted more reasonably and accurately the thermal-hydraulic behaviors during LOCA than one-dimensional (1-D) model. Therefore, in accordance with KINS-REM, uncertainty evaluation using multi-D MARS models is performed and the peak cladding temperature (PCT) with 95% confidence and 95% probability level is compared with that of 1-D model for APR1400 plant. The uncertainty parameters and their ranges associated with plant operation and design, heat transfer model in MARS code and etc. are selected based on the experience, engineering judgments, and reference documents. Also, the sensitivity analysis is carried out in order to identify the influence of loss coefficients on the PCT in the internal of reactor vessel and study on the feasibility of an uncertainty parameter in the multi-dimensional model. (authors)