Development and application of KEPRI realistic evaluation methodology (KREM) for LB-LOCA

A realistic evaluation method for LB-LOCA of a PWR, KREM, is developed and its applicability is confirmed to a 3-loop Westinghouse plant in Korea. The method uses a combined code of CONTEMPT4/MOD5 and a modified RELAP5/MOD3.1. RELAP5 code calculates system thermal hydraulics with the containment backpressure calculated by CONTEMPT4, exchanging the mass/energy release and backpressure in every time step of RELAP5. The method is developed strictly following the philosophy of CSAU with a few improvements and differences. Elements and steps of KREM are shown in Figure this paper. Three elements of CSAU are maintained and the first element has no differences. An additional step of 'Check of Experimental Data Covering (EDC)' is embedded in element 2 in order to confirm the validity of code uncertainty parameters before applying them to plant calculations. The main idea to develop the EDC is to extrapolate the code accuracy which is determined in step 8 to the uncertainties of plant calculations. EDC is described in detail elsewhere and the basic concepts are explained in the later section of this paper. KREM adopts nonparametric statistics to quantify the overall uncertainty of a LB-LOCA at 95% probability and 95% confidence level from 59 plant calculations according to Wilks formula. These 59 calculations are performed in step 12 using code parameters determined in steps 8 and 9 and operation parameters from step 11. Scale biases are also evaluated in this step using the information of step 10. Uncertainties of code models and operation conditions are reflected in 59 plant calculations as multipliers to relevant parameters in the code or as input values simply. This paper gives the explanation on the overall structures of KREM and emphasizes its unique features. In addition, its applicability is confirmed to a 3-loop plant in Korea. KREM is developed for the realistic evaluation of LB-LOCA and its applicability is successfully demonstrated for the 3-loop power plants in Korea. It is developed following the philosophy of CSAU with an additional step of EDC, to confirm the uncertainty parameters reproducing the code accuracy. Overall calculational uncertainty is quantified from 59 plant calculations using non-parametric statistics and its validity is confirmed by performing as many as 640 calculations. 30 phenomena of 11 components are taken into consideration and 27 uncertainty parameters are used for sampling plant calculations. The biases due to the scale distortions of ECC bypass and steam binding are independently evaluated, and the uncertainty of 10 K due to plotting frequency is additionally considered. When the uncertainties are taken all together, final PCT is 1319 deg K (and local oxidation and core-wide hydrogen generation are found to leave sufficient margins to the acceptance criteria