28th Symposium of AER on VVER Reactor Physics and Reactor Safety (2018, Olomouc, Czechia)
Reactor physics experiments and code validation (benchmarks)
Branislav Vrban, Štefan Čerba, Jakub Lüley, Filip Osuský, Vladimír Nečas (Slovak University of Technology)
The criticality safety criteria require that the effective multiplication factor of the investigated system is less than the defined limits. Therefore, the key issue in any criticality safety problem is to estimate all possible uncertainties and to predict consequent deviation of the calculation from reality. If the calculated value is not equal to its true value bias occurs. The fundamental assumption is that the computational bias is mostly caused by errors in the cross-section data. In addition, the use of random variables in the calculation introduces a non-random bias in the computed result as well. The American National Standards are utilized to predict and bound the computational bias of criticality calculations. These standards require the validation of the analytical methods and data used in nuclear criticality safety calculations to quantify the computational bias and its uncertainty. This paper presents a method for determining the computation bias and bias uncertainty for SCALE KENO-VI code focusing on VVER-440 technology. For this analysis a 3D core model was developed by B&J NUCLEAR ltd. company. Several calculation steps are used to address bias estimation method including sensitivity analysis, uncertainty analyses and cross-section adjustment method. In addition, the neutronic similarity of VVER-440 core to several hundred critical benchmark experiments is evaluated by the use of three integral indices. The database of the benchmark experiment is based on the selection and processing procedure VALID provided by the Oak Ridge National Laboratory and specified in the IHECSBE. The results of all analyses performed are given and discussed in the paper.