Quantitative Study of Geometrical Effect on CANDU Channel Power by Estimating Real Variance

In this study, the difference between the results using different core boundaries will be reported in which for the reliability and accuracy, the McCARD code will be used for the numerical simulations. However, because the McCARD code is based on the Monte Carlo method, which is a probabilistic theory, it accompanies probabilistic errors, or a so-called variance. Although the McCARD code provides us with the diverse analyses regarding a real variance such as the Hystoy-based Batch(HB) method and Fission Source Distribution(FSD) method, the most fundamental and accurate way is to simulate a certain problem infinitely and calculate its real variance. Because the difference between different boundaries can lie along the interval of the standard deviation, an accurate estimation of variance is required and thus 100 simulations are conducted in this study for channel power error estimation. Originally, more realistic problems and larger number of problems should be solved. However, in this research, only two problems, which are a mathematical initial core problem and a practical initial core problem are solved due to a limitation of the study scope. In this study, a comparison of the results using different shapes on the boundary is done and the real variance of the channel in CANDU core is also calculated. The results indicate that the boundary shape effect cannot be ignored and the real variance of the channel power in CANDU core is 5-7 times larger than that of the apparent variance. This results supports the rationales for the development of finite element method (FEM) based code such as the DEFENS code, because the FEM does not have limitations about the geometry