PWR pin-homogenized cross-sections analysis using big-data technology

Abstract

To accurately and reliably predict the behavior of the nuclear reactor core, pin-by-pin fuel management calculation is becoming the next most possible methodology for the Pressurized Water Reactor (PWR). The few-group constants parameterization, however, would be the most challenge problem due to the large number of few-group constants and the complicated pin-cell states during the entire operation history of the reactor core. In this study, the big-data analysis technologies were employed to find a possible relationship between the pin-cell homogenized few-group constants and the pin-cell homogenized nuclide densities. Firstly, correlation analysis demonstrated a strong multicollinearity between different nuclide densities, which implies a significant internal structure and interaction among them. Secondly, by using the factor analysis, the large number of nuclide densities were reduced to a small group of nuclide density factors that can reflect the main information of their variation trends. Thirdly, multiple regression analysis between the nuclide density factors and few-group constants verified the possibility of using nuclide density in form of driven factors as the state parameters to predict the few-group constants.