Research On Uncertainty Analysis Methods Of Core Physics Calculations

With the U.S. Nuclear Regulatory Commission (USNRC) regulatory changes, the traditional method of analysis has been best estimate plus uncertainty analysis methods(BEPU) replaced with a 95% / 95% tolerance limits of the analytical results have been IAEA(IAEA) are supported. The best estimate for the study abroad plus uncertainty analysis has been carried out a lot, but the country is still little research in this area, the need for more research into the study of this aspect.Uncertainty analysis methods mainly on analytical methods to determine the analytical methods and statistical analysis methods to determine the statistical theory of ratio analysis method has more advantages, more and more popular. This paper analyzes the uncertainty analysis using statistical methods. Which Latin hypercube sampling method using analytical methods based on non-parametric statistical methods Wilks theory.This paper analyzes the sources of uncertainty in the core physical computing systems,the preparation of the core physics calculations needed for uncertainty analysis procedures.Before uncertainty analysis, sensitivity analysis of the core physics calculations. Use CASMO for component calculations, analyze the impact of changes in component parameters for the components of the calculation results that sensitivity. Multi-parameter studies conducted to study the correlation between parameters. SIMULATE be calculated using the core to analyze the effect of the core parameters on core computing.Uncertainty analysis module homogenized computing. Made from hot full power (HFP)to the cold conditions of uncertainty calculations to get cold from HFP process due to the uncertainty caused by the component parameters on the reactivity of uncertainty. The critical factor for the proliferation of the system is estimated uncertainties component parameters. In addition to calculating the multiplication factor of uncertainty, the correlation coefficient was also calculated, which represents the similarity between the two critical systems. The uncertainty analysis of whole core diffusion calculation to obtain the information in there is uncertainty affecting the tolerance limits.Finally, the last chapter discuss the acceleration method that agency model technology,compared the effects of various agents in dealing with the core model of physical computing problems, which mainly include the response surface models, multivariate adaptive regression splines model, Kriging model, artificial neural network model.The results show that the order of non-parametric statistical methods to meet the needs of the core physics calculations uncertainty analysis. Enrichment is the most critical component design parameters, the parameters are relevant,assembly homogenization calculate the uncertainty brought uncertainty to subsequent calculations. Calculation of the reflective layer reflective layer parameters are not sensitive. PWR components in the high similarity, the uncertainty component decreased ’self-stabilization’ of the physical phenomena. The uncertainty of the physical characteristics of the core are within an acceptable range, the central power core large uncertainty. In the uncertainty analysis of the core physical computing, Kriging surrogate model works best.