Study on nonparametric statistic method applied to nuclear criticality safety analysis of spent fuel rack

Abstract

Nonparametric statistic method is a statistic method which does not depend on a certain distribution. Nonparametric statistic method can calculate tolerance limits under certain tolerance percentile and confidence level through sampling methods. The number of samples of the nonparametric statistic method is related to the tolerance percentile and confidence level instead of the number of sampling parameters. The tolerance limit of nonparametric statistic method is compared to the result of conservative method and that of nominal parameters’ values. The ordered sequence of parameter sensitivity is obtained from multiple parameters sensitivity analysis based on the correlation coefficients. In this paper, nonparametric statistic method is used in the spent fuel rack model to analyze the uncertainty of manufacturing parameters, nuclide composition from assembly depletion calculation and reaction cross sections in criticality calculation code. The statistical analysis of calculation results demonstrates that the 95/95 tolerance limit of nonparametric statistic method is smaller than the value of conservative method, which proves that the method of calculating the uncertainty from every single parameter is conservative and shows the margin of conservation. R2, the square of linear correlation coefficient of each scatter diagram, is used to quantify the sensitivity of each parameter to keff. The sequence of parameter sensitivity is stable for a certain condition of a certain model which will be different if the working condition is changed. The sensitive parameters are obtained from sensitivity analysis and should be paid close attention to in the process of designing or manufacturing relevant equipment.