LAUSR

Abstract A method of using small samples to achieve high-precision related variable sequence sampling was proposed. This method mainly uses two techniques: Latin hypercube sampling method for normal distribution and transform the covariance between variables of standard normal distribution sampling samples as unit. Applying this method, the fast reactor uncertainty analysis and sensitivity analysis program, named SUFR1.0, was developed, and conducted verification analysis. Using the NJOY2016 program, 33-group covariance libraries based on ENDF/B-VIII.0 and ENDF/B-VII.1 were developed for uncertainty and sensitivity analysis for fast reactors BN-600. Using the above 33-group covariance libraries, the uncertainty of the effective multiplication factor keff and the uncertainty of the power distribution of core were analyzed. It was found that the covariance matrix values between ENDF/B-VIII.0 and ENDF/B-VII.1, for some nuclides, such as 238U, 239Pu and 56Fe within some neutron energy range, are quite different. At the same time, these relatively different cross section types will also cause a large difference in the relative uncertainty of keff. For the relative uncertainty of power, the calculation result of ENDF/B-VIII.0 is significantly smaller than that of ENDF/B-VII.1.