Nuclear data uncertainty propagation to integral responses using SANDY

Abstract

SANDY is a nuclear data sampling code compatible with nuclear data files in ENDF-6 format. Exploiting the basic theory of stochastic sampling, SANDY generates random nuclear data samples that reproduce the covariance information stored in the ENDF-6 files. Such random data are rewritten in perturbed ENDF-6 or PENDF files and can be used as nuclear code inputs to produce perturbed responses. After the statistical analysis of the sampled responses, the sample mean and variance are quantified. Not only can SANDY be used for the study of the response sample variance, but it can also estimate global sensitivity indices for correlated and uncorrelated input parameters using a variance-based decomposition method.SANDY was tested for the generation of random ENDF-6 files with perturbed cross sections and resonance parameters. These files were used for the uncertainty quantification of integral responses, such as the one-group integrated cross section and the resonance integral of several isotopes. Then, the response variance was apportioned to the several inputs and the parameters producing the largest impact were identified.