The sampling based GRS uncertainty and sensitivity software XSUSA (Cross Section Uncertainty and Sensitivity Analysis) has been developed for the use with nuclear data uncertainties as contained in the nuclear covariance matrices. To this end, many calculations for the problem under consideration are performed with varied input data. The variations of the input data are generated randomly from the given probability distributions of the parameters including possible correlations between them.
When dealing with nuclear data and their uncertainties as contained in the covariance matrices, only the first two moments of the distributions (expectation values and covariances) are known. Due to this lack of information, the distributions are assumed to be of Gaussian shape. After performing all the calculations (typically 100 or more), the output quantities of interest are statistically analyzed, and their uncertainty ranges and sensitivities to the input parameters are determined.
As code system for the nuclear calculations, the SCALE-6 system is chosen. It has the advantages that it contains comprehensive sets of up-to-date nuclear cross sections as well as covariance data and almost all codes needed for the desired calculations, codes and data are well matched, and both data and source code are open. The tools necessary for varying the nuclear data corresponding to the sampling results, as well as handling the nuclear data in all steps of the calculation chain, have been developed within the XSUSA system. As transport solvers, the 1D SN code XSDRN so far has been used for pin cell criticality calculations, the 2D general geometry SN code NEWT for spectral calculations, including few-group cross section generation, and the 3D Monte Carlo code KENO for stationary reactor core calculations. The propagation of uncertainties through depletion calculations with TRITON/NEWT is in preparation and has already been tested on a fuel assembly depletion calculation.
All these transport codes are part of the SCALE-6 system. In certain cases, also the 3D Monte Carlo code MCNP in multi-group mode has been applied.