Applications of SCALE 5 TSUNAMI for Critical Experiment Benchmark Design, Interpretation, and Estimation of Biases and Uncertainties for Space Power Reactor Designs and Safety Analyses

摘要

Sensitivity and uncertainty (S/U) analysis computational software was released by Oak Ridge National Laboratory (ORNL) in June 2004 within the SCALE 5 software sequences named TSUNAMI-1D and TSUNAMI-3D (Tools for Sensitivity and UNcertainty Analysis Methods Implementation). The software is based upon first-order linear perturbation theory. An additional software sequence named Javape?o permits the ready interpretation and comparison of sensitivities and uncertainties in a graphical format. These tools were applied to the design of critical experiment benchmarks that have nearly identical nuclide neutron-energy-wise reaction sensitivities compared with specific space power reactor (SPR) designs and related safety analyses. Sensitivities for all of the nuclide neutron-energy-wise reactions available from the SCALE 5 ENDF/B-V 238-energy group library were simultaneously determined by TSUNAMI-3D, a three-dimensional KENO-V–based Monte Carlo sensitivity analysis computational sequence. Comparisons and interpretations of the sensitivity results and uncertainties due to Monte Carlo calculations and cross-section data permitted the estimation of biases and uncertainties for materials of construction for two SPR design concepts and their safety analyses. The two SPR concepts analyzed with TSUNAMI-3D were low-power (500-kW) and high-power (2-MW) liquid-lithium- metal-cooled, rhenium and niobium–1% zirconium-alloy-clad (Nb-1Zr) highly enriched uranium-nitride-fueled reactors that were reflected with beryllium or beryllium oxide. Design of critical experiment benchmarks, referenced as generic physics experiments (GPEs), was undertaken utilizing TSUNAMI-3D. The designs were developed from available fissile materials at the Los Alamos Critical Experiment Facility (LACEF). Due to the immediate unavailability of rhenium and the limited availability of LACEF for experimentation, only one designed GPE was performed - an experiment to characterize Nb-1Zr in a particular SPR safety evaluation design that included partial water flooding. Preliminary analyses of that critical experiment, relative to the 2-MW SPR design considered, were performed to estimate the anticipated bias and associated uncertainties (i.e., experimental and computational) that may be attributed to Nb-1Zr. This paper describes these applications of the ORNL TSUNAMI software to SPR design evaluations, and benchmark design and interpretation, as well as the estimation of biases and uncertainties related to SPR concepts.