Nuclear data verification based on Monte Carlo simulations of the LLNL pulsed-sphere benchmark experiments (1979 & 1986) using the Mercury code

摘要

Livermore's nuclear data group developed a new verification and validation test suite to ensure the quality of data used in application codes. This is based on models of LLNL's pulsed sphere fusion shielding benchmark experiments. Simulations were done with Mercury, a 3D particle transport Monte Carlo code using continuous-energy cross-section libraries. Results were compared to measurements of neutron leakage spectra generated by 14MeV neutrons in 17 target assemblies (for a blank target assembly, H{sub 2}O, Teflon, C, N{sub 2}, Al, Si, Ti, Fe, Cu, Ta, W, Au, Pb, {sup 232}Th, {sup 235}U, {sup 238}U, and {sup 239}Pu). We also tested the fidelity of simulations for photon production associated with neutron interactions in the different materials. Gamma-ray leakage energy per neutron was obtained from a simple 1D spherical geometry assembly and compared to three codes (TART, COG, MCNP5) and several versions of the Evaluated Nuclear Data File (ENDF) and Evaluated Nuclear Data Libraries (ENDL) cross-section libraries. These tests uncovered a number of errors in photon production cross-sections, and were instrumental to the V&V of different cross-section libraries. Development of the pulsed sphere tests also uncovered the need for new Mercury capabilities. To enable simulations of neutron time-of-flight experiments the nuclear data group implemented an improved treatment of biased angular scattering in MCAPM.