Fast Reactor Calculations with MCNP, PSG/Serpent and ERANOS

Evaluated reactor physics benchmark exercises were calculated with both stochastic Monte Carlo and deterministic codes over the period of the NETNUC project. The Monte Carlo codes MCNP and PSG (early version of Serpent) were used to calculate the benchmark of the Japanese sodium-cooled fast reactor JOYO, after which some features of the ZPR-6/7 (Zero Power Reactor at Argonne, USA) benchmark were calculated with the deterministic ERANOS code system. The JOYO calculations were performed to determine the keff for two separate core configurations, control rod reactivity curves for the two regular rods, sodium void reactivity for seven various fuel assembly locations and the isothermal temperature coefficient. The calculated results were compared between the codes and to the experimental results. The codes mostly yielded mutually pretty consistent results, but in some cases the discrepancy with the experimental observations was significant. Also the limitations of the Monte Carlo method due to its statistical uncertainty became obvious, when small reactivity differences were present. The steady-state fast reactor physics code package ERANOS was introduced to extend VTT's deterministic calculation capabilities to fast neutron spectrum device. It was employed to calculate parts of the ZPR-6/7 benchmark with objectives to determine how it behaves along with JEFF-3.1 and -3.1.1 nuclear data against the experimental values, and how the different functions and parameters of the code package perform against each other. The method and geometry choice proved to be a significant factor in some cases, meanwhile the minor parameters did not generally have any major effect.