Analysis of simplified BWR full core with serpent-2/simulate-3 hybrid stochastic/deterministic code

Highlights: •SERPENT-2 is evaluated as a tool to generate nuclear data libraries. •BWR model is used in comparison between CASMO-5 and SERPENT-2. •The effect of ADFs and diffusion coefficients definition is dominant. -- Abstract: This paper presents an evaluation of a Monte Carlo (MC) neutron transport code as a tool to generate nuclear cross section data for nuclear reactor core simulators. The goal is not to replace deterministic lattice calculations with MC simulations but to provide an audit tool. A boiling water reactor (BWR) assembly is modeled with a deterministic lattice physics code CASMO-5 and with a MC code SERPENT-2 for a simplified set of histories and branch cases. The work reported in this paper is a continuation of authors' previous analysis of a BWR in the reference 13 and SERPENT-2 is used instead of SERPENT-1 to overcome excessive memory usage and execution time of the latter. CASMO-5 and SERPENT-2 are compared with those of a hybrid CASMO-5 + SERPENT-2 scheme which uses the depletion solver of CASMO-5 and the transport solver of SERPENT-2. Their results are compared for an assembly model, in terms of k-inf, macroscopic cross sections, assembly discontinuity factors and diffusion coefficients. The differences are less than 1% except for diffusion coefficients of 5% difference. Subsequently, a nuclear cross section data library is generated for SIMULATE-3 core simulator. A full core depletion calculation is performed and k-eff, axial and radial power distributions are compared. The discrepancies of k-eff between CASMO-5/SIMULATE-3 and SERPENT-2/SIMULATE-3 are smaller than 250 pcm, and the differences of axial relative power fraction (RPF) root mean square errors are smaller than 8%. The differences of radial RPF root mean square errors are smaller than 1.5%.