Evaluation of the OSCAR-5 nodal diffusion approach to the MIT BEAVRS PWR experimental benchmark

Abstract

BEAVRS is a benchmark of a Pressurized Water Reactor (PWR) located in the USA. The experi-mental benchmark was established to test and validate high-fidelity transport codes. In this work the benckmark is used as a verification tool for the Overall System for the CAlculation of Reactors (OSCAR). OSCAR is a code suite developed at Necsa; this latest version of the code (OSCAR-5) envisions a unified modelling scheme for PWR type reactors to be deployed to various codes for multiple applications. A unified, heterogeneous model of the BEAVRS PWR was constructed in OSCAR-5, which was then homogenized; the homogenized nodes were used to construct a 3D nodal model. The heterogeneous model was deployed to the Serpent Monte Carlo code to conduct control bank worth calculations for validation purposes. It was discovered that the greatest differ-ence between the BEAVRS experimental control bank worth and Serpent is less than 100 pcm. As a result, the unified model built in OSCAR-5 is deemed to be sufficient. In the construction of the homogeneous nodal diffusion model, infinite lattice calculations were performed on fuel compo-nents, using the Serpent lattice code as well as the collision probability transport code, HEADE. In particular, the infinite lattice calculations were done to verify the validity of using HEADE for deterministic calculations as opposed to Serpent. Both these codes produce nodal cross-sections comparable to published results using infinite lattice approximations. The 3D nodal model was used to determine the power distribution in the reactor core. The 3D nodal model yielded an aver-age relative power error of 7.97 % as compared to the 3D heterogeneous model, with the greatest error of 18.44 % occurring in one of the assemblies at the boundary of the reactor core. These errors show that the infinite lattice approximation is not sufficient for analysis of the BEAVRS core and various potential improvements are discussed. Overall, OSCAR-5 is quite capable of constructing a unified model of BEAVRS and the unified modelling approach can be applied to PWR geometries, as is evident from comparison to published results.