Computational fluid dynamics (CFD) round robin benchmark for a pressurized water reactor (PWR) rod bundle

Abstract

This study examined the capabilities and limitations of steady Reynolds-Averaged Navier–Stokes (RANS) approach for pressurized water reactor (PWR) rod bundle problems, based on the round robin benchmark of computational fluid dynamics (CFD) codes against the NESTOR experiment for a 5×5 rod bundle with typical split-type mixing vane grids (MVGs). The round robin exercise against the high-fidelity, broad-range (covering multi-spans and entire lateral domain) NESTOR experimental data for both the flow field and the rod temperatures enabled us to obtain important insights into CFD prediction and validation for the split-type MVG PWR rod bundle problem. It was found that the steady RANS turbulence models with wall function could reasonably predict two key variables for a rod bundle problem – grid span pressure loss and the rod surface temperature – once mesh (type, resolution, and configuration) was suitable and conjugate heat transfer was properly considered. However, they over-predicted the magnitude of the circumferential variation of the rod surface temperature and could not capture its peak azimuthal locations for a central rod in the wake of the MVG. These discrepancies in the rod surface temperature were probably because the steady RANS approach could not capture unsteady, large-scale cross-flow fluctuations and qualitative cross-flow pattern change due to the laterally confined test section. Based on this benchmarking study, lessons and recommendations about experimental methods as well as CFD methods were also provided for the future research.