Assessment of RANS turbulence model performance in tight lattice LWR fuel subchannels

Reynolds-averaged Navier–Stokes simulations have been performed of the turbulent flow (Re=48000) through a tight lattice (P/D=1.194) subchannel typical of those found in nuclear thermal–hydraulic applications. A wide range of turbulence models (linear eddy-viscosity, non-linear eddy-viscosity and stress-transport) and the use of different near-wall treatments, including low-Re models, standard wall-functions and the more advanced analytical wall-function, have been considered. Results indicate that only those models capable of reproducing some degree of stress anisotropy are able to correctly predict the presence of secondary motion and only the low-Re near-wall treatment was able to reproduce the correct qualitative behaviour of the wall shear stress along the rod surface, in agreement with experimental results by Hooper (1980). They also demonstrate both the failure of the standard wall-function approach in reproducing flows with significant near-wall convection and pressure-gradient effects, and the limitations of using linear eddy-viscosity approaches in flows with even moderate anisotropic effects. Although models utilizing a low-Re near-wall approach demonstrated superior overall performance, the analytical wall-function tested offers improvements over the standard ‘log-law’ based formulation.