Calibration Of High fidelity bare rod bundle simulations for various PRANDTL fluids

One of the key factors in the design of liquid metal cooled fast reactors (LMFRs) refers to the detailed knowledge of the coolant flow in the sub-channels of the fuel assemblies in the reactor's core. In fact, a thorough experimental investigation of the flow mixing and heat transport in such sub-channels is often impossible or quite costly to be performed. Due to these limitations, computational fluid dynamics (CFD) has recently become a valuable tool in the study of the dynamics of the flow within the sub-channel regions of typical nuclear fuel assemblies. In general, RANS (Reynolds Averaged Navier-Stokes) approaches are used to analyze the flow and mixing processes within fuel assemblies. In some cases LES (Large eddy simulation) could also be used. Nevertheless, both RANS and LES modeling are associated with numerical errors derived from the averaged/filtered Navier-Stokes equations, which in turn impose significant limitations especially in the modeling of low Prandtl fluids, used as coolant in LMFRs. In addition, the lack of experimental database or additional reference data, e.g. fully detailed high fidelity DNS, often limits the reliability of traditional RANS and LES approaches in modeling flow and heat transfer in liquid metal cooled reactors. In this paper, through a calibration procedure of the mass flow rate of a sub-channel geometry characteristic of LMFRs fuel assemblies, an approach is developed to obtain the required resolution for a computationally affordable high fidelity DNS of sub-channel flow. As a first step, the sub-channel geometry of a well-documented case is adopted; the hydraulic experiments performed by Hooper on a tight lattice bare rod-bundle configuration which uses air as the working fluid. Subsequently, a set of URANS simulations are performed in which the experimental mass flow rate was systematically scaled in order to estimate a feasible resolution to perform a high fidelity DNS of Hooper's sub-channel flow. Finally, on a sub-channel configuration with P/D=1.32, a set of URANS simulations for different working fluids (air, water, and liquid lead) are additionally performed. Despite the large variability of the fluid properties, results show that the overall flow physics typical of sub-channels, e.g. large-scale macroscopic pulsations, is identical in all cases provided the bulk Reynolds number is the same. In general, these results show that the detailed DNS data to be obtained for Hooper's case will provide a valuable resource reference for turbulence modeling validation of liquid metal sub-channel flows in LMFRs fuel bundles.