Detailed knowledge of a coolant flow in a fuel assembly of a reactor core has always been a major factor in the design of new nuclear systems. In this regard, traditionally adopted subchannel analysis codes cannot take into account local phenomena, which are quite essential. On the other hand, Computational Fluid Dynamic (CFD) is being recognized as a valuable research tool for thermal-hydraulics phenomenon in the fuel assembly geometries. Because of the high Reynolds number and geometric complexities, the practical CFD calculations are mostly limited to pragmatic Reynolds Averaged Navier-Stokes (RANS) type modelling approaches. A good prediction of the flow and heat transport inside the fuel rod bundle is a challenge for such RANS turbulence models and these models need to be validated. Although the measurement techniques are constantly getting improved, however, the CFD-grade experiments of flow mixing and heat transfer in the subchannel scale are often impossible or quite costly to be performed. In addition, lack of experimental databases makes it impossible to validate and/or calibrate the available RANS turbulence models for certain flow situations. In that context, Direct Numerical Simulation (DNS) can serve as a reference for model development and validation.
The aim of this work is to design a numerical experiment in order to generate a high quality DNS database for a tight lattice bare rod bundle, which will serve as a reference for the validation purpose. The considered geometric design is based on the well-known Hooper experiment, which contains a bare rod bundle with pitch-to-diameter ratio of P/D = 1.107. Performing a DNS computation corresponding to the Hooper experiment requires a huge computational power. Hence, a wide range of unsteady RANS (URANS) study has been performed to scale-down the Reynolds number such that it is feasible for a DNS computation and at the same time it still preserves the main flow characteristics. In addition to the flow field, a parametric study for three different passive scalars is performed to take into account the heat transfer analysis. These passive scalars correspond to the Prandtl numbers of air, water and liquid metal fluids. The heat transfer of these three fluids has been studied in combination with two different boundary conditions at the walls, i.e. a constant temperature and a constant heat flux. Finally, the obtained URANS results are used to compute the Kolmogorov and Batchelor length scales in order to estimate the overall meshing requirements for the targeted DNS.
Direct numerical simulation of turbulent flow and heat transfer in a hexagonal rod bundle