CFD analysis of low Prandtl number fluid with internal heat generation flowing in a rod bundle at low reynolds numbers

The aim of this study is to evaluate the performance of commonly used RANS turbulence models in the context of heat generating liquids of moderately low Prandtl numbers of 0.1 and 0.2. This situation is of particular interest in the context of novel nuclear reactors using liquid metal fuels, such as the prototypical Dual Fluid Reactor described for example in ( Arnold et al., 2025 ; Sierchuła et al., 2019 ; Weissbach et al., 2020 ). This reactor uses a liquid metal coolant (Lead) and a liquid metal fuel (Uranium-Chromium eutectic) in two separate circuits. This concept can be considered as a mixture of a molten salt reactor and a lead cooled fast reactor. The liquid metal fuel and -coolant allow a fast neutron spectrum as well as high operating temperatures. The liquid fuel enables continuous online reprocessing and also offers the possibility to implement passive safety systems such as subcritical drain tanks. The current study investigates fundamental thermal-hydraulics of a triangular-lattice rod bundle arrangement with a pitch-to-diameter ratio of 1.4. The focus lies on the low Reynolds number regime of flows ( Re ≤ 5000), which is of particular interest in case of natural circulation. Since the streamwise- as well as the crossflow direction are widely used and in their nature quite different, both situations are analysed. The study is performed using the Nek5000 DNS code as well as the Ansys Fluent code, which was used for RANS simulations. The investigated Prandtl numbers of 0.1 and 0.2 were selected since they are the highest and lowest Prandtl numbers expected to occur in the liquid fuel of the dual fluid reactor studied in ( Arnold et al., 2025 ). The DNS results are compared with two RANS models, namely the 7-equation Reynolds stress model (RSM BSL) and the 2-equation k-omega-SST model. • DNS and RANS simulations of a rod bundle were performed at low- to intermediate Reynolds numbers for streamwise and crossflow conditions. • Prandtl numbers of 0.1 and 0.2 were used to investigate the effects of heat transfer in this sparsely studied range of Prandtl numbers. • k-omega-SST and RSM-BSL models were found to be in reasonable or even good agreement with the DNS data, however, further simulations are found to be necessary.