Reduced RANS approach validation to model passive heat removal due to inter-wrapper flow

Abstract

Most fuel assemblies in a liquid metal fast reactors (LMFRs) are surrounded by wrapper tubes, usually arranged in a hexagonal arrangement. The heat transfer across these wrapper walls between sub-assemblies, due to the inter-wrapper flow (IWF), can play a vital role in reducing peak cladding temperatures during flow blockage or flow reduction scenarios, particularly during the passive heat decay removal. This is specific to each LMFR design. CFD analyses of different flow and power conditions may provide essential understanding of the physics and efficacy of heat removal due to IWF. However, the partially- or fully-blocked fuel assemblies lead to highly transient nature of the flow, which can dramatically increase the computational requirements of the simulations making it infeasible to compute using well resolved CFD approaches. In this context, a reduced-resolution RANS model of the influence of IWF is validated in this paper.The three 7-pin hexagonal bundles of the KALLA facility at KIT in Germany are used to model the IWF experimentally. In the computational model, three adjacent bundles along with the inter-wrapper region between them are modelled, where the pins in each bundle are spaced using wire wraps. Wall-resolved RANS simulations are first performed as reference simulations. As the modelling of natural convection is more sensitive to mesh resolution, only this case is used for the resolution sensitivity study. In order to further reduce the computational effort of the sensitivity study, only one wire-pitch length of the inter-wrapper region with two adjacent bundles is used. The resolution is reduced separately in the bundle, the inter-wrapper region and the solid structures, to better understand the effect of mesh resolution. It is demonstrated that the reduced-resolution approach may be used to predict the temperatures in the inter-wrapper region fairly well. After the reduced-resolution approach is optimised for the natural convection case, the approach is validated against experimental data of the KALLA facility.Experimental cases representing different flow blockages have been reproduced using reduced resolution RANS for the purpose of validation. Both qualitative and quantitative comparison and analyses are presented in terms of temperature field. The difference between the model and the experimental data is attributed for and quantified. It is noted that the temperature fields in the bundle and inter-wrapper region are predicted fairly well by the present computational model. However, the local heat flux to the inter-wrapper region is under-predicted by roughly 15–18%, while the total power transported by the IWF is under-predicted by about 6–11% for the present cases. The present model is observed to over-predict the maximum temperature hot-spot in the bundle by roughly 18–33%. It is demonstrated that this gross over-prediction is due to the absence of modelling of conjugate heat transfer in the rods and wires. If the solid wires and pin claddings are modelled, this over-prediction is reduced to roughly 12%. It is shown that a reduced-resolution scheme substantially reduces the computational costs, with a minor difference in prediction of heat transfer in comparison to a wall-resolved mesh.