Thermal design optimization of passive cooling capability in a dry-storage system by adding wall undulations or semi-circular fins

•A new optimized geometry for the Nuclear dry-storage System is suggested.•Ability of advanced RANS models to predict buoyancy driven flow is carried out.•Verification and Validation examining models’ implementation is discussed.•A new dry-storage geometry returns convective heat rate 3 orders higher.•Validation limited to temperature, hide differences between models’ predictions. Thermal design investigation of a novel liner inner wall surface geometry alteration of a ventilated concrete cask used for spent nuclear fuel storage is carried out through three-dimensional Computational Fluid Dynamics (CFD) simulations. The present study reveals that the new proposal returns values for the convective heat transfer coefficient that are almost three orders of magnitude higher than the base case. Hence, proving that even making such a rather simple alterations to the smooth walls geometry, i.e. adding undulations or semi-circular fins, can result in a significant convective heat transfer enhancement, therefore, further reducing the risks of structural damages caused by the dry-storage system overheating. Before conducting the design optimization study mentioned above, two validation test cases have been considered in order to examine the Reynolds Averaged Navier-Stokes (RANS) models’ capability to predict the flow under buoyancy driven and low turbulence levels such is the case in the passively cooled dry cask system. The test cases considered in the validation study sections are the experimental measurements of Betts and Bokhari for turbulent natural convection in an enclosed tall cavity and the temperature measurements in concrete-shielded spent fuel storage cask system reported by McKinnon. The models selected for the validation study are namely the k-ω SST model, the v2-f model, the low-Reynolds k-ε model, and the recently formulated elliptic blending k-ε model. It has been found that out of these models, the low-Reynolds k-ε model was found to provide a good combination in returning both; physically realistic and conservative solution. The present validation using well-resolved RANS simulations has shown that a more quantitative and detailed comparison of the predictions is required in order to clearly highlight the discrepancies between the models’ predictions in terms of velocity, turbulent kinetic energy and, in particular, convective heat transfer coefficient profiles. The commonly adopted methodology in limiting the validation process to temperature pro