Heat transfer correlations for curved downward-facing surfaces in forced and mixed convection regimes

This study investigates single-phase heat transfer on curved, downward-facing surfaces under forced and mixed convection conditions, focusing on applications in external reactor vessel cooling for severe nuclear accident scenarios. Existing correlations, primarily designed for flat surfaces, inadequately predict heat transfer on curved geometries due to the complex flow separation, transitional effects, and mixed convection interactions unique to these surfaces. A custom-designed experimental flow loop was employed alongside computational simulations to capture these behaviors over a range of Reynolds, Prandtl, and Buoyancy numbers. Key findings reveal that flow separation and recirculation on curved surfaces lead to a reduction of up to 22% in local Nusselt numbers compared to flat surfaces under forced convection, where transitional flow dynamics are critical. In the mixed convection regime, turbulence suppression and laminarization were observed around Bo¯≅1, causing significant impairment in heat transfer with reductions of up to 60% in the Nusselt number ratio (Nu/NuF) as buoyancy forces dominate and disrupt turbulence production. To accurately model these effects, multiple turbulence models were tested, with the v2−f and Speziale, Sarkar and Gatski Reynolds Stress Model (RSM-SSG) models outperforming others with a maximum error around 10%. These models are more successful in capturing boundary layer changes due to turbulence anisotropy and turbulence suppression effects. While the inclusion of buoyancy production and dissipation terms of turbulent kinetic energy seemed insignificant to the overall performance. The resulting correlations for local Nusselt numbers recommend different approaches based on the Buoyancy number. For Bo¯10−3, a pure forced convection correlation applies with an account for local position, while for Bo¯10−3, a combination of both the forced convection and another equation that accounts for buoyancy effects is used. These insights and correlations offer a framework for integrating curved-surface heat transfer models into nuclear safety codes, enhancing the predictive accuracy for cooling scenarios that are critical to reactor pressure vessel integrity.