DNS of transient mixed convection in a vertical differentially heated turbulent channel

Advanced reactor designs strongly emphasize passive safety technology. An important aspect of advanced reactor safety analysis is an understanding of the transient heat transfer phenomena, particularly those involving transitions between forced and natural convection during events such as loss-of-flow (LOF) scenarios. This study presents direct numerical simulation (DNS) of transient flow from forced to mixed convection in an idealized downcomer-like differentially heated channel. The working fluid is modeled with a unity Prandtl number, enabling detailed examination of the coupled momentum and thermal transport processes. Utilizing Nek5000/NekRS, scalable spectral element method (SEM) computational fluid dynamics (CFD) solver, we modeled a sudden decrease in driving pressure – mimicking a LOF event and tracked the resulting evolution of Reynolds and Nusselt numbers, boundary layer and turbulence statistics. Key findings include the systematic thickening and eventual asymmetry of velocity and temperature fields under increasing role of buoyancy; “memory” lag in Reynolds shear stress (RS) and turbulent kinetic energy (TKE) profiles when compared to steady state simulations at a matching Reynolds number; pronounced transient Nusselt number enhancement driven by thermal boundary layer inertia and residual eddy mixing. The dataset provides detailed high-fidelity benchmark information on transient buoyancy-affected turbulence and heat transfer that can support future model development and validation efforts.