In order to increase an efficiency of thermo-fluids systems, enhancing heat transfer performance of heat exchangers in turbulent thermal convections is on huge demand. Shape optimization of heat transfer surfaces for convective heat transfer has been achieved with an adjoint-based shape optimization theory. Since linearization on the momentum equation is not applicable for turbulent flows due to strong nonlinearity, one main approach uses a steady RANS-based adjoint analysis with turbulent models. Existing turbulence models, however, may not always provide good estimates especially when flow separation and reattachment are present, so that the resultant optimal shape would not guarantee the performance enhancement. In this study, we propose a new adjoint-based approach which uses the eddy viscosity and the eddy diffusivity obtained from scale-resolved unsteady simulations such as DNS or LES.
View full abstract