Multifidelity Topology Design for Turbulent Heat Transfer Problems Using a Darcy Flow Model Incorporating Pseudo-Boundary Layer

Abstract Topology optimization for turbulent heat transfer problems is challenging due to the multimodality of the solution space caused by the strong nonlinearity of the physical model. In this study, we utilize the framework of multifidelity topology design (MFTD), which is effective for such computationally hard problems. The framework aims to yield promising design solutions by dividing the original problem into low-fidelity (LF) and high-fidelity (HF). In the first step, we aim to generate exhaustive solutions in the solution space by executing TO using a LF model under a wide variety of optimization conditions. The best design is selected from these solutions by evaluation using a HF model. The performance of the final design depends on whether the LF model can generate promising solutions. Therefore, the construction of an appropriate LF model is critical for the MFTD. In this study, we propose a Darcy flow model incorporating pseudo-boundary layer. This model achieves higher accuracy than the conventional Darcy flow model by considering viscous effects. We use the proposed flow model as a LF model of the MFTD and confirm the effectiveness by showing that promising solutions to turbulent heat transfer problems exist among the various solutions generated by the proposed flow model.