Topology optimization of natural convection heat transfer using SEMDOT algorithm based on the reduced-order model

Due to the strong nonlinear nature and the mutual effect between the temperature field and the velocity field, it is very difficult to solve the natural convection heat transfer problem. To this end, the Gauss Seidel iterative algorithm based on the reduced-order model is proposed to decouple temperature and pressure. In this paper, the governing equations are discretized by finite element method using bilinear shape functions. To suppress the oscillation of the solution of the governing equations, Streamline Upwind Petrov-Galerkin method (SUPG) is adopted. The main idea of the Gauss Seidel iterative algorithm is: given an initial temperature, pressure and temperature are obtained successively by solving only two nonlinear equations, which greatly improves the computational efficiency. Compared with full order Navier-Stokes model, the Gauss Seidel iterative algorithm for reduced order model has similar results. Then, in order to generate accurate boundaries, Smooth-Edged Material Distribution for Optimizing Topology (SEMDOT) based on density method (SIMP) is studied. Finally, some numerical examples demonstrate the feasibility and effectiveness of the proposed scheme.