The non-Boussinesq convection in a ventilated square cavity is studied numerically. The flow is driven by a prescribed constant inlet velocity at the top-left wall and an outlet located at the bottom-right wall. The prescribed temperature difference, between the inlet flow and the walls of the cavity, involves buoyancy forces which modify the flow. The density variations are considered by means of a low-Mach-number formulation, which is discretized by a projection scheme based on a second-order finitedifference discretization. The steady solutions associated to forced and mixed convection are presented. A discussion based on the flow and heat transfer features is proposed. Especially, the deviation from the Boussinesq model is highlighted in the case of mixed convection.
This paper presents an analysis of a multigrid defect correction to solve a fourth-order compact scheme discretization of the Poisson's equation. We focus on the formulation, which arises in the velocity/pressure decoupling methods encountered in computational fluid dynamics. Especially, the Poisson's equation results of the divergence/gradient formulation and Neumann boundary conditions are prescribed. The convergence rate of a multigrid defect correction is investigated by means of an eigenvalues analysis of the iteration matrix. The stability and the mesh-independency are demonstrated. An improvement of the convergence rate is suggested by introducing the damped Jacobi and Incomplete Lower Upper smoothers. Based on an eigenvalues analysis, the optimal damping parameter is proposed for each smoother. Numerical experiments confirm the findings of this analysis for periodic domain and uniform meshes which are the working assumptions. Further numerical investigations allow us to extend the results of the eigenvalues analysis to Neumann boundary conditions and non-uniform meshes. The Hodge-Helmholtz decomposition of a vector field is carried out to illustrate the computational efficiency, especially by making comparisons with a second-order discretization of the Poisson's equation solved with a state of art of algebraic multigrid method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.