A comparative study of boundary conditions for the density-based solvers in the framework of OpenFOAM. (30th August 2018)

• Record Type: Journal Article
• Authors: Dong, Yidao; Deng, Xiaogang; Gao, Xiang; Xiong, Min; Wang, Guangxue
• Is Part Of: Computers & fluids. Volume 172(2018)
• Page Start: 284
• Page End: 300
• Publication Date: 2018-08-30
• Subjects: Boundary condition -- Riemann method -- Ghost cell method -- Finite wave model -- OpenFOAM; Fluid dynamics -- Data processing -- Periodicals; 532.050285
• DOI: 10.1016/j.compfluid.2018.04.031
• Languages: English
• ISSNs: 0045-7930

Abstract

Highlights: Extending the nonreflective boundary condition based on the finite wave model to unstructured grid. Implementing the strong- and weak-imposition approaches for the open source software OpenFOAM. Demonstrating the performance of different boundary condition treatments through systematic comparisons. Abstract: In the framework of OpenFOAM, two methods for the boundary condition treatments are discussed and compared for the density-based solvers, including the strong- and the weak-imposition approach. For the weak-imposition approach, the Riemann method is implemented. For the treatment of subsonic inlet and outlet in the Riemann method, the nonreflective boundary condition based on the finite wave model is adopted and extended to the unstructured grid. With the Riemann method, the boundary condition treatments are transformed into the acquisition of the numerical flux with approximate Riemann solvers at the boundary face, same as that of interior faces. And the information of propagating waves at the boundary is correctly reflected through the usage of approximate Riemann solvers. What is more attractive is that the formal accuracy near the boundary as well as the global accuracy can be preserved without destroying the numerical stability of the simulation. For the strong-imposition approach, the ghost cell method is implemented. In spite of its simplicity, the robustness is not good enough on the unstructured triangular grid. Besides, it has been found that the formal accuracy near the boundary is only first order, thus influencing the solution quality and the global accuracy of the flowfields. Some typical inviscid examples are presented to evaluate the performance of these two methods.