关键词: 非结构网格, 并行计算, 区域分裂, Navier-Stokes方程, 有限体积方法

Abstract: A parallel computing strategy for a 3D hybrid unstructured Navier-Stokes solver has been developed. The Navier-Stokes equations are solved by a face based finite volume method which is suitable for arbitrary mesh type. The Spalart-Allmaras one-equation turbulence model is implemented to evaluate the turbulent viscosity. Parallel computation is based on the domain decomposition method and load balance is achieved by using METIS system. The Message Passing Interface (MPI) library routines are utilized to pass information between processors and non-blocking communications are used to minimize the communications overhead. The dynamic allocation feature of FORTRAN 90 is fully exploited to reduce the memory requirement. Finally the parallel code is applied to simulation of viscous flow around the DLR-F6 geometry(wing-body-pylon-nacelle) to validate its accuracy, high parallel performance and the ability to deal with complex geometries.

Key words: unstructured grid, parallel computation, domain decomposition, Navier-Stokes equation, finite volume method