【title】：Chimera-based parallelization of an implicit Navier-Stokes solver with applicationsChimera-based parallelization of an implicit Navier-Stokes solver with applications
【作者】：Morgan, P. E.^Visbal, M. R. (USAF, Research Lab., Wright-Patterson AFB, OH)^Tomko, K. (Cincinnati, Univ., OH)
【Corportate Author】：【Catalog】：7502 Aerodynamics (1975-)|【Keywords】：THREE DIMENSIONAL FLOW; IMPLICIT METHODS; NAVIER-STOKES EQUATION; COMPUTATIONAL GRIDS; PARALLEL PROCESSING (COMPUTERS); MISSILE SYSTEMS; VECTOR PROCESSING (COMPUTERS); SUPERCOMPUTERS; FLOW DISTRIBUTION; HIGH ANGLE OF ATTACK【Publish Year】：
【Language】：English【Record id】：2001-1088【File Size】：1.57M
【Come From】：AIAA, Aerospace Sciences Meeting and Exhibit, 39th, Reno, NV, Jan. 8-11, 2001, 14 p..【AAIA ID】：A0116872
【Abstract】：The parallelization of a Chimera version of a 3D, time-accurate, Navier-Stokes solver is described. ^The baseline vector code is parallelized by distributing the overset grids to separate processors and incorporating MPI communication with the existing inter-grid communication. ^This approach provides a simple parallel implementation with low communication overhead. ^The code's performance has been assessed on two supercomputers, the IBM SP3, and the Silicon Graphics Origin 2000. ^Parallel efficiencies above 90 percent are achieved on both platforms using up to 64 processors. ^Performance of the current parallel scheme is also compared with a previous parallel implementation based on 2D multi-partitioning. ^The solver is validated for Couette flow formation, and for both steady and periodic flows past a circular cylinder. ^The capability of the new code for large-scale simulations is demonstrated through application to ogive-cylinder and finned-missile flow fields at high angle of attack. ^(Author)
|
外文论文导航
