Unsteady separated flow simulations using a cluster of workstations

Abstract

The possibility of predicting the full three-dimensional unsteady separated flow around complex ship and helicopter geometries is explored using unstructured grids with a parallel flow solver. The flow solver used is a modified version of the Parallel Unstructured Maritime Aerodynamics (PUMA) software, which was written by Dr. Christopher Bruner as part of his doctoral thesis at the Virginia Polytechnic Institute and State University. The efficiency and accuracy of PUMA at resolving several steady state solutions and a fully three-dimensional unsteady separated flow around a sphere were studied in order to determine if it was a suitable platform to base future work on. The COst effective COmputing Array (COCOA), a powerful 50-processor Beowulf cluster, was also built and tested as a part of the effort to make all this possible at a very economic cost. Unstructured grids were utilized in order to maximize the number of cells in the area of interest, while minimizing cells in the far field. A high level of clustering is required to solve viscous unsteady problems, and unstructured grids offer the least expensive method to ensure this. NASA’s VGRID package was used to generate the unstructured grids.