High Performance Parallel Algorithms for Improved Reduced-Order Modeling; Final rept. 1 Jul 2005-30 Nov 2007

摘要

We describe the development of a reliable parallel algorithm and software tools that utilize flow-adapted KLE/POD representations and that are able to take advantage of distributed data formats on cluster/grid computer architectures. The associated module functions efficiently within the context of current best practices of fluid flow simulation. Additionally, we describe methods that lead to greater predictive capability and extend the range of flows for which KLE/POD-based methods are effective. Model reduction methods used for linear input-output systems based on a rational Krylov framework were also studied. We propose new investigations informed both by approximate inertial manifold approaches and by energy-based turbulence modeling/LES approaches capable of accounting for the effect of small scale dynamic structures.