Locally-Adaptive Tabulation of Low-Dimensional Manifolds using Bezier Patch Reconstruction

Abstract

An efficient tabulation strategy for turbulent combustion simulations is proposed. Using a locally adaptive arrangement of structured Bezier patches–often used in computer graphics–the combustion manifold can be efficiently tabulated; thus reducing the table size by over an order of magnitude while maintaining high accuracy. A hybrid search algorithm is implemented that uses a one-step quadrant analysis followed by a linear search to minimize the cost of the data retrieval. The present tabulation has been successfully applied to flamelet/progress variable approach (FPVA) and trajectory generated low-dimensional manifold (TGLDM). The error, computational size and search-and-retrieval time are quantified and compared against classical tabulation approaches.