A Non-Intrusive Acceleration Technique for Compressible Flow Solvers Based on Dynamic Mode Decomposition
Artikel i vetenskaplig tidskrift, 2016
There are several well-established methods to achieve solver speed-up for finite-volume solvers for compressible flows. These methods are more or less difficult to implement and more or less suitable for implementation in a parallel unstructured type of solver. This article presents a non-intrusive solver acceleration technique applicable for steady-state problems. The proposed technique is based on Dynamic Mode Decomposition (DMD) and does not depend on data format or mesh structure and is thus as straightforward to implement in an unstructured parallel code as in a structured sequential one. The main idea behind the method is that it is possible to use the information available in global flow-field modes to find a correction that will bring the solution closer to a steady-state condition. The DMD-based acceleration technique has been implemented in a massively parallel block-structured finite-volume Navier-Stokes solver for compressible flows and tested on a turbine cascade case with promising results.
Dynamic Mode Decomposition
Computational Fluid Dynamics