Improved delayed detached eddy simulation (IDDES) results of a backward facing step flow are analyzed using dynamic mode decomposition (DMD). Different flow variables and the time-resolved skin friction coefficient are investigated and compared to a spectral analysis of the wall pressure fluctuations. Although the flow field does not contain single dominant modes, two distinct flow features can be extracted and visualized using the DMD mode shapes. A low frequency flapping motion of the shear layer is found in the mode decomposition of the pressure, the wall-normal velocity and the skin friction coefficient. At higher frequencies, a wake mode similar to a von Karman vortex street is identified in the streamwise velocity, the pressure and the vorticity field. This work uses a modified version of the original dynamic mode decomposition algorithm that enforces a sparse solution with a user-defined number of modes. It is shown that the algorithm extracts the most important flow features reliably across different flow variables and that sparse DMD can be applied in situations where no single dominant mode is present.
展开▼
