Modal Decomposition of Pulse Burst PIV Data

Dynamic Mode Decomposition (DMD) has recently evolved as a useful modal decomposition technique for time-resolved data. In this work, DMD is applied to 2-D velocity datasets obtained from Pulse Burst PIV measurements of flow over an open cavity at free-stream Mach number, M_∞ = 0.8*. Resolved modes are obtained by considering increasing number of datasets. Further, rank reduction is done to include higher energy modes. Spatio-temporal characteristics of these modes have been compared with modes obtained from application of Proper Orthogonal Decomposition (POD) to the same datasets. The modal shapes of wall-normal velocity fluctuations obtained by these methods closely relate to each other at specific Rossiter modes, which are the dominant tonal frequencies from surface pressure and velocity spectra. Reduced order representations of velocity fields are reconstructed at frequencies close to Rossiter modes revealing that flow dynamics can be captured by considering only a few modes at these frequencies. This comparative analysis helps in relating DMD and POD as modal reduction techniques for time-resolved velocity datasets.

*Beresh et al., AIAA Paper 2016-1344