This study discusses a general framework to identify the unsteady features of a flow past an oscillating aerofoil in deep dynamic stall conditions. In particular, the work aims at demonstrating the advantages for the design process of the Spectral Proper Orthogonal Decomposition in accurately producing reliable reduced models of CFD systems and comparing this technique with standard snapshot-based models. Reynolds-Averaged Navier-Stokes system of equations, coupled with k−ω SST turbulence model, is used to produce the dataset, the latter consisting of a two-dimensional NACA 0012 aerofoil in the pitching motion. Modal analysis is performed on both velocity and pressure fields showing that, for vectored values, a proper tuning of the filtering process allows for better results compared to snapshot formulations and extract highly correlated coherent flow structures otherwise undetected. Wider filters, in particular, produce enhanced coherence without affecting the typical frequency response of the coupled modes. Conversely, the pressure field decomposition is drastically affected by the windowing properties. In conclusion, the low-order spectral reconstruction of the pressure field allows for an excellent prediction of aerodynamic loads. Moreover, the analysis shows that snapshot-based models better perform on the CFD values during the pitching cycle, while spectral-based methods better fit the loads’ fluctuations.
Inverse methodology as applied to reconstruct local textile features from measured pressure field
