Abstract
A combined data-assimilation and linear mean-flow analysis approach is developed to estimate coherent flow fluctuations from limited mean-flow measurements. It also involves Reynolds-Averaged Navier–Stokes (RANS) modelling to efficiently tackle turbulent flows. Considering time-averaged Particle Velocimetry Image (PIV) measurements of the near-stall flow past a NACA0012 airfoil at an angle of attack of \(10^{\circ }\) and in the chord-based Reynolds number range \(4.3 \cdot 10^4 \le Re \le 6.4 \cdot 10^4\) , data assimilation is first employed to correct RANS equations that are closed by the Spalart-Allmaras model. The outputs of this procedure are a full mean-flow description that matches the PIV data and a consistent turbulence model that provides not only a mean eddy-viscosity field but also the perturbations of the latter with respect to mean-flow modifications. Global stability and resolvent analyses are then performed based on the so-obtained mean flow and model to satisfactorily predict near-stall low-frequency phenomena, as confirmed through comparison with the Spectral Proper Orthogonal Decomposition (SPOD) of the PIV measurements. This comparison highlights the benefits in taking into account variations in the turbulent eddy-viscosity over a frozen approach for the correct estimation of the present coherent low-frequency oscillations.
Graphical abstract
