Reduced-Order Modeling of Transonic Buffet Aerodynamics

In the present work, a reduced-order modeling framework based on nonlinear system identification is extended and applied concerning the prediction of transonic buffet aerodynamics. For this purpose, the external dynamic filtering approach combined with both a recurrent neuro-fuzzy model and a multilayer perceptron neural network is employed. In order to calibrate the model, training data are provided by means of a forced-motion unsteady Reynolds-averaged Navier-Stokes simulation. The intention of the developed model is the efficient computation of time-varying integral quantities such as aerodynamic force and moment coefficient trends in contrast to the resolution of detailed flow effects. From an identification-based point of view, the challenge lies in the reproduction of the self-sustained unsteadiness of the buffeting flow that is present even if no external forcing or excitation is active. Finally, the performance of the reduced-order model is demonstrated for predicting the air loads with respect to a case including predominant buffet phenomena. In this regard, the methodology is tested by considering the NACA 0012 airfoil at transonic freestream conditions undergoing a forced pitching motion beyond the buffet-critical angle of attack. A comparison with the full-order reference solution shows that the essential characteristics of the nonlinear aerodynamic system are captured by the proposed model.