Model-based adaptive reduced basis methods for unsteady aerodynamic studies

The present work investigates the problem of performing accurate reconstructions of vortex-dominated unsteady flows by means of RBM (Reduced Basis Methods). When faced with the necessity of reconstructing tile flow field over a specified time window, the proposed method aims at automatically and adaptively selecting the most accurate reduction technique among the classical snapshot POD and more recent techniques such as SPOD, DMD and Recursive DMD. The method is paired with an equation-based approach which guides the adaptive selection through a residual evaluation of the different RBM solutions, using the initial set or governing equations in a discrete sense (Finite Volume Discretization). Problems of high relevance to the aerodynamics field are considered as test-cases to evaluate tile performances of the Model-Based Adaptive RBM with respect to pure POD, such as the impulsive start of 20 and 30 lifting bodies.