Identification of Reduced-Order Models for Flutter Analysis Using CFD-Based Aerodynamic Data

This work presents a comparative evaluation of reduced-order modeling approaches for transonic aeroelastic stability analysis. Unsteady aerodynamic data are obtained by solving the Euler equations and used to construct reduced-order models with Rational-Function Approximations (RFA), Eigensystem Realization Algorithm (ERA), and Dynamic Mode Decomposition (DMD). These models are coupled with a common structural model to form an aeroelastic system, and flutter onset is identified through eigenvalue analysis. Interpolation is applied when necessary to refine the resolution of the instability point. Computational effort is evaluated by decomposing the total time into CFD-based data generation, ROM construction, and flutter analysis, illustrating how resources are distributed across the workflow. Results indicate that all ROMs can predict flutter onset speed within approximately 6% of the CFD-based reference, while significantly reducing computational cost. The study also highlights that each ROM presents a distinct trade-off between accuracy and efficiency, offering practical strategies for rapid and reliable aeroelastic stability assessment in transonic flows.