Digital Twin-Enhanced Identification of Flutter Derivatives of a Bridge Deck Section

Flutter derivatives are critical parameters for investigating self-excited force characteristics and predicting the flutter response of a long-span bridge. However, wind tunnel tests and computational fluid dynamics (CFD) simulations of a bridge deck section, which are two commonly used methods for determining flutter derivatives, have their own merits and shortcomings. This paper thus proposes a digital twin-enhanced identification of flutter derivatives of a bridge deck by integrating CFD simulation with forced vibration tests in a wind tunnel. A bridge deck section tested in a wind tunnel serves as a physical model, while the CFD simulation of the deck section is taken as a virtual counterpart. An algorithm is then developed to map the virtual model to the physical model by fusing the data collected from the physical model with the virtual model to form a digital twin. The established digital twin eliminates the shortcomings of both wind tunnel tests and CFD simulations and enhances the identification accuracy of flutter derivatives. By using the digital twin, the attack angle-dependent and amplitude-dependent characteristics of the derivatives are investigated. A global digital twin in terms of individual digital twins and Kriging interpolation is also developed for predicting flutter derivatives and applied to a full aeroelastic bridge model for predicting its flutter response. The results show that the digital twin-enhanced method is more accurate and efficient than wind tunnel tests or CFD simulations alone.