The Research On Flutter Derivatives Of Long-span Bridges Based On Time-domain Identification

This paper focuses on identification of flutter derivatives of bridge decks in time domain in smooth flow or turbulence flow. The main studies are presented as follows:1) Identification methods to obtain flutter derivatives, as well as the research history and the up-to-date development are reviewed. The free vibration test method, the forced vibration test method and stochastic test method, as well as the advantage and disadvantage are evaluated systematically.2) Identification of flutter derivatives by using the Eigensystem Realization Algorithm and the Fast ERA with Data Correlation are presented. Flutter derivatives of the idealized thin plate are extracted based on numerical simulation applied to the two methods. It is indicated that under noise free condition, the results obtained from the two methods are in fairly good agreement with the Theodorsen analytical solutions. When high level noise is included in response data, however, the results from ERA show significant deviation from Theodorsen analytical solution, while at high reduced wind speeds, results from FERA/DC still show stable agreement with Theodorsen analytical solution, which suggests that the FERA/DC is robust to noise and is more suitable to identify flutter derivatives of bridge decks.3) Sectional model wind tunnel tests are carried out to identify flutter derivatives of a thin plate and a real bridge deck. The results from ERA and FERA/DC are presented and compared. The effectiveness of FERA/DC and its robustness to noise are confirmed.4) The harmonic superposition method is employed numerically to produce a turbulent wind field. Horizontal and vertical time histories of flucatating wind are generated at several locations with same height. After comparing the spectra of those time histories with the target spectra, the effectiveness of the present method is validated. The precdures as to obtain wind-induced buffeting force are formulated.5) The theory of Covariance-driven Stochastic Subspace Identificaton CSSI is presented. Based on numerical simulation with this method, flutter derivatives of the idealized thin plate are identified. It is found that CSSI is viable and more suitable to identify flutter derivatives of bridge decks under natural wind. Sectional model wind tunnel tests are performed as to identify flutter derivative of a thin plate. It is followed by the comparation of results from smooth flow and from turbulent flow.6) The CSSI method is applied to identify flutter derivatives of tandem bridge decks, in which strong aerodynamic interference is involved. Based on wind tunnel tests, the effect of aerodynamic interference on flutter derivatives of tandem bridge decks is evaluated. For bridge wind-resistant practice, the developed method provides a reasonable way to identify flutter derivatives of bridges regarding strong stochastic aerodynamic interference.