Nonlinear buffeting response of suspension bridges considering time-variant self-excited forces

In the last thirty years, the aerodynamic nonlinearities related to the slow variation of the angle of attack produced by large-scale atmospheric turbulence and their impact on the buffeting response of long-span suspension bridges have been a hot topic in wind engineering research. Self-excited forces accounting for such an effect of turbulence have been crucial in predicting the dynamic response of bridge sectional models and long-span suspension bridges subjected to multi-harmonic gusts and the turbulent wind, respectively. Despite several nonlinear aerodynamic models produced by the scientific community throughout the last years, only few studies on full suspension bridges nonlinear buffeting response in realistic turbulent flows are available. This doctoral work addresses this topic, aiming to enlarge the understanding of the effects of turbulence on the suspension bridge buffeting response. The first contribution of this work concerns nonlinear aerodynamic load modelling. Indeed, large-scale atmospheric turbulence produces large-amplitude low-frequency fluctuations of the angle of attack that can significantly change the self-excited and the external buffeting forces acting on a bridge deck. Assuming that the angle of attack varies slowly compared to the bridge motion, a time-variant linear model relying on Roger’s rational function approximation (RFA) of the force transfer function is proposed for modelling self-excited forces. In particular, an existing model is improved by a flexible fitting of the RFA directly in the multivariate space of reduced velocity and angle of attack. Another contribution of the present work is setting up an experimental procedure based on bi- or multi-harmonic forced-vibration tests to underscore the variation of magnitude and phase of self-excited forces under a time-variant angle of attack. These wind tunnel tests also allowed a sound experimental validation of the proposed model, considering two quite different bridge deck cross-sections as case studies. Aerodynamic derivatives for various angles of attack were measured to determine the model parameters. Despite its simplicity, the model yields accurate results up to relatively fast variations of the angle of attack, and it can reproduce the complicated behaviour of the self-excited forces revealed by the experiments. The model performance strongly depends on the goodness of the RFA-based fitting of aerodynamic derivatives, and the excellent results obtained were possible thanks to the high flexibility of the proposed method. Then, the nonlinear external buffeting forces are formulated to achieve a reasonable compromise between the conflicting needs of modelling both nonlinear and unsteady effects of wind velocity fluctuations. Then, the proposed 2D RFA model for self-excited forces and the nonlinear buffeting forces are incorporated into a stochastic time-variant state-space framework to assess the nonlinear buffeting response of a suspension bridge. The most important feature of this model is the modulation of the self-excited forces due to the spatio-temporal fluctuation of the angle of attack produced by low-frequency turbulence. Such an angle of attack accounts for the spatial wind correlation along the bridge girder. The model is applied to the Hardanger Bridge in Norway, considering different wind conditions. Indeed, the aerodynamic derivatives of this bridge deck cross-section present a strong dependence on the mean angle of attack. Moreover, a novel approach is suggested, diversifying the cut-off frequencies for the considered input motion components in the self-excited forces. In the wake of this, the thesis also investigates the sensitivity of the response statistics to the model cut-offs used to separate the low-frequency and the high-frequency turbulence band. The results emphasise the significant impact on the buffeting response and flutter stability of considering time-variant self-excited forces, though in specific cases the classical linear time-invariant approach is found to provide accurate predictions of the bridge vibrations.