Reliability analysis and parametric studies of the buffeting performance of long-span bridges under multiple sets of random variables

Buffeting-induced accelerations and displacements of bridge deck girders commonly drive the bridge design’s comfort, operational, and strength limit states. The scattered nature of the main wind characteristics and bridge responses recorded in multiple monitoring campaigns make deterministic approaches insufficient to assess the bridge’s performance along its life span. This study reports comprehensive sensitivity and reliability studies conducted to unveil the influence of multiple parameters controlling long-span bridges’ buffeting responses. The impact of several sets of random variables on the reliability of the Great Belt Bridge is systematically studied. A detailed treatment of the uncertainty of flutter derivatives consisting of combining their frequency-dependent random definition with their experimentally defined correlation is proposed. Results show the drastic impact of uncertainty in the flutter derivatives, the vertical turbulence intensity, the mean wind velocity, and the definition of the buffeting loads, particularly the slopes of the force coefficients and the aerodynamic admittance, on the buffeting-induced accelerations. The influence of aerodynamic admittance on the results is analyzed in the context of random definitions of mean velocity, turbulent intensities, length scales, structural damping, and aerodynamic characteristics. The computational efficiency of gradient-based reliability methods is discussed, showing its potential to address high-dimensional problems within design frameworks. • The bridge buffeting response was analyzed considering multiple random variables. • Reliability indices of buffeting in high-dimensional problems were identified. • Several parametric studies support the interpretation of the reliability results. • The most relevant random variables and their influence were discussed.