Site-specific data-driven probabilistic wind field modeling for the wind-induced response prediction of cable-supported bridges

In this study, full-scale wind velocity measurements are conducted at eight locations on the Hardanger Bridge girder to investigate the possibility of a probabilistic representation of the turbulence field along the bridge span. Using appropriate assumptions, the two-dimensional turbulence field along the structure is reduced to six turbulence parameters, which are considered as correlated lognormally distributed random variables. The directionality and wind speed dependence of the parameters are demonstrated by means of wind roses and scatter diagrams. Depending on the wind speed and direction, simulations of the turbulence field were carried out using random number generators. The performance of simulated wind fields in capturing the variability and correlation structures of an actual wind field at a site is tested by detailed comparisons with the measurement data. For the sake of illustration, simulations were also performed for the design wind speed of the Hardanger Bridge using the current model and another model from the literature. The resulting probabilistic model is suitable for implementation in reliability-based frameworks and long-term extreme response analysis.