Reconstruction and Simulation of Non-Synoptic Wind Fields for Nonlinear Aeroelastic Analysis of Long-Span Bridges

Non-synoptic extreme winds such as thunderstorms and downbursts exhibit pronounced non-stationarity, with rapid variations in wind speed, turbulence characteristics, and angle of attack that challenge conventional stationary wind-field models and linearized aeroelastic formulations. This study proposes a measurement-driven numerical framework to reconstruct and simulate fully non-stationary wind fields from sparse full-scale monitoring data on long-span bridges, providing physically consistent inputs for nonlinear time-domain aeroelastic analyses. Monitored records are processed to extract time-varying wind parameters, which are then spatially reconstructed along the deck and used to generate non-stationary wind fields through an EPSD-based synthesis. The resulting wind inputs are employed in nonlinear time-domain aeroelastic simulations and compared against a standard stationary frequency-domain benchmark based on an equivalent synoptic wind representation. The comparison highlights the limitations of stationary assumptions in capturing transient and spatially localized response features induced by non-synoptic events. To assess the generality of the observed mechanisms, the analysis is extended to multiple wind records with different degrees of non-synoptic character and to multiple suspension bridges. The results show that the response features are primarily driven by the non-stationary nature of the wind, while structural properties mainly influence their magnitude and spatial distribution.