19th Asia-Pacific Vibration Conference (APVC2021)

In the wind resistance analysis of flexible structures such as long-span suspension bridges, cable-stayed bridges and transmission towers, except for the mean wind and longitudinal component, the effects of lateral and vertical components of fluctuating wind on structures can also not be ignored. To model the stochastic turbulence wind field in detail, Tubino and Solari expressed the turbulent wind field with n space points as a 1D-3nV weak stationary stochastic process with zero mean, and proposed the conventional double proper orthogonal decomposition (DPOD) which essentially belongs to Monte Carlo simulation. However, when the number of space points is too large, it has to confront huge computational expense and numerical instability to conduct the POD and even fail to work. To this end, stochastic wind turbulence field is described as a continuous model, i.e., two-dimensional and three variables (2D-3V) stochastic field in the present study. The concept of wavenumber spectral density (WSD) matrix of three-dimensional turbulent wind field is proposed, and the theoretical basis is a hybrid model of wavenumber spectral representation (WSR) and spectral decomposition (including Cholesky decomposition and POD), which makes it only necessary to decompose the three-dimensional matrix and simplifies the derivation process and intermediate variables. Meanwhile, the dimension reduction method is utilized to realize the purpose of the finely simulating three-dimensional turbulence field with only three basic random variables, which avoids the cumbersome sampling high-dimensional random variables in Monte Carlo simulation. Finally, a numerical example of wind turbulence field on a bridge desk is implemented to verify validity of the proposed hybrid model.