Stochastic simulation of wind-related processes with intermittency

In structural engineering, turbulent flows are critical as most structures are exposed to winds in atmospheric boundary layer with high Reynolds number. Due to the prohibitive cost of solving Navier-Stokes equations, statistical methods are popular in analyzing structural response subject to wind related excitations. Conventional strategies of simulating wind processes are usually restricted to utilizing the second-order statistics with implied Gaussian assumption, which neglects some of the features of turbulent flows. For a typical turbulent flow, the process consists of a wide range of eddies at different scales. The nonlinear interaction transfers the velocity fluctuations from larger eddy scales down to smaller scales through a cascade process which is referred to as a self-similar property. In order to improve the fidelity of turbulence simulation, a novel scheme based on a Log-Poisson model is applied to the simulation in this study. This scheme extends the second-order statistics to the high-order statistics of the velocity increments, implemented with the wavelet expansion. The relationship between the wavelet coefficients and the velocity increments is identified in this paper. Further detailed information, such as the multifractal structure and the intermittency, can be appropriately presented utilizing wavelet coefficients. The wind effect on the structure is re-examined using the turbulent wind with intermittency. It is noted that the turbulent wind results in a higher peak factor than the Gaussian wind process. Hence, ignoring the effects of intermittency in turbulence may face underestimation of loads and risk.