Nonstationary Extreme Winds: Field Measurement, Modeling, Simulation And Wind-induced Response Of High-rise Buildings

In the nature, nonstationary extreme winds such as the typhoon or hurricane, tornado,and downburst are great challenges to the wind-resistant design of structures in the civil engineering. With the development of society, some skyscrapers or long-span structures such as Guangzhou Tower, Shanghai Tower, National Stadium, Bridge across the Strait of Joan,and Strait of Messina Bridge have been built or planned. The stiffnesses of these structures are so small that their natural vibration periods can even reach 10-20 s. These features will result in a large transient structural response under nonstationary extreme winds. Although some achievements have been obtained in the research associated with nonstationary extreme winds and the corresponding structural responses to these extreme winds, there are still many problems that needed to be solved. In this paper, the field measurement, modeling and simulation of nonstationary extreme winds and the wind-induced response of high-rise buildings are investigated. The major contents and achievements are summarized as follows.A wireless high-frequency anemometer instrumentation system for field measurement is developed. Specifically, the architecture of the proposed instrumentation system, measured data transmission and treatment, main features and its practical application are presented.Compared with traditional systems, the developed system has some considerable advantages such as real-time transmission, low cost, no manual collection of data, and convenience.Hence the developed system can facilitate the field measurement in remote areas such as mountain valleys.Based on the engineering background of Puli Great Bridge in a mountainous area, the wind characteristic in the mountainous area is investigated. With the measured data ready,the wind characteristic in the mountainous area is classified into several wind types. For different wind types, the mean wind characteristic is analyzed individually. Finally,the fluctuation characteristic of downburst winds is highlighted. Results show that the thermally developed wind is the major wind type in the mountainous area. Furthermore, the thunderstorm downburst happened many times during the field measurement. The 10-min mean wind speed of the measured downburst winds can reach 13.3 m/s, and the mean wind directions of the measured downburst winds are of great discrepancies. In addition, the majority of mean wind elevations of the measured downbursts are positive. The fitted wind profile is different from the existing wind profile model of downbursts. Regarding the fluctuation of the selected downbursts, the turbulent intensity of the vertical wind speed is overall higher than that of the horizontal wind speed. Finally, the noticeable time-varying features have been found in the estimated coherence functions of both horizontal and vertical wind speeds, which the former is overall larger than the latter.A stochastic wave-based simulation approach for the multivariate nonstationary random process is proposed. Central to the proposed approach is the transformation of the simulation of a multivariate nonstationary random process along a straight line to that of a nonstationary one-dimensional stochastic wave. The application condition of this approach is that the coherence function should be a function of only the separation distance and frequency. The approach contains two schemes. The first scheme relies on the direct summation of cosine items. This scheme is still not very efficient although the spectral matrix decomposition can be avoided. After using the proper orthogonal decomposition (POD) to factorize the time and space-dependent decomposed two-dimensional evolutionary power spectral density (2D EPSD) of the converted stochastic wave, the simulation of nonstationary and nonhomogeneous stochastic waves has been transformed to the summation of a series of the simulation of stationary and homogeneous stochastic waves. Hence, 2D fast Fourier transform (FFT) can be employed to further increase the simulation efficiency. Numerical examples show that the simulation accuracy of the proposed approach can be satisfactory.More importantly, the proposed approach may enhance the simulation efficiency of stationary and nonstationary processes dramatically.Starting from the Sigma-oscillatory process theory, the Wold-Cramer decomposition model is derived. The relationship between these two models is discussed in detail. Due to the easy application, the latter is employed to model the evolutionary spectra-based time-varying coherence function. Furthermore, the POD-based fast simulation method is modified slightly to take into account the influence of time-varying coherence functions.Finally, two sets of measured downburst winds were used to verify the introduced theory and proposed method. Results show that the estimated coherence functions have obvious time-varying characteristics whose variation trend is consistent with the according time-varying standard deviation. In addition, the coherence functions decrease obviously with the increase of separation distances. When the EPSDs along different heights admit large discrepancies, the POD-based method fails to remain the enough accuracy.With the aid of the established theoretical framework of time-varying coherence functions, the alongwind response analysis framework of tall buildings to nonstationary winds is extended. Numerical examples were employed to verify the above theoretical framework and compare the structural response under the time-varying and corresponding time-invariant coherence functions with the consideration of different time-varying vertical profiles. Results show that the root mean square (RMS) responses of the tip displacement under the case of time-varying coherence functions is noticeably larger than those under the case of time-invariant coherence functions for the building with height of 300 m. The biggest difference can even reach 41%. When it comes to the mean extreme of the tip displacement,the response difference between time-varying and time-invariant coherence functions tends to be reduced. However, the largest difference in terms of the estimated mean extreme response can also be 10%.