Buffeting And Its Extreme Response Of Bridge Structures Under Non-gaussian Wind Field

Modern bridge structure has the trend to be longer,thinner and more flexible,which will inevitably lead to the result that the structure is more sensitive to the effect of wind.Hence the study on wind-induced vibration of bridge structure is becoming more and more important.For wind-induced vibration,flutter and buffeting have always been the focus of the research.For a long time,pulsating wind has been regarded as a stationary Gaussian random process.However,the environment of bridge structure is often complicated.As a result,pulsating wind sometimes has non-Gaussian characteristics.The non-Gaussian characteristics of wind load may increase the extreme response of structure and accelerate the fatigue failure of structure.Therefore,it is of great engineering significance to accurately describe the probabilistic characteristics of pulsating wind and evaluate the influence of non-Gaussian pulsating wind on structural buffeting and related extreme response.In view of this,buffeting and extreme response of bridge structure under the action of non-Gaussian wind field is investigated in the thesis.The main work is summarized as follows:According to the power spectral density,skewness and kurtosis of non-Gaussian pulsating wind,the simulation of non-Gaussian pulsating wind field was realized based on Hermite polynomial model.By comparing with the target value,the reliability of the non-Gaussian wind field simulation method is verified,which provides input for the follow-up calculation of bridge buffeting response.It is found that the amplitude and standard deviation of non-Gaussian wind speed are larger than that of Gaussian wind speed,and stronger non-Gaussian characteristics correspond to larger amplitude and standard deviation of wind speed.The finite element model of the bridge structure is established by ANSYS,and the corresponding wind load is obtained based on the simulated non-Gaussian pulsating wind speed time history.Then the buffeting response of the structure is calculated.According to the buffeting response time history,the power spectrum,root mean square,skewness,kurtosis,probability density and other statistical characteristics of the response were evaluated.The results show that the structure response power spectral density is greater than that of Gaussian wind field at low frequency under the action of non-Gaussian wind field.Meanwhile,under the same average wind speed,the amplitude and root mean square of response under non-Gaussian wind field are larger.The stronger the non-Gaussian characteristic of the wind field is,the larger the RMS value of the response is.The difference of the results under strong non-Gaussian and Gaussian pulsating wind speed can reach 21.5%.Therefore,when the wind field has nonGaussian characteristics,the buffeting response results based on the Gaussian wind field underestimate the actual response.The skewness and kurtosis of structural response tend to the result of Gaussian distribution under different non-Gaussian pulsating wind field,and the probability characteristics tend to be Gaussian distribution.Based on ACER method,the short-term extreme value distribution of structural response under a certain return period wind speed is estimated.Based on the assumption that the shortterm extreme values are mutually independent,and combined with the probability distribution of the annual maximum average wind speed,the long-term extreme value distribution of the structural response under pulsating wind field with different non-Gaussian characteristics are obtained,and the long-term extreme values of the given return period are finally determined.The results show that both short-term and long-term extreme responses caused by non-Gaussian pulsating wind are greater than those caused by Gaussian pulsating wind,and the stronger the non-Gaussian characteristics are,the greater the extreme response is.For the 100-year return period,the strongly non-Gaussian case was 27.1%larger than the Gaussian case.The weakly non-Gaussian case is 23.1%larger than the Gaussian case.