| ABSTRACT:For a long time, the extreme value of wind pressure was estimated based on the quasi-steady hypothesis and the peak factor method proposed by Davenport. Based on the hypothesis, both the wind speed and wind pressure comply with Gaussian distribution. However, researchers now find out that when wind blows across buildings, turbulence can be generated on side walls, leeside walls and roofs which will cause Non-Gaussian pressure at these locations. In this case, the peak facor estimated based on quasi-steady hypothesis will be smaller than the actual value and in turn lead to unsafe design. So it is necessary to research the Non-Gaussian wind load and provide references for the future structural design.In this thesis, wind load features on cladding and components of cylinder roofs were researched rely on the result of wind tunnel tests. The modified Hermite Moment Model was used to estimate the peak factor and a zoning process in terms of the extreme wind pressure was provided. The zoning result was compared with Chinese and Japanese wind load codes as well. The main contents are as follows:(1) Brief descriptions of different extreme wind pressure estimate methods were presented, including three classical methods and two transform methods. By discussing the advantages and disadvantages of each method, the modified Hermite Moment Model was choosed at last to estimate the peak factor and extreme wind pressure.(2) Wind tunnel tests in terms of cylinder roof structures were carried out. Wind pressures on roofs with different rise span ratios and under different wind directions were researched based on tests; Non-Gaussian features of wind pressures were discussed as well. On the basis of the test result, effects of the rise span ratio and the wind direction on wind pressures were analysed. It was found that the rise span ratio had influences on the value of wind pressure, the distribution pattern of small scale vortexes at wind front and the characteristic parameter of the columnar vortex and the conical vortex. Also, the wind direction had influences on the distribution pattern of wind pressure, the peak value of wind pressure coefficient and the lift coefficient.(3) Peak factors and extreme wind pressures were estimated based on the modified Hermite Moment Model. Features of the extreme wind pressure on each model under different wind directions were analysed as well. Based on the analysis, it was found that the distribution pattern between the peak factor and the wind pressure was quite different and there was no relevance between these two. Besides, the distribution pattern of the extreme wind pressure was influcenced by the mean wind pressure, the fluctuant wind pressure and the peak factor.(4) A zoning process and the extreme wind pressure coefficient in each zone were given based on clustering method. And the zoning result was compared with Chinses and Japanese wind load codes. The comparision showed that the extreme wind pressure coefficient derived from this research was close to the suggestion value in the Japanese code but was quite different from the suggestion value in the Chinese code. |