Research On Probability Characteristics And Extreme Value Estimating Method Of Wind Pressure On Surface Of Large-span Roof

Large-span roof structures have been extensively applied in the construction of public structures,such as stadiums,exhibition centers,railway stations,etc.In order to maximize the seating capacity and meet the demands of novel architectural design,truss and grid structures are frequently employed for the supporting system.This type of structure roof is characterized with large horizontal span,low weight,and sometimes openings and cantilevers,which inevitably increases its sensitivity to wind-induced damages.Wind-induced damage is often considered to be one of the dominant factors that cause the failure of large-span structures.Therefore,the accurate estimation of extreme wind pressures on large-span structures has been subject to increasing concerns in the field of wind engineering.In this dissertation,wind tunnel experiments were conducted on the basis of five selected models featured with different rise-to-span ratios,and a stadium project.The state-of-the-art of commonly used probability distribution functions and extreme value estimation methods were revisited,and the characteristics of surface mean and fluctuating pressure coefficients,as well as the probability statistics and distribution of extreme values,were examined and discussed in detail.The major contents included in this dissertation are summarized as follows:(1)Introduction of wind tunnel testing and analysis of wind pressure characteristics.In the first and foremost,the general background of the conducted wind tunnel tests at Hunan University was introduced and the basic methods associated with data processing were presented.In the second part,the distributions of surface pressure coefficients on wind tunnel models were analyzed,in which a good agreement was obtained with respect to the variation patterns of mean and fluctuating pressure coefficients.The model roof was mostly subjected to negative pressures,and flow separation existed near the roof bulge tends to produce high negative pressure.For the cantilevered parts of the roof and walls,superimposing effects were observed on both the upper and lower surfaces in the upwind direction.The distribution of surface pressure was found to be a function of rise-to-span ratio,and the distinct feature of conical vortex near the roof corner vanished as the rise-to-span ratio increases.Last but not least,the characteristics of pressure coefficients on walls and local shape coefficient were investigated.The results indicated that the pressure coefficient near the corner of walls was under the influence of flow separation,which in general has high negative values,whereas the changes in the rise-to-span ratio reveal little effect on pressure coefficient.As for the estimation of local shape coefficient on model roof,the recommended values in Chinese standards for flat-roof model were found to agree well with the results derived from wind tunnel tests.However,for the arch-roof model,particularly near the roof ridge area,the Chinese standards underestimate the magnitude of local shape coefficient by a percentage about 14 to 30,as compared to wind tunnel results.For the local shape coefficients of walls,the values derived at both windward and leeward regions matched with those given in the standards,while significant deviations were obtained on the sides.The derived shape coefficients of internal pressure were about-0.10 to 0.16,which were well located in the range specified in the standard.(2)Statistical analysis of surface pressure probability of large-span structure.Firstly,the distribution characteristics of the third and fourth order statistical parameters of roof surface pressures was discussed,and their non-Gaussianity were examined.Meanwhile,a number of common probability density functions were applied to describe the distribution of surface pressures at the typical measurement locations.It was shown that,the pressure distributions at typical measurement taps can generally be well fitted using Gaussian,generalized extreme value(GEV),as well as 3-parameters Gamma distribution functions.Nevertheless,no unified distribution function is currently available to produce desirable results for all measurement taps.(3)Review of extreme value theories and corresponding extreme value estimation methods.In the first place,emphases were placed on the review of the theoretical basis of zero up-crossing theory and classical extreme value theory,and the widely-accepted extreme value estimation methods were discussed and modified.Furthermore,comparative analysis was implemented associated with these extreme value estimation methods,and the most optimized method for this study was identified.(4)Analysis on the distribution of extreme wind pressure of a large-span structure and discussion of an engineering practice.The distributions of extreme wind pressures on roof surface were estimated using an appropriate method,and the dependence of surface extreme wind pressure on the variation of geometrical parameters of models was identified.It was shown that,the short side direction for the cantilevered parts of the roof tends to produce high negative pressure,while presenting a weakening effect in the long side direction of model roof as the rise-to-span ratio increases.With a constant rise-to-span ratio,the extreme pressure increases for single wall opening,which is unfavorable to the roof,but it present an opposite effect for the double symmetrical wall openings.In addition,detailed investigation on surface pressures of Meizhou stadium project was presented.