Study on Wind Effect Uncertainty and Wind-resistant Design of Large-span Roofs
|School||Harbin Institute of Technology|
|Keywords||large-span space structures wind effect uncertainty repeated wind tunnel tests universal equivalent static wind load|
Reliability based wind-resistant design is an important development directionin the research of wind engineering field. The probabilistic assessment of windloads and load effects on structures, which considers various uncertainties, isessential for a reliability-based risk-consistent structural design to strong winds.Wind-resistant design of large-span structures is characterized as: complex time andspatial distribution of fluctuating wind; multi-mode vibration of wind-inducedresponse; multiple equivalent objectives to calculate equivalent static wind load(ESWL), which makes the effect of uncertainties more significant. Therefore, it hasimportant theoretical value in wind-resistant design and practical significance in thedevelopment of large-span space structures to carry out thorough research onuncertainty of wind response of large-span space structures by both theoretical andexperimental methods. The main contents in this paper are shown as following:1. The research framework of wind effect uncertainty analysisBased on the concept of wind load chains proposed by Davenport, the researchframework of wind effect uncertainty analysis is established. Probability plotcorrelation coefficient (PPCC) method and maximum likelihood estimation methodare adopted to identify the appropriate marginal probability distribution of thestatistics of wind effect. Then global sensitivity analysis is performed in order todiscuss the mechanism of uncertainty propagation. The extreme wind load andresponse under a certain guaranteed ratio are calculated by estimation method basedon extreme value theory.2. Uncertainty analysis of wind load of large-span roofMore than one thousand independent runs of wind tunnel experiments arecarried out in order to obtain a large amount of independent samples. Uncertaintyanalysis is performed with regard to statistics of wind load such as mean value, rootmean square, skewness and kurtosis. Based on these work, corresponding probabilitymodel is established. Then the probability model of extreme wind pressure isproposed based on extreme value theory and the relationship between probabilitydistribution of extreme wind pressure and signature turbulence is discussed. Then theresults from traditional peak factor method are examined. It is found that it isinfeasible for peak factor method to provide distribution of extreme load pressure with consistent guaranteed ratio and the most unfavourable negative pressure isunderestimated, with an error rate of20%~30%.3. Uncertainty analysis of wind-induced response of space structuresBased on the repeated wind tunnel tests, global parametric sensitivity analysis isperformed in order to quantify the strength and relevance of the inputs in determiningthe variation in wind-induced response. The propagation law of uncertainty betweenwind load and wind-induced response is discussed. It is found that1) the uncertaintyof wind-induced response is mainly controlled by wind load uncertainty;2) theparameter sensitivity of wind-induced response is related to the ratio of resonantresponse to the total response. The sensitivity value is larger with the increase of theratio. Further, the parameters of generalized extreme value distribution of the largestdynamic response on typical wind-sensitive structures are established. The proposedprobability model is compared with peak factor method. It is found that peak factormethod could not provide distribution of extreme wind-induced response withconsistent guaranteed ratio.4. Determination of design wind load for cladding under a certainguaranteed ratioIn order to determine design wind load for cladding under a certain guaranteedratio, a procedure for statistical estimation of extreme wind pressures and wind-induced response is proposed by improving Cook’s "method of independent storms".Autocorrelation analysis is carried out to eliminating any shorter period maximawhich are due to other mechanisms. The new approach usually estimates the extremepressure/response more accurately than the peak factor method. Furthermore, ananalytical solution to the quantiles of pressure/response could be obtained.5. Determination of design wind load for main structure under a certainguaranteed ratioIn order to determine design wind load for main structure under a certainguaranteed ratio, a new approach to determine the universal ESWL that reproduce thelargest load effects of large-span roof structures simultaneously is proposed. First, amulti-target equivalent equation is constructed to make the ESWL reproduce thelargest load effects of structure at the same time. Then a constraint equation relatedto the direction of wind load is constructed in order to reduce the dimension ofESWL vector and to exclude some unrealistic and curious distributions.