WIND LOAD MODEL FOR LARGE-SPAN FLAT ROOFS AND ITS APPLICATION IN DATABASE-ASSISTED DESIGN

Wind load database is useful to provide some information for determining wind loads on large-span roofs during the design stage. Wind load data from different source were aggregated into wind load database for analyzing and predicting use. As wind load data accumulate, data mining is necessary to preserve analyze and apply these data, and the modelling of wind load is the most important step to carry out which also can be viewed as a compression of wind load data since wind load model can reflect the wind load characteristics and can be applied in reproduction of fluctuating wind load data. In this paper, the wind load data was modeled as 4 statistical parameters: mean, RMS, skewness and kurtosis of wind pressure coefficients; and 4 spectral parameters that depict the characteristics in frequency domain according to the reduced wind pressure spectral curve shapes: dominant frequency fin, descending slope of reduced auto-spectrum curve k2, peak value of reduced auto-spectrum Sm and decay coefficient of coherence function kc. From the distribution of these parameters, the action of body-induced turbulence can be reflected. These parameters are expressed in zoning partition and can be used to rebuild the wind pressure field by multivariate non-Gaussian stochastic simulations. The feasibility of the wind load models is verified by applying the simulated wind pressure data into wind-induced responses analysis and extreme wind pressure estimation and comparing with results based on wind tunnel tests.. Finally, large-span flat roof is taken as an example to illustrate the procedure of modeling and application of wind load.