A comparative study on methods for estimating extreme wind-induced response of structures

To evaluate the accuracy of some engineering methods for estimating the extreme values of wind induced response, which are usually based on the Gaussion distribution assumption, repeated wind tunnel tests and wind-induced response analysis with over 1000 samples were carried out on a single-layer spherical lattice shell structure, and more than 1000 samples of the structural response time-history were obtained. Secondly, the extreme value of the structural response was analyzed by the method of classical extreme value theory based on the large number of samples, and can be regarded as the reference value. Then, the Peak Factor method, the Modified Hermite Model method, and the Sadek-Simiu method are used seperately to estimate the extreme values of structural response based on only one sample. Finally, statistical analysis is carried out on these estimated values, and the stablity and accuracy of these three methods are investigated. The results indicate that the estimation results from the Peak Factor method have small deviation, but most are underestimated; the estimation results of the Modified Hermite Model method vary greatly, and most are overestimated; the estimation results of the Sadek-Simiu method are close to the reference value. Thus in practice it is recommended to estimate extreme values by the Sadek-Simiu method.