A COMPARATIVE STUDY ON MOMENT-BASED TRANSLATION PROCESS METHODS FOR THE PEAK FACTOR ESTIMATION OF NON-GAUSSIAN WIND PRESSURES

Cladding design requires accurate estimation of the peak values or peak factors of non-Gaussian wind pressures. Among different translation process methods for estimating the peak factors, the moment-based Hermite polynomial model (HPM), moment-based Johnson transformation model (JTM) and moment-based shifted generalized lognormal distribution model (SGLD) are three commonly used ones. The peak value or peak factor is determined by the parent probability density function (PDF) tail. Currently, the differences of the predicted parent PDF of non-Gaussian process based on the same first four statistical moments by those models are not clear, which makes the selection of those models for peak value estimation ambiguous. To compare the differences of the three models and provide certain selection principles, this paper systematically compares the accuracy on the peak factor estimation of non-Gaussian wind pressures by the three methods. We numerically compare the difference of the predicted PDF and peak factors. Very long time wind pressures from a wind tunnel test are then used to validate the numerical observations and examine the accuracy of the predictions. The results showed that the HPM generally gave the most accurate estimations.