Extreme wind pressure estimation based on the r largest order statistics model

This paper presents a procedure for statistical estimation of extreme wind pressures using the r largest order statistics (r-LOS) model, which includes a joint generalized extreme value (GEV) model and joint Gumbel model. Methods are devised to extract r-LOS vectors of independent peaks from individual time histories, choose optimum r, and discriminate between the r-LOS GEV model and r-LOS Gumbel model, respectively. The procedure is applied to analyze the pressure data obtained on the rigid model of a low-rise industrial building. When multiple pressure time histories are used to estimate the extreme pressure coefficients, the r-LOS Gumbel model is superior to the r-LOS GEV model and the classical Gumbel model. When a single time history is used, the r-LOS Gumbel model usually estimates the extreme pressure coefficients more accurately than the peak factor method based on the Modified Hermite Model and Sadek-Simiu procedure, and it is applicable when the wind pressure is non-Gaussian; furthermore, the r-LOS Gumbel model gives an analytical solution to the quantiles of extreme pressure coefficients. The paper concludes that the procedure based on the r-LOS Gumbel model is an effective alternative for the estimation of extreme wind pressures when either multiple pressure time histories or a single time history is available.