Analysis and Characterization of Hurricane Winds

To understand and model the effects of natural hurricane winds on residential buildings, full-scale field testing and analysis are needed. Toward this goal, a group of researchers at Florida Tech, in collaboration with the Florida Coastal Monitoring Project team, have deployed sensors in a real storm scenario and collected roof wind pressure and local wind data. Because the behavior of a hurricane is subject to vagaries of nature, the current research effort is oriented to develop a rational basis for analyzing the data. To investigate this, the variable time averaging is applied to the data collected by the Florida Coastal Monitoring Project from three storms: Hurricane Gordon, Hurricane Isidore, and Hurricane Rita. The wind data correspond to a 10-m-high tower in different exposures at a 10-Hz sampling rate. In the wind load specifications worldwide, a wide range of averaging times have been used, 3–60 s to define gusts and 10–60 min to define mean speeds. This is seemingly more heuristic for the local hurricane wind field for which ergodicity and stationarity are not well established. In this study, variable averaging times from 1 to 1,000 s were applied to the wind speed time series data. Then three-dimensional statistical maps of the mean, RMS, skewness, and kurtosis distributions are obtained as functions of the averaging period. The effect of the averaging time on the mean is evidenced by a large variance for short averaging periods and a small variance for long averaging periods for all three hurricanes. The variation of the mean reaches a minimum at approximately 300 s. Therefore, all the higher-order moments are also calculated with respect to the 300-s mean. The skewness values are different (nonzero) for different storms, indicating a non-Gaussian behavior for the fluctuations, although the kurtosis values range mostly from 2.5 to 3.5.