Generation of Large Directional Wind Speed Data Sets for Estimation of Wind Effects with Long Return Periods

Abstract

For structures sensitive to wind directionality, methods for the estimation of wind effects require the use of time series of directional wind speeds that cover time periods exceeding the length of the mean recurrence interval (MRI) of interest to the design. This study proposes a procedure for generating such time series from relatively short wind data sets. First, an algorithm is developed for estimating the parameters of the distributions of the directional wind speeds, given that the size of the data sample within one or more directional sectors can in some instances be too small for statistical inference purposes. Once the distribution parameters are estimated, a simple Monte Carlo procedure is used for data generation. The wind speed data being generated can be used within the framework of the database-assisted design approach to determine wind effects on buildings by accounting for wind directionality.