The exceedence probability of a desing wind load in influenced by two contributions with a distinct random character; the intensity of a storm and the fluctuations over the length of a storm. In recent years, a number of papers suggest that beside the Gumbel distribution also type III extreme value distribution seem to be an appropriate model for both the extreme wind speeds and the extreme action or action effect co-efficients. The paper discusses in its first part what fractile values of the extreme wind speeds and the extreme coefficients are to be combined to obtain the target overloading risk in dependence of hte characteristic vlaues of the contributing extreme value distributions. The second part of the paper deals with the problem of estimating extreme rare events for confined samples sizes. For confidence intervals of 75
展开▼