Specification of the design wind load based on wind tunnel experiments

Once a high-quality wind tunnel experiment with no major flaws on the experimental side has been performed and the data have been processed/translated appropriately to the design-decisive variable, the final step is to estimate the appropriate load or load effect coefficient for the specification of the design wind load. Four questions arise: what is the appropriate length of a single run in the wind tunnel, what is the appropriate fractile of the observed extreme coefficients, how many independent experiments are required to estimate this fractile and what should be the target confidence interval. The paper tries to give some answers to these questions and discusses the findings in regard to some wind load codes.