Best-fit probability models for wind loads and its effect on the structural reliability

Wind loads are very important in the design of buildings as the load combinations, including the wind load, often govern the design. Compared to other loads, such as gravity load, wind loads are more uncertain and therefore bring about a lower safety index as the ratio of wind to gravity loads increases. Wind load, which is evaluated in terms of wind pressure, is a product of several factors. Wind speed is the main parameter. Other factors depend on the structural shape and geometry. In order to perform a realistic and accurate reliability analysis, it is important to find a suitable model for wind pressure. Previous studies have used Extreme Type I for modeling the wind load. In this study, based on wind speed data for 105 stations located in non-hurricane regions, this wind pressure model is evaluated and then modified. The best-fit models investigated were Lognormal and Extreme Type I. For these, the probability distribution parameters were averaged and general models were then obtained. Results show that lognormal probability distribution could better model wind pressure, although the Extreme Type I model is also close to Lognormal. It is shown that using lognormal probabilistic distribution leads to more conservative reliability indices than those from Extreme Type I.