Shape dependence of wind pressure peak factor statistics in hyperbolic paraboloid roofs

The complex shape dependence of wind pressure's peak factors (PFs) makes it difficult to predict their values for all geometries by using analytical models. Thus, experimental studies are necessary to validate available or new analytical models for each specific shape and geometry. Very little information is available in the literature for the hyperbolic paraboloid geometry, which is frequently used for cable net and membrane roofs that are particularly appropriate for covering large spans. This paper focuses on wind pressure's PFs for this specific geometry. It presents a statistical analysis of experimental data taken from wind tunnel experiments on eight different geometries, obtained through the combinations of squared and rectangular plan shapes, high and low models, and two different curvatures of the roof parabolas. Roof zones for which a non-Gaussian behavior is predominant are identified. A comparison is made between PF statistics obtained experimentally and those estimated by the Davenport, modified Hermite, and Translated Peak Process (TPP) models, in terms of means and standard deviation errors estimated over the entire roofs. It is concluded that the modified Hermite and the TPP models provide the best estimates of the PF means and standard deviations, respectively.