Uncertainties in wind-induced loads on hyperbolic paraboloid roofs: Wind-tunnel tests and analytical models

The design coefficients provided by a wind-tunnel campaign on in-scale models, e.g. the pressure coefficients discussed in this paper, are always affected by physiological and unavoidable uncertainties. On the other hand, wind-induced loads at the real scale (e.g. peak pressure coefficients) may be provided, with a prescribed confidence level, by analytical models that include stochastic characteristics of the wind field (mean values, standard deviation, skewness, kurtosis) obtained by wind-tunnel tests. Based on aerodynamic tests on hyperbolic paraboloid models in the boundary layer wind-tunnel of the CRIACIV (Italy), this paper discusses a typical example of experimental uncertainties, i.e. the asymmetries measured for geometries and wind directions nominally symmetric, and investigates their effects on peak pressure coefficients provided by analytical models. To this aim, experimental peak pressure coefficients and their analytical prediction are compared for symmetrical geometries under wind flow direction parallel to the symmetry axis, and it is noted that asymmetries shown by experimental tests may be reduced by appropriate analytical models, that can therefore contribute to filter out the effects of experimental uncertainties for design purposes.