Probabilistic descriptions of wind effects and wind-load factors for database-assisted design

Database-assisted design (DAD) for wind results in structures with reduced material consumption and failure risks, i.e., in improved safety and sustainability. The paper presents a DAD-compatible probabilistic procedure for estimating: (1) Wind effects corresponding to any mean recurrence interval of the wind speeds, and (2) wind-load factors. Unlike the ASCE 7 Standard and earlier structural reliability estimates, the procedure accounts for the inherent randomness in the estimation of peaks and for relevant wind-related knowledge uncertainties. On the basis of a selection of uncertainty parameters believed to be reasonable, it is concluded that: (1) The extreme wind speed distribution type significantly affects the estimates of the load factors. Our estimates, based on the Extreme Value Type III Reverse Weibull distribution, yield load factors consistent with values specified in the ASCE 7 Standard-in contrast to estimates published in the 1980's. (2) The variability of the ratio between wind speeds averaged over different time intervals can significantly affect the estimated wind-load factor. (3) An increase in size of the extreme wind data set from, say, 30 to 50 years, has a marginal effect on load factor estimates. (4) For low-rise buildings in open terrain, wind-load factors are typically affected weakly by even large uncertainties with respect to terrain roughness length. Additional research on the effect of terrain roughness uncertainties for buildings in built-up terrain may require an expansion of the aerodynamic databases currently available. It is noted that consideration of uncertainties in isolation, as commonly performed by some analysts and wind tunnel operators, especially for peak effects estimation, can be misleading; for structural applications, uncertainties need to be considered collectively.