Uncertainties in the estimation of local peak pressures on low-rise buildings by using the Gumbel distribution fitting approach

The lack of a standard accepted method to estimate local peak pressure coefficients from wind tunnel data can lead to inconsistent definitions and interpretations, particularly because cost and time constraints associated with wind tunnel tests of low-rise buildings necessitate relatively short (equivalent full-scale) durations. This paper focuses on a Gumbel distribution fitting method widely used in practice. Because the sources of uncertainty regarding estimated peaks include the use of short-duration records (in practice) and the assumption that the observed peaks from wind tunnel pressure data are Gumbel-distributed, this is quantified in detail in terms of the parameters determining the required minimum record length. It is shown that 15 observed peaks can lead to local peak pressure estimates with adequate precision for many design scenarios. However, the conversion of peak coefficients from a short duration to those of a longer duration requires an increase in the number of observed peaks to maintain precision.