Uncertainty in the dynamic properties of tall buildings and propagation to the wind-induced response

The response of tall buildings to wind actions is commonly assessed through the quasistatic approach considering mean, background, and resonant components of the action. The latter accounts for the amplification due to resonance and depends on the dynamic properties of the buildings, that is, modal mass, frequency, and damping ratio. Selecting appropriate values of the modal parameters of tall buildings is not immediate and is usually done using predictive models. These contain uncertainty, which eventually propagates to the dynamic response. The main aim of the paper is the assessment of uncertainty in the dynamic response of flexible buildings to wind action arising from a not perfect knowledge of their dynamic properties. The paper explicitly refers to the dynamic models given by Eurocode 1, but the approach is general, and the analyses can be repeated selecting any other model. It is found that the bias in the dynamic factor is always less than one, with values on average between 0.86 and 0.98. This indicates that the approach of Eurocode 1 is conservative. The only exception is that of the acrosswind response of steel buildings with an high aspect ratio, in which case the bias can be as large as 1.16. As to randomness, the coefficient of variation of the alongwind dynamic factor is very seldom found to exceed 10%, with average values around 5%. Such values are much lower than those of the coefficient of variation of damping, which is in the order of 50% or more. This indicates that uncertainty attenuates when it propagates to the response. On the other hand, the coefficient of variation of the acrosswind and torsional dynamic factors reaches values of 20% or more, indicating that such attenuation is much lower in that case.