Characterizing lower tail components of multivariate non-Gaussian wind loads under typhoon climate: Using the Huang-Cui triangular pulse function model

Wind loads at the edge and corner areas of the roof surface of low-rise buildings under extreme winds exhibit intensive non-Gaussianity, especially a high degree of negative skewness. The cumulative effect of negative wind loads below a specific threshold u0 in the time domain is the direct cause of structure destruction under extreme winds. Therefore, this study proposes a triangular pulse function model to quantitatively characterize lower tail components of wind loads with negative skewness via equal cumulative integral on time. In this model, the two critical factors are the intensity for the absolute value of the threshold u0 and the minimum continuous duration T0. Additionally, this article investigates the dependence structures of the lower tail components from multivariate wind loads, and their values vary from 0.65 to 0.80 at quantiles (0.25,0.25). In characterizing the peak locations, a position function combining Poisson pulses and periodic pulses is established, and the period T for generating periodic pulses is a constant. Through analysis, the median value of storm lengths varies from 7.0 to 8.5, and 95 % are in the interval (5.0, 15.0). Finally, an example is given to illustrate the effectiveness of the continuous integral by the proposed model. For wind loads measured from different locations, the dependence of pulse locations, lower tail storm length, and intensity are investigated comprehensively. In the end, a physical perspective to understand the generation mechanism of lower tail storms is provided via the fractal dimension. The proposed model and analyzed results are of significant engineering guidance value in time-domain numerical simulation of multi-point synchronous non-Gaussian wind loads to improve simulation accuracy. •Wind load samples are non-Gaussian, especially with significant negative skewness.•A triangular pulse model is proposed for characterizing the negative non-Gaussian components.•The pulse location function is composed of Poisson pulses and periodic pulses.•Model parameters can be solved based on equal temporal integral.•The temporal dependence of lower tail storms from different locations is analyzed.