Validation of a multivariate non-Gaussian and non-stationary stochastic wind pressure model driven by stationary wind tunnel data

With growing interest in probabilistic assessments of structures subject to extreme winds, there is a need for stochastic simulation schemes that can capture the inherently non-Gaussian and non-stationary (NG–NS) features of wind loads during the simulation of path-dependent phenomena such as inelasticity, fatigue, and cladding damage. Currently, no validated NG–NS stochastic load model exists that can be calibrated to standard wind tunnel data. A recently developed model proposed the stochastic simulation of direction-wise pressures based on standard wind tunnel data, followed by a filter-based transition scheme for generating NG–NS wind load records. Recognizing the importance of validation in creating confidence in a model, this work focuses on extensive experimental validation of the NG–NS stochastic load model. Validation is achieved using data from specifically designed wind tunnel tests, whereby NG–NS records are obtained through continuous rotation of the turntable. Six different rates of change in wind direction, consistent with those expected in hurricanes, are considered with validation based on quantitative time–frequency error analysis using spectrograms, as well as the examination of temporal trends in statistical properties. The results show that the scheme correlates well with the experimental data, with spectrogram-based errors remaining within 10% across all pressure taps and test conditions.