Multivariate stochastic simulation of wind pressure over low-rise structures through linear model interpolation

Abstract

The study focuses on the simulation of time histories of roof loading on low-rise structures for the purpose of determining design loads. Database assisted design (DAD) utilizes datasets from wind tunnel studies conducted over a variety of building shapes and in various roughness length conditions to provide an online database of time histories of structural loads. With the development of the DAD approach, it is of some importance to investigate simplified interpolation procedures in order to produce accurate pressure time histories for building shapes that have not been tested in the wind tunnel. It was hypothesized that the load time histories from the tested buildings could be used to extrapolate, interpolate, or otherwise infer the time histories of an untested building. This study investigates the efficacy of an interpolation scheme to infer the spectral and probabilistic descriptors of non-Gaussian correlated variables representing roof pressure coefficients based on those measured from wind tunnel tests on comparable but not identical building shapes. A multivariate non-Gaussian simulation method is used to generate time histories based on these interpolated models. Peak aggregate uplift statistics from the simulated loads compare favorably with experimental results. Limitations of the method are discussed.