Abstract :
[en] The non-Gaussian nature of turbulent wind loading has been long accepted. Nevertheless, although this might have significant influence on design quantities, it seems to be very often ignored when it comes to practical applications, specially when dealing with real civil engineering structures, ranging from medium to quite large dimensions.
Unfortunately, the need for significantly high computational power has been the main challenge to overcome. However, things have evolved quite fast in that domain over the last few years, possibly providing better tools for its effective computation.
Proper Orthogonal Decomposition (POD) and Reduced Order Model (ROM) techniques have also helped scientists and practitioners, in several fields including Wind Engineering,
to tackle the complexities of the considered problem. Using physical and mathematical tools, POD can help at reducing the problem dimensionalities, and making it more affordable to be solved numerically, whenever analytical formulations are not derivable.
This works aims at presenting a new framework for tackling the Bispectral problem applied to large civil engineering structures, under non-Gaussian wind loads. To do so, we leverage both mathematical and numerical tools. We will provide a novel formulation of application of POD techniques to the Bispectral problem. Also, a novel algorithmic arrangement for tackling efficiently the computation of non-Gaussian features of the wind load, i.e. the Bispectrum, will be presented.
Coupling these two aspects, application of Bispectral analysis to real civil engineering examples will be provided. This will lead to the final effort of providing a reasonable answer to a fundamental question: when exactly do we need a Bispectral analysis?
