Search

We perform Computational Fluid Dynamics (CFD) analyses of a low-rise building immersed in a turbulent boundary layer. The Silsoe 6 m cube is chosen as the main case study aiming at representing the key features of the typical wind conditions experienced by low-rise buildings. The turbulent inflow condition is generated by employing a precursor technique. A sensitivity study is conducted to investigate the effects of eight subgrid-scale (SGS) models on the flow around the cube. The incoming turbulence intensity and mesh sizing for each of the SGS models are also explored. The results in terms of time-averaged pressure fields for relatively well-resolved simulations are well compliant with available experimental data and insensitive to the selection of the SGS model. We demonstrate that, while the effect of the aforementioned factors has a limited impact on the first-order statistics of the pressure coefficients, their influence becomes much more pronounced when analysing higher-order statistics (i.e. variance, skewness and kurtosis) and extreme values. The extent of their relative importance depends on the location of the point under investigation and the quantity of interest. The current study provides guidelines on the choice of factors which can strongly affect the results obtained employing CFD simulations when the local peak pressures are of interest, for example, as in the case of cladding design.

In this article, the accuracy of standard computational fluid dynamics techniques and turbulence models in predicting the critical flutter speed of streamlined and bluff deck sections is investigated. It is well known that such numerical tools can provide inaccurate prediction of the flow field around bluff bodies which reflects in biases in their aeroelastic behavior. Aiming at systematically study the effects of such inaccuracies, in this article, six deck sections are analyzed by numerically extracting flutter derivatives and calculating the critical flutter speed for a set of representative test structures. It is shown that the flutter onset velocity is mainly underestimated but cases showing opposite behavior are identified and discussed. Two simulation strategies are adopted: the differences in the flutter mechanism between such two approaches are taken as an indicator of possible strong inaccuracies.