Space-time characteristics of building wake and their relationship to wind load generation

Abstract

Tall buildings are known to change the wind environment and consequently pollution dispersion and pedestrian comfort. In addition, wind loading and wind-induced dynamic responses of tall buildings must be taken into account during the design stage. The simplified method for computing wind loads on building structures in most wind codes is based on the quasi-steady (QS) approach. Many studies have shown the approach as over-simplified and a fuller understanding of the velocity-pressure-load relationship is required.

The main objective of this study is to investigate more comprehensively the relationship between building wake and wind load generation. Whole-field flow measurement using particle image velocimetry (PIV) is performed around building models in the wind tunnel, simultaneously with concurrent measurement of wind pressures on building surfaces. Statistical methods, including proper orthogonal decomposition (POD), conditional sampling and cross correlation, are used to analyze the space-time characteristics of the velocity-pressure data. POD is applied to velocity or pressure data separately and also as a combined set of velocity-pressure data to extract the deterministic characteristics by projecting the high-dimension data to lower dimensions. The building wake patterns typical for generation of peak wind pressures and wind loads are explored by conditional sampling and interpretation of instantaneous flow events. Consequently, the validity of the QS theory is assessed.

For the velocity-pressure cascade process, the conditional sampling results are not sufficient to reveal the typical wind flow patterns for producing peak forces on a building but the averaged speeds in selected zones shows significant differences from the mean values during peak load events. For instance, during a peak minimum lift force on a building side face, the averaged wind speeds around the building show significant reduction from the mean flow. Inspection of single peak load events is able to suggest a clearer connection between the occurrence of peak loads and the vortex shedding activities and proximity of shear layers to the building side face.

The space-time characteristics of building wake are described using POD analysis. The mean value and probability distribution of the building wake lengths are compared with the reconstructed results. On the central longitudinal vertical plane, the first POD mode is suggested to be connected with pulsations of the mean recirculation bubble, while the two higher modes are probably connected with vortex activities and distortion of the recirculation bubble. On the horizontal planes, mode 1 is clearly contributed by alternative shedding of vortices from the building side walls, while mode 2 features symmetric vortices in the building wake. POD analysis is applied to study the space-time distribution of wind pressures on tall buildings with different cross-section shapes. The first POD mode is found dominant in reconstructing the crosswind force signals. The excellent reconstruction is revealed by the high correlation between the two side face crosswind forces. The presence of recessed cavities on the tall building is found to result in a slight reduction in the dominance of the first POD mode. POD analysis is also applied to study lift force generation on a large cantilevered roof.