In the study, the modified discretizing and synthesizing random flow generation (MDSRFG) was adopted to generate an anisotropic boundary layer inlet for large-eddy simulation. The mean velocity, turbulence intensity and turbulence length scale distributions at inlet, were defined according to the measurements at TKU wind tunnel. The von Kármán model was used as the target spectrum. Wind tunnel pressure measurements on a square prism model with aspect ratio of 3 was used for validation of numerical simulation. Results show that turbulence energy is well maintained from the inlet to the downstream. The relative differences between the measurement and predicted results are 3.4% (mean drag coefficient), 11% (fluctuating drag coefficient), 25.6% (fluctuating side force coefficient) and 4.7% (Strouhal number). The simulated mean and fluctuating pressure distributions showed good agreements with the experiments. The averaged differences between measurement and predicted results are 14.49% (mean pressure coefficient) and 13.74% (fluctuating pressure coefficient). This indicates that the adoption of a reasonable process based on the MDSRFG method is an effective tool to generate a spatially correlated atmospheric boundary layer flow field.
Journal of Wind Engineering and Industrial Aerodynamics 146, pp.51–58