Comparisons of Statistics for Wind Speeds Around an Isolated Building Determined by Wind-tunnel Experiment and Numerical Simulation

Accurately predicting low-occurrence wind speeds around urban structures is vital for proper building design as well as pedestrian safety. Although Large-eddy simulation (LES) is often used as a high-fidelity model compared to Reynolds-averaged Navier–Stokes (RANS) simulation, current validation processes mainly compare fundamental statistics of the mean and standard deviations of the wind speed. Discrepancies between LESs and wind-tunnel experiments (WTE) remain unclear, particularly regarding physical quantities that characterize the unsteadiness of turbulent flows, such as probability densities, power spectral densities, and high-order moments of the wind speed. This study evaluates LES's results in predicting unsteady wind behavior around a 1:1:2 block model. Two advection schemes in LESs, including first-order upwind and second-order linear were investigated. The results show notable discrepancies, particularly in high-order statistics with WTE consistently exhibiting higher energy levels across all frequencies. These findings underscore the need to refine numerical models to improve their predictive accuracy.