EVALUATING THE ACCURACY OF LARGE-EDDY SIMULATIONS IN PREDICTING TURBULENT STATISTICS AROUND A 1:1:2 BLOCK MODEL:

A COMPARISON WITH WIND TUNNEL EXPERIMENTS

Abstract

Accurate prediction of the low-occurrence wind speeds around urban structures is critical for secure building design and pedestrian safety. In general, Large-eddy simulations (LESs) are preferred over Reynolds-Averaged Navier-Stokes (RANS) simulations for the prediction accuracy in the turbulent statistics. However, previous validations focus mainly on fundamental statistics such as the mean and standard deviations of velocity components, leaving discrepancies in other statistics characterizing the unsteadiness of the flow unclear. Therefore, this study evaluates the accuracy and reliability of LESs in forecasting unsteady wind patterns around a 1:1:2 block model. By comparing the statistics determined by various advection schemes in LESs with those by wind-tunnel experiment (WTE), significant discrepancies were found, particularly in high-order statistics of the wind speeds. These findings underscore the necessity of to refine the procedure to employ LESs for better predictive accuracy, especially for the prediction of the low-occurrence wind speeds impacting urban wind environment and safer building designs.