SINGULAR VALUE DECOMPOSITION ANALYSIS FOR WIND ENVIRONMENT PREDICTION AT PEDESTRIAN LEVEL IN SIMPLIFIED CUBICAL ARRAY USING LARGE-EDDY SIMULATIONS

Abstract

The effect of urban geometries of simplified urban-like arrays on the turbulent airflow at the pedestrian level has been investigated by wind tunnel experiments and computational fluid dynamics. In these experiments, there are the limitations of the spatial and temporal sizes of stored datasets due to the high spatial and temporal resolutions. Therefore, in the present study, the airflow over simplified urban-like arrays is determined by large-eddy simulations using external force accelerating the flow with the periodic boundary conditions and the principal component analysis based on the singular value decomposition is conducted. The characteristics of the singular modes are compared using different sizes of the original datasets to clarify the effect of the sizes of the datasets on the resolvable turbulent properties. Additionally, low-rank approximation analysis was performed on the wind speed fields at pedestrian level to understand the dominant singular modes on the gusty phenomena.