Assessing urban ventilation using large-eddy simulations

The ventilation of nine neutral boundary layers over three representative urban geometries (in-line, staggered, and long-canyon building layouts) under three different geostrophic wind conditions was investigated using large-eddy simulations. Pollutants were injected into a well-developed boundary layer, and the replacement of pollutants with fresh air was simulated in the nine cases. In all cases, the domain-averaged concentration first decreased linearly, following the overall flow, and then decreased slowly owing to the trapping and intermittent drainage of canopy-layer pollutants. During drainage, downstream in-canopy pollutant concentrations were the highest in the long-canyon layout owing to more building-induced vertical recirculations, more isolated flows, and more trapping of pollutants. In contrast, the in-line layout had the lowest downstream in-canopy pollutant concentrations. Pedestrian-level ventilation was investigated by analyzing the changes in 2-m pollutant concentration over normalized time, considering the overall flow and building-induced turbulent friction together. The time series of the pedestrian-level pollutant concentrations in the nine cases were divided into three types based on the building layout rather than geostrophic wind speed. Additionally, the normalized ventilation time, that is, the normalized time for the initial concentration to decrease to 1/e, was calculated and compared to the simple ratios of the 2-m wind speed and geostrophic wind speed. While the velocity ratios and non-normalized ratios were grouped by geostrophic wind speed, the normalized ventilation times were grouped into three groups according to the building layout, showing that differences in geometry (e.g., long-canyon to in-line) can decrease ventilation time by more than 9.7%.