Air Quality Modeling in Dense Urban Areas at Ground Level—CFD, OSM or Gauss?
Paper i proceeding, 2021
There is an ongoing intensive urban densification in many cities. The need for advanced modeling methods that realistically simulate dispersion at ground level in complex urban areas have increased. As more modeling is done outside usual model communities, guidelines supporting the selection of suitable models is needed. Dispersion of air pollutants in street canyons is affected by width, height, length and building structure. Studies show that concentration of NO2 in street canyons can become 5-times higher, compared to open conditions. Since Gaussian models cannot simulate dispersion at ground level in dense urban settlements models that include 3D information (buildings) is therefore required, such as CFD or OSM models to obtain realistic results. The purpose of the project is to investigate the effect from building on air quality and giving recommendations to authorities and consultants helping with model choices. This is based on calculations with three different model types, CFD, OSM and Gaussian for four urban environments, from dense to open suburban areas using the same input. Validation was done with continuous measurements of NO2 at each site. The CFD modeling gave best results at all sites, the OSM modeling quality varied between the sites, with slight to about 50% underestimation. The Gaussian modeling generally underestimated the concentration with 50%. The aspect ratio H/W ≥ 0.7 favors development of vortices and lateral dispersion of pollutants in street canyons. It was shown that with at least this aspect ratio locally emitted NO2 frequently becomes at least 2–3 times higher compared an open road case. A methodology has been developed visualized in a flow chart, for help choosing an appropriate model for each environment. Most important factors are; street canyon structure; local concentration level; proximity to high emissions/risk of leakage into calculation area.
Street canyon dispersion
Choice of models