Dispersion of air pollutants around buildings: A review of past studies and their methodologies

The review shows that on-site measurement has the benefit of obtaining real life data under real atmospheric boundary layer conditions, but it is too difficult to capture the real atmospheric conditions and too expensive to conduct such experiment as too many samplers and receptors are needed, particularly for investigating the air pollutant issues with regard to building arrays. This difficulty can be solved by physical-scale modelling, which can provide more controllable airflow boundary conditions, but accuracy is compromised by its inability to simulate real boundary layer turbulence conditions and buoyancy effects. Water channel is advantageous in buoyancy experiments but maintaining realistic viscosity and density properties of the medium could be challenging. Direct numerical simulation can provide accurate results; but it is impractical to simulate flows in large domains due to high computational requirements. Reynold-averaged Navier-Stokes numerical simulation is the most commonly used model for pollutant dispersion and infection control analysis but generally over-predicts surface concentrations at the leeward side of a building. Although large eddy simulation can over-predict the lateral pollutant concentration in the wake region of the building, flow unsteadiness and intermittency can be solved.