Numerical evaluation of pollutant dispersion in the built environment: Comparisons between models and experiments

Steady-state RANS Computational Fluid Dynamics (CFD) simulations of pollutant dispersion in the neutrally stable atmospheric boundary layer are made with the commercial code Fluent 6.1 for three case studies: plume dispersion from an isolated stack, low-momentum exhaust from a rooftop vent on an isolated cubic building model and high-momentum exhaust from a rooftop stack on a low-rise rectangular building with several rooftop structures. The results are compared with the Gaussian model, the semi-empirical ASHRAE model and wind tunnel and full-scale measurements. It is shown that in all three cases and with all turbulence models tested, the lateral plume spread is significantly underestimated. It is suggested that transient simulations might be required to achieve more accurate results. The numerical results are quite sensitive to the value of the turbulent Schmidt number. The comparisons however cannot clearly indicate which Schmidt number is most suitable for which type of flow due to the large number of other error sources in the simulations, including steady-state RANS modelling, turbulence modelling, near-wall treatment limitations and unintended streamwise gradients in the turbulent kinetic energy profiles.