Measurements, Numerical Simulations and Fast Modelling of Urban Winds and Pollutants


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date6 May 2022


Urban air quality is influenced by many complex factors, e.g., urban geometries, inhomogeneous pollutant sources and meteorology. These factors have been widely studied by numerical modelling where idealisation is made. This thesis characterises the extent to which urban air quality is sensitive to the effect of more realistic scenarios.

First, the effect of urban obstacles on pollutant dispersion is studied by field measurements. Previous studies have shown that the pollutant concentration decreases from ground level within street canyons. One may expect lower pollutant concentrations inside elevated walkways, on account of the increased distance from traffic-related emissions. The primary objective is to measure whether pedestrians on the elevated walkway experience lower exposure than those utilizing the sidewalk below. Results suggest that this is not always true.

Second, the effect of source inhomogeneity on pollutant dispersion is investigated by numerical simulations. In real cities, vehicle emissions are a function of space and time. The relationship between traffic volume and air quality is complicated. The measurements indicate that in a busy neighbourhood pollutant concentrations are determined not only by local vehicle emissions, but also by nearby busy road emissions. Computational fluid dynamics (CFD) simulations are conducted in order to clarify how source locations and strength influence the pollutant concentrations over urban-like arrays and realistic urban neighborhoods. Within a regular building array pollutant concentrations are sensitive to the source locations and source strengths, especially for the smaller wind angles, while with the increase of wind angle to \ang{45}, the sensitivity becomes weaker. The effect of source inhomogeneity on pollutant dispersion is also investigated in realistic urban topographies. The pollutant concentrations are highly affected by the nearby main road emissions instead of local emissions, which is consistent with the field measurement conclusion. Additionally, in order to improve urban air quality, one may expect that the air quality would be significantly improved after removing traffic volume from pedestrianized roads. However, simulation results show that the concentration only decreases ~ 26% after the pedestrianization.

Third, fast models are developed to predict urban winds and pollutant distribution. Field measurements cannot isolate multiple influencing factors and are affected by many uncontrolled factors. CFD simulations are computationally expensive. Thus, fast models are developed. A simplified but accurate model is developed for predicting urban wind profiles by exploiting the close connection between the velocity and the vorticity, i.e, vortex method. Mean wind profiles within urban canopies may be predicted by assuming that the velocity is controlled by intense layers of vorticity, i.e., by solving the three-dimensional Poisson equation for a set of discrete vortex sheets. The results show that the vortex method performs well for a wide range of wind directions and urban geometries, with the relative error $\sim$ 30\% within the canopy, and the vortex method is around five orders of magnitude faster than CFD, that is very important for air quality models, especially for the emergency dispersion model. A fast pollutant dispersion model, based on predicted mean wind profiles from vortex method, is developed by solving the time-dependent three-dimensional advection-diffusion equation. The predicted wind profiles induced by the actual building geometry from vortex method perform better than the spatially uniform winds, indicating the importance of predicted mean wind profiles. Since only the predicted wind profiles are used in fast models, it is not expected that scalar dispersion could be accurately predicted. Thus, fluctuation parameterization and effective diffusivity are proposed, representing complementary approaches to the mean flow field. Compared with the Gaussian plume model, this fast model has better agreement with CFD, and normalized mean square error (NMSE) is reduced by ~ 80%.

This work investigates complex influencing factors by considering more realistic scenarios and in more efficient approaches. It can contribute to a better understanding of pollutant dispersion and improve setting more realistic configurations in CFD models, thus providing better suggestions for urban design, traffic control policy formulation and urban environmental assessment.

    Research areas

  • CFD, Urban ventilation, LES, Air quality, Pollutant concentration, Vehicle emissions