Abstract

Air pollution is a major concern in industrialized countries. In dense urban areas, the most common sources of pollutants are the exhaust stacks, ventilators, and cooling towers located on top of buildings. Depending on wind characteristics and flow re-circulations induced by adjacent buildings, effluents can be transported toward fresh air intakes and contaminate indoor air causing health problem to the buildings’ occupants. This particular urban pollution case is known as re-entrainment of pollutants. Unfortunately, the available dispersion models are not adapted to analyse such problems, since they were developed for an isolated building configuration. The present research aims to investigate pollutant aerodynamics and re-entrainment potential for non-isolated building configurations using Computational Fluid Dynamics (CFD) techniques.
To do so, the steady Reynolds-Averaged Navier-Stokes (RANS) approach was evaluated and compared with wind tunnel data and ASHRAE-2011 dispersion model results. The best numerical model possible was defined by performing a sensitivity analysis on the effect of meshing, turbulence model, convergence criteria and turbulent Schmidt number (Sct). For passive scalar transport, it was observed that RANS underestimates dilution when using the standard Sct = 0.7, perhaps due to the inherent incapacity of RANS in reproducing unsteadiness of flow. However, a sensitivity analysis showed that a better agreement is obtained with Sct = 0.3, which is within the range of values suggested in the literature.
Furthermore, a comparative performance evaluation of steady and unsteady approaches was carried out. Three unsteady modelling techniques were compared: unsteady Reynolds-Averaged Navier-Stokes (URANS), Detached Eddy Simulation (DES) and Large Eddy Simulation (LES). The flow pattern within the wake of a two-building configuration was evaluated and dispersion of pollutants compared against wind tunnel data. The influence of meshing size, time step and inlet boundary conditions was discussed. URANS using the Realizable k-ɛ model, fails to reproduce unsteadiness, and dilution values converge to the same RANS results but DES captures well the unsteadiness of the flow. LES dilution predictions are not satisfactory in all locations, perhaps because the mesh used was not sufficiently refined near the walls. It was concluded that under these specified computing conditions, DES showed results closer to experimental data than all other approaches considered.
Finally, RANS was selected to perform a series of simulations for three non-isolated building configurations: a building located upstream of an emitting building, a building located downstream of an emitting building and an emitting building between two tall buildings. After performing a parametric analysis of geometric characteristics of adjacent buildings, a guideline for safe placement of intakes on buildings façades was proposed.
In line with the previous results, this thesis provides three relevant contributions. First, in terms of numerical simulation, the thesis contributes with insights concerning computational simulation for pollutant dispersion in urban areas. Second, additional information in terms of normalized dilution values, contours and streamlines for different building configurations (isolated and non-isolated) is given in order to better comprehend the pollutant dispersion in the urban environment. Third, the thesis offers a guideline with practical recommendations regarding safe placement of intakes to avoid pollutants re-ingestion. These results are also a source of data to code and standard writing bodies.