Section 3 - Results
Section 4 - Previous Work
1. Stathopoulos and Baskaran use the k-ε turbulence model to evaluate the flow regime around a cluster of buildings in downtown Montreal. On completion of the computational simulation a scale model of the control volume was constructed and evaluated under Reynolds similarity assumptions in a wind tunnel. Stathopoulos et al. conclude that "Computed results generally agree with experimental data for most locations."
2. Selvam also investigates the environmental flow around a large building comparing results from two versions of the k-ε computational model to those measured in a wind tunnel and actual experimental reading taken around the structure modelled. Selvam notes "the computed pressures from both turbulence models are almost the same except at the leeward roof.
3. Leuzzi and Monti use computational technique to evaluate the dispersions of a pollutant around a bluff building using a Lagrangian stochastic model(LS). This simulation is considered a good approximation to the release of pollutants from motor vehicles. Leuzzi et al. state that "the agreement between predictions and observations is qualitatively and quantitatively satisfactory".
4. Leitl and Meroney (1997) investigate a similar problem, however the model involves two bluff buildings forming a channel. This simulation reflects accurately the geometry of terraced house. In conclusion the paper states that although the computational models could "simulate the flow field in urban street canyons", discrepancies as high as 90% existed between the simulated and experimental results, independent of which turbulence model was selected.
5. Huser, Nilsen and Skåtun (1997) used a standard k-ε model to predict medium scale flows over complex terrain. The simulation results were compared to experimental results obtained from road planning in Drammen, Norway. In the simulation the inlet is assigned a velocity, temperature and turbulent kinetic energy profile as well as the turbulent dissipation rate (ε). The simulation was run and the normalised eddy viscosity plotted in order to compare computational and experimental results. It was noted that the computational model "over predicted compared to measured values".
6. Theodoridis, Karagiannis and Valougeorgis (2002) investigate the claim that where previously the k-ε model for turbulent diffusion has been used, for very complicated urban structures the Reynolds Stress Model (RSM) should be selected instead. However when comparing the k-ε, Reynold Stress and experimental results, the Reynolds Stress predictions do no show any significant difference.
7. Baskaran and Kashef (1996) simulated single and multiple buildings acting as bluff objects in the incident flow. It was concluded that singular building examples could be fully validated but multiple building validity could not be proved as there was a lack of credible experimental data to allow comparison.
8. Davidson, Snyder, Lawson and Hunt (1996) performed a series of wind tunnel simulations to demonstrate the behaviour of plumes when they encounter groups of bluff obstacles.
9. Berkowicz, Ketzel, Vachon, Louka, Rosant, Mestayer and Sini (2002) performed an analysis of experimental results gathered on and around the Rue de Strasbourg, Nantes, France.
10. Longley, Gallagher, Dorsey, Flynn, Allan, Alfarra, and Inglis (2002) carried out an experiment to collect and model the distribution of particulate sizes within a street canyon. The data collect was performed on Princess Street, Manchester.
11. Coppalle, Delmas and Bobbia (2001) performed analysis of NOx concentrations at ground level within the centre of a metropolitan area. The study included equal weightings of all wind directions. The research noted that for low wind speeds <1m/s it was not always possible to observe an effect on the direction of pollutant transport.