Wind direction effects on urban flows

In the research on urban flows, two trends can be distinguished: some studies based on real urban environments, either in the field or on reproductions; and some on simplified models. One such model, a staggered cube array, has been investigated extensively, both experimentally in wind tunnels and numerically using LES and DNS, but always with a flow direction normal to the faces of the cubes. We are here investigating the effects of different wind directions on the flow within and above the cube array. As no other study has been published on the subject, both numerical and experimental investigation methods have been used. For the numerical side, large-eddy simulations were conducted using the finite volume method implemented in commercial software (StarCD, CD-adapco). The half-channel flow simulated was driven by a momentum source equiva- lent to a pressure gradient. The results are shown to agree well with those of wind-tunnel experiments conducted in parallel. For the experimental side, the boundary layer flow is measured in two dimensions using particle-image velocimetry and the drag using pressure tapped cubes and a drag balance. The results show that the mean flow field within the canopy is only weakly dependent on the direction of the flow above. Above the canopy, the mean velocity profile, time av- eraged and spatially averaged in horizontal layers, is shown to include an inertial sublayer where a logarithmic profile can be fitted. The fit proves to be best by considering for zero-plane displacement the height at which the mean drag acts and points arguably in the direction of a flow dependent value of the Von Karman constant. At the top of the numerical domain, the mean flow is also deviated from the direction of the forcing. We show that this deviation is directly related to previously unseen lift forces applied on the cubes. On balancing the different forces in the numerical domain, the viscous contribu- tions to both the lift and the drag forces are shown to be larger than anticipated from previous studies. Although these results could not be validated by the experiments as the viscous forces remain small compared to the experimental errors, such contributions raise concerns with regards to the Reynolds number independence of the flow and to the need for more complex wall conditions for LES. Overall, the variations of the flow pa- rameters are not negligible and prove to be non-linear and non-monotonic with the flow angle which highlights the importance of considering the flow direction when studying or parameterising urban landscapes. 1.EThOS - Electronic Theses Online ServiceGBUnited Kingdo