Canopy structure is one of the most important factors that has significant influence on flow pattern in canopy, including aspect ratio, building shape, building orientation, etc.. Since wind flow field is strongly influenced by building configurations and building surface heating in urban area, we investigated systematically the effect of a long street canyon on wind field under five different approaching wind direction (included angles between wind flow and model’s long side are 0° as parallel flow, 22.5°, 45°, 67.5°, and 90° as perpendicular flow), wall surface heating conditions (ground, leeward and windward wall heating), and different section of canyon (inlet, middle and outlet). Wind tunnel experiments were conducted using PIV (Particle image velocimetry), observing both vertically and horizontally. At inlet and middle section of neutral conditions, every direction of flow except parallel flow formed a vortex in the center of canopy. With the decrease of angle, at outlet section, the vortex became weaker until disappeared. For parallel, two parallel counter rotating vortexes were formed. There is a downward flow in the center of canopy, which induces the outside air going inside. In heating cases, a strong buoyancy flow generated and effected flow pattern and air exchange between inside and outside of the canyon.

In the past two decades significant efforts have been devoted to the characterization of turbulence in the roughness sublayer (RSL) over cities, in particular to enhance prediction of wind, dispersion and pollution in the urban canopy layer (UCL) it encompasses. However, due to difficulties in both conducting representative measurements and in performing robust simulations over such complex geometries, the role of dispersive stresses, transport and pressure effects is still inadequately understood. A characterization of such terms would allow to guide/improve simple models such as 1-D urban canopy parameterizations. The physics of UCL and RSL flow and turbulence are here investigated through large-eddy simulations (LES) of the airflow above a realistic urban geometry in the city of Basel, Switzerland, where extensive tower-measurements are available from the Basel Urban Boundary Layer Experiment (2001-02). We rely on a double averaging approach to separate turbulent fluctuations from dispersive terms, in order to determine how spatial variations of time-averaged quantities affect turbulent kinetic energy (TKE). A Lagrangian scale-dependent LES model is adopted to parametrize the subgrid-scale (SGS) stresses and buildings are taken into account adopting an immersed boundary (IBM) approach, with the geometry taken from a highly accurate digital building model. We consider fully developed flow over a 512x512x128m computational domain (mean building height is 15.3m) and adopt a 1m stencil. In agreement to measurements, TKE in the RSL is found to be primarily produced at roof-level. Here, turbulent production overcomes dissipation by SGS stresses and the excess in TKE is dislocated down into the cavities of the UCL (street canyons, backyards) and upwards into higher parts of the RSL by turbulent transport and dispersive transport terms. Turbulent and dispersive transport terms are comparable in magnitude and act as a sink of TKE in the upper RSL and as a source term in the lower RSL and UCL. The spatial heterogeneity of mean velocities and Reynolds stresses in the lower RSL and in the UCL results in a significant wake production rate of TKE. Moreover, pressure transport is found to be a significant source of TKE in the lower UCL, whereas transport by SGS stresses is negligible throughout the RSL.

Wind is important for ventilation and pollutant dispersion in urban areas. We first experimentally studied some 25-row and 15-column aligned cubic or square building arrays (the building width B=72mm and building heights H=B or 2B) in a closed-circuit boundary layer wind tunnel at Laboratory of Ventilation and Air Quality, the University of Gävle, Sweden. The working section of this wind tunnel is 11 m long, 3 m wide and 1.5 m tall. The approaching wind is parallel to the main streets. Effect of building area densities (lp=0.11, 0.25, 0.44 or street width W=B, 2B, 0.5B) and building height variations (H=B or 2B) on flow adjustment (velocity and turbulence profiles by hotwires) and drag forces of individual buildings were first measured in wind tunnel experiments, then the sectional drag coefficient and detailed flow mechanisms in adjustment region, interior region (or fully-developed region) and exit region etc were quantified by computational fluid dynamic (CFD) simulations. The distribution of pressure difference between windward and leeward walls was used to indicate natural ventilation potential in buildings.

We find that wind speed decreases quickly through idealized building arrays due to strong drag produced by buildings. For 25-row urban models with building area density of lp=0.25, the fully-developed region starts at about the 12th building, but not the 4th or 5th one as discussed in the literature. The adjustment length varies due to various building packing densities. For all 25-row urban models with uniform building heights (H=B,lp=0.11, 0.25, 0.44), denser urban models produce lower drag force by individual buildings and attain smaller velocity in urban canopy layers, which produce weaker urban ventilation capacity and less natural ventilation potential through buildings. For 25-row urban models (lp= 0.25, 0.44) with building height variations (H=B or 2B), taller buildings (H=2B) always produce much stronger drag force than lower ones (H=B). Drag force by the latter was small (lp= 0.44) or a little negative (lp= 0.25) because they were significantly sheltered by the former. Vertical profiles of sectional drag coefficient were also analyzed the show natural ventilation potential at different height levels. Results show that, taller buildings usually attain better building natural ventilation potential than lower ones. For urban models with the same building packing density, there is a dilemma that it is difficult to ensure the urban ventilation capacity and building natural ventilation potential at the same time. But ultilizing smaller building packing density can improve both.

The research communities of building physics and wind engineering have steadily focused on the effect of strong wind force caused by an isolated high-rise building on pedestrians for the last decades. In contrast, various researchers of urban climatology have investigated the turbulent boundary layer over urban building arrays for accurate prediction of urban climate. Under these circumstances, this study intends to examine the aerodynamic interaction between the before mentioned two research targets; the wake flow structure observed behind an isolated high-rise slender building and the turbulent boundary layer develops over urban roughness based on a wind tunnel experiment.

The experiment was conducted in a low-speed single-return wind tunnel with a test section of height 1m, width 1.5m and length 8m. The spatial uniformity of the inflow is carefully controlled by means of a honeycomb layer and several mesh screens with an open-area ratio larger than 0.57. To simulate the development of the urban boundary layer, the floor of a wind tunnel with a streamwise length of about 3m is covered by a staggered cubical array with a height of 25mm (hereafter, H), which means the fetch length is 130H, resulting in the boundary- layer height of about 6H (0.15m). In addition, a quarter-elliptic, constant-wedge-angle spire with a height of 32H (0.8m), which is higher than the depth of the boundary layer, is installed in the upwind region of the wind tunnel to generate the wake flow which is analogous to the flow leeward of an isolated slender building.

The detailed distribution of streamwise velocity in lateral¬-vertical planes which cover both within and above the boundary layer developed by an underlying block array is measured in leeward positions of the spire using a hot wire anemometry under a condition of a reference stream velocity of approximately 8ms-1. And the authors investigate how the turbulence characteristics of the wake of the spire are affected by the boundary layer of the block array based on the measurement data. In order to validate the experimental setting, the wake flows by the spire were then compared with the characteristics of two dimensional (2D) wake flow derived theoretically based on the gradient-diffusion model.

The followings were revealed: (1) The streamwise changes of the half width and the velocity deficit are similar to that of the 2D wake flow above the boundary layer generated by both a smooth wall and a block array; however, those within or near the outer edge of boundary layer shows totally different treads from that of 2D wake flow. (2) The spanwise development of the wake from generated by an isolated spire is oppressed in the wall boundary layer.

A wide diversity of urban fabrics exists worldwide, especially in traditional neighborhoods. They are generally linked with typical spatial configurations of buildings. In particular, open, compact and attached forms were developed depending on the local geography, climate and culture. The different levels of porosity that characterize these urban forms involve different very-local micro-climates and internal wind patterns. Indeed, the very-local interactions between mean winds and the shape and layout of urban structures determine to a large extent the flow patterns that develop in the urban canopy layer (UCL). These aerodynamic phenomena affect in turn, among other things, urban ventilation processes, wind and thermal comfort as well as climate at the upper scale of the city.

At the building and street canyon spatial scales, specific air flow structures develop within courtyards and other urban internal outdoor spaces depending on their orientation in relation to the mean wind incidence, as well as openness, i.e. whether these outdoor spaces are partially or totally surrounded by a building or contained inside a building group, or not. In this paper, we thus present the study we carried out to analyze and better understand the aerodynamic processes leading to these flow structures that develop around buildings and inside urban blocks depending on their topological features. Only forced convection processes are considered, and the effects of the horizontal openness of courtyards and internal open spaces of urban blocks are more specifically examined.

The study is based on numerical experiments, which are performed using computational fluid dynamic (CFD) models and the commercial software Ansys Fluent, versions 14.5 and 15. The model was preliminarily validated by comparison with high quality reduced-scale experimental data from wind-tunnel tests (1) as well as detailed numerical predictions (2). The accuracy of steady Reynolds averaged Navier-Stokes (RANS) predictions was evaluated in cases of an isolated and a regular array of rectangular obstacles. Actual simulations were thus performed using the steady RANS Reynolds stress model (RSM). In comparison with e.g. steady RANS k-ε turbulence models, the RSM accounts for anisotropic effects of turbulence on the mean flow.

Simulations were run for different building and urban block generic types, which were specifically designed for that purpose. These morphological types are based on an analysis and abstraction of urban textures that exist in different regions of the world, as well as an identification of the urban morphological factors that affect aerodynamic processes that develop in the UCL.

In cases of isolated building types, results show quite different flow patterns and recirculation phenomena in courts depending on their horizontal openness and orientation in relation to the approach flow incidence. Even clearer differences in flow patterns are observed in the internal open spaces of urban blocks, showing ventilation paths or entrapped recirculation phenomena, which more or less interfere with the surrounding flows. These different flow patterns are characterized by different levels of vorticity and mean velocities, creating higher wind speeds zones or rather sheltered regions.

Providing information on the wind conditions next to buildings and on ventilation processes, such observations linking urban morphological properties to physical phenomena would support a better understanding of the urban heat island phenomenon on larger urban scales, as well as a more integrated and bio-climatic design of buildings and urban areas on smaller scales.

(1) Datasets from the Compilation of experimental data for validation of micro-scale dispersion models (CEDVAL, Hamburg University)

(2) Predictions issued from lattice Boltzmann method – large eddy simulation (LBM-LES, Obrecht et al., 2014)