The flows in urban street canyons represent particularly large-scale cavity geometries. The EU-funded project TRAPOS (Optimisation of Modelling methods for Traffic Pollution in Streets) was established to examine some of the key aspects.

The objective was the improvement of modelling tools:

CFD (street canyon turbulence modelling).

Practical empirical urban air quality models (simplified street geometry and wind flow regimes, modified Gaussian plume dispersion).

The studies examined the effects of:

Street architecture.

Traffic-produced turbulence and heat.

Solar produced building wall heating.

The studies of geometry included examination of the wind flow in nominally 2-D canyons with varying streamwise width to height (W/H) ratio. Here, comparisons were made between wind tunnel data and those obtained from the k-e model CHENSI (developed at Ecole Centrale de Nantes, France). Some of these comparisons are shown below:

The graph below shows a comparison between CFD and Experiment of the (X,Z) location of the centre of the main vortex (for the four main cases studied of W/H = 0.5, 0.7, 1 and 2).

This study showed that for all the cases under consideration the skimming flow regime was observed, indicating that the transition to wake interference flow takes place at an aspect ratio of W/H > 2.0. The agreement between the wind tunnel measurements and the numerical study, using the same inlet boundary conditions was reasonably good, although CHENSI tended to underpredict the magnitudes of the velocities within the canyon. The main flow features, in terms of the large vortex structures, were predicted extremely well. However, the centre of the main vortex was 5%-15% higher and more downstream in the cavity in the predictions than in the experiments, indicating a tighter and less diffuse vortex.

Further work was undertaken to assess the influence of solar heating of the downwind canyon wall on the flow in a canyon with W/H=1 under low wind speed conditions. It was postulated that the buoyancy forces generated by the heating may, at low Froude numbers, induce a significant change in the flow regime from a single to a double vortex structure. This would have implications for the dispersion from the canyon of traffic-produced pollutants.

The results from the wind tunnel study are shown below. Several different conditions of wall heating and approach flow velocity were investigated (different Froude Numbers) as well as the neutral case without wall heating (i.e. infinite Froude number).

The results from the present study indicate that the heating of the windward-facing wall does appear to have some influence on the generation of a very weak secondary flow close to the ground of the canyon at very low Froude numbers. However, so far there is little evidence that the buoyancy forces induce a widespread upward motion, except in a very thin layer near the heated wall, as also noted from field experiments in Nantes, France. Hence, it is not possible to clearly state that the effect of wall heating will be significant in terms of the canyon flow field and the motion and dispersion of pollutants. Further work is planned to examine the possible three-dimensionality of the canyon flow regime with wall heating, together with 3-D simulation, based on the CFD code CHENSI for predicting such flows.