Mean Flow and Turbulence Measurements Around a 3-D Array of Buildings in a Wind Tunnel

In order to predict the dispersion of harmful materials released in or near an urban environment, it is important to first understand the complex flow patterns which result from the interaction of the wind with buildings and, more commonly, clusters of buildings. Recent advances in the application of computational fluid dynamics (CFD) models to such problems have shown great promise, but there is a need for high-quality data with which to evaluate CFD models. This study was performed to fill that need for a limited range of conditions.

High-resolution measurements of the three components of the mean and turbulent velocity statistics were previously obtained around a 2-D array of model buildings in the USEPA meteorological wind tunnel. The current results extend the earlier study to include detailed flow measurements around a 3-D array. Seventy-seven cubical buildings (H=0.15m) were placed in the simulated atmospheric boundary layer of the wind tunnel and arranged in an array 11 wide and 7 deep, separated by one building height. A pulsed-wire anemometer was used to measure mean velocity and turbulence statistics. Measurements extended from 3.3H upstream of the building array to 7.5H downstream.

The flow structure around the 3-D building array was found to be very similar to that measured around the 2-D array in terms of the gross characteristics of the flow field. Flow approaching the 3-D building array decelerates more slowly than the flow approaching the 2-D array, although the depth over which the deceleration takes place is essentially the same (about 1H) in both cases. The upstream reverse flow region is smaller for the 3-D case, extending to about 0.25H upstream and 0.1H in the vertical, as opposed to about 0.5H upstream and 0.25H in the vertical for the 2-D array. At higher elevations (1.5-3H), the flow over the 2-D array clearly accelerates, but little, if any, acceleration is apparent for the 3-D array. Flow over top of the first building of the 3-D array shows separation taking place, albeit rather shallow compared to the 2-D array. Indeed, at 0.1H above the first building in the 3-D array, there is no mean reverse flow, whereas the 2-D array showed a mean reverse flow over the entire top of the building at that height. Velocity profiles in the canyons downstream of the buildings are essentially identical for the 3-D array. The longitudinal velocity component shows reverse flow to a height of about 0.7H, but the reverse flow near the surface is considerably weaker for the 3-D array than for the 2-D array. Velocity profiles downstream of the 3-D array show reverse flow until approximately 1.2H downstream of the last building, whereas reverse flow extended slightly more than 3.5H downstream of the 2-D array. For both arrays, the velocity profiles farther downstream relax toward their undisturbed values and show little difference between the 2-D and 3-D arrays beyond 5.5H downstream.

The data from this study and the 2-D array measurements are contrasted with results from other field and laboratory studies, and some conclusions are drawn regarding the representativeness of the results for the proposed application.