Abstract
Many experiments and numerical simulations have been conducted for decades to investigate the relationship between urban flow characteristics and urban geometries (e.g. Kubota et al. J. Environ. Eng., AIJ. vol. 43, 2008).
Meanwhile, most of the studies in the past assumed the real urban geometry as idealized block arrays; however, real urban surfaces generally include diverse and complicated topographies derived from balconies and penthouses. A few pioneering studies investigated the impact of complex surface topographies on flow fields. For example, Reynolds-averaged Navier-Stokes (RANS) simulation for 2D street canyon with flat eaves carried out by Mohamad et al. (Building and Environment vol. 44, 2008) presented the drastic effects of eaves on temporally averaged flow fields around building models. With this curious investigative result, the authors have conducted a wind tunnel experiment with a time-resolved particle image velocimetry (TR-PIV) system for further investigation about the nature of turbulent flows around 2D street canyons with flat eaves.
In this study, 2D street canyons of four different shapes were employed. A canyon aspect ratio (ratio of the street width to canyon roof height) of 3 and 7 for canyons with eaves and without eaves are used. The images of turbulent flow fields were photographed at a frequency of 1, 000 Hz and a camera magnification of 0.17 mm/pixel. A direct cross-correlation algorithm is employed for image analysis with the combination of a recursive cross-correlation method and an image transforming method (window deformation iterative multigrid, WIDIM). An appropriate setting for the size of an interrogation window is important in the WIDIM method for improving the data accuracy and spatial resolution. Thus, the authors optimized the size of the interrogation window size based on the histograms of the particle displacement rate so that peak locking problems do not arise. In this study, the following four results are obtained.
In a narrow canyon, the vector maps of temporally averaged flow fields indicate that the flow regime inside the canyon was drastically changed by eaves causing a complicated flow field, which contains three vortexes of similar sizes. In contrast, in a wide canyon, the flow regimes of the canyons with eaves have a few points in common around the center with that of the canyons without eaves. However, eddies appear under the upward and downward eaves causing a change in local flow field structures.
The standard deviations of streamwise and vertical turbulent components and the Reynolds stress become small inside a narrow canyon with eaves. However, they become slightly large in the area where the high-speed downward flow directly reaches from above the canyon. In contrast, in a wide canyon with eaves, the standard deviations and Reynolds stress become small in the upstream region while the maximum peaks appear in the downstream region.
The canyon ventilation rate of a narrow canyon with eaves is only 11% when compared to that of a narrow canyon without eaves, whereas, the ventilation rate is maintained at 70% in a wide canyon with eaves. Judging from the 2 point correlation function of the vertical velocity component, in a narrow canyon with eaves, the penetration of a large-scale turbulent structure into the canyon is prevented by the eaves reducing the vertical air mixing. In contrast, in a wide canyon, the impact of eaves on the large-scale turbulent motion is small; thus, the air exchange rate is maintained.
Large-scale upward and downward turbulent motions cause air mixing in the spanwise direction in canyon. The intensity of spanwise air mixing with a large turbulent motion is close to the canyon ventilation in the narrow canyons; however, it is weaker than vertical air mixing in the wide canyons.
