In this study the simulation of street canyon flow was investigated. Since the same numerical method was used and the solutions were grid-independent, it can be concluded that the main differences between the simulations were caused by the turbulence modeling. Unfortunately none of the used turbulence closure schemes led to a fully satisfying result demonstrating that predicting the flow in street canyons is a challenging problem for RANS calculations. The k-e model showed a qualitatively reasonable solution, but was quantitatively relative poor in comparison with the LDV measurements. Counterintuitively, the more advanced SST and SSG models did not produce better results. The rooftop recirculation zone was over-estimated severely, causing a qualitatively different flow field inside the canyon. As a consequence, one should be very careful in interpreting the results of a RANS calculation for this kind of flow. The variation of solutions between the different turbulence models is so large that there is a very high uncertainty factor. In conclusion, we recommend that within a series of RANS parameter studies, ke simuations, if possible, are either validated by means of wind tunnel experiments or LES. Alternatively, to strike a balance between inexpensive RANS calculations and high-fidelity but costly LES, it might be worthwhile to employ hybrid LES/RANS methods. Such methods have shown great promise for the simulation of complex turbulent flows, see the review of Fröhlich and von Terzi [2008]. Furthermore, some of these methods are already available in commercial CFD software.
Scrutinizing turbulence closure schemes for predicting the flow in street canyons
2006
8 Seiten, 8 Bilder, 1 Tabelle, 8 Quellen
Aufsatz (Konferenz)
Englisch
Prediction of traffic noise shielding by city street canyons
TIBKAT | 2004
|Modeling traffic-related air pollution in street canyons of Beijing
Tema Archiv | 2000
|Online Contents | 1996