Simulating turbulent flows in a city of many thousands of buildings using general high-resolution microscopic simulations requires a grid number that is beyond present computer resources. We thus regard a city as porous media and divide the whole hybrid domain into a porous city region and a clear fluid region, which are represented by a macroscopic k- [graphic removed] model. Some microscopic information is neglected by the volume-averaging technique in the porous city to reduce the calculation load. A single domain approach is used to account for the interface conditions. We investigated the turbulent airflow through aligned cube arrays (with 7, 14 or 21 rows). The building height H, the street width W, and the building width B are the same (0.15 m), and the fraction of the volume occupied by fluid (i.e. the porosity) is 0.75; the approaching flow is parallel to the main streets. There are both microscopic and macroscopic simulations, with microscopic simulations being well validated by experimental data. We analysed microscopic wind conditions and the ventilation capacity in such cube arrays, and then calculated macroscopic time-averaged properties to provide a comparison for macroscopic simulations. We found that the macroscopic k- [graphic removed] turbulence model predicted the macroscopic flow reduction through porous cube clusters relatively well, but under-predicted the macroscopic turbulent kinetic energy (TKE) near the windward edge of the porous region. For a sufficiently long porous cube array, macroscopic flow quantities maintain constant conditions in a fully developed region.
展开▼
