The performance of different unsteady Reynolds-averaged Navier-Stokes models mined by Fuzzy C-means algorithm with Fourier coefficient-based distance on the constructed vortex structures around a single building

Abstract

The Fuzzy C-means (FCM) method with Fourier coefficient-based distance was firstly used to construct the vortex features and analyze the large-scale unsteady flow pattern around the building by the transient ensemble velocity magnitude and Q criterion value. The performance of the unsteady Reynolds-averaged Navier-Stokes (URANS) turbulence models, including Standard$k - \varepsilon $model, Realizable$k - \varepsilon $, Renormalization Group (RNG) theory$k - \varepsilon $and Shear-Stress Transport (SST) $k - \omega $models on the flow structures around a high-rise building, was examined and compared by data analysis. The transient data was analyzed to identify the root of the different flow structures. The vortex structure was used to distinguish the different turbulence simulation performance directly. The result indicated that the RNG$k - \varepsilon $model shown the best simulation results on the mean velocity and total turbulent kinetic energy fields. The RNG $k - \varepsilon $ model could simulate the modest scales of vortices which approximates the experiment data around the building, while the SST$k - \omega $model could simulate the abundant scales of vortices which led larger turbulence viscosity coefficient. In addition, the nature pressure difference ventilation partition method was proposed and verified in simulating the nature pressure difference around a single building.

Keywords

• vortex
• q criterion
• transient flow characteristic
• unsteady rans turbulence model
• nature pressure difference