Large eddy simulation of a cross-ventilated building: validation and impact of inflow generators

Abstract eng:
In the present paper the cross-ventilation of an isolated building immersed in a neutral atmospheric boundary layer (ABL) with openings on the windward and leeward facades is investigated by means of numerical simulations at wind tunnel scale. Specifically, the large eddy simulation (LES) approach is employed, with a wall shear stress model at the bottom of the computational domain in order to take into account ground roughness. Three different inflow methods, i.e. a precursor method, the vortex method and the spectral synthesizer, are employed to generate the turbulent velocity profiles at the inlet of the computational domain. The numerical simulations are validated through wind tunnel test data available in literature and a statistical analysis is performed using common validation metrics. Results show that the choice of the inflow generator has a qualitative and quantitative influence on the predicted flow field and time-averaged concentration. Specifically, the precursor method and the vortex method provide similar results in terms of mean streamwise velocity, whereas the spectral synthesizer differs from the other two inflow generators in terms of predicted flow features (most notably a higher inclination of the ventilation jet). Slightly larger differences are observed for turbulence kinetic energy and time-averaged concentration, which are most accurately predicted by the precursor method. Nevertheless, despite the higher computational cost (+33%) of the precursor method, the performance difference with especially the vortex method is limited. In addition, both precursor method and vortex method outperform the spectral synthesizer for both turbulence kinetic energy and time-averaged concentration in most metrics. In conclusion, the vortex method can be considered the most costeffective among the inflow generators employed for the case examined.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket