CFD simulation of non-isothermal mixing ventilation in a generic enclosure: Impact of computational and physical parameters

Ventilation is required to provide a healthy and comfortable indoor environment for building occupants. However, proper design of ventilation strategies requires detailed information about the indoor airflow, which can be obtained using computational fluid dynamics (CFD) simulations. CFD simulations have to be performed with the utmost care, as they are very sensitive to the chosen approach and computational and physical parameters imposed. This paper presents a systematic analysis of the impact of computational and physical parameters on non-isothermal mixing ventilation in an enclosure with a heated floor. The results from CFD simulations are validated with experimental data from literature. Six computational and physical parameters are tested: (1) computational grid resolution, (2) Reynolds-averaged Navier-Stokes (RANS) turbulence models, (3) inlet velocity, (4) inlet turbulent kinetic energy, (5) near-wall treatment, and (6) discretization schemes. It is shown that indoor velocities are more sensitive to the change of computational and physical parameters compared to air temperatures and that none of the five tested RANS turbulence models clearly outperforms the other models for both velocities and temperatures. The largest influence on the results is observed for the prediction of mean velocity using different types of near-wall treatment. In addition, if the inlet velocity is decreased by 25% and 50%, the direction of the recirculation cell changes from clockwise to counter-clockwise.