Influence of inlet boundary conditions on 3D steady RANS simulations of non-isothermal mechanical ventilation in a generic closure

Inlet boundary conditions at the supply opening play an important role in the accuracy and reliability of computational fluid dynamics (CFD) simulations for indoor airflow. A non-exhaustive overview of past CFD simulations of ventilation flow in generic enclosures indicates that uniform hypotheses of inlet boundary conditions are commonly used. However, due to thermal effects and a complex air supply system geometry, constant values of inlet airflow quantities can be insufficient for an accurate simulation of non-isothermal ventilation flow. In addition, this can lead to biased conclusions on the performance of turbulence models and other computational settings. To assess this issue, a well-documented experiment on non-isothermal mechanical ventilation with detailed information of inlet conditions from the literature is used in this study. Different turbulence models and inlet boundary conditions are validated, including seven steady Reynolds-averaged Navier-Stokes (RANS) models, six methods of imposing inlet air velocity, three methods of specifying air temperature, and a wide range of turbulence quantities. The parametric study demonstrates that in this particular case, i.e., cold air supply from a single round nozzle diffuser close to the ceiling, ω-based turbulence models perform better than the ε-based turbulence models. The impacts of different methods to specify inlet air temperature profile and turbulence quantities on the simulation results are not significant. However, if the inlet airflow direction is not taken into account appropriately, significant deviations are observed between CFD simulations and experimental data.