Evaluation of two-eddy viscosity turbulence models to predict airflow in enclosed environments

Abstract

Prediction of different air distribution parameters is required to design healthy and comfortable indoor environments. Computational Fluid Dynamics (CFD) is the most advanced technique to model the indoor airflow. The robustness and reliability of CFD models and their results strongly depend on the experience and judgment of users to properly define several aspects of CFD models. This process is not intuitive, thus new users shall develop skills on CFD modelling. The paper aims to provide more complete information on CFD modelling via evaluating five k-ε and k-ω models for basic natural and mixed convection cases. The main results show CFD results might vary significantly among turbulence models in the boundary zones. Convergence is not always given for granted for natural convection case, and it strongly depends on relaxa-tion factors whose setup is not intuitive. The five evaluated turbulence models perform well predicting air ve-locity and temperature. However, their accuracy depends on the convection case and quantity predicted.