COMPARISON OF DIFFERENT K-EPSILON MODELS FOR INDOOR AIR-FLOW COMPUTATIONS

In this article, five k-epsilon, two-equation models are studied: the standard k-epsilon model, a low-Reynolds-number k-epsilon model, a two-layer k-epsilon model, a two-scale k-epsilon model, and a renormalization group (RNG) k-epsilon model. They are evaluated for their performance in predicting natural convection, forced convection, and mired convection in rooms, as well as an impinging jet flow. Corresponding experimental data from the literature are used for validation. It is found that the prediction of the mean velocity is more accurate than that of the turbulent velocity. These models are neither able to predict anisotropic turbulence correctly nor to pick up the secondary recirculation of indoor air flow; otherwise the performance of the standard k-epsilon model is good. The RNG k-epsilon model is slightly better than tire standard k-epsilon model and is therefore recommended for simulations of indoor air flow. The performance of the other models is not stable.