New turbulence models for indoor airflow simulation

Abstract

It is important to predict indoor environment in order to design thermally comfortable and healthy indoor spaces. Heating, Ventilating and Air Conditioning (HVAC) design engineers and architects widely use the Computational Fluid Dynamics (CFD) technique for indoor environment predictions. The CFD technique requires a turbulence model to correctly calculate indoor air distribution. However, the currently available turbulence models in the literature are either inaccurate or inefficient for the indoor environment predictions. To solve the problem, this thesis proposes two two-layer turbulence models and a zero-equation turbulence model. The two-layer models use a one-equation (k) model for the near wall region and the "standard" k -£ model in the outer region. The zero-equation model calculates turbulent viscosity based on local velocity and a length-scale. The near wall models have been developed with the aid of the data of natural and forced convection flows by Direct Numerical Simulation (DNS), while the zero-equation model has been proposed empirically. One of the two-layer turbulence models is used for predicting natural convection in rooms. The other two-layer model and the zero-equation model can be used to predict forced, natural, and mixed convection in rooms. These three new models have been applied to predict different types of indoor airflows. The corresponding DNS or experimental data were used to validate the models. This study concludes that the two-layer models can predict airflows most accurately, better than many k -E models. The computing cost is significantly lower than that of the low Reynolds number k-E models and is only slightly higher than that of the "standard" k-E models. The zero-equation model is at least ten times faster than the "standard" k-E model and it is numerically stable and can predict indoor airflow with acceptable accuracy.