This work aims to construct reliable low-dimensional ventilation models with aid of CFD simulation data. Once constructed, these models can predict indoor pollutant concentration distributions very fast and efficiently, further facilitating on-line monitoring and control of ventilation systems. In this work, ANSYS FLUENT and an open-source CFD package OpenFOAM are employed as CFD simulation tools. For CFD simulations, the Reynolds-averaged Navier¨CStokes (RANS) approach is used. In a first step, we investigate the accuracy of RANS modelling for indoor ventilation at transitional slot-Reynolds numbers, with focus on turbulent inlet boundary conditions on indoor air-flow characteristics and pollutant dispersion. A simple benchmark ventilation case under constant-density conditions is considered. Two turbulence closure models are included in the study, i.e., a low-Re k-¦Å model, and the SST k-¦Ø model. When looking at velocity fields, we find that the influence of turbulent length scales at the inlet boundary on the indoor flow field is small. The influence of turbulence intensity (ranging between 2% and 30%) is considerably larger, in particular affecting the separation point of the inlet jet along the top wall. When further investigating the effect of turbulent conditions at the inlet on pollutant dispersion, we find that variations of inlet turbulent length scales lead to differences in pollutant concentration of up to 20%. Variations due to changes in inlet turbulent intensity lead to differences up to a factor 2. These findings strongly emphasize the importance of imposing realistic boundary conditions for turbulence models. In a second step, we investigate the development of linear low-dimensional ventilation models for stationary pollutant source conditions. The RANS simulation models investigated earlier, serve as a benchmark testing ground. We employ a discrete Green¡¯s function approach to derive a linear low-dimensional ventilation model directly from the governing equations for indoor ventilation (i.e. the Navier¨CStokes equations supplemented with a transport equation for indoor-pollutant concentration). It is shown that the flow equations decouple from the concentration equation when the ratio ¦Á of air mass-flow rate m ?_a to pollutant mass-flow rate m ?_p increases to infinity. A low-dimensional discrete representation of the Green¡¯s function of the concentration equation can then be constructed, either based on numerical simulations or experiments. This serves as a linear model which allows for the reconstruction of concentration fields resulting from any type of pollutant source distribution. A ventilation case under constant-density conditions is again considered. Discrete linear ventilation models for the concentration are then derived and compared to coupled RANS simulations. An analysis of errors in the discrete linear model is then presented. In a next step, the validity and applicability of linear ventilation models for heavy-gas dispersion (density/buoyancy effects) is investigated. The effect of buoyancy forcehas been taken into account in turbulent production term to obtain correct diffusion behaviour. A low-Re k-¦Å model is employed and the generalised gradient diffusion hypothesis is used for buoyancy source term. It is concluded that the flow equations decouple from the concentration equation when the mass flux ratio ¦Á is 10 times higher than the range without density differences ¦Á>10
展开▼
