Modeling airflow and contaminant transport in enclosed spaces with advanced models.

摘要

Unstable flows with relatively low mean air velocity and very high turbulence intensity have been found in laboratories, operating rooms, airplane passenger cabins, trains, and buses. This type of flows is an important mechanism of contaminant transport and deposition. The flow is probably created by low Reynolds number jet, thermal buoyancy generated from occupants, and/or separations caused by a large quantity of furniture in such spaces. This type of flows is difficult to model using available turbulence models based on the Navier-Stokes equations. This investigation studied the complex features of such flows, and developed a new turbulence model to predict correctly the unstable flows. The study also applied advanced turbulence simulation techniques such as LES and DES to simulate contaminant transport in enclosed spaces, and obtained better result than using traditional turbulence models.;The investigation first experimentally studied how low velocity and high turbulence were generated. The experiment measured mean velocity and turbulence level in an 8 ft x 8 ft x 8 ft model room, which represents a section of a half of a two-isle airliner cabin. The measurements were for (1) isothermal forced convection from a ceiling jet in the empty model room; (2) isothermal forced convection from a ceiling jet in the model room with a box that represents furniture; and (3) mixed convection from a cool ceiling jet in the model room with a heated box that represents furniture and occupants. The experimental data showed that the inlet jet generated a high turbulence level in the room. The wall had a strong damping effect on the turbulence in the room air. A box that simulated furniture in the room acted as an obstacle and can decrease the velocity and turbulence levels in the room air due to its wall damping effect. The buoyancy generated from the heated box enhanced the air recirculation thus the velocity and turbulence levels.;The second part of this research was to evaluate eight popular turbulence models for such complex flows by the experimental data. The models include six Reynolds-Averaged Navier Stokes (RANS) equation models, one Large Eddy Simulation (LES) model, and one Detached Eddy Simulation (DES) model. By comparing the data of the three experimental cases with the computed results from these turbulence models, this study found that the RNG k-epsilon and RSM models had a good performance among the six RANS models tested, but their accuracy suffered as the flow became more complex. The LES model was the best among all the models tested. The performance of the DES model was stable.;To improve the accuracy of airflow and turbulence prediction, this investigation developed a DES model, the semi-v2f/LES model. By solving the transport equation for k and epsilon, and an algebraic equation for wall normal stress v 2, this model calculates the turbulence viscosity in the RANS region and subgrid-scale turbulence viscosity in the LES region. The semi-v2f/LES model was validated by the experimental data from the three cases and from the literature. The new model performed better than the other models for predicting the turbulence and temperature. The semi-v2f/LES model is recommended for indoor airflows at transitional Reynolds number.;With the improved turbulence models, this research further investigated the performance of contaminant transport models. This investigation used the experimental data from two steady-state cases as well as one transient particle dispersion case in evaluating the performance of five (one steady and four transient) airflow models with the Eulerian and Lagrangian methods. The transient models obtained the mean flow and particle information by averaging them over time. The Eulerian method performed similarly for all five airflow models. The Lagrangian method predicted incorrect particle concentrations with the RANS and Unsteady RANS (URANS) methods, but did well with the LES and DES models. For unsteady-state particle dispersion, the LES or DES models, along with the Lagrangian method, showed the best performance among all the models tested.;To predict particle deposition onto indoor surfaces, this study also developed a Lagrangian particle deposition model with DES. The computer model was validated with experimental data for particle deposition in a cavity with natural convection and with air velocity, air temperature, and particle concentration data from a four-row, twin-aisle cabin mockup. The validation showed that the model performed well for the two cases. Then the model was further used to study particle deposition in the cabin mockup with seven sizes of particles. The particles were assumed to be released from an index passenger due to breathing or talking at zero velocity and due to coughing at a suitable jet velocity. This study can provide quantitative particle deposition distributions for different surfaces and particles removed by cabin ventilation.