Modeling of airflow and contaminant transport in enclosed environments

Zhao Zhang, Purdue University

Abstract

Outbursts of airborne infectious diseases directly cause death and illness, while air pollution strongly associates with people's mortality and morbidity in the long term. Since modern people spend most of their life time in enclosed environments, good understanding and accurate prediction of air distribution and people's exposure to air contaminants is crucial to our safety and health. This dissertation advances the modeling techniques in predicting airflow and contaminant transport in enclosed spaces. This study first surveyed the recent development of turbulence modeling in computational fluid dynamics (CFD) and the applications in predicting indoor airflows. The survey identified eight turbulence models potentially suitable for enclosed environmental studies. These models can be categorized into Reynolds averaged Navier-Stokes (RANS), hybrid RANS and large eddy simulation (or detached eddy simulation, DES), and large eddy simulation (LES). We tested and evaluated their modeling performance against experimental data in four interior airflow scenarios. It was shown that the LES has reasonable accuracy while its computing cost is much higher than RANS models. Among the RANS models, the RNG k-ϵ and a modified [special characters omitted] model have the best overall performance over the cases studied. With the turbulence models, an Eulerian method was used and a Lagrangian method developed to predict the transport and distribution of gaseous and particulate contaminants in enclosed environments. Both methods were validated by comparing with experimental data in various ventilated rooms and in an air-conditioned aircraft cabin. Our results showed that both the methods can predict particle distributions with acceptable accuracy and the corresponding computing time required is affordable. The Lagrangian method is better than the Eulerian method for transient particle transport. To further explore the capability of Lagrangian method, we developed a particle-eddy interaction model to simulate particle depositions onto indoor surfaces. By using this model, we substantially improved the modeling reliability of predicting particle deposition onto indoor surfaces compared with other commonly used particle models in the literature. By validating the method with experimental data, we concluded that the Lagrangian method with the new interaction model can predict indoor particle deposition with reasonable accuracy provided the near-wall turbulence is correctly modeled.

Degree

Ph.D.

Advisors

Chen, Purdue University.

Subject Area

Civil engineering|Mechanical engineering

COinS