High performance simulation of air pollution, 1999

Air pollution is of major concern in urban cities and industrialized areas. The adverse effects air pollution has on humans, the ecosystem and the environment warrant action to help prevent it. Knowledge about contaminant dispersion near pollutant sources is critical in preventing toxic contaminant levels. This is because contaminant dispersion is inversely related to air toxicity. Thus, pollution levels generally decrease as dispersion rates increase. The aim of this research is to use fast and robust computational fluid dynamics tools to simulate contaminant dispersion of stationary and mobile pollutant sources. Using high performance computational (HPC) methods, computer animated simulations were generated to visually predict variations in contaminant dispersions as other effective factors such as wind speed and emission rate are varied. Such simulations will allow the eventual prediction of air pollution levels near pollutant sources. Two simulations were developed in this research. The first was a stationary source simulation involving contaminants emitted from an imaginary smokestack. The same principles were then applied to a mobile source simulation of contaminants emitted from automotive vehicles traveling on roads in the downtown area of Atlanta, Georgia. The simulations were carried out in three main steps: mesh generation, computation and visualization. The computation was based on finite element methods developed and implemented on a parallel supercomputer by Dr. Shahrouz Aliabadi. The methods used in this research to create HPC simulations of contaminant dispersion were successful. Geometric models of pollutant source areas were produced, and surface and 3D volumetric meshes were generated. Simulations under physical atmospheric conditions yielded flows that were reasonably possible. Furthermore, the effects altering the magnitude of these physical parameters in the system yielded appreciable dispersion deviations consistent with what is logically expected. The simulations may be altered or expanded to give users approximations of pollution levels in a region based on real geographical and atmospheric data