Subgrid-scale modeling of concentration variance in large-eddy simulation for reactive pollutant dispersion

Air pollution is a matter of concern in densely developed urban areas. Air quality issues with the photochemical oxidant, particle matters, and other pollutants are increasing. Many scientists and engineers have been trying to investigate and predict the flow and pollutant dispersion in complex urban configurations, and they have been employing the computational fluid dynamics (CFD) technique as a powerful tool for their research works.

The recent increase of computing power is making it possible to conduct CFD simulations of phenomena of complex mixture of physical and chemical processes with high spatial and temporal resolutions. The large-eddy simulation (LES) therefore has become a very useful approach to analyze dispersion problems within complex urban flow fields. In addition, there have been recent attempts to extend LES to the transport of reactive pollutants. Because photochemical oxidant or ozone, which is one of major pollutants in urban areas, is generated and dissipates in the atmosphere, LES is a key technique to develop an advanced predicting system for the air pollution in urban areas.

However, LES has a limitation in terms of accuracy in the evaluation of chemical reaction rate of pollutants. Although LES can simulate detailed processes of dispersion and mixture of reactants in the atmospheric turbulence which is solved on computational grid coordinates (grid scale, GS), variance at small scales (subgrid scales, SGS) are not computed explicitly owing to the filtering operation of LES. Reaction rate depends on not only the property of reaction itself but also mixing rate of reactants in turbulence. The elimination of SGS variance of concentration therefore leads to under/overestimation of reaction rates.

Following these back ground, in this study, we introduce a method to analyze SGS variance of reactants concentration in LES, and validate the accuracy of the method comparing results of LES with experimental data. In our method, to evaluate the SGS variance of the concentration, a transport equation of the SGS variance is computed with the transport of GS concentration.

In this study, we first conducted an experiment of dispersion of a passive scalar in a test chamber which modeled an urban street canyon of aspect ratio of unity. A tracer gas was released from a line source on the center bottom of the chamber, and mean and fluctuation intensity of the concentration in the chamber were measured using a fast ionization detector. Secondly, we performed large-eddy simulations of the passive scalar dispersion in the test chamber using the transport equation of the SGS variance. The results found that the concentration fluctuation at the GS in the LES became as small as half of the full-scale fluctuation in the experiment in the downwind portion of the line source. However, the fluctuation at the SGS evaluated from the transport equation made it possible to complement the underestimated intensity to agree well with the experiment's results. This result showed that the method had enough accuracy for the evaluation of the SGS variance of pollutant concentration for the application to the dispersion of pollutants considering the chemical reaction in the atmospheric turbulence.