On possibilities to estimate local concentration variations with CFD-LES in real urban environments

Applied studies with Large Eddy Simulation (LES) of hazardous gas dispersion around buildings in cities have become increasingly feasible due to rapid advancements in computing technology. However, there is little extant literature investigating how each model's results compare with others, as well as their ability to predict near-field dispersion in a real city. In this study, three typical LES sub-grid-scale models are used to simulate gas dispersion, utilizing alternatively constant values and synthetic turbulence at inflow boundaries. The results are compared with data from the Joint Urban 2003 Atmospheric Dispersion Study in Oklahoma City. Flow and turbulence statistics of the simulation is presented at two probe locations, one inside the city-core and one outside. In addition, comparisons with the measured mean concentration and maximum concentration values are conducted. It was found that in the core of the city, simulated turbulence is mainly determined by buildings and their configurations, and is only weakly affected by model type and assumed turbulence at the inflow boundaries. On the other hand, outside and upwind the city center the turbulence set at the inflow boundaries is very important if realistic turbulence statistics is to be achieved. Downstream of the source, all tested models produce similar predictions of maximum concentration values, which in turn are similar to the experimental data. Thus, the results indicate that it could be better to use the LES calculated maximum-concentration instead of the calculated mean-concentration when developing methods for hazard area estimation.