Estimating the time evolution of urban pollutant concentrations using offline CFD

Abstract

${\mathrm{NO}}_2$, ${\mathrm{NO}}_x$ and ${\mathrm{PM}}_{2.5}$ concentrations within two densely built-up areas are predicted using offline steady-state CFD simulations by adapting the WA CFD-RANS methodology introduced by Parra et al. 2010. Statistical metrics for the pollutants indicate good agreement with a two-week measurement campaign and air quality monitoring data for 2021. The performance is assessed with respect to the assumptions required for pollutant concentrations to be estimated from RANS simulations for different inflow wind directions. It is shown that these assumptions are satisfied by many different configurations over a wide range of meteorological conditions since background concentrations play an important role, the turbulence time scale is short, the impact of wind-direction fluctuations is limited, and the flow is fully developed. Extensive sensitivity testing confirms that good agreement is still obtained for a small set of offline simulations, averaging intervals much shorter or longer than one hour, different definitions of inflow and in situ velocity scales, and substitution of LES for RANS. This work elucidates the conditions under which urban pollutant concentrations can be estimated from offline RANS and demonstrates that the methodology’s applicability may extend beyond typical air quality applications.