Predictions of nanoparticles concentrations at Marylebone Road in London (UK)

The contribution of nanoparticles (referring to those below 300nm in diameter to represent majority of particle number concentrations, PNC) in deteriorating urban air quality has shown to urge attention given that traffic-produced particles can contribute up to 90% of the total PNCs. Nevertheless, their prediction in real urban environment is still challenging given the intrinsic difficulty of simulating complex geometries at high resolution and particle transformation as well as the scarcity of data necessary for validation of numerical models. In this study, we focus on Computational Fluid Dynamics (CFD) modelling for transport and diffusion of nanoparticles in real street canyons using the Marylebone Road as the site for validation. Marylebone Road is a major arterial route (A501) for traffic and pedestrians within the City of Westminster. Central London Roadside buildings create an asymmetric street canyon with a variable aspect ratio (i.e. width-to-height up to 2) surrounded by secondary street canyons with smaller aspect ratios. Specifically we used data available from the UK-DAPPLE project (2002-2006; 2006-2010) and those from the Automatic Urban and Rural Network (AURN) operated by UK-DEFRA. Simulations were made using a Reynolds-Average-Navier-Stokes (RANS) solver with the realizable k-epsilon model for turbulence closure coupled with scalar transport equation to study flow and PNC in the area around the Marylebone Road and Gloucester Place intersection covering an area of 600m × 600m explicitly resolving 29 buildings. An aspect of novelty in our simulations has been the use of a specific turbulence diffusion coefficient for nanoparticles that was included in the calculation of the Schmidt number by user defined functions. Analyses of results have clarified the role of vortical structures stretched and advected along the main street axis as the main mechanism of pollutant removal at pedestrian level. Predicted annual PNCs using average meteorological conditions for the site are in good agreement with annual measured data. Attention will be given to the quantification of errors associated with the various simplification procedures and recommendation will be provided for the use of CFD for PNC predictions in real scenarios.