Evaluation of a Gaussian plume model for urban canopy using CFD data

There are various methods for predicting plume dispersion, e.g. wind tunnel technique, numerical simulation technique by computational fluid dynamics, and Gaussian plume model. It is well known that wind tunnel experiments are a reliable tool for predicting plume dispersion behaviors under local topography and/or building conditions. With the rapid development of computational technology, CFD also has been recognized as a helpful tool. The Gaussian plume model is widely used for regulatory purposes due to its simplicity and low cost. This theoretical model composes of various variables such as source strength, representative wind speed, dispersion parameters, and source height. The traditional Gaussian model is effective for a flat or nearly flat terrain but has limited applicability for densely built-up urban areas. Therefore, many studies on the development and improvement of urban dispersion models have been conducted, using observed data from urban field experiments. McElroy and Pooler (1968) suggested that initial horizontal and vertical plume spreads are 40m due to turbulent mixing motions by urban buildings near the plume source. Hanna et al. (2003) mentioned that initial plume spreads can be approximated by the half of the average building height. Here, urban morphology parameters are generally evaluated over an area of 1km×1km. However, urban surface geometries are highly inhomogeneous and covered with low- and high-rise buildings. This random arrangement of urban buildings induces turbulent flows with a strong three-dimensionality, which significantly influences the transport and dispersion of airborne contaminant. Therefore, important issues remain in appropriately determining the representative wind speed and dispersion parameters within urban canopy. In this study, first we carried out LESs of plume dispersion in various urban areas and investigated canopy wind speed, lateral and vertical plume spreads. Our LES model consists of two computational domains are set up. One is a driver region for generating a target approach flow and the other is a main analysis region for LESs of plume dispersion in urban areas. In the driver region, a neutral atmospheric boundary layer flow is produced by the recycling technique and setting up various roughness obstacles. This turbulent inflow data is imposed at the inlet of the main region at each time step. In the main region, the size of the domain where an urban surface geometry is resolved is 1.5km (streamwise) × 1.0km (spanwise) with the depth of 1.0km. The lateral grid spacing is 4.0m. The vertical grid spacing is stretched from 1.3m to 53m. The study sites are four actual urban areas in Central Tokyo (low-rise buildings area, street-canyon area, complex of high-rise and low-rise buildings area, and high-rise buildings area). Then, we estimated the representative canopy wind speed and plume spreads by sensitive analysis based on the LES data. Our objective is to evaluate the urban-type Gaussian plume model using those representative parameters estimated by sensitive analysis.