Skip to main content
eScholarship
Open Access Publications from the University of California

Large eddy simulation of atmospheric boundary layer flow in urban terrain : implications for transport of pollution and heat

Abstract

A three-dimensional immersed boundary method was implemented into a Large Eddy Simulation (LES) with advanced subgrid-scale modeling capability. In this way, obstacles in the urban atmospheric boundary layer such as buildings and hills could be represented without changing the Cartesian grid. These numerical methods are applied in two urban environment investigations. The first explores the effect of hilly urban morphology on dispersion characteristics in the urban boundary layer. The second investigate the application of wall functions for building convection heat transfer in large eddy simulation. Air flow and dispersion in urban areas are strongly affected by the presence of buildings. In natural settings hills strongly impact dispersion. Five simulations of flow over building arrays over flat terrain and witch of Agnesi hills with maximum slope of 0.26 were conducted to study turbulence and dispersion properties in and above the canopy. While the small hill reduces the shear stress and velocity variance above the urban canopy compared to the flat urban array, the shear stress increases for larger hills. The TKE in the canopy downwind of the hill decreased below the value for the flat urban case, but canopy ventilation for the hilly cases was several times larger than for the flat case, especially near the hill crest. Therefore, urban dispersion models should account for these relatively moderate terrain changes to produce accurate results. In urban energy balance models, convection heat transfer model is often over-simplified by using a uniform convection heat transfer coefficient (CHTC) for each building surface. We consider more complex flow patterns by implementing a wall function to calculate the local CHTC from local velocities provided by LES. Simulations consisting of single building, 3 x 3 building arrays and 6 x 6 building arrays with neutral and unstable conditions were performed. Validation showed similar results as a low Reynolds number simulation resolving the viscous region, but both simulations disagreed with measurements in a wind tunnel. The log-law relation, which is a fundamental assumption underlying many wall models, was found to be accurate for the roof surface velocity and temperature for high building density, but it does not apply to windward and leeward surfaces. Density of buildings also acts as one of most important factors in determining the temperature distribution and buoyancy force in the urban canyon and roughness layer

Main Content
Current View