Fine Scale Modeling of Urban Structures: Dependence on Grid and Model Features

Computational fluid dynamics (CFD) models are being increasingly used to predict fine scale features of dispersion in an urban environment. The accuracy of such modeling is highly dependent on grid resolution, the subgrid scale model used, and the characterization of the boundary layer flow. The goal of this study is to document the degradation of the modeled features as the fidelity of the modeling choices are gradually relaxed. In this way, the tradeoffs involved between capturing the details of the flow versus using a resolution that enables modeling larger domains is evaluated.

Different levels of CFD modeling are applied to flow about a cubical structure at a Reynolds Number typical of atmospheric flows. A very high resolution grid with fully turbulent subgrid models and a Detached Eddy Simulation (DES) is the base case. This base case is compared to measured data and both the time averaged and unsteady components of the flow carefully characterized. Then the grid and modeling features are systematically relaxed and the characteristics of the modeled features are documented.