Abstract
A better understanding of urban canopy flows can provide insights into city planning, especially for green buildings integrated with wind turbines or solar panels. This study investigated the use of urban roughness parameterization methods to reduce the computing resource cost of computational fluid dynamics simulations for assessing wind power generation capacity in high density urban cores. Ten potential rooftop wind turbine siting locations in a core area of interest in the city of San Francisco were selected for unidirectional wind potential assessment, to compare the efficacy of simplified approximation models versus traditional, computationally expensive geometrical models. The simplified models replaced the geometry of some groups of upstream buildings from the baseline model with sand-grain roughness wall boundary conditions using two urban roughness parameterization methods, the Davenport classification method, and the Lettau morphological parameterization method. The simplified approximation models were both assessed to work well for urban wind assessment when averaging error between measurement sites. The average normalized velocity magnitude error across all test locations was found to be less than 3.2% for either parameterization method. Whencomparing the two roughness parameterization methods, the Lettau method yielded 19.6% lower errors than the Davenport method. Overall computation time was reduced by up to 50% over the full-scale model when employing the Davenport method and up to 59% in cases that used the Lettau method.