Abstract
Estimating the wind velocity field in an urban area is important for pedestrian comfort and safety, as the local wind velocities dictate the transport of heat and air pollution in urban environments. Therefore, it is an essential requirement to assess wind patterns when designing urban areas. Computational Fluid Dynamics (CFD) numerical solvers are usually employed to estimate the wind comfort in an urban area under different wind intensities and directions. However, CFD simulations are expensive in terms of time, especially when many scenarios are addressed to ensure safety and comfort for multiple conditions. Here, a CFD framework based on an immersed boundary approach to discretize the urban topography is developed and validated against experimental data in a wind tunnel and two different standard body-fitted mesh codes. The new solver employs a structured cartesian octree grid automatically generated from Lidar data of urban topographies. The advantages are eliminating the complex and time-consuming pre-processing of urban topographies and making the framework accessible to urban planners without CFD expertise. Furthermore, the code is equipped with GPU parallelization that further reduces the computational time. Provided that the best practice guidelines for urban simulations are satisfied, in particular at least 10 cells between two buildings, the code shows very good agreement in all the tests comprising of a simplified and real urban neighborhood highlighting the importance of accurately solving the complex terrain topography of an urban region when non-negligible elevation changes of the order of 20/40 meters are present.