Large Eddy Simulations of Urban Turbulent Flows Using a GPU Accelerated Atmospheric Boundary Layer Environment-Lattice Boltzmann Model (ABLE-LBM)

The ABLE-LBM was developed for microscale (1 – 100 m space, seconds-minute in time) resolution large-eddy simulation for atmospheric boundary layer flows with any kind of terrain feature (steep terrain, urban, forest canopy, or combined). In this presentation, we will report urban validation studies of the ABLE-LBM using two laboratory observed data sets. The first set is from PIV (particle imaging velocimetry) observation of flows around two buildings of different heights. The second set is from an MRV (magnetic resonance velocimetry) measurement of an urban model with 49 buildings. The ABLE-LBM produced results similar to both laboratory observation data sets, indicating that the LBM method is a viable simulation method for the microscale atmospheric boundary layer flow simulations. The ABLE-LBM system was recently updated to accelerate the computation using the GPU (graphics processing unit). The model computation speed using an Nvidia GPU card is 300 times faster than a single CPU, meeting the real-time application requirement for a small computation domain (200 x 200 x 100 grid points).