Comparison of AERMOD, CFD-RANS and Wind Tunnel Dispersion for Simple Building Configurations

摘要

AERMOD is a Gaussian plume dispersion model and is the EPA preferred model for regulatory applications. The model includes building downwash algorithms and is therefore widely used for building configurations ranging from simple to complex. AERMOD was designed based on EPA wind tunnel data collected for simple rectangular buildings. Past work has shown that AERMOD under- and overpredicts contaminant concentrations for building configurations that it was not designed for, such as very wide buildings, short buildings with large footprints and rotated buildings that cause increased plume downwash due to corner vortices. At the other end of the spectrum, Computational Fluid Dynamics (CFD) models are used with increased frequency for complex dispersion applications. However, detailed validation studies are rare. Processor speeds are increasing, and RANS and LES models can be run on a laptop computers. While numerous CFD software packages are available for anyone to purchase online, methods and guidance on how to accurately simulate atmospheric dispersion is usually lacking from the software literature. This study compares dispersion model performance for three simple building configurations upwind of a single tall stack. Ground-level concentration predictions from AERMOD and a RANS-CFD model coupled with a Lagrangian stochastic (LS) particle model were evaluated and compared to a data from a wind tunnel experiment. The dispersion models are simple and straightforward for this study, so sensitivity to specific building shapes is obvious. The same approach wind profile and neutral stability is used in all models. This paper assesses building downwash characteristics by evaluating ground-level concentration distributions downwind of a stack located adjacent to different building shapes. Our goal is to evaluate AERMOD performance for building configurations that it was not designed for, and to determine if a RANS-CFD model can produce the dispersion effects of corner vortices. AERMOD shows significant sensitivity to the selection of meteorological conditions and the downwind concentrations predicted by AERMOD are very similar for all building shapes evaluated. The RANS-CFD model seems to capture the flow field around the rectangular buildings well and the concentrations predicted by the RANS-CFD driven Lagrangian stochastic particle model are comparable to the wind tunnel simulations for the rectangular buildings. The CFD model was unable to produce the dispersion effects of comer vortices for the rotated building.