CFD simulation of neutral ABL flows

Abstract

This work is to evaluate the CFD prediction of Atmospheric Boundary Layer flow field over different terrains employing Fluent 6.3 software. How accurate the simulation could achieve depend on following aspects: viscous model, wall functions, agreement of CFD model with inlet wind velocity profile and top boundary condition. Fluent employ wall function roughness modifications based on data from experiments with sand grain roughened pipes and channels, describe wall adjacent zone with Roughness Height (Ks) instead of Roughness Length (z0). In a CFD simulation of ABL flow, the mean wind velocity profile is generally described with either a logarithmic equation by the presence of aerodynamic roughness length z0 or an exponential equation by the presence of exponent. As indicated by some former researchers, the disagreement between wall function model and ABL velocity profile description will result in some undesirable gradient along flow direction. There are some methods to improve the simulation model in literatures, some of them are discussed in this report, but none of those remedial methods are perfect to eliminate the streamwise gradients in mean wind speed and turbulence, as EllipSys3D could do. In this paper, a new near wall treatment function is designed, which, in some degree, can correct the horizontal gradients problem. Based on the corrected model constants and near wall treatment function, a simulation of Askervein Hill is carried out. The wind condition is neutrally stratified ABL and the measurements are best documented until now. Comparison with measured data shows that the CFD model can well predict the velocity field and relative turbulence kinetic energy field. Furthermore, a series of artificial complex terrains are designed, and some of the main simulation results are reported.