Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Civil and Environmental Engineering


Girma T. Bitsuamlak


Aerodynamic drag in the atmospheric boundary layer (ABL) is affected by the structure and density of obstacles (surface roughness) and nature of the terrain (topography). In building codes and standards, average roughness is usually determined somewhat subjectively by examination of aerial photographs. For detailed wind mapping, boundary layer wind tunnel (BLWT) testing is usually recommended. This may not be cost effective for many projects, in which case numerical studies become good alternatives. This thesis examines Computational Fluid Dynamics (CFD) for evaluation of aerodynamic roughness of the built environment and complex terrain.

The present study started from development of an in-house CFD software tailored for ABL simulations. A three-dimensional finite-volume code was developed using flexible polyhedral elements as building blocks. The program is parallelized using MPI to run on clusters of processors so that micro-scale simulations can be conducted quickly. The program can also utilize the power of latest technology in high performance computing, namely GPUs. Various turbulence models including mixing-length, RANS, and LES models are implemented, and their suitability for ABL simulations assessed.

Then the effect of surface roughness alone on wind profiles is assessed using CFD. Cases with various levels of complexity are considered including simplified models with roughness blocks of different arrangement, multiple roughness patches, semi-idealized urban model, and real built environment. Comparison with BLWT data for the first three cases showed good agreement thereby justifying explicit three-dimensional numerical approach. Due to lack of validation data, the real built environment case served only to demonstrate use of CFD for such purposes.

Finally, the effect of topographic features on wind profiles was investigated using CFD. This work extends prior work done by the research team on multiple idealized two-dimensional topographic features to more elaborate three-dimensional simulations. It is found that two-dimensional simulations overestimate speed up over crests of hills and also show larger recirculation zones. The current study also emphasized turbulence characterization behind hills. Finally a real complex terrain case of the well-known Askervein hill was simulated and the results validated against published field observations. In general the results obtained from the current simulations compared well with those reported in literature.