Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Doctor of Philosophy


Civil and Environmental Engineering

Collaborative Specialization

Scientific Computing


Bitsuamlak, Girma


The lack of Computational Fluid Dynamics (CFD) guidelines for structural wind load evaluation significantly hinders its crucial application in research and practice. The basis for such guidelines is understanding how CFD parameters, such as mesh size, affect the simulated physical input and output parameters, such as velocity field and building aerodynamics. This dissertation addresses this gap by first establishing a relationship in modeling the approach flow and building aerodynamics that culminates into a set of CFD requirements for cladding wind load evaluation. Cladding load on low-rise buildings is selected for two main reasons. Cladding loads on low-rise buildings is critical when dealing with extreme wind hazards, such as hurricanes and tornados, where CFD can be applied. It is also the most error-sensitive compared to global wind loads, which makes it a conservative choice for validation and guideline developments.

The first problem addressed in this dissertation is to evaluate the influence of Large-Eddy Simulation (LES) mesh size and duration on the spectral frequency limits of the modeled velocity field. A relationship between the mesh size in the approach flow region of the LES domain and the resulting cut-off frequency in the modeled velocity spectra is proposed. A method based on a stationarity test is proposed to obtain the optimal duration for converged aerodynamics results. The proposed relationship was used to design the mesh throughout the thesis. The resulting velocity spectra show that the expression can accurately predict the cutoff frequency prior to running the simulation. For the optimal duration, results show that the optimal duration can be significantly shorter than full-scale 1-hour.

Second, the implications of LES filtering and its subsequent amplified viscosity on the modeled effective Reynolds number are investigated. The effective viscosity for separated reattaching flow over a forward-facing step (FFS) is defined as the mean sub-grid scale (SGS) viscosity maximized along the centerline of the separated shear layer (SSL). The FFS is placed in a uniform-smooth and turbulent-sheared inlet at varying levels of effective viscosity and corresponding effective Reynolds number. The Reynolds number effect on the separation bubble size, surface pressure statistics, flow profiles in the separation bubble, and characteristics of the SSL are used to evaluate the proposed definition compared to the conventional definition and vast data from the literature. The results show that the proposed Reynolds number definition based on the SGS viscosity characterizes the LES data more accurately than that defined based on the laminar viscosity.

The third part of the dissertation provides a set of CFD requirements, recommended workflow, and a benchmark validation for cladding load on a low-rise building. The requirements are based on the findings in this dissertation and the existing wind tunnel guideline, ASCE 49-21. New additions that are unique to CFD are also included. The recommended LES workflow is applied to a benchmark validation case for the cladding load on a flat-roof low-rise building. The results are evaluated according to the proposed requirements. The validation results show an excellent agreement with the target wind tunnel data with some explainable differences. The methodologies and results illustrate that, with thorough consideration of the physical requirements, CFD can be utilized to accurately simulate cladding wind loads.

Summary for Lay Audience

Understanding how buildings endure powerful winds, like tornados and hurricanes, is crucial for safety. Computer simulations, called Computational Fluid Dynamics (CFD), are relatively new methods to study wind effects on buildings. The lack of clear guidelines for CFD has hindered their broader application. This dissertation aimed to fill that gap by figuring out how different factors in these simulations affect the accuracy of predictions. The specific focus of this study was on the part of buildings called cladding, which is the general outer layer facing the wind. The type of CFD applied in this dissertation is called Large-Eddy Simulation (LES).

One of the features of LES is to simplify the nature of wind by approximating the wind movement in small imaginary boxes instead of every point in the air. When the approximating boxes get smaller towards the realistic points, the accuracy of the results improves, but the computational demand increases. The first contribution of this dissertation is to formulate a method to predict the accuracy of the simulated wind from the size of the box used before executing the simulation.

Another side effect of the box approximation in LES is the increased viscosity of the simulated air. Viscosity tells us how easily any fluid moves when disturbed. For example, honey does not flow out of a cup as easily as water does because honey has a higher viscosity than water. When the wind effect on a building is simulated in LES, the simulated viscosity becomes higher than the actual wind, which distorts essential features for accurate results. The second contribution of this study is a method to estimate the simulated viscosity and measure the extent of the distortion in the results.

Overall, this research proposes new guidelines for using computer simulations to evaluate wind loads on buildings. These guidelines were then tested on virtual models of low-rise buildings, and the results were compared to data from laboratory tests. The study showed that with the right adjustments, computer simulations can be a reliable way to predict how wind affects buildings during extreme events.

Available for download on Saturday, April 25, 2026