Computational modeling of wind turbine wake interactions
Date
2012
Authors
Davis, Cole J., author
Venayagamoorthy, S. Karan, advisor
Heyliger, Paul R., advisor
Maloney, Eric D., committee member
Journal Title
Journal ISSN
Volume Title
Abstract
The rapid expansion of the wind energy market necessitates the need for advanced computational modeling and understanding of wind turbine aerodynamics and wake interactions. The following thesis work looks to study turbulence closure methods widely used in computational fluid dynamics (CFD) and their applicability for modeling wind turbine aerodynamics. The first investigation is a parametric study of turbulence models and their performance on geometries of stationary in-line turbines and disks spaced at different intervals. A variety of Reynolds-averaged Navier-Stokes (RANS) closure schemes (Spalart-Allmaras, Standard k-ε, k-ε Realizable, k-ε RNG, Standard k-ω, k-ω SST) were studied as well as a large eddy simulation (LES) with a dynamic Smagorinsky-Lilly sub-grid scale (SGS) model. The simulations showed the grid refinement to be inadequate for LES studies. The RANS closure schemes did not indicate a dominant model. However, relevant literature on separating flows has shown the k-ω SST model to be preeminent. The second investigation uses only the k-ω SST RANS closure scheme to model wake development and resolution for both a single fully resolved rotating turbine as well as two in-line fully resolved rotating turbines. These simulations were successful in predicting wake development and resolution, as well as predicting velocity deficits experienced by the downstream turbine. Vorticity results also showed an accurate wake structure and helical tendencies. In the third investigation, a grid independence study was performed to gain an accurate pressure distribution on the blade surfaces for a separate, collaborative, non-linear, structural study of wind turbine blades. This study showed a strong asymptotic relationship of the maximum pressure on the blades to the predicted Bernoulli pressure on the blade. The results of this research show clear wake development, structure and resolution. The velocity deficits found translate directly in to power deficits for downstream turbines and the vorticity translates directly into increased fatigue experienced by the blades. In contrast to the vast super-computer simulations found in literature, all simulations in this thesis work were calculated using four parallel processors. The accuracy was achieved through assumptions, which were designed to maintain the desired physics while simplifying the complexity of the problem to the capabilities of desktop computing. This research demonstrates the significance of model design and capabilities and accuracy achievable using desktop computing power. This has vast implications of accessibility into academia and the further development of the wind power industry.
Description
Rights Access
Subject
CFD
wind power
wake interaction
RANS