A Computational Analysis of Wind Turbine and Wind Farm Aerodynamics with a Focus on Dual Rotor Wind Turbines
Is Version Of
This dissertation serves to summarize my research into wind farm and wind turbine aerodynamics.
Included in this thesis is a summary of the methods I use as well as the four research problems that I investigated.
Motivation is provided for my research as well as an overview of the computational
methods that I use. These methods include analytical methods such as blade element
momentum (BEM) theory and the vortex lattice method as well as computational fluid dynamic methods like
the Reynolds averaged Navier-Stokes (RANS) equations and large eddy simulation (LES). These methods are used
to investigate wind turbine and wind farm aerodynamics. In particular, I use these methods to confront the
various forms of loss that wind turbines and wind farms experience. They include the losses that individual turbines
experience due to swirl, induction, and viscosity as well as the loss that wind farms experience due to turbine-wake interaction.
Horizontal axis wind turbines (HAWTs) suffer
from aerodynamic ineffciencies in the blade root region (near the hub) due to several non-aerodynamic
constraints. Aerodynamic interactions between turbines in a wind farm also lead to signifcant loss of wind
farm efficiency. A new dual-rotor wind turbine (DRWT) concept is proposed that aims at mitigating these two
losses. A DRWT is designed that uses an existing turbine rotor for the main rotor, while the secondary rotor
is designed using a high lift-to-drag ratio airfoil. Reynolds Averaged Navier-Stokes computational fluid
dynamics simulations are used to optimize the design. Large eddy simulations confirm the increase energy
capture potential of the DRWT. Wake comparisons however do not show enhanced entrainment of axial momentum.
I extend the prescribed wake vortex lattice method (VLM) to
perform aerodynamic analysis and optimization of dual-rotor wind turbines.
The additional vortex system introduced
by the secondary rotor of a DRWT is modeled while taking into account the
singularities that occur when the trailing vortices from the secondary
(upstream) rotor interact with the bound vortices of the main (downstream)
rotor. Pseduo-steady assumption is invoked and averaging over multiple
relative rotor positions is performed to account for the primary and
secondary rotors operating at different rotational velocities. This
implementation of the VLM is first validated against experiments and blade
element momentum theory results for a conventional, single rotor turbine. The
solver is then verified against RANS CFD
results for two DRWTs. Parametric sweeps are performed using the proposed VLM
algorithm to optimize a DRWT design. The problem with the algorithm at high
loading conditions is highlighted and a solution is proposed that uses RANS
CFD results to calibrate the VLM model.
In addition to wake losses, aerodynamic interaction between
turbines in wind farms leads to surface flow convergence . This phenomenon has been
observed in field tests with surface flux stations. A hypothesis is
proposed to explain this surface flow convergence phenomenon - incomplete
pressure recovery behind a turbine leading to successive pressure drops in
tightly-spaced turbine arrays leads to drop in overall pressure deep inside
a wind plant; this low-pressure acts as an attractor leading to flow
convergence. Numerical investigations of the phenomenon of surface flow
convergence are carried out that support this hypothesis. An actuator disk
model to represent wind turbines in an LES
CFD solver is used to simulate hypothetical wind plants. The flow
convergence phenomenon reflects as change in flow velocity direction and is
more prominent near the ground than at turbine hub height.
Numerical simulations of wind plant aerodynamics are conducted with various
approximations to investigate and explain the flow convergence phenomenon.