| Summary: | <p> With the ever growing electrical demand, utility scale providers have looked, and have in some cases turned to expand current generation capacities to renewable energy generation. Currently wind turbine arrays utilize a selfish control algorithm to optimize the power production of each individual turbine. However, with the increasing demands for greater production capacity the array of wind turbines must increase in size. The increase in array size can present several challenges when optimizing with current control algorithms. These problems include turbine wake interaction and required spacing between turbines. Due to the variable nature of wind direction and speed, upwind turbines can obstruct the wind ahead of the downwind turbines. Avoiding inefficiencies due to the aerodynamic interaction within the turbine array can help reduce operation and maintenance costs as well as decrease the spacing required between turbines themselves. </p><p> To do this it is important to understand the wake that develops behind the wind turbine in order to accurately predict any losses that will be incurred by the aerodynamic interaction. The wake developed behind a turbine is dependent on several factors; terrain roughness, boundary layer, and oncoming wind characteristics. The wake generated by the wind turbine itself can be described as three different regions; the near wake, the intermediate wake, and the far wake. This has spurred experimental research wind tunnel testing and given way to numerous methods to approximate a turbine wake. All of these models have their advantages and disadvantages when approximating the downwind characteristics of the generated wake. These models are all used for micro-siting turbines to maximize the generation capacity of the turbine array while minimizing wake interaction between wind turbines. Although micro-siting wind turbine arrays will decrease the interaction between the turbines, shadowing effects can still occur between turbines. By utilizing a dynamic wake approximation, in conjunction with a feed forward control system, the wakes produced by the turbines can be mitigated to further increase the generation capacity and decrease the operation and maintenance cost of the entire array. This thesis seeks to bridge the gap between dynamic numerical models and low order empirical models in order to evaluate the interaction between wind turbines for varying states of wind. </p><p> Current wake models can be segregated into 4 separate categories including: empirical models, Linearized Reynolds-averaged Navier-Stokes (RANS), Nonlinear RANS, and Large-Eddy Simulation (LES). These categories increase in computational time and accuracy respectively. Industry typically utilizes the empirical and linearized RANS simulations to micro-site wind turbines while the latter two categories are used almost exclusively for research. This thesis will describe the first three models, as well as compare them with data taken wind tunnel testing done by Smith and Taylor's data.</p><p>
|
|---|
