Modeling, control and design of extreme scale wind turbines

Abstract

Wind turbines are among the top renewable energy producers, having carved out a large chunk of the renewable energy market as a result of their economic viability and the abundant wind resource available throughout the United States and world. This green push over the past two decades has seen the nominal size of wind turbine rotors increase nearly three fold, being one of the main contributors to the economic viability of wind energy. This increase in rotor size has not been realized without advances in available sensors, actuators, and control systems responsible for maintaining and operating these turbines. Turbines were originally operated by constant speed generators and had passive aerodynamic stall speed regulation, however, with the advent of variable speed generators and active pitch control the application of advanced control architectures has further enabled increased turbine performance and longevity. Wind turbines are governed by nonlinear dynamics, benefiting from non-linear control architectures for optimal operation across a large operating envelope. Control systems are responsible for optimal tip speed ratio tracking during below-rated operation, speed regulation during above-rated operation, yaw control for inflow direction changes and shutdown procedures during unsafe operating conditions. In addition to these baseline metrics, control systems are being used to provide load alleviation during power production operation while maintaining maximum power capture. The ability of the control system to maintain a wind turbine within a safe operating envelope allows designers to reduce design loads, in turn, reducing the mass and cost of wind turbine structural components. Advances in control tuning algorithms and control architectures lead to cheaper, more reliable turbines which reduce the levelized cost of energy (LCOE). This dissertation presents work completed on the development of advanced turbine modeling and control algorithms that facilitate systematic and reproducible turbine design, in addition to providing increased wind turbine performance in terms of load alleviation during operation and shutdown procedures. A linear parameter varying (LPV) approach is used during the model development and control synthesis stages to account for plant non-linearities, identify variations in plant eigen-frequencies across the operational envelope, and to facilitate multi-objective control and optimization. The LPV models are incorporated into a generalized control synthesis algorithm that drastically reduces the time required to obtain an optimal controller in addition to providing targeted eigen-frequency attenuation during closed-loop operation and easy model augmentation for multi-input-multi-output (MIMO) control synthesis. Finally, the modeling and control algorithms are utilized in the design of a segmented ultra-light morphing rotor (SUMR) 50MW wind turbine with an associated levelized cost of energy (LCOE) analysis comparing values for currently existing on and offshore wind farms showing that our design process results in an economically competitive wind farm.