|Multi-objective coordinated control of wind turbines in wind farms based on active yaw control
|Yang, Hong Xing (BEEE)
Lu, Lin (BEEE)
Hong Kong Polytechnic University -- Dissertations
|Department of Building Environment and Energy Engineering
|xi, 156 pages : color illustrations
|The urgency to address the energy crisis and meet the targets set by the Paris Agreement has led to a growing emphasis on transforming the energy utilization landscape. Renewable energy sources, such as wind energy, offer promising solutions to reduce primary energy consumption, curb carbon dioxide emissions, and foster the development of carbon-neutral communities. However, the field of wind energy faces significant challenges, with wake losses in wind farms being a prominent concern. The wake flow generated by one or more upstream wind turbines is characterized by higher turbulence intensity and reduced wind speed. This phenomenon results in fluctuating fatigue loads and substantial reductions in power generation for downstream wind turbines. In order to mitigate the impact of wake effects on the power output of downstream wind turbines, yaw control has emerged as one of the most promising active wake control techniques. By deliberately adjusting the yaw angle, the wake can be intentionally deflected away from the downstream wind turbines, minimizing the detrimental effects and optimizing their power generation.
To begin with, a groundbreaking anisotropic three-dimensional wake model is developed based on a multivariate Gaussian distribution. This innovative model presents a closer approximation to reality by considering two independent dimensions, namely the horizontal and vertical directions, instead of relying solely on circular radius. As a result, the model incorporates a generalized expression for the wake expansion rate in each dimension. In comparison to conventional models, this novel approach demonstrates a commendable capability to reasonably predict the wake distribution of an individual wind turbine. Furthermore, the model boasts the advantages of low computational cost, eliminating the need for additional numerical calculations or trial computations. Therefore, the wake velocity at any given spatial position can be accurately and swiftly calculated. Validations are conducted from near wake to far wake with relative error within 5% in most cases.
In order to achieve precise prediction of yawed wakes for effective yaw control, an advanced three-dimensional yaw wake model is introduced, building upon the previously developed wake model. To validate the model, the deflected wake model and yawed wake distribution are rigorously tested using well-known public measurements and wind tunnel experiments conducted by the authors utilizing particle image velocimetry (PIV). The comparison results showcase the excellent performance of the proposed model in accurately predicting both the wake distribution and the trajectory of the centerline of a deflected wake. The model's incorporation of the elliptical wake cross-section, with a short horizontal axis resulting from yawing operations, ensures the capture of the physical significance of the proposed yawing wake model. Results shows that both the trajectory of deflected wake centerline and yawed wake distribution can be accurately predicted with small NRMSE. This validation underscores the reliability and applicability of the proposed model in the field of yawed wake prediction and control.
Afterwards, a multiple yaw wake model is formulated by combining the newly developed three-dimensional yaw wake model with the classical sum of square (SoS) superposition model. This model enables the independent adjustment of the yaw angle for each wind turbine within a wind farm. By optimizing the yaw angles individually, this model facilitates the identification of optimal solutions for wind farms. Furthermore, the model successfully reproduces the interaction between wake flow and secondary wake steering effects. The accuracy and reliability of the model are validated through comparisons with particle image velocimetry (PIV) measured data, further confirming its effectiveness in capturing the intricate dynamics of wake flow and secondary wake steering effects.
In order to optimize yaw control for maximizing power output and minimizing fatigue loads, a machine-learning based power and fatigue prediction framework is proposed. To accomplish this, a comprehensive set of inflow conditions, wind turbine layouts, and yaw control angles is carefully chosen to conduct high-fidelity simulations. Features and outputs are reasonably selected to reflect the environmental and operational conditions and the corresponding fatigue and power performance. Then, a machine learning algorithm, support vector regression (SVR), is employed to train and evaluate the prediction model. The performance of the selected prediction method is assessed by comparing it with another algorithm using multiple metrics. This comparative analysis serves to validate the superiority of the chosen prediction method in accurately predicting the power output and fatigue loads associated with different yaw control configurations. Through this rigorous evaluation process, the reliability and effectiveness of the machine learning-based prediction framework are demonstrated. The prediction takes less than 1s and most outputs have prediction errors within 3.5%.
Subsequently, a comprehensive fatigue analysis is conducted to assess the combined effects of wake flow and yaw control on wind turbines. The analysis encompasses various load factors, including bending moments at blade root, yaw bearing, and non-rotating components like the tower base. Different wake conditions, ranging from full wake to partial wake, are generated to explore the interplay between wake flow and yaw control. The analysis reveals distinct load trends and variation rates in response to the different scenarios considered. By examining these combined effects, a more comprehensive understanding of the fatigue behavior of wind turbines can be obtained.
In the final stage of this study, a multi-objective coordinated yaw optimization strategy is proposed with the aim of simultaneously maximizing the total power output and minimizing the fatigue loads of wind turbines. Leveraging the validated superimposed yaw wake model, the required inputs for the optimization process, such as the nine-point inflow profile, turbulent intensity, and yaw angle, can be readily prepared. With the assistance of a well-trained support vector regression (SVR) framework, the fatigue loads and power output under various yaw control scenarios can be accurately predicted. These predicted values serve as the objective functions for the optimization process. By considering different inflow conditions and wind turbine layouts, the coordinated yaw optimization can be performed. The objective is to identify the optimal yaw angles that lead to the maximum power output and minimum fatigue loads. By iteratively exploring and evaluating different yaw angles, the multi-objective optimization algorithm aims to identify the most favorable configurations that optimize both power output and fatigue load reduction.
The present study introduces several novel contributions. Firstly, a novel anisotropic 3D Gaussian wake model is developed, which is further extended to a yaw wake model and a superimposed yaw wake model for multiple wind turbines. This model takes into account the anisotropic nature of wake expansion and accurately predicts the wake distribution under yawed conditions. Furthermore, a machine-learning-based framework is developed for fatigue and power prediction under yaw conditions. This framework utilizes inputs from the proposed multiple yaw wake model and employs machine learning algorithms to accurately predict fatigue loads and power generation. The framework provides a practical and accessible approach for assessing the performance of wind turbines under yaw conditions. Additionally, a comprehensive fatigue analysis is performed, considering the combined effects of wake flow and yaw control on critical turbine components. This analysis provides valuable insights into the fatigue behavior of turbine components and aids in the development of strategies to mitigate fatigue damage. Lastly, a multi-objective coordinated yaw optimization framework is proposed. This framework aims to maximize the total power generation while mitigating the associated fatigue loads. By considering various inflow conditions and wind turbine intervals, the optimization process identifies optimal yaw angles that optimize power output and minimize fatigue damage.
|All rights reserved
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: