Dynamic modelling of wind farms for closed-loop control

Jaime Liew*

*Corresponding author for this work

Research output: Book/ReportPh.D. thesis

1282 Downloads (Pure)


This PhD thesis presents research on dynamic wind farm modelling for wind farm control applications. As wind energy continues to grow as a viable and scalable carbon-neutral energy source, the technology for wind farms must advance to meet increasing demand. One area of improvement is in wind farm flow control, in which wind turbines are manipulated to achieve a shared goal such as maximising power, power output tracking, or minimising structural loads. While some wind farm flow control strategies show notable benefits, the adoption of wind farm flow control in full-scale wind farms is still in its early stages. Detailed wind farm simulations are needed to test the response of a wind farm to these advanced strategies, but current simulation tools are either too computationally expensive or lack the required dynamic details to accurately capture the complex interactions within a wind farm.

The aim of this thesis is to develop dynamic wind farm modelling techniques and apply them to wind farm flow control applications. The research addresses two main questions: 1) How can the wind farm dynamics relevant to wind farm flow control be accurately and efficiently captured in a simulation environment? and 2) How can closed-loop wind farm flow control be performed in constantly changing, partially observable wind conditions? The study combines theoretical analysis, simulation experiments, and case studies to answer these questions. A major outcome of the study is the development of HAWC2Farm, a time-stepping wind farm simulator that utilises state-of-the-art dynamic models, including aeroelastic wind turbine simulations and the dynamic wake meandering model, to recreate complex dynamic scenarios in a wind farm. HAWC2Farm belongs to a unique genre of dynamic wind farm simulations that require fewer computational resources compared to computational fluid dynamic simulations, while still providing more detail than steady or quasi-dynamic wind farm simulations. The results of the study demonstrate that dynamic wind farm modelling can accurately capture the short-term dynamics of wind farm operations, including fatigue loads in wind turbines and the impact of wind farm control strategies on power output and structural loads.

The study also shows the feasibility of closed-loop wind farm flow control in constantly changing, partially observable wind conditions using a combination of model-based and data-driven control methods. Several use cases for the dynamic wind farm simulation methodology are presented, including both open and closed-loop controller design and implementation, with a focus on accumulated fatigue damage to each of the wind turbines in a farm. The closed-loop control strategy presented in the study uses wake steering as a means of controlling the wind flow and incorporates dynamic mode decomposition and reinforcement learning methodologies.

The findings of this research contribute to the advancement of dynamic wind farm modelling techniques and their use in wind farm flow control and optimisation. The thesis contains a series of journal and conference articles that present the research findings and are organised into three chapters that summarise and synthesise the results.
Original languageEnglish
Place of PublicationRisø, Roskilde, Denmark
PublisherDTU Wind and Energy Systems
Number of pages153
Publication statusPublished - 2022


Dive into the research topics of 'Dynamic modelling of wind farms for closed-loop control'. Together they form a unique fingerprint.

Cite this