NSF backs AI model to enhance safety of offshore wind turbine installation
Between late 2019 and early 2020, workers were installing offshore wind turbines (OWTs) off the coasts of Germany in the North Sea that would provide 200 megawatts of offshore wind power.
Amid rough ocean waves and high winds, a jack-up crane vessel would lift a blade several hundred meters high to connect it to the top of the tower, one of the most critical stages in OWT construction. Before the blade could be attached to the hub, the tower and nacelle, which houses the components that convert wind into electricity, would shake erratically, forming complex orbits.
Such motions put workers’ safety at risk and could cause delays in the installation process and increased costs. Incidents like these involving large tower-top motions during offshore wind farm development are not uncommon, yet there are no resources for predicting precisely when they will occur and to what extent.
Through scaled model testing and generative artificial intelligence (AI), Amrit Verma, University of Maine assistant professor of mechanical engineering, aims to develop a digital twin that will help make building fixed-bottom OWTs, particularly blade installation, safer and more cost-effective.
“A digital twin represents a virtual copy of a physical object or system that closely resembles and operates like the real-life version. By integrating the data from sensors on the physical system with computer simulations, digital twins can forecast how a system will respond to different real-world situations in near real-time.” Verma said.
Verma is leading this effort to create a generative AI model that can forecast tower top motions in real time under different wind and wave dynamics and OWT configurations. Yifeng Zhu, professor and chair of the UMaine Department of Electrical and Computer Engineering, and Andrew Goupee, associate professor of mechanical engineering, are collaborating with Verma on the project, which was awarded $299,960 through the National Science Foundation’s EArly-Concept Grants for Exploratory Research (EAGER) funding mechanism. EAGER supports exploratory work in the early stages on untested but potentially transformative research ideas or approaches that are considered high-risk, but high-reward.
Verma and Goupee are also faculty affiliates of UMaine’s Advanced Structures and Composites Center (ASCC), which has been a global leader in developing next-generation offshore wind technologies for nearly two decades. Three UMaine students, Saravanan Bhaskaran, Max Kruse and Aenor Codjo, are also contributing to the research.
“The installation of an OWT is a complex and costly process restricted by favorable weather conditions. During the installation, one of the most critical phases is the blade mating process during which the blades are connected to the hub,” Verma said. “Due to the increasing hub heights of the modern OWTs, the blades have to be lifted well over 100 meters using crane vessels, which could result in large motions at the blade root. The mating process requires a very high degree of precision, and damage to blade root has been reported in the past. Currently, there is relatively limited knowledge about the nature of tower top motions during installation. There is a clear need for in-depth studies and development of predictive tools to reduce the risks and uncertainty involved in the installation process.”
The team will first conduct an experiment in which they place a scale model of a wind turbine in a wave tank and characterize the tower top motions it experiences. The nature of the tower top motions can change at different wave heights, directions and periods, and across diverse turbine configurations. The experiment, which will be complemented with physics-based numerical simulations, will help make clear the specific drivers behind the orbital motions that the turbines experience, as well as support the data for the creation of a predictive model.
With data from the experiment, the model will be able to predict the tower top motion experienced by full-scale turbines at an even greater variety of sea conditions and technical configurations in real-time. Generative AI will allow the tool to forecast different scenarios turbines may experience during construction. Researchers will train the model to handle complex data distributions.
“One of the main outcomes of this project is an open-source AI-backed digital twin model, which will be able to accurately predict the tower top motions in near real-time based on the local weather forecast data. This will ensure that the people in the industry can foresee any complications that could arise during an installation, thereby reducing the risk to human life and improving the overall efficiency of the process,” Verma said.
The project also offers several educational opportunities for aspiring scientists and engineers. Data will be used to create new course modules and an open-source textbook. The graduate students participating in the project will receive mentorship from UMaine faculty as well as researchers from the National Renewable Energy Lab’s (NREL) campus in Golden, Colorado; Equinor, a global leader in offshore wind farm development based in Norway; and Kartorium, an Alaska-based digital twin company. Additionally, Verma and his colleagues will recruit high school students from Bangor High School to participate in various stages of the project.
“I’m eagerly looking forward to working on this project as it is a fantastic opportunity to collaborate and learn from experts in the industry as well as from top researchers in this field,” said Bhaskaran, a Ph.D student in mechanical engineering. ”As one of the two main graduate students involved in this project, I hope to contribute significantly towards producing an impactful research output.”
Creating new resources and experiences that offer academic and professional development in the field of offshore wind is nothing new to Verma. With support from the Governor’s Energy Office’s Clean Energy Partnership program, Verma has been spearheading the development of new courses, micro-credentials and an undergraduate concentration in offshore wind energy. He has also been devising new research and educational exchanges between UMaine and the Norwegian University of Science and Technology, the country’s largest university.
“I am very excited to work on this project with Professor Verma. I look forward to the manufacturing process of our model and the testing once we are finished,” said Kruse, an undergraduate student in engineering. “The work I have done thus far has been relevant to classes I have already taken and courses that I will have in the coming semesters. I plan on focusing my career around offshore wind once I graduate. I believe this internship will create a solid foundation for future work.”
Verma’s motivation for the project derived from his participation in UMaine’s Enhanced Mentoring Program with Opportunities for Ways to Excel in Research (EMPOWER), which supports faculty seeking to achieve significant professional growth and advancement, including success in research and scholarly activities. Goupee served as Verma’s mentor during the program.
Contact: Marcus Wolf, 207.581.3721; marcus.wolf@maine.edu