Optimizing your design increases revenue and reduces land requirements
According to an article, the wind industry could benefit from the use of artificial intelligence (AI) in the design and deployment of wind farms. natural energy Written by researchers at the U.S. Department of Energy's National Renewable Energy Laboratory (NREL).
The researchers developed an AI-based surrogate model called the Wind Plant Graph Neural Network (WPGNN). The model was trained on simulations of more than 250,000 randomly generated wind farm layouts under different atmospheric conditions, plant designs, and turbine operations. The simulation data was generated by his FLOw Redirection and Induction in Steady State (FLORIS) tool, another model developed by NREL. AI uses that information to determine the optimal design for wind farms. AI facilitates calculation of ideal plant layout and operations to achieve a variety of outcomes, including reduced land requirements and increased revenue.
This study focused on a strategy called wake steering. This optimizes the amount of energy that the plant can produce by controlling the wake that moves from the upstream turbine away from the downstream turbine.
Using AI, researchers were able to determine the impact of wake maneuvers on three different objectives: land use, cost, and revenue.
Although the benefits of wake steering have previously been demonstrated at the plant level, most studies have been limited in the spatial scale and range of optimization objectives considered. His WPGNN, used by the NREL team, efficiently represents wake interactions as a directed graph and provides comprehensive information about optimal settings for both turbine position and nacelle yaw across a national wind energy portfolio. made the investigation possible.
“Previously, studying site-specific wake steering optimization was extremely difficult, but WPGNN's graphical representation dramatically improves the ability to represent flexible layouts, changing wind directions, and perform slope-based optimization. ,” said co-author Ryan King. paper, “Nationwide evaluation of land use and wake steering economic effects by optimizing wind power plants using artificial intelligence”
This cross-cutting effort included researchers from the institute's Center for Strategic Energy Analysis, Center for Computational Sciences, and National Wind Technology Center.
King is a senior scientist at the Center for Computational Science, and co-author Andrew Grouse is a researcher in applied mathematics at the center. They co-authored the paper with two colleagues, Dylan Harrison Atlas and Eric Lantz, who have since left NREL. Previously he was Group Research Manager at NREL and is currently Director of the Department of Energy's Office of Wind Energy Technology.
The use of wind as a source of renewable energy is expected to become increasingly important in decarbonizing the country's electricity sector, but restrictions on where wind turbines can be installed in some regions pose obstacles. still remains. The AI-driven scenario considered building 6,862 plants across the country with a cumulative capacity of 721 gigawatts, with the goal of reducing carbon emissions from the energy sector by 95% by 2050. Ta.
Adopting a wake steering strategy could reduce the land required for future wind farms by an average of 18%, and in some cases by as much as 60%. Nationwide, the land savings total approximately 13,000 square kilometers, or 28% of the U.S. wind energy footprint.
Wake steering is valuable because simply widening the turbine is often not enough to avoid wake losses, and some wind farms lack the necessary space for further expansion. In addition, wind farms optimized for wake steering can have more concentrated turbines, thus meeting the desire of some communities to limit the amount of land available to industry. The ability to install more turbines in a smaller footprint provides greater flexibility from a site planning perspective, potentially allowing developers to take advantage of economies of scale on larger projects.
The researchers also found that the use of wake steering consistently reduced energy costs for wind power generation. Thanks to AI, researchers were able to uncover differences in the regions best suited to implement strategies.
“We found that different regions of the country are more or less susceptible to benefiting from wake steering, and that the benefits can be realized in different ways,” Graus said. “This could be important in understanding where and how to invest in this new technology.”
NREL's high-performance computing resources enabled researchers to train WPGNN. DOE's Wind Energy Technology Office funded the study.
