How artificial intelligence (AI) can boost community support, financial returns, and performance of agrivoltaics projects
Agrivoltaics may be a relatively new concept, but many of its benefits are based on concepts farmers have grasped for centuries. Agrivoltaics is when solar photovoltaic (PV) panels are installed above crops so that the same land can be simultaneously used for energy and food production. When panels provide partial shade to crops, the shade reduces water evaporation and bolsters soil moisture levels by protecting the soil’s microbiome and productive capacity. Some crops can also increase solar production through increased albedo (reflected sunlight) and evaporative cooling.
Long before anyone imagined agrivoltaics, farmers planted shade trees, mixed tall and short crops to provide shade, and used cloths and nets to limit exposure to sunlight. While the basic concepts behind agrivoltaics are nothing new, technological advances can improve everything from where projects are sited to the day-to-day operations and potential long-term value of co-locating farming and solar energy production.
A recent EPRI technology innovation spotlight, Artificial Intelligence and Agrivoltaics, examined how AI and machine learning (ML) can help pinpoint communities likely to welcome projects and optimize their energy and food production once built.
Already, AI is being deployed to improve the production of individual solar power plants. This is no surprise: AI has extraordinary power to comb through massive data sets and pick out valuable trends and insights. For example, AI can analyze meteorological data and forecasts to adjust solar panels installed on a tracker to maximize energy production based on factors like temperature, cloud cover, and the intensity of sunlight. AI and ML can also bolster predictive maintenance by monitoring data for signs of everything from wiring problems to panel degradation and failure to inverter problems. By detecting potential problems before they materialize, AI and ML can trigger proactive maintenance that allows a solar generation facility to avoid disruptions that reduce its power output and revenue.
Similarly, AI and ML are being applied widely to improve agriculture production. For example, when combined with sensors that collect data on a farm, AI and ML can guide decisions about irrigation, planting, and harvesting based on soil quality and the health of crops. A recent study by the consulting firm McKinsey & Company estimates that AI can drive about $250 billion in value by improving crop yields, reducing costs for labor and inputs like fertilizers and pesticides, and by increasing sales and overall operational efficiencies.
Uniquely Suited to Complex Data Problems
If AI can benefit solar and agriculture individually, can it also be helpful when the two are combined in agrivoltaics projects? Bailie Neary, an energy systems and climate analysis engineer at EPRI, investigated that question to produce a technological innovation spotlight. Going into the project, Neary thought it reasonable that AI could be helpful to agrivoltaics. “Agrivoltaics and AI makes sense because it’s a complex optimization problem,” Neary said. “It felt like AI could benefit deployment.”
In researching the topic, however, Neary could not find any specific examples of real-world projects combining AI and agrivoltaics. However, AI’s potential to improve agrivoltaics has attracted the interest of the U.S. Department of Energy (DOE). Earlier this year, the DOE issued the report, AI for Energy. The report identified a wide range of areas where AI could accelerate decarbonization and grid resilience, including its capacity to forecast renewable energy production and streamline planning and permitting for transmission and other projects.
The DOE report identified the co-location of renewables and agriculture as one promising use for AI. “One specific example of such co-location is agrivoltaics, the combination of solar and agriculture, but other examples may include co-location with wind or geothermal or the use of agriculture for biofuels,” the report said. “AI can also translate land use challenges, from siting, permitting, and optimizing how to incorporate other economic uses, into opportunity maps for renewable energy developers.”
A Tool for Scaling Agrivoltaics
Agrivoltaics could use the potential boost AI offers. “True agrivoltaics where you have crop production—and that could be specialty crops, food crops, herbs—is difficult to scale to the size of traditional utility-scale solar facilities,” said Terry Jennings, an EPRI principal team leader whose research focuses on addressing environmental challenges of large-scale renewable energy development. “You see it mostly at what I would call commercial and industrial (C&I) scale projects. They’re typically five megawatts or less.”
Scaling agrivoltaics means tackling several challenges, including reducing upfront project costs, finding sites where communities will welcome projects, and operating sites to balance energy and agriculture production properly.
Here’s how AI could be beneficial at different stages of an agrivoltaics project:
Site Selection. One of the main barriers to expanded renewable energy development is community opposition. According to researchers at the Massachusetts Institute of Technology (MIT), 53 utility-scale wind, solar, and geothermal projects in 28 states were delayed or blocked between 2008 and 2021 due to various opposition sources. EPRI collaborated with Louisville Gas & Electric-Kentucky Utilities (LG&E-KU) to research potential vegetation management savings from grazing sheep at agrivoltaics facilities. The work was partly driven by a desire to address community concerns that solar degraded local ecosystems and made land less suitable for agriculture.
Proactively addressing community concerns and working together to ensure solar projects benefit citizens can be a good approach for project developers. However, AI could also potentially improve the initial site selection for agrivoltaics projects and boost the likelihood of successful community engagement. “It would be possible to have some data that allows us to determine which sites might be best based on some sort of census tract level social acceptance or availability of farmland,” Neary said. “It could be a way to filter many potential site options and come up with a map of the most promising sites.”
Building a Business Case and Attracting Financing. Project financing is a challenge for any solar project. But when solar is combined with agriculture, modeling the financial returns investors, farmers, and solar developers want to see becomes extremely complex. AI can be used to assess the financial viability of an agrivoltaics project by considering electricity demand, wholesale market prices, and how revenues can be shared equitably among stakeholders to ensure long-term collaboration.
Part of building a business case for agrivoltaics also relates to site selection. AI can help evaluate a potential site based on metrics of potential community support and can also factor in agricultural economics. “This is very site and region-specific,” Jennings said. “You could combine some form of AI or ML into a GIS tool and look at the market value of crops in that area to help identify potential sites.”
Project Development and Operation. Limiting the capital and operational expenses of agrivoltaics projects while maximizing their energy and food production is critical to their financial viability. Decisions about project design, equipment selection, and day-to-day operations involve combing through large amounts of data, a process that can be accelerated and improved with the help of AI.
For example, agrivoltaics projects can take advantage of sensors to provide a steady stream of real-time data. Because solar and crop production can be improved by adjusting to changing weather, soil, and other conditions, applying AI to all that collected data can inform better operational decisions. For example, temperature monitors could trigger a shift in tracking panels when an array is overheated and producing less electricity; that movement of the panels could benefit crops by allowing needed sunlight to reach plants.
“Already, agrivoltaics research sites have installed sensors to track soil moisture, evaporative cooling, solar irradiance, and power production,” Neary said. “The next step would be to analyze the data with AI and ML and use the insights to inform and possibly automate real-time decisions that improve the site’s performance.”
In 2023, EPRI collaborated with the New York Power Authority (NYPA) to identify leading practices in developing agrivoltaics projects. Among the main findings of the research was that a major hurdle to agrivoltaics projects is the fear among farmers of losing valuable food-producing land.
Without the support of farmers, agrivoltaics projects will be rare. AI and ML, however, provide an opportunity to proactively address the concerns of farmers and communities and support the development of high-performing projects that deliver financial benefits to everyone involved.
“Agrivoltaics is still in its infancy, and lots of people are trying to figure out the best way to pursue these projects to benefit everyone,” Jennings said. “So much data needs to be understood and acted on to site, build, and operate these projects in the best way possible. I think that is where AI can really help.”
EPRI Technical Expert:
Bailie Neary
For more information, contact techexpert@eprijournal.com.
