Would you be surprised to learn that there are about 55 gigawatts of solar generation behind-the-meter (BTM) in the United States? That’s one heck of a lot of solar generation to be BTM, but there’s more to the story. The figure came from a U.S. Energy Information Administration (EIA) study found on the EIA’s website. It makes for some very interesting reading, starting with the statement that over one-third of the U.S. solar power capacity comes from the small-scale solar category. In case you’re wondering, small-scale solar is defined as being less than one megawatt in capacity.
Residential small-scale solar systems typically are installed on rooftops. Commercial and industrial systems can be found on both rooftops and/or ground locations because they tend to have larger capacities than homeowners need. The EIA report started off saying the installed small-scale was 44 gigawatts in 2023. It continued saying that when the 2024 figures were available, EIA estimated that the installed capacity would grow to roughly 55 gigawatts. The final figures for 2024 aren’t available yet, but the third quarter looked good.
The Behind-the-Meter-Challenge
According to EIA, in 2023, residential customers made up 67% of the small-scale solar capacity. The commercial segment accounted for 27%, and the industrial sector was 6%. If the 44 gigawatt figure was used, the residential sector’s small-scale solar capacity was about 29.5 gigawatts. That figure jumps to 36.9 gigawatts if the 2024 estimate is correct and it’s probably close given all the BTM solar activity reported as 2024 ended. When we’re kicking around double-digit gigawatt figures, it probably doesn’t matter. What does matter is it’s getting attention!
When the experts gather, it’s being discussed and a question crops up. What if all of this BTM solar generation was available to improve the distribution network? That may sound like a challenge, but is it? Some of this small-scale solar has already been combined. It’s being sold to utilities by these aggregators using traditional methods despite their limitations. Of course, that begs another question. What impact would integrating a dynamic technology like advanced artificial intelligence (AI) into conventional microgrids have?
Well, that’s a trending topic making the rounds at smart grid gatherings worldwide. It touches so many elements of grid modernization like decentralization and decarbonization for example. Standard microgrids excel with small and localized energy systems, but they need help when it comes to merging and managing thousands if not tens of thousands of these small-scale solar installations simultaneously. For that matter, other distributed energy resources (DERs) must be included too, and this is where it really gets interesting.
Microgrid and Management
Have you heard of AI-powered microgrids, AI-driven microgrids, and AI-enhanced microgrids? These are three that come up continuously in web searches. The terms are being used interchangeably, but they really should not. There are differences, but that tends to be overlooked. Basically, AI-enhanced implies AI is a minor upgrade used to expand a system. AI-powered takes it a step further with AI being a key component of a system. The last one, AI-driven uses AI as the predominant force behind the system’s functionality.
When dealing with gigawatts of small-scale solar and a variety of DERs, AI-driven microgrids is the technology of choice. This application is needed for microgrid technology handling multiple DERs efficiently. By definition, an AI-driven microgrid is a “decision-making engine.” It controls the majority of the system’s activities and operations in real-time. Actually with an AI-driven microgrid, it’s about the management control system, which differs extensively from legacy microgrids.
Keeping it simple, traditional microgrid struggled managing the massive amounts of data coming at them 24/7. In today’s big-data environment that is not a good characteristic. When the data is coming from an assortment of multiple DERs it’s critical for the microgrid to be able to blend the data and respond as a single entity. That’s why AI-driven microgrids are trending, they are designed for this type of activity. Their management systems look at the data from DERs making up the microgrid and beyond to include every database interacting with the microgrid.
These include a variety of real-time databases like market prices, energy consumption, power grid conditions, environmental conditions like weather conditions, etc. An AI-driven microgrid management system sorts through all of this real-time data and predicts energy demands, recognize system vulnerabilities, and restore the system quickly during outages and they do it autonomously! If there are areas missing in the massive big-data being supplied, the AI technology can extrapolate the data it has and fill in the blanks. These abilities make this breed of microgrid more responsive when it comes to deciding when it’s time to generate, store, or sell electricity.
AI’s Impact on the What-Ifs
There have been numerous studies published about the benefits of AI on the power delivery system and microgrids. They all point out AI technology provides higher reliability, better flexibility, improved power quality and superior responsiveness, but there are other paybacks that are hard to quantify. The statistics vary depending on the parameters of studies and what procedures were followed. Given the complexity of the subject and the wide diversity of the studies being compared, it might help reduce confusion by comparing ranges rather than numerical results.
Several groups reported that AI-driven microgrid management systems can substantially reduce energy consumption between 10% to 20%. There were also reports saying that AI-driven energy demand techniques can reduce energy waste by 15% to 20%. Another cluster said AI-driven smart grids can reduce energy distribution losses by up to 30%, but it’s improving energy utilization that fits this discussion. All of these AI-driven microgrids provide characteristics that offer a number of solutions to the “what-ifs” popping up whenever regulators, utilities, and grid operators meet.
The what-ifs are those simple questions that popup when the evaluation process starts. It’s a progression we go through when making assessments, like what if a number of homeowners want to combine their respective rooftop solar panels to get a better price for selling their excess power? Or, what if a battery energy storage system is added? How about, what if they want to turn all these DERs into a community microgrid? What if electric vehicles were added to these microgrids?
Rapidly Changing Landscape
Suddenly the variety of options make a simple concept a lot more tricky, which is another reason interest in AI-driven microgrids is growing. Integrating the diversity of clean energy DERs into microgrid systems are a key element for a reliable energy supply. There is still a great deal of debate going on about the pros and cons of using AI-driven microgrids, but AI usage is rapidly expanding. It does present cybersecurity issues along with its autonomous decision making abilities. There are also some who would wait until AI is a noncontroversial technology – lots of luck with that.
When rooftop solar first came on the scene, there were similar discussions about embracing that technology, but the customer wasn’t listening. They made a massive commitment to solar, and they have added energy storage systems too. A few decades later we have 55 gigawatts of small-scale solar BTM that are ready for aggregators to power up an array of AI-driven microgrid systems. Taking it a step further, consider AI-driven microgrids managing a variety of the multiple DERs that have been discussed. What would be the potential benefits of melding these AI-driven microgrids and using them as a virtual power plant (VPP)?
This would be challenging with traditional microgrids, but not these AI-driven microgrids. Combined with advanced communication technologies and sophisticated cloud computing, AI-driven microgrids advance the VPP concept. We’ve talked about non-wire technologies being a solution to an aging and congested transmission system. AI-driven microgrid powered VPPs would be like a grid enhancing technology on steroids. It’s not that there aren’t VPP projects operating, but we need more of them and AI-driven microgrids would certainly take us to the next level.
A recent press release from Con Edison said their customers were continuing to invest in clean energy. As a result, those customers produced 647 megawatts of solar energy capacity from more than 72,500 solar installations last summer. They pointed out that represents “five times more energy than needed to power Times Square.” Con Edison said it “helped reduce energy peaks, impacts to the environment, and costs.” Imagine the AI-driven microgrid fueled VPP these 72,500 installations could bring about.
There are so many potential applications for AI-driven microgrids and VPPs are only one of them. What about taking advantage of the AI-driven microgrid’s to look beyond current conditions and identify a pending outage? Critical loads would be islanded and served via the microgrid without any human intervention. In a power grid striving to improve reliability and resilience, grid enhancing technologies like AI are needed, but understanding how they work is paramount. Equally important is understanding what they cannot do, but that keeps us involved and learning!
