In 2009, the U.S. Department of Energy rolled out a $7.9 billion effort to invest in new smart grid technology. Most of those billions went into advanced metering infrastructure — two-way smart meter systems that could allow utilities to collect tons of data from the grid edge and promised a slew of benefits for the electricity system.
Nearly a decade and a half later, many of those promises remain unfulfilled. More than 100 million smart meters have been installed in the U.S., but industry experts tend to agree that the full potential of smart meters hasn’t yet been realized.
Early adopters weren’t able to manage the large amounts of data they pulled in, a 2021 paper from Guidehouse Insights noted, and tended to use the promising tech for simple processes like automated billing. And they generally didn’t share relevant data with customers, leaving billions of dollars of energy-efficiency savings on the table.
Today, the industry is experiencing a kind of déjà vu. A new generation of grid-edge technologies, equipped with artificial intelligence and advanced analytics capabilities, could be poised to bring fresh benefits to the electricity system. And as utilities grow aware of these benefits, spending on grid-based smart meter analytics is estimated to increase at a 13.3 percent compound annual growth rate between 2021 and 2030, according to the Guidehouse paper.
Of course, the industry landscape has changed a lot since the first wave of smart meters. There’s been a broader shift to distributed solar and battery storage in homes and businesses, as well as an influx of electric vehicles on the roads. These changes have made grid planning more complicated, but also offer a lot of potential in terms of demand flexibility and decarbonization.
This all means that the industry has a real opportunity for a smart meter “do-over.” And as the industry works towards aggressive clean energy targets, getting this right is even more important now than a decade ago. But getting the most out of these technologies will also involve nailing the cost-benefit equation, as well as a clear-eyed approach to regulatory buy-in.
‘It’s not just a sensor’
Multiple companies, like Itron and Schneider Electric, are taking a closer look at smart grid monitoring products and other devices that could be placed at the grid edge.
At the forefront of this wave is energy software company Utilidata, which has created a distributed AI platform called Karman— powered by AI leader NVIDIA — that can be attached to meters. The chip uses a combination of on-chip and cloud-based software, as well as distributed AI, to pull in complicated grid data that can provide utilities with a clearer picture of how the system is working in real time, identify and manage distributed energy resources, and offer a host of other capabilities.
“It’s not just a sensor at the edge of the grid or something that can run a few algorithms — it’s a real AI-ready distributed operational platform,” Jess Melanson, Utilidata’s president and chief operating officer, said.
Assuming that this technology is deployed at scale, utilities may soon be able to fully understand and predict what’s happening on the grid in real time, and have better control over how their assets, and their customers’ assets, are operating.
This information has transformative potential. It could help utilities reduce the cost of operating the grid, and more easily adopt electric vehicles, solar, and energy storage. It could help tighten up interconnection processes, and allow utilities to collect more information about, and communicate better with, the distributed energy resources on its system, according to Utilidata.
And it could bolster the grid’s overall reliability, Melanson said, “whether that’s shedding load in a resource constrained event, or just being able to really precisely and in real time locate outages and respond to them.”
Oregon-based Portland General Electric, for instance, plans to pilot 500 of Utilidata’s smart chips by mid-2024.
“Integrating smart grid chip technology and grid-edge computing will contribute to PGE’s overall strategy, helping us add more clean energy to the grid, reduce power outages, enable quicker storm recovery, and ultimately lower the cost of grid operations,” Matt Hubbard, PGE’s manager of distributed device strategy, said.
Adding real-time advanced data analytics at the edge of the grid will improve system visibility, he added, and ultimately support PGE’s decarbonization transition by enabling new and optimized DERs like customer solar and batteries.
The cost question
But the question remains: how will these new technologies will avoid the pitfalls that plagued the first wave of advanced metering infrastructure?
Some 73 percent of U.S. households and businesses were equipped with a smart meter in 2021, the Edison Foundation’s Institute for Electric Innovation found. But a decade after the 2009 passage of the American Recovery and Reinvestment Act — which led to the first wave of advanced metering infrastructure — only 2.9 percent of federally funded smart meters had real-time data features enabled, according to a report from Mission:data Coalition, a group of energy technology companies.
First, there’s the issue of costs. For a utility, deploying Utilidata’s smart grid distributed AI platform at scale represents a multi-million dollar investment. Precisely how many millions, the company said, will depend on how many customers a utility has.
“We’re advocating for a major increase in the investment in computing power at the edge of the grid,” Melanson said. But these utility investments are still way cheaper than upgrading transformers because they don’t have enough capacity, or installing a separate device to communicate with DERs, he said.
Viewed from the broader perspective of grid modernization, Melanson added, “the absolute best investment you could make is data capture and software at the edge.”
Utilidata’s initial cost-benefit analysis estimates that the value of the smart chip is more than ten times its cost. Much of this value comes from avoided infrastructure upgrade costs.
Elizabeth Cook, director of advanced grid systems and grid modernization at Duquesne Light Company, said that smart chip technologies could make sense from a cost perspective, assuming the industry stops viewing them as individual devices to swap out.
Instead, she added, the chips should be considered one element of building a new platform to drive all utility operations. The data they collect can inform and help drive down costs in utility operations and maintenance work, capital deployment, and manual labor requirements, and help automate multiple functions.
“Having that bottoms-up data and driving insights across the whole utility I believe pays for itself,” Cook said.
Regulatory skepticism endures
Meanwhile, smart chip companies and utilities will also need to convince utility regulators to greenlight investments in these technologies. Regulatory skepticism in the next wave of smart meter technology is justified, Melanson said.
“I think working with regulators to say ‘there are better solutions, there are better approaches,’ is really important,” he said. “That will involve trusting utilities to invest in more modern technology.”
Having that bottoms-up data…pays for itself.
In a certain sense, this requires faith in the utility sector’s ability to change its ways. Traditionally, utilities have tended to be slow in adopting new technologies, including AI.
Melanson outlined certain measures regulators and legislators can take to help get these technologies deployed at scale, including expected avenues like federal funding.
The sector also needs to see regulatory mechanisms that ensure utility investments in technologies like smart chips continue adding more value over their lifetime, according to Melanson. Essentially, regulators need to differentiate between software and hardware in how they tackle improvements and value creation.
Under the hardware status quo, a utility might make a business case for certain technology, and then report back to regulators a year later to prove it worked. But with something like a smart chip, Melanson said, a utility could instead come back to regulators year after year to demonstrate new ways it is driving value for customers, without additional investments.
Duquesne’s Cook sees regulatory skepticism as an obstacle, “but obstacles do not block the path, they are the path,” she said. It comes down to quantifying all of the costs and benefits associated with smart grid chips, as well as highlighting qualitative customer benefits, she said.
PGE is coming at the problem from a similar angle. The utility emphasizes “beginning with the end in mind,” Hubbard said. Practically, this means identifying business process changes before deploying the technology, so it can better track the results.
“These testing, implementation, and review processes will help PGE – and other utilities using similar best practices – ensure a smooth application of new technology while always learning to improve,” he added.
According to Melanson, the technology must follow the business case in order to be successful at scale. Building out a robust business case that keeps in mind the costs, as well as current and future benefits of smart chip technology, will be key if the industry hopes to apply the lessons of the first generation of smart meters.
As for what this means in practice, Melanson pictures an edge platform that isn’t just “a meter that does a few other tricks,” but rather a ground-up distributed AI platform, the likes of which NVIDIA has deployed in other industries.
Editor’s note: This story was updated on October 19 to reflect that Utilidata’s distributed AI platform is branded Karman.
