VPPs, batteries, AI shove US power sector into 'fourth paradigm,' but rules lag: experts

The proliferation of virtual power plants, battery storage and artificial intelligence has put the US power sector in a "fourth paradigm" of development in which decentralized grids enhance resilience, but regulators may not have kept pace, experts said at a conference in San Antonio.

Speaking at the Energy Thought Summit at San Antonio's Boeing Tech Port Arena on May 22, Khalil Shalabi, Lower Colorado River Authority executive vice president for strategic initiatives and transformation, said electric utilities have gone through three paradigms, driven by technology and regulation.

"We spent 100 years with these big central stations as a traditional network," Shalabi said during a panel discussion entitled "VPPs, AI and Storage: Which will save us, which will doom us?"

"Then in the 1990s, with a different regulatory model and coming up with a combined-cycle gas turbine type technology -- which has lower initial capital cost and high operating costs -- we were able to have all these deregulated markets, and that was the second phase," Shalabi added. "The third phase was incentivizing a lot of renewables, right? Between that technology really advancing and regulatory models that allow it, that's the third one. Now we're struggling with building more resources in the fourth paradigm. I think the tech is advancing, but we're not seeing the sort of regulatory model that allows you to go into this kind of fourth paradigm. So I'm waiting for that."

Moderator Brent Sitterly, founder and CEO of Fixed Point Analytics, a data management company, asked how the power sector can "start to use VPPs, storage and AI to optimize our energy usage over time and do it in a way that prioritizes community?"

Using the 'most important asset'

Panelist Jeremy Adelman, manager of emerging technologies and innovation at Arizona's Tucson Electric Power, said utilities "have moved from a one-way static valuation environment of our assets to a two-way dynamic valuation environment for our assets."

"Everything has a market on the grid, even on the distribution level," Adelman said. "We need to start to think about the value of our systems differently and more dynamically as a result of both the storage and the VPPs that go on top of them to optimize them on behalf of our customers. The other piece of that, then, is how do all of us as utilities continue to be on this journey of getting much, much better with bringing our customers into these conversations as the core of the system, right? The customers are the most important asset we have."

Panelist Anissa Rodriguez, CEO of Pecan Street, a company that provides data research and product testing for energy, water, transportation and agriculture sectors, acknowledged that centralized control of the grid has been reliable, "but decentralized control is what is going to help us to build resilience."

"I look forward to this imaginary Utopian future where things are adaptive, they're inclusive, they're community-driven," Rodriguez said. "These technologies help the grid to learn faster and in real time in a way that is inclusive of everyone."

Rodriguez said its energy monitoring equipment network, which includes solar and storage customers and electric vehicles drivers, covers 10 states plus Puerto Rico and treats project customers as "participants in that project with us."

"The projects that we design are coalitions between universities, utilities, regulatory agencies, other community-based organizations," Rodriguez said. "The most important of that coalition I just think is probably the community-based organizations because they help us to co-create solutions that are hyper-localized to that community. That's how we assure them that the data we're collecting through these behind-the-meter devices that we're installing in their homes is going to be safe and that it's going to be secure and that the research is for a good purpose. And it's even given back to them in the form of this amazing user-intuitive platform so that they can actually adapt their behavior, even if it's just for ... lowering their electricity bill."

Panelist John Choi, vice president for strategic partnerships and inorganic growth at OCI Energy, which develops utility-scale solar and battery storage projects, said "it's extremely important" that AI helps power remain affordable for consumers below the poverty line.

"One of the ways AI can do in the near term is probably just price transparency at the level of the residential customer, so they can decide how it will help drive their habits," Choi said.

Machine learning can fall short

LCRA's Shalabi and TEP's Adelman differed over how helpful AI can be for utilities.

"We run drones, we take pictures, and now we have humans go through thousands of pictures," Shalabi said." We've been trying to train a model with this vendor, but we just don't have enough defects for the model to start finding out what a defect is and telling us what's there, so we don't rely on it. We just have to keep going back to basically supervisor viewing the drone pictures and saying, 'Okay, this is a broken insulator, this is rusty bulge.' ... So I think that part of the risk is maybe overselling AI or just having the wrong model right now."

Adelman acknowledged that AI needs enough data to "identify the outliers ... so there may be some use cases where you just can't make machine learning work yet."

"But I think there are a lot of use cases such as propensity modeling around EV adoption or community growth, or even looking at predictive modeling for distribution systems, which are becoming more and more prevalent in terms of grid planning," Adelman said. "At least we have a propensity for the idea that some parts of the system are more risky than other parts of the system."