Under the hood of PG&E's summer VPP pilot with Sunrun

During some of the hottest months of 2023, 8,500 residential batteries in California discharged back to the grid for two hours every evening. This routine helped their utility to manage the challenge of after work peaks, combined with unusually high temperatures.

Those homes were part of a much anticipated, first-of-a-kind virtual power plant pilot operated by PG&E and Sunrun, a discrete program that stemmed from a directive by the California Public Utilities Commission in response to the state’s summer capacity emergencies.

From August to October, this “permanent load shift” program provided the grid with an average of 27 megawatts of power every evening, reaching a peak output of nearly 32 MW, which the companies said could power more than 20,000 homes. 

The VPP’s output, however, was lower than the 34 MW PG&E was anticipating, and the utility said it’s not looking to replicate the program quite yet — though the utility’s head of distributed generation programs Larsen Plano told Latitude Media that the experience yielded “some interesting learnings.”

“We would not do this program in exactly the same way that we did in 2023,” he added. “It’s interesting to see that there’s a lot of gremlins in this process that still need to be worked out.”

Technical snafus

As the largest utility in the United States, PG&E has experimented with a variety of VPP models, including through a grid stabilization partnership with Tesla. 

The “permanent load shift” pilot with Sunrun, though, represented a novel discharge and compensation model, one that offered scheduled power injections and gave customers a flat $750 participation fee. Dubbed the “Energy Efficiency Summer Reliability Program,” it received authorization in early 2023 from the CPUC for $10.5 million in funding, and the VPP was up and running within six months.

It was run entirely in-house on Sunrun’s network operations platform, without any software interaction between Sunrun and PG&E, and metered directly at the battery level.

However, despite the fact that the program had a static schedule that wasn’t based on real-time pricing or flex alerts, Plano said a number of unresponsive batteries dampened the anticipated output each day.

“This wasn’t anything to do with the customers or anything they have control over,” he said. “It’s just that somewhere in the chain of central dispatch origin down to the actual device on the customer premise, something wasn’t lining up.” 

There wasn’t any apparent rhyme or reason to those failures, Plano said — they occurred with different batteries and at different times across the program. At the outset, the program was supposed to deliver 34 MW of nameplate capacity each day, assuming all batteries were responsive. During a typical month of the program, daily capacity averaged under 30 MW, Plano said.

But that wasn’t solely due to unresponsive batteries.

“There’s a variety of different [battery] manufacturers that each of these customers might have, some of them are size X and some of them are size Y,” Plano said. “You have to make an assumption early on about the mix of different batteries we’re going to get.”

PG&E’s starting assumption of how many megawatts would be up for grabs didn’t pan out, he added. 

“I would say that those technical gremlins in the dispatch were the biggest factor in terms of the day-to-day megawatt numbers not meeting what we were shooting for,” Plano said.

That will ultimately inform ways in which PG&E might “tweak” or “sharpen” this model, he added. One very simple way to build on the pilot’s learnings will be to downgrade capacity assumptions based on their new and better understanding of battery response rates. 

Customer calculations

Sunrun’s business model made the process of enrolling program participants relatively smooth, said Chris Rauscher, the company’s head of grid services and VPPs. The residential solar giant already owned and operated the solar and battery setups in every enrolled home, he added.

“We can quickly optimize that fleet using our software provider Lunar to continue providing time of use bill savings, and also hit that grid services VPP dispatch for the utility,” Rauscher told Latitude Media.

“It’s a strength of our third party ownership model that we can operationalize VPPs at scale so quickly,” he said, adding that it was also easy to keep customers enrolled; opt-out rates were below 10%. 

Those low opt-out rates alone make the project a unique one, said Latitude Intelligence managing director Matt Casey. 

“We’ve heard with other VPP platforms that resemble more traditional demand response or have voluntary participation, participation rates have been very low,” Casey said. In some of those cases, he added, fewer than 50% of enrolled participants responded to an event. 

PG&E and Sunrun’s low opt-out numbers likely come down to convenience and compensation, Casey said. This particular program only called on batteries for two hours during the day, and customers’ existing relationship with Sunrun likely reduced some of the friction points for enrollment, he said. The lure of $750 certainly didn’t hurt either.

“Sunrun is probably starting to understand where people are potentially falling through the cracks, and honing in on program designs that reduce churn,” Casey said.

Plano pointed to the positive customer response as a key takeaway and something of a surprise for the utility. “That’s for me a fundamental learning about customer willingness to turn over control of their batteries,” he said. 

For PG&E, perhaps the biggest benefit to the pilot was the customer data Sunrun was able to provide. According to Plano, the “smooth flow of data” allowed the utility "to do really good analysis to show precisely what was the benefit that the program provided.”

At the same time, refinement is still needed to make the model replicable, Plano said: “The incentive was quite generous, and as a result this program did not stack up well in terms of cost effectiveness.” 

Building a blueprint

Plano said he’s thinking about the program as a “self-contained step in the journey towards the grid of the future,” and that PG&E will be pulling its lessons into other projects, like demand response and emergency load reduction programs.

Sunrun is also bringing the pilot’s results in-house, to apply to current and future VPP programs. 

“We were very pleased with the performance, and we also learned a bunch of lessons about a fleet of this size operating for this type of season,” Rauscher said, adding that those lessons give the company a higher degree of confidence in its performance forecasts. 

But unlike PG&E, Sunrun is already looking to replicate the summer reliability program elsewhere.

“Not only would we replicate it today, but we’re proactively seeking new contracting parties and new operating programs,” Rauscher said. 

Sunrun is offering a load modification product, structured very similarly to the PG&E pilot of last summer, to California Community Choice Aggregators. The product harnesses the company’s immense fleet — as of late September, Sunrun has deployed 76,000 solar and storage systems nationally — to help CCAs manage peak demand and daily peak load curve.

In a tender released in late January, Sunrun touted the “astounding and positive response” from operating the VPP with PG&E. It outlined an offering in which CCAs can select any two-hour period beginning at 5pm PT to hedge against peak demand and pricing in their territory, with the option to modify the dispatch period across seasons.

The inherent flexibility in the aggregator VPP model is one of its most valuable selling points, Rauscher said, pointing to the success of another VPP program Sunrun is operating with a very different utility, Puerto Rico’s Luma Energy. 

That’s the future of the distributed grid, Rauscher said: a VPP structure that can work for the country’s largest and most complex utilities, as well as smaller ones and CCAs.

“The most exciting thing for me is that every utility in the country sits somewhere on the spectrum between PG&E and Luma,” Rauscher said. “That means that we can do these VPPs at scale, providing value to utilities and their customers today, anywhere in the country.”