Camus Energy: Grid software for small utilities with big clean energy ambitions

Keeping a grid running on 100 percent solar power is a technically challenging feat. It’s particularly hard when you’re a utility covering remote, mountainous terrain serving as few as five customers per mile of power line, as is the case with Kit Carson Electric Cooperative.

But the 23,000-member, customer-owned electric utility in north-central New Mexico has been able to steadily move ahead on its goal of 100 percent daytime solar power by 2022 — not just from a handful of central solar farms, but from a dozen different distributed solar sites connected to its low- and medium-voltage grid.

“We have a 500-kilowatt system here, a 15-megawatt system there, a battery integrated here and there, all around our territory,” CEO Luis Reyes said. ​“Today we have full communities that are 100 percent daytime solar.”

That wasn’t supposed to be possible, according to the U.S. Department of Energy experts Kit Carson consulted when it set its goal a few years ago. ​“We talked to Los Alamos [National Laboratory] and Sandia [National Laboratories], and even the National Renewable Energy Laboratory, and asked what happens at different penetration rates,” he said. ​“Initially the feedback was [that] the sky will fall if we go above 35 percent.”

But so far, the utility has been able to balance reverse power flows, voltage sags and surges and other solar-generated grid disruptions with conventional grid technology, he said — largely due to the help of software partner Camus Energy.

“In the end, our business is data-driven,” he said. ​“With the Camus product, the more data I get, the more opportunities I have to come up with a solution.”

Since its 2019 launch, Camus Energy has deployed its hardware-agnostic cloud computing platform to a roster of cooperative and municipal utilities with 400,000 collective customers across the U.S. West. Noteworthy customers include Kit Carson and Colorado-based Holy Cross Energy, two co-ops with ambitious clean energy goals.

On Tuesday, Camus raised a $16 million Series A round led by Park West and joined by Congruent Ventures and other investors. The funding comes on top of a $3 million seed round in late 2019 led by Congruent and Wave Capital, and a friends-and-family round that allowed the San Francisco–based startup to build a prototype to deploy with a handful of test customers.

Astrid Atkinson, Camus’ CEO and co-founder, described the startup’s technology as a grid-facing version of the site reliability engineering built by Google, where she led the teams responsible for the search giant’s web serving layer and social media products.

“The core of that model for Google was basically this idea of unreliable hardware with reliable software,” she said in an interview. In its early days, Google bought ultra-cheap servers to meet its massive scale-up needs, and its site reliability engineering team was tasked with keeping them all running, to the point that servers ​“could literally catch on fire, and we would fix it with software.”

“That’s what we were looking to bring into the grid space,” she said. Power grids face similar challenges in terms of connecting a multitude of hardware devices on the grid or behind customers’ meters, ranging from decades-old analog switchgear to the latest digital smart meters and solar and battery inverters.

One of the key goals of Camus’ small utility customers, she said, is connecting distributed energy resources, or DERs — solar panels, batteries, inverters, electric vehicle chargers and building energy control platforms.

“We provide an [advanced distribution management system] equivalent real-time dashboard for their grid, and we’ll directly incorporate utility or customer-owned DERs for that,” she said.

A system of systems for utilities that can’t afford top-shelf grid software

An advanced distribution management system (ADMS) is software designed to give utility operators visibility and control over their sprawling distribution grids. They’re made by global corporations like Schneider Electric, Siemens, General Electric, ABB and Emerson, and they come with implementation costs and complexities that largely limit their usefulness to larger investor-owned utilities — not smaller rural co-ops like Kit Carson, which has only 129 employees.

“Our pricing ends up in the ballpark of a low-end ADMS,” or about $600,000 to $800,000, compared to as high as $100 million for a full-scale implementation at the largest utilities, Atkinson said. Camus charges an integration fee and a subscription fee for its cloud computing services, which aligns well with publicly owned utilities that lack the investor-owned utility incentive to earn rates of return on capital investment into their computing infrastructure.

Beyond the cost and time involved in implementing them, ADMS platforms aren’t necessarily built to be integrated with other utility software systems for managing electric meter communications, customer billing and services, and other core operations, Reyes said.

“Like a lot of utilities, I probably have 20 software programs, most of which don’t talk to each other. Camus puts those all together, so the same information my engineering department is looking at, my customer service department is looking at,” he said.

In that sense, Camus is ​“more of an aggregator of aggregators,” Atkinson said. For example, customer Holy Cross Energy has ​“a number of programs they’re putting in place or want to put in place,” including community solar systems, microgrids, and behind-the-meter solar, batteries and EV chargers.

“Each of those comes with a software platform,” she said, “[and] it’s hard to then go to each dashboard and individually schedule all the things to accomplish any goal at all.” Camus’ single dashboard allows the 44,500-member co-op to pick and choose from the entirety of these DERs to shift or shed energy use to reduce costly peak electricity demands, or store solar power in batteries for later use, she said.

Integrating these kinds of disparate software systems is a perennial utility challenge, but it’s become more complex with the rise of distributed energy. Camus’ reliance on cloud computing to manage the immense data collection, storage and processing needs involved has allowed it to manage those kinds of tasks in real time, Atkinson said.

“Our team is pretty good at talking to obscure computer interfaces,” she added, with a suite of ​“connector libraries” for a host of different vendors’ supervisory control and data acquisition (SCADA) systems that manage grid communications and controls, the advanced metering infrastructure systems for smart meters, and the various protocols used by solar and battery inverters.

Those integration and real-time analysis chops have played a role in allowing Kit Carson to exceed experts’ expectations of the limits of solar penetration. ​“They knew they would be entering into an environment where there may be impacts they couldn’t foresee and might cause reliability issues,” she said.

By connecting to the SCADA at distribution substations, ​“we have that head-end view of the substation and circuits below it. We also connect to the smart meters below the line all along it to put together a voltage profile all along the circuit.”

By tapping the utility’s interconnection data, ​“we know where rooftop solar is located all along the circuit,” and can use modeling and forecasting techniques to predict each system’s output at any given time.

The end result is ​“a circuit profile that maps out, what are the available voltage readings across this circuit, what are the solar systems on this circuit [and] a deeper understanding of thinking about the settings for the equipment they have in place.”

Building visibility into a shared-resources grid

These kinds of insights have allowed Kit Carson to fine-tune its existing grid equipment to manage the voltage and power flow fluctuations that can come from an all-solar-powered grid, and might otherwise trip protective relays or otherwise cause grid failures, Reyes said.

That, in turn, has opened the door to much greater penetration of local solar for the communities it serves, including six municipalities and two tribal nations — the Pueblos of Taos and Picuris — as well as for individual customers and the University of New Mexico’s Taos campus.

“One of the things we’re trying to create is this model — anyone who wants to go to 100 percent local distribution solar generation, here are the things that didn’t work, and here are the things that corrected those actions,” he said.

Along with its big push into distributed solar, Kit Carson has switched wholesale energy providers to expand its access to renewable energy, he noted. But it’s already looking ahead to the need for batteries to store its solar power to mitigate peak loads, as well as the likelihood of far greater EV adoption, now that the Ford electric F-150 pickup has hit the market — ​“rural adoption of EVs was kind of lukewarm until the trucks started coming.”

That process will necessarily involve its customer-members in the decision-making that goes into where to locate these DERs, as well as how to control them to minimize any costly disruptions to the grid’s overall operations, he said.

“We see Camus as giving our members total visibility into the system,” Reyes added, including data on voltage levels from moment to moment, and circuit capacity for new DERs to share with developers. ​“If you’re going to have a true DER communication and energy flow grid, anyone who participates needs access to those tools. Most vendors don’t look at it that way — it’s proprietary, it’s controlled by the utility, and maybe you can get information from a portal.”

Camus has built in hefty cybersecurity to support these kinds of data-sharing use cases, Atkinson said, including guidance from CTO Cody Smith, a former member of Google’s intrusion response team, to employ ​“zero-trust” methods that secure individual components in its system.

It’s also working with the DOE’s Pacific Northwest National Laboratory and Kit Carson on a federal grant-funded project to apply machine learning techniques such as singular spectrum analysis and multi-resolution wavelet spectra analysis to fill in the gaps from inaccurate or incomplete data. That’s a common problem on grids with less-than-perfect connectivity and data flows from smart meters, grid sensors or behind-the-meter DERs, and could serve larger-scale utilities as well, she said.

An investor-owned utility that joined Camus’ recent investment round is working on applying the company’s technology to its grid operations, Atkinson said. While she wouldn’t name the utility in question, she did say that it was ​“looking at building this capability internally, and got as far as staffing it up and seeing what it would cost,” before learning of Camus and agreeing to work with the startup.

For the next few years at least, however, Camus plans to remain primarily focused on publicly owned utilities with fewer than 1 million customers, she noted. ​“They have flexibility to explore these models,” whereas larger investor-owned utilities face more regulatory and economic disincentives to move quickly in ways that could disrupt their existing operations.

There’s also value in proving out its techniques via projects like its DOE-funded project with Pacific Northwest National Laboratory and Kit Carson, she said. ​“Without the data, it’s hard to take risks.”

Utilities around the world are struggling to chart a pathway to obtaining most, if not all, of their electricity from renewable energy without causing major disruptions to their grids, Atkinson noted. The name she and her co-founders chose for their startup highlights that fact — it’s from ​“The Myth of Sisyphus,” the essay by French existentialist philosopher Albert Camus, which asks the question, ​“how do you create meaning and purpose in the face of the universe as it is,” she said, an endless struggle with an uncertain outcome.

“The answer is: Pick something. Do it,” she said. ​“There’s joy to be found in doing the work, and the joy is the struggle itself.”

(Lead image courtesy of Kit Carson Electric Cooperative)