For years, the Energy Department’s National Renewable Energy Laboratory (NREL) has been working on technologies that could turn distributed energy resources (DERs) — specifically, smart solar inverters and batteries — into tools for keeping the grid stable.
It’s a tough task, requiring second-by-second coordination of behind-the-meter generators, batteries and loads to prevent them from throwing grid voltage and frequencies out of whack — or beyond that, to tap their flexibility to respond positively to grid disruptions.
As DERs scale up from hundreds to thousands to millions of devices at the grid edge, any attempt to control them with a centralized system is bound to be overwhelmed by communications and computational constraints, given the rapid responses required.
That means any scalable system for managing a grid that’s shifting from one-way electricity delivery to a more bi-directional network will have to break down key parts of the complexities involved, taking care of locally arising needs with split-second, automated decision-making. And, of course, it will need to give grid operators the tools to monitor and control everything on a grand scale.
This concept of a “hierarchical” grid control system has guided NREL’s work on an ARPA-E research agency grant-funded project launched in 2016. Now it’s yielded a working software platform, called Real-Time Optimal Power Flow (RT-OPF), that’s already at work coordinating DERs in Colorado, and is about to be put to use to coordinate community solar inverters on a distribution circuit in upstate New York.
Last month, grid voltage optimization company Utilidata announced it has secured the rights to commercialize RT-OPF in real-world deployments. The Providence, R.I.-based company’s unusual combination of real-time power flow modeling and real-time grid controls — and its ongoing integration with equipment from major grid vendors including Siemens, Itron and Landis+Gyr — could allow RT-OPF to become part of the toolkit for utilities struggling with DER challenges.
Utilidata’s first test of RT-OPF’s capabilities will come in a newly announced pilot project with the New York State Energy Research and Development Agency (NYSERDA), aimed at testing its ability to coordinate solar inverters to maintain reliability and improve voltage optimization.
The pilot will also study how utilities and DER owners can agree on how to use these edge-of-grid controls to reduce interconnection costs — and how solar systems can be properly compensated for the services their inverters can provide, without harming their revenues. That last issue is a serious problem for tapping smart solar inverters, since actions they take to solve grid problems can reduce the power output that they get paid for.
“We know there’s going to be an evolution of the commercial arrangements” to tap smart inverters for grid use, Jess Melanson, Utilidata’s chief operations officer, said in an interview last month. “But there’s too much value there for something like a solar panel not to capture it.”
Distributed computing to solve grid edge problems
So how does RT-OPF help solve these DER challenges?
“Let’s start with the acronym,” Marissa Hummon, Utilidata’s chief technology officer and former NREL researcher, said in last month’s interview.
“RT — real time — means, in this case, every second,” in how fast a utility gets information collected from DERs, although devices behind-the-meter are acting much faster to manage sub-second changes.
“OPF means optimal power flow. This optimization sits on the grid side as well as on the DER side; it faces both directions.”
Beyond NREL's own testing, the first real-world implementation of RT-OPF is with Colorado energy cooperative Holy Cross Energy at a newly built, all-electric powered affordable housing project in Basalt, Colorado, featuring solar panels, batteries and electric vehicle chargers. NREL added microcontrollers running RT-OPF to each in-home device, as well as to the distribution grid transformers serving the homes.
“The part that faces the DERs knows which kinds of DERs are there, and assesses their capability: how flexible they are, what their operating constraints are, how a customer uses it,” Hummon said. “The technology that faces the grid takes all that information and presents the voltage, VARs, and power — reverse power flow, pushing or pulling. When you combine that with your neighbors’ data, you can get a picture of what’s really going on.”
Holy Cross and NREL ran several tests, including setting the co-op’s preferred load profiles at the transformer and asking DERs to match them and running a demand response event to minimize on-site loads. They also ran a 24-hour experiment to see what would happen in conditions where no power was flowing across the transformer, with homes neither drawing power from the grid or feeding power back.
Andrey Bernstein, NREL, senior researcher and co-inventor of RT-OPF, said that this is one of the more technically challenging set of conditions for distributed energy controls “because most of the elements inside the community are highly variable and uncertain.”
Passing clouds can reduce solar panel output unexpectedly, residents can turn air conditioners and appliances on and off without warning. “If you don’t’ have real-time systems to react to these changes, you will not be able to implement zero power,” Bernstein said in an interview.
Beyond keeping generation and consumption in balance, RT-OPF can tap both real and reactive power capabilities of connected smart inverters to correct for voltage and frequency mismatches between grid and home circuits that can cause protection equipment to trip offline or otherwise cause problems. That’s something that demand response or on-site generation controls that are focused on active power can’t do, he said. (NREL has reported extensively on the challenges real-world technology trials have run into on these fronts.)
“Another difference is in the fact that our platform is really designed for this highly distributed system,” Bernstein added. “It’s basically an edge computing platform. You have all these computations going on at the edge of the system, in people’s houses, on transformers."
"Maybe there is some coordination you need to do in the SCADA system, or in the cloud, or some other central location, but most of the computation is going on at the edge.”
From technical optimization to economic solutions
The new project with NYSERDA will take a similar approach to instrumenting and controlling solar inverters, Hummon said, in this case, at a community solar project in Clifton Park, N.Y., connected to a distribution circuit that Utilidata is already managing voltage on for utility National Grid. NREL is participating, as is Standard Solar, the owner/operator of the community solar site, and inverter vendor Chint Power Systems.
Beyond optimizing voltage and Volt-Ampere Reactive (VAR) control to maintain grid stability, the project will also be studying how those control capabilities could increase the hosting capacity of distribution circuit, or defer the need to make upgrades to them to allow more renewable energy to be interconnected to them. “We’re going to do some modeling upfront with NREL on what are the financial impacts to a solar developer, and how are those mitigated through optimization," Hummon said.
For C.J. Colavito, VP of engineering at Standard Solar, that’s a much more interesting problem to solve. “Inverters are really sophisticated solid-state devices,” he said in an interview. Multiple pilot projects have shown they can inject or absorb reactive power to manage grid voltage fluctuations and other imbalances, whether they’re caused by the solar PV systems themselves or not.
But “the only way to do it — and this is painful for us as a system owner — is to give some control of the asset you own to the utility, and then trust that they’ll use that control appropriately and wisely,” Colavito said. The problem is, every unit of reactive power a solar inverter provides to stabilize grid voltage reduces the amount of real power they put out — and solar systems get paid for real power, not reactive power.
Under its Rule 21 regulations, California already requires all new PV installations to come with smart inverters that comply with IEEE 1547 standards. Other state regulations, such as the New York Standardized Interconnection Requirements, are following suit. But the “smart” features being tapped today include only relatively simple automated features such as low-voltage ride-through, to prevent them from tripping off during relatively common voltage excursions and other avoidable problems.
Smart inverters also must be able to communicate with utility systems under IEEE 1547. But before utilities can start using those communications links to tap inverters’ more advanced reactive power controls, they’ll need to work with regulators and the solar industry to assure that those instructions don’t undermine the economics of solar systems.
“That’s part of this study — how to set up a range, or bracket, around what they can and cannot control, for an owner whose primary, or often only, source of revenue is generation,” Colavito said.
That kind of analysis will help determine how much control can be handed over to utilities; how it could replace some of the more expensive equipment upgrades that could otherwise be needed to allow solar systems to interconnect in the first place; and possibly how solar systems could be compensated for the reactive power they’ll be asked to provide.
“We want that integration, we want to see lower interconnection costs, higher penetration of solar on feeder and distribution circuits across the U.S. — but we want to know what that cost is," Colavito said.
States like Hawaii, California and Arizona, which are facing significant challenges from high solar penetration, are already pushing to incorporate these kinds of capabilities into grid operations, Utilidata’s Melanson said.
“Requiring smart inverters is a great thing. But unless you connect that to a system that can talk to and optimize them, you’re not getting the full value out of them.”